Laboratory techniques in biochemistry and molecular biology
6
LABORATORY TECHNIQUES IN BIOCHEMISTRY AND MOLECULAR BIOLOGY Volume 6 Edited by
T.S. WORK - N.I.M.R.,Mill Hill, London E. WORK - ‘East Lepe’, 60 Solent View Road, Cowes, Isle of Wight
Advisory board G. POPJAK - U.C.L.A. S . BERGSTROM - Stockholm K. BLOCH - Harvard University P . SIEKEVITZ - Rockefeller University E. SMITH - U.C.L.A. E.C. SLATER - Amsterdam
NORTH-HOLLAND PUBLISHING COMPANY AMSTERDAM . NEW YORK . OXFORD
Part I
Part I1
R. Hinton and M . Dobrota DENSITY GRADIENT CENTRIFUGATION T. Chard A N INTRODUCTION TO RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
1978 NORTH-HOLLAND PUBLISHING COMPANY AMSTERDAM . N E W YORK.OXFORD
0 ElsevierfNorth-Holland Biomedical Press, 1978 AN rights reserved. No parts of this publication may be reproduced, stored in a retrieval system. or transmitted, in any form or by any means, electronic. mechanical, photocopying. recording or otherwise, without the prior permission of the copyright owner.
ISBN - series: 072044200 I - volume 6 : 072044221 4
Published by: ELSEVIERINORTH-HOLLAND BIOMEDICAL PRESS 335 JAN VAN CALENSTRAAT, P.O.BOX 21 1 AMSTERDAM, THE NETHERLANDS
Sole disrributorsjor the U.S.A. and Canada: ELSEVIER/NORTH-HOLLAND INC 52 VANDERBILT AVENUE NEW YORK. N.Y. 10017
Printed in The Netherlands
Editors’ preface
Progress in research depends upon development of technique. No matter how important the cerebral element may be in the planning of experiments, a tentative hypothesis cannot be converted into an accepted fact unless there is adequate consciousness of the scope and limitation of existing techniques; moreover, the results may be meaningless or even positively misleading if the technical ‘know how’ is inadequate. During the past ten or fifteen years, biochemical methods have become specialized and sophisticated to such a degree that it is now difficult for the beginner, whether undergraduate, graduate or specialist in another field, to grasp all the minor but important details which divide the successful from the unsuccessful experiment. In order to cope with this problem, we have initiated a new series of Laboratory Manuals on technique. Each manual is written by an expert and is designed as a laboratory handbook to be used at the bench. It is hoped that use of these manuals will substantially reduce or perhaps even remove that period of frustration which so often precedes the successful transplant of a specialized technique into a new environment. In furtherance of this aim, we have asked authors to place special emphasis on application rather than on theory; nevertheless, each manual carries sufficient history and theory to give perspective. The publication of library volumes followed by pocket paperbacks is an innovation in scientific publishing which should assist in bringing these manuals into the laboratory as well as into the library. In undertaking the editing of such a diverse series, we have become painfully conscious of our own ignorance but have been enV
VI
EDITORS PREFACE
couraged by our board of advisers to whom we owe many valuable suggestions and, of course, by our authors who have co-operated so willingly and have so patiently tolerated our editorial intervention.
T. S. & E. Work Editors
Contents of parts I and I1
PART I DENSITY GRADIENT CENTRIFUGATION. Richard Hinton and Miloslav Dobrota Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter I . Introduction to zonal centrifugation . . . . . . . . . . . . . Chapter 2 . Theoretical aspects of centrifugal separations . . . . . . . . . . Chapter 3. Conditions for a centrifugal separation . . . . . . . . . . . . . Chapter 4 . Centrifugation in conventional rotors . . . . . . . . . . . . . . Chapier 5 . Centrifugation in zonal rotors . . . . . . . . . . . . . . . . . Chapter 6. Assay of fractions separated by density gradient centrifugation . . . Chapter 7. Applications of density gradient centrifugation . . . . . . . . . Chapter 8. Artifacts arising during centrifugal separations . . . . . . . . . . Chapter 9. Future prospects for density gradient centrifugation . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix I1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix 111 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix IV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Subject index . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
3 8 46 70 97 120 197 205 243 254 262 263 265 269 271 274 287
PART I1 AN INTRODUCTION TO RADIOIMMUNOASSAY AND RELATED TECHNIQUFS. T. Chard . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291 Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293 299 List of abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . Chapter I . The background to radioimmunoassay . . . . . . . . . . . . . 301 Chaprer 2. Requirements for a binding assay - purified ligand . . . . . . . . 329 Chapier 3. Requirements for binding assays - tracer ligand . . . . . . . . . 343 VII
Vlll
CONTENTS OF PARTS I AND 11
Chapter 4. Requirements for a binding assay .the binder . . . . . . . . . . Chapter 5. Requirements for a binding assay .separation of bound and free ligand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 6. Requirements for a binding assay extraction of ligand from biological fluids . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chapter 7. Requirements for binding assays - calculation of results . . . . . . Chapter 8. Characteristics of binding assays - sensitivity . . . . . . . . . . Chapter 9. Characteristics of binding assays - specificity . . . . . . . . . . Chapter 10. Characteristics of binding assays - precision . . . . . . . . . . Chaprer 11. Characteristics of binding assays - relation to other types of assay . Chapter 12.Automation of binding assays . . . . . . . . . . . . . . . . Chapter 13. Organisation of assay services . . . . . . . . . . . . . . . . Appendix I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix I1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix I11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix IV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix V . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Subject index . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
311 401
~
427 440 446 463 419 49 5 504 510 518 520 521 522 523 521
531
DENSITY GRADIENT CENTRIFUGATION Richard Hinton and Miloslav Dobrota Wol/son Bioanalytical Centre, University o f Surrey,
Guildford, Surrey GU2 S X H , U.K.
This Page Intentionally Left Blank
Contents
Chapter 1 . Introduction to zonal centrifugation . . . . . . . .
8
1.1. The first applications of centrifugation in biology . . . . . . . . . . . 1.2. Centrifugal techniques . . . . . . . . . . . . . . . . . . . . . . 1.2.1. Analytical ultracentrifugation . . . . . . . . . . . . . . . . . . 1.2.2. Differential pelleting . . . . . . . . . . . . . . . . . . . . . 1.2.3. Rate-zonal centrifugation . . . . . . . . . . . . . . . . . . . 1.2.4. lsopycnic zonal centrifugation . . . . . . . . . . . . . . . . . 1.3. The development of centrifuges and rotors . . . . . . . . . . . . . . 1.3.1. The centrifuge . . . . . . . . . . . . . . . . . . . . . . . . 1.3.2. Rotor materials . . . . . . . . . . . . . . . . . . . . . . . . 1.3.3. Rotorshape . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.4. Zonal rotors . . . . . . . . . . . . . . . . . . . . . . . . . 1.4. Uses and limitations of centrifugal techniques . . . . . . . . . . . . . 1.5. Design of a centrifuge laboratory . . . . . . . . . . . . . . . . . . 1.6. Safety ofcentrifuges . . . . . . . . . . . . . . . . . . . . . . . . 1.6.1. Causes and prevention of rotor failure . . . . . . . . . . . . . . 1.6.I .1 . Overspeeding . . . . . . . . . . . . . . . . . . . . . . . 1.6.1.2. Corrosion . . . . . . . . . . . . . . . . . . . . . . . . 1.6.1.3. Fatigue . . . . . . . . . . . . . . . . . . . . . . . . . 1.6.1.4. High gradient densities . . . . . . . . . . . . . . . . . . . 1.6.1.5. Vacuum failure . . . . . . . . . . . . . . . . . . . . . . 1.6.1.6. Freezing of rotor contents . . . . . . . . . . . . . . . . . . 1.6.2. Other precautions in operation of centrifuges . . . . . . . . . . . 1.6.2.1. Access to spinning rotor . . . . . . . . . . . . . . . . . . 1.6.2.2. Aerosol formation . . . . . . . . . . . . . . . . . . . . . 1.6.2.3. Electrical safety . . . . . . . . . . . . . . . . . . . . . . 1.6.2.4. Balancing and assembly of rotors . . . . . . . . . . . . . . 1.6.2.5. Compatability of centrifuge and rotor . . . . . . . . . . . . . 1.6.2.6. Water leaks . . . . . . . . . . . . . . . . . . . . . . . 1.6.2.7. Flammable solvents . . . . . . . . . . . . . . . . . . . .
8 9 9 11 13 15 16 16
3
18
19 20 24 27 29 29 30 32 32 33 33 33 34 34 34 35 35 35 35 36
4
DENSITY GRADIENT CENTRIFUGATION
1.7. Care of rotors . . . . . . . . . . . . . . . . . . . . . . . . . . 1.7.1. Choice of rotor . . . . . . . . . . . . . . . . . . . . . . . . . 1.7.2. Properties and care of materials used in rotor construction . . . . . . 1.7.3. Washing and drying rotors . . . . . . . . . . . . . . . . . . . 1.7.4. Assembly, disassembly and storage . . . . . . . . . . . . . . . 1.7.5. Care of individual rotors . . . . . . . . . . . . . . . . . . . . 1.8. Guarantees on rotors . . . . . . . . . . . . . . . . . . . . . . .
36 36 36 40 41 41 43
Chapter 2 . Theoretical aspects of centrifugal separations
46
. . . .
2.1. Theory of rate-zonal separations . . . . . . . . . . . . . . . . . 2.1.1. Theory of sedimenting particles . . . . . . . . . . . . . . . . 2.1.2. Stability of the sample zone . . . . . . . . . . . . . . . . . . 2.1.3. Stability of a sedimenting zone . . . . . . . . . . . . . . . . 2.2. Theory of isopycnic banding . . . . . . . . . . . . . . . . . . . . 2.2.1. Shape of zones separated by isopycnic banding . . . . . . . . . 2.2.2. Redistribution of density gradient solutes in a centrifugal field . . . 2.2.3. Influence of the rotor on gradient shape . . . . . . . . . . . . 2.3. Effects of density gradient solutes on subcellular structures . . . . . .
Chapter 3 . Conditions for a centrifugal separation
. . .
. . .
46 47 51 57 59 59 60 64 64
. . . . . .
70
. .
3.1. Choice of approach . . . . . . . . . . . . . . . . . . . . . . . . 70 3.2. Choiceofrotor . . . . . . . . . . . . . . . . . . . . . . . . . . 75 3.3. Density gradient solutes . . . . . . . . . . . . . . . . . . . . . . 79 3.3.1. Salts of alkali metals . . . . . . . . . . . . . . . . . . . . . . 80 3.3.2. Small hydrophilic organic molecules . . . . . . . . . . . . . . . 81 3.3.3. High molecular weight organic compounds . . . . . . . . . . . . 86 3.3.4. Other types of density gradient solutes . . . . . . . . . . . . . . 87 3.3.5. Choice of gradient material . . . . . . . . . . . . . . . . . . . 89 3.4. Choice of gradient . . . . . . . . . . . . . . . . . . . . . . . . 90 3.4.1. Gradient for rate-zonal separations . . . . . . . . . . . . . . . 90 3.4.2. Gradient for isopycnic separations . . . . . . . . . . . . . . . 91 3.4.3. Design of complex gradients . . . . . . . . . . . . . . . . . . 93
Chapter 4 . Centrifugation in conventional rotors 4.1. Rate-zonal centrifugation . . . . . . . . . . 4.1 .1. Preparation of the density gradient . . .
. . . . . . .
97
........... . . . . . . . . . . . .
97 97
CONTENTS
4.1.2. Layering of sample on to the gradient . . . . . . . . . . . . . . 4.1.3. Centrifugation . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.4. Recovery from the gradient . . . . . . . . . . . . . . . . . . 4.1.5. Monitoring the displaced gradient . . . . . . . . . . . . . . . . 4.2. Isopycnic zonal centrifugation . . . . . . . . . . . . . . . . . . . 4.2.1. Preparation of the density gradient . . . . . . . . . . . . . . . 4.2.2. Layering of sample on to the gradient . . . . . . . . . . . . . . 4.2.3. Centrifugation . . . . . . . . . . . . . . . . . . . . . . . . 4.2.4. Displacement and monitoring of the gradient . . . . . . . . . . .
5 103 105 108 112 115 115 117 118 118
Chapter 5 . Centrifugation in zonal rotors . . . . . . . . . . . . 120 5 . I . Conventional. non-reorienting zonal rotors . . . . . . . . . . . . . . 5.1.1. Construction of rotors . . . . . . . . . . . . . . . . . . . . . 5.1.2. Types of zonal rotors . . . . . . . . . . . . . . . . . . . . . 5.1.2.1. A-XI1 rotor . . . . . . . . . . . . . . . . . . . . . . . 5.1.2.2. Z-15rotor . . . . . . . . . . . . . . . . . . . . . . . . 5.1.2.3. HS rotor . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.2.4. B-IVrotor . . . . . . . . . . . . . . . . . . . . . . . . 5.1.2.5. B-XIV and B-XV rotors . . . . . . . . . . . . . . . . . . 5.1.2.6. B-XXIX and B-XXX rotors . . . . . . . . . . . . . . . . 5.1.2.7. Interchangeable batch and continuous-flow rotors . . . . . . . 5.1.2.8. Continuous-flow rotors . . . . . . . . . . . . . . . . . . . 5.1.2.9. Other rotors . . . . . . . . . . . . . . . . . . . . . . . 5.1.3. Operation of zonal rotors . . . . . . . . . . . . . . . . . . . 5.1.3.1. Preparations for a run . . . . . . . . . . . . . . . . . . . 5.1.3.2. Gradient loading . . . . . . . . . . . . . . . . . . . . . . 5.1.3.3. Sample loading . . . . . . . . . . . . . . . . . . . . . . 5.1.3.4. Acceleration spin and deceleration . . . . . . . . . . . . . . 5.1.3.5. Unloading . . . . . . . . . . . . . . . . . . . . . . . . 5.1.3.6. Monitoring and fraction collecting . . . . . . . . . . . . . . 5.1.4. Ancillary equipment . . . . . . . . . . . . . . . . . . . . . . 5.1.4.1. Gradient makers . . . . . . . . . . . . . . . . . . . . . 5.1.4.2. Pumps . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.4.3. Refractometers and density meters . . . . . . . . . . . . . . 5.1.4.4. Spectrophotometers, colorimeters and UV meters . . . . . . . 5.1.4.5. Flow-through cells . . . . . . . . . . . . . . . . . . . . 5.1.4.6. Thermometers . . . . . . . . . . . . . . . . . . . . . . 5.1.4.7. Perfusor pumps . . . . . . . . . . . . . . . . . . . . . . 5.1.4.8. Stroboscopic lamps . . . . . . . . . . . . . . . . . . . . 5.1.4.9. Recorders . . . . . . . . . . . . . . . . . . . . . . . .
122 122 130 130 134 134 135 136 139 140 141 142 143 144 153 156 163 168 174 180 180
182 183 185 185 186 186 186 186
6
DENSITY GRADIENT CENTRIFLIGATION
5.1.4.10. Cooling equipment . . . . . . . . . . . . . . . . . . . . . 5.1.4.11. Minor accessories . . . . . . . . . . . . . . . . . . . . 5.1.4.12. Fraction collectors . . . . . . . . . . . . . . . . . . . . 5.1.4.13. Integrator . . . . . . . . . . . . . . . . . . . . . . . . 5.2. Reorienting zonal rotors . . . . . . . . . . . . . . . . . . . . . .
187 187 188 189 189
Chapter 6 . Assay of fractions separated by density gradient centrifugation . . . . . . . . . . . . . . . . . . . . . . 197 6.1. 6.2. 6.3. 6.4.
Enzyme and chemical assays on fractions . . . . . . . . . . . . . . Electron microscopic examination of fractions . . . . . . . . . . . . Assessment of results from density gradient separations . . . . . . . . Calculation of sedimentation coefficients . . . . . . . . . . . . . . .
197 198 200 202
Chapter 7. Applications of density gradient centrifugation . . . . 205 7.1. Separation of living cells . . . . . . . . . . . . . . . . . . . . . . 7.2. Separation of cell organelles from mammalian tissues . . . . . . . . . 7.2.1. Subfractionation of a crude nuclear fraction and separation of large sheets of plasma membranes . . . . . . . . . . . . . . . . . . 7.2.1.1. Subfractionation of purified nuclei . . . . . . . . . . . . . . 7.2.2. Subfractionation of the mitochondria1 fraction . . . . . . . . . . 7.2.3. Subfractionation of the lysosomal fraction . . . . . . . . . . . . 7.2.4. Subfractionation of microsomes . . . . . . . . . . . . . . . . . 7.2.5. Fractionation of ribonucleoprotein particles . . . . . . . . . . . . 7.2.6. Fractionation of chromatin . . . . . . . . . . . . . . . . . . . 7.3. Separation of subcellular structures from plant cells . . . . . . . . . . 7.4. Separation of subcellular structures from unicellular organisms . . . . . 7.5. Fractionation of macromolecules . . . . . . . . . . . . . . . . . . 7.5.1. Nucleic acids . . . . . . . . . . . . . . . . . . . . . . . . . 7.5.2. Proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.6. Other applications of density gradient centrifugation in biochemistry . . . 7.6.1. Fractionation of serum lipoproteins . . . . . . . . . . . . . . . 7.6.2. Separation ofviruses . . . . . . . . . . . . . . . . . . . . . . 7.7. Other applications of density gradient centrifugation . . . . . . . . . .
205 208 210 211 212 214 217 222 225 226 229 232 232 235 237 237 239 242
7
CONTENTS
Chapter8 . Artifacts arising during centrijugal separations . . . 243 . . . . . . . . . . . . . . . . . . 8.1.2. Damage due to high hydrostatic pressure . . . . . . . . 8.1.3. Damage due to high concentration of the gradient solute .
8 . I . Damage to particles during centrifugation 8.1.1, Damage caused by pelleting . . . .
. . . .
. . . .
. . . .
. . . .
. . . .
8.2. Factors affecting the accuracy of assays performed on fractions from density gradients . . . . . . . . . . . . . . . . . . . . . . . . . 8.2.1. Reaction of the gradient solute with reagents used in the assay of separated constituents . . . . . . . . . . . . . . . . . . . . . 8.2.2. Interference with the performance of analytical equipment . . . . . 8.2.3. Interference in the sensitivity of assays . . . . . . . . . . . . . . 8.3. Uncertainties in estimates of particle density and sedimentation coefficient 8.3.1. Particle density . . . . . . . . . . . . . . . . . . . . . . . . 8.3.2. Sedimentation coefficient . . . . . . . . . . . . . . . . . . . .
243 244 245 245 246 241 248 249 250 250 251
Chapter 9 . Future prospects for density gradient centrifugation . 254 9.1. 9.2. 9.3. 9.4.
Centrifuge design . . . . . . . . . . . . . . . Developments in centrifuge rotors . . . . . . . Developments in ancillary systems . . . . . . Uses of centrifugal methods . . . . . . . . . .
. . . .
. . . .
. . . .
. . . .
. . . . . . 255 . . . . . . . 257 . . . . . . . 259 . . . . . . 260
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . .
262
Appendix I . Manufacturers and suppliers of centrifuges. ancillary equipment and special chemicals . . . . . . . . . . 263 Appendix II . Glossary of terms used in density gradient centrifugation . . . . . . . . . . . . . . . . . . . . . 265 Appendix III . Density and sedimentation coefficients of rut liver cell organelles . . . . . . . . . . . . . . . . . . 269 Appendix IV . Theory of preparation of density gradients . . . . 271 References
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
274
Subject index . . . . . . . . . . . . . . . . . . . . . . . . . . .
287
CHAPTER 1
Introduction to zonal centrifugation
1 .l. Thefirst applications of centrifugation in biology A pleasing surprise given to writers of this type of article is the discovery of useless, but to them novel, facts, like the use of centrifuges for separating biological structures by Miescher in 1872. Little attention was paid at that time to the structure of cells, and for many years the use of centrifuges was restricted to applications such as the separation of milk, the collection of precipitates and (from about 1930 on) the separation of large particles such as nuclei. At that latter period most biochemists directed their attention to the separation of purified fractions - especially enzymes - rather than to analysis of the structure of the cell. Accordingly, sophisticated analytical ultracentrifugeswere developed for testing the homogeneity of purified fractions, but preparative centrifugeswere chiefly used for the collection of precipitates. Two major advances paved the way for a more wide-spread use of centrifugal techniques. Firstly, the development of alloys with a high strength in relation to their density permitted centrifugation at high speed of much larger quantities of material. Secondly, the development of methods for examining biological specimens under the electron microscope (see Palade 1971) revealed the complexity of the internal structure of cells. Earlier workers had attempted to purify nuclei (Behrens 1932) and mitochondria (Bensley and Hoerr 1934) but light microscopy was the only method of assessing these preparations. Given the complexity of the structure of the cell and the essentially arbitrary nature of the fractions which were being 8
Ch. 1
INTRODUCTION TO ZONAL CENTRIFUGATION
9
separated, what was needed was not ‘pure’ preparations, but a systematic and quantitative study of the distribution of subcellular particles and of enzymes between different fractions. De Duve (1971) has stated his belief that the insistence of workers such as Claude and Schneider and Hogeboom (1951) on the necessity for quantitative experiments ensured that the better understanding of the internal structure of the cell was rapiddly followed by an understanding of the biochemical role of the component parts. These early studies were all performed using the technique which was later called differential centrifugation or, more properly, differential pelleting (Reid 1972b). This method was refined by De Duve and his colleagues, and the ‘mitochondrial’ fraction was resolved into a heavy fraction containing mainly mitochondria and a ‘light’ fraction enriched in lysosomes (de Duve and Berthet 1954; de Duve et al. 1955). The importance of these advances is seen in the vast volume of work on the fractionation of different types of cell which followed. While differential pelleting has been an enormously useful technique for cell fractionation, a number of workers realised that only particles differing considerably in size could be separated in this way. Alternative techniques, rate and isopycnic zonal centrifugation were, in fact, proposed at about the same time as the differential pelleting scheme was developed, but the application of these methods was limited by the apparatus available (Anderson 1956; Allfrey 1959; de Duve et al. 1959). The full potentiality of density gradient centrifugation only began to be realised with the development of high capacity swing-out rotors and zonal rotors during the early 1960’s.
I .2. Centrifugal techniques 1.2.1. Analytical ultracentrifugation As we have mentioned, early high speed centrifuges were mainly used for the study of ‘exthcts’ from cells. Both this use and metallurgical limitations meant ‘that the volume of specimen had to be minimised. Rotors were designed with transparent windows so that the distriSirhiri I iiidc,x p 297
10
DENSITY GRADIENT CENTRIFUGATION
bution of particles in the centrifugal cell could be examined during centrifugation. A typical example of such an analytical rotor is shown in Fig. 1.1. The cells in such a rotor are filled with a uniform suspension of the mixture to be analysed. The rotor is
Fig. 1 . 1 . (left) A rotor for an analytical ultracentrifuge. (right) A high-speed preparative centrifuge rotor (Beckman type 65) is shown for comparison.
accelerated to its operating speed. Particles move at a rate determined by their size and shape and by the centrifugal force. Thus, if the cell was initially filled with a uniform suspension (Fig. 1.2A1) containing only one type of particle, a clear zone will appear to move slowly down the cell as the particles that were in that region sediment away (Fig. 1.2B1). If a mixture of particles were initially present, each type of particle will sediment at its own speed, so that after centrifugation the distribution will be as shown in Fig. 1.2B2. If some generalised property such as refractive index or ultraviolet absorbtion is measured, then patterns similar to those shown in Fig. 1.2C will be obtained. If an optical system sensitive to changes of refractive index is used (Schlieren optics) then the output will show a series of peaks (Fig. 1.2D). The latter is the form of output normally chosen, but it is important to realize that these peaks do not represent a zone of particles moving down the cell through a clear supporting medium, but the ‘back end’ of a sedimenting block of
Ch. 1 la1
1
11
INTRODUCTION TO ZONAL CENTRIFUGATION lbl
IC)
.:*.. . ..-.. .’.*..
2m Centrifuaal field
I
Fig. 1.2. Separations in an analytical rotor. The particles are initially distributed uniformly through the cell (A). As centrifugation proceeds, the particles sediment down the cell. Each type of particle will sediment at a distinctive rate. Thus as each group of particles sediment, a series of interfaces will form (B). These interfaces will appear to sediment through the cell at the same speed as the particles with which they are associated. The interfaces may be detected either by measuring the distribution of particles through the cell using some property common to all the particles such as ultraviolet absorbance (C) or by using an optical system (Schlieren optics) which provides an output which is related to the rate of change of refractive index (D). The latter system is less sensitive, but the output is more easily interpreted.
particles. As will beseen later, the use of an initially uniform suspension and the measurement of the clearance from the cell circumvents many problems. Analytical ultracentrifugation has been greatly developed from this essentially simple basis, but not in such a way as to fall within the scope of this article. Readers who are interested will find more extended accounts in books and articles by Schachman (1959), Trautman (1964) and Bowen (1970). 1.2.2. Differentialpelleting Differential pelleting is similar in principle to separations in an analytical ultracentrifuge. The centrifuge tube is filled with a uniform suspension.During centrifugation particles move down the centrifuge tube and pellet on the bottom. Ideally, centrifugation is continued S ~ h Ipr,rd<.\1, 287
12
DENSITY GRADIENT CENTRIFUGATION
for just long enough to pellet all of the largest class of particles (Fig. 1.3C). This will yield a supernatant free from one type of particle which may then be centrifuged at a higher speed to separate the next largest type of particle and so on. Unfortunately, the pellets are not homogeneous. Suppose that of the three particles shown in the illustration (Fig. 1.3) the middle and the smallest-sized sediment at respectively half and one tenth of the rate of the largest particles. As the particles are initially uniformly distributed, one half of the middle-sized particles and one tenth of the smallest sized particles will be found in the pellet if centrifugation is continued for just long enough to sediment all of the largest particles (see Fig. 1.3C).
Fig. 1.3. Fractionation of particles by differential pelleting (from Anderson 1966a). For details see text.
Ch. 1
13
INTRODUCTION TO ZONAL CENTRIFUGATION
More prolonged centrifugation would result in even greater contamination. The separation achieved by differential centrifugation may be improved by 'washing' the pellets. The pellets are resuspended in the homogenisation medium and recentrifuged under the same conditions as in the original pelleting. In the example used above, the pellet C contained half of the middle sized particles and one tenth of the smallest sized particles. If this pellet were resuspended in the same volume of liquid as the original suspension and were recentrifuged in the same way asbefore, all the largest particles would be recovered in the pellet, but only one half of a half (i.e. 25%) of the middle sued particles and one tenth of a tenth (i.e. 1%) of the original amount of the smallest particles. Hence, differential pelleting with washing of the pellet is an efficient way of separating particles which differ greatly in size, but not for separating particles of similar size. It is for this reason that differential pelleting has been effectively limited to the separation of the five fractions described by de Duve and colleagues in 1955. If greater resolution is required, other techniques must be employed. 1.2.3. Rate-zonal centrifugation The technique of rate-zonal centrifugation was first proposed by Brakke (1951). In essence, the technique is very simple. A small volume of a suspension is layered over a shallow density gradient. The latter is required to stabilize the sedimentation of the particles (see 9 2.1.3). On centrifugation, particles move away from the starting zone with velocities determined both by their size and shape and by the centrifugal force to which they are subjected. After centrifugation for a certain time, particles will be found in a series of zones spaced according to the relative velocities of the particles (Fig. 1.4). In this way particles differing in sedimentation rate by 20% or less can be separated without undue difficulty. Rate-zonal centrifugation thus complements differential centrifugation. As with most simple-sounding techniques there were a number of problems which limited the usefulness of the technique especially SlIhl'Y I Ill",.\
p 287
14
DENSITY GRADIENT CENTRIFUGATION
Fig. 1.4. Fractionation of particles by rate-zonal centrifugation in a swing-out rotor. The gradient and sample layer are introduced at rest. The tubes are attached to the rotor and accelerated, swinging out to a horizontal position (B). Zones of particles move down the tube at their characteristic rates. Centrifugation is continued until an adequate separation is obtained. The rotor is then decelerated to rest, and the gradient and separated zones: sample zone SZ, small particle zone (SPZ) and large particle zone (LPZ) recovered, for example by dripping out the gradient through a small hole punctured in the bottom of the tube (C) (redrawn from Anderson 1966a).
in the ten years following its introduction (see Brakke, 1960, for a survey of work carried out in this period). Rate-zonal separations cannot be carried out in angle head rotors as the sample mixes with the gradient during acceleration of the rotor (0 3.2) and until recently, the capacity of swing-out rotors was severely limited. As the sedimenting zones are as broad as, or broader than, the starting zone, the volume of material which can be loaded onto a rotor is limited if any degree of separation is to be achieved. In addition, the concentration of material in the sample must not be too high, or the entire band will mix with the top part of the gradient (see 0 2.1.2). Rate-zonal centrifugation was initially used mainly for analytical separations such as the analysis of the size distribution of samples of polysomes (McQuillen et al. 1959) or of RNA (Nomura et al. 1960) although very soon after the introduction of the technique
Ch. 1
INTKODUC'TION TO ZONAL CKNTRIFUGATION
15
Thomson and his colleagues (Thomson and Mikuta 1954; Thomson and Klipfel 1958) used rate sedimentation to separate mitochondria and lysosomes. Improvement in the design of swing-out rotors and especially the introduction of zonal rotors has greatly extended the application of rate-zonal centrifugation and it is with this method that we shall be mainly concerned. Vertical tube rotors (see p. 261) may also be used.
I .2.4. Isopycnic zonal centrifiigation Subcellular particles may differ not only in size but also in density, as was first demonstrated by Harvey (1931, 1932). A suspension containing living cells was layered over a solution of greater density than the cells. On centrifugation, the cells banded at the interface and, when they were examined in the microscope, the contents of the cell appeared to have separated into a number of layers. This separation was not lethal to the cells, but if the speed of centrifugation was increased, the cell broke in two, the nuclei being associated with the denser fragment. These density differences were exploited in order to purify nuclei by flotation in organic solvents (Behrens 1938) and, much later, by sedimentation through dense sucrose (Chauveau et al. 1956). Most early separations were performed by suspending the particles in a liquid of defined density and centrifuging. Particles less dense than the medium floated, particles more densepelleted. Fractionations by stepwise flotation are tedious, especially when more than one particle type is to be separated. Later investigators therefore developed the technique of isopycnic zonal centrifugation (Meselson et al. 1957; de Duve et al. 1959). A suspension of the particles to be separated is either layered over a density gradient or the particles are actually suspended in the solutions used to make the gradient. On centrifugation, particles either rise or sediment until they reach a liquid of their own density. Here they have no weight and do not move any further regardless of the time of centrifugation. The particles may be recovered as a series of zones, each particle at its own density. This technique, first used in analytical ultracentrifuges, is free from orh,'ll
,,,
16
DENSITY GRADIENT CENTRIFUGATION
many of the problems of rate zonal centrifugation. In 1959, Beaufay and colleagues, using this technique, showed that a group of enzymes involved in the formation and breakdown of hydrogen peroxide which sedimented like lysosomes on differential centrifugation were, in fact associated with quite distinct particles the microbodies (or peroxisomes). Since that time, isopycnic zonal centrifugation has been widely applied to biological separations and will be discussed in some detail below (Chs. 2, 3 and 8). However, the technique has one major disadvantage as compared with rate-zonal centrifugation. During isopycnic separations, particles are inevitably exposed to high concentrations of the gradient solute. This may cause damage to the particles and the possible appearance of extra bands which correspond to the damaged particles. ‘Separations’ may thus be achieved which bear no relation to differences which actually exist in the living cell. These problems are considered in some detail in later sections.
1.3. The development of centrifuges and rotors I .3.I . The centrifige Since we are primarily concerned with high-speed centrifugation, we will not discuss the relatively slow-speed bench-type centrifuges in any detail. These have been commercially available from the middle of the 19th century in the form of hand driven machines and (from 1911-1912) electrically driven. These machines were similar to present day bench centrifuges and capable of similar speeds, i.e. up to 3000 revs/min. The feature of most interest is the use of a swingout four place head, a design which was not adopted for high-speed centrifugation until much later. In order to appreciate the problems, let us examine the major design requirements for a modern high-speed centrifuge. 1) a reliable and accurately controllable high-speed drive system is needed ; 2) since air friction at speeds above 20,000 revs/min causes excessive heating, a rotor operated above 20,000 revs/min must be spun in a partial vacuum; 3) the rotor must be made of a material strong enough
Ch. 1
17
INTRODUCTION TO ZONAL CENTRIFUGATION
to withstand the very high centrifugal fields generated at such speeds. Finally, the machine must have an effective armoured shield to contain the rotor fragments in the event of a blow-up. Although centrifuges have improved over the years, it is probably fair to say that the technology to produce high-speed centrifuge drives (i.e. capable of generating over 250,000 g with a suitable rotor) has existed since the 1920's. Evidence for this claim is apparent from the report by Beams, Pickels and Weed in 1934 that their air turbine centrifuge had attained speeds of 1,200,000 revs/min. The limitation in this early period in the development of centrifuges was almost certainly the lack of metallurgical know-how to produce a rotor material strong enough for these high centrifugal forces. The rotor used in the experiments mentioned above was only 9 mm in diameter. What was needed was a material with an exceptionally high strength to weight ratio. As described in 5 2.1.1, the rate of sedimentation of a particle in a gravitational field is related to the size of the particle by Stokes law. This relationship was used by a number of workers to determine the size of small particles (see Svedberg and Pederson 1940). It was rapidly realised that the weakness of the earth gravitational field limited the size of particles which could be examined in this way, but that measurements could be made if the particles were subjected to a centrifugal field. The first attempts towards this end were made by Dumanskii (1 913) using ordinary laboratory centrifuges, but rather poor results were obtained, probably due to convection currents. The technique of ultracentrifugation was further developed in the 1920's and 1930's. By 1926,rotors reaching 5,OOOg had been developed which were used for measurement of the molecular weight of haemoglobin (Svedberg and Fahraeus 1926). By 1934, rotors capable of reaching 900,000 g had been developed (Svedberg et al. 1934) but these exploded after a few runs, and were very small. Larger rotors run at lower speeds providing centrifugal fields of about 260,000 g were found more suitable for routine work. The development and the testing of these early centrifuge systems is described in a fascinating and now classical book by Svedberg and Pederson 'The UltraS,,hl'< I
Ill< ! p
287
18
DENSITY GRADIENT CENTRIFUGATION
centrifuge’ (1940). These early machines were designed mainly for the study of sedimentation rates and measurement of the molecular weights of macromolecules such as proteins and could handle only very small volumes. Early analytical ultracentrifuges were all driven by some form of turbine, either air turbines (Bauer and Pickels 1940) or oil turbines (Svedbergand Pederson 1940). By contrast, the preparative centrifuges of the period were essentially scaled-up bench centrifuges with a shaft directly driven from an electric motor. These simple machines were the direct ancestors of the high-speed preparative centrifuges developed from 1940 onwards. Some machines have retained a mechanical drive through a gearbox, but there has been an enormous improvement in the speed control systems and in the rotors. More recently the air and oil turbine drives have been reintroduced on machines manufactured respectively by Electro-nucleonics and Sorvall. It will be interesting to see how successful such machines are (although it is likely that the noise of air turbines will make them unacceptable except for special purpose machines which can be housed in a separate laboratory). 1.3.2. Rotor material The rotors used by Svedberg and his colleagues were made of high tensile steel. However, other workers in the same period preferred to use aluminium alloy since this was cheaper, easier to machine and, due to its lower mass, caused less damage to the armoured shield when it did finally blow up (Bauer and Pickels 1940). Both of these materials were only strong enough for the relatively small analytical rotors but since then stronger aluminium alloys which can withstand higher stresses have been developed. The development of techniques for refining and machining titanium have now permitted its use for centrifuge rotors. Titanium alloys have a considerably greater strength to weight ratio than even the best aluminium alloy, so that a titanium rotor may withstand almost twice as large a centrifugal field as an equivalent aluminium rotor. In addition, titanium has good chemical resistance and reasonably good fatigue resistance. In spite of its
Ch. I
INTRODUCTION TO ZONAL CENTRIFUGATION
19
expense, difficulty in working and difficulties of supply, titanium has become a favourite rotor material and all the major manufacturers now offer a range of titanium rotors. The properties of titanium and aluminium rotors are discussed further in 5 1.5. Possible developments in centrifuge rotors in the future are considered in Ch. 9. 1.3.3. Rotor shape The first high-speed preparative rotors were introduced about 1940 and had a series of tubes held at an angle to the axis - hence the name angle head rotor (Pickels 1943). These rotors were designed for maximum strength and are primarily suitable for pelleting (differential centrifugation) although they may also be used for isopycnic banding (see 0 2.2.3). Modern angle head rotors show considerable variations in shape, the tubes being held at angles of between 14" and 40" to the axis of the rotor. In general, the highest speeds are developed with \the rotors whose tubes are most nearly vertical, but the best separations are achieved with rotors whose tubes lie at the greatest angle to the axis of rotation (but see 5 3.2). At high centrifugal fields, tubes in angle head rotors are subjected to considerable strain due to the outward pressure of the inner wall of the tube. This tends to result in the collapse of the tube, especially if it is not completely filled with liquid. This problem has become especially serious with the introduction of very high-speed centrifuge rotors (up to 75,000 revs/min) and, in fact, the speeds of angle rotors may well be limited not by the strength of the rotor material but by the strength of the plastic used to make the centrifuge tube. It was early realised that the sedimentation of particles in anglehead rotors could be subject to many artefacts such as convection currents. Although later experiments (Anderson 1969b) have shown that the importance of these artefacts can be overrated, at the time they were important in leading workers to examine the possibilities of using swing-out buckets, as in the early bench centrifuges, for high speed centrifugation (Kahler and Lloyd 1951). Hogeboom and Kuff (1954) later demonstrated that swing-out rotors are also suitable for analytical separations. Today a wide range of these rotors is S u C n f mi<,\ p 287
20
DENSITY GRADIENT CENTRIFUGATION
available, from the large scale 6 x 30 ml rotor to those with especially elongated tubes for high resolution rate-zonal separations or the small 3 x 4 ml rotors which are especially suited for rapid isopycnic banding of small particles, The construction of all these types of rotor is essentially similar. When the rotor is at rest the bucket is supported on either a steel pin (Fig. 1.5a and b) or rests in a socket on the rotor (Fig. 1 . 5 ~ ) When . the rotor is accelerated, the tube swings up on the pin to a horizontal position and locks into a socket in the main part of the rotor. The weight of the tube when the rotor is spinningat full speed is taken by the socket, not by the pin. If, however, the pin is bent or damaged in some other way, the tube may not swing up properly with disasterous consequences. The precautions to be taken in using swing-out rotors are discussed in more detail in (3 1.5. Angle-head rotors are used mainly for pelleting, swing-out rotors for separations on density gradients. 1.3.4. Zonal rotors
Once the potentiality of density gradient centrifugation for the separation of biological structures was realized, it became clear that the technique could be scaled up if the individual tubes could be replaced by a hollow chamber, thus making much better use of the internal volume of the rotor. Technically this was difficult to achieve and it took from 1955, when the validity of the ideas behind zonal rotors were first demonstrated, until 1964 for the group at the Oak Ridge National Laboratory under N.G. Anderson to produce rotors suitable for large-scale production. The developments and early experiments on the use of zonal rotors are described in a monograph published by the National Cancer Institute (Anderson 1966a) and in articles by Anderson (1962, 1966b). Commercially available zonal rotors and their operation are described in chapter 5 . Initially, the 'tubeless rotors' developed at Oak Ridge had only a single channel which led to the outside of the rotor. The sample had to be sprayed onto the surface of the gradient by a syringe (Fig. 1.6). This arrangement was obviously unsatisfactory and, after many trials, the coaxial Rulon seal (see Ch. 5) was developed. This seal was self-
Ch. I
INTRODUCTION TO ZONAL CENTRIFUGATION
21
Fig. 1.5. Somemodern swing-out rotors withdifferent methods of attaching the bucket to the rotor. In the older 3-place rotors (A) the bucket is held by a steel pin which is passed through the flange a t the top of the bucket and is screwed into the rotor. In modern 6-place rotors the geometry rules out this type of attachment and the buckets are attached to the rotors either by hooks which pass over steel pins in the rotor (B) or the top of the bucket is machined so as to fit into a spherical cup in the body of the rotor ( C ) . (Figures by courtesy of M.S.E. Ltd. The rotors illustrated are the 3 x 23 ml, the 6 x 16.5 ml and the 6 x 14 ml).
polishing, and the two channels allowed direct access to the centre or the edge of the rotor. The success of zonal rotors is almost certainly due to the development of this seal which is still used in almost its original form. S ~ VI N r,idc,.r p 287
22
DENSITY GRADIENT CENTRIFUGATION
Fig. 1.6. Schematicdrawingoftheoperationofanearlyzonalrotor(A-111). Thedensity gradient was fed into the rotor through the annular ring in the centre of the rotor and thence through small tubes to the rotor edge. The sample layer was then manually layered on the central (light) edge of the gradient. After separation of the particles, the gradient was recovered by displacement toward the central drain tube using a dense sucrose solution introduced to the rotor edge (from Anderson 1966a).
The workers in the Oak Ridge Laboratory originally concentrated their attention on two types of zonal rotor, the slow disc-like A series and the high-speed B series (Fig. 1.7). The first B type rotors were very tall in relation to their width and had to be supported by bearings at both top and bottom. This meant that the Rulon seal had to be incorporated into the top bearing which was also vacuumtight. The B-IV commercial batch type rotor which embodied this design is now obsolete. The same design principle was used for the early continuous flow rotors which were designed, above all, for the collection of viruses from large volumes of tissue culture medium. Although the B-IV rotor was produced commercially for some years, it required the continual attention of highly skilled staff for safe operation. It has therefore been superceded by the B-XIV and B-XV rotors. These rotors are much shorter in relation to their width than
Ch. 1
INTRODUCTION TO ZONAL CENTRIFUGATION
23
Fig. 1.7. An illustration of the variation in size and shape between different zonal rotors. From left to right the rotors are B-XIV, B-XV, B-IV, HS, A-XII. All are the models manufactured by M.S.E. Ltd with the exception of the B-IV rotor, which is the Beckrnan model.
the B-IV rotor (Fig. 1.7) and hence do not require a top bearing. The liquid feed head, which incorporates the Rulon seal, can therefore be constructed in such a way that it may be removed after loading the sample. The lid of the centrifuge may then be closed and the machine run in the normal way. The great advantage of this system is that the rotors can be used in the standard highspeed machines of today. Thus the B-XIV and the B-XV are the most widely used of all zonal centrifuge rotors. Another approach to zonal rotors, also developed by the group working under N.G. Anderson, has depended on loading gradient and sample into a stationary rotor and then reorienting the gradient by slow and controlled acceleration (see 0 5.2). The beauty of this system is that no rotating seals are required with the exception mentioned below. Re-orienting rotors have not been much used, and there are still some doubts about the usefulness of the procedure with rate separations where the resolution depends on avoiding any spreading of the sample band. Possibly as a result the only really successful reorienting zonal rotors in use today are the very large K Seh/rrr iida p 287
24
DENSITY GRADIENT CENTRIFUGATION
and RK continuous flow zonal rotors which are used for vaccine separation on a commercial scale. These rotors are continuous flow, and separation is necessarily based on isopycnic banding so that problems connected with the spreading of the sample band on reorientation are irrelevant.
I .4. Uses and limitations of centrifugal techniques It is useful now to discuss the types of separation which may be attempted by zonal (density gradient) centrifugation. If particles are to be separated by centrifugation, they must differ either in their density o r in their sedimentation coefficient. It should be remembered that there may be considerable variations in size and in density between two cells of the same type and similar variations occur with in the larger cell organelles. Thus, for example, the ‘sedimentation coefficient’ or ‘density’ of rat liver mitochondria, is an average. Actual mitochondria will form a distribution around this average but any two may differ considerably from each other. In some cases the sizes and densities of two functionally distinct groups of particles will overlap: thus large lysosomes are the same size and may be similar in density to small mitochondria. It will not be possible to separate such overlapping groups completely by centrifugation. Such heterodispersity within groups, and hence overlapping between groups seems to increase with increasing size of particle. Thus the size of macromolecules is usually precisely defined, and there is moderate overlapping between the different organelles of a cell, but there is very severe overlapping in size and density between the different types of cells in a tissue. Thus although it is, in theory, possible to separate whole cells by centrifugation, in practice this has so far proved useful in only a few cases (but see Ch. 7). Whole cells are probably the largest biological structures which may be effectively separated by zonal centrifugation. The lower limit on the size of particle for whose separation zonal centrifugation is the method of choice is governed by the increase in importance of the broadening of bands by diffusion. The broadening of the
Ch. 1
INTRODUCTION TO ZONAL CENTRIFUGATION
25
sedimenting bands limits the resolution which can be obtained by density gradient centrifugation. The most important alternative technique, polyacrylamide or agarose gel electrophoresis is, however, limited by the difficulty in handling the very dilute gels required for the separation of molecules with a molecular weight of more than one or two million although small polysomes have been resolved using mixed polyacrylamide-agarose gels (Dahlberg et al. 1969). The relative resolution of RNA and ribonucleoprotein particles by density gradient centrifugation and by electrophoresis in agar or agarose gels is illustrated in Fig. 1.8. Cytoplasmic RNA (MW 40,000-1.3 million daltons) is much better resolved by gel electrophoresis than by density gradient centrifugation. Ribosome subunits (MW about 1.2 and 3.1 million daltons) are almost equally well resolved by the two techniques whereas polysomes are much better resolved by centrifugation than by gel electrophoresis. We would conclude that density gradient centrifugation is not, generally, the best analyricul method for separating particles with a molecular weight of less than one million. However, there are cases where centrifugation may provide a useful method for analysing a mixture of smaller particles. This is especially so if it is necessary to carry out several different measurements to determine the distribution of separated components of the mixture. It is very much easier to split the liquid fractions from density gradients into several aliquots for analysis than it is to slice and extract an electrophoresis gel. Thus it may sometimes be convenient to use centrifugal methods even where gel electrophoresis would provide better resolution. The discussion above applies to analytical separations. When the aim is not to analyse the composition of a mixture, but to prepare one component, then density gradient centrifugation may be the preferred technique for smaller particles. It is difficult to obtain the resolution in preparative polyacrylamide gel electrophoresis which can be readily attained with analytical gels and there is often difficulty in recovering the material (see Gordon, this series Vol. I). Thus density gradient centrifugation is routinely used for the separation of haemoglobin messenger RNA (MW 210,000) (Williamson et al. F!d~,c< r erdck p 3 7
26
DENSITY GRADIENT C E N T R I F U G A T I O N
A260 nrn
A26Q nrn
4
Sedimentation
Electrophoresis
n
Sedimentation
0.75
-
lc)
Electrophoresis IC‘I
I
0.5c
0.25
C Sedimentation
Electrophoresis
Fig. 1.8. Comparison of the resolution of a.a’ RNA, b,b’ native ribosome subunits and c.c’ polysomes by density gradient centrifugation and by electrophoresis on either (a’ and b‘) agar gel or c’ agarose gel. The conditions for electrophoresis of RNA are given by Tsanev er al. (1965). for ribonucleoprotein particles by Dessev et al. (1969). Centrifugation wascarriedout either in the tubes o f a n M.S.E. 3 x 23 ml swing-out rotor (a and c) o r a B-XIV zonal rotor (b). Sucrose gradients ranging from 0.5-1.0 M were employed. The gradient solutions contained either a) 0.01 M Tris pH 7.4 or (b and c) 5 m M Tris pH 7.4 and 1 mM MgC12.
1971) and has been successfully used in the purification of S-nucleotidase from liver plasma membranes (Evans and Gurd 1973a, b). These two molecules are probably about the smallest for which it is worth while using density gradient centrifugation as a preparative
Ch. I
INTRODUCTION TO ZONAL CENTRIFUGATION
21
method. For smaller molecules gel filtration (Fischer 1970, Vol. 1 of this series) gives better resolution. For example tRNA and 5 S ribosomal RNA (MW 26,000 and 36,000) are better separated by gel filtration than by centrifugation (Galibert et al. 1965). It must be reemphasized, however, that it is possible to separate particles by density gradient centrifugation only if they possess different sizes, densities or both. Particles identical in size and density cannot be separated by centrifugation and some other technique must be employed. The magnitude of the difference in the sedimentation coefficient* between two particles which is necessary for separation by zonal centrifugation will depend on the skill of the operator. An experienced worker could probably separate particles whose sedimentation coefficient differs by 5% (No11 1969). Less experienced workers would probably find that difference of 10-20% in analytical separations and of 20-30"/, in preparative separations would be as much as they could achieve. Relatively better resolution can be attained in isopycnic gradient separations, for example DNA molecules differing in density by as little as 0.005 g/ml can be separated (Flamm et al. 1972). However, most large subcellular structures do not have precisely defined densities, but rather band over a range of 0.01-0.02 g/ml. Thus the separations which can be achieved in practice by isopycnic gradient centrifugation may be little better than can be achieved by rate separation. In conclusion, therefore, density gradient centrifugation is the method of choice for separating particles between the sizes of whole cells and of the largest macromolecules.
1.5. Design of a centrifuge laboratory Centrifuges can be positioned in the laboratory either centrally with access from all four sides or against a wall with access to three sides. * The sedimentation coefficient of a particle is a function related to the size and shape of the particle, as described in 5 2.1.1. I t describes the speed at which that particle will sediment in any particular centrifugal field. St,hw'~r , a / l , i 11 287
28
DENSITY GRADIENT CENTRIFU(IATI0N
The former may require overhead supply of water and power while the latter lay out is probably more practical in spite of the reduced access. The distance between machines should, if possible, be sufficient to take a trolley with the monitoring equipment. Most modern centrifuges will consume nearly 30 amps of current. The correct connection is a permanent, fused, and switchable outlet rather like an electric cooker outlet. For extra safety there should also be a separate isolator switch, preferably one for each machine, outside the laboratory so that in the event of a serious accident power can be switched off from outside. A cold water supply is needed for most high-speed machines to cool various components which generate frictional heat and also for the diffusion pump condenser. The supply must be constant in flow rate and pressure. The return pipe must be securely mounted to prevent a possible flood. For a recirculating system see Ch. 5. The laboratory must be well ventilated since centrifuges produce considerable quantities of heat. If only half the total power (i.e. 20 amps = 5 kW) is dissipated to the room as heat this represents about 23 kW per machine. Therefore, 5 centrifuges may generate 123 kW of unwanted heat. The exhaust from the centrifuge vacuum pump can produce an oil mist during a long run. This unpleasant and probably harmful contaminant should be removed either by an oil mist filter or via a manifold leading outside. The laboratory floor must withstand the loading imposed on it; as an example, the MSE SS75 machine weighs 1,900 Ib. or nearly 900 kg. Although not essential for normal centrifuge laboratories, a zonal laboratory should ideally have a continuous sealed floor coved at the corners. If possible a floor drain should also be fitted, thus allowing the whole floor to be hosed down in case of serious contamination. Washing facilities must be adequate, especially for zonal rotors. A stainless steel double sink (one for washing the other for rinsing) supplied with hot and cold water is useful. The size should be sufficient for the largest rotor available, i.e. the A-XI1 zonal rotor needs a sink of about 20 x20 ins. We do not recommend the installation of
Ch. 1
INTRODUCTION TO ZONAL CENTRIFUGATION
29
wooden racks for draining and drying of rotors since this can lead to the bad practice of always leaving them out and exposed. All rotors should be wiped dry before storage. Since it can be inconvenient to store rotors in their individual boxes they are best stored in a large cupboard. A large upright fridge or a nearby cold room are needed for precooling of rotors. Ideally the only access to such a laboratory should be from a corridor and not from another laboratory. No one should be permanently based in a centrifuge laboratory.
I .6. Safety of centrifuges Centrifuges can be dangerous if wrongly used or improperly maintained. The safety factor is strongly linked to care of machine and rotors, and it is therefore logical to discuss the two together. I t cannot be emphasised too strongly that carelessness will increase the chance of injury and lead to higher expenditure on repair and replacement. Fortunately centrifuge manufacturers do ensure that the possibility of injury is minimised. To avoid accidents all users of centrifuges should be persuaded to read the manuals and other publications referring to care and safety (Price 1972; Burns 1972) as part of a training scheme. It is also useful to have a centrifuge expert in the department who can give advice, train newcomers, do simple troubleshooting and be generally responsible for the machines and rotors. Such a person can often save money by doing simple repairs while also doing a favour to service engineers who hate being called out for such jobs as replacing fuses. In some cases the servicing department may be glad to give a course on maintenance. 1.6.1, Causes and prevention of rotor,fuilure
Since a rotor bursting at high speed could cause serious injury, all the major manufacturers fit armour-plated bowls designed to contain the fragments. No manufacturer quotes the ‘safety factor’ on high speed centrifuge rotors, presumably because it cannot be met. St,hw I stdc,\ p 287
30
DENSITY GRADIENT CENTRIFUGATION
According to the British Standard this factor should be 2, which means that a rotor must be tested at twice the maximum operating RCF for 10 min and should not stretch more than 0.1%. Testing of prototype rotors to destruction gives an indication of the strength and hence the safety margin. After this initial development regular checks are made by testing to destruction rotors from production batches. Since the destructive test is the only reliable one known, obviously individual rotors cannot be tested. The containment of rotor fragments was extensively studied in the early days by Svedberg and Peterson ( 1 940) and more recently by Anderson et al. (1969a). In our own experience, armoured bowls do work, and indeed give a feeling of security (see Fig. 1.19). The momentum of the spinning rotor as it bursts and suddenly comes to a halt (in fragments), may result in the centrifuge itself jumping violently. If a worker happened to be leaning on the machine at the time of a blow-up he or she could be thrown across the laboratory. Special anti-skid feet which reduce this movement in the event of a blow-up, are available (MSE Ltd.). This problem is overcome in the latest Beckman machines by not mounting the armour bowl rigidly, thus allowing it to spin with the fragments, and so absorb the torque of the blown up rotor. However, there are a number of older machines, whose armour plate may not be strong enough, and one can never be absolutely sure that a blow-up will not cause injury. Besides, a blow-up is likely to produce an expensive repair bill and should be avoided. I h . 1 .I. Overspeeding The specified maximum speeds of rotors are designed to give a wide margin of safety. The speed at which a rotor may explode is probably 25% higher than the recommended maximum. Most lower speed machines (upto about 18,000 revs/min) have a built-in safety factor, since the air resistance becomes so great that a rotor can not be driven much faster than its specified maximum. Also these rotors are rugged and the safety margin is greater than with the high speed rotors. All the high-speed centrifuges are fitted with an overspeed control
Ch. 1
INTRODUCTION TO ZONAL CENTRIFUGATION
31
Fig. 1.9. This rotor disintegrated at about 38,000.revs/min. The safety offered by the armour bowl (fitted to all modern ultracentrifuges) is evident from the fact that the explosion was contained and indeed would have been even with the fastest rotor. However, this should not encourage carelessness.
mechanism which switches the machine off, or slows the rotor down. Obviously a zonal rotor must not stop during a run, and machines which cut out entirely rather than decelerate a zonal rotor to unloading speed, are not designed with the zonal user in mind. With most ultracentrifuges the correct speed for a particular rotor is selected manually by the operator, and no damage will be caused if too high a speed is selected as the ‘overspeed’ will automatically stop the rotor. However, it is more satisfactory to have positive Sahrcl s i d r r p . 287
32
DENSITY GRADIENT CENTRIFUGATION
selection of the correct maximum speed for each rotor. On their SS40 and SS50 machines MSE use a system of levers and microswitches to select the correct maximum speed, e.g. with a 21,000 revs/ min rotor only that speed can be selected and the rotor physically prevents the levers being moved into any other speed position. The Superspeed 65 and 75 (MSE) use a mechanical means of measuring adisc, at the bottom of each type of rotor. The machine automatically translates the disc diameter into maximum speed for a given rotor as part of the starting sequence. 1.6.1.2. Corrosion Rotor corrosion is probably the most common cause of rotor failure. The manner in which corrosion takes place and the chemicals involved are covered in Q 1.7.2. Corrosion is a serious problem with aluminium rotors since these are attacked by acids and alkalis ; titanium is much more resistant. Corrosion should not occur if a rotor is well cleaned, dried, polished and stored. A rotor should not be used if there is any sign of corrosion, since it is unsafe and since the guarantee will be automatically invalidated. The first signs of damage to the anodising on aluminium rotors should be checked by the manufacturer, and the rotor should be reanodised if it is not damaged. Titanium rotors are usually painted black, but this is not a protective coat, and small chips in this paint are not serious. 1.6.1.3. Fatigue During acceleration a rotor stretches and on deceleration it contracts; it is essentially these two processes which cause fatigue. This physical property of the rotor results in the effective life of the rotor being related to the total number of runs. An indication of the ‘life’ of a rotor may be given by the length of the guarantee. In this respect one is very much in the hands of the manufacturers, some of whom appear to play very safe (see 5 1.8). It is important to record all runs, so that if a rotor does fail one has a full record of the rotor’s usage. Beckman take this very seriously and in their zonal manual actually define a run as ‘any operation
Ch. 1
33
INTRODUCTION TO ZONAL CENTRIFUGATION
of the rotor to 75% or more of maximum speed, no matter how brief. 1.6.1.4. High gradient densities Centrifuge rotors are not designed to contain liquids whose average density is greater than 1.2 g/ml. Higher densities introduce heavier loads on the rotor and the maximum speed must therefore be lowered according to the formula
New max. revslmin
=
1.2 x max. revs/min new density
This fact is well documented by all the major firms. With sucrose solutions it is unusual to have average densities of over 1.2 but other gradient materials such as CsCl and NaBr can give densities which require the maximum speed to be reduced. If this is not done the rotor could blow up. 1.6.1.5. Vacuumfailure Vacuum systems are generally reliable and non-hazardous under normal conditions. Zonal centrifugation, particularly with the removable feed head arrangement, does require that the lid of a vacuum-type centrifuge be open so that the spinning rotor is accessible for gradient loading and unloading. This is a system contrary to normal practice in modern machines, which use the vacuum system to initiate the whole process of starting a rotor. With some modern machines, however, it is possible accidentally to admit air into the centrifuge bowl while the rotor is spinning at high speed. In this case the rotor usually lifts off the shaft with disastrous consequences. Ideally of course it should not be possible to do this at any speed other than the loading, unloading speed. 1.6.1.6. Freezing of rotor contents It is possible in some cases to freeze the contents of a zonal rotor and cause the rotor to burst. The conditions in which this can occur are surveyed under temperature control in 0 5.1.3.4, where control systems are described in some detail. Siihirrr indm p 287
34
DENSITY GRADIENT CENTRIFUGATION
1.6.2. Other precautions in operation of centrifuges There are other aspects of centrifugation which are even more potentially dangerous than the bursting of a rotor. 1.6.2.1. Access to spinning rotor An effective lid lock should (although in some cases it does not) prevent the lid from opening or being opened while the rotor is spinning at high speed. Filling and emptying of zonal rotors is done while the rotor is spinning, and is the most critical and hazardous part of the procedure. In industry, rotating machines such as lathes or grinding wheels usually have effective guards, automatic cut-outs, and proniinently displayed warnings. We have not seen such a warning on a zonal centrifuge which can be run with the lid open. Although unlikely, rotating parts of the feed head could possibly catch and tear loose ends of clothing, particularly ties; or, even more dangerously, long hair. 1h.2.2. Aerosol formation A serious problem sometimes encountered with zonal rotors is the formation of aerosols (or atomised leaks). In the case of radioactive samples or infectious viruses, stringent control must be observed. If the rotating seal leaks, most of the material is thrown out onto the guard tray and the centrifuge bowl, forming a distinct ring which can later be washed or decontaminated. However, an unknown quantity of the leaked material will undoubtedly atomize and spread quickly in the atmosphere. An aerosol containing radioactive isotopes (typically soft and hard j-emitters such as used for biological tracer work) is only moderately hazardous and can be tolerated provided the machine and apparatus are well decontaminated. However, with pathogenic organisms and viruses, the hazard can be great and an aerosol must be contained. In this case the whole centrifuge and ancillary apparatus should be installed in a sterile cabinet. Such systems are quoted by Cho et al. (1966), Birnie (1967) and Webb et al. (1975). Electro-Nucleonics, in their K and RK systems filter and scrub the exhaust gas from the turbine drive in case this is contaminated.
Ch. 1
35
INTRODUCTION TO ZONAL CENTRIFUGATION
1.6.2.3. Electrical safety Most machines are well-designed and circuits are well insulated and protected. Users who attempt to do small repairs on their machines should however be careful, since there are always exposed terminals inside the centrifuge. No work should be done unless the machine is completely isolated electrically. Turning off the main power supply on the centrifuge itself is not sufficient, it should either be unplugged from the mains or disconnected (see also fi 1.5). 1.6.2.4. Balancing and assembly of rotors Zonal rotors do not require balancing, but the tubes of angle and swing-out rotors must be balanced, usually to within 0.1 g (but check with individual rotor instructions). With angle rotors the two balanced tubes must be placed opposite one another. The only other serious precaution is to ensure that the rotor lid is properly located and tightened. When balancing tubes for a three place swing-out rotor, all the three tubes must be balanced within 0.1 g. With six-place rotors, balanced pairs can be placed opposite one another. Generally swing-out rotors must be used with all their buckets in place, but some six-place rotors can be run with only one or two pairs in place. The manufacturer's instructions must be followed at this point. 1.6.2.5. Compatibility of centrifuge and rotor It most cases, safety features such as the overspeed protection do not allow low-speed rotors to be spun in high-speed machines at higher than the intended maximum speed, while allowing for a reasonable degree of interchangeability. Naturally this interchangeability applies only if both centrifuge and rotors are from the same manufacturer. A simple colour code, put on as dots of same colour on the rotor and appropriate centrifuge, can help to identify the correct centrifuge and speed. We personally prefer a foolproof system which would not physically permit any errors (i.e. MSE's SS 65 and 75, see 9 1.6.1.1). I .6.2.6. Water leaks High-speed centrifuges require coolant water for various bearings and gears and for the condenser of the diffusion Sirhpcr
I , I ~ C Yp
. 287
36
DENSITY GRADIENT CENTRIFUGATION
pump. The water supply needs to have a specified pressure and flowrate, and obviously all connections should be secured with pipe clips. 1 d.2.7. Flammable solvents Organic zonal gradients are usually made from halogenated compounds which are not flammable. Naturally these organic solvents cannot be used in the low-speed Perspex (lucite) rotors. The use of flammable solvents in low-speed centrifuges may result in a fire since these machines are electrically driven, and themotor brushes may spark. The centrifuge manufacturer must be consulted to check that such solvents can be used safely.
I .7. Care of rotors 1.7.1. Choice of rotor An integral part of rotor care should be the initial planning and choice of rotor. It should be remembered that the total investment in rotors could well be in excess of the price of the centrifuge. In that case, the whole system could well be chosen for the merit of the rotors. Questions the potential buyer should consider when choosing new rotors are : 1) What is the guaranteed life of the rotor? 2) What is the price related to the life? 3) What are the conditions of the guarantee? 4) Is there a free inspection service or testing of rotors? 5 ) What is the replacement price if an old rotor is traded in for a new one? 6 ) What are the specifications of compakble rotors of different makes? Once the choice has been made, the correct care and handling of a rotor will not only extend its life, but will have a direct bearing on safety. 1.7.2. Properties and care of materials used in rotor construction Aluminium is readily attacked by both acids and alkalis. Although anodising gives some protection, aluminium rotors are only safe in
Ch. 1
INTRODUCTION TO ZONAL CENTRIFUGATION
37
theneutralorphysiologicalpHranges6.5to 7.5. Aluminium is attacked very readily by alkaline detergents (e.g. Decon 90, RBS 25), some of which carry no warning of their alkalinity. The Beckman zonal manual (12-TB-059) actually recommends a non-alkaline detergent, Ivory Liquid (Proctor and Gamble), and mentions that most laboratory detergents are 'too strong' (presumably too alkaline) for anodised coatings. Caesium chloride causes invisible intercrystalline cracks after prolonged exposure. Therefore, it may only be used provided that all contact between aluminium and the caesium salt solutions is avoided, or at least minimised, and the rotor is thoroughly washed if a spillage is suspected. However it is worth noting that the Beckman manual states that aluminium zonal rotors should only be used with sucrose gradients in the pH range of 6.5 to 7.5; the guarantee is invalidated if other solutions are used. Aluminium is also attacked by citrates, tartrates and other salts. MSE advise that, since their aluminium rotors are heat treated for increased strength, they should not be subjected to temperatures above 70°C as this will weaken the metal. This means that they may not be autoclaved and should be sterilised with ethylene oxide, formaldehyde or ethanol. Beckman rotors may be autoclaved at temperatures up to 125 "C,but their life will be prolonged if alternative (chemical) methods of sterilisation can be used. IEC advice on the effect of heat is simply that their rotors must not be subjected to dry heat above 100 "C. This conflicting information from the three major manufacturers probably means that the rotors are made of slightly different alloys which require different annealing temperatures. Titanium. This metal has recently become popular for centrifuge rotors, because of its favourable strength to weight ratio and its good chemical resistance. Its disadvantages, the difficulties encountered in purifying and machining it, are reflected in the price of titanium rotors, which is about twice that of aluminium rotors. For rotor construction it is used as a titanium alloy which contains a substantial proportion of copper. The presence of copper may be SiihIect WdcX p 287
38
DENSITY GRADIENT CENTRIFUGATION
a disadvantage since it is known to inhibit a number of enzymes. The chemical resistance of titanium allows it to be used safely with CsCl gradients, and with acid and alkaline solutions within the pH range 3.5-1 1. The MSE manual on the use of zonal rotors (B-XIV, B-XV publication No. 49) gives a list of chemicals which may attack titanium viz. strong formaldehyde solutions, warm oxalic acid even at low concentrations, dilute mineral acids at concentrations above 5%, and some halogenated hydrocarbons which may cause stress corrosion. Titanium is also heat-treated for strength, but at much higher temperatures than aluminium and can therefore be safely autoclaved at about 1 2 0 k Noryl is a trademark of the General Electric Co. for polyphenylene oxide, a thermoplastic of good chemical resistance. It is used for the cores and septa in the various zonal rotors, and its strength, ease of machining and relatively low cost make it ideal for this purpose. The Beckman zonal manual contains a comprehensive list of chemicals which may or may not be used with noryl. It is resistant to such substances as the common acids and alkalis, most inorganic salts, and also acetone, ethyl acetate, cyclohexane, ethyl and propyl alcohols, hydrogen peroxide, ozone, phenol. It is not resistant to benzene, toluene, carbon tetrachloride, chloroform, heptane or petroleum ether. Rotors fittedwith noryl septa must not be spun above 5,000 revs/min if they are empty or only partly full as noryl is not strong enough to withstand the high centrifugal fields unless it is supported by liquid. This can be a disadvantage if only noryl cores are available, since occasionally trial runs with empty rotors can be quite useful (although in practice it is not too difficult to fill a rotor just for a ‘dummy’ run). Also it may not be possible to use organic solvent gradients if only noryl septa are available. Perspex. This transparent plastic (polymethymethacrylate) goes also under the trade mark Lucite or Plexiglas. It is attacked by nearly all organic solvents except the lower alcohols. It is ideal for the construction of low-speed rotors, but is relatively soft and can be scratched rather easily. Its use in MSE A-XI1 and HS rotors produces
Ch. 1
INTRODUCTION TO ZONAL CENTRIFUGATION
39
one operational peculiarity which must be emphasised. Perspex has a higher coefficient of expansion than aluminium, and therefore at room temperatures it is held so tightly in the aluminium clamp ring that it cannot be removed. The whole rotor must be cooled to about 5 "C, allowing the Perspex to contract, before dismantling (see also 5 1.7.4). Rulon is a trade name (Du Pont) given to a filled PTFE plastic. The filling is jeweller's rouge which gives it a self-polishing property when in contact with metal. It is chemically a most resistant plastic, and no special precautions need to be taken if organic gradient solutions are used. It is relatively soft and so the seal face can be damaged easily, even with a fingernail. However, this same softness allows it to be repolished easily to restore it to perfect condition. '0'-rings. These are usually made of synthetic butyl rubber, which has good resistance in most inorganic salts, but not to organic solvents. 0-ringsof Viton (a fluorinated elastomer which is resistant to organic solvents) can be supplied if organic gradients are used. Care should be taken when using CsCl gradients in B-type zonal rotors. Since this salt gives very heavy solutions which are only slightly higher in viscosity than water, blemishes or scratches will allow the caesium chloride to leak out of the rotor. This commonly occurs when the rotor is at maximum speed and the hydrostatic pressure on the outer edge O-ring is greatest. For safety, a new O-ring should be used with every CsCl run. In any case, all O-rings should be carefully greased with silicone grease to prolong their useful life. Other materials. The stationary part of the seal is normally made of stainless steel (in the case of Beckman, oxide-coated metal): but the steel seems to have been chosen more for its compatibility with the Rulon rotating seal than for its stainless quality. Its resistance to corrosion is very low. At the end of a run, the two surfaces of the seal should be washed carefully and thoroughly dried. If the seal is left wet, even overnight, signs of corrosion can be seen on the stainless stationary part. The modern feed-head in which the Rulon and stationary seal parts are both contained, must be dismantled at the end of each run and cleaned. With the earlier system, the action of SuhIc<~rude\ p 287
40
DENSITY GRADIENT CENTRIFUGATION
taking off the feed head actually parted the two seals, thus exposing them and leaving them visibly in need of washing and drying. During manufacture the four tubes connected into the stationary seal are pressed in, and then soldered. This solder has been known to break. It should be remembered that solder is composed of tin and lead, and if samples and gradients come into contact with it, the question of metal contamination should not be ignored. I .7.3. Washing and drying rotors All rotors should be washed with lukewarm water immediately after use and dried. A water wash is usually quite sufficient, but if a detergent must be used then it must not be alkaline. In such a case it is best to scrub a rotor well with a mild neutral detergent using a soft nylon brush, rinse well in water to remove all traces of detergent, and finally rinse with distilled water. It is very bad practice indeed to leave rotors soaking in a detergent bath in the hope that this will finally clean and decontaminate them. After rinsing, the rotors must be dried. The usual method is to leave the rotors to dry at room temperature, this being the most convenient practice, but it is much more satisfactory to wipe them dry and perhaps use a hair dryer to dry the inaccessible parts. The Perspex plates and aluminium rings (of the A-XII, HS and Z-15 rotors) should be washed thoroughly in cold water and dried rapidly by wiping with paper towelling. The centre core can be left in the sink with hot water running over it. The bowl of the B-type rotors can be similarly washed in a sink with warm water (below 70 "C) running over it for a few minutes. This washes the core and the rotor and leaves it warm ready for the next step. While still warm, the core is washed with acetone or ethyl alcohol which dry quickly, leaving no wet traps inside the channels for possible bacterial or fungal growth, and at the same time sterilise these otherwise inaccessible parts. These 'mild' organic solvents do not seem to effect the Rulon part or O-rings. If the grooves of the O-ring are not dry, bacteria or fungi may grow under the O-rings. To prevent this, the core and O-rings must be removed at regular intervals (preferably after
Ch. 1
41
INTRODUCTION TO ZONAL CENTRIFUGATION
every run) and the grooves cleaned, dried and greased lightly with silicone grease. After drying, the outside of rotors may be polished with a wax to protect the surface. Car polishes are not suitable since they contain a fine abrasive and a degreasing agent which may attack aluminium anodising. Obviously only the outside of zonal rotors should be polished. I .7.4. Assembly, dismantling and storage When fully assembled and ready for use, all zonal rotors must be tightened to the correct torque to avoid leakage of gradient from the outer edge of the rotor. If rotors are tightened and then stored for long periods, the edge seal O-rings will collapse and will need replacing. To avoid this, rotors should be left loosely assembled and the correct torque should only be applied immediately before use. Manufacturers recommend that rotors be stored dismantled, but this is rather impracticable. We find it most convenient to store them in their boxes, dry, assembled, but not fully tightened. We then pre-cool immediately before a run (see 0 5.1.3.1), tighten up to correct torque and use. The A-XI1 and HS rotors cannot be assembled while at room temperature, they may be tightened only when cold, otherwise the torque settings will be incorrect. For storage and precooling of the HS, Z-15 and B rotors it is best to keep the vacuum cap on, to seal the rotor and keep the inside clean and dry.
1.7.5. Care of individual rotors There are special points, characteristic of the various zonal rotors which need more detailed examination. A-XZZ. The centre core with the septa is a complex Perspex disc, with the feed to the edge of the rotor formed as four channels underneath (see Fig. 5.7 of HS rotor which is very similar). The centre part is sloped to allow the sample to be fed in evenly and form a band. This results in a piece of Perspex with many sharp corners in which fungal growth soon appears unless carefully cleaned and dried after each run. If the two aluminium clamping rings are Suhwl
w&,Y&
287
42
DENSITY GRADIENT CENTRIFUGATION
not thoroughly dried after washing the bolt holes form an ideal water trap for electrolytic corrosion. These holes should be washed with acetone then dried, and the bolts should be regularly greased with engine oil or grease (not silicone, which is not a good lubricant). H S and Z-15 rorors. The above points apply to these two rotors as well. The HS rotor has a large number of rather small Allen-headed (hexagonal recess) bolts. The stainless steel bolts are relatively soft, allowing the head to stretch so much, after some use, that the Allen key no longer fits. We have replaced these bolts with high-tensile steel bolts, which have proved to be much more satisfactory. They are not stainless, but are regularly greased and show no sign of rusting after four year's use. B rotors. These are of simple construction and are so easy to clean and maintain that they can be washed, dried and put away in 5 min. Silicone grease should be applied only sparingly to the centre core O-rings, as excess grease can block the four holes leading down the vanes. If these holes do get blocked or contaminated with fungal growth, they can be cleaned effectively with a pipe cleaner wetted with a detergent and pushed down the whole length of the vane. After each run, the threads on the bowl and lid should be dried and greased with a special graphite silicone or lubricant grease (Beckman recommend a special grease, Spinkote). Other rotors. The general remarks on care of rotors apply to most rotors even those of the more sophisticated B-IV and K series. The upper and lower shafts on the tall cylindrical rotors should be treated with respect and great care as they form an integral part of the stabiliser and vacuum bearings. Feed head, Guard tray. The operation of the feed head, covered in 8 5.1.3.1, should be referred to for a fuller description of its maintenance. The major items of the feed heads which require special attention are: 1) The two parts of the seal, Rulon and stainless steel, which must be dry and, preferably, kept apart, leaving the sealing surfaces separated during storage. 2) The locating bearing needs to be in good condition. It can become contaminated with
Ch. 1
43
INTRODUCTION TO ZONAL CENTRIFUGATION
gradient material and seize. When in bad condition, it is usually very noisy and vibrates, often causing the seal to vibrate and leak. Spare bearings should always be available in case of seizure. 3) The spring which holds the two seal elements pressed together can corrode or just weaken with age. It must be examined at intervals and replaced if faulty. 4) The vacuum cap for sealing B-type rotors is a simple device, the only moving part being a small bearing. This allows the top to be gripped while the bottom part is still spinning with the rotor. This cap is usually very reliable, but it should be examined at regular intervals.
I .8. Guarantees on rotors Although all the manufacturers give what appear to be very comprehensive guarantees, it is worthwhile to compare them and examine them in more detail. We will summarize the guarantees given on swing out and zonal rotors, but the manufacturers must be consulted for the precise conditions. Beckman rotors are guaranteed, subject to conditions, for 1000 runs, o r 2500 hr or 5 years (whichever frst) a t maximum speed. After that the guarantee is extended to a further 1000 runs or 2500 hr at 90% of the maximum speed. If defective, a rotor may be replaced at a special reduced price equal to the current list price x (no. runs/2000) or x (no. hours/5000). Christ ‘guarantee’ their rotors for a specific number of runs. These runs are defined as having a value of 1 for a run of maximum speed. For any speed less than maximum, a value lower than 1 can be determined by referring to a set of tables. This value is called the ‘Run Valency Figure’ (RVF). The titanium rotors are given the RVF of 1 irrespective of speed and are guaranteed for 5,000 runs. Aluminium rotors may be guaranteed for a small number of runs, i.e. the 8 x 20 angle rotor, used at 60,000 revs/min has the maximum number of 500 runs, while at 24,000 revs/min it has an unlimited life. It is interesting to note that Christ do not use the term guarantee or warranty, but state that rotors have ‘Maximum permissible Sehl‘rr
l I l d < ~ Yp
287
44
DENSITY GRADIENT CENTRIFUGATION
number of runs’. In practice, it appears to work like a normal guarantee. ZEC give the simplest guarantee. ‘Each IEC rotor is supplied with an unlimited warranty. It may be used at its maximum speed indefinitely, regardless of hourly or cyclic use, provided the rotor is maintained in accordance with a set of conditions’. Damaged rotors may be exchanged for a new rotor of the same type at a discount of 10% of the list price on the original rotor. MSE rotors, whether aluminium or titanium, carry a 4-year guarantee, again depending on a set of conditions. After that period, there is no derating scheme. If a replacement rotor is bought within 6 months of the expiry date, a discount of 10% is given. The sets of conditions given by all the firms are similar, and contain such points as: correct use according to manual, no corrosion, damage or excess heating, mean density of contents must not exceed 1.2. The warranty period, although different in every case does in fact give a reasonable number of runs. Presuming that a rotor is used 4 times a week, regularly, this means about 200 runs a year, or 1000 runs in 5 years. In practice it is extremely difficult to keep up this heavy, regular usage for any long periods, and the figure after 5 years is more likely to be 500 runs or less. This means that a 5-year guarantee does not really give the full usage, but the manufacturers presumably feel that, after such a period, corrosion etc. would have weakened the rotor enough to make it advisable to derate it. The IEC unlimited guarantee is obviously the most attractive. The Beckman system gives another 1000 runs or 2500 hours at 90% of the maximum speed after the first 5 years and combined withquiteagood replacement price, seems to be a reasonable warranty. Since the maximum speed on Beckman machines is selected manually by each individual worker (who can make mistakes), we are not sure how this scheme works out in practice. The choice of 1000 runs or 2,500 hr seems to be one which only Beckman can justify. The stress put on a rotor is due to a run (which they define very precisely) and yet a run at maximum speed lasting 24 hr would effectively
Ch. 1
INTRODUCTION TO ZONAL CENTRIFUGATION
45
count as 10 runs. Since we are led to believe that the stress put on a rotor is due to a ‘run’, the relationship raises difficulties for laboratories accumulating a lot of hours with only a few isopycnic runs. Sorval have adopted a Beckman type warranty for their swing out and zonal rotors available for the new OTD machines.
CHAPTER 2
Theoretical aspects of centrifugal separations
Intelligent design of experiments depends on some acquaintance with the underlying theory. Many workers have been discouraged from examining the theory of gradient centrifugation because of the complex mathematics required to give an accurate model of a sedimenting zone and the limited usefulness of such models. One must, however, distinguish between the objectives of the experimental worker and the theoretician. The former wishes for some generalisations which will help him improve his practical results, the latter wishes to construct a mathematical model which will fully represent the phenomenon which he is studying. The ensuing discussion will be purely concerned with the former objective. Readers interested in the theoretical aspects should consult the quoted references. This chapter is divided into three main sections. The first and second concentrate on the theory of rate and isopycnic zonal separations. The third section discusses the properties of the actual density gradient materials and how these will affect the sedimentation of subcellular organelles and macromolecules. The discussion at this point will be limited to changes which affect centrifugal separation. Analytical problems arising from the effect of density gradient materials will be considered in Ch. 8.
2.1. Theory of rate-zonal separations A brief discussion is given of the theory of sedimentation of particles through a density gradient. In practice, particles may not sediment exactly in the way predicted by this theory. This occurs because 46
Ch. 2
THEORETICAL ASPECTS OF SEPARATIONS
47
the concentration of the particles has exceeded the ‘capacity’ of the gradient. We are concerned, therefore with how to calculate the maximum amount which can be loaded on to a density gradient and on the circumstances under which an initially stable band can become unstable during centrifugation. We also discuss the effect of diffusion on the shape of a sedimenting band and the limits which this places on the size of particles which can be separated by rate-zonal centrifugation. 2.1 . I . Theory of’sedimenting particles
No attempt is made to give more than an elementary discussion of the theory of centrifugation. Extended accounts are given in articles by Schumaker (1967), Trautman and Cowan (1968) and Bowen (1 970). When particles are to be separated by rate-zonal centrifugation, a thin layer of the initial mixture is layered on top of a density gradient. On centrifugation each type of particles moves down the gradient at a characteristic velocity. If the velocities of two types of particles are different, then the zones become separated. The apparent centrifugal force experienced by a particle of volume V and of density d when centrifuged at an angular velocity of w radians/second (a full circle is 27t radians, so w = 27t x (RPM) + 60) is V d r w2 where r is the distance of the particle from the centre of rotation. As the particle displaces some of the density gradient medium, Archimedes principle indicates that there will be an apparent upthrust on the particle equal to the weight of liquid displaced, i.e. V p r 02 where p is the density of the medium. Thus the net outward force C is given as
If the particle is assumed to be spherical with a radius R, and it is moving with a velocity v, then by Stoke’s law the frictional force opposing motion is 6 7tRqv where q is the viscosity of the medium. If the particle is not spherical, then an ‘asymmetry’ factor n must SuhicvI uidm p. 287
48
DENSITY GRADIENT CENTRIFUGATION
be added to the right-hand side of the of the equation. Thus the frictional force F is F = 6 n Rvva.
(2.2)
A particle of the size of cell fragments sedimenting in a liquid medium may be assumed to attain its terminal velocity instantaneously. This means that, as the particle is not accelerating, the inward and the outward forces on it are the same i.e. F = C or 6 n Rqva so that the velocity of sedimentation
I)
=
V(d-p)rw*.
(2.3)
is
(2.4) If the medium of centrifugation is defined, eq. (2.4) shows that the velocity of sedimentation is proportional to the centrifugal acceleration rw?. Eq. (2.4) can therefore be rewritten v
= s rwz
(2.5)
where
As it stands, eq. (2.5) is of no use for density gradient centrifugation as s is dependent on the density and the viscosity of the medium. If, however, we can assume that the size and the shape of the particles are not altered by changes in the medium, then, if any particular medium has a density of p , and a viscosity of rt, and a second medium has a density and viscosity of p? and %, the value of s in the second medium is S? =
V 6n Ra
__
d-P 2 %
(2.7)
Ch. 2
THEORETICAL ASPECTS OF SEPARATIONS
49
It is most important to remember that eq. (2.7) rests on the assumption that the size and the shape of the sedimenting particle is independent of the composition of the medium. If we do make this assumption, then we can define a constant sTM (the sedimentation coefficient) which is the value of s at a standard temperature T in a standard medium M . Inspection of eq. (2.6) shows that the sedimentation coefficient thus defined is related to both the size, the shape and the density of the particle. The standard medium is normally taken as water at 20 “C, the sedimentation coefficient then being written as s20,w.Thus, by combining eq. (2.5) and (2.7) we can show that the velocity of sedimentation of a particle in a medium m at a temperature t
where p,,,,, p2n,wr ‘I,, and ‘ I ~ ~ are , ~ the density and viscosity of the medium and of water at 20 “C respectively. Inspection of eq. (2.8) will show that szo,whas the dimensions of (Time). As the value of szo,win sec is inconveniently small for all biological particles, it is conventionally multiplied by In this case, the sedimentation coefficient is written in equations with a capital S , for example as Szo,w. The unit of sedimentation is a Svedberg (conventionally abbreviated s).Thus eq. 2.8 is written as
where S2,,,wis the sedimentation coefficient in Svedbergs. Inspection of eq. (2.9) shows that the *velocity of sedimentation is governed by four factors: 1) the sedimentation coefficient of the particle; 2) the density of the particle; 3) the density and viscosity of the centrifugation medium; 4)the centrifugal force. Up to this point, the necessity for a density gradient has been simply assumed. However, as can be seen from the preceding discussion, the existance of a gradient is a considerable inconvenience when we try to predict what will happen in a centrifugal separation. Siihlcwr rider p 287
50
DENSITY GRADIENT CENTRIFUGATION
Initially, a gradient does not seem essential for separations based on the difference in the sedimentation rates of the particles; however, if a particle is to sediment in a medium, it must be more dense than that medium. As the denser particles will displace some of the lighter medium, the overall density of the zone containing the particles will be higher than the rest of the liquid. If a density gradient is not present, the whole zone will tumble to the bottom of the liquid column. A density gradient in the medium, providing that it is steep enough, will prevent this tumbling by raising the density of the medium on the centrifugal side of the zone above the combined density of the particles and medium inside the zone. The capacity of the gradient is, however, strictly limited, as will be discussed later. Thus, a density gradient is necessary to support the sedimenting zones of particles. This means that as the particles move further into the gradient, their speed of sedimentation falls, for as the density p of the medium increases, the term d-p in eq. (2.9) falls. While it is convenient, for some theoretical purposes, to consider particles sedimenting in a uniform centrifugal field, this is not the case in practice with particles separated in preparative centrifugal rotors. The centrifugal ‘force’is proportional to r o 2so that the further a particle is from the centre of rotation, the greater the force which it experiences. This effect is of much greater importance with zonal rotors than with swingout rotors. For example the top of the tubes of an MSE 3 x 23 ml swing-out rotor are 5.88 cm from the centre of rotation, the bottom at 12.87 cm. The centrifugal force is therefore 2.2 x greater at the bottom of the tube than at the top. Corresponding figures for the Spinco SW 40 rotor are 6.67 cm and 15.87 cm giving an increase in the centrifugal force of 2.38-fold. With the B-XV zonal rotor, on the other hand, the sample, if placed in its normal position just clear of the rotor core, is 2.3 cm from the centre of the rotor. The outside of the gradient space is 8.8 cm from the centre, therefore the centrifugal force is 3.8 x greater at the periphery than at the centre. As a result, particles moving in a nonviscous gradient will tend to move faster as they sediment into the gradient.
Ch. 2
51
THEORETICAL ASPFCTS OF SEPARATIONS
So far the gradient material has been assumed to have negligible viscosity. However, the viscosity of solutions of sucrose and of most other density gradient materials rises sharply with increasing concentration. Thus, the acceleration caused by the increasing centrifugal force as the particle sediments down the gradient will be opposed by a retardation due to the increase in viscosity. Although this retardation is usually simply a problem to be overcome by increasing the centrifugation time, on occasions it can be exploited to improve a separation as will be discussed in Ch. 3. 2.1.2. Stability of the sample zone Before we consider in detail the effect of the gradient on the sedimentating zones of particles, we must consider under what conditions a band of material layered on top of a density gradient will be stable. The first requirement for stability is quite simple. It is that the overaN density of the sample band must be less than the density to separate polysomes directly from a deoxycholate-treated 20,000 g, that the overall density is the important factor here, for it is easy to underestimate the contribution which the suspended biological particles make to the overall density. For example, we might wish to separate polysomes directly from a deoxycholate-treated 20,000 g, 20 min supernatant fraction prepared from a 20% homogenate in 0,25 M sucrose of rat liver. As about half the protein of the liver homogenate is present in the deoxycholate fraction, and liver contains about 20% protein, the fraction will contain about 20 mg protein per ml. 0.25 M sucrose has a density of 1.032. As proteins have a density of about 1.3, they will raise the density of the medium by about 0.005 g/l, so that the density of the liver suspension will beabout 1.037. Therefore the sample will only be stable if the gradient starts a sucrose concentration of about 0.3 M (density 1.040). It may be objected that the meth’od of calculation used in the last paragraph is extremely crude in that all the constituents of the homogenate other than protein have been neglected. This approximation is, however, not nearly as serious as it sounds. Nucleic acids and carbohydrates only make up a small proportion by weight of the Sirhje
rrldcx p 2x7
52
DENSITY GRADIENT CENTRIFUGATION
cell and so will only have a minor effect on the overall density. In case of doubt it is easy enough to measure the density of a sample directly. Unfortunately it has been found that, in practice, the preceding analysis is not adequate. Experimental results have confirmed that although sample bands close to the theoretical limit are initially stable (Spragg and Rankin 1967) the outer edge of the band spreads much more rapidly than can be explained by diffusion (Brakke 1964; Nason et al. 1969; Sartory 1969; Meuwissen and Heirwegh 1970; Meuwissen 1971, 1973). This phenomenon has been called ‘droplet sedimentation’ (Brakke 1955) or ‘streaming’ (Anderson 1956) and should be distinguished from ‘tumbling’ of the whole sample zone which has been described earlier. Droplet sedimentation is due to differences in the diffusion rate of the gradient material and the material in the sample. The interface between the sample and the gradient is illustrated in Fig. 2.1. Thisinterfacewill never becompletely smooth, as vibration, convective currents and Coriolis forces will create small wavelets (Fig. 2.2a). The tips of these wavelets will have a very large surface area for their volume. As the concentration of density gradient material must be lower in the sample zone than at the start of the gradient, as we discussed earlier, the density gradient material will tend to diffuse back in to the sample region. The sample material will, indeed tend to diffuse out but this will happen much more slowly, if, as is usual, the sample is of higher molecular weight than the density gradient material. The overall density inside the ‘wavelet’will therefore be higher than the overall density of the beginning of the density gradient and the wavelet will therefore tend to break away and sediment a short distance into the gradient (Fig. 2.lb, c). The same process will be happening all around the sample band and will cause a slow streaming forward of the whole leading edge. This effect of droplet sedimentation is one of the most important causes of sub-optimal resolution in zonal centrifugation. Meuwissen (1971, 1973; Meuwissen and Heirwegh 1970) have shown that, with samples layered on sucrose gradients, droplet sedimentation
Ch. 2
.... .'.'..:...
53
THEORETICAL ASPECTS OF SEPARATIONS
0'. * 0 ' . ' 0 -0.e..
.i *
*
0.0
...i..::..;.:.
..:. . ..... -: .. ........................ ..................... .. ....... .. *... . (a) . .. ..... ... .. ... .. ... .. ..... .. .. ......... ...
.......................... .. .:..:.;. . . . .. . ..... . .. ... .. . . . . . . . . . . . . . . . .. . . .-. -. ..
(b)
.. .. . . . ....... .. ...... .. ... .. .. ..... ..':: .......... :.* . . . . . .......... . . :. ....: . . . . . . . . . . :....... .*.:
Fig. 2.1. Diagrammatic representation of the initiation of streaming from a shelflayered sample zone. For details see text.
occurs, even at the lowest sample concentrations, if the sample is layered directly on to the gradient (Fig. 2.2a); but if the sample is applied in an inverse gradient (Fig. 2.2b) as described in 5 4.2 then up to 20% of the theoretical capacity can be obtained before streaming occurs. Very much better results are obtained if a high molecularweight density gradient solute such as Ficoll is used. In this case no stpeaming occurs as the diffusion rates of the sample and the density gradient material are the same, as the following discussion will show. Even sample zones layered in inverse gradients show anomalous broadening at high concentrations, but there is no visible evidence that streaming, as discussed in the last paragraph, is taking place and it is suggested that the broadening is due to another form of Sub/ccr llld<\
/?
287
54
DENSITY GRADIENT CENTRIFUGATION
20
1.0. shbilizwl albumin
zone
0.8.
&280, 01
initial distribution
-
18
distribution after 30 min l g
-
CP mglml
I
0.6
2
I I
a c d a r a t e d to 40,000 g
'
0.4
'
8
0.2
'
I
0
.
2
4
ml
6
0
Fig. 2.2. Comparative stability of (left) shelf-layered and (right) inverted gradient layered zones in sucrose gradients. The concentration distribution of albumin zones after 30 min at 1 g and after are shown at the time of introduction (--------), 60 min with an interposed acceleration to 40,000 revs/min and deceleration. The total amount of albumin applied to the inverted gradient was 85% of that layered as a shelf. Both samples were well below the critical load for the density gradient used (from Meuwissen 1971).
hydrodynamic instability called by Meuwissen (1973) broadening due to bulk flow; this is essentially its description. It is due to variations in the concentration of the gradient solvent (normally water) across the sample zone. To understand this phenomenon one must remember that molecules of gradient solute or of material in the sample zone will displace water molecules. Hence the concentration of water falls along a rising density gradient. However a much more complicated situation is found at the dense edge of a sample zone or a zone of sedimenting particles. The macromolecules and small particles which are separated by density gradient centrifugation are normally intrinsically light (e.g. proteins have a partial
Ch. 2
55
THEORETICAL ASPECTS OF SEPARATIONS
specific volume of 0.73 as compared with 0.61 for sucrose and 0.25 for CsCl) and hence the concentration of water in the sample suspension will be lower than in a solution of the gradient solute with the same overall density. Further, a considerable proportion of the water in the sample zone may be immobilised in the hydration spheres of macromolecules. Hence when the concentration of sample is high, the concentration of water may actually rise across the leading edge of the sample zone (Fig. 2.3).
Av arbitrary units
Fig. 2.3 The distribution of gradient solute (sucrose) (-.-.-.-,), sample (albumin) ) across a heavily loaded zone. It is assumed that the sample has been layered in an inverse gradient.
( _ _ _ _ _ __ _ _ -) and water (
Once one considers this possibility the mechanism for broadening becomes clear. In a stable zone (Fig. 2.4a) the gradient material tends to diffuse back into the sample zone, but this is counterbalanced by an outward diffusion of water which results in the volume of the sample zone remaining constant. However, if the fall in water concentration across the sample zone is very low or, as in the example discussed, there is actually a rise, then there will be nothing except the very slow diffusion of the sample material to balaqce the increase in volume of the sample zone due to the influx of gradient solute and, in the extreme case, of water. Hence the volume of the sample zone will rise; also, the lower the diffusion rate of the gradient solute, the less will be the broadening of the sample zone due to bulk flow. Suh,r
UIL/PX
p 2x7
56
DENSITY GRADIENT CENTRIFUGATION
Composition and voluma fluxes of doscanding ronr boundary
-.-
--
-
Campanant 0 (Watar) Component 1 (Albumin) Componant 2 (Sucrora)
A Laad 10% of maximum (nabla)
B Lord 100% of maximum (unstablr)
Fig. 2.4. Diagrammatic representation of hydrodynamic instability. (A) illustrates a stable boundary. Diffusion of the gradient solute back into the sample zone is balanced by diffusion of solvent (water) from the sample zone, there is therefore no change in volume. (B) An unstable situation. The concentration of sample is so high that theconcentration of solvent (water) increases on moving away from the centre of the sample zone. Therefore both the density gradient solute ( ) and the solvent (__________) d’iffuse into the sample zone, so increasing its volume. v and J are the partial specific volumes and diffusion coefficients of the three components (from Meuwissen 1973).
It follows therefore,that much more sample can be layered on to a Ficoll gradient than can be layered on to an equivalent sucrose gradient. Mathematical analysis of broadening by ‘bulk flow’
Ch. 2
THEORETICAL ASPECTS OF SEPARATIONS
51
(Meuwissen 1973)indicates that the sample zone will be unstable when dC, __
dr
5, D, dC,
a-.-.-
v, D, dr
(2.10)
where dC,/drand dC,/dr are the concentration gradients of the sample and gradient solutes respectively and V, and D,,V, and D ,are the partial specific volumes and diffusion coefficients for the sample and gradient solutes. A similar relationship is suggested by the experiments of Halsall and Schumaker (1971). It should be noted that bulk flow, unlike droplet sedimentation, can occur during sedimentation as well as during sample loading. The theoretical limitations on the amount of material which can be loaded on to a density gradient can, then, be summarised as follows. The density of the sample must be less than the density of the top of the gradient. If this is not so then the sample will tumble through the gradient until it reaches its own density and will broaden considerably in the process. Ifthe amount of material is close to the capacity of the gradient and if a low molecular weight gradient solute is used the band will broaden by droplet sedimentation. This can be diminished by loading the sample in a gradient which is, however, a somewhat inconvenient procedure. If one is not aiming at a very high loading or very high resolution it is usually sufficient to layer the sample directly on to the gradient and not to worry about the small loss in resolution which this will entail. If the resolution is not sufficient for the purposes required one should start worrying about streaming.
2.1.3. Stability of a sedimenting zone Normally, if the initial zone is stable, the zones of particles which sediment away from that zone will remain stable. This is especially so if a simple linear gradient is used. However, it is possible for an initially stable band to become unstable during centrifugation and one should be aware of the conditions under which this can occur. Consider a zone of a finite width sedimenting through a gradient, Sirhlrrr Imlcu 11 2x7
58
DENSITY GRADIENT CENTRIFUGATION
the outside edge of the zone is subject to a greater centrifugal force than the inside edge thus tending to expand the zone, but the particles at the leading edge are also encountering a liquid of greater viscosity and density, both of which will tend to slow down the faster sedimenting particles. The balance of the centrifugal force favouring expansion of the zone and of the liquid properties which tend to narrow it will depend on the shape of the gradient and the properties of the gradient material. Equations describing changes in the width of a band during sedimentation have been developed by Berman (1966). If the band remains ‘thin’ during centrifugation (i.e. the width of the band is small compared with the distance of the band from the centre of rotation) and the shape of the particles is not altered by changes in the concentration of the density gradient solute, the initial width A r and the final width Ar‘ of the band at distances r and r‘ from the centre of rotation are given by -Ar’ - - d (d‘-p’)qT’
Ar
d(d-p)q%
(2.1 1)
where d and d‘ are the density of the particles forming the zone in the initial and the final media and p and p’ and q and q’ are respectively the densities and the viscosities of these media. It is important to note that allowance is made for the changes in the density of the particles with changes in the concentration of the density gradient solute (see 0 2.3). As was discussed earlier, the capacity of a gradient is dependent on the slope of that gradient. Thus the capacity of a gradient will vary with distance from the starting band for all but linear gradients. In addition, changes in the width of the sample band will cause changes in the concentration of material in the sedimenting zone and therefore changes in the total capacity i.e. the total amount of material which can be held in a given zone rather than the highest concentration. The ratio of the maximum capacities M and M’ of zones with centres 7 and 7’from the centre of rotation, calculated by Berman ( 1 966) is
Ch. 2
THEORETICAL ASPECTS OF SEPARATIONS
59
(2.12) where (dpldr) and (dpldr)’ are the slopes of the gradient at r and r‘ and the other symbols have the same meaning as in eq. (2.1 1). Analysis of eq. (2.11) and (2.12) shows that bands contract on steep gradients but that this is more than compensated for by the increased capacity. The ‘danger point’ is where a band which has been narrowed by passage through a steep region of the gradient enters a shallow region. The capacity of the gradient at this point may fall to a small percentage of its original value. In general this is not a serious problem because during centrifugation the material will have separated into several bands, each of which will contain only a small proportion of the total weight of sedimenting particles. Also diffusion will have caused some broadening of the bands, which will again reduce the concentration at any given point. A problem could arise if a sample containing a high proportion of a single type of particle, for example monomeric ribosomes, were separated on a gradient containing a sharp step.
2.2. Theory of isopycnic banding In isopycnic banding, the approach of particles to their equilibrium positions is very slow even with non-viscous gradient materials so that prolonged centrifugation is required to band even quite large particles. As the width of the sample zone has no influence on the width of zones separated by isopycnic banding, none of the limitations discussed in Q 2.1.2 have any importance. The major point of concern in isopycnic banding is that the gradient solute will tend to redistribute itself under the influence of prolonged centrifugation so that the gradient recovered after fractionation will not necessarily be the same as the gradient initially loaded into the rotor.
2.2.1. Shape of zones separated by isopycnic banding If it is assumed that the concentration of particles is low compared with the concentration of the gradient solute, then it is very simple Sehicc‘l‘rindrx p. 287
60
DENSITY GRADIENT CENTRIFUGATION
to predict the shape of the zone. A particle close to the centre of the zone will be suspended in a liquid of its own density and hence subject to no centrifugal force. However, thermal diffusion will tend to carry the particle away from the centre of the zone. Once it has moved any significant distance, it will move into a liquid of different density and, under the influence of the centrifugal field, will be returned to the central part of the zone. Mathematical analysis (Meselson et al. 1957) shows that the balance between spreading by diffusion and contraction under the influence of the centrifugal field will result in a band of Gaussian shape with a standard deviation 0 (measurable as half the width of the band at 6110th~of the peak height) given by (2.13) where R is the gas constant, T the temperature in "K, dpldr the slope of the density gradient across the band, pi the density at the centre of the band, w2rs the centrifugal force at the centre of the band and M the molecular weight of the particles which form the band. From eq. (2.13) it follows that the width of the band is inversely proportional to the centrifugal force. That is the higher the speed of centrifugation, the sharper the bands, the steeper the gradient and the greater the risk of damage due to high hydrostatic pressure (see Q 8.1.2). The width of the bands is also inversely proportional to the molecular weight and can be used as a measure of the latter provided that the particles which form the band are homogeneous in density. It is, however, unusual to be able to make this assumption with any degree of confidence. 2.2.2. Redistribution of density gradient solutes in a centrifugal field As molecules of a density gradient solute are denser than the solvent (water) they will tend to sediment in a centrifugal field. Generally, however, the gradient material does not pellet as diffusion will tend to redistribute the solute through the solution to equalise the con-
Ch. 2
THEORETICAL ASPECTS OF SEPARATIONS
61
centrations through the column. It can readily be shown that this will lead to the production of a density gradient. Consider a small section ofa density gradient column. Assuming that the concentration at a point r cm from the centre of rotation is C then from eq. (2.4) the mass dm, of particles of gradient solute crossing the boundary X X’ in time dt is dm,
=
CAsw?rrlt
(2.14)
where A is the cross sectional area of the tube, s the sedimentation coefficient of the gradient solute and o the angular velocity of the rotor. By Fick’s law, the mass dm, of particles diffusing across the boundary is dm,
=
dc -DA-dt dr
(2.15)
where dc/dr is the concentration gradient. The liquid column will be stable (i.e. there will be no net movement across the boundary when dm,
=
dc dm, i.e. C A sw2r = DA dr
(2.16)
i.e. when the concentration gradient is given by dc s -- _. Cr o2 dr D
(2.17)
From eq. (2.6) we can write
where p is the density of the solvent, q the viscosity of the solution, V the volume of the solute molecule, x its diameter a an asymmetry factor and d the density of the solute molecule. It should be noted that d represents the density in solution which will probably not be the same as the density of the solid solute. Sihjoci LiOm p. 287
62
DENSITY GRADIENT CENTRIFUGATION
As the mass of a molecule of the solute is V . d , the molecular weight M of the solute is
M
=
V*d.N
where N is Avogadro's number. Hence s =-. M 6nxaN
(I-vp) V
(2.18)
where V is the partial specific volume of the solute molecule defined as lld. Subject to the same assumptions as were made in 9 2.1.1 when discussing the derivation of the sedimentation coefficient it can be shown that (2.19) where R is the universal gas constant and T the temperature in so that
_s _ M(l-ifp) D
RT
O K .
(2.20)
The derivation and limitations of eq. (2.17) and (2.20), which are of great importance when using the analytical ultracentrifuge, are discussed in detail by Bowen (1970). While it is possible to demonstrate redistribution of low molecular weight organic solutes such as sucrose in a centrifugal field, the effect is not of great practical importance as the approach to equilibrium is so slow. However when density gradient solutes such as the salts of the alkali metals, which give mobile solutions and in which the solute has a low partial specific volume, are employed then a major redistribution of material will occur during high-speed centrifugation. Under such conditions any preformed gradients will become distorted after a few hours' centrifugation even if the equilibrium gradient is not reached.
Ch. 2
THEORETICAL ASPECTS OF SEPARATIONS
63
Study of eqs. (2.17) and (2.20) shows that the slope of the equilibrium gradient depends on three factors : the molecular weight of the density gradient solute, the partial specific volume of the density gradient solute and the centrifugal field. In practice the influence of the first two factors means that gradients of very different slope are formed at the same speed of centrifugation by different gradient materials. However, as discussed in Ch. 3 the choice of gradient solute is usually governed by other considerations. The equilibrium gradient in long tubes will not be perfectly linear as the centrifugal field, and hence the slope of the gradient, will increase along the length of the tube. Thus when particles very similar in density, such as DNA molecules of different base composition, are being separated, increasing the speed ofcentrifugation is likely to decrease the resolution for, with increased speed the slope of the gradient will increase and hence the distance between bands will decrease. While the bands formed during centrifugation will be correspondingly narrower, the small distance between the bands will increase the risk of crosscontamination during displacement. While too high a speed of centrifugation may result in poor resolution, too low a speed will require very prolonged centrifugation to establish the gradient and to band the particles with proportionately increased risks of degradation. In the past it has been necessary to establish some compromise speed, but the relaxation technique introduced by Anet and Strayer (1969a) promises to allow both rapid separation and a wide spacing of bands. In the relaxation method, the gradient is centrifuged at high speed to establish a gradient and to band particles at their isopycnic densities. All the bands at this time will be concentrated in the centre of the tube. Once banding is complete, the speed is lowered to that calculated to give a gradient which will spread the bands out through the tube. This speed is maintained for 6-8 hr. The gradient flattens by diffusion and the bands of particles are carried apart by the bulk flow without any more broadening than would be expected from the lower speed of centrifugation.
64
DENSITY GRADIENT CENTRIFUGATION
2.2.3. Influence of the rotor on gradient shape As will be clear from the above discussion, the slope of the gradient formed during centrifugation is determined by the properties of the gradient solute and by the speed of centrifugation. When a swingout rotor is used, the relative positions of the liquid zones do not change during deceleration. If an angle head rotor is used, the situation is very different, The gradient formed during centrifugation is determined by the position of the liquid when the rotor is at speed and is effectively equivalent to the gradient which would be formed in a very short, but very large diameter tube in a swing-out rotor. When the rotor is decelerated, the liquid re-orients in such a way that the physical distance between the different zones is much increased. The risk of cross contamination during displacement is therefore minimised. The main risk with this procedure is that the separated bands would become spread during deceleration. This does not seem to happen (Flamm et al. 1966, 1972) and isopycnic banding in the tubes of angle head rotors has been used very successfully in banding DNA (Fig. 2.5). However, it is important to remember the small radial distance between the meniscus and the bottom of the tube, hence the total density range covered by the gradient is quite small. Thus, angle-head rotors are most useful for separating material such as DNA, where the fractions differ only slightly in density. In such cases the small radial distances are an advantage, for the time taken for particles and gradient solute to reach their equilibrium positions will be proportional to the distance which they must move and hence the shorter the tube, the more rapid the equilibration.
2.3. Effects of density gradient solutes on subcellular structures Reversible changes occur which affect the sedimentation rate or the isopycnic banding density of macromolecules or subcellular structures. Density gradient solutes can also cause irreversible damage and this problem is discussed in chapter 8. As discussed in 9 2.1.1 the velocity of sedimentation of a particle in a centrifugal field is
Ch. 2
65
THEORETICAL ASPECTS OF SEPARATIONS
100
'"I I
t
f
.Mouse DNA
E E
. v)
c V
1.62
1
Bottom
2
3
Volurne,ml
1
100
Dense
Volume, ml
Light
Meniscus
Fig. 2.5. (left) Gradients obtained after centrifuging a CsCl solution of initial density 1.72 g/cm' in a fixed angle Spinco Type 40 rotor or an SW 39L rotor. Centrifugation was 60 hr at 33,000 revs/min. B) (right) Absorbance (0-0-0) and radioactivity (C--- 0--- 0)profiles from a mixture of (I4C) DNA from E. coli and unlabelled D Y A from mouse liver centrifuged to equilibrium in fixed angle (A) and swing-out (B) CsCl density gradients (from Flamm et al. 1972).
dependent (among other factors) on the density and on the size and shape of the sedimenting particle. Little is known about possible changes in shape during sedimentation, but it is easier to assess changes in density. Obviously the effect of density gradient solutes on subcellular structures will depend on the structure of the particle. For the purposes of this discussion two models will be discussed, which respectively approximate to macromolecules or nucleoprotein particles and to the larger membrane-bound structures of the cell. A macromolecule or a small particle may be considered as a fine mesh through which water and possibly other low molecular weight compounds can percolate freely. Associated with this 'core' will Strh,cr/ rrrdcr p. 287
66
DENSITY GRADIENT CENTRIFUGATION
be a certain amount of 'bound water' attached to the organic core by hydrogen bonds. As the tertiary structure of macromolecules is typically fairly loose, a large amount of water may be bound. Naturally occuring macromolecules are, with the exception of lipids, all denser than water. In dilute solution the hydration sphere must be considered as part of the macromolecule or particle and hence the particles will behave in dilute solutions as though their density were much lower than the anhydrous density of their organic core. Theoretically there are two ways in which a density gradient solute could affect such a structure. Firstly the hydrophilic solute molecules could abstract water from the hydration sphere. Secondly the gradient solute could penetrate into the hydration sphere displacing bound water. In practice both effects would result in an increase in the apparent density of the structure. Ultimately if the concentration of gradient solute in the hydration sphere is the same as in the bulk of the medium, the particles will band at their anhydrous densities. As far as the authors know, there is no simple method of determining which of the two mechanisms is of more importance in practice. What is certain is that the density gradient medium does have a major effect on the banding density of macromolecules and of nucleoprotein particles (4 3.3). To the authors it would seem likely that penetration of gradient solute into the structure of the macromolecules is of greater importance than dehydration because of the different effects of gradient solutes which have similar structure and would be expected to be almost equally hydrophilic. Thus, DNA bands at a lower density in caesium sulphate or oxalate than in caesium chloride (Zolotor and Engler 1967). This would be explained if the larger anions failed to penetrate the structure of the DNA. Electrostatic forces would then limit the penetration of Cs' ions. A more clear-cut instance is found with variations in the apparent densities of nucleoproteins in non-ionic gradients ; ribosomes (which have an anhydrous density of 1.56 (Petermann 1964)) behave in sucrose solutions as though they had a density of 1.43 (Petermann 1964) and band in Metrizamide (MW 742) at a density of 1.26 (Hinton et al. 1974b). This would be explained if sucrose partly penetrates the
Ch. 2
67
THEORETICAL ASPECTS O F SEPARATIONS
structure of the particle, but Metrizamide scarcely penetrates a t all. Similarly, chromatin b&ds at a densityof 1.3 in CsCl (Hancock 1970), 1.55 in chloral hydrate (Hossainy et al. 1973), 1.37 in sucrose/glucose gradients (Raynaud and Ohlenbusch 1972) and 1.20 in Metrizamide (Rickwood et al. 1973). Therefore, in considering the centrifugal behaviour of macromolecules and small particles, one should remember that the density and mass of the particles may well change during the course of centrifugation due to the penetration of density gradient solute. Thus, the isopycnic banding density will vary widely depending on the extent to which the gradient solute can penetrate the particle. Paradoxically, the effects of density gradient solutes on membranebound particles, although more complex, are better understood. The model normally used is that developed by De Duve and his colleagues (de Duve et al. 1969; Beaufay and Berthet 1963; Beaufiy et al. 1964) in which the particle is divided, for the purposes of calculation into three 'compartments'. The first compartment may be thought of as the solid part of the particle. This includes water firmly bound to the macromolecules which make up the particle together with low molecular weight solutes which cannot escape through the membrane. The volume of the 'solid' space is taken as 4dand its density as p,,. The second compartment represents the areas of the particle which are accessible to water, but not to the gradient solute. These compartments will fill with water until the osmotic pressure exerted by the low molecular weight solutes mentioned above is equal to the osmotic pressure of the solution in which the particle is placed. If there are CI moles/g of osmotically active solids then the volume of liquid in the space will be
v,
(2.21) where p, is the density of water and m is the molality* of the density gradient solute while the mass M , is
*
The molality of a solution is the concentration in g/kg solvent. SIIII,'~~ I !,,,/<,,-
,'
287
68
DENSITY GRADIENT CENTRIFUGATION
M,
=
VOPW
(2.22)
The third compartment is accessible to both the solvent and the density gradient solute and hence will have a density equal to the density p, of the suspending medium. If the volume of this space is fl @d then the mass of material will be p Qd p,. Thus the total mass of the particle pp will be (2.23) and the volume V, will be (2.24) SQ
that the density of the particle will be (2.25)
Values for c1 and P for rat liver mitochondria and lysosomes are given by Beaufay et al. (1964). Eq. (2.25) predicts a curvilinear relationship of pd on pm if CI f 0 and is capable of predicting the density of mitochondria and lysosomes in sucrose solution with great accuracy (Fig. 2.6). It should be noted that the density of the particles changes very considerably with the density of .the medium and this must be taken into account when calculating sedimentation coefficients or attempting to predict the movement of membrane bound particles through density gradients. The isopycnic banding density p, for a particle which behaves as described by eq. (2.25) will be the density at which the particle density pp is equal to the density of the medium, i.e. pp = pm = p,, or, by substituting into eq. (2.25) and rearranging (2.26) Thus the regions of the particle which can be penetrated freely by
Ch. 2
THEORETICAL ASPECTS OF SEPARATIONS
69
I
Fig. 2.6. Variation in density of rat liver subcellukdr structures with the density of the pucrose solution in which they are suspended. a) Microbodies as indicated by ( 0 ) catalase, (0)D-aminoacid oxidase, (A) urate oxidase. b) Lysosomes as indicated by ( 0 )acid deoxyribonuclease, (A) acid phosphatase and mitochondria as indicated by ( 0 )cytochrome oxidase (from Beaufay 1966).
the density gradient solute have no influence on the isopycnic banding density and, indeed, when the particle is totally permeant as are rat liver microbodies (Fig. 2.6) so that CI = 0 then the isopycnic banding density will be equal to p,, the density of the solid matrix.
CHAPTER 3
Conditions for a centrifugal separation
The simplest way of choosing conditions for a centrifugal separation is to look up someone else’s work. However, it may be necessary to attempt to fractionate a mixture which has not been examined before. We will discuss approaches to this problem. The design of conditions for the separation of particles by density gradient centrifugation is more an art than a science, and there may be many equally valid approaches. The reader may, therefore, find differences between the approaches recommended by workers from other laboratories (Cline and Rye1 1971; Birnie 1973; Price 1972) and the approach discussed in this chapter.
3.1, Choice of approach As discussed in previous sections, centrifugal separations may be accomplished in three ways. Firstly, one can exploit differences in thesizesofthe particles i.e. use rate-zonal or differential centrifugation. Secondly, particles may be separated by isopycnic banding in which case the separation depends solely on their buoyant densities. Thirdly, one may, by using complex gradients, exploit simultaneously differences in density and in sedimentation coefficient. If during a rate separation the densities of one or more of the components of the mixture lie within, or only just outside, the density span of the gradient then the separation will depend on both the size and the density of the particles (9 3.4.3). In addition, with many lipid-containing particles, one may choose between layering the particles over the gradient and separating by differences in the sedimentation rate, I0
Ch. 3
CONDITIONS FOR SEPARATIONS
71
or by making the overall density of the sample greater than the density of the particles to be separated, layering the sample under a density gradient and separating by differences in the flotation rate. As the separability of particles by density gradient centrifugation is determined by differences in their size and density, the first step is to tabulate the size (or sedimentation coefficient) and density of all the particles which are thought to be present in the mixture (see Appendix 3 for data from rat liver). These values do not have to be very accurate and one may often extrapolate from the organelles of one type of cell to those of another if electron microscopy shows the particles to be similar in size and construction. One must also decide on the markers* for each of the particles in the mixture. If there are no data in the literature on the desired properties of some of the particles, then these should be determined in preliminary experiments (see Ej 3.4). The next step in planning a separation is to draw up an 'S-p' diagram. This is a simple graphical method of presenting the data on particle size, usually plotted as the Sedimentation coefficient (S) and the density ( p ) which are tabulated in Appendix 3 (Fig. 3.1). Where particles of greatly different size and thus greatly differing sedimentation coefficient are present, it is convenient to use a logarithmic scale for the sedimentation coefficient. From the relationship of particles in the S - p diagram one can guess at the optimum conditions for a separation. In general, rate sedimentation is to be preferred for analytical separations as being quicker than isopycnic separations (see Ej 2.1.1). During this shorter centrifugation there will be less time for particles to autolyse. Also particles will be exposed to much lower concentrations of the density gradient medium, high concentrations of which may be damaging to the particles (see 9; 8.1.3). Where there is little risk of damage to the particles by autolysis or by reaction with the
* A marker is an enzyme or chemical component which is ideally found in only one organelle of the cells of a given tissue. The choice of markers is discussed by Reid (1972) and Hinton and Reid (1975). Suh/r
72
DENSITY GRAD1 ENT CENTRIFUGATION
Glycogen
1
c
r
.. 0
Nuclei
-Polysomes
1.3
-E>
t Prote,ns
c
$
1.2
1.1
1 .o
-
0
Microbodies
4 fl:ochonciria
Endoplasmic reticulum fragments
?-&
*----
5-
I
Whole cells
1
0
Red blood cells
- -. : ,_-_------------.________2 -______-----(.-_/
t
-
Plasma membrane fragments
I
I
I
I
I
I
I
1
Fig. 3.1. An ‘S-p’ didgram showing the approximate sedimentation coefficients and isopycnic banding densities of particles in a rat liver homogenate when separated in sucrose gradients. The densities of ribonucleoprotein particles are estimated from the effect of changes in the medium on their sedimentation coefficients (Petermann 1964) and the density of glycogen from its banding density in Metrizamide gradients (Mullock and Hinton 1973). Note that the banding density of both the latter particles in CsCl gradients is considerably higher. (Adapted from Anderson 1966.)
gradient solute, isopycnic banding may be preferred for preparative separations, as much more material can be separated in a given volume of gradient (see 9: 2.2). If particles differ in sedimentation rate by factors of ten or more there is little point in using density gradient centrifugation for, as discussed in 6 1.2.2, differential pelleting will give adequate results. Where particles differ only in sedimentation rate (e.g. polysomes) or only in density (e.g. lysosomes and microbodies), clearly rate sedimentation and isopycnic banding respectively must be employed. However, when a mixture contains particles similar both in size and density, then complex gradients must be employed. Thus the three component mixture whose S-p diagram is shown in Fig. 3.2 will not be separated by simple rate-zonal separation, but can be fractionated on a steep gradient which approaches the isopycnic banding density
Ch. 3
13
CONDITIONS FOR SEPARATIONS
Fig. 3.2. A diagram showing a simple hypothetical three-component mixture which cannot be fully resolved by rate sedimentation o r by isopycnic banding.
of the particles. Rather more complicated, but 'real life' examples of such difficult separations are described by Boone et al. (1968) and Prosper0 and Hinton (1973). One usually tends to avoid separation by flotation because of the damage that high concentrations of the gradient material may cause to biological particles (see Q 8.1.3). However, rate flotation is very useful where the density of a group of particles is inversely proportional to their size (i.e. the smaller the particle, the denser it is). An example is serum lipoproteins, an S-p diagram for which is shown in Fig. 3.3a. If such a mixture were separated by rate sedimentation, many hours would elapse before an adequate separation was obtained (Fig. 3.3b) for the highest density component (HDL) must pass through the band of LDL with all the attendant risks of the smaller particles getting tangled in the band of larger particles. If, on the other hand, the particles are separated by flotation, the bands separate permanently very quickly after the start of centrifugation (Fig. 3.3~). If particles are identical in both sedimentation coefficient and density then obviously they cannot be separated directly by centrifugation. However, if the size or density of some of the components can be modified, separation can be achieved. For example, in a postSUh,'< I ardc, ?,
2x7
14
DENSITY GRAD1 ENT CENTRIFUGATION
1
I
0.9 0
100
1
I
200
300
Flotation coefficient lin Na Br. density 1.20)
.il,/ c
,
,
b
6
r
a
t
E
F
r I
0
10 Vblume lml)
20
Fig. 3.3. a) Diagram showing the approximate flotation coefficient (Appendix 2) and density of the major classes of lipoprotein in human serum. The flotation coefficient used here is a measure of how fast the lipoprotein will float in sodium bromide of density 1.2. b) Computed movement of serum lipoproteins when layered over an NaBr gradient ranging in density from 1.063 to 1.4 in the tubes of an MSE 3 x 23 ml swing-out rotor. Centrifugation is at 30,000 revs/min. c) Computed flotation of serum lipoproteins from a suspension adjusted to a density of 1.45 and layered under a L D L ; -.-.-. VLDL. gradient similar to that described above. ---, HDL; -,
lysosomal supernatant, fragments of the smooth endoplasmic reticulum and of the plasma membranes overlap in size and density. If, however, MgCl, is added to the density gradient medium free ribosomes bind to the smooth endoplasmic reticulum fragments and increase their density so permitting some separation from plasma membrane fragments (El-Aaser et al. 1966; Dobrota, unpublished). Even better separation is achieved by adding small amounts of lead nitrate (Hinton
Ch. 3
15
CONDITIONS FOR SEPARATIONS
et al. 1971) which increases the density of the endoplasmic reticulum fragments while leaving plasma membrane fragments relatively unaffected. An alternative approach to the problem of selectively changing the properties of some of the components of a mixture is to change the density gradient medium. As discussed in 5 2.3, density gradient materials differ in the ease with which they penetrate biological structures. The overall density of a particle is a function of the amount of ‘solid’ in the particle and of the volume and density of the fluid associated with the particles. Thus the density of the particles will depend on the ease with which the internal spaces are penetrated by the solute. As the proportion of these internal spaces will vary from particle to particle, so the relationship between the isopycnic banding densities will change with changes from a penetrating to a nonpenetrating density gradient solute. A final method of altering the density relationships of a group of particles is to selectively damage one type of particle. Thus, plasma membranes from unperfused liver and mitochondria both band at a density of about 1.18 in sucrose gradients. If, however, mitochondria are suspended in sucrose of a density greater than 1.18, their outer membranes are damaged and they band at a density of 1.21 (Leighton et al. 1968). As plasma membranes are not affected, they can be separated from mitochondria by isopycnic flotation, but not by isopycnic sedimentation.
3.2. Choice of rotor Any discussion of the choice of rotor for a given separation is likely to be idealistic, for the normal problem is how to carry out the separation given the rather unsatisfactory range of rotors available in the laboratory. Nevertheless, for those planning the equipment of a centrifuge laboratory, we will discuss briefly the choice of rotor. As discussed in 4 1.3.3 the tubes of different swing-out rotors vary considerably in their shape from long and thin to rather dumpy. One must consider how these variations in shape will affect the Suh,c
,IlC/‘,\
p 2x7
16
DENSITY GRADIENT CENTRIFUGATION
resolving power. Resolution in rate zonal separation is usually limited by the width of the sample band which will be larger than the volume of the sample due to mixing (see 4 2.1.2). In zonal rotors, there will be less disturbance to the starting band during acceleration, for the angular acceleration of a zonal from its loading speed is less than the acceleration of an initially stationary rotor. Nevertheless ‘jetting’ during loading of the sample (§ 2.1.2) and other effects, may cause the sample band to be broader than expected (Price and Kovacs 1969). The increase in width of the sample band is independent of its total width. Thus if expressed in terms of sample volume, the percentage broadening will be less with a long narrow tube than with a short fat one. The potential separation on rate centrifugation will therefore be better with the narrower tube. Also, a much greater range of particles can be separated in a long, thin, tube, than in a short, fat one, as the absolute width of the sample band will be the same in both cases. A similar analysis would apply to zonal rotors if there were the same choice as with swing-out rotors. With high-speed separations, one’s choice is limited to either the B-XV or B-XIV type of rotor, the B-XV has a rather greater diameter than the B-XIV but is not able to reach such high speeds. When separating large particles such as lysosomes or mitochondria, use of HS or A-XI1 rotors may be a great advantage (see 4 7.2.1 and 2). A general guide to the suitability of the different types of zonal rotor is given in Fig. 3.4. It should be emphasised that the comments above apply only to analytical rate-zonal separations. When the aim is to purify one component of a mixture, a different strategy is adopted. In such preparative separations, the aim is to obtain the greatest possible yields of the particle concerned in the minimum time. As the concentration of material in the sample band must be strictly limited (§ 2.1.2), one can increase yields only by increasing the volume of the sample. If a large volume of sample is applied to the long narrow tubes recommended for analytical separations, much longer centrifugation or a much greater accumulated centrifugal force will be needed than if the same volume of sample were applied to a
Ch. 3
I7
CONDITIONS FOR SEPARATIONS
Use/i/ working range of batch zonal rotors (in context of rate sedimentation)
HS B-XV (AI) A-XI1 B-XIV (Ti) RNA
RIBOSOMES POLYSOMES
LYSOSOMES
MITOCHONDRIA
NUCLEI
PLASMA MEMBRANE
MICROSOMES
WHOLE CELLS _ _ _ _ _ _ ~ ~
100
10'
102
103
I 04
105
106
107
106
SEDIMENTATION COEFFICIENT Fig. 3.4. Suitability of different types of batch zonal rotor for the separation of liver subcellular components by rate sedimentation (from Dobrota and Reid 1971).
tube of greater diameter. With long tubes, particles may be damaged by the hydrostatic pressure (see 5 8.1.2). As the volume of gradient which can be centrifuged at a particular speed is limited by the strength of the rotor material, a wider tube must necessarily be shorter. In preparative experiments where the simultaneous separation of particles of widely different sedimentation is not required, the reduction in length is no disadvantage. So far, nothing has been said about the choice between zonal and conventional rotors. We believe that the two types of rotor do not differ significantly in their ability to separate subcellular particles, as shown in Fig. 3.5. The major difference is that zonal rotors can handle very large volumes of gradient and that only one sample can be separated at a time. This means that zonal rotors are the ideal instruments for large-scale preparative separations. They are Suh/rcr d e ' i p 287
78
DENSITY GRADIENT CENTRIFUGATION
Centrifugal field
\ I
n
A2MInm
0.41-
Gradient volume
-
Gradient volume
-
Fig. 3.5. Comparison of the separation achieved by rate zonal separation in the tubes of swing-out and zonal rotors. a) Pattern obtained after centrifugation of a crude mitochondria1 + lysosomal fraction from rat liver for 40 min at 9,000 revs/min in the tubes of an MSE 3 x 40 ml swing-out rotor. b) The pattern obtained after centrifugation of a similar fraction for 45 min at 9,000 revs/min in an HS zonal rotor. 0--- 0--- 0 , protein; 0-0-0, acid phosphodiesterase. c) Pattern obtained after centrifugation of ribonucleoprotein particles, prepared by precipitation with 0.05 M MgCIz (Leitin and Lerman 1969) for 5 hr at 27,000 revs/min in the tubes of a Spinco SW 27 rotor. d) Pattern obtained after centrifugation of a similar fraction for 180 min at 47,000 revs/min in a B-XIV zonal rotor. In both cases the gradient used extended from 0.5-1.0M sucrose and, counting from the left hand side, the first four peaks correspond to i) low molecular weight material, ii) free small subunits, iii) free large subunits, iv) monomeric ribosomes.
also invaluable for analytical separations, especially when a large number of assays is to be performed on the separated fractions. The large amounts of material which can be loaded on to zonal rotors allow the use of conventional assay methods, whereas if tubes are used, no more than one or two assays can be performed on each
Ch. 3
CONDITIONS FOR SEPARATIONS
19
fraction unless micro-methods are developed. As it is simple to transfer a fractionation method between different types of rotor, it may be convenient to develop methods using swing-out rotors for then several samples may be run at a time. When an apparently satisfactory separation is achieved, the separation may be scaled up to a zonal rotor, and a full set of assays performed on the fractions so separated. The choice of rotor for isopycnic banding is much simpler than for a rate separation. One uses the fastest rotor available subject to three provisos. Firstly, the volume of gradient must be sufficient to band all the particles, the capacity of gradients used for isopycnic banding being high but not unlimited (see (j 2.2.1). Secondly, the hydrostatic pressure must not be so high as to damage the particles (see (j 8.1.2), this problem only becomes serious at speeds greater than 30,000 rev/min. Thirdly, more rapid banding will be obtained by use of rotors with short relatively wide tubes; this is an advantage, as the long times often required to obtain isopycnic banding can result in degradation of the particles. While swing-out or zonal rotors must be used for rate sedimentation, angle-head rotors may be used for isopycnic banding (see (j 2.2.3). Because of the relatively higher capacity of gradients used for isopycnic banding, it is less necessary to use zonal rotors for large scale separations. Zonal rotors can, however, be most useful when large volumes of sample must be processed, for isopycnic banding may be used simultaneously to fractionate and concentrate particles from a dilute suspension ((j 7.2.3 and (j 7.6.2). The rotors designed by Beaufay (9 5.1.2.9) are especially suited for isopycnic banding and have been shown to give much better results than swing-out tubes (Leighton et al. 1968). These rotors are however, not commercially available and appear to be difficult to operate.
3.3. Density gradient solutes The criteria for an ideal density gradient medium have been set out by Hartman et al. (1974). It should a) form solution covering the Sehjecr indvx p. 287
80
DENSITY GRADIENT CENTRIFUGATION
density range needed for the particular application, b) form solutions of low viscosity, c) possess some property, such as refractive index, by which its concentration may be measured, d) not damage the material which is being separated, e) be readily removable after the separation and f) not interfere with the analysis of the separated fractions. Water-soluble compounds which will form solutions covering the densities required for the separation of subcellular particles fall into three major classes, salts of the alkali metals, small neutral hydrophilic organic molecules such as simple sugars and hydrophilic macromolecules such as polysaccharides and proteins. A number of other compounds have also been proposed. The proporties of the different classes of density gradient solute are discussed below and the most commonly used compounds are listed in Table 3.1. 3.3.1. Salts of alkali metals Solutions of alkali metal salts have low viscosity and their concentrations may easily be measured by refractometry. Unfortunately, solutions of a useful density are of high ionic strength and tend to disrupt protein-protein and nucleic acid-protein bonds. Salt gradients are, therefore, used mainly for banding macromolecules such as DNA or RNA. The most widely used compound for banding DNA is caesium chloride (Flamm et al. 1972); other salts reported are sodium iodide (Anet and Strayer 1969b) and caesium sulphate, formate and acetate (Zolotor and Engler 1967) but the latter forms rather viscous solutions (Szybalski 1968). A given DNA will band at a different density in each of these media, probably due to variation in the amount of water which remains bound to the macromolecules (Szybalski 1968). The banding density in CsCl is essentially that of the anhydrous molecule (Petermann 1964). As discussed in Chapter 2, the binding of water reduces the separation between different types of DNA. The anhydrous density of RNA is higher than the density of saturated CsCl, but double-stranded RNA can be banded on caesium sulphate gradients and both single and double-stranded RNA on mixed gradients of CsCl and Cs,SO, (Szybalski 1968).
Ch. 3
CONDITIONS FOR SEPARATIONS
81
Bonding between macromolecules is generally disrupted by solutions of high ionic strength. Certain viruses can, however, be separated from fragments of the host cells by banding on CsCl (Anderson and Cline 1967) or potassium tartrate gradients (Reeve and Alexander 1970). Other nucleoprotein structures, such as chromatin and ribosomes can only be fractionated after fixation in glutaraldehyde or formaldehyde (Spirin et al. 1965; MacGillivray et al. 1972). CsCl is the material usually chosen for the banding of such fixed particles, as they appear to band at their anhydrous densities, so maximising density differences (but see McConkey 1974). CsCl is, however, expensive, and for large scale separations it may be worth while investigating other solutes. One class of structures which are little affected by high concentrations of salt are the serum lipoproteins. These have isopycnic banding densities varying between 1.2 and 0.96 and can be fractionated either by isopycnic banding or rate flotation (Lindgren et al. 1972). NaBr is the density gradient material normally chosen as being the cheapest which will cover the required density range. Finally one should remember that degree of hydration, and hence the banding density of any particular structure, will depend on how effective the gradient material is as a dehydrating agent. 3.3.2. Small hydrophilic organic molecules Sucrose is the most popular compound in this class. It will form solutions which cover the density range of all the larger constituents of cells. It has little effect on intermolecular bonding and is very cheap. Commercial sucrose, unless otherwise specified, is contaminated by other organic material including enzymes like ribonuclease (Hinton et al. 1969 inter alia). Most ultra-violet absorbing material, including all the ribonuclease present, may be removed by treatment ofthe sucrose solutions with activated charcoal (Steeleand Busch 1967). The procedure is very simple. The sucrose is dissolved to form a stock solution (we find that 2 M sucrose is a convenient concentration) and mixed with 50 g of activated charcoal per kg of sucrose for about 20 min. The charcoal is then removed by filtration in a Buchner funnel (at least two filtration cycles are needed and even Suhtrcr rr~dr\1’ 287
82
DENSITY GRADIENT CENTRIFUGATION
then a few very small particles may remain in the solution). As an alternative, the ribonuclease may be destroyed by treating the sucrose solutions with diethylpyrocarbonate (Williamson 1971). This reagent is added to the sucrose solutions at about 1 drop/250 ml. The solutions are allowed to stand for a few minutes when the diethylpyrocarbonate decomposes to ethanol and carbon dioxide. Solutions should therefore have their pH adjusted after treatment. Two points should be noted. Firstly diethylpyrocarbonateappears to be very effective against ribonucleases of plant origin, but does not inhibit some mammalian ribonucleases. Secondly, some people become sensitised to diethylpyrocarbonate which should therefore be handled with care. As an alternative to these procedures specially purified sucrose may be purchased at considerably greater expense. Sucrose itself inhibits enzymes when high concentrations are present in the assay medium and it will damage complex structures like mitochondria and whole cells (see Ch. 8). Sucrose solutions of a useful density have a high osmotic pressure and cause a considerable loss of water from membrane-bound organelles. Sucrose itself will form solutions with densities up to 1.29; densities of up to 1.37 may be achieved by the addition of glucose (Raynaud and Olenbusch 1972), or by dissolving the sucrose in D,O (Kempf et al. 1972), and such gradients have been used for the banding of nucleoproteins (see Table. 3.1). Other gradient materials in this class show little advantage over sucrose and, with the exception of glycerol, chloral hydrate and deuterium oxide, may be neglected. Glycerol gradients are sometimes preferred to sucrose gradients as the glycerol appears to protect enzyme activity (Freifelder 1973). Glycerol penetrates almost all biological membranes (Wallach 1967). For reasons discussed in 5 2.3 membrane-bound particles will band at greater densities in glycerol gradients than in sucrose gradients. The advantage of this is not apparent at the present time, but applications may be found at some future date. The high viscosity of concentrated glycerol solutions will, however, mean that prolonged centrifugation will be necessary. Chloral hydrate forms solutions with density up to 1.91 and has
0
TABLE 3.1 Properties of the more common density gradient materials. Material
Mol. Wt.
Max. density ofaq. soh.
uv
Ionic strength ofsolutions
Viscosity of 20", solutioni
absorbance
high
,
low
Price' fjl0Og
Uses'
J' W
References
~-
Caesium chloride
361.9
1.8
high
Sodium bromide
102.91
I .5
high
Potassium tartrate Sucrose
+ glucose D,O
~
5 P
1.888
Caesium sulphate
Sodium iodide
2
169.4
Glycerol
149.9
235.3 342.3
I .9
I ,485
I .3
high
high non-ionic non-ionic
~
i
-
-
2.954(5
-++
low
8.984
low
0.15
high
t
)
1 I. I 54 Banding of nucleic acids
low low
low
0.55
0.37 0.03 0.545 -
~
92.09
1.26
non-ionic
ti
low
0.13
and nucleoprotein Banding of DNA and RNA Fractionation of lipoproteins Bandingof DNA Banding of DNA and RNA Banding of viruses Very many Banding of chromatin Banding of ribonucleoprotein particles Banding of membrane fragments Rate zonal separation of proteins Fractionation of nuclei
WolfandBrown(1964)* see text Szybalski (1968)* Szybalski (1968) Wolf and Brown (l964)* Lindgren et al. (1972) Hinton et a1 (1974) Aneldnd Stidyer(I96Yb) Anet and Strayer (l969b) Birnie ( 1972) Reeveand Alexander (1970) Wolfand Brown(lY64). Dobrota (1971): Raynaud and Ohlenbusch (1972) Kempf et al. (1972)
r, 0
-U 0 =!
f
-n
0
2
rn
;zw z
2
p
v)
Wolland Brown (1964)* Wallach (1967) Schrcier and Staelielin ( I 973) Johnston and Mathias ( 1972)
00 W
Material
Mol. wt.
Max. density ofaq. s oh.
Ionic strength of s ohtions
Viscosity of20% solutionl
UV absorbance
Price' fjl0Og
Sorbitol Heavy water (D,O) Ficoll6
20 400,000
1.11 1.23
non-ionic non-ionic
+++
1.31
low low
Dextran
(72,000)'
1.05
non-ionic
t+l
low
69.000
1.12
non-ionic
++t
high
Bovine serum albumin
Ludoxg Renografinlz Urografinl3
3.7 6.05
Fractionation of microsomes 15.908 Separation of whole cells
1.219
non-ionic
+
low
_I I
614"
1.45
low
+
high
16
I .45
low
+
high
Banding of trachoma agent Rate sedimentation of yeast mitochondria Separation of actin Very many
6.62
-10
61415
uses-'
-1 5
Hypaquel4
Separation ofwhole cells and large organelles Separation of cells Separation ofcells Separation ofchromatin
Isopaque17
62815
1.45
low
f
high
-16
Separation ofcells
Conray's (Na iothalamate)
61415
1.6
low
+
high
-16
Separation of nucleoprotein
References
00 P
Clineand Dagg (1973) Neal et al. (1970) Trinick and Rowe (1973) Harwood (1974) Pretlow et al. (1969)* Wolf and Brown (1964)* Graham (1972a. b) Wolfand Brown(1964)* Mateyko and Kopac (1963) Miller and Phillips (1969) Harwood (1974) Pertoft and Laurent (1969)*
O
rn
f
-1
<
n s
>
E! rn
z -I
n rn
z -I E71 C
Ilinton and Mullock (1Y76)* Hinton and Mullock (1976)' Hinton and Mullock (1976)' Hinton and Mullock (1976)* Hinton and Mullock ( I 976)*
n
-
21 -
0
z
+
high
-16
+
low
0.24
Metrizamide19
789
1.46
non-ionic
I
Chloral hydrate
165.4
1.91
low
-t
Fractionation of chromatin Separation of nuclei Separation ofcells and ribonucleoprotein particles Separation ofchromatin
Rickwood et al. (1973) Mathias and Wynter (1973) Rickwood (1976)*
0
Hossdiny et al. (1973)'
*
References to properties of medium. It is difficult to locate exact figures for the viscosities of any density gradient solutes. + indicates a solution almost as mobile as water, + + a solution with a viscosity similar to a sucrose solution of the same concentration and + + a solution similar in viscosity to Ficoll at the same concentration. 2 The approximate prices are for the purest grade readily available in January 1974 and are for comparison only. 3 The list of uses is not exhaustive. We have tried to give more references for materials which we think may have some unexploited potential than for materials which are already widely used. 4 A s a 60% aqueous solution. 5 Especially purified RNase-free grade. 6 Manufacturer. Pharmdcia. 7 A relatively low molecular weight preparation. Dextrans of mean molecular weights between 50.000 and 300,000 are available. 8 Cohn Fraction V. 1
+
Manufacturer Du Pont lnc. 10 Ludox is an especially finely divided form of colloidal silica. Hence Mol. Wt. meaningless. 11 Very cheap, enquire from manufacturers. 12 Mixed sodium and methylglucamine salts of diatrizoic acid, manufactured by E.R.Squibb and Son and the Mallinckrodt Chemical Company. 13 Similar to above, manufactured by Schering Chemicals Ltd. 14 Sodium salt of diatrizoic acid, available from Winthrop laboratories Ltd. 15 Molecular weight of the free acid. 16 Enquire from the manufacturers. 17 Mixed sodium and methylglucamine salts of metrizoic acid, manufacturer by Nyegaard and Co. A/S, Oslo. Norway. 18 Mixed sodium and methylglucamine salt of iothalamic acid, available from May and Baker Ltd. Also available as free acid from Mallinckrodt ChemicalCo. 19 2(3-acetamido-5-N-methylacetamido-2.4.6,tniodobe~amido)-2-deoxy-D-glucose. Manufactured by Nyegaard and Co. A/S. 9
$
f9 3'
a
2 9
8z 0
z,
n 0 P m m n P
P
>
2f
00 VI
86
DENSITY GRADIENT CENTRIFUGATION
been used for the isopycnic banding of chromatin (Hossainy et al. 1973). Its effect on other subcellular structures is not known. 3.3.3. High molecular-weight organic compounds One solute stands out above all others. Ficoll (Pharmacia Ltd., Uppsala, Sweden) is a co-polymer of sucrose and epichlorhydrin with an average molecular weight of about 400,000 and was especially developed for density gradient centrifugation. It has little effect on biological particles and, unlike sucrose, does not inhibit enzymes even when it is present in high concentrations (Hartman et al. 1974). Ficoll, as purchased, is contaminated by some low molecular weight material which should be removed by dialysis before use. The viscosity of Ficoll solutions is higher than that of sucrose solutions of the same density, but this is counterbalanced by the fact that Ficoll does not penetrate biological membranes and only exerts a tiny osmotic pressure, so that membrane-bound particles band at lower densities in Ficoll than in sucrose. However, the tonicity of Ficoll rises exponentially with concentration, so that the tonicity of concentrated solutions is very high (Bach and Brashler 1970). With rate-zonal separations, the use of high molecular-weight solutes such as Ficoll allows a greater concentration of material in the sample zone than can be maintained with a solute such as sucrose as ‘sedimentation in droplets’ does not occur (see 0 2.1.2). Glycogen gradients have been used for the isopycnic banding of mitochondria and lysosomes (Beaufay et al. 1964). Glycogen gradients are difficult to handle as glycogen sediments markedly at quitea low centrifugal force and may be considered obsolete. Dextran gradients have been used for the isopycnic banding of microsomes (Graham 1972, 1973). The separations achieved are equivalent to those obtained in other tissues with Ficoll gradients. As the cost of dextrans and of Ficoll is about the same, there would seem little to choose between the two materials. WhileFicoll haslittleeffecton the functionof isolated cell organelles, it may be damaging to some living cells (Mathias et al. 1969; see also Ch. 7). A number of materials have been investigated for sepa-
Ch. 3
CONDITIONS FOR SEPARATIONS
87
rating such sensitive cells, including colloidal silica and iodinated aromatic compounds, both of which are considered below. Bovine serum albumin is perhaps the safest of all, but solutions of sufficient density are difficult to prepare and, when formed, are very viscous. The authors have no personal experience of albumin gradients, the strong solutions required for isopycnic banding are, said to be, difficult to prepare (Mateyko and Kopac 1963). However, albumin can be very useful for rate sedimentation of living cells (Ch. 7). 3.3.4. Other types ojdensity gradient solute While most materials suitable for use as density gradient solutes can be fitted into the three categories listed above, a number of other interesting materials have been proposed (Table 3.1). A finely divided form of colloidal silica (Ludox, duPont, Wilmington, Delaware) has been used for the banding of whole cells (Wolff and Pertoft 1972a), mitochondria (Lagercrantz and Pertoft 1972) and lysosomes (Wolff and Pertoft 1972b) and several other purposes. This material has a number of advantages over other gradient materials. Ludox solutions have no appreciable osmotic pressure and do not penetrate biological membranes. Unlike Ficoll solutions, Ludox solutions are very mobile, so that much shorter centrifugation times are needed. The size of Ludox particles is sufficiently large for gradients to be ‘self-forming’ (4 2.2.2) at moderate centrifuge speeds. Probably for the latter reason the angle-rotor will give very satisfactory separations (Wolff and Pertoft 1972b). Unfortunately, Ludox has two disadvantages. Firstly, it cannot be used in centrifugal fields of more than about 100,000 g as the silica particles will pellet. Secondly, concentrated Ludox solutions are not stable between pH 4 and 7.5 which is the region of greatest stability for most biological structures. The properties of Ludox as a density gradient material are reviewed by Pertoft and Laurent (1969). Iodinated aromatic compounds, such as are used as X-ray contrast media, have also been used as density gradient media. The most commonly used contrast media are mixtures of the sodium and methylglucamine salts of three derivatives of triiodobenzoic acid, I,,/>,‘,I ,,,d<’I p 287
88
DENSITY GRADIENT CENTRIFUGATION
diatrizoic acid, metrizoic acid and iothalamic acid (Hinton and Mullock 1976). These media have been widely used for the separation of blood cells and of microorganisms at different stages of their cell cycle; but not for the separation of subcellular structures. The compounds are somewhat difficult to handle as the free acid precipitates if the solutions are adjusted to an acid pH or on contact with solutions of a high ionic strength. The properties and uses of these compounds are discussed by Hinton and Mullock (1976) and Rickwood (1976). While the componds mentioned in the previous section are salts, a recently introduced X-ray contrast medium. Metrizamide (2-(3acetamido-5-N-methylacetamido-2.4.6.-triiodobenzamido)-2-deoxyD-glucose) is a completely covalent molecule and, as such, forms solutions of a markedly lower tonicity than its electrovalently bonded relatives. Metrizamide formssolutionswithdensitiesof up to 1.45g/ml. These solutions are somewhat more viscous than solutions of an equal density prepared from the ionisable contrast media, but are considerably more mobile than equivalent solutions of sucrose or Ficoll. The solutions are stable over a wide range of pH and ionic strength, and have been very widely used for example in the fractionation of unfixed deoxyribonucleoprotein (Hell et al. 1972) ribonucleoprotein (Hinton et al. 1974b) and of liver cells (Munthe-Kaas and Seglen 1974). The properties of subcellular particles in Metrizamide gradients are similar to those in sucrose gradients, although banding densities appear to be slightly lower. Nucleic acid and nucleoproteins band at remarkably low densities in Metrizamide gradients indicating a high degree of hydration. Protein molecules on the other hand, band at densities similar to those found in CsCl gradients and it has been suggested that this may be explained by binding of the Metrizamide to the protein. This binding would, however, appear to be freely reversible, for the inhibition of certain enzymes which is observed when assays are carried out in the presence of Metrizamide may be reversed by dilution. Further information on these and other points may be found in Rickwood (1976) and it is clear that Metrizamide and related compounds will be very useful in density
Ch. 3
89
CONDITIONS FOR SEPARATIONS
gradient centrifugation in situations where more conventional media give poor results. Metrizamide seems to be the most attractive of all the X-ray contrast media, and is not cytotoxic (Munthe-Kaas and Seglen 1974). Finally one should mention density gradient solutes that cause denaturation. Thus, gradients of dimethylsulfoxide and hexadeuterodimethylsulfoxide and of methanol and methoxy ethanol have been used on the fractionation of RNA. These compounds are chosen to break up any aggregates which may be formed during the separation of RNA and also reveal ‘hidden breaks’ (Parish 1972;see also Hastings et al. 1968) in the RNA chain. Gradients of organic solvents have been used for the purification and subfractionation of nuclei prepared in non-aqueous media (Shakoori et al. 1972). It should be emphasized that if zonal rotors are to be used with non-aqueous media, than the seals and O-rings must be made of solvent-resisting material. The centrifuge manufacturer should be consulted before any experiments are undertaken.
3.3.5. Choice of gradient material There is no density gradient material which is ideal for all separations. For rate-zonal separation of subcellular components, one would normally choose sucrose, for it has little adverse effect in relatively dilute solutions which are normally used. High molecular weight compounds, such as Ficoll have theoretical advantages (Q 2.1.2) but the cost for routine use is high. The choice of gradient material for the rate separation of living cells is complicated by the difference in the tolerance of different cell types to different gradient materials (see Ch. 7). Obviously the most important requirement for a density gradient material is that it should not damage or alter the material which is to be separated. Thus the salt media, which have most of the physical properties required for a medium on which to carry out isopycnic separations, cannot be used for many applications. In general, high molecular weight solutes such as Ficoll, when used in the high concentrations needed for the isopycnic banding of cell S,,h,cr I
,,,‘/<,A
,I
2x7
90
DENSITY GRADIENT CENTRIFUGATION
organelles, cause less damage to subcellular organelles than sucrose. However, all the high molecular weight materials are considerably more expensive than sucrose. An additional consideration is that methods have been worked out for the separation of organelles from a large number of different tissues on sucrose gradients, but much less work has been carried out on separation in Ficoll gradients. The banding densities of particles in sucrose and Ficoll gradients are not simply related; if Ficoll is used as the gradient medium, preliminary experiments will usually be necessary to establish the optimum conditions for a separation. As regards the new density gradient materials already described, a user must be prepared to do the spadework of a pioneer.
3.4. Choice of gradient 3.4.1. Gradient for rate-zonal separations For most purposes, a simple linear gradient suffices for rate-zonal separations. The density of the light end of the gradient is usually determined by the need to support the sample and to minimize ‘sedimentation in droplets’ (4 2.1.2). If membrane-bound particles are to be separated, than the sample must be resuspended in 0.25 M sucrose, or some other isotonic medium, to avoid lysis. Thus the gradient must start at a density equivalent to at least 0.35-0.4 M sucrose. The density of the heavy end of the gradient should be less than the density of the least dense particle in the sample, otherwise the pattern obtained after centrifugation may be confused by the isopycnic banding of the lightest particles. The least dense fragments in most tissue homogenates - apart from lipid droplets are fragments of the Golgi apparatus with a density of about 1.12; equivalent to about 0.9 M sucrose. Within this limit it has been demonstrated that the greater the ‘slope’ of the gradient, the better the resolution which will be obtained (Neal and Florini 1972). Thus a linear gradient ranging from 0.5 to 1.0 M sucrose (about 15-30% w/w) is used for many separations. When macromolecules or small particles, such as RNA or ribosome subunits, are to be
Ch. 3
CONDITIONS FOR SEPARATIONS
91
separated, it is not necessary to suspend the sample in an isotonic medium, so that the concentration of the light end of the gradient may be reduced. A 5-20% w/w gradient has often been employed (McConkey 1967). The reduction in the mean viscosity as compared to the 15-30% gradient results in a more rapid separation. The only other type of gradient which has been much used for ratezonal separations in swing-out rotors is the isokinetic gradient (Noll 1967, 1969; Steensgard 1970). Isokinetic gradients are designed so that the increase in density and viscosity of the gradient material just balances the increase in centrifugal force along the length of the centrifuge tube. As a result, particles move a t a constant speed and sedimentation coefficients can be calculated provided that the sedimentation coefficient of one particle in the mixture is known. Isokinetic gradients are also reported to give better resolution than linear gradients (Noll 1967, 1969; Steensgard 1970). There are, however, two problems. Firstly, an isokinetic gradient is calculated for one particle density only. If the particles to be separated differ in density, much of the value of the isokinetic gradient will have been removed. Secondly, isokinetic gradients must be calculated for each type of centrifuge tube. The authors have no personal experience of the use of isokinetic gradients, but are not, a t the present time, convinced that they offer significant advantages over linear gradients. The same comments apply to isovolumetric gradients for use with zonal rotors (Pollack and Price 1971). These gradients are designed to compensate not only, as with isokinetic gradients, for the increasing centrifugal force along a gradient, but also for the radial dilution which occurs with bands separating in zonal rotors. 3.4.2. Gradient ,for isopycnic separations Isopycnic separation may be performed either on preformed gradients, or by exploiting the density gradients which are formed by many solutes after prolonged centrifugation of an initially homogeneous solution in an high centrifugal field (5 2.2). The shape of such equilibrium density gradients is determined by the properties of the solute and the centrifugal field. With low molecular weight density gradient solutes S,,h,c~I ,,,
92
DENSITY GRADIENT CENTRIFUGATION
equilibrium gradients form very slowly. Hence, when the size of the particles to be separated is such that they can be banded after centrifugation for 18 h or less, preformed gradients must be used. The preparation of such gradients is discussed in 8 4.1.1. Sucrose and glycerol do not sediment significantly even in high centrifugal fields, hence preformed gradients must always be employed with these solutes. The authors have no information on the behaviour of Ficoll or dextran gradients on prolonged centrifugation. When very prolonged centrifugation ( > 36 hr) is required, there is probably no advantage in preformed gradients, as the gradient solute will redistribute to give the same final gradient as would have formed from an initially uniform solution of the same average density. Ludox rapidly forms equilibrium gradients even under moderate centrifugal fields (Pertoft and Laurent 1969) and hence one does not usually bother to make preformed gradients. The design of preformed gradients for isopycnic separations is very simple - the density range should cover the density range of the particles in the solution which is to be fractionated. If the sample is incorporated in one of the gradient solutions (6 4.2.1), or if there are known to be no low molecular weight contaminants, then one usesa gradient such that the least dense particles band close to the top of the tube (or just outside the core in the case of a zonal rotor) and the densest particles band just above the curved portion at the base of the tube or at the edge of the zonal rotor. If low molecular weight contaminants are suspected, then one would normally layer the sample over the surface of the gradient, as in rate separations, and design the gradient so that the lighter particles band about one-third of the way down. The upper third of the gradient serves as a barrier to small molecules diffusing from the sample zone. Three factors influence the design of ‘equilibrium’ density gradients, the nature of the solute, the centrifugal field and the initial density. The first two factors have been discussed above and in § 2.2.2. The initial density should always be slightly higher than the isopycnic banding density of the most dense component of the mixture to be fractionated (Flamm et al. 1972). If this precaution is not
Ch. 3
CONDITIONS FOR SEPARATIONS
93
observed. the densest particles may pellet before the density gradient has formed. 3.4.3. Design of complex gradients When a mixture contains morphologically distinct particles which are of similar size, but differ in density and in addition particles which are of similar density but differ in size, then either the fractionation must be carried out in two stages - firstly separating by particle size, then by particle density (or vice versa), o r complex gradients must be employed. An example of such a complex mixture is the crude nuclear fraction prepared from a rat liver homogenate. By use of a suitable gradient all except the smallest particles can be separated from each other. Inspection of the S-p diagram (Fig. 3.1) shows that all the larger particles (plasma membrane, red blood cells, nuclear and whole cells) differ from each other in density. The size range of particles is too great for them all to be separated by rate sedimentation. On the other hand, plasma membrane sheets and endoplasmic reticulum fragments differ only in size. Hence the larger components of the mixture must be fractionated by isopycnic banding, the smaller by rate separation. One therefore designs a twopart gradient with a fairly flat initial section followed by a steep outer section ; the larger organelles are separated by their isopycnic banding densities in the outer portion of the gradient, and are separated from mitochondria and smaller particles by their different sedimentation rates through the flatter earlier part of the gradient. While complex gradients of the type illustrated can be most useful, they cannot always be applied. For some mixtures two-stage procedures must be employed if complete separation is to be obtained. An example of such a mixture is the M + L fraction prepared from rat hepatoma homogenized in the presence of Ca2' (Fig. 3.6). In this case the plasma membrane sheets are not sufficiently different in S value from the mitochondria to be separated by rate sedimentation, but they can be separated by isopycnic banding. However, before this separation can be achieved, the mitochondria must pass right through the plasma membrane band. By the time the smallest Si,hp,c/ ,,nfc\ p 287
94
DENSITY GRADIENT CENTRIFUGATION
"l
t
I
Plasma membrane fragments
I
I
I
I
101
id
103
irp
I
105 Sedimentation coefficient
I
(S)
106
1 107
Fig. 3.6. S-p diagram showing the approximate density and sedimentation coefficients of the major components of a crude M L fraction prepared from rat hepatoma.
+
mitochondria have sedimented clear of the plasma membrane band, the largest lysosomes will have sedimented into this region (Fig. 3.7). Hence a two-stage procedure must be employed to obtain complete fractionation of the mixture, although plasma membrane sheets may be separated in a single step as their low density and relatively large size means that they will float more rapidly from dense sucrose than any other component of the mixture. Complex gradients are also useful when particles with a very wide range of sedimentation coefficient are to be fractionated in a single experiment. Thus, liver lysosomes range in sedimentation coefficient from 1000 to 7000 S . If a simple linear gradient is used these cannot be separated simultaneously from mitochondria and microsomes. If, however, a two-part gradient is used the first part being flat to separate small lysosomes from microsomes, while the latter part is steep to slow down the sedimentation (hence maintain the separation) of large lysosomes and mitochondria, then a good separation can be achieved (Fig. 3.8). Another method which can be used for slowing down the sedimentation (or flotation) of the larger particles in a mixture is to change in the middle part of the gradient to a different gradient material which forms solutions of higher viscosity
Ch. 3
CONDITIONS FOR SEPARATIONS
95
1
0
20
10 fnctia b.
.-.-.,
Fig. 3.7. Resolution of the different classes of particle from a hepatoma M f L fraction after centrifugation for 75 min at 9,000 revs/min in an HS zonal rotor. a) -- -- - - _ _ _- , density , protein; b) 0-0-0, 5’-nucleotidase (plasma membrane fragments); acid phosphatase (lysosomes); +--- +--- + succinate dehydrogenase (mitochondria); 0--0--I 3 glucose-6phosphatase (endoplasmic reticulum fragments) (from Prosper0 1973).
than the first. We have attempted to use this approach in fractionating serum lipoproteins (Mallinson and Hinton 1973 ; Hinton et al. 1973b), but have obtained less good resolution than we had hoped, possibly due to unequal diffusion of the two gradient materials causing inS ! d > j ~rridca.rp. l 287
96
DENSITY G R A D I E N T CENTRIFUGATION
Fig. 3.8. Subfractionation of a post-nuclear fraction from rat liver by centrifugation in an HS zonal rotor. Use of a convex gradient permits the simultaneous separation from mitochondria ( x - . - - . x ) and microsomes of lysosomes .--@---@ (0.. . . 0.. . . 0). -------- distribution of protein (redrawn from Burge 1973). The marker enzymes were the same as were used in the experiment illustrated in Fig. 3.7.
stability in the gradient similar to the sedimentation in droplets described in 5 2.1.2. As was stated at the beginning of this section, gradient design is more of an art than a science, and as such is difficult to communicate. Computer modelling of the movement of particles through a density gradient (Steensgard et al. 1973, 1974) may help to avoid actually performing so many zonal runs, but it is still necessary to design the gradients which will be tested on the computer model. In summary, all that can be said is that if any class of particles in a mixture has a unique combination of size and density, then that class can be separated from all others by density gradient centrifugation. We hope that the discussion in this section will have helped readers to select the best conditions.
CHAPTER 4
Centrifugation in conventional rotors
As has been discussed in Ch. 2 we believe that, providing operating techniques are optimised, it is possible to obtain equally good resolution in conventional swing-out rotors and in zonal rotors. The only difference is that much greater amounts of material can be separated in the zonal rotor. However, as the techniques for operating conventional and zonal rotors do differ markedly, it is convenient to discuss practical methods in separate chapters.
4 .I . Ra te-zonal centrifugation 4.1 . I , Preparation of the density gradient
Most density gradient separations in the tubes of swing-out rotors are performed on simple linear density gradients (see Q 3.4.1). If the aim is to study the distribution of particles through the gradient rather than prepare a specific fraction, the gradient is normally underlaid with a sufficient volume of a cushion of dense and viscous liquid to fill the curved portion at the bottom of the tube and to prevent any loss of particles in to a pellet. The apparatus used to prepare linear density gradients is simple (Fig. 4.1). Two cylindrical vessels of equal diameter are connected together at their base via a tap. One of the vessels is fitted with a stirrer and an exit port through which the gradient is drawn off. Several commercial gradient makers have been constructed on this principle. Theoretical analysis of the operation of such gradient makers shows that the gradients formed will not be precisely linear, as the two columns of liquid will not have exactly 91
Strhiarr rrrdcjr y
287
98
DENSITY GRADIENT CENTRIFUGATION
Stirrer
CL
Fig. 4.1. A simple gradient maker to prepare linear gradients.
the same height. The deviation from linearity is related to the difference in density across the gradient and with the 15-30% sucrose gradients commonly used for rate zonal separations the deviation is very small. A significant deviation is, however, found with the steep salt gradients used in the separation of lipoproteins. Exactly linear gradients can be obtained by using one of the more expensive commercial gradient makers (see Table 4.l), but it is debatable whether the extra precision has any significant effect on the separations achieved. Density gradients are not used singly, but in groups of 3 or 6 , depending on the number of tubes which can be centrifuged in a particular rotor. The type of apparatus discussed above can readily be adapted to prepare 3 gradients simultaneously provided that 3 or 6 tubes can be fed simultaneously with the same gradient. This can be done by using a multi-channel peristaltic pump. The pump must be fitted with 3 pieces of pump tubing which give exactly the same flow rate. The wide tolerance to which such tubing is manufactured makes the selection of suitable pieces time-consuming. Immediately after the exit from the gradient maker, the liquid stream is divided into three or six using a suitable junction and each part
TABLE4.1 Commercially available density gradient makers.
c,
P P
Manufacturer'
M.S.E.
Beckman
Buchler
1
P
9
Model
0
Types of gradient formed
Linear
Gradient former
Linear
5.25 or 70 mi 60 mi
Linear
50 ml
Phoenix Searle instruments Isco
Varipump Searle gradient former3
Fully programmable Exponential
Model 570
Convex, linear or concave
Isco
Dialagrad Model 382
Fully programmable
> 100 m14
LKB
Ultragrad
Fully programmable
> 100 mls
100 ml
50 mi 80 ml
5
4
5
1
For addresses see Appendix 1.
2
See Fig. 4.1. Similar to apparatus in Fig. 4.5.
h,
9
3
Comments
-
Small tube gradient maker2
2-5 102 and 2-5 109
Max. volume (per tube)I
4
Triple outlets to enable simultaneous production of 3 gradients when a suitable pump is used Programmed syringes - so will form linear gradients even when solutions differ widely in density Conical chambers allow the preparation of gradients varying from 5-50 ml with the same unit 2-5104 has triple outlets as with M.S.E. Gradient drawn on template Supplied with calculator for producing isokinetic gradients Automatic system which allows a series of centrifuge tubes to be sequentially filled with identical gradients Gradient dialed up in 11 points. A linear gradient is formed between each point Gradient drawn on template
Flow rate up to 640 ml/hr. Our reservations on use of the Ultragrad for making gradients for zonal rotors (9: 5.1.4.1) do not apply in this case.
n 0 Z
m e
r!
0 Z
r 7a
2z
W
100
DENSITY GRADIENT CENTRlFUGATION
is pumped through one channel of the pump. This will provide 3 or 6 identical gradients. In general, complicated gradients are not used with swing-out rotors. Some commercially available gradient makers, are listed in Table 4.1. Some will only produce linear gradients as described above; curved gradients may be prepared by use of a closed mixing chamber (see end of this section); uses of complex gradients are discussed in 4 3.4. The most complicated gradient makers listed in the table are even more versatile, but it is debatable whether the facilities they offer are, in practice, worth the high price. The gradient must be introduced into the tube with the minimum of disturbance. The properties of the plastic walls of the centrifuge tube determine how the gradient should be introduced. Polypropylene is reasonably wettable by aqueous solutions and it is therefore simplest to touch the end of the tube leading from the gradient maker on to the wall of the centrifuge and allow the stream of liquid to run slowly down the wall on to the top of the forming gradient (Fig. 4.2a). In this case the gradient should be prepared ‘heavy end first’ and the cushion placed in the bottom of the tubes before starting the preparation of the gradient. This method is not suitable for use with polycarbonate or polyallomer tubes. These plastics are not easily wettable by water so that the gradient material tends to gather into large drops around the tip of the delivery tube coming from the gradient maker. These drops finally detach, and cannon down the side of the tube and would disturb (or even destroy) the gradient. If Polyallomer tubes are soaked in spent chromic acid for 24 hr and then washed extensively in distilled water (Wallace 1969) the plastic becomes readily wettable. This also works with polycarbonate. Alternatively one can avoid such drastic measures by feeding the gradient to the base of the centrifuge tube through Mine delivery tube (Fig. 4.2b). In this case the gradient should be built up light end first (i.e. the light solution is placed in the moving chamber) and the cushion should be introduced at the end of the gradient. As mentioned earlier, the simple two cylinder gradient maker described above does not produce precisely linear gradients. An
Ch. 4
101
CONVENTIONAL ROTORS
From gradient maker
From gradient maker
Ice I
Ice bath -
I
I (a)
(b)
Fig. 4.2. Loading of gradient into a centrifuge tube. a) With a wettable tube. The tube is held at an angle of about 75" by a spring clip and is immersed in an ice bath. The tube carrying liquid from the gradient maker is held so that it just touches the side of the centrifuge tube. b) With a non-wettable tube. The tube is held vertically in an ice bath and the gradient is pumped through a probe leading to the bottom of the tube.
alternative design, which in theory should avoid this problem, is described by Roodyn (1972). Here the solution is pumped from the reservoir to the mixing vessel by a peristaltic pump, and the gradient is extracted from the mixing vessel at precisely twice the rate. Theoretical analysis shows that this will give an exactly linear gradient (see Appendix 4). The gradient makers described above are suitable only for linear gradients. As discussed in Q 3.4, isokinetic gradients (No11 1967, 1969) may occasionally be required. These can be approximated by the 'exponential' gradients formed when a fixed-volume mixing vessel is employed. One design of such an apparatus is shown in Fig. 4.3. Liquid is fed from the reservoir C, into the air-tight mixing vessel which initially contains a second liquid C , .Providing that the mixing Si,h,c< t ,mh
Ip
287
102
DENSITY GRADIENT CENTRIFUGATION
Fig. 4.3. Normal method for the preparation of an exponential gradient. The less dense solution (Ci) is placed in an air tight mixing chamber. The denser liquid (C,) is placed in a reservoir. When the tap from the reservoir is opened the liquid from the reservoir enters the mixing vessel and displaces some of the liquid already there. This is either allowed to flow out freely (as in diagram) or is pumped out. The liquid ih the mixing vessel is mixed by a magnetic stirrer. The syringe serves to pressurise the mixing vessel to minimise the initial surge of liquid from the reservoir when the tap is opened. The gradients formed by this type of gradient maker are shown in Fig. 4.5. (from Noll 1969).
vessel is completely air-tight, the rates of flow in to and out of the mixing vessel are identical and gradients of the forms shown in
Ch. 4
103
CONVENTIONAL ROTORS
5
10 15 20
5
10
15
20
Volume (ml) Fig. 4.4. The shape of ‘exponential’ gradients formed by the type of apparatus illustrated in Fig. 4.3. The total volume of gradient formed in each case is 20 ml. The liquid initially present in the mixing vessel is assumed to have a density of 1.0 g/ml. that i n the reservoir a density of 1.2 g/ml. The four curves illustrate the effect on the shape of the gradient of variations in the volume of solution initially placed in the mixing vessel.
Fig. 4.4are produced. As can be seen, one can achieve a considerable variation in shape by varying the ratio of the volume of liquid in the mixing vessel to the volume of gradient required. The calculation of the concentration of liquids A and B and the volume of liquid in the mixing vessel so as to obtain an isokinetic gradient is described by No11 (1970). It should be noted that the gradient maker shown in Fig. 4.3 will not give precisely the calculated gradient due to the surge when the tap from the reservoir is opened. Nevertheless this design gives satisfactory results in practice. Use of a completelyfilled mixing vessel creates problems in mixing solutions which can only be solved with elaborate apparatus such as that designed by Anderson and Rutenburg (1967). 4.1.2. Layering of sample on to the gradient In rate-zonal separations the layering of the sample on to the gradient and the acceleration of the rotor are the two most critical steps. The sample must be stable according to the criteria of 8 2.1.2, Stihicir
i n i h
p 387
104
DENSITY GRADIENT CENTRIFUGATION
alsoonemustavoidmixing the sample with the top part ofthe gradient. Everybody develops their own favorite method for layering samples, but it is advisable to practice layering dyes before attempting real separations. The authors know, from personal experience, how easy it is to become satisfied with separations which are much worse than could have been achieved with a little extra care. The technique preferred by the authors is illustrated in Fig. 4.5. It is much easier to obtain a good starting band with polypropylene tubes or polycarbonate tubes treated as described in 8 4.1.1 than with tubes made from water-repellent plastic.
(a)
(b)
(C)
Fig. 4.5. Method for layering a sample on to a density gradient. The steps are a s follows: a) The tube is slightly titled and the pipette containing the sample is touched against the meniscus at the wall of the tube. b) The pipette is moved slightly upward to leave a thin channel of liquid between the pipette and the miniscus. c) The sample is run slowly out of the pipette.
In (j 2.1.2 we discussed the theoretical advantages of layering the sample itself in a density gradient. This may be done simply as described by Williamson (1970). The sample is drawn up into a relatively large pipette, for example, if a 0.1 ml aliquot were to be layered, one would draw up 0.11 ml aliquot into a 1 ml pipette. The pipette is then immersed in the less dense of the gradient solutions and a volume of liquid equal to the volume of sample to be layered
Ch. 4
105
CONVENTIONAL ROTORS
(in the example 0.1 ml) allowed to flow into the pipette. Mixing in the pipette forms a small gradient. The liquid is then layered on the main gradient in the normal way. Although this technique is useful where small volumes of concentrated solutions are to be layered onto density gradients, for theoretically layering in a gradient is of advantage when a separation is limited by spreading of the sample band due to sedimentation in droplets. The authors have found that samples are rarely so concentrated that this is the case. When other factors limit resolution, there is no advantage to be gained by layering in a gradient. Thus we would recommend anyone using rate sedimentation routinely to try running in parallel samples layered as described in the first paragraph of this section and samples layered in a gradient to see whether they obtain any improvement by using the latter technique. Up till now we have assumed that the sample is to be layered on top of the gradient. However, as described in 8 3.1, rate flotation can sometimes be a useful alternative to rate sedimentation. In this case the sample must be introduced under the gradient. The technique which we use is illustrated in Fig. 4.6. It must be emphasized that on no account must air bubbles be allowed to enter the gradient, for these may pick up small portions of the sample and carry them through the gradient. A second point is that when, as in the illustration, a liquid underlay is used, ‘sedimentation in droplets’ of the samples into the underlay will cause severe spreading of the sample band (for discussion see Q 2.6.1). It may therefore, be preferable to fill the curved portion of the gradient with an epoxy plug as described by Lindgren et al. (1972) and not to use any liquid underlay. 4.1.3. Centrifugation The centrifuge rotor should be accelerated slowly and decelerated without use of the brake. The powerful motors of modern centrifuges give very high initial accelerations. The resultant Coriolis force can cause considerable mixing of the sample with the upper part of the gradient. Naturally, these effects are more pronounced with lowviscosity salt gradients (Fig. 4.7), but also occur on sucrose gradients. S#rh,c% I I,ld<,, I’
2x7
Fig. 4.6. Method for layering a sample under a density gradient. If necessary, the density of the sample should be adjusted by the addition of solid gradient solute until it is denser than the densest portion of the gradient. The tube containing the density gradient should be held firmly in a stand. We use the displacement apparatus marketed by MSE Scientific Instruments (Fig. 4.14). The steps illustrated are as follows: a) Sleeve a piece of flexible tubing (narrow silicone tubing is most suitable) on to a thin length of stainless steel tubing (metal probe) and attach a 1 ml syringe. b) Lower the metal tube through the gradient to the bottom of the tube and clamp in place. c) Suck until the gradient is just drawn into the bottom of the syringe. Clamp the silicone tubing with a pair of artery forceps. d) Suck up the sample in to a syringe. Attach to the tubing and suck very gently to remove all bubbles from the line. e) Inject the sample to the bottom of the tube. f) Clamp the tube. Replace the sample syringe with a syringe full of underlay. Suck to remove all bubbles from the line. Inject the underlay. Withdraw the injection tube.
Ch. 4
107
CONVENTIONAL ROTORS
*?I irn
'280 nm
1
s
(bl
0.E
\ 1
0
10 Volume (ml)
I
20
C
20
10 Volume (ml)
Fig. 4.7. Damage to a separation due to over-rapid acceleration and deceleration. The experiments illustrate the results obtained when separating human serum lipoproteins in NaBr gradients of densities 1 to 1.2 g/ml. in the tubes of 3 x 23 ml swing-out rotor as described by Hinton et al. (1974). a) The rotor was accelerated freely and decelerated using the brake. b) The rotor was accelerated slowly and decelerated without use of the brake.
The greatest angular acceleration, and hence the greatest damage to the sample zone, occurs in acceleration through the first few hundred revs/min and cannot be controlled by the normal speed control fitted to ultracentrifuges. If the centrifuge does not have a control permitting slow acceleration, a rheostat should be fitted in the motor circuit. This should only be done with the advice and approval of the centrifuge manufacturer. An alternative approach is to use a rotor with a horizontal as well as a vertical pivot so that the line of buckets does not have to lie along a radius of the rotor. In this case the bucket may swing so that the resultant centrifugal field will always act along the axis of the bucket so that there will be no force tending to cause mixing. Such a modified rotor has been made for use with I.E.C. centrifuges (Leif et al. 1972). As the bands of separated particles always broaden somewhat during centrifugaSuhp I
,,I&\
p 287
108
DENSITY GRADIENT CENTRIFUGATION
tion, controlled deceleration is less important than controlled acceleration. Nevertheless, we prefer to decelerate without use of the brake. 4.1.4. Recovery from the gradient Essentially, there are four ways in which one can recover gradients. Three of these are illustrated in Fig. 4.8, the fourth, which we do not recommend is to chop up the whole gradient column, tube and all. Alternatively, one may use one of the special Strohmaier cells marketed by Christ, but the authors cannot see how these will give any significant advantage over more conventional methods of displacement. Systems for fractionating density gradients may either be brought complete (Table 4.2) or be assembled partially or completely in the displacing
Gradient out
Gradient out
Air or water in
(a)
Fig. 4.8. Methods for recovering fractions separated by density gradient centrifugation. a) A dense liquid may be pumped to the base of the gradient through a probe. The gradient is then displaced upwards and channelled out through a special cap. b) The base of the tube may be pierced and the gradient allowed to run out. The rate of displacement can be controlled by placing a cap with a suitable valve on top of the tube. c) A cap may be fitted on top of the tube and either air or a light liquid pumped in. The gradient is displaced through a probe which leads to the base of the centrifuge tube.
TABLE 4.2 Commercially availahle apparatus for fractionating densiry gradients. M anufacturcrl
Modcl
FIX USC
Detection
witti
Mode o f
rommenrs
dirphuemm L -~
a ) Simple gradient fractionators Beckman
~
not including pumps or UV monitors Beckman
-
Upward or downwards displacement Tube piercing only
M.S.E.
M.S.E.
Christ
Tube piercer
Buchler
2-5110 2-5150
2-5162
Upwards or downards displacement Tubz piercing
Christ I " x 3"
4'' x 2" tubes
Tube piercing only Upward displaccrncnl
Spinco o r IEC Any Any
hpualiQi1
Tlir ~r.tcliciiLia suckrd U N thruugh a
probe which follows the liquid surface b) Gradient fractionation system IK O 640
lSCC0
Y r'
:
Giltbrd
I
Upward displacemenl
All Beckman & IEC. UVColorirlleler
Upward displacemenl
2480
Most others Beckman wit1 fit some o t h m
Upward displacement
?
For addresses see Appendix 1.
$2-
U V Colorinieler
183
%
2
down the lube
All Beckman & IEC. Most others
Spectrophorometer
Franion collwlor included. \'ill take sciiitltlation vials. Fraction 'miinrled OF by viping delivery lube befiirc advancing collcclor A simpler sptcm. Fraction collector not included Very sniall hold-up beforc flow cell. 5 . 5 4 flow cell
110
DENSITY GRADIENT CENTRIFUGATION
laboratory or workshop. Purpose-designed instruments such as the ISCO density gradient fractionator will probably give better results than home-built systems (Fig. 4.9), for the designer of the purposebuilt system is able to minimize the lengths of flow lines and avoid sharp curves which will cause mixing in the gradient. The designer of the home-made system on the other hand, is limited by the geometry of the monitoring and collection equipment which is being used. However, the cost of purpose-built systems will rule them out for many laboratories. We will therefore concentrate mainly on design of home-made systems. The apparatus used for fractionating density gradients by upward displacement (Fig. 4.9) consists of a cap with a conical interior which fits on top of the centrifuge tube. At the apex of the cone is a tube
Fig. 4.9. The apparatus used in our laboratory for fractionating density gradients. For the sake of clarity the flow cell is shown separate from the spectrophotometer. Although a peristaltic pump, as shown here, is usually quite satisfactory, a motorised syringe would be better if very viscous gradients are to be fractionated. Note that the tubing between different components of the system is, inevitably, rather longer than in a purpose built apparatus.
Ch. 4
111
CONVENTIONAL ROTORS
to carry away the displaced gradient. Through the cap passes a tube which reaches to the base of the centrifuge tube. The cap is sealed to the tube either by the top of the tube pressing against a flat O-ring (Fig. 4.10a) or by a plug fit (Fig. 4.1Ob). The first design is now incorporated into several commercial gradient fractionation devices (Table4.2). The second design is easier to make in a workshop, but the sealing can be affected by distortion of the tubes during centrifugation. The O-ring shown in the drawing compensates for this to some extent, but does not completely prevent problems. Whatever the design of the displacement cap, the gradient is displaced by pumping dense liquid to the base of the gradient. For this, either a pulse-free peristaltic pump, or a motorised syringe (such as is used for perfusion and for applying samples to zonal rotors (Q 5.1.3.3) may be used. The latter is preferable, as the flow rate is much less affected by variations in the back pressure. Essentially the same apparatus can be used for the fractionation of gradients by downward displacement. The apparatus manufactured by Beckman-Spinco is, in fact, deliberately designed to be used either Gradient out O-ring 1
O-ring
1 (a)
Displacing fluid
(b)
Fig. 4.10. Designs of caps for use in the fractionation of gradients by upward displacement. a) The top of the centrifuge tube is sealed by clamping against an 0 ring. The drawing shows the lay-out used in the MSE density gradient fractionator. b) A simpler 'plug fit' in which the tube is sealed by the pressure of the O-ring against the walls of the tube. Sirhlccl rtrdat p 2x7
112
DENSITY GRADIENT CENTRIFUGATION
way. When fractionating by downward displacement, great care must be taken to avoid any pellet forming during centrifugation, or material washed from its surface may contaminate the upper parts of the gradient. The probe is lowered to the very base of the tube and air pressure applied at the surface of the gradient (Fig. 4.8~).The curved portion of the centrifuge tube acts in the same way as the funnel in upward displacement devices in directing the gradient to the exit tube, but it is less efficient. The gradient is inverted during passage up the displacing tube with the resultant risks of mixing, and there are disadvantages in monitoring, as will be discussed in the next section. In fact, this is the least satisfactory of the three methods discussed. The earliest method for recovery of a gradient was to pierce the base of the tube and allow the gradient to drip out. The obvious disadvantage of the method is that a new centrifuge tube must be used for each experiment. In addition, there are problems, shared with the previous method, when the optical density of the displaced gradient is to be monitored. The apparent advantage of the method is that it minimizes disturbance to the gradient, but even so the resolution appears to be less good than that obtained by upward displacement. Fractionation by upward displacement thus appears to be the best method for fractionating density gradients. 4.1 S . Monitoring the displaced gradient In practice, the only monitoring which is carried out on material separated in the tubes of swing-out or angle-head rotors is the measurement of the extinction, usually at 260 or 280 nm, to determine the overall distribution of material through the gradient. Such monitoring is usually performed during displacement of the gradient (Beckman market quartz centrifuge tubes, which may be spun up to about 400,00Og, in which the distribution of the separated fractions can be monitored in situ). The geometry of the flow cell is most important. Ideally a tubular cell, should be used. Cells of this shape are fitted into UV monitors such as the ISCO UA5 monitor, or the
Ch. 4
113
CONVENTIONAL ROTORS
LKB Uvicord, but cannot, in general, be used with general-purpose spectrophotometers which require rectangular flow cells. Gradient inversion which may cause mixing and turbulance in the flow cell, must be avoided at all costs. As a second choice, a cell allowing liquid te enter at the bottom and leave at the top is to be preferred. If the geometry of the spectrophotometer demands that liquid should enter and leave at the top of the cell there should be a short ‘upwards’ section leading to the light path. The least favourable design is the simple ‘U’ tube. Readers are strongly recommended to check the geometry of flow cells with manufacturers before they order. Any reasonable spectrophotometer or column monitor can be used to monitor displaced density gradients provided it accepts a flow cell, although in some cases the manufacturer will not have provided adequate access to the cell compartment so that special holes must be drilled into the sample compartment if flow lines are not to be inordinately long. The spectrophotometer should either give an output linear with extinction, or be connected to a recorder through a logarithmic converter. Ideally, the instrument should be so designed that the flow cell can be mounted directly above the displacement cap as with most commercial density gradient fractionation systems. The volume in the flow cell should be as small as possible to minimize mixing. When monitoring a gradient, one must observe certain precautions if the separation achieved in the rotor is not to be spoilt during measurement. Firstly, mixing in the flow lines should be minimized by keeping them as short and narrow as possible. Secondly, gradient inversions must be avoided. For example if the gradient is being fractionated by upward displacement (i.e. light end first) the flow lines should run upwards all the way to the flow cell. It is for this reason that upward displacement is preferable when monitoring, for the whole liquid stream must move upwards through the flow cell ifair bubbles are to be effectively removed. One cannot emphasize too strongly that air bubbles are the major cause of trouble in monitoring liquid effluents and every care should be taken to avoid their entering the liquid stream. If the gradient is displaced dense end Siihlri I wdv\ p
287
1 I4
DENSITY GRADIENT CENTRIFUGATION
first, there is a tendency for the dense liquid to enter the cell and remain as a puddle in the base of the cell so interfering with the measurement of later displaced portions of the gradient. This is not so important with sucrose gradients, where the viscosity of the solutions ensures efficient clearance, but can cause major degradation of separations when low viscosity salt gradients are employed. Fractions from density gradients may be collected either by hand or with a fraction collector. In the latter case, collection for constant time coupled with the use of a motor driven syringe for displacement will probably give fractions more nearly equal in volume than a drop counter. A peristaltic pump may be used, but the flow rate will fall slightly as the concentration (and viscosity) of the displaced gradient increases. In choosing a fraction collector, one must ensure that either the collector is equipped with an efficient flow arrest device or that movement from one tube to another is rapid compared with the rate of displacement. For general work we prefer to use a small ‘bench top’ collector, but when fractions are to be collected solely for the measurement of their radioactivity, a collector which collects directly into scintillation vials, is most useful. When volumes of less than 0.5 ml are to be collected, serious problems can arise due to the size of drops no longer being small compared with the volume of the fraction. Problems may arise both because the desired, fraction voldme goes not correspond to an integral number of drops and because drops are lost between tubes at the moment ofchanging fractions. An adaptor suitable for attaching to any time-operated fraction collector was designed by No11 (1969). This operates by flushing the delivery tip with a predetermined volume of water immediately before the collector is advanced to the next tube. The apparatus also includes facilities for adding scintillation fluid, if required, directly to the vials (see No11 1969 for a full description).
Ch. 4
115
CONVENTIONAL ROTORS
4.2. Isopycnic zonal centrvugation The procedures for preparing and fractionating small tube gradients after isopycnic separations are very similar to those used for fractionating particles by rate-zonal centrifugation. In this section, we will only consider cases where the procedures differ, or where precautions suggested with rate-zonal separations can be relaxed, o r should be tightened. Isopycnic zonal centrifugation is commonly applied to two types of problem; firstly the fractionation of large structures, such as whole cells or large subcellular organelles, using gradient materials which give solutions of low ionic strength and relatively high viscosity e.g. bovine serum albumin, sucrose or Ficoll; secondly, the fractionation of small particles or macromolecules on salt gradients. The procedures used in the two cases differ at several points. 4.2.1. Preparation of rhe density gradient Gradients of the viscous materials used for the fractionation of large subcellular particles and of whole cells must be formed by gradient makers such as are described in § 4.1, Much simpler techniques can be used with salt gradients. The high diffusion rate of non-viscous density gradient solutes such as caesium chloride allows one to form gradients simply by layering the solution corresponding to the light end of the desired gradient over a solution corresponding to the heavy end. Even with swing-out rotors, a gradient approaching the ‘equilibrium’ gradient will be formed within 12 hr by diffusion and sedimentation of the solute. With an angle-head rotor, a linear gradient is rapidly formed during acceleration by shearing as the two liquid layers reorient, and if this gradient is steeper than the equilibrium gradient (see next para.) then the gradient will shift rapidly towards the equilibrium gradient during centrifugation (Fig.. 4.11). Even the preparation of an initial step gradient is, in fact unnecessary with caesium chloride, sodium iodide and Ludox (colloidal silica), since such solutes redistribute during centrifugation to form a concentration gradient (4 2.2.2 and 3.3). This gradient reaches equilibrium when the movement of gradient material downward by sediS u h p I s r c h\
11
287
116
DENSITY GRADIENT CENTRIFUGATION
(a)
A
1.40
-
X
3..-W
H
t K
1.35
------
I I
I
I
1.33
I
I
L
1.5
Fig. 4.11. Flattening of CsCl gradient in the tubes of a 10 x 10 ml angle head rotor. A) Initial gradient (-----); gradient recovered after acceleration to 47,000 revs/min B) Initial gradient (-----), gradient recovered and immediate deceleration (-). after centrifugation for 7 hr at 47,000 revs/min (-).
mentation is balanced by the movement of material upward by diffusion. The slope of such equilibrium gradients depends on the
Ch. 4
117
CONVENTIONAL ROTORS
geometry of the centrifuge tube and on the centrifugal force; the greater the centrifugal force, the steeper the gradient. Much shallower gradients are formed in angle head rotors than in swing-out rotors (8 2.2.3). The time taken to reach equilibrium depends on the radial distance between the ‘top’ and the ‘bottom’ of the tube during centrifugation (van Holde and Baldwin 1958), so that equilibrium is more quickly established in angle-head than in swing-out rotors (Flamm et al. 1972). The use of ‘equilibrium’ gradients or preformed gradients depends on the size of the particles to be separated. If the particles will band rapidly compared with the time needed to establish the equilibrium gradient, as is the case with ribonucleoprotein particles on caesium chloride gradients, it is clearly better to form gradients by diffusion from at least two solutions as outlined in the first paragraph of this section. If the particles require very long centrifugation for isopycnic banding, as with sheared samples of DNA, then the easiest technique is to simply adjust the sample to a density slightly higher than the expected mean density of the material and centrifuge. Specialised articles (eg. Flamm et al. 1972) should be consulted for further details. 4.2.2. Layering of sample on to the gradient One must first estimate the densities of the particles in the mixture - usually by reference to earlier published experiments. The choice of rotor and the speed of centrifugation will depend on the density range which is required (5 2.2.2). The sample solution is then adjusted to a density very slightly higher than the density of the most dense particle in the mixture, the mixture placed in the centrifuge and spun for the required time. The high initial density is chosen to avoid pelleting of dense particles during the early stages of centrifugation before the gradient has had time to form. The separation achieved in isopycnic separation is independent of the initial distribution of the sample (Ch. 2). The volume of the sample and its initial distribution through the gradient are therefore unimportant unless, as is the case with mitochondria (de Duve et al. 1959) one or more of the particles present in the sample is SuhpI
IiidPv p
2x7
118
DENSITY GRADIENT CENTRIFUGATION
damaged by exposure to solutions of the density gradient medium more dense than its isopycnic banding density. Subject to this proviso, one can select the most convenient method for the introduction of the sample. 4.2.3. Centrifirgation Although it is not necessary to accelerate the rotor slowly when separating particles by isopycnic zonal centrifugation, it is strongly recommended that the brake should not be used in deceleration, especially with salt gradients. The low viscosity of solutions of salts such as caesium chloride makes them very susceptible to mixing and, if the rotor is decelerated too fast, swirling under the influence of Coriolis forces can seriously degrade the separation. A useful way of reducing the time required for banding an ‘equilibrium gradient’ is the ‘relaxation technique’ introduced by Anet and Strayer (1969a). The sample is centrifuged at a speed much higher than that which will produce the optimal gradient. This will result in rapid equilibration of the sample, but also in the formation of a very steep gradient so that the separated zones lie very close to one another. The centrifuge speed is then reduced to that which will produce the optimal gradient. The gradient will flatten due to diffusion and the zones of banded material will move apart. After about 6 hr, the new gradient is established and the rotor can now be slowed down and the gradient fractionated. One should note that the separation takes place during the initial high-speed stage of centrifugation. The movement apart of the zones during the ‘relaxation’ phase is due to bulk movement of the whole liquid zone and is not dependent on the size of the particles. The purpose is to obtain the maximum physical separation in order to minimize contamination during pumpoff. 4.2.4. Displacement and monitoring of the gradient Great caremust also be taken to avoid ‘gradient inversion’ in displacing the gradient, for example if the gradient is to be fractionated by upward displacement, no flow line should run downwards as there
Ch. 4
CONVENTIONAL ROTORS
119
would almost certainly be some mixing in the liquid stream over such sections. This is especially so with solutions of low viscosity, so that when displacing and monitoring such gradients especial care should be taken to keep the flow lines from the gradient to the monitor and from the monitor to the fraction collector as short as possible.
CHAPTER 5
Centrifugation in zonal rotors
The mysterious nomenclature given to zonal rotors, i.e. A-XII, BXXIX, etc. stems from the original numbering used at Oak Ridge. The letter identified the rotor as belonging to either the slow speed A class, with maximum speed usually below 6,000 revs/min or the high speed B class, with maximum speed commonly up to 50,000 revs/min, while the numbers in Roman numerals were given in sequence to each new design. In some cases there may also be a suffix Ti or A1 to signify the material. Anderson (1968a) gives a table of characteristics of A and B series rotors. Most commercially available zonal rotors still carry the original Oak Ridge nomenclature since they are direct copies, the construction and internal dimensions being the same as those of the original designs. This applies to MSE, Christ. and IEC*, the internal dimensions being close enough to the original for the Rutenberg (1966) formula (see Appendix 6) relating radius and volume in the B-XIV and B-XV to be correct. Beckman rotors differ and this formula does not apply as the internal dimensions are different. As a rough guide, rotors bearing the prefix B are probably identical to the Oak Ridge designs, while rotors with prefix Ti or A1 (e.g. Beckman Ti-14, A1-15) are not exactly the same as the originals. IEC, while retaining the original dimensions, have redesignated their B rotors B-29, B-30. Presumably they feel it is time to replace the rather cumbersome Roman figures, i.e. B-XXIX.
*
see Appendix 1 for addresses of manufacturers 120
Ch. 5
ZONAL ROTORS
121
Rotors designed and produced by various manufactures independently of Oak Ridge have a variety of different identification numbers, i.e. HS (MSE), Z-15, CF-6 (IEC), JCF-Z (Beckman). Of these four rotors only the CF-6 is a new concept, the other three being descendants of Oak Ridge designs. Materials used for rotor construction range from plastic to titanium. The low speed rotors do not produce enough centrifugal force to require very high strength materials and are usually made of aluminium and Perspex. High speed rotors were originally made of aluminium alloy but a full range of titanium rotors is now available from all the major manufacturers. These titanium rotors can be spun on the very high speed machines, various angle rotors are capable of speeds as high as 75.000 revs/min while the highest speed batch zonal rotor.can be spun at 60,000 revs/min (Beckman, 2-60). Barringer (1966) has examined the ideal shapes for optimum stability and also the properties of various materials for rotor construction. The main conclusions were that flat disc-shaped rotors and tall cylindrical rotors are stable, while rotors with diameter equal to the height are inherently unstable. This is seen in practice to-day with disc-shaped rotors like the A-XII, B-XIV, B-XV running quite smoothly supported on a single shaft and the cylindrical B-IV and K series supported with stabiliser mountings and bearings at the top and bottom, while there are no ‘square’ (diameter equal to height) rotors in use. The shape of a rotor is usually dictated by the capacity required. In view of the stresses imposed at high centrifugal speeds disc-shaped rotors can obviously be of only a certain limited diameter and are therefore of relatively low capacity (the largest in practice being 1,700 ml). The largest rotors are the cylindrical K series, 2 ft. 6 in. (76 cm) high with a maximum capacity of nearly 8 1. The aim of this section is to describe in detail the construction and operation of typical research laboratory zonal rotors. This narrows the field to batch-type and smaller continuous-flow rotors and precludes the large scale continuous-flow rotors, although some of these will be mentioned to summarise what is available. Batch rotors are the A-XII, HS, Z-15, B-IV, B-XIV, B-XV, B-XXIX (see S u h w 1 uid‘ \ p 287
122
DENSITY GRADIENT CENTRIFUGATION
Table 5.1) and the Sorval SZ-14 reorienting rotor (see $5.2). The JCF-2 (Beckman) and K-series (Electro-Nucleonics) can be used as either continuous-flow or batch rotors. Continuous-flow rotors are the B-XVI (which is a B-IV with a special core), B-XX (MSE), CF-32Ti (these two are the B-XV modified for continuous flow work), and finally CF-6 (IEC). The operation of zonal rotors is a relatively complex technique and will be covered in detail. The early rotors were difficult to use, but in our experience zonal rotors are getting easier to operate and this is certainly to be expected as the manufacturers develop more efficient equipment. An encouraging note on rotor operation appears in the IEC manual on the use of the CF-6 continuous flow rotor (Page 29.0, IM-169) ... since the rotor offers broad new approaches to large volume handling it should not be approached with caution. The individual components were designed for maximum operator use and are virtually trouble-free. The entire system should be challenged to perform to your requirement. We would certainly agree with the general theme that rotors are getting easier to use, but would still advise any newcomer that zonal rotors should be approached with some caution. This chapter on zonal rotors should in no way be taken to mean that we personally have undertaken a thorough comparative study of zonal rotors from different manufacturers, but we have made an effort to contact users of zonal rotors and are much indebted to them for their comments (see acknowledgements). We have tried to be as objective as possible but offer no guarantee that we have succeeded.
5.1. Conventional, non-reorienting zonal rotors The materials used in the construction of rotors are described in $1.7.2. 5.1.I. Construction of rotors A typical zonal rotor consists of a hollow bowl or cylinder usually made in two halves so that it can be easily opened for cleaning,
Ch. 5
ZONAL ROTORS
123
servicing etc. (Fig. 5.1). This rotor bowl is machined to very close tolerances and is accurately balanced so that it will spin on its own axis without vibrating. The centre of the rotor has a a r e and radial septa which are designed to distribute a gradient as it is pumped into the rotor either to the edge or to the centre. A feed head with a special rotating seal is used to pump the gradient into the spinning rotor, This description tits the so-called B-type rotor quite well and although the composite rotors made of aluminium and perspex look
Rotor core
Fig. 5.1, Cut-away diagram of a B-XV zonal rotor, showing details of all the functional parts. Note that the feed-head illustrated is of the early type (Compare Fig. 5.3) (By permission of MSE Ltd.). StrhCr'r huicr p 2x7
124 DENSITY GRADIENT CENTRIFUGATION
Ch. 5 ZONAL ROTORS
M
x i;:
125
126
DENSITY GRADIENT CENTRIFUGATION
muchmorecomplicated they are basically the same. The various rotors and their functional components are described in detail (55. I .2) and the characteristics of batch rotors are summarised in Table 5.1. The properties of materials used in rotor construction are given in 41.7.2. At this stage it is useful to give a short summary of the operation of a typical batch zonal rotor. With the rotor spinning at the loading speed (Fig. 5.2a) a gradient is pumped into the rotor; usually the light end is pumped through a fluid line leading to the rotor wall. With the gradient increasing in density (Fig. 5.2b) the lighter layers are displaced towards the centre. Finally a dense solutior, (to prevent pelleting on rotor wall), called cushion or underlay, is pumped in to fill the rotor. When the rotor is full, a gradient, increasing with radius, has been established. A sample, followed by an overlay (a liquid less dense than the sample) is then loaded into the centre (Fig. 5.2c), the rotor is accelerated and spun for the time required to achieve the separation (d & e). It is then decelerated to unloading speed (same as loading speed) and (f) a dense solution (same as cushion) is pumped to the wall of the rotor displacing the separated bands through the centre tube. During unloading and collection of fractions (g) the gradient is passed through a gradient profile monitor (refractometer) and a spectrophotometer, thus obtaining a record of the gradient shape and the location of components. Construction of the feed head and rotating seal. The feed head locates the rotating seal onto the rotor (Fig. 5.3). Details of design vary considerably throughout the range of equipment available. However, the most important part is the seal and fortunately this is, in principle, similar on all the modern zonal rotors. Rotating seal. A metal piece and a plastic piece rotate against one another while pressed together by spring tension. The metal part is normally stainless steel but in the Beckman system it is oxide coated, presumably for better wear. The plastic piece is usually Rulon, but in the Beckman JCF-Z rotor it is graphite. Most rotors use a ‘coaxial seal’ (Fig. 5.4) which allows separate connections to the centre and the edge of the rotor to pass through the same seal and, ideally, not to contaminate each other. The metal and plastic parts
Ch. 5
Z O N A L ROTORS
127
Fig. 5.3. The MSE B-XIV titanium rotor, feed head. guard tray +vacuum cap. Note the new type feed head (Compare Figs. 5.1 and 5.5) (By permission of M.S.E. Ltd.).
have matching holes in the centre and a number of holes (usually six) arranged in a circle round the centre. Normally the centre hole leads to the centre of the rotor, the concentric series being connected via their annular groove to the wall. The most usual arrangement is to
Fig. 5.4. Coaxial rotating seal. Top part is the stationary metal piece which is kept firmly pressed against the rotating Rulon seal. (By permission of Beckman.) Suhiccr itidex p . 287
128
DENSITY GRADIENT CENTRIFUGATION
have the Rulon part as the rotating seal while the metal part is stationary. The most serious problem encountered with the coaxial seal is cross-leakage (see definition in p.1.3.3). The K series rotors use a simple rotating seal with matching central holes in the stationary and the rotating elements; separate feeds to the edge and centre of the rotor are achieved by placing two ofthese seals at opposite ends of the rotor, thus eliminating crossleakage and cross-contamination. In all cases, the two elements of the seal are kept together by spring tension which can be adjusted. The rotating faces of the seal do generate some frictional heat and therefore cooling is required. Thefeed headassembly is composed of the following parts (Fig. 5.5) : the stationary seal element which also has four tube connections (manifold assembly; one to the wall, one to centre of the rotor and two for coolant), the rotating seal (Rulon), is sprung to keep the seal elements pressed together, a bearing (so that the Rulon part spins with the rotor) and a housing assembly. The compact design of the removable feed heads, allows for easy operation and maintenance. Apart from the rotating seal the only other moving part is a bearing which spins with the rotor and locates the feed head on the spinning rotor. Once on, the feed head is locked into place on the guard tray (Fig. 5.3) or locating ring. The feed head then effectively becomes a part of the rotor and is not mounted rigidly on the guard tray but is allowed to move and ‘vibrate’ with the rotor. The bearing is usually prepacked with grease, some systems using a completely sealed bearing, while others use an open type which can be cleaned out and regreased. The early B high-speed rotors, with removable feed heads, all used the basic design which originates from the Oak Ridge laboratory. This system used the Rulon part of the seal in the rotor core and the stationary part in the feed head, so that the two sealing surfaces had to be brought together or separated each time the feed head was put on or removed. This arrangement proved rather unpractical since the Rulon parts could not be removed for repair while the rotor was spinning. Beckman redesigned the feed head so that both
Ch. 5
129
ZONAL ROTORS
8
SCREW
STATIONARY SEAL WITH PRESS-FIT SEAL GUIDE ROTATING RULON SEAL I
BEARING ASSEMBLY
-BEARING HOUSING
Fig. 5.5. Thecomponent partsof the feed head assembly. (By permission of Beckman.)
stationary and Rulon parts were within the feed head which then locked directly into the rotor. After seeing the advantages of this system other manufacturers soon followed suit. Now we have MSE, Sub,',' I rrr&\
p . 287
130
DENSITY GRADIENT CENTRIFUGA'I'ION
IEC and Beckman all using basically the same layout. Christ appear to have redesigned their system in a slightly different manner, with the Rulon part stationary in the feed head while the stainless steel part becomes the rotating seal. Guard tray (rotor shield). This is normally a composite perspex and aluminium tray which screws or clips into the centrifuge bowl just above the rotor (Fig. 5.3). It usually has provision for adjusting the height and lateral location. The Beckman system has a support band ('Belly band') into which the rotor shield (guard tray) is clipped, by means of a spring loaded bar. The feed head (seal assembly) is then carefully located on the spinning rotor and locked into the rotor shield. Lateral movement of the feed head is possible since it does not lock rigidly into the shield but is allowed to follow the rotor. The MSE system does not allow so much movement and therefore centering of the guard tray is much more important, but it does have the advantage that it seals the centrifuge bowl better so that less condensation collects in the bowl. Rotor cores and septa. Referring to the original designation of the B rotors one sees that the centre core and the septa were two separate parts (see Fig. 5.1). This design requires the use of O-rings on the rotor core to provide a good seal to the feed lines in the septa. This arrangement is still used by MSE and Christ, while Beckman and IEC core and septa are all in one piece. The latter system is obviously simpler and should be less prone to leakage but it may be more difficult to clean. Seals and O-rings. O-rings of synthetic butyl rubber are normally used with aqueous and non-corrosive gradients. Special O-rings of Viton can be supplied for use with organic solvent gradients. 5.1.2. Types of zonal rotors 5.1.2.1 A-XII rotor An earlier version of this rotor was made by IEC while it was still under development and was then called the A-IX. Some of these were available commercially and the use of an A-IX is quoted by Boone et al. (1968). The A-XI1 is presently available from MSE. It is a large rotor weighing about 70 Ib
Ch. 5
ZONAL ROTORS
131
(31 kg), e$ternal diameter 18 in (46 cm), composed of Perspex plates held together with a large aluminium clamping ring (Fig. 5.6). This ring has to be tightened accurately to a specified torque and for this purpose a torque wrench is provided. This rotor is so heavy that a special lifting extractor has to be used, unless the operator is exceptionally strong. Its diameter is large relative to the height resulting in a stable rotor requiring no upper bearing. The usual loading and unloading speed is 400-500 revs/min, while the maximum operating speed is 5,000 revs/min. The figure shows clearly the perspex construction, including the four septa, which in all the zonal rotors prevent swirling and form the feed lines to the wall of the rotor. The transparent nature of the rotor makes it ideal for observing a separation during the run, particularly if a fluorescent light (available for the
5.6. The A-XI1 zonal rotor (M.S.E. Ltd.). Strhlrrt irdrx p . 287
-
TABLE 5.1 Characteristics of batch zonal rotors. Rotor
Material
Capacity Max. speed ml revs/min.l
Max. RCF at edge6
Max. path cm (core taper to
Approximate g of sample at max. speed
5,000 12,000 9,100 90,900 91,000 91,300 62,000 48,100 165,000 151,000 186,500 171,800 101,800 89,500 121,800 102.000 48.100 101.800
13.0 7.7 8.0 3.3
1,400 4,100
Manufacturer's Oak Ridge designation designation A-XI1
HS 2-15
B-IV B-XIV AI-14 B-XV AI-15 B-XIV B-XIV 8-14 (B-30) Ti-14 B-XV B-XV B-15 (B-29) Ti-15 8-29 core B-XXI$ B-XXIX B-29 B-29 cores4 8-29 core,' 8-30 B-29 core5 2-60
A-XI1 -
B-IV B-XIV
(Al. and Perspex) (Al. and Perspex) (Al. and Perspex) A1.
Al.
B-XV
Al.
B-XIV
Ti.
B-XV
Ti.
-
Al Ti Ti. Ti.
-
B-XXIX B-XXIX
1,300 695 780 1,720 640 665 1.660 1,675 640 650 659 665 1,660 1.650 1,674 1,675 -
1.420
1,480
B-XXX B-XXX
Al . Ti.
-
Ti.
5,000 10,000 8,000 40,000 35.000 35,000 25.000 22,000 47,000 45,000 50,000 48,000 32,000 30,000 35,000 32.000 22.000 32,000
-
570
-
330
h) W
4.3. 5.3 6.5 7.5 4.3
121.800 102.000 91.300 186,500 171.800 256,000
34,000 84.000
5.1 5.1
5.3 6.5 7.4 7.4 7.5
66.000
-
~
5.2
6.0 35,000 32.000 35,000 50.000 48.000 60,000
46.000 47.000
66.000 66,000
4.0
47,000 84,000
4.8
-
~
Manufacturer2 Approx. wt : kg.
MSE MSE IEC Beckman MSE Beckman MSE Beckman MSE Christ IEC Beckman MSE Christ IEC Beckman Beckman MSE Christ IEC Beckman Beckman IEC Beckman Beckman
31 13 -
6 10
10
18
10
18 18
6 10 -
Obviously this maximum speed applies only if the rotor is used on the fastest machines, i s . MSE’s Aluminium B-XIV can be spun at 35,000 revs./min. on the SS65 and SS75 machines and only at 25,000 revs./min. on the HS 25. 2 See Appendix for full names and addresses. The manufacturer should be consulted on the centrifuge to be used for each rotor. Aluminium and titanium B-XIV and‘B-XV rotors are also available from Janetzki. Leipzig. G.D.R. 3 The MSE figure is the path length from a position clear of the tapering core (representing the true sample position) and in the case of the B-XXIX to the beginning of the edge taper. Other manufacturers quote either the maximum path or a wrongly measured minimum path. hence the discrepancies. 4 Beckman call the edge loading, unloading core ‘B-29 type Core’ for the A l , or Ti, 14 or 15 and do not differentiate between the B-XXIX and B-XXX (see also (j 5.1.2.6). 5 M S E s edge unload rotor is now offered only as a conversion with tapered inserts for the titanium B-XV. 6 The maximum usable RCF, as opposed to the quoted value, is slightly lower because of the rounded corners at the edge of the rotor. With the B-XXIX rotors this highest usable RCF is even lower because of the edge taper. I
9 wt
N
0
e W
134
DENSITY GRADIENT CENTRIFUGATION
M.S.E. Mistral 6L) is fitted under the rotor. It must be cooled before assembly and dismantling. The rotor has the radius in cm engraved on one of the plates and this may be used to relate radius to volume during centrifugation. The non-removable feed head remains on the rotor during the spin and is secured by an antirotation arm. The normal feed head of the A-XI1 (MSE) has the centre line of the seal leading to the edge of the rotor, while the outer feed line, passing through the outer holes of the coaxial seal, leads to the centre. None of the other zonal feed heads have this arrangement. (see also 45.1.3.3). 5.1.2.2. Z-15 rotor This rotor was developed by IEC as the logical scaled-down version of the A-XI1 rotor. Its construction (aluminium and Perspex) is very similar to the A-XI1 and HS (MSE). The Z-15 can be used in the IEC B-20 or the B-60 high speed centrifuges, has a capacity of 780 ml, maximum speed of 8,000 revslmin and a maximum RCF of 9,100 g at the edge of the rotor. The feed head is of a type common to all IEC zonal rotors (except the AXII), the delicate parts (Rulon and stationary seal) can be removed while the zonal rotor is spinning. The application of this rotor was first reported by Szabo and Avers (1968). 5.1.2.3. H S rotor This rotor (Fig. 5.7) is MSE’s equivalent to the Z-15 rotor. It is like the A-XI1 and Z-15 is constructed of aluminium and Perpex and has a capacity of 695 ml. This allows it to be used at a higher maximum speed of 10,000 revs/min which gives a maximum RCF at the edge of 12,000 g . It can be used in the MSE HS 18 and Prep Spin 21 centrifuge. In our experience, the most suitable speed for loading and unloading is approximately 1,500 revs/min. This speed is difficult to control unless the machine is fitted with a slow speed zonal control accessory which is available from MSE. The only constructional difference between the A-XI1 and the HS is in the feed head, which on the HS rotor is removable and is placed in position only during loading and unloading. During a run a removable cap is used to seal the rotor. A guard tray is locked
Ch. 5
ZONAL ROTORS
135
5.7. Cut-away diagram of the HS zonal rotor (M.S.E. Ltd.) 'A' Perspex plates, 'B' Perspex centre piece with septa and edge feed channels, 'C' Aluminium clamping rings, 'D' Centre core, 'E' Feed head, 'F' Guard tray.
in place in the centrifuge bowl just above the rotor and this locates the feed head on the rotor. This rotor, like the A-XII, must be cold for assembly or dismantling. The HS rotor is rather a tight fit in the HS 18 centrifuge and there is not enough space to fit a light under the rotor. Blyth et al. (1973) and Burge and Hinton (1971) have demonstrated the usefulness of this rotor. 5.1.2.4. B-IV rotor This rotor was the first of the high speed batch
rotors, which could be loaded and unloaded dynamically. Although complex in design and operation and now considered obsolete, several are still in regular use. The B-IV is not now generally available as the B-XIV and XV series have succeeded it, being cheaper and easier to use and requiring no special modification to standard centrifuges. However, we believe, Beckman will still make one to order if a customer is sufficiently enthusiastic and if he possesses a ZU or Siihwi ifides p. 287
136
DENSITY GRADIENT CENTRIFUGATION
LA centrifuge. The feed head is not removable and incorporates a complex vacuum tight seal. Generous cooling and lubrication is needed for the bearing and the seal which stays in position during the whole run (which means speeds of 40,000 revslmin.). The rotor is made of aluminium, its capacity is 1720 ml, and the maximum speed is 40,000 revs/min (maximum RCF 91,000 g). B-IV rotors made by Beckman were used in the special ZU centrifuge, and later in the L4 machine. The fact that the sample position is at 46,000 g when the rotor reaches 40,000 revs/min. combined with a relatively short centrifugal path of 2.9 cm makes the B-IV particularly useful for isopycnic runs. 5.1.2.5. B-XIV and B-XV rotors
These are probably the most common zonal rotors in use to-day. Their development and properties were documented by Anderson et al. (1967). The two types of rotors are similar in shape and differ only in size, B-XIV (Figs. 5.3 and 1.1) having a capacity of about 640 ml, B-XV about 1670 ml. Each is available in aluminium or titanium. The aluminium rotors should only be used with non-corrosive gradient materials, such as sucrose and Ficoll buffered within the physiological ranges. The titanium rotors can be used with nearly all gradient materials within the pH range of 4-10. B-XIV and B-XV aluminium and titanium rotors are available from Beckman, IEC, MSE and Christ. Recently Janetzki, Leipzig, have introduced aluminium and titanium versions of both rotors; we only mention their existence, as at the time of writing few details are available. The actual construction of B rotors is very simple, consisting of the bowl, lid, septa and core (Fig. 5.1). The IEC and Beckman rotors have the septa and core all in one piece while on the MSE and Christ rotors they are separate. The bowl and lid have machined threads and are simply screwed together; the correct torque is usually indicated on the lid and the bowl in the form of two marks which have to be lined up. Rotors from different manufacturers vary somewhat in detail, such as internal dimensions and maximum speeds, e.g. Christ 45,000 revs/min (151,00Og), IEC 50,000 (185,000 g). IEC and Christ manuals contain
Ch. 5
ZONAL ROTORS
137
graphsrelating radius to volume. Septaare madeeither ofa hard plastic, Noryl, or of the same metal as the rotor. When Noryl septa are used the rotor must not be spun partially full at any speed higher than the loading and unloading speed. In earlier models, septa feed lines were horizontal, now they lead diagonally to the lower corner, thus allowing reorientation of the gradient (see $5.2). The latest feed heads on all these rotors are similar (see $5.1.1.2). MSE, IEC, Christ and Beckman use the B series nomenclature with slight variations, i.e. B-15 (B-29) or B-XV, Ti-I5 and A1-15. The first Beckman rotors were Oak Ridge designs and were designated B-14, B-15, A1 or Ti. However, the present series are called A1-14, A1-15 and Ti-14, Ti-I5 and were designed by K.E. Jacobson of Beckman. Their construction and dimensions differ from the original Oak Ridge design. B rotors were the first developed specifically for use in standard ultracentrifuges. The feed head is therefore designed to stay in position only during loading and unloading and is removed during centrifugation, after which the rotor is capped with a vacuum tight cap and the vacuum lid is closed. The guard tray, which locates the feed head, stays in position during the whole run. The IEC rotors B-14, B-15 and Z-15 are all the same height and use the same feed head. This means that one guard tray and one feed head can be used with any of the rotors without adjustment. The IEC feed head arrangement is more versatile than any of the other systems. Part of the rotor core is integral with the feed head and as such is also removable. This means that different cores with differently arranged channels may be used during the same run. The B-XXIXa rotor has a second series of channels in the septa emerging about a centimeter from the wall of the rotor (Fig. 5.8). This system, called MACS (Multiple Alternate Channel Selection) by IEC, and described by Price and Casciato (1974), allows unloading if a pellet has blocked the normal wall line. It also means that a short gradient for isopycnic work formed between the two edge feed lines can remain undisturbed while a number of samples or even complete 'rate' gradients have been loaded on top of it (see Contractor 1973). 5L,h,'Yl IIldPrp 2x7
138
DENSITY GRADIENT CENTRIFUGATION
a
b
C
d
e
Fig. 5.8. The IEC ‘MACS’ system. Upper Figs. shows the B-29 (in effect a B-XV) rotor with only the removable plug illustrated instead of the complete feed head. The feed lines are conventionally arranged, centre and edge tubes on the feed head leading to the centre and edge of the rotor respectively. A, B, C and D represent four access points along the septa (radius) of the zonal rotor. ‘A’ is the centre feed line while ‘D’ is the edge line. ‘B’ and ’C’ are two different points positioned between ‘A’ and ‘D’. The lower Figure shows a complete set of ‘plugs’. a) The plug represents the reverse of normal, centre of feed head leads to edge of rotor and vice-versa. b) A different plug with access between points ‘B’ and ‘D’ along the radius. c) This is the reverse of b) giving access between ‘B’ and ‘A’.d) Access now is between ‘C’ and ‘D’. e) The reverse of c) with access between ‘C’ and ‘A’.
Ch. 5
139
ZONAL ROTORS
To do this practically requires changing plugs on the feed head. There is no reason why such a system could not be developed even further to give a number of access points along the septa, with the corresponding set of plugs. 5.1.2.6 B-XXIXund B-XXXrotors
These rotors are further developments of the B-XIV and XV, allowing centre or edge loading or unloading, and are described by Anderson et al. (1969b) (for pros and cons of the edge or centre loading-unloading principle see $5.1.3.3 and $5.1.3.5). The B-XIV and B-XV have the edge wall vertical, the B-XXIX and B-XXX rotors have this wall sloping. B-XXIX is the edge load-unloading version of the B-XV while the smaller B-XXX corresponds to B-XIV. Because of the sloped outer wall of these rotors it must be remembered that the effective centrifugal path is considerably reduced as is the maximum g. The rotors, available from the four major manufacturers (see Table 5.1) are worth considering in detail as there are differences in construction. Beckman use a liner in the standard B-XIV or B-XV bowl to convert the vertical edge to a slope, together with a modified core (Fig. 5.9); this is an economical and practical approach. IEC also use an insert to convert a normal zonal to the edge load-unload rotor. MSE and Christ produce their B-XXIX as a distinct rotor, not interchangeable with B-XV, the slopes at the edge being machined in the main body of the rotor.
BBD b
Fig. 5.9. Three different edge load/unload B rotors. a) Construction of the original Oak Ridge B-XXIX rotor. MSE and Christ use this same design. b) IEC use two tapered aluminium inserts to convert a standard B rotor into the B-29. c) In the case of Beckman’s 8-29 a single tapered insert is used. Sl,hic
140
DENSITY GRADIENT CENTRIFUGATION
5.1.2.7. Interchangeable batch and continuous-flow rotors This is a group of rotors which may be used for either continuous flow or batch work. Since zonal rotors are a considerable investment these rotors are obviously attractive, combining pure research facility and commercial scale production capability. Beckman JFC-Z rotor. This new rotor is designed for use in the Beckman model 5-21 centrifuge; externally it bears a strong resemblance to the B-XV. The total volume for batch work is 1900 ml, while the maximum RCF of 40,000 is generated at 20,000 revs/min. It is made of titanium with interchangeable Noryl cores to adapt it for normal zonal use reorientation or continuous flow work. Further description of this rotor can be found in Chevrenka and Elrod (1972). K Series. These rotors were developed by the Oak Ridge team as high capacity, continuous flow machines for harvesting virus on a commercial level. Their development and application is described by Anderson et al. (1969a) and Cline (1971). However, since these rotors are not the concern of this monograph we will only mention the three main types for the sake of completeness. The series comprises three designs, K, RK and J rotors, which are now commercially made by Electro-Nucleonics Ltd. We refer to the whole group as the 'K' series because they are all similar in construction and operation. K and RK stand for two distinct centrifuges, respectively designed for the tall K series rotors and the RK rotors which are approximately half the height of the K. The J-1 rotor with its maximum speed of 60,000 revs/min is used on the RK machine. The principle of operating these rotors is shown in Fig. 5.23. K rotors. These are very tall cylindrical rotors 30 in high (76 cm) and 6 in (15 cm) diamter, consisting of an aluminium or titanium cylinder and a variety of different cores (designated by numbers) designed for different functions. The K rotor set up for large-scale production can process up to 120 1 of culture per hr. Maximum speed is 35,000 revs/min. RK rotors. Again a standard cylinder in titanium, 14 in (36 cm) high and 6 in (15 cm) in diameter, can be used with any of the different RK numbered cores. The maximum speed of the RK is
Ch. 5
ZONAL ROTORS
141
also 35,000 revs/min and it can process up to 60 1 of culture per hr. J-I roror. This is a cylinder made in titanium, 16 in (41 cm) high but only 4.5 (11 cm) in diameter. It is the smallest, but fastest (60,000 revs/min.) of the series, giving a maximum RCF of 160,000g , and for batch work it can be applied to the same range of separations as a titanium B-XIV. Rotors of the K series have the connections to the centre and edge at opposite ends of the rotor, thus eliminating any danger of cross leakage. For batch work these rotors are just slightly more complex to use than the more common B-XIV and B-XV. Being considerably heavier, the K series rotors are normally handled with the aid of special lifting hoists. Once in the centrifuge, they are loaded and unloaded much as any other zonal rotor. The major difference is that these two operations are done with the rotor stationary. For the run, the rotor is accelerated and decelerated very slowly, allowing reorientation of the gradient (Fig. 5.23). It should be emphasised that the K series represent the most expensive zonal systems. Their purchase can only be justified by the need for the large scale continuous flow facility. 5.1.2.8. ContinuousTflow rotors The true continuous-flow rotors are
designed for harvesting and purifying cultures using conventional centrifuges as opposed to the purpose-built K installations. Their use is mainly for pilot or even full-scale production of vaccines and since they are rarely used for pure research they are merely listed briefly. B-XVZ rotor. This was the first of the continuous-flow rotors to be commercially available and is virtually identical to the B-IV except for the design of the core. Like the B-IV it is now considered obsolete. B-XX rotor.This rotor (MSE) and the CF-32 Ti (Beckman) rotor are virtually identical in specification. They are both basically B-XV rotors, which have been modified for continuous-flow operation by using a new core and a new feed head system. CF-6 rotor (IEC). This continuous-flow rotor is of a unique Strhlc
142
DENSITY GRADIENT CENTRIFUGATION
design and bears no resemblance to any other continuous-flow rotor. Its maximum speed is 6,000 revslmin. It is claimed to be capable of separating particles with S values as low as 104. It is therefore more suitable for harvesting whole cells (yeasts, bacteria, spermatoza etc.) and large cell organelles than for harvesting viruses, which is the usual application of the other continuous flow rotors. None of the true continuous-flow rotors can quite match the capacity or performance of the K series rotors which can also be used as batch rotors. 5.1.2.9. Other rotors Rotors listed here are not necessarily available, but are worth mentioning to complete the picture. Most of them were developed at Oak Ridge and represent some fairly advanced projects, and may in the future appear as commercially available hardware. IEC high speed ‘B-14’ rotor. This company produced a titanium zonal rotor similar to a B-XIV in dimensions and volume. The construction was of an advanced design allowing the rotor to be spun at 75,000 revs/min. (nearly 500,000 g ) . Although a number were made and are still in use, the demand was so low that it was never manufactured. However, IEC (at the time of writing) are willing to produce the rotor if there are enough customers. Anyone sufficiently interested in these rotors should contact the Applications Director, IEC . The 2-60,This rotor was recently introduced by Beckman as the fastest zonal rotor commercially available (see Table 5.1). It is operated exactly like a normal B-XIV rotor. At the time of writing we are not aware of any published applications of this rotor. C-series. These were very high speed preparative batch rotors capable of speeds as high as 150,000 revs/min. Some were actually produced at Oak Ridge but no details were released. D-series. Anderson (1968a) describes these as ‘rotors for exploring the limits of available high centrifugal fields’. They were reputedly designed for ultra-high speeds of up to 400,000 revs/min. F-series. Rotors for rapid centrifugal freezing were first reported
Ch. 5
143
Z O N A L ROTORS
by Anderson et al. (1966b). These were used for rapid freezing of cells and homogenates in which small droplets are centrifuged at high speed through liquid nitrogen. The Beaufay rotor. Although never produced commercially this interesting rotor is still in use in Dr. Beaufay's laboratory. It was developed exclusively for isopycnic work and therefore has a very short sedimentation path of approximately I cm. The maximum speed is about 39,000 revs/min, the chamber is annular (rather like that of a conventional zonal), has a volume of about 20 ml and is spun on a modified Beckman Spinco centrifuge. It has a single feed channel, used both for feeding and unloading. An unusual feature of the rotor is the method of unloading which is done by applying nitrogen at a pressure of 15 atm. to the centre. When the rotor has decelerated to 9,000 revs/min the gas pressure overcomes the hydrostatic pressure of the liquid in the rotor, and unloading begins. The rate of unloading is controlled entirely by the rate of deceleration and the gradient is completely displaced by the time the speed has fallen to 6,000 revs/min. The prototype rotor was first described by Leighton et al. (1968) but a number of rotors have since been made for use in Louvain and other laboratories (Seymour et al. 1974). The work of Wattiaux (1974) on the effect of hydrostatic pressure on subcellular organelles has drawn attention to the fact that this rotor deyelops the lowest hydrostatic pressure of any centrifuge rotor in use to-day. This observation is probably responsible for the revived interest in the Beaufay rotor in spite of its complex and temperamental feed head arrangement. Beaufayet al. (1974) now refer to the rotor as the E40 and also describe a new and larger version with a larger sedimentation path (3.4 cm) designed for rate sedimentation work and designated S25. 5.1.3. Operation of zonal rotors
We shall now give a set of detailed instructions and advise on operation of zonal batch rotors. The section is written primarily for the beginner, but also contains details extracted from manufacturers' manuals and gained by hard experience, which might prove useful Sllhi'll Ilrdt 1 p
2&7
144
DENSITY GRADIENT CENTRIFUGATION
to every zonal user. Although the principle of zonal centrifugation has already been extensively covered in review articles, Anderson (1966b and 1968), Birnie (1967 and 1969), Anderson (1966a) and Price (1 972), the only articles (apart from the manufacturers manuals) to have concentrated on the technique and methods of zonal operation are those by Cline and Rye1 (1971) and the Beckman publication by Chevrenka and Elrod (1972). The first attempt at a zonal run is nearly always a failure, but this should be viewed philosophically as an initial familiarisation with the equipment and technique. It may be useful for some beginners to write a concise set of instructions in their own style, around their own equipment, and perhaps using the tables in this section as a framework. It should be made clear to a potential newcomer to the zonal field, that, apart from the centrifuge and zonal rotor only a modest range of equipment is required, i.e. pump, homemade gradient maker (see Fig. 5.21 and s . l . 1 ) and refractometer. Although we describe a more lavish layout of equipment, this is done only for the sake of completeness and to demonstrate that such equipment is most useful to a laboratory whose work entails the frequent use of zonal rotors.
5.1.3.1. Preparations for a run This section describes all the preparation up to the point of starting to load the gradient. Table 5.2 gives a sequence of instructions, which are described below in more detail. a ) Planning the run and checking the centrifuge. In view of their large capacities, zonal rotors can take considerable time to fill with gradient. This filling time (Table 5.3) is related to rotor volume and to the flow rate which can be obtained. The zonal rotor and other equipment should be carefully watched during loading and therefore any sample preparation must be done before or after gradient loading. The ideal way is to synchronise the two, and have the sample and gradient ready at the same time, but in practice this needs two pairs of hands.
Ch. 5
145
ZONAL ROTORS
TABLE5.2 Preparations for an experiment using a Zonal rotor. -~
-
Plan the experiment. Ensure that the rotor is clean and precooled. Prepare gradient solutions. Prepare gradient maker. Check that the rotor is tightened to the correct torque. Check ancillary equipment. Connect up all the feed lines. Connect up the coolant for the seal. Check that the feed lines have been correctly connected. Check feed-head bearing. Carefully examine both elements of the rotating seal. Wash out feed lines with distilled water, empty. Place rotor in centrifuge. Locate the guard tray and the anti-condensation shield. Switch on centrifuge. Ensure that refrigeration is also switched on. Check that the correct loading speed has been reached. Locate feed head on the spinning rotor (only if removable type). Check that all the anti-condensation seals are in place. Ready to start loading gradient.
Since every zonal run can be extremely valuable in terms of effort and information obtained, every precaution should be taken to ensure that the centrifuge performs faultlessly during the whole zonal run. It is neither practical nor necessary to have a service before important runs, but in some cases a short dummy run beforehand could save TABLE5.3 Approximate times required to fill rotors with gradients, using flow rates of 4&50 ml/min A-XI1 - 45 min
HS,Z.I5 - 35 min B-XIV - 30 min B-XV - 1 hour These times are on the generous side and give some leaway for possible accidents, but do not allow for other preparation steps. Si,hico iiak.rp 287
146
DENSITY GRADIENT CENTRIFUGATION
effort. Unfortunately it is not possible for us to list all the centrifuge checks, as there are too many different machines on the market. A brief look at the fault finding chart (in the manufacturers manual) can give an indication of what to look for during this dummy run. After being placed in the centrifuge a zonal rotor is accelerated to its loading or unloading speed. These speeds should be about : A-XI1 - 400-600 revs/min; HS or Z-15 - 1500-2500 revs/min; B-XIV, B-XV - 2,000-5,000 revs/min. If problems arise these speeds can easily be adjusted. Firstly, ensure that the feed head is correctly adjusted and in perfect working condition. While loading a dummy gradient, adjust the speed until the feed head is seen and heard to run smoothly and. quietly without any sign of vibration. It is advisable at the same time to check the backpressure in the edge line since some rotors can generate some hydrostatic back pressure. Therefore choose a compromise load/ unload speed at which the pressure is reasonably low and the feed head is behaving well. In practice this speed is set during the initial installation and will not normally need r,esetting. If the particular centrifuge is heavily used, ensure ahead that it is available and book it for the correct length of time. 6) Check ancillary equipment. Expensive gradient makers usually incorporate a pump, but if a home-made gradient maker is used, a pump is required to load the gradient (and later unload it). The requirements for a suitable pump are listed in v.1.1 and §5.1.4.2. Here it should suffice to remind the users that this pump should be checked. This applies particularly to the tubing of peristaltic pumps which should be changed if signs of wear appear, and to the valves of piston pumps which may become blocked with crystallised gradient material if the pump had not been properly washed. To follow the profile of the gradient, it is useful to use a flowthrough monitor (See $5.1.3.6). If such an instrument is available it should be connected in the loading line. The instrument and the recorder should be suitably zeroed, and the chart and pen be made ready for the start.
Ch. 5
ZONAL ROTORS
147
During both loading and unloading it is often necessary to stop the flow and clamp the tubing(to prevent syphoning). This is best done with artery forceps which are very quick to use and do not damage PVC tubing. Alternatively, any other tubing clamps may be used. To store information from each zonal run it is useful to use printed zonal forms, which should contain as much information on the run as possible, i.e. : sample preparation, gradient, sample volume, acceleration, deceleration, duration of spin, details of unloading, assays etc. An example (for perfectionists) of such a sheet is given by Cline and Rye1 (1971). c) Rotor. The rotor should be clean, dry and at the correct temperature. Washing and drying have already been discussed in $1.7.3. The simplest way of cooling rotors is to leave them in a cold room overnight. The smaller rotors like the B series can be left in a refrigerator or, for rapid cooling, they can be placed in a bucket of ice for about half an hour before the run. During cooling, the vacuum cap should be used to seal the rotor and prevent condensation collecting on the inside. The rotor (B types which are used in vacuum chamber centrifuges) should be placed in the centrifuge only when all the preparation is complete and all that remains to be done is to put on the feed head and switch on the gradient pump. This is done to keep the period of 'lid open' to a minimum and thus avoid excess condensation collecting in the bowl. (This is further explained in $5.1.3.4.) The centrifuge drive shaft should be smeared with light oil (not silicone grease) before the rotor is placed on it in order to facilitate easy removal at the end of the run. d ) Fluid line connections. All tubing connections should be as short as possible. The tubing should be of a fairly narrow bore (2-3 mm), to minimise mixing during pumping, but this does mean that cold, viscous solution will cause some back pressure, and connections must be tightly secured. Standard plastic connectors which are readily available, can be used to join short lengths of tubing. The tubing (also discussed in $5.1.4.11) should remain flexible even at 5 "C and should be transparent enough to show up air bubbles or dirt. Cheap Sl,hlcc/ ozrlei p 287
148
DENSITY G R A D I E N T C E N T R I F U G A T I O N
PVC tubing is adequate for routine work, except for connections to the feed head where soft silicone rubber is preferable. Silicone rubber tubing, although expensive, and easily damaged with artery forceps, has the advantage of being suitable for peristaltic p m p s . When the connections from the gradient maker to the feed head have been established, as shown in Fig. 5.10, this whole line should be washed well with distilled water and emptied. The line from the gradient maker must be connected to the tube on the feed head which leads to the rotor wall. The central tube on the feed head, which leads to the centre of the rotor, should have a short piece of tubing, dipping into a beaker of water. Finally, water should flow through a coolant tube (Fig. 5.12) connected to the seal. This tube is wound round the stationary seal element and exits either to drain or, if a recirculating system is used, back to a tank. Either of the tubes may be used for in-flow or out-flow. D
Fig. 5.10. Schematicrepresentationof Gradient Loading. Thegradient makerA pumps the gradient through the recording refractometer B, then through the cooling coil C , past a bubble trap D, a three-way tap E, thermocouple measuring cell F, and finally enters the rotor (G) at the edge. Air displaced by the gradient appears from the centre line.
Ch. 5
ZONAL ROTORS
149
e) Gradient maker and pump. In this section we presume that the operator knows how to make a gradient (see g.1.1 and (34.2.1). It must be remembered that a gradient loaded from the edge of the rotor (to the rotor wall), must start with the lightest solution and then progressively increase in density during the loading. The line will have been washed out with distilled water prior to the loading of the gradient. No water should be left in the tube so that when loading begins it is pushed into the rotor, since this will destroy the leading edge of the gradient, also the sample will tumble right through this water layer, and will spread considerably. The gradient line should at the start be either completely empty or full of the gradient itself. f) Coolant. Seal coolant can be tap water, which is sufficiently cold for gradients of 15 "C or above. If stringent temperature control, at say 5 "C is needed, then ice-cold water must be used, circulating in any convenient reservoir such as a plastic water bath, box or bucket containing an ice/salt mixture. The most suitable pump for circulating the coolant seems to be a simple centrifugal pump (see 85.1.4). Ice-cold water can also be obtained by using a refrigerated water cooler, but these are usually bulky items which would occupy precious space near the centrifuge. The Beckman L4 centrifuge had a built-in ice-cold water supply. To ensure that the gradient enters a zonal rotor at the correct temperature, it must be precooled. The temperature can be then checked with a thermocouple, embedded in a special connector (Fig. 5.11) which is placed directly on the edge tube on the feed head as near to the rotor as possible. The gradient is pumped through a stainless steel coil, dipped in an ice bath, and the thermocouple temperature is continually followed. The temperature is then controlled by either stirring the ice round the coil, or by adding more salt to the ice, or by reducing the flow rate. If salt is used, this should be added after the gradient loading has begun, otherwise the liquid in the coil may freeze and block the system. In Fig. 5.12 we show a layout, used in our laboratory for a common coolant for seal and gradient. To save space and keep the tubes as short as possible, Suhpv~wdc\ p 2U7
150
DENSITY GRADIENT CENTRIFUGATION
Fig. 5.1 I . The thermocouple 'cell'. This is made entirely of PVC tubing. Pieces of tubing of appropriate dimensions are welded together with cyclohexanone (dotted lines) and the thermocouple is inserted at one end.
Fig. 5.12. The essential connections to the feed head A and the common coolant bath for cooling the seal and the gradient. The centrifugal pump B pumps iced water to the seal and hack into the plastic box. For the sake of clarity the thermocouple cell and the three-way tap have been omitted from the line leading from the cooling coil C to the feed head.
Ch. 5
151
ZONAL ROTORS
the coolant container is placed on the centrifuge, as near to the rotor bowl as possible. It is essential to secure the coolant return tube in the bath since the centrifuge could be flooded if this tube accidentally fell out. g) Feed head and seal preparation. The feed channels must be clean and free from obstruction. It is most important to check that the feed-head bearing is in good condition. There are a number of different types of bearing in use, open, sealed on one side, or completely sealed and prepacked with grease. These bearings can be washed in hot water, in a degreasing solvent (petrol, trichloroethylene etc.), then hot water again, followed by thorough drying and finally repackingwith grease. Spare bearings should always be kept in reserve, but in case of emergency the above method will usually ‘revive’ an old bearing enough to allow an important run to be completed. Before every run, the two elements of the seal should be carefully examined. The metal part normallyrequireslittlemaintenance,but may show signs of corrosion if it was not dry during storage. It should ideally have two highly polished concentric rings, but if pitting is noticed on these rings, as shown in Fig. 5.13, then the seal can be
Fig. 5.13. The damage caused to the stationary seal of an A-XI1 rotor by corrosion. This was left for about 48 hr before being washed clean of the remaining sucrose. S,,h,r11 r n k r
p 287
152
DENSITY GRADIENT CENTRIFUGATION
restored to almost new condition by the method described below. On the Rulon, the two sealing surfaces (which rub against the metal part, and produce the two polished rings), should be absolutely flat, unscratched, and shiny. If there are scratches, or if the seal had leaked during the last run, then it may be reconditioned. Polishing either seal element is carried out as follows. Find a piece of perfectly flat glass (a thin-layer plate is ideal) on top of this place a sheet of ‘wet and dry’ emery paper (grade 600) and wet it liberally with water. Holding the Rulon (or the metal part) between thumb and third fingers, with the sealing face pointing down flat on the emery paper, use the forefinger to press it down firmly onto the emery paper. With a fairly rapid to and fro motion (similar to using a rubber) rub the seal until all the scratch marks have disappeared, and a matt finish is obtained. While rubbing, make sure that the seal is kept perfectly flat. With the very soft Rulon, this is done very quickly, while with the metal part it takes much longer. Now repeat the whole procedure, but this time use a much finer dry emery paper (Crocus or Buffing grade), rub to obtain a shiny finish. This last step can be used routinely between runs, or just before very crucial runs. When the feed head is placed on the rotor, the guard tray prevents it from spinning freely with the rotor (and so twisting up all the tubes). With the A-XI1 rotor, the feed head is attached to the rotor and stays in place during the whole run. To prevent the stationary part from spinning, the anti-rotation arm must be secured against the top of the bowl. h) Guard tray. The guard tray is fitted after the rotor has been placed in the centrifuge (in the case of removable feed head rotors). Since this guard tray normally locates the feed head, it also determines its height and centering in relation to the rotor. The height is critical for correct spring pressure on the seal, and must be adjusted whenever necessary. The Beckman and IEC feed heads are selfcentering; thus the guard tray location is not so critical, but the MSE and Christ guard trays must be centered by lateral adjustment. With the MSE machines it is particularly important to centre the tray initially; it should then require little, if any, adjustment. However,
Ch. 5
ZONAL ROTORS
153
if the machine has had a major repair involving the removal of the drive shaft housing, the shaft is nearly always displaced and the guard tray must be recentered and relevelled. With MSE’s S S 65 and 75 machines, the guard tray has a spring loaded plunger, which operates a microswitch and thus ensures that a zonal rotor can be spun with the lid open only after the guard tray has been installed. This plunger must be carefully located; otherwise the zonal mode will not be selected. In order to minimise condensation forming on the bowl when the centrifuge refrigeration is switched on during loading, the various rubber sealing rings must be used to seal the bowl as well as possible. These seals also help to prevent any accidental spillage from entering the bowl. i) Positioning the,feedhead on the spinning rotor. This is undoubtedly the most anxious moment for a beginner, but is in fact a very simple and rapid operation. The feed head should be held firmly (with two hands in some cases) and as level as possible. It should then be slowly lowered towards the rotor and carefully lined up. Once contact is made it should be pressed down firmly and deliberately without hesitation or jerks. It is then turned (usually left, anticlockwise) until it locks in the guard tray and the operation is complete. For removal of feed head see 55.1.3.5. Once in place the four tubes connected to the feed head should be secured so that they do not vibrate, kink, and cannot be accidentally pulled off. Beckman and Christ machines are provided with tubing blocks for this purpose. 5.1.3.2. Gradient loading a ) Pumping the gradient into the rotor. When all the preparations outlined in the previous section and in Fig. 5.10 are complete, the gradient pump can be switched on to feed the gradient into the spinning rotor. In our experience, the flow rate during the loading should not exceed 50 ml/min with sucrose gradients and 30 ml/min with viscous materials like Ficoll but 50 ml/min or more can be used with salts like CsCl or NaBr (for filling times see Table 5.3). .Yr,h,c c I r,n/c\ p 35’7
154
DENSITY GRADIENT CENTRIFUGATION
TABLE 5.4 Instructions for gradient loading. Check that the feed head is in place and the rotor is spinning. Start to pump the gradient. - Ensure that air is being displaced from the rotor. - Ensure that the flow rate is not higher than 60 ml a minute. - Check the temperature of the ingoing gradient and adjust it to the required value. - With gradient loaded, switch off pump and clamp both feed tubes, (not coolant tubes). Transfer pump inlet tube to a beaker of 'cushion'. - Unclamp tube, start to pump cushion at reduced rate since it is usually viscous. - Carefully watch the centre line. - When liquid appears from centre line, rotor is full. - Switch off pump. - Clamp both feed lines. -
~
The coolant must always be circulating while the feed head is in position on the rotor. To control the temperature of the gradient, stir the ice round the cooling coil or add salt to the ice (see Fig. 5.12). To ensure that the temperature of the gradient in the rotor is correct a number of precautions must be taken. If the temperature is to be 5 "C,the rotor which has been precooled to 5 "C is placed in the centrifuge whose temperature control is set at 5 "C, and the gradient is then controlled at 5 "C at the probe, as it goes into the rotor. Lastly the coolant water used to cool the seal must be below 5 "C.Tap water (usually about 18 "C)must not be used for cooling the seal if the gradient is to be maintained at 5 "C, as the seal will then equilibrate between 5 "C (due to the gradient) and 18 ' C , and the actual gradient temperature will be nearer to 18 "C than 5 "C.If the initial temperature of the rotor and contents is much higher than the required gradient temperature, the evacuated bowl will cause the refrigeration system to take a considerable time to cool the rotor down to the selected temperature. It is therefore important to start with the rotor and contents at the correct and known temperature. These stringent conditions apply particularly if the zonal rotor is to be
Ch. 5
ZONAL ROTORS
155
used as an analytical tool when the gradient temperature must be known to calculate the sedimentation coefficient. From the centre of the rotor the line leads into a beaker of water (Fig. 5.10). The appearance of bubbles will indicate that air is being displaced and that the gradient is entering the rotor. If any liquid appears in this line before the rotor is full, the gradient is leaking into the centre line (cross-leak see $5.1.3.3). A slight cross-leak at this stage can be tolerated provided a) that it does not contaminate the centrifuge (see §5.1.3.4), b) that the volume of the gradient is not reduced significantly, and c) that the fault can be remedied before the sample is loaded. Cross-leakage can also be caused by a blocked feed line to the rotor wall. An indication of this can be given by using a short piece of soft silicone tubing anywhere in the line from the pump to the feed head. This will retain its shape normally, but will begin to swell if there is a blockage. b ) Gradient making and monitoring. With all zonal rotor systems it is inadvisable to allow any material to pellet on the rotor wall, as this may block the wall feed lines and prevent unloading. To prevent pelleting it is necessary to use a ‘cushion’ of high density gradient solution. When the rotor is full, the light end of the gradient will appear in the centre line. The pump must be stopped immediately and both of the tubes clamped firmly. Although an automatic switch-off device could be used at this point there is little to be gained since the operator must be present. The total volume of the rotor is known and the volume required to fill the rotor can be calculated. Monitoring of the gradient profile during loading can indicate any serious errors in the gradient shape. It is relatively simple to stop the run at this stage, clean the rotor and start again. In the absence of a gradient monitoring device the gradient is usually checked at the end of the run, by measuring the refractive indices of the individual fractions. Any error in the gradient is only detected after the completion of the whole run and a precious sample may have been lost. It may help to use a pressure gauge to monitor the pressure in Sd>,u<1 md<J\p 28x7
156
DENSITY GRADIENT CENTRIFUGATION
the input line. This should ideally not exceed 15 psi (1 atm.) The maximum permissible pressure can easily be measured by blocking off the centre line through which air is displaced and noting the pressure at which the seal begins to leak. If the pressure during loading approaches this value, the pump rate, and hence the pressure, should be reduced. If all the channels are clean there should be little back pressure in the feed line even when pumping ice cold 2 M sucrose. c ) Generalprecautionsduring loading. Although it should have been checked during the preparation, the feed head bearing may become noisy during loading and could finally seize up. This is particularly serious with the B rotors since a bearing seizure will almost certainly stop the rotor with serious consequences. During loading the vacuum chamber lid must not be closed as this will also damage the feed head. When loading the HS rotor (MSE Ltd), in the HS 18 centrifuge which is not fitted with the slow-speed zonal control, the loading speed 1,500-2,000 revs/min must be constantly checked and usually needs adjusting to keep it in that range. Heat from the centrifuge motor may be conducted up the drive shaft into the core of a zonal rotor. In order to counteract this effect in the case of MSE's A-XI1 rotor operated in the Mistral 6L centrifuge in which the distance between motor and rotor is very short, it is necesary to select a temperature of - 10 "C if a seal temperature of about 5' is wanted. 5.1.3.3. Sample loading (Table 5 . 9 ) a ) Methods of' introducing the sample. The simplest and commonest method is to inject the sample with a hand-operated syringe (see Fig. 5.14). Convenient sizes of syringes covering the range of most likely sample volumes are 20 ml and 50 ml (plastic disposable syringes are suitable). The tubing recommended earlier for the feedhead connection is of the correct bore to be a good push fit onto a standard Luer syringe tip. A more convenient way of attaching the syringe is to use a three-way tap (on the end of the centre line) with standard syringe connectors and lock the sample syringe into
Ch. 5
157
Z O N A L ROTORS
TABLE5.5 Instructions for sample loading. Unclamp the edge line and position it over a beaker of water. Attach the syringe containing thc sample to the rotor centre line. Sample must be of a density slightly lower than the light end of the gradient - Unclamp the centre line and inject the sample slowly and smoothly. - Observe that the dense cushion is displaced from the edge. - By watching the drops of displaced cushion from the edge line the rate of injection can be carefully controlled. - Fill syringe with overlay (less dense than the sample). Unclamp centre line and inject correct volume of overlay. - Ensure that no air bubbles enter thc rotor. - Remove feed head, and put on the running cap (if using removable feed head rotor). - Close chamber lid, and accelerate rotor to desired speed. ~
-
~
~
this (see 45.1.4). The centre line should be unclamped (or the tap opened) and the sample injected slowly, at a rate of about 5-10 ml/min. It is particularly important to start the sample injection slowly in order to avoid mixing the leading edge of the sample band with the gradient. The same applies to the injection of overlay. If the
Fig. 5.14. Sample loading in a zonal rotor. The sample A is injected with a syringe into the centre of the rotor B. The displaced heavy end of the gradient appears from the edge line C and is seen to sink through the water in the beaker D. st!h/c
,I.
2x7
158
DENSITY GRADIENT CENTRIFUGATION
rate of cushion displacement from the edge line is watched the sample can be injected at an even, jerk-free rate. Care should be taken not to inject air bubbles with the sample. When the injection is complete, the tube should be clamped before removing the syringe. This normal method of sample injection is almost completely reliable with the modern seal systems. If a cross leak should occur it may be cured by one of the methods listed below (b). A simple gadget can be used to produce, manually, an even flow with a syringe. The syringe barrel is mounted on a stand and the piston is connected to a threaded rod with a cranking handle. The handle is turned to push the syringe piston and load the sample evenly (see $5.1.4). The sample can also be loaded with a perfusor or a syringe pump which have a range of very slow accurate flow rates, one of which is usually suitable. A number of these pumps are commercially available. A peristaltic pump with a low flow rate can be used in much the same way as the perfusor, although there is a risk that the sample may be warmed in its passage through the pump. It may not be necessary to inject the sample into the rotor if the flow of the gradient pump can be reversed, and reduced (i.e. as with the IEC gradient pump, see $5.1.4). The centre line is dipped into the sample and the cushion sucked from the rotor wall so that the sample enters the centre of the rotor. For some critical rate separations it may be necessary to apply the sample as a ‘reverse wedge’. This means that sample is incorporated actually into the density gradient as opposed to being simply a step in say 0.3 M sucrose. At the same time the sample concentration is lowest at the leading edge and highest at the trailing edge. For a full description of how to make a ‘reverse wedge’ sample see below (c). b ) Leakage and cross-leakage. Anderson (1968) writes ‘It is unfortunate that the high-speed seals commercially available at this writing violate most of the above principles and almost invariably leak’. Although seals have improved since that was written, the warning is still relevant today as the leakage associated with the coaxial seals can still cause problems, especially for the beginner. Since cross-leakage
Ch. 5
ZONAL ROTORS
159
arises as a most serious problem during sample loading it seems appropriate to deal with it here. As the sample is injected into the centre of the rotor, ideally only cushion material displaced from the edge should flow from the edge line. If the sample cross leaks, in preference to entering the rotor, it leaks across the sealing face of the rotating seal, and emerges from the edge line and mixes in with the cushion, as shown in Fig. 5.15. Once this occurs, it is difficult to estimate how much sample is lost. Cross-leakage is relatively easy to detect in the case of sample loading (see below) but is much more serious and difficult to detect if it occurs during unloading (see 5.1.3.5). Detection of cross-leakage. 1) The simplest way to detect crossleakage in the actual run is as follows. The edge line should be dipped in a solution of exactly the same concentration as the cushion in the rotor. During sample injection the displaced cushion will mix
t
I t I
'L
Fig. 5.15. Cross-leak. Thick arrows show where the sample leaks between the faces of the rotating seal into the edge line containing the cushion. Suhiwr index p. 287
160
DENSITY GRADIENT CENTRIFUGATION
with this solution perfectly and invisibly. Any sign of streaming of two solutions of different refractive indices (the eye can detect very small differences), will indicate that the sample (which has a much lower gradient solute concentration than thecushion) has cross-leaked into the edge line. A coloured sample will produce coloration in the beaker containing displaced cushion. 2) Set up a dummy run. For this the rotor can be partly filled with water before assembly to save time, but the last 25% of the volume should be cushion (2M sucrose). Prepare a sample of diluted ink or some other coloured soluble dye, inject it into the rotor and observe if any of the coloration appears in the displaced cushion. 3) An alternative method, which is much more sensitive, is to inject a radioactive sample. If the activity of the sample is known, the cushion can be quickly tested for radioactive contamination. If the test run shows there is no cross-leakage, it is unlikely to occur during unloading. Causes of cross-leakage: Cross-leakage, although less common with modern seals can still be caused by misuse. Chevrenka and Elrod (1972) extensively discuss cross-leakage and how it can be avoided. The following list gives the common causes, with the most common first. - Worn or damaged seal. The two elements of the seal are delicate and should always be in perfect condition. Any radial scratches or marks, on the Rulon in particular, should be polished out (as described in 4 5.1.3.1). - Spring tension is set too low. In order to keep the two seal elements tightly pressed together it is important to adjust the spring tension. This pressure will depend on the initial strength of the spring used, but as a general rule we have found that if the spring is compressed by 4-5 mm the pressure should be sufficient. To set this, scribe a line on one of the tubes on the feed head, flush with the top of the feed head, with the spring uncompressed, then as the tension is increased, the distance between the line and the top of the feed head will increase until it reaches the optimum &bout 4-5 mm. Once set this should not require readjustment unless either of the seal elements are replaced.
Ch. 5
ZONAL ROTORS
161
- Back pressure. In the case of a blocked or even slightly obstructed channel in the rotor, core or septa, the pressure created in the feed head could get high enough to part the two elements of the seal. This will, undoubtedly, result in both cross-leakage and leakage. Enough back pressure may also be created to provoke a cross-leak, if air bubbles are admitted into the rotor. - Vibration of the seal. If the feed head is, 1) not located squarely on the rotor, 2) badIy out of centre, 3) set with low spring pressure, or 4) if the cone in the stationary seal is not perfectly central, the two faces of the seal may not stay firmly in contact, but may vibrate or patter, (the screws and the four tubes will appear as a blurr). This will almost certainly produce a serious cross-leak. The only cure lies in re-adjusting the feed head and guard tray. - Damaged O-rings. Various O-rings in the rotor core act as seals between the two crucial lines, namely edge and centre, and crossleakage can also occur if these rings are damaged. They should be examined and replaced regularly. Manufacturers usually provide a comprehensive set of spare O-rings in the kit with a new rotor. c) Sample loading related to different applications. The maximum permissible concentration and volume of sample depends on whether the separation will be on a rate sedimentation or isopycnic basis. With rate sedimentation it is important to understand the meaning of band capacity (see 52.1.2). Zonal rotors can be used just like angle or bucket rotors to prepare large pellets. A B-XV rotor for example can be completely filled with nearly 1700 ml of post-mitochondria1 supernatant and the microsomal ‘pellet’ can be scraped off the rotor wall (Nelson et al. 1971). The K and RK rotors can be fitted with special K-11, RK-11 cores, designed particularly for this kind of operation. For very exacting rate sedimentation work, it may be necessary to use a ‘reverse wedge’ sample principle (Meuwissen, 1971). The principle and its advantages are described in 42.1.2. A simple linear gradient maker, such as is commonly used for swing-out tube gradients (see Fig. 4.1), may be used to make a ‘reverse wedge’. The mixing chamber contains a solution of similar composition to Sirhiecr rndey p 2x7
162
DENSITY GRADIENT CENTRIFUGATION
the lightest part of the gradient (i.e. 0.3 M sucrose) whilst the reservoir contains the sample in a lower density solution (i.e. 0.2 M sucrose). The resultant gradient will decrease linearly in sucrose concentration (from 0.3 to 0.2 M sucrose) and at the same time, increase linearly with sample concentration (from zero to the concentration in the reservoir) as presented graphically in Fig. 2.3. Ideally the overlay should also be made as a gradient thus establishing the sample band in a continuous gradient. d ) Special features of different rotors. B-XXZX. This rotor offers the facility of loading the overlay sample and gradient from the edge. The principle is attractive and appears to offer some advantages. However, it has not been accepted as widely as was first anticipated, and relatively few publications quoting the application of the B-XXIX have appeared. The edge of the rotor where the sample first appears, at a typical loading speed of 3,00&4,000, is already at a high centrifugal force (up to 5,000 g). This means that even at the loading speed, some sample may be pelleted on the wall and may then be washed off as the rest of the gradient follows, thus spreading the sample. Contractor (1973), Cook and Apsey (1970) and Brown et al. (1973) have reported the use of the B-XXIX rotor and the difficulties encountered. A-XZZ rotor. This large rotor expands during acceleration and will draw air into the centre core if the centre line is not clamped. Air should not be admitted into this rotor as serious imbalance will occur. The most satisfactory way to deal with this problem is to attach a small reservoir, containing about 100 ml of water or overlay, to the centre line. This is connected after the sample and overlay are loaded and the tube is left unclamped, allowing the liquid to be drawn into the rotor (Fig. 5.16). The illogical arrangement of the feed lines of an A-XI1 feed head (MSE Ltd.) has already been mentioned (p. 134). This arrangement, causes a practical problem for some applications, (nuclei particularly), since the sample is in effect homogenised as it passes through the outer holes of the coaxial seal. MSE can supply a rotor core with the holes drilled in a conventional manner, thus allowing the
Ch. 5
163
ZONAL ROTORS
Fig. 5.16. Before accelerating the A-XI1 rotor the centre line is left connected to a reservoir of overlay, allowing this to be sucked in as the rotor expands during acceleration. The edge line is closed off at the three-way tap.
sample to pass through the centre line (low shear). However, it is quite unnecessary to go to this expense as the A-XI1 can be assembled with the centre perspex piece (carrying the septa) upside down. The channels then all match up to convert the feed lines to the conventional layout. H S rotor. With our own HS rotor, we regularly encounter high back pressure when injecting the sample. After applying all the known remedies, the back pressure is still present, and we therefore conclude that it must be a characteristic of the rotor. 5.1.3.4. Acceleration, spin, and deceleration (Table 5.6) If sedimen-
tation coefficients need to be calculated, it is necessary to know the total integrated centrifugal field. This value can be worked out by noting the speed at regular intervals during acceleration and deceleration, together with the time spent at maximum speqd. Alternatively the value can be automatically computed with an electronic integrator fitted to the centrifuge. The MSE AX11 rotor, while being accelerated (or decelerated) may Sihi
p 287
164
DENSITY GRADIENT CENTRIFUGATION
TABLE 5.6 Instructions for acceleration, spin and deceleration. - Remove the feed head, or if non-removable see that it is well secured. - Place the vacuum-tight rotating cap on the rotor (if vacuum centrifuge). - Close chamber lid (if vacuum centrifuge). - Accelerate rotor to operational speed. - Set the auto-timer for the required run duration. - During spin observe rotor temperature and vacuum. - Decelerate rotor to unloading speed. - Open lid.
vibrate between 600 to 1,000 rpm. The rotor will stabilise once the speed goes above 1,000 rpm, or below 600 rpm. It seems to be a resonant unstable speed range, which may initially cause concern for the beginner, but is in fact not dangerous. In the case of other zonal rotors this instability is hardly noticeable and occurs at a speed lower than the loading/unloading speed. Reliable and reasonably accurate temperature control is necessary for temperature control during the spin, for two reasons: 1) As already emphasised in 41.6.1.6, the contents of a rotor must not freeze, since this can cause the rotor to burst. 2) It is possible to calculate sedimentation coefficients, and use the zonal as an analytical rotor, only if the temperature of the rotor contents is known. The slow-running zonal rotors operated in non-evacuated centrifuge bowls normally use a rather simple thermistor or thermocouple to measure the temperature in the bowl, and control it by switching the refrigeration unit on or off. This works reliably but cannot be claimed to be accurate. With the AX11 rotor used in the Mistral 6L (MSE) it is difficult to measure the temperature of the contents, but it is probable that in fact there is a temperature gradient, with the centre being warmest and the edge coldest. Fortunately the majority of the applications of the slower rotors do not seem to require accurate temperature control. However, the criticisms of temperature control apply not only to zonal rotors but to all rotors used in these centrifuges. S u h p r inder p . 287
Ch. 5
ZONAL ROTORS
165
In the case of high speed zonal rotors operated in evacuated centrifuges, temperature control becomes a topic of even more controversy. For accurate temperature control the most important requirement is a good vacuum which does not fluctuate during the spin. The bowl vacuum can be seriously affected by condensation or any liquid contamination, such as an accidental spillage. During loading of the gradient and sample, the refrigerated centrifuge bowl can collect a lot of condensation. Comparison of the systems available on the present high-speed centrifuges gives some idea of the problems and how they may be solved. MSE’s old system as used on the SS40 and SS50 machines, measures the temperature in such an indirect way that accurate control is not possible. This system is therefore not suitable for zonal rotors. The present IEC, and Beckman machines and the current MSE machines use an infra-red sensor to measure rotor temperature. A refrigeration or heating cycle is then used to control it. IEC and Beckman position the probe in the bottom of the bowl quite near to the rotor shaft. This central position and the fact that the probe is covered by a protective window, means that this probe is not affected seriously by a poor vacuum. However, the MSE probe (on the SS65 and SS75 machines) is placed inch away from the outer edge of the zonal rotor, and is not covered by a shield. This position is therefore directly in the path of the greatest turbulence and friction. We have previously mentioned the problems of such a system (Dobrota 1971). Christ use a completely different principle. The temperature sensor is a thermistor embedded in the rotor shaft and hence in close contact with the rotor. The refrigeration unit is unusually powerful and is designed to trap any contamination as a frozen solid (much like a freeze dryer trap). Therefore it allows the vacuum to be nearly ideal in spite of contamination in the bowl. The rotor temperature is then controlled by an infra-red heater positioned near the rotor. However, such a system could be subject to serious temperature gradient problems. The skin temperature of a rotor made of titanium (a poor conductor) could be high compared to the centre where the measurement is being taken. In practice this problem does S v h ~ c cwdc\ ~ p 287
166
DENSITY GRADIENT CENTRIFUGATION
not seem to be serious, since users of Christ zonals appear to be quite happy with the temperature control. With a Beckman Ti-14 rotor, spinning at 48,000 in an L-652B the indicated temperature will read 5 "C, even with a relatively poor vacuum of 100 microns. Under identical conditions with an MSE B-XIV spinning at 47,000 in a SS65 (MK11 specification)and the same vacuum, the rotor temperature would read about 15 "C (although the real temperature is 5 "C).This marked difference we think is due only to the sensor position. This idea is confirmed by repositioning the SS65 sensor to a region of less turbulance when the indicated temperature reads about 8 "C, or by reducing the speed down to about 30,000 revs/min at which the temperature reads about 5 "C in spite of a bad vacuum. Reliable temperature control of the MSE B-XIV rotor at maximum speed is only possible if the vacuum is better than 10 microns. Both MSE and Beckman (and IEC) systems have serious disadvantages. Firstly the latter system does not measure the temperature at the edge of the rotor. Thus with a poor vacuum the indicated temperature may still be correct at the position of the probe while the real temperature at the edge of a zonal rotor or the tip of a swing-out bucket will in fact be much higher. With the MSE system the erroneous reading of the probe is at least some indication of the temperature at the edge of the rotor. However, since the indicated temperature is higher than the real temperature the refrigeration stays on continually (attempting to cool the rotor to the set temperature) with the consequence that the rotor contents could freeze. In spite of a poor vacuum, it is possible to achieve some temperature control of the MSE system by regular readjustment during the spin and resetting the selected temperature to the indicated temperature. The rotor must be precooled to 5 "C and precautions taken to ensure that it is still about 5 "C when the lid is closed. When it accelerates to maximum speed and the probe 'thinks' and indicates that it is 15 "C (due to poor vacuum), the required temperature is reset to 15 "C, and the system will then control at an apparent
Ch. 5
167
ZONAL ROTORS
‘15 “C’, which is in fact 5 “C. It almost certainly will need readjustment during the spin since the indicated temperature will fluctuate with the vacuum. Also there will be a temperature gradient set up in the rotor. However, this fiddly method will prevent the contents from freezing. The manuals advise that condensation or spillage must not be allowed to contaminate the bowl. With many years experience we have discovered that this is quite unrealistic and almost impossible. Therefore we find it refreshing that Christ have accepted the ‘fallibility’ of the zonal user, and have designed their system to cope with contamination. We therefore come to the conclusion that for accurate temperature control the vacuum in the centrifuge bowl must be good; ideally between 1 to 5 microns (about 0.001-0.005 torr). Any contamination must be either rendered harmless (by trapping it) or be prevented from getting into the bowl. We quoted the use of a cold trap for removing condensate from the centrifuge bowl (Mullock et al. 1971). This glass trap, installed in the vacuum line of a SS50 (MSE) between the bowl and vacuum pump was of a two piece construction with the lower part removable for emptying and cleaning. It was protected by a cage since breakage during a spin could have been disastrous. Although it proved successful a slight snag was that the solid CO,/Cellosolve mixture would last only about 12 h without topping up. The Christ system solves this problem by using its powerful fridge as a trap, while the other manufacturers seem reluctant to provide such a facility. The cost of such an improvement would add relatively little to the price of a centrifuge. We feel that this approach to the vacuum problem is more realistic than preventing contamination, especially condensation, from entering the bowl. The method recommended for cooling the K, RK rotors is to place the rotor, which is at room temperature, directly in the centrifuge. The refrigeration is switched on only after the bowl has been sealed and evacuaied. Ice cold buffer, or distilled water, is then pumped through the rotor to cool it. Only then is it ready for use. The main aim of this complex procedure is to prevent any moisture Slihiccr index p
287
168
DENSITY GRADIENT CENTRIFUGATION
contaminating the vacuum system, which would otherwise upset the temperature control. Aqueous contamination, which seems to be the great enemy of temperature control, can be removed to some extent by the machine's own vacuum pump. For this purpose regular overnight flushing runs should be done, particularly on machines used for zonal work. However, for this to be effective the temperature of the centrifuge bowl must be above ambient, i.e. 30* and the refrigeration should not be switched on. Under these conditions a two-stage vacuum pump, such as that fitted to a SS65 or 75 machine (MSE), is capable of clearing up to 200 ml of water per hr. Trinick and Rowe (1973) reported that a low speed run of one hour, before accelerating the rotor (in this case a B-XIV) to its operating speed, was necessary to clear the condensation from the bowl of an MSE Super-speed 65 and so obtain a good vacuum. We have, on occasions used this method, but with success only if the temperature was set to at least 30 "C. This, however, may be unacceptable for some applications since there is no way of telling how much the temperature of the rotor contents has risen or how long it takes to cool down once the temperature is reset to 5 "C. We have tried to keep the atmosphere in the bowl dry by blowing C 0 2 in through the drain of the bowl. It appeared to reduce the condensation but did not stop it entirely. Apart from keeping an eye on the temperature and perhaps the vacuum, the spin and deceleration are fully automated and require little attention. However, the time can be well used to prepare for the unloading. This preparation can take up to 30 min with the more elaborate unloading layout. The detailed steps of preparation are listed in the next section. 5.1.3.5. Unloading a ) Preparation for unloading. Before attempting to displace the contents of the rotor, the flow lines should be connected as shown in Fig. 5.17. The layout is much more complicated than that used in loading. Both lines are in use and the large number of connections
Ch. 5
ZONAL ROTORS
169
TABLE 5.1 Instructions for unloading. ~~~~
~
~~~~
~~
Remove vacuum cap. - Check the seals and the feed-head. - Place feed-head (if removable type) on rotor. - Ensure that coolant is being pumped through the coolant lines. Check that the edge line is full of displacing solutions and is completely free of air bubbles. Connect the centre line to a flow-through cell of a spectrophotometer. If available, also use a flow-through refractometer or density monitor in this line. Fill the centre line with distilled water and zero the monitoring instruments. - Check that the monitoring instruments are connected to the chart recorder. - Zero the recorder and ensure that the pens are writing. - Connect outlet of the centre line to a fraction collector. - Check that the event marker from the fraction collector is connected to the recorder. - Switch on the recorder chart drive. - Select the appropriate flow-rate and start to pump displacing solution. - During pumping carefully watch recorder trace and check that all systems are working correctly. - With gradient unloaded stop the pump and remove feed-head. - Wash out all lines and flow-through cell with warm water and then formalin (0.1%). - Switch off recorder and other instruments. - Stop the rotor, remove it from the centrifuge, wash, dry and put it away. - Wipe off condensation in the centrifuge bowl. Wipe any spilt gradient solutions and clear up. -
~
~
~
~
~
must be perfectly secure. Constant attention must be paid to the displacing pump, monitoring equipment, recorder and fraction collector. The success of this complex procedure will also depend on thorough preparation. It is imperative that the rotor should not stop during unloading (see $1.6.1). Although safeguards are incorporated in each centrifuge, minor modifications can improve reliability for zonal work. For example the centrifuge may stop if a fault is registered as part of a ‘fail’ sequence which normally cannot be overridden. To overcome this kind of problem, our own SS65 machine has been provided with an override sequence allowing unloading even if a fail Si,h,arr o d e v p 287
Fig. 5.17. Diagramatic view of unloading set-up. Displacing solution (A) is pumped to the rotor edge (B). The gradient is pushed out first through the ‘thermocouple cell’ (C), then the recording refractometer (D) and the spectrophotometer (E) and finally to the fractionating device (F) and the collector tubes (G).
signal does occur. There is no reason why other manufacturers should not fit such a modification provided they are assured that only well-trained operators will use it. All the tubing connectors, especially in the line from the displacing pump to the rotor must be very secure as the pressure may get as high as 30 psi (2 atm.). This line must be completely filled with displacing solution. If this is not done and water or other low density solution is pumped to the edge of the rotor, it will stream up through the gradient and destroy it. It is also essential to clear all air bubbles from this line. This is done by pumping overlay (or even water), to the centre of the rotor exactly as in sample loading until all the bubbles in the rotor core feed-head and cushion line are flushed out by the cushion solution. If this is not done, air bubbles may cause sectorial unloading (see c. in this section). Threeway taps, fitted with standard syringe connections, can be very
Ch. 5
ZONAL ROTORS
171
convenient for diverting flow, stopping it completely and attaching syringes, in both the displacing and the centre line (see Fig. 5.17). Artery forceps (Spencer-Wells) should also be available for clamping tubing in case of emergency. Ideally the line from the rotor centre through which the gradient will be pumped, should be connected through the monitoring instruments and finally to the fraction collector. The importance of these instruments is discussed in detail in 55.1.3.6 and 55.1.4. Before starting to unload, the line from the centre of the rotor to the fraction collector should be tilled with water so that the monitoring instruments and recorder may be zeroed. For details of fraction collecting refer to 55.1.4. Only after checking all equipment and ensuring that none of the lines are clamped or blocked should the displacing pump be switched on to begin the unloading. 6 ) Main features of unloading. In unloading, there are many things which can go wrong, and constant attention is necessary. The operator should keep an eye on the following: - There must be no excessive build-up of pressure in any of the fluid lines (soft silicone tubing will swell up quickly and thus give visual indication of high pressure). - No leaks occur at the numerous joints. - No cross-leakage at Rulon seal. - Ensure that the displacing solution reservoir does not run dry (air bubbles would enter the edge line). - Fraction collector and volume sensing device. - Feed-head, seal and bearing. - Temperature of the outflowing gradient. - Supply of coolant to the seal. At the end of unloading, the equipment must be switched off and thoroughly cleaned (spilt sucrose must be wiped off). First, wash and dry the feed-head; the importance of this is emphasized in Fig. 5.13. The displacing line and the centre line plus the monitoring instruments can be directly connected using the pump to flush the whole system with warm water followed by 0.1% formalin. Suhlrcr mdrxp 287
172
DENSITY GRADIENT CENTRIFUGATION
c ) Problems during unloading. 1) Cross-leakage. If cross-leakage is detected (see Fig. 5.18 which shows a typical refractometer trace) during unloading, the following measures should be taken. With the later type feed-heads, firstly remove the feed-head, dismantle, examine the faces of the seal and repolish quickly (see 55.1.3.1). If this does not cure the cross-leak, then the cause is probably a ‘block’ in one of the lines or channels in the rotor. The line from the rotor should therefore be checked for possible obstructions. If none are found, the blockage is probably in the rotor in which case little can be done except to reduce the pump rate.
Fig. 5.18. This refractometer trace ( ) shows a typical crossleak during unloading. First spiky region is due to the crossleak which was later rectified and then gave the smooth trace (dotted line is the expected gradient trace). Absorbance at 650 nm is also shown (---), Zero absorbance is at the top.
With older type seals a cross-leak caused by a damaged Rulon cannot be cured since the Rulon is located in the spinning rotor. However, in some cases it may be possible to prevent cross-leakage for a short period (but long enough to complete the unloading) by
Ch. 5
173
ZONAL ROTORS
equalising the flow-rates into and out of the rotor. We have used such a method successfully (Dobrota 1971)on a rotor which otherwise tended to cross-leak regularly. Two pieces of identical pump tubing (autoanalyser or silicone) are placed in the same peristaltic pump ; one is used to pump the displacing solution to the edge of the rotor while the other sucks the gradient out of the centre at exactly the same flow-rate. Although this method should not be necessary with the modern seals it is worth remembering as a last resort. 2) Sectorial unloading. If the flow-rate in all the septa channels is not identical the cushion will enter unevenly, causing one or two compartments to unload slowly. This results in distorted bands (not perfectly circular) which will leave the rotor spread in a larger volume than normally. This phenomenon is illustrated in Fig. 5.19. It is usually caused by air trapped in the rotor core or by a block in the septa channels. To prevent it, all air must be removed from the core and the displacing line just prior to unloading. If it occurs once unloading has started it may not be practical to remove the air but the bands can still be smoothed out to a certain extent by reducing the pump rate and increasing the rotor unloading speed. While this phenomenon can be easily seen on the transparent A-XI1 rotor as a wobbling band it almost certainly will not be detected on the non-transparent rotors such as the B series. d) Rotors with special features. B-XXIX and B-XXX. This type of rotor called ‘edge load/unload rotor’ can be unloaded either by displacement with a dense solution as described above, or by pumping water (or any low density solution) to the centre and collecting the gradient from the edge. This facility offers a number of advantages. Firstly a part of the gradient containing a band which is not needed can be removed and then replaced with a different portion of gradient for a further spin. This is particularly advantageous since the ‘dense’ part (near the wall of the rotor) of the gradient can be removed and then replaced by a new gradient. A typical application is quoted by Contractor (1 973). Secondly, if an expensive material is used for the gradient it can be unloaded using water rather than an expensive displacing solution. Siih/c(I
rirl I
p 2x7
174
DENSITY GRADIENT CENTRIFUGATION
Fig. 5.19. Sectorial unloading. The band is obviously not circular demonstrating that the segments are being unloaded at different rates. Photograph was taken while the A-XI1 zonal rotor (MSE) was illuminated with a stroboscopic light. Exposure time was about 1 sec.
5.1.3.6. Monitoring and fraction collecting In this article the term ‘monitoring’ is generally used to describe the continuous recording of band position and gradient shape by use of flow-through instruments during loading or unloading. A typical recorder trace obtained in our laboratory (Fig. 5.20) shows refractive index (gradient shape), absorbance (peaks representing bands) and also the position of each fraction collected.
Ch. 5
175
ZONAL ROTORS
Fig. 5.20. A typical trace of refractive index (gentle curve) and spectrophotometer trace (sharp peaks, A,,o) obtained during unloading. The event marker trace at the top shows the position of each fraction.
Parameters which may need to be monitored 1. Density gradient, profile usually measured as refractive index. 2. Absorbance, to locate the separated bands. 3. Temperature of the gradient and the rotor. 4. Viscosity, when mixed gradients are employed. 5 . Vacuum. Summary of monitoring requirements during various steps of a zonal run. Gradient loading: gradient shape ; gradient temperature. During thespin: rotor temperature; centrifugevacuum; rotor speed. Unloading : gradient shape ; absorbance ; gradient temperature ; viscosity (perhaps). Siihicrr
mdcr p 287
176
DENSITY GRADIENT CENTRIFUGATION
The equipment needed for monitoring and other zonal work is listed and described fully in the next section. a ) Density or refractive index. The importance of a recorder trace which gives an immediate visual check on the gradient profile has already been mentioned. Furthermore if a cross-leak should occur during unloading, the only way it can be detected instantly is by using a continuous flow monitor as shown in Fig. 5.18. Although it would be logical to monitor density directly, it is in fact more usual to monitor refractive index because a number of continuous-flow recording refractometers are available, whilst there is a lack of suitable density meters. Most of the gradient materials commonly used do have a measurable refractive index, i.e. sucrose, Ficoll, dextran, Urografin, Metrizamide, NaBr, CsCl, K citrate, etc. However, while the refractive index of a solution of any particular compound is simply related to the concentration and hence to the density, there is no general relationship between refractive index and density. Hence the refractive index profile of mixed gradient solutes would be meaningless. Since refractometers measure the deflection of a light beam as it passes through the liquid (in the flow-through cell), readings may be disturbed when a very turbid peak of suspended particles passes through the cell. The range of refractive indices of the most common gradient solutes would require such an instrument to respond to a difference of 0.1 above water. Hence the range of 1.333 (water) to 1.433 would in terms of density represent a range of 1 to 1.265 for sucrose and 1 to 1.9 for CsCl. b ) Absorbance (extinction). An intriguing approach to the problem of locating the bands was described by Anderson et al. (1966a). A light source and a photocell were mounted around and moved along the radius of an A-XI1 rotor. The bands were located and recorded while still in situ in the rotor. Thus by repeated scans during the run, the rate of movement of each band could be followed in much the same way as with an analytical ultracentrifuge. The detection of the separated bands during unloading can best be done with a colorimeter or spectrophotometer. Naturally these in-
Ch. 5
ZONAL ROTORS
177
struments must be equipped with a flow-through cell and must have an analogue recorder output. The choice of the appropriate instrument can pose some problems. Firstly it may be dictated by budget or availability. At this point let us make it clear that the nature of the materials separated on the transparent zonals like the A-XI1 usually makes them readily visible by eye. Thus the success of the separation can be judged visually just as well as with the most expensive spectrophotometer. However, because of the need for a permanent record of the peaks related to each fraction, we would recommend the use of such an instrument. A cheap colorimeter is adequate for detecting turbid material (larger organelles : - mitochondria, nuclei, etc.) while a spectrophotometer (with UV facilities) will be more suitable for detecting smaller particles and macromolecules. The turbidity trace at 650 nm is not likely to be quantitative with either a colorimeter or a spectrophotometer so that the cheaper colorimeter might as well be used. If, however, the separated material is relatively pure RNA, DNA or protein, then a UV trace at 260 or 280 nm would be more quantitative, thus making the spectrophotometer the better instrument. Chromatography UV monitors are usually designed with tiny flowthrough cells and very narrow connections dictated by the relatively low flow rates. If a suitable cell which can cope with flow rates of up to 50 ml/min (with ice-cold sucrose solution) is available, then it can be used for peak detection. It may be difficult to estimate in advance the relative heights of the peaks and thus ensure that they all fall within the absorbance range available on the particular spectrophotometer. Although the transmission scale will cope with this, peaks above the absorbance of about 0.8 (16% transmission) will be compressed. It is therefore preferable to use the absorbance scale, but now we come up against the problem of scale expansion which can be solved in two ways. Firstly the more sophisticated spectrophotometers, are capable of fairly accurate measurements above absorbance of 2. If during unloading, a peak is seen to go off scale, a new range can be quickly selected to bring the peak back on scale. This approach unfortunately SiihIccr mdcrp 287
178
DENSITY GRADIENT CENTRIFUGATION
depends on the availability of an expensive and often cumbersome instrument. An alternative way is to change scales by reducing the light path of the flow-through cell, during unloading. This can only be done with special flow-through cells of which two types are available from Beckman-RIIC. The first is a complex cell whose light-path is fully adjustable between 0-5 mm. The micrometer light-path can be adjusted only if the cell compartment is opened. This in turn means that the displacing pump must be stopped during adjustment otherwise a sharp peak may be lost. The other which is simpler and cheaper has four fixed light paths (10, 5, 2.5 and 1.25 mm), each selected by moving the cell up or down (Fig. 5.21 and $5.1.4.5). Since both types of cell are very sensitive to air bubbles, a bubble trap must be used in the line just before the cell. The particular advantage of this system is that the spectrophotometer is working within an optimum and accurate range. c) Viscosity. This must be known for calculating sedimentation coefficients. If the gradient is a pure solute, i.e. sucrose, the viscosity can be calculated from the gradient concentration. Therefore continuous monitoring of viscosity is in most cases unnecessary. d ) Temperature. To achieve good temperature control during a zonal run (see 5.1.3.4.) it is essential that both rotor and gradient are precooled to the correct temperature. The gradient temperature must be continually checked during loading and unloading by means of the small thermocouple inserted in the flow line (see Fig. 5.11). In view of the possible discrepancy between the real and the indicated rotor temperature, there is little point in recording the indicated rotor temperature. However, this temperature should still be checked from time to time as it can give an indication of possible faults (i.e. non-functional fridge). e) Vacuum.A check on this can give useful information on the state of the temperature control (see 5.1.3.4.). f ) Collecting offractions. In our experience an operator should be free during unloading to check the equipment and cope with any emergency. In the past (see Dobrota and Reid 1971), a number of workers have advocated that manual fraction collection is preferable
Ch. 5
ZONAL ROTORS
119
to automatic. We have found that if there was only one operator, manual collection meant that the operator was so busy collecting that he had little time to check all the items listed earlier. We therefore think that automatic collection with manual supervision is more satisfactory with inexperienced operators. The decision on the volume and number of fractions to collect will depend on a number of factors. If two bands are well resolved but in terms of volume are only 40 ml apart, then obviously the fractions should be less than 40 ml. For the study of the distribution patterns of various marker enzymes, it is quite adequate to collect between 30 to 40 fractions. This means that approximately 20 ml fractions are collected from the smaller zonal rotors (HS, Z-15, B-XIV) and 40 ml from the larger rotors (A-XII, B-XV). If one particular region requires special attention, then it may be advantageous to collect smaller fractions over that region. Obviously this cannot be done automatically. Conversely, if the purpose of the experiment is to prepare just one component, there is no need to collect any fractions except around the essential band. One of the major problems in collecting fractions from zonal gradients is that the progressive increase (or decrease) in density and viscosity makes the accurate metering of the fraction volume difficult. The following methods can be used for collecting zonal fractions : syphon, drop counter, timer, liquid level detector. If a syphon is used, it must have a wide delivery tube to allow the rapid exit of the liquid and it must be scrupulously cleaned before every experiment. Even so, with viscous gradient solutions, the volumes collected will increase with the rise in gradient concentration. This occurs because the fractions are being filled (the pump cannot be switched off) during the delivery which takes progressively longer with the increase in viscosity. The use of a timer in conjunction with a constant flowrate during unloading also has drawbacks. The flowrate of most pumps will in fact decrease with the increase in viscosity of the gradient. This is especially marked with peristaltic pumps. Also it is sometimes necessary to reduce the flowrate (or even switch off the pump) in the case of a leak or crossleak thus making Sithicr I iiidcr p 2x7
180
DENSITY GRADIENT CENTRIFUGATION
it rather awkward to readjust the time interval in order to maintain the correct fraction volume. Drop counters are similarly not suitable since the drop size depends on surface tension (which changes with the gradient), and anyway zonal rotors are usually unloaded at flowrates which are so high that the gradient pours out as a continuous stream. The only method other than collecting manually, which we have found satisfactory, is to use an Isco Volumeter or similar ‘electronicsyphon’. For further notes on the equipment available refer to 95.1.4. If the fractions need to be kept cold during collection, the whole collector can be mounted inside a fridge or deep freeze. Alternatively a special collector with either a cold water jacket or a built-in fridge unit can be used. The cold water for such a collector can be supplied from the common cooling unit used for the feed-head. 5.1.4. Ancillary equipment
Although this section might appear to contain a formidable list of useful instruments, zonal centrifugation can be done successfullywith only a few accessories, i.e. a pump and a manual refractometer. However, other ancillary equipment can be productive, especially if automated and versatile so that it can meet other laboratory requirements. The addresses of the manufacturers quoted in this section are listed in Appendix 1. 5.1.4.1. Gradient makers
1) Simple gradient makers can be constructed in the laboratory on a limited budget (see Fig. 5.21). These can produce exponential or linear gradients and have already been described in detail (Hinton and Dobrota 1969). 2) Gradient makers suitable for fixed profiles. MSE Ltd., Model W 101-A is designed to produce linear gradients. It is sold as a kit which includes a vibrating stirrer (effective but noisy) and a pump and represents comparatively good value for money. IEC offer two instruments (no model numbers given). The first
Ch. 5
181
ZONAL ROTORS
TO ZONAL ROTOR
REFRKTOMETER
,--_---_ --------_->-------------------1
Fig. 5.21. A versatile and inexpensive home-made gradient maker. The major component is a pump with two separate and adjustable channels. (From Hinton and Dobrota 1969.)
is a very simple exponential gradient former which was first described by Anderson & Rutenberg (1967). It needs a supplementary pump. The second is a more elaborate instrument with a built-in pulse-free piston pump and a magnetic stirrer. It can produce gradients which are either linear or exponential or a combination of the two. The pump can also be used to load the sample by reversing the flow and withdrawing the cushion out (at a reduced flowrate) while the centre line is dipped in the sample which is sucked into the rotor. Although extremely well made and designed, its price is such that it has to compete with the more versatile fully variable gradient formers. Sorvall GF-2. In specification is very similar to the complex IEC instrument. Although cheap it does need a separate pump. Buchler Zonal Varigrad 2-5 18 1. Suitable for linear and exponential gradients. Suh/<wr nrdc., y 287
182
DENSITY GRADIENT CENTRIFUGATION
3) Complex, commercially-availableinstruments which can form any gradient profile. These are naturally more expensive than .the previous group but are the most versatile. Beckman model 141. The shape of the required gradient is cut out on a sheet of phosphor bronze which then controls the piston pump stroke and therefore the relative amounts of light and heavy solutions pumped. In spite of being one of the oldest available, the precise metering of the piston pumps makes it still one of the most suitable for zonal work. MSE, model Z-100A. This also requires the exact shape of the gradient to be cut on a sheet metal template, which then controls the relative flow-rates of two pumps. A slight snag with the instruments lies in the use of peristaltic pumps whose flow-rates require to be calibrated rather frequently. LKB Ultrograd. Represents a novel approach to gradient making. It meters the relative quantities of light and heavy solutions simply by the length of time a solenoid valve remains open to either solution. This valve is in turn controlled by a ‘graph reader’ on which the gradient is cut in black paper. A pump (which is a separate item) then draws either light or heavy solution, mixes them and pumps the resulting gradient into the rotor. Problems could well arise due to the pump having to suck the viscous heavy solution through the valve and also in the efficiency of the mixing. Isco Dialagrad. The only instrument which does not require the profile to be plotted on a graph or a metal template. The gradient is ‘dialled’ in ten points along a percentage scale, 0 being the light and 100 being the heavy solution. Two pumps then meter the appropriate amounts of the two solutions. Between the preset points the gradient is electronically smoothed to give a curve rather than a series of steps. 5.1.4.2. Pumps The pumps needed for zonal work fall into two
categories: those suitable for pumping gradients and those for circulating coolant round the rotating seal or other equipment.
Ch. 5
ZONAL ROTORS
183
The requirements for a pump for gradients are as follows: it should be able to handle liquids of different viscosities, it must cope with backpressures as high as 35 psi (2 atm.), the flow-rate should be adjustable in the range of (r60 ml/min and the hold-up volume of the pump must be as small as possible. Most piston and diaphragm pumps satisfy these criteria, but the pulsing flow may in rare cases cause leakage at the seal. The pulsing can be reduced by either a pulse suppressor or, simpler still, by using a piece of soft silicone rubber in the line between the pump and the rotor. Peristaltic pumps give a more even flow but are much more affected by variations in back pressure and viscosity of the liquid being pumped. Of the piston pumps available, we have" found the Micro Pump series 2 (Metering Pumps Ltd.) most suitable. It offers positive displacement (up to 250 psi.), adjustable, reasonably accurate flow-rates and most important, the facility to mount up to six pump-heads on to one drive, motor. We have described the application of such a pump, fitted with two pump heads (see Fig. 5.21) for a simple home-made gradient maker (Hinton and Dobrota 1969). To ensure that accurate flow-rates are maintained with this and other piston pumps, the outlet must at all times be positively pressurized. If this is not done, the solution can syphon through the pump since the zonal rotor is often at a lower level than the gradient reservoir on the bench. To prevent this, either a backpressure valve should be fitted in the outlet line or, place the outlet higher than the inlet. The ISCO Model 300 diaphragm pump should also be suitable. Although peristaltic pumps cannot deliver accurate flow-rates or pump against a high backpressure, they can nevertheless be quite good for zonal work. The most popular peristaltic pump for this application seems to be the Hiloflow (Metering Pumps Ltd.). Cheap centrifugal pumps are adequate for circulating coolant round the rotating seal and other ancillary equipment. A good example of such a pump is the Grants model P2. 5.1.4.3. Refractometers and density meters One reasonably good flow-through refractometer is no longer available. This was the Hilger Siihiwr bidex p. 287
184
DENSITY GRADIENT CENTRIFUGATION
& Watts model 550 which has been discontinued. Our own instrument has proved quite rugged over the last seven years of continual use (and misuse). The major problem with the instruments available today stems from the fact that they were primarily designed for use as detectors in chromatography. Consequently, sensitivity is extremely high (down to 16 x 10-8 RI) and the very narrow connections and the tiny flow-cell(50 pl or less) cannot take the high flowrates normally used in zonal work. Therefore, a difficult modification has to be made to reduce the sensitivity and only a part of the gradient to be monitored can be passed through the cell. The latter means that a stream splitter and a second (accurate) pump may be needed to sample only a part of the gradient. This portion of the gradient must either be fed back into the main line or be wasted, in which case its volume has to be known. A range of these instruments is available from Waters Ass., Atago Ins., Laboratory Data Control, Perkin-Elmer, Phoenix Precision Inst. Co., Varian, Winopal, etc. The Winopal refractometer has been used successfully for zonal work by Bachofen (1974). A number of density meters are available but they are all large industrial instruments not readily adaptable for small laboratory use. We will therefore not list them here. The instrument with the nearest specification to ideal seems to be the Anton-Paar DMA-10 density meter. However it has no recorder output but only a digital display suitable for single readings (although a whole zonal gradient can be pumped through the measuring cell). We believe that a new instrument with an anologue output, is now available but at present have no details. Such instruments measure density and are not affected by changes of gradients solute. The /3-ray absorption density meters described by Atherton et al. and Cope and Matthews (1973) are both affected by different solutes and therefore need to be calibrated for every class of solute. In practice this problem of changing scales is not serious as most zonal workers routinely use only one gradient material. Of the two above instruments the Cope & Matthews one appears simpler and therefore cheaper although neither are commer-
Ch. 5
ZONAL ROTORS
185
cially available. However, we believe that the Atherton instrument can be made to special order. Some while ago the idea of weighing a loop of tubing with a strain gauge whose voltage response was linear with density was investigated (see Dobrota and Reid 1971) but nothing has appeared as hardware. To sum up, although none of the instruments seem ideal we hope that the above information will prove useful for the interested user. 5.1.4.4. Spectrophotometers,colorimeters and UVmeters. The range of instruments suitable for locating the separated bands is so large that we could not possibly list them all. As a general point they must have provision for a good flow-through cell, if possible have a reasonable range of wavelengths and must have a recorder output, and also should be compact enough to fit onto a trolley. Manufacturers include Pye Unicam, Perkin-Elmer, Carey-Varian, Beckman, Gilford, Cecil, etc. We particularly like the compact Cecil 404 model and some of the Beckman instruments. The more sophisticated, and more expensive, instruments such as the Gilford can cope with very high absorbance readings. The recorder output of most colorimeters and some spectrophotometers will be linear with transmission and may therefore need converting to absorbance. This can be done with a suitable logarithmic converter. Chromatography UV monitors like the LKB Uvicord, Gilson, and Isco may also be used. While the Gilson, Isco (which reads in absorbance) and Uvicord I represent good value for money, the new Uvicord 111(although suitable) is as expensive as a spectrophotometer. Its purchase solely for monitoring zonal gradients could hardly be justified.
5.1.4.5. Flow-through cells. Beckman-RIIC market two flowthrough cells. Type BTF 5 has a variable light-path while the ‘multi-path flow cell’ has four fixed light-paths (see Fig. 5.22). The latter cell can only be fitted in a Beckman spectrophotometer (although the outside dimensions are the Same as a standard 10 mm cuvette). However, a cheap non-adjustable flow-through cell, such as Sehiecr nrdi,r p 287
186
DENSITY GRADIENT CENTRIFUGATION
Fig. 5.22. The Multipath flow cell. (By permission of Beckman-RIIC.)
the Hellma type 154, with a 10 mm lightpath, a hold-up volume of about 1 ml and fitting a standard cuvette holder is ideally suited for zonal work. Similar cells are also available from Scientific Supplies. 5.1.4.6. Thermometers. For measuring the in-line temperature of
gradients suitable thermocouple electrical thermometers can be obtained from Gallenkamp (Cat. No. TK-075)and Electroplan (Comark range). 5.1.4.7. Perfusor pumps. The Sage perfusor pumps (sold by A.R. Horwell) and the Braun perfusor (Shandon-Southern) can be useful for pulse-free injection of precious zonal samples and also for unloading swing-out tube gradients. 5.1.4.8. Stroboscopic lights. These can be extremely useful parti-
cularly with the transparent rotors. Firstly they are ideal for revealing particle aggregation if it should occur while to the eye the bands might have appeared as crisp well defined bands. Secondly they can be used to check the rev/min of any spinning rotor. Suppliers of strobes, amongst many include Dawe Instruments, Electronic Applications Ltd., Electroplan, General Radio, etc. 5.1.4.9. Recorders. In view of the need to monitor two parameters, a two channel (or even more) recorder is most suitable. The two pens
Ch. 5
Z O N A L ROTORS
187
should provide the trace on the same chart paper, and there must also be an event marker pen linked to the fraction collector. Good examples of such recorders are the Rikadenki Model 241 (TEM, Sales) and the two-channel Smiths Servoscribe. Preferably the recorder should have a fairly wide range of input, i.e. 5 mV to 5V (DC), and a reasonable adjustment on both zero and full span so that odd (e.g. 85 mV) inputs can be matched up to full-scale deflection on the recorder chart. The speed of pen response does not need to be fast since both gradient and absorbance traces are gentle curves. 5.1.4.10. Coolingequipment. Although a supply of ice should suffice, a permanent supply of ice-cold-water from a bath may be advantageous. A suitable refrigeration unit is available from Grants, model CC-15, and can supply enough cold water (with a suitable circulating pump) for the seal, gradient cooling coil and the fraction collector jacket. Most high-speed centrifuges need a supply of cooling water for the diffusion pump condenser and various bearings. For example a Super Speed 65 or 75 (MSE) requries water (2l/min) at not less than 20 psi (1.5 atm.). With some Beckman machines there is a maximum pressure which must not be exceeded otherwise a flood will result from a burst diaphragm in the pressure switch. If the continuous use of a cold water supply is too wasteful (in practise the tap tends to be left on all the time), a simple recirculating system can be constructed. This would need a tank, a car-type radiator, tubing and an electriccentrifugal pump (as in a washing machine) to circulate the water from the tank to centrifuge then to radiator and back to the tank.
5.1.4.1I . Minor accessories. Tubing suitable for zonal flow lines should be not less than about 3 mm bore since it could cause excessive pressure build up. Cheap PVC tubing such as No. 6H (Portex Ltd.) can be used for most of the lines but since it becomes rather rigid when cold, soft silicone rubber tubing (Bsco-Rubber Ltd.) is preferable for the connections to the feed-head. Simple plastic Siihpw inct~~x p. 287
188
DENSITY GRADIENT CENTRIFUGATION
connectors for joining up lengths of tubing can be bought from most laboratory suppliers. Tubing clamps, for securing tubing connections are obtainable from Uniclip Ltd. Plastic threeway taps, which are most useful for the two feed lines to the feed-head since they allow reversal of flow and easy connection of syringes, are supplied by Henley Medical Supplies Ltd. and Portex Ltd. (through agents). Artery forceps (also called Spencer-Wells,and obtainable from any surgical instrument supplier), are invaluable for rapid and effective clamping of tubing in case of a blown connection or ather emergencies. The gradient cooling coil can be made from stainless-steel tubing of approx. 3 mm bore. This can be supplied by the Oxford Instrument Co. 5.1.4.12 Fraction collectors. As with spectrophotometers, the range
of fraction collectors is wide and well known to biochemists who often have personal preferences. Apart from stating that compactness, cooling facility, a capacity for large tubes (up to 50 ml.) and the provision of an event marker are all important we leave the choice to the reader. The major problem of collecting fractions of equal volume (see 85.1.3.6.0 can be solved by using an Isco Volumeter. This device detects the level of the liquid in a burette-like vessel and then opens a solenoid-operated tap allowing the liquid to drain into the fraction collector tube. A useful modification to allow very rapid drainage into the fraction collector tube is to increase the bore of the Teflon delivery tap. If this is not done the volumes of the fractions can get progressively bigger, with the increase in gradient concentration. An alternative approach is to use a fraction collector of the Gilson Escargot type which detects the volume in the collector tube which is a molded uniform plastic tube. In view of the relatively high flow rates used in zonal work, some spillage may occur between tubes as the fraction collector changes.
Ch. 5
ZONAL ROTORS
189
Although this problem does not arise with the Volumeter, with other systems it may be serious enough to warrant the instalation of a flow-arrest device, such as that available for the LKB Ultrograd. 5.1.4.13. Integrator This accessory, which is electrically connected
to the centrifuge, calculates and displays the time intergral of the square of rotor velocity (w’t). Although expensive, it can be useful for exacting analytical work, especially since it can be preset to switch the centrifuge off at any selected value. It is available for the MSE. Beckman and Christ machines.
5.2 Reorienting zonal rotors When the tubes of a swing-out rotor move from a vertical to a horizontal position as the rotor accelerates, they mechanically move the tube contents into the new position dictated by the centrifugal field (see also 81.3.3). Thus the whole tube is reoriented, leaving the contents undisturbed. However, if a gradient is loaded into a zonal rotor at rest and the rotor is then accelerated, the gradient will in fact reorient from a vertical to a horizontal position (see Fig. 5.23). This will occur in any system, including angle tubes, and can be successful provided that the rotor shape is correct and accurately controlled rates of acceleration and deceleration can be achieved. Anderson et al. (1964) and Fisher, Cline and Anderson (1964) were among the first to report the successful reorientation of density gradients when they described the use of angle rotors for isopycnic banding of DNA and phage on CsCl gradients. Later, the first reorienting zonal rotor was described by Elrod, Patrick and Anderson (1969). Gradient profiles were checked before and after the spin and found to be almost identical, thus confirming that gradients in zonal rotors would, under the correct conditions, reorient successfully. The merits of this technique are that if the rotor can be loaded and unloaded successfully, no rotating seals, such as the dynamic Rulon seals, are needed. Also, loading and unloading need not be done in the centrifuge, they could be done on the bench or in the Suhlc~irfriCr p 287
190
DENSITY GRADIENT CENTRIFUGATION
Ch. 5
ZONAL ROTORS
191
Fig. 5.23. Sequence of events in operating an RK rotor, showing reorientation and continuous-flow banding. A density gradient is loaded into the rotor at rest (A). The gradient reorients vertically as the rotor is carefully accelerated, up to its operating speed (B). A sample is pumped in at the top. As it flows over the top of the gradient, particles sediment into the gradient and are ‘captured’. while the eMuent Suhjrcr index p
2x7
192
DENSITY G R A D I E N T CENTRIFUGATION
flows out through the bottom seal (C). Sample flow is stopped and the captured particles are allowed to equilibrate to their isopycnic point (D). The rotor is decelerated (E) and the gradient reorients to its original horizontal position (F). This process does not disturb the particle bands. The rotor is unloaded, at rest, by applying air or water pressure to the top of the rotor and using a pump to control the flow ( G ) .
cold room. Therefore the whole technique of zonal centrifugation could be enormously simplified. Anderson (1966a) reported that a small reorienting rotor of 100 ml capacity had been tested at 141,000 revs/min. Most of the problems are purely technical. Firstly a smooth and controlled rate of acceleration and deceleration is needed, especially in the low speed range, i.e. up to 500 or 1000 revs/min. This is most easily achieved when the centrifuge is driven by an air or oil turbine, as in the K, RK series of reorienting continuous-flow rotors. With electric-drive centrifuges we have encountered a slight problem (see below). Since the rotor is loaded and unloaded at rest and thus the gradient is not stabilised by centrifugal force as in the classical zonals, the channels for loading and unloading must be uniform and of precisely the same diameter. A further, but only theoretical objection, is the greater likelihood of droplet formation at rest before a sufficient g is reached to prevent it (see $2.1.2). Reorientation in zonal rotors is already in common use. Rather remarkably the rotors used (K, RK and J) are tall cylinders, which would appear to be quite unsuitable, and yet according to many workers notably Cline, there is apparently little loss in resolution. It is therefore evident that the technique can be made to work, but its usefulness for a wide range of separation problems (especially rate sedimentation) remains to be demonstrated. Reorienting rotors: Sorval SZ-14. This rotor was developed by Dr. P. Sheeler and has been commercially available for some time now. It can be operated in a number of Sorvall centrifuges, including the RC2-B in which it has the maximum operational speed of 19,500 (maximum g 40,500). The total capacity is 1,400 ml and the usable centrifugal path is 5.3 cm. The minimum radius of the rotor is 4.2 cm meaning that
Ch. 5
ZONAL ROTORS
193
a reasonably high g is available in the sample region thus making the rotor suitable for isopycnic work. We have no personal experience of operating this rotor, but have no reason to suspect any major difficulties. The rotor shape is shown in Fig. 5.24.
Fig. 5.24. Diagramofthe SZ-14 Sorvall reorientingrotor.Upper Fig. shows the sample application and, on the right half of the rotor, the separated bands. Lower Fig. shows the rotor at rest after reorientation ready to be unloaded.
The maximum g attainable would put this rotor in terms of its applicability somewhere between the slow speed A, HS and the B-XV zonals. Unfortunately it is still difficult to assess its absolute usefulness mainly because relatively few publications have quoted it, and of these the majority seems to originate from the team which designed it. They have demonstrated (Sheeler and Wells, 1971; Wells, Sheeler and Gross, 1972) that it can be used for separating mitochondria, lysosomes, nuclei, liver glycogen and even rough and smooth membrane vesicles. The one worry which we would have about this rotor is purely theoretical and not founded on any practical experience. This concerns Sirhirrr irrde.~p. 287
194
DENSITY GRADIENT CENTRIFUGATION
Fig. 5.25. Reorientation of sucrose gradients in a B-XV rotor. Graphs 1-5 show the profiles of step gradients, 300 ml water, 400 ml 0.5 M sucrose, 400 ml 1.5 M sucrose, approx. 560 ml of 2 M sucrose, monitored during loading (----) and during unloading -( ). In each case the gradient was left at about 3,000 rpm for 15 min.
Ch. 5
195
ZONAL ROTORS
the method of unloading. The rotor has a core and six septa which are provided with channels leading to the tapered bottom of the rotor. Having experienced sectorial unloading with an A-XI1 rotor (see Fig. 5.19) at approximately 500 g, we suspect that under 1 g there could be differential unloading through these six channels, thus resulting in a loss of resolution. K , RK series. These rotors have already been described in $5.1.2.8. They are all used as dynamiczonals or as true reorienting rotors. While the large K and RK are better suited for continuous-flow work, the J rotor can be applied in very much the same way as a B-XIV zonal batch rotor (Cline 1972, personal communication). B-XZV and XV as reorienting rotors. Beckman and MSE rotors can be supplied with diagonal septa channels for reorientation work. Klucis and Lett (1970) have used a B-XXV zonal rotor (a special version of a B-XV designed specifically for DNA work) in a reorienting mode; although they do not discuss the reorientation it must have been successful. In view of the scanty information available, we decided to test the feasibility of this method, using an aluminium B-XV, running in a SS65 (MSE) centrifuge, which was fitted with a slow accelerate/decelerate control unit. (When using this control, the motor brushes must be in perfect working order, otherwise acceleration and deceleration will be uneven.) These tests consisted of simply loading and unloading complex step gradients under various conditions (summarized in Fig. 5.25), and continually monitoring the gradient profiles on a recording refractometer. between loading and unloading. 1) Rotor used in normal zonal manner, gradient loaded and unloaded with the rotor spinning. 2) Gradient was loaded with rotor spinning. Rotor was then decelerated to rest, left for 15 min and accelerated to unloading speed, and unloaded. 3) Exactly as ( I ) except that the 300 ml of water was not pumped in as part of the gradient but was injected to the centre as with a normal sample. 4) Gradient was loaded with the rotor stationary. It was then accelerated and after 15 min, unloaded while spinning. 5 ) Gradient was loaded with rotor spinning. After deceleration it was unloaded at rest, by pumpingcushion to the bottom anddisplacingthegradient upwards. 6 )Alineargradientwasloadedwiththe rotor at rest. After acceleration it was unloaded normally with the rotor spinning. .Svhii,ir
in&.r
p. 287
196
DENSITY GRADIENT CENTRIFUGATION
From the results two distinct conclusions can be drawn. First, let us state categoricallythat gradients are reoriented successfully, indeed the shape of the gradient is in some cases preserved better than with dynamic loading. When a sharp step gradient is introduced dynamically and then unloaded, the first step becomes almost a smooth curve, while the step is much better defined if the identical gradient is loaded statically and then reoriented. The most likely reason for this is that when the layers of gradient enter the rotor, under dynamic loading, their surface areas at the wall periphery are so large (radial dilutions effect) that solute diffusion destroys the step. The surface area of the bottom of rotor - the chamber - is much smaller than that of the wall thus explaining the better preservation of profile in the case of static loading. However, the second conclusion is that there will always be some loss of resolution when a B-XIV or B-XV rotor is unloaded at rest. The reason is that the top of the rotor chamber is completely flat, and as the discrete layers of gradient are pumped up (or down), they are held up at the corners and mixed in with the successive layers. Were the top of the rotor slightly tapered much like a tube-unloading device, unloading could be without loss of resolution. In emergencies, when the rotor has stopped and cannot be restarted, smooth gradients may be recovered without disastrous mixing if displaced at a very slow rate (less than 10 ml/min).
CHAPTER 6
Assay of fractions separated by density gradient centrifugation
6.1. Enzyme and chemical assays on fractions An article of this size cannot include all the methods used in the assay of fractions separated by density gradient centrifugation. To assess a centrifugal separation, it is often necessary to assay markers for the various subcellular structures present in samples. We have discussed the choice of such markers elsewhere (Reid 1972; Hinton and Reid 1975). There are two general points which can usefully be made. Firstly, with the number of fractions normally obtained from each separation (20-60) it is hardly worthwhile to set up a separate AutoAnalyser channel for each assay. When chemical constituents such as protein are to be assayed, the fractions can be stored until material from several separations has accumulated, and it does then save time to set up an Auto-Analyser manifold. In some cases, such as enzymes releasing p-nitrophenol, a single AutoAnalyser manifold may be used for the assay of several enzymes. It is, however, tedious to have to perform several assays on up to 60 fractions without any analytical aids. Unstable components, especially enzymes such as glucose-6-phosphatase, are best assayed manually but using dispensors and repeating syringes as much as possible. However, care must be taken in using sampler diluters as some do not give reproducible results when used with fractions containing appreciable amounts of sucrose (Reid 1972~). We have discussed the Auto-Analyser because this is the only analyser which we ourselves have used. It would be interesting to know how useful modern discrete analysers (see Roodyn 1971) such 197
Siihjeil eiCr 1’.
287
198
DENSITY GRADIENT CENTRIFUGATION
as the Vickers D-300 would be for this application. While no single group could justify the purchase of such a machine, their speed and the ease with which they may be switched from one assay to another might make then suitable as a joint purchase by several research groups. Another analyser which would be of interest to a centrifugation laboratory, if only for its mode of operation, is the GeMSAEC analyser (Anderson 1969b). In this machine the reactions take place in a centrifuge rotor and are monitored continuously through glass windows inset in the rotor. The problem with this analyser is that the high density of some fractions separated from density gradients causes difficulty in mixing even at 1 g, and one would expect very severe problems to arise if, as in the GeMSAEC, one tried to mix sample and reagents during centrifugation. Another important point is that inert density gradient solutes such as sucrose may in high concentrations, interfere severely with the assay of enzymes and other tissue constituents. We consider such interference in more detail in Ch. 8.
6.2. Electron microscopic examination of fractions There is a tendency to underrate the usefulness of the light microscope in the examination of subcellular fractions. In fact, with particles the size of mitochondria or larger, light microscopy, especially with phase contrast, gives more information on the composition of fractions than electron microscopy. This is because one can rapidly inspect a representative sample of a fraction using a light microscope and recognise intact particles more easily than in the thin sections used in electron microscopy. However, particles smaller than mitochondria can only be resolved by the electron microscope. The great problem with electron microscopy of fractions separated by density gradient centrifugation is to obtain a representative sample. When the particles are pelleted from the separated fractions, the pellet will naturally tend to be non-uniform with the largest particles at the bottom of the tube. When the pellet is broken up during fixation and embedding, the orientation of the fragments will be lost.
Ch. 6
ASSAY OF SEPARATED FRACTIONS
199
In addition, fragments of the pellet will tend to lie flat at the base of the capsule and consequently the tendency will be to cut across a pellet rather than through its thickness. Any particular section may, therefore, appear homogeneous, in spite of considerable striation in the original pellet. There are a number of ways of minimising or circumventing this problem. When fractions are collected by pelleting one should, 1) dilute to give the thinnest possible pellets, 2) break up the pellet as little as possible during fixation and embedding, 3) cut across the thickness of the flakes, and 4) cut sections at several different depths in the block. Alternatively, one may avoid pelleting and separate the particles from the medium by filtration rather than by centrifugation (Baudhuin et al. 1967). This gives more even sampling than pelleting and makes it simpler to maintain the orientation of the specimen during fixation and embedding. Finally, one may abandon sectioning altogether and use negative staining methods (Whittaker et al. 1964). The disadvantage here is that the material must be examined immediately after separation and the images are more difficult to interpret than those obtained after thin sectioning. A second problem in electron microscopic examination of subcellular fractions is to identify the particles. For example, while it is simple to recognise intact mitochondria, damaged or fragmented mitochondria may be difficult and even impossible to recognise. The answer lies in specific cytochemical staining methods for enzyme activity, but there are many problems associated with their use although excellent results have been obtained by Leskes et al. (1971) and El-Aaser et al. (1973). Poor penetration of the substrate into jelly-like pellets can be avoided by resuspending the particles, doing the cytochemical reaction in free solution and pelleting after fixation with glutaraldehyde. The pellets can be stained with osmic acid and embedded in the usual way. Because the particles are collected by centrifugation after the cytochemical reaction, one must again be aware of striation.
200
DENSITY GRADIENT CENTRIFUGATION
6.3. Assessment of resultsfrom density gradient separations The presentation of the resultsof analytical separations usually involves the study of the distribution of a particular cell component - often an enzyme - among the separated fractions. The patterns can then be compared with those of the established ‘markers’ (see Reid 1972; Hinton and Reid 1975). The most usual (and convenient) method of presenting results is to plot the total activity (or amount) per fraction against fraction number, but if the fractions are not of equal size the activity (or amount) per unit volume should be used. As well as the distribution of each component across the gradient, one also wishes to gauge the reliability of the experiment. We find two control values especially useful. Firstly comparison of the enzyme activity of the homogenate with the activity obtained with a similar tissue in earlier experiments serves as a control on systematic errors in the assay methods. One rapidly learns to distinguish the normal variation between animals from the truly ‘odd’ result which requires further investigation. The specific activity of the homogenate (i.e. the activity per mg. protein) is usually more constant than the activity per gm. wet weight as it is less subject to errors arising from poor homogenisation and from adhesion of liquid to the surface of the tissue. The second essential quality control figure is the recovery, the sum of the activities in the fractions divided by the activity in the material loaded onto the gradient. The latter is estimated from a sample, taken before separation, stored at the temperature used in centrifugation and then assayed in parallel with gradient fractions. If the recovery is much more or much less than loo%, then one must exercise great caution in interpreting the results until an explanation for the activation or loss in activity is discovered. While presentation of the results of analytical separations simply as activity per fraction is usually sufficient, sometimes other methods of plotting will give better results. After separation by isopycnic banding one may wish to correct the results for the effect of nonlinear gradients. This may be done simply by dividing the activity in any fraction by the density difference across that fraction and plotting
Ch. 6
ASSAY OF SEPARATED FRACTIONS
201
the resultant figure (activity/unit density difference) against density. The result of this sort of transformation can be seen in Fig. 7.7. However, it is usually better, in such a case, to repeat the separation on a linear gradient. The plotting of specific activity (activity/mg protein) or of purification (the specific activity/specific activity of the same enzyme in the homogenate) as used in presentation of the results from differential pelleting (de Duve 1966) is dangerous. Thus the experiment shown in Fig. 6.1 would appear to show 5’-nucleotidase
. c
‘E
Fraction no.
Fig. 6.1. An illustration of the dangers of using the specific activity of an enzyme as a guide to its distribution. A rat liver post-nuclear fraction was fractionated by centrifugation in an HS zonal rotor as described by Burge and Hinton (1971). When the specific activity of 5’-nucleotidase is plotted (A) a ‘peak’ is shown in the central (lysosome-rich) part of the gradient. The actual distribution (B) shows that a negligible proportion of the 5’-nucleotidase is actually recovered in the lysosome-rich region. Upper part : 5‘-nucleotidase specific activity. Lower part: 0__ 0, 5’-nucleotidase: . . . . . ,acid P-glycerophosphatase; . . . . . , protein. ~
.
Slrh,w/ b v k . ~ p . 287
202
DENSITY GRADIENT CENTRIFUGATION
in the lysosomes when the specific activity is plotted. In fact, further fractionation gives no evidence for this and in any case the amount of 5’-nucleotidase activity in the lysosome-rich region is tiny. One must remember that de Duve’s method for presenting the results of differential pelleting experiments was developed specifically for the localisation of lysosomal enzymes, and while very useful for this purpose can be very misleading outside its proper context. In our experience, the most valuable alternative method for plotting results from separation experiments is to plot the activity in each fraction as a percentage of the total activity recovered from the gradient. This has the effect of reducing all the distributions to a common scale and makes it easier to distinguish small, but significant differences in distribution. Thus in the experiment illustrated in Fig. 6.2 it would be difficult to be sure that acid ribonuclease was differentlydistributed from acid phosphatase, from the plot of activity, but plotting the results as a percentage of total activity makes the difference clear. For preparative separations, reporting requirements are that results be given for markers of potential contaminants as well as for the particles being separated, and that the results for each marker should be presented, a) as a percentage of the homogenate activity recovered in the final preparaion (and preferably at intermediate steps) and b) as a purification (i.e. the ratio of the specific activity in the final fraction to the specific activity in the homogenate). These figures will permit direct comparison with the results of other workers. Specific activities are meaningless in the absence of data on the homogenate.
6.4. Calculation of sedimentation coefficients Although ‘simple’ methods have been proposed for the calculation of the sedimentation coefficients of particles fractionated by rate zonal centrifugation (McEwan 1967; Halsall and Schumaker 1969; Young 1972; Funding and Steensgard 1973), the authors doubt their value. Fortunately there are a number of well developed programs
Ch. 6
203
ASSAY OF SEPARATED FRACTIONS
uMolssImin E260/min
10.4
Fig. 6.2. An illustration of the advantages of plotting all results on a common scale. A mitochondria1 + lysosomal fraction from the livers of rats injected 3 days before death with Triton WR-1339 was fractionated by centrifugation in an HS zonal rotor as described by Burge and Hinton (1971). Acid B-glycerophosphatase (A-A) is concentrated in lysosomes which sediment slightly faster than those which contain acid ribonuclease (0----0). This difference is shown more clearly when activities are plotted as a “6 of the activity recovered from the gradient (B) than when absolute activities are plotted (A).
for these calculations, and if a computer is available, one can usually find a program which will run on that particular machine. It is, however, frequently not neccessary to use a computer program. For reasons that will be discussed later (0 8.3.2) we feel Suhjrcr index p . 287
204
DENSITY GRADIENT CENTRIFUGATION
that one cannot, with present centrifuge equipment, use density gradient centrifugation for the primary determination of sedimentation coefficients, for this purpose one must use an analytical ultracentrifuge. What can be done is to compare sedimentation coefficients, so that if the value for one component of a mixture is known, one may calculate the coefficients of other components. If isokinetic or, with zonal rotors, isovolumetric gradients (see 5 3.4) are used, one may carry out this calculation by linear interpolation. With the linear 0 . 5 1 .O M (1 5-3073 gradient which is commonly used with swingout rotors, even the errors introduced by linear extrapolation are not very large, providing that the particles whose sedimentation coeffcient is known are not too dissimilar in size to the ‘unknown’ particle. Much greater problems arise with zonal rotors because of radial dilution as well as non-linear sedimentation, so that it is usually impossible to use linear interpolation. However, one of the authors having dabbled with the problem, suspects that ‘simple’ methods for calculating sedimentation coefficient are usually more fun for the inventors than useful to other workers. The choice is really between proper computing of the sedimentation coefficient and obtaining a rough estimate by comparing the radial distance moved by the ‘unknown’ particle with the distance moved by a particle of known sedimentation coefficient. As the most frequent use of the sedimentation coefficient is as a label for particles which can be identified in no other way, a rough estimate is often quite sufficient. An accurate estimate of the sedimentation coefficient is only required when constructing mathematical models of density gradient separations.
CHAPTER I
Applications of density gradient centrifugation
We shall not attempt to give a complete guide to all the problems to which the technique of density gradient centrifugation may be applied, much less to list all the approaches. Rather we will try to indicate the major areas in which gradient centrifugation may be useful and to discuss briefly the approaches which may be most helpful and the problems which may arise. Detailed information on separation methods is scattered through the literature, but the monograph on zonal rotors edited by Anderson (1966a), Volumes 1, 3 and 4 of ‘Methodological Developments in Biochemistry’ (Reid 1971, 1973 and 1974) and the book edited by Birnie (1972) form useful entries to the literature together with Chevrenka and Elrod (1972). In all cases the actual articles are biased very strongly towards the use of zonal rotors, but the introductions and discussions contain references to methods used with conventional rotors.
7.1. Separation of living cells In general, biochemists have tended to pay lip service to the biological evidence of the tissues of higher organisms, but have not been able to separate the different cell types in sufficient quantity and purity to enable any extensive biochemical studies to be carried out. There are two major steps in the separation of the different types of cell which make up a tissue. Firstly, the individual cells must be separated from each other by a procedure sufficiently gentle for the separated cells to retain their function. Secondly, a method must be found to separate the different types of cell from each other. The first topic 205
.Sirhii,ir
nid‘*.rp 287
206
DENSITY GRADIENT CENTRIFUGATION
is outside the scope of this article and it must suffice to say that the majority of published methods rely on the perfusion of the tissue with an enzyme such as collagenase (Munthe-Kaas and Seglen 1974; Crane and Miller 1974; Glick et al. 1974), pronase (Romrell et al. 1975) or trypsin (Aspray et al. 1975) which will break down intercellular connective tissue. A chelating agent such as sodium citrate or EDTA is sometimes included to facilitate disruption of the tight junctions. The most popular method for separating the released cells, and for the separation of free living and blood cells, is density gradient sedimentation either in a centrifugeor at unit gravity (Harwood 1974). Whole cells may be fractionated either by rate sedimentation or by isopycnic banding. It is hardly necessary to centrifuge when separating by rate sedimentation as cells sediment significantly at 1 g (Miller and Phillips 1969; Denman and Pelton 1973), Cells may also be fractionated by isopycnic banding, but one must take great care when using centrifugal fields greater then 15,000 g as cells may actually be pulled apart under their own weight (Mateyko and Kopac 1963). However, the greatest problem in the separation of living cells is the selection of the density gradient medium (see Table 3.1). As mentioned in 5 2.3, only density gradient solutes of a fairly high molecular weight are suitable for the separation of whole cells. Probably the most widely used compounds have been Ficoll and bovine serum albumin, their use is discussed in some detail by Harwood (1974). Both compounds are well suited for the formation of the supporting gradients used in rate sedimentation, for their high molecular weight will minimise instabilities in the sample zone (see 5 2.1.2). They are, however, less well suited for separations by isopycnic banding; the high viscosity of concentrated serum albumin solutions means that prolonged centrifugation is needed for the cells to reach their isopycnic banding densities, while the tonicity of Ficoll solutions rises sharply, and in a non-linear fashion, at high concentrations (Bach and Brashler 1970). Hence there has been some interest in novel density gradient media. Ludox (colloidal silica) has been used for cell separations (Pertoft and Laurent, 1969) but was found to damage spermatozoa, causing gelation and reducing motility (Bene-
Ch. 7
207
APPLICATIONS
dict et al. 1967). Dextrans have also been employed (Harwood, 1974) but may be expected to give the same problems as Ficoll. The most interesting group of compounds is the X-ray contrast media whose properties are described in +henext paragraph. The requirements for an X-ray contrast medium, low toxicity coupled with the capacity to form solutions of a high density, are very similar to the requirements for a density gradient solute. The commonest X-ray contrast media have all been used in density gradient centrifugation (see 9 3.3.4 & Table 3.1) although Metrizamide has been recently studied Litensively. The other media, notably Renografin (a mixture of sodium and methylglucamine diatrizoate), have been used for some time for the separation of microbial and blood cells (Hadden and Nester 1968. Tamir and Gilvarg 1966; see also refs in Hinton and Mullock 10%) Metrizamide would seem, on balance, to be the most suitable of the X-ray contrast media for use in density gradient centrifugation (Hinton and Mullock 1976), although none of them appear to cause serious damage to living cells. No further details will be given here, as separations in gradients of iodinated media have been covered extensively in articles in Rickwood (1976). All these compounds are expensive, so that financial considerations may prohibit their use in zonal rotors. To summarise, one may fractionate cells by rate o r isopycnic centrifugation. In the former case, bovine serum albumin or Ficoll are the media of choice. These media may also be used for isopycnic banding, but iodinated media such as Metrizamide are probably more suitable, as used by Munthe-Kaas and Seglen (1974) in the separation of liver cells (Fig. 7.1). Where neither simple rate separation nor isopycnic banding can separate the mixture, one can use the two techniques sequentially. Alternatively one could experiment with steep gradients approaching the banding densities of the cells as used by Boone et al. (1968); in this way it is possible to exploit differences both in density and in size to optimise separations between two populations of cells. In such cases computer simulation of the separation, as used by Boone et al. (1968), may be a great help in designing the gradient and deciding on the optimal centrifugation time. Sllh,'
p 287
208
DENSITY GRADIENT CENTRIFUGATION
-1.17
;;
- O N
- 1.15 -
-5 0
-1.13
-
-1.11
5
- 1.09 > t - I n
-
1.07
- 1.05 BOTTOM
FRACTION NO.
TOP
0
Y
Fig. 7.1. Distribution of rat liver cells in a Metrizamide density gradient. Freshlyprepared rat liver cells were incorporated into an 8 ml Metrizdmide gradient (density 1.05-1 .I6 g/cm’) and centrifuged for 20 min at 5000 revs/min at 4 in the tubes of a Beckman SW 40 rotor. A) Osmolarity; B) Density ( X----x ) ; parenchymal cells (0-0); non-parenchymal cells (0-0): Note 10-fold scale difference (from Munthe-Kaas and Seglen 1974).
7.2. Separation of cell organelles from mammalian tissues Because of the severe limitations on the amount of material which can be loaded onto a density gradient (see 9: 2.1.2), one does not normally attempt to separate an entire tissue homogenate by density gradient centrifugation. Rather the homogenate is first fractionated by differential pelleting. The fraction so obtained may then be further separated by rate o r isopycnic zonal centrifugation. The initial fractionation has the advantages both of simplifying the mixture and, in preparative experiments, of reducing the amount of contaminants so enabling more material to be processed in a single centrifugation.
Ch. I
APPLICATIONS
209
In spite of the comments in the previous paragraph one may sometimes wish to fractionate a whole homogenate, especially when dealing with a tissue in which the size of the various particles is not well established. An A-XI1 zonal rotor is probably the most suitable instrument for this purpose. The gradient we have tested is constructed in two parts, a fairly shallow initial section occupying about 2/3 of the diameter of the rotor followed by a steep final section (Hinton 1972). (Earlier descriptions of this gradient (Hinton et al. 1970 and 1971) contain errors.) Mitochondria, lysosomes and microsomes separate in the early part of the gradient on the basis of their sedimentation rate; in the latter part of the gradient, nuclei, fragments of connective tissue and red blood cells and plasma membrane sheets (which all sediment so fast that, if the separation were purely on the basis of sedimentation rate, would have pelleted on the wall of the rotor before the smaller organelles have left the starting zone) are separated on the basis of their isopycnic banding densities (Fig. 7.2). Such a separation may be useful as a first step in examining the distribution of some enzymic or chemical components among subcellular particles. If one is used to handling the A-XI1 rotor, the fractionation is much quicker than working through the scheme for differential pelleting and, except in the case of microsomes and cytosol, more information on the distribution of the unknown component is obtained. Generally, however, density gradient centrifugation is used for the further fractionation of organelle fractions separated by differential pelleting. 7.2.1. Subfractionation of a crude nuclear fraction and the separation of large sheets of a plasma membrane The crude nuclear fraction (4000 g/min pellet) separated from the liver homogenate by classical differential pelleting contains, in the case of liver, not only nuclei but sheets of plasma membrane, partially broken cells, aggregated material and large amounts of smaller organelles trapped in the pellet. These various structures are difficult to separate from each other by differential pelleting but may be separated successfully with an A-XI1 zonal rotor using a similar Siihieil s d r \ y . 287
210
200-
-.C 0 c 0 0
t
-2
100.
.-al C
c
2
a.
Fraction No. Fig. 7.2. Separation of particles from a rat liver homogenate in an A-XI1 zonal rotor. The sample comprised 20 ml of a homogenate (equivalent to 4 g of liver) prepared in 0.25 M sucrose, 5 mM NdHC03 pH 7.5 by 3 strokes of a PotterElvehjem homogeniser. The rotor was loaded with a complex sucrose gradient (Hinton 1972). The sample was injected and overlaid with 50 ml of 0.08 M sucrose. During acceleration, a further 70 ml of overlay solution was taken into the rotor. Centrifugation for 60 min at 3,700 revslmin at 4 (from Hinton 1972).
gradient to that used in the fractionation of the whole homogenate (Hinton et al. 1971; Hinton 1972. nb The molarity of the sucrose solution 'a' is wrongly printed as 1.164 in the first reference, the correct figure is 1.25 M. The density given for this solution and all other figures are correct. We would, however, now strongly recommend that as suggested in the second reference 0.3 M sucrose be used as the starting solution in place of the 0.25 M sucrose specified in the first reference). Apart from the nuclei themselves, the most interesting components of the crude nuclear fraction are the sheets of plasma membrane. If these are to be purified, most of the red blood cells must be removed before the crude nuclear fraction is prepared. This may be done either by perfusing the tissue prior to homogenisation or by homogenising
Ch. 7
APPLICATIONS
21 1
in a hypotonic medium. If a significant number of intact red blood cells are present in the 4000 g/min pellet, they will aggregate with plasma membrane sheets (Hinton 1972). Similar methods have been used for separating kidney plasma membranes (Price et al. 1972) and could, no doubt, be used for the separation of membranes from many other types of tissue where the intercellular links are strong enough to maintain the plasma membranes as large sheets. These methods cannot be used for the separation of plasma membranes from tissues which have weak intercellular links as in this case the cells are torn apart during homogenisation resulting in plasma membrane fragments which are too small to be separated from mitochondria by rate sedimentation (Hinton 1972). This is the case with many transplantable hepatomas (Prosper0 and Hinton 1973). The single step procedure discussed above depends on the use of an A-XI1 zonal rotor. If this is not available, a two-stage procedure must be used to separate plasma membrane sheets. Firstly microsomes are removed either by repeated washing or by sedimentation in a B-type zonal rotor. Simple washing is an effective procedure here as there is a very large difference in size between plasma membrane sheets and microsomes (see 9 3.1). Plasma membrane sheets are then separated from mitochondria, nuclei and other large components of the nuclear fraction by isopycnic flotation. These procedures were originated by Neville (1960) and Emmelot et al. (1964) and have since been applied to a large number of other tissues (Hinton 1972; de Pierre and Karnovsky 1973). 7.2.1.1. Purified nuclei As nuclei tend to aggregate on pelleting in 0.25 M sucrose, nuclei for use in subfractionation studies must be separated directly from the homogenate by pelleting through 2.2 M sucrose. Purified nuclei may then be fractionated on the basis of their size by rate sedimentation in an A-XI1 zonal rotor (Johnstone et al. 1968; Johnstone and Mathias 1972). The number of subfractions will depend on the tissue and the age and species of animal used. Fractions corresponding to diploid, tetraploid, octaploid and hexadecaploid nuclei can be separated from the livers of old mice (Fig. 7.3) .7lll>/'?I ,,vk.\ p 287
212
DENSITY GRADIENT C'ENTRIFUCiATION
To P
Effluent vol (mt)
Bottom
Fig.7.3. Separation of liver nuclei from mice of the NIH strain. Nuclei were purified by pelleting through 2.2 M sucrose. The purified nuclei were resuspended and layered over a 20-50% w/w sucrose gradient in an A-XI1 zonal rotor. The gradient contained 1 mM Mg Cll and was adjusted to pH 7.4 with NaHCO,. Centrifugation was for 1 h at 600 revs/min. The sharp peak at 200 ml shows the position of the sample. Further zones, from left to right, show diploid, tetraploid, octaploid and hexadecaploid nuclei (from Johnston and Mathias 1972).
and the results are an elegant illustration of the resolution which can be obtained with rate-zonal centrifugation. With less homogeneous tissues than liver, more complex patterns will be obtained as the nuclei will differ in size and shape (Austoker et al. 1972). Such methods may prove most valuable in exploring the mechanism by which different parts of the genome become activated during morphogenesis. 7.2.2. Subfractionation of the mitochondrial fraction The mitochondrial fraction separated from liver by differential pelleting contains, in addition to mitochondria, a variety of other subcellular components. Important among these are large sheets of endoplasmic reticulum such as surround mitochondria in many living cells and adhere to mitochondria after cell breakage unless chelating agents are included in the homogenisation medium (Chappel and Hansford 1972), lysosomes, peroxisomes and, if the tissue was homogenised gently, the Golgi apparatus. Some fragments of the plasma membrane may also be present, especially in homogenates of tissues
Ch. 7
213
APPLICATIONS
where intercellular bonds are weak. Secretion vacuoles may also be present (see Fleischer and Packer, 1974). Subfractionation of the mitochondrial components, or examination of the possible heterogeneity among mitochondria themselves or separation of fragments of broken mitochondria. Fig. 3.1 indicates that of the other cell components that are similar in size to mitochondria, and hence sediment in the mitochondrial fraction, lysosomes and peroxisomes are best separated by rate sedimentation, for their density is similar to that of rough endoplasmic reticulum fragments; fragments of the Golgi apparatus and of the plasma membrane, being heterogeneous in size, are best separated by isopycnic banding. Fragments of the Golgi apparatus band at a density of about 1.12 and may be readily separated from other components of the mitochondrial fraction (Morre et al. 1970, 1972, 1974). Plasma membrane fragments band at a density of 1.16 (Prospero and Hinton 1972) as compared with a density of 1.18 for undamaged mitochondria (Beaufay et al. 1964) and cannot be completely separated unless the density of the latter is increased by the addition of 2 mM CaCl, to the homogenisation medium (Prospero andHinton 1972).Thechoiceof rotor and gradient in such experiments will depend on the objectives (see § 3.5). If this is simply to isolate a single structure, then a step gradient in a swing-out rotor may be most convenient (Morre et al. 1972). Although mitochondria are remarkably uniform in density, they are heterogeneous in their sedimentation coefficients and can be subfractionated by rate sedimentation. Equally good results are achieved with the dynamically loaded A-XI1 zonal rotor (Swick et al. 1967), the reorienting SZ - 14 zonal rotor (Wilson and Casciato 1972) or with a swing-out rotor (Storrie and Attardi 1973; see Fig. 7.4). The choice depends on the amount of material to be fractionated. There have been reports that, contrary to what is stated above, mitochondria can be divided into two subfractions by isopycnic banding (Pollak and Nunn 1970); these results can, however, probably be explained by the greater susceptibility of some mitochondria to high hydrostatic pressure (Wattiaux 1974). Siih,ccr
iiidc,T
p 287
214
DENSITY GRADIENT CENTRIFUGATION
r
1
I
Fraction No. Fig. 7.4. Subfractionation of Hela mit;chondria. A crude mitochondria1 fraction, derived from a homopenate of 2-4 x 10 cells labelled for 20 min with S'H-uridine, was layered on a linear density gradient extending from @-200/, dextran and containing 0.48 M sucrose and 0.1 mM Tris pH 6.7 (25 ) in the tube of a SW-27 swing-out rotor (Beckman-Spinco). A cushion of 1.7 M sucrose, 0.01 M Tris pH 6.7 was placed at the bottom of the tube. Centrifugation was for 2 hr at 7,000 revs/min. Note the separation of mitochondria synthesising RNA and the mitochondria containing malate dehydrogenase (MDM) activity from those containing cytochrome oxidase (From Storrie and Attardi, 1973) N.B. In this figure the direction of sedimentation is from right to left.
Mitochondria are extremely complex structures, surrounded by two distinct membranes and possessing distinctive DNA and ribosomes. The mitochondrial inner and outer membranes have been found to have markedly different densities and to be readily separable by isopycnic banding on sucrose gradients (Sottocasa et al. 1967; Schaitman et al. 1967), while the inner membrane may itself be split into fragments of differing density (Werner and Neupert 1972). Methods for separating mitochondrial DNA and ribosomes are mentioned in Ch. 3 and cj 7.2.5. 7.2.3. Subfractionation of the lysosomal fraction Purification of lysosomes is extremely difficult (Beaufay 1969 ; Reid 1972) because of the inherent inefficiency of differential pelleting
Ch. 7
215
APPLICATIONS
(9 1.2.2) the 'classical' L or lysosomal fraction is neither pure nor, necessarily, representative. We feel that it is better to start with an M + L fraction prepared by centrifuging the post-nuclear supernatant for about 150,000 g/min. Lysosomes may be separated from such a fraction by rate zonal sedimentation (see Fig. 3.10); either a zonal o r a swing-out rotor may be used (Fig. 3.6). The lysosomal fraction prepared by rate sedimentation will be free from mitochondria but will contain microbodies, which are of the same size as lysosomes (see Fig. 3.1.), a few plasma membrane fragments and large densely-coated fragments of rough endoplasmic reticulum which tend to co-sediment with lysosomes (M. Dobrota and J.T.R. Fitzsimons, unpublished experiments). A partial separation of these components from the lysosomes may be achieved by isopycnic banding (Fig. 7.5). This is most conveniently carried out by loading the whole fraction separated by rate sedimentation onto a short isopycnic gradient (Dobrota and Hinton 1974). This procedure circumvents problems connected with the fragility of lysosomes on pelleting and resuspension. Lysosomes separated in this way are still contaminated with endoplasmic reticulum vesicles. These cannot be removed by centrifugal methods and if very pure lysosomes are required they must be separated by free-flow electrophoresis (Stahn et al. 1970; Henning et al. 1974). Lysosomes are an extremely heterogenous class of organelle. In addition to the distinction between primary lysosomes originating from the Golgi apparatus and secondary lysosomes, formed by fusion of primary lysosomes with phagosomes or autophagic vacuoles, some types of cell, notably circulating phagocytes, manufacture more than one type of primary lysosome. Secondary lysosomes may readily be separated from tissue homogenates after the injection of compounds suchas Triton WR-1339 (Wattiaux et al. 1963), dextran, iron-sorbitolcitrate complex (Arborgh et al. 1973) o r colloidal gold (Henning and Plattner 1974). These componds are taken up into lysosomes and modify their density so that they may be separated from other cell components by isopycnic banding (Beaufay 1969; Reid 1972). It is much more difficult to separate the different types of granules S,,h,'YI ,,,,I',\
p 287
216
DENSITY GRADIENT CENTRIFUGATION
FRACTION No
FRACTION No
Fig. 7.5.Isopycnic banding of lysosomes. A B-XIV zonal rotor was filled with 400 ml of sucrose gradient. The sample was 150 ml of the lysosome-rich region separated by centrifugation in an HS zonal rotor. Centrifugation for 150 min at 47,000 revs/ min. ---, density. Upper diagram, 0-0, acid phosphatase; A-A, acid acid phosphodiesterase; +p galactosidase. ribonuclease; v-v, Lower diagram, __ , protein; -0, 5’-nucleotidase; L O , glucose-6phosphatase; xx catalase; and, to facilitate comparison, 0-0, acid phosphatase.
+,
present in normal liver. No separation is achieved on isopycnic banding, but some separation can be obtained by rate sedimentation (Rahman et al. 1967; Burge and Hinton 1971). Canonico and Bird
Ch. 7
APPLICATIONS
217
(1970) achieved a similar fractionation of lysosomes from muscle; they interpret their results in terms of the separation of lysosomes from different cell types. As mentioned earlier, circulating phagocytes may possess more than one type of granule containing hydrolytic enzymes ; these may be separated by rate sedimentation (Baggiolini et al. 1969). Sometimes additional resolution may be achieved by isopycnic banding of the fractions separated by rate sedimentation (Baggiolini et al. 1970). Attempts have been made to separate the membranes of Triton WR- 1339-loaded secondary lysosomes (Thines-Sempoux 1973) and of ‘normal’ lysosomes from rat liver (Dobrota and Hinton 1974) by isopycnic banding after the membranes have been ruptured by vigorous homogenisation. However it is not clear whether the fractions so separated contain only the lysosomal membrane or whether they also contain the postulated lysosomal matrix. Similar comments apply to membranes prepared from lysosomes separated by free flow electrophoresis (Henning et al. 1974) although here the risk of contamination of the preparations by membranes deriving from other subcellular structures is much reduced.
7.2.4. Subfract ionat ion of microsomes The microsomal fraction is highly heterogeneous. The term rnicrosomes simply indicates the small particulate fraction which will not pellet at 150.000 g/min but is pelleted at about lo7 g/min. The subcellular source of these fragments depends on the tissue. In the case of liver, fragments of the rough and smooth endoplasmic reticulum are the major components, but fragments of the plasma membrane and of the Golgi apparatus are also present (Fleischer and Fleischer, 1970; Hinton 1972) as well as outer membrane fragments from broken mitochondria (Amar-Costesec 1974) and small lysosomes. As normally prepared, the microsomal fraction will also contain the free polysomes and glycogen particles together with a proportion of free ribosome subunits and of ferritin particles. The proportion but not the number of the various constituents will vary between different cell types. Sirhlrcr nidcr 11 287
218
DENSITY GRADIENT CENTRIFUGATION
Clearly the fractionation of such a complex mixture is not simple. The dense ribonucleoprotein particles, glycogen and ferritin granules may be separated by centrifugation through a layer of 2 M sucrose, although care must be taken to avoid the small particles tangling in the layer of membrane which forms over the 2 M sucrose. This may be avoided by using liver from fasted animals, by fractionating very dilute suspensions (Cow et al. 1970) or by recentrifuging the microsomes in 2 M sucrose in which case the membranes and the small particles will move in opposite directions. Glycogen granules can be fractionated according to their size by rate sedimentation on sucrose gradients and then separated from co-sedimenting contaminants by banding on CsCl gradients (Barber et al. 1966). However, it is not possible to fractionate the membrane-bound vesicles which make up the greater proportion of the microsomal fraction by rate sedimentation. A better separation can be achieved by isopycnic banding in sucrose gradients of a microsomal fraction prepared by differential centrifugation in either the tubes of swing-out rotors (see Tata 1972), highspeed zonal rotors (see Norris et al. 1974) or the special rotors developed fur isopycnic banding by Beaufay (Amar-Costesec et al. 1974). Similar results are achieved in all cases. Marker enzymes for the varioud membranous components of the microsomal fraction show very broad and overlapping distributions (Fig. 7.6).* One should be very careful about using non-linear gradients for this type of work as apparent separations may be achieved which are, in fact, gradient artefacts (Ch. 8). Thus in the experiment illustrated in Fig. 7.7, mathematical analysis shows that the apparent separation achieved on the step gradient is, in fact, exactly the same as the separation achieved on the linear gradient. Better separations between a proportion of the plasma membrane fragments and the endoplasmic
*
In the experiment illustrated in Fig. 7.6 the sample was applied at the dense end of the gradient and the particles separated by isopycnic flotation. A very similar pattern is obtained on isopycnic sedimentation although the degree of separation between the different types of particle is somewhat less.
Ch. 7
219
APPLICATIONS
1.2 I I I I
I
6 m
->
'2 0" 4.
1.1
20
10
30
1.0
Fraction no.
30 // / I / - ] 1.2 / /
'OI >
4-
>
P
2
15
4-
c
0 Fraction no.
Fig. 7.6. Subfractionation of microsomes by isopycnic flotation. A microsomal fraction. prepared from a rat liver homogenate by differential pelleting, was resuspended in I .8 M sucrose and loaded under a sucrose gradient in a B-XIV zonal rotor. Centrifugation was for 150 min at 47,000 revsimin. The gradient contained 5 mM Tris pH 7.4. (from Hinton et al. 1971). Sl,l,jl,'~lesk*\ S,,I~,?'l bsk.\. 1'. 287
220
DENSITY GRADIENT CENTRIFUGATION
Fig. 7.7. An illustration of the problems of interpreting separations on non-linear gradients. A microsomal fraction from rat liver was centrifuged for 3 hr at 40,000 revs/min in a B-IV zonal rotor in a gradient containing 5 mM MgC12 using (A) a linear gradient and (B) a gradient containing a flat step between two linear portions. and glucose-6-phosphatase (0-.-.-.0) In B zones rich in 5’-nucleotidase (0-0) appear to be clearly separated. However when the results from B are processed to show the density distribution of the fragments (C), the separation is nearly identical to that obtained in experiment A (from Prosper0 1973).
reticulum fragments is achieved when isopycnic flotation is used in place of isopycnic sedimentation, but the plasma membrane fragments which sediment with the microsomal fraction are themselves heterogeneous (Norris et al. 1974 ). With microsomes from Ehrlich ascites cells, much better results are achieved by isopycnic banding on Ficoll gradients than are achieved with sucrose gradients (Wallach 1967). No systematic comparison appears to have been carried out with liver microsomes,
Ch. 7
22 1
APPLICATIONS
but brief reports that plasma membrane and endoplasmic reticulum fragments form sharp bands after isopycnic banding on Ficoll gradients (House and Weidemann 1970) as against the rather broad (Hinton et al. 1971;Norris et al. 1974) bands obtained after separating in continuous sucrose gradients suggests that Ficoll may have advantages for the isopycnic banding of liver microsomes. Other high molecular weight solutes may be equally useful, as demonstrated by Graham (1972, 1973) who used Dextran gradients for the isopycnic separation of enzymically distinct membrane fragments from cultured kidney cells. While the different particles which make up the microsomal fraction can be partially separated by isopycnic banding in sucrose or Ficoll gradients, the fractions separated are still very heterogeneous. When the entire post-lysosomal supernatant is to be applied to a density gradient, the separation of endoplasmic reticulum fragments from other membraneous particles may be improved by the inclusion of Mg2+ions in the density gradient (El-Aaser et al. 1966). The degree of separation is not altered by changing the Mg concentration at least at concentrations greater than 1 mM (Hinton et al. 1967). The effect would appear to be due to attachment of free ribosomes to 'smooth' endoplasmic reticulum membranes, for the addition of Mg2+ ions to a suspension of microsomes has little effect on the density of the particles. The addition of Pb2+ions, in very low concentrations does, however, cause a large increase in density of endoplasmic reticulum fragments, and, providing that the aggregates which form are dispersed by sonication, does permit the separation of plasma membrane fragments from endoplasmic reticulum-derived vesicles (Hinton et al. 1971). This potentially powerful method for isolating one particular component of the microsomal fraction by using cytochemical methods to bind heavy metal ions to one type of membrane (Leskes et al. 1971) uses the resultant increase in density of that particular component to obtain its separation by isopycnic banding. Care must be taken to remove the non-specifically bound lead, if specific separations are to be obtained (Hinton et al. 1971). Enzymes do not seem to be +
Stihwr rrr&\
p 287
222
DENSITY GRADIENT CENTRIFUGATION
distributed evenly over the surface of microsomal fragments derived from the endoplasmic reticulum. Both rough and smooth microsomes can be broken into smaller fragments by sonication. These fragments may be separated by rate sedimentation on sucrose gradients in the tubes of swing-out rotors. Enzymes associated with the oxidation of NADH' show a different distribution to those connected with the oxidation of NADPH (Dallner et al. 1972). This suggests some mosaicism in the membrane.
7.2.5. Fractionation of' r ibonucleoprotein particles The fractionations of polysomes and of ribosome subunits by rate sedimentation on sucrose gradients were among the earliest applications of the technique (McQuillen et al. 1959). Separations may be carried out either on swing-out (No11 1969) or zonal rotors (Birnie 1972). The separation of polysomes, in particular, is one of the best tests for technique in density gradient centrifugation, as polysomes form a nicely graded series in which each particle is slightly more similar in size to the next largest than it is to the next smallest. Thus, trimers (154 S) are 25% larger than dimers (123 S) but tetramers (183 S) are only 18% larger than trimers and so on (Puderer et al. 1965). Polysomes up to the 12-mer have been resolved as separate peaks (Norman, 1971) but most workers are satisfied if they can separate up to the 7-mer or 8-mer. Density gradient centrifugation can also be used to reveal slight differences between newly formed small ribosome subunits and ribosome subunits recycled from polysomes and to fractionate ribonucleoprotein particles extracted from the nucleoplasm (Lukanidin et al. 1972) or the nucleolus (Prestayko et al. 1972). Ribosome subunits may also be fractionated according to their density by isopycnic banding in CsCl gradients (Perry and Kelley 1966), provided that they have previously been fixed with formaldehyde. This technique is useful for exploring heterogeneity among the subunits and in separating mRNA-containing particles from dissociated polysomes (Henshaw 1968) (see Fig. 7.8) or from the other small ribonucleoprotein particles sedimenting free in the cytoplasm
Ch. 7
223
APPLICATIONS
300
Tdbe number
Fig. 7.8. CsCl equilibrium gradient centrifugation of rat liver ribosome subunits. Rat liver polyribosomes were treated with EDTA and fixed with 6:; formaldehyde. Separation was by centrifugation for 64 hr at 40,000 revs !I an a Spinco SW 50 swing-out rotor. The peak at density 1.46 represents mRNA-containing particles. The peaks at about I .52 and I .59 are due respectively to small and large ribosomal x , AZbOnn1; 0-----C. RNA labelled for 40 min in vivo with subunits. x ''C-orotic acid: 00 . RNA labelled for 44 hr in vivo with 'H-orotic acid (from Henshaw 1968). ~
(Henshaw et al. 1967; Ayuso-Parila et al. 1973). One must, however, take great care in interpreting results obtained from ribonucleoprotein particle preparations contaminated by any significant amount of cytoplasmic protein (Fig. 7.9). A considerable amount of protein may be bound during fixation resulting in a much lower banding density than would have been obtained with the pure particles. Furthermore, recent experiments (McConkey 1974) have cast some doubt on whether there is the strict relationship, hitherto assumed, between the density of ribonucleoprotein particles in CsCl and their chemical composition. Unfixed ribonucleoprotein particles may be banded on Metrizamide SuhpI
lltde\
p 287
224
DENSITY GRADIENT CENTRIFUGATION
A200 nm
0.Sr
Fraction no. A200 nm
I
Fraction no.
Fig. 7.9. Artefacts introduced by the presence of high concentrations of extraneous proteins during fixation. Polysomes were separated by the method of Leitin and Lerman (1970) and fixed with 6% formaldehyde buffered according to the method of Perry and Kelley (1968) either a) alone or b) in the presence of about 1 mg/ml of rat liver cytosol protein. Centrifugation was for 18 hr at 50,000 revs/min in an MSE 3 x 6.5 ml swing-out rotor using preformed CsCl gradients. (B.M.Mullock and R.H. Hinton, unpublished experiment).
gradients (Mullock and Hinton 1973; Hinton et al. 1974b), and Buckingham and Gros succeeded in separating mRNA-containing particles. Ribonucleoprotein particles have lower densities in Metriz-
Ch. 7
APPLICATIONS
225
amide solutions (Hinton et al. 1974) than in either sucrose (Petermann 1964) or CsCl gradients (Perry and Kelley 1965). This could be due to variation in the degree of hydration of the RNA and protein molecules (Hinton et al. 1974). A similar dependence of apparent density with medium is observed with chromatin (see Chs. 2 8z 3). 7.2.6. Fractionation of chromatin Probably the greatest gap in our knowledge of cells is our lack of understanding of the processes which control the development of the tissues of higher organisms. These changes may be largely expressed by means of synthesis of new messenger RNA to code for the novel proteins needed for the development of specialised tissue. Comparison of the structure of active and inactive regions of chromatin should shed some light on the processes by which the many segments of DNA needed to code for the tissue specific proteins may be activated. There has, therefore, been a rapid increase in interest in methods for fractionating intact Aromatin. Most centrifugal methods for preparing or fractionating chromatin depend on isopycnic banding, although Bhorjee and Pederson (1973) have used rate edimentation through concentrated sucrose gradients for separating chromatin from the smaller ribonucleoprotein particles. Chromatin is too dense to band in sucrose gradients and is dissociated by CsCl (MacGillivray et al. 1972). Unfixed chromatin may be banded in sucrose-glucose (Raynaud and Ohlenbusch 1972) or in chloral hydrate gradients (Hossainy et al. 1973). However, the most suitable medium for banding chromatin appears to be Metrizamide, in which separation of chromatin from other nuclear constituents (Birnie et al. 1973a) and subfractionation of sheared chromatin (Rickwood et al. 1974) have been demonstrated. Chromatin isolated by such methods contains RNA in addition to DNA and protein. A simple method for separating these three classes of macromolecule has been devised by Monahan and Hull (1974). The chromatin is treated with the detergent sarcosyl (sodium dodecyl sarcosinate) and centrifuged to equilibrium on a gradient of Cs,SO, containing sarcosyl and dimethylsulphoxide. Siihiecl a d c r p 287
226
DENSITY GRADIENT CENTRIFUGATION
7.3. Separation of subcellular structures from plant cells The presence of a tough cell wall makes difficult the breakage of plant cells without extensive damage to intracellular structures. This has meant that much less work has been carried out on the separation of plant subcellular structures than is the case with mammalian tissues. In the following paragraphs we shall outline only the separation of the larger subcellular structures. Methods for the separation of polysomes, ribosomes and ribosome subunits from plant cells resemble those used with mammalian cells (e.g. Beevers and Poulson 1972; Jones et al. 1973) but allow for differences in the homogenisation medium and require considerable precautions against RNA degradation. When separating plant ribosomes it is necessary to remember that, as well as the cytoplasmic ribosomes, the mitochondria and chloroplasts have their own protein synthesis systems which include distinctive ribosomes. Electron microscopy of plant tissue is more difficult than of mammalian cells and establishment of markers for the various subcellular structures has proved much more difficult. Some of the problems are discussed by Halliwell (1974). Further confusion has been introduced by the giving of names to imperfectly characterised structures. Thus glyoxysomes and microbodies have been spoken of as two distinct structures when, in fact, the former appears to be only a specialised form of the latter (de Duve 1969). When one eliminates ‘doubles’ of this type it becomes evident that plant cells contain the same range of structures as mammalian cells, namely nuclei, mitochondria, lysosomes (although in a very adapted form), microbodies, Golgi apparatus, endoplasmic reticulum and plasma membrane. The only novel structures are the chloroplasts and their precursors. Some tissues also contain storage granules (Schnarrenberger et al. 1972b) and zymogen granules (Cohen et al. 1971). The largest subcellular structures, apart from nuclei, in plant cells are the chloroplasts. The outer membrane of the chloroplasts of spinach leaves (Rocha and Ting 1970) and of other tissues is easily broken during homogenisation or pelleting. Intact chloroplasts can,
Ch. 7
APPLICATIONS
221
however, be separated by rate sedimentation in an A-XII* zonal rotor (Still and Price 1967) or in the tubes of a swing-out rotor (Rocha and Ting 1970). Intact spinach leaf chloroplasts have a sedimentation coefficient of 200,000 S or more and band in sucrose gradients at a density of 1.21. Broken chloroplasts sediment more slowly and band at a density of 1 .I7 (Rocha and Ting 1970). While the mature chloroplasts may be easily recognised from their chlorophyll content, it is more difficult to recognise their proplastid precursors. Proplastids from the cotyledons of sunflower seeds germinated in the dark band at a density of 1.26 along with microbodies, but retain a characteristic enzyme content (Schnarrenberger et al. 1972a). Plant mitochondria and microbodies are most readily separated from broken chloroplasts by isopycnic banding on sucrose gradients. Excellent results are obtained both with spinach leaf homogenates (Rocha and Ting 1970; Donaldson et al. 1972) and with sunflower cotyledons (Donaldson et al. 1972). An example of such a separation is shown in Fig. 7.10. It should be noted that the separation, as shown, depends on the chloroplasts being broken either by homogenisation or during pelleting. Intact chloroplasts are of the same density as mitochondria, but may be removed by rate sedimentation. Plant mitochondria band in sucrose gradients at a density of 1.21, microbodies band at a density of 1.25 (Rocha and Ting 1970; Donaldson et al. 1972). Spinach leaf mitochondria have a sedimentation coefficientoftheorderof 3000 S, the microbodies a sedimentation coefficient of the order of 9000 S (Rocha and Ting 1970). The evidence for the existence of lysosomes in plants has been reviewed by Matile (1969) and Gahan ( I 973). Both authors conclude that, with plant cells one must think in terms of a system, corresponding to the Golgi-endoplasmic reticulum-lysosome system postulated in animal cells (Novikoff et a]. 1971). In plants, what are thought of as typical lysosomal enzymes are found in the endoplasmic reticum, in the dictyosomes of the Golgi apparatus, in pre-vacuolar vesicles * The experiments described in the reference were actually carried out using an A-IX zonal rotor, an earlier form of the A-XI1 zonal rotor. S u h p I w d e ~11 287
228
DENSITY GRADIENT CENTRIFUGATION
41r=4 sucre..
g a'
20
10
I
n
Volume Lmll
Fig. 7.10. Distribution of subcellular organelles from spinach leaves. A large particulate fraction was prepared by differential centrifugation and then further fractionated by centrifugation for 2 hr at 30,000 revs/min in a B-XXX zonal rotor. Markers are as follows: chlorophyll for chloroplasts, catalase for microbodies and cytochrome oxidase for mitochondria. NADH-cytochrome c reductase is present in both mitochondria and endoplasmic reticulum fragments (from Donaldson et al. 1972).
and in vacuoles. The last two structures would correspond to the lysosomes of mammalian cells, but play a much more elaborate part in the metabolism of plant cells than is performed by mammalian lysosomes. The small vactoles present in the meristematic root tip
Ch. 7
APPLICATIONS
229
cells of corn seedlings have been isolated by the isopycnic banding of a ‘mitochondrial’ fraction on a sucrose gradient (Matile 1968). The size, and consequent fragility, of the vacuoles of parenchymatous cells has so far prevented their isolation. Granules containing storage proteins have been isolated from sunflower cotyledons (Scharrenburger et al. 1972b) and from the reserve tissues of seeds (see Matile 1969). The latter granules appear to contain hydrolytic enzymes. ‘Zymogen bodies’ containing a starch debranching enzyme in inactive form may be isolated from pea seeds by isopycnic banding on a Ficoll gradient (Cohen et al. 1971). Plant microsomes have been little studied. The microsomal fraction as separated by differential pelleting is very heterogeneous, containing fragments of the endoplasmic reticulum, the Golgi apparatus and the plasma and vacuole membranes. In the absence of markers for structures other than the endoplasmic reticulum (Halliwell 1974) it is not possible to draw any definite conclusions about the purity of the various subfractions which may be separated by isopycnic banding (Lord et al. 1973).
7.4. Separation of subcellular components from unicellular organisms In bacterial cells the major role of density gradient centrifugation has been in the separation of the DNA or the ribosomes. DNA separation is discussed later (0 7.5.1.2). The techniques for separating bacterial polysomes, ribosomesandribosomalsubunitsarevery similar to the methods used with the corresponding structures from mammalian cells. It should be noted that very rapid cooling of the growing cells and careful lysis by digestion with lysozyme in the presence of a detergent is essential if intact polysomes are to be separated (Godson and Sinsheimer 1967; No11 1969). The separation of ribosomal subunits from bacteria is very simple. Up to 2 g of subunits can be separated at any one time by the use ofa steep sucrose gradient (Eikenberry et al. 1970; Sypherd and Wireman 1974). Free-living eucaryote cells have at least as wide a range of intraSuhlerf index p 287
230
DENSITY GRADIENT CENTRIFUGATION
cellular structures as mammalian and plant cells. Like the latter, the cells of free-living eucaryotes are surrounded by a strong cell wall and one of the major difficulties in separating the intracellular structures is to break the cell wall without simultaneously breaking all the intracellular membranes. Another similarity with plant cells is the difficulty of finding reliable markers for the various intracellular structures. Recently, however, a comprehensive discussion of possible marker enzymes for the larger subcellular structures of a wide range of eucaryotic micro-organisms by Lloyd and Cartledge (1974) has partly clarified the field. The most commonly used method for fractionating homogenates from protozea and yeasts has been isopycnic banding on sucrose gradients. In most cases, work has been concentrated on the larger subcellular particles. The following short summary is taken from the article by Lloyd and Cartledge (1974) which should be consulted for further references. Among protozoa, the most extensive work on the isolation of subcellular structures has been with the ciliate Tetrahymena pyriformis. Lloyd et al. (1971) and Muller (1972) have reported the patterns obtained by isopycnic banding of the large particulate fraction on sucrose gradients. Mitochondria, microbodies and endoplasmic reticulum fragments were well separated, but lysosomal enzymes showed a very complex pattern which indicated the presence of at least three populations with distinctive enzymology. There is considerable cross contamination between the various fractions. Little work appears to have been carried out on the fractionation of Tetrahymena homogenates by rate sedimentation, but there are indications that it is possible to obtain a considerable degree of purification of lysosomes in this way (Lloyd et al. 1971; Cooper and Dobrota 1974). Other protozoa have been studied in less detail. The soil amoeba Hartmanella castellanii gave results very similar to those obtained with Tetrahymena pyriformis except that there was less separation between mitochondria and microbodies. A serious problem in all work on micro-organism is illustrated by the experiments of Wiener and Ashworth (1970) on the myxamoeba of Dictyostelium discoideum. The
Ch. 7
APPLICATIONS
23 1
isopycnic banding densities of the lysosomes and microbodies and their enzyme composition varied markedly with the conditions of growth. The parasitic flagellate Trichomonas foetus has been fractionated by isopycnic banding in the E 40 isopycnic banding rotor developed by Beaufay (Muller 1973). The results are interesting because of the unsuspected enzyme composition of what appears, morphologically, to be a fraction highly enriched in microbodies. These results emphasise again the necessity for establishing markers for each species of organism which is studied. The results obtained with yeasts are, in general similar to those obtained with protozoa. It has been possible to separate intact cells of the fission yeast Schizosaccharomyces pombe in different stages of growth by centrifugation in a HS zonal rotor using rate sedimentation on a sucrose or Ficoll gradient and by isopycnic banding in a B-XIV zonal rotor (Poole and Lloyd 1973). The centrifugal behavior of subcellular particles from Saccharomyces cerevisiae has been extensively studied by Cartledge and Lloyd (1972). On isopycnic banding, mitochondria were split into two subfractions, but it is not clear whether this is due to intrinsic heterogeneity or to damage such as can occur during the isopycnic banding of liver mitochondria in a swing-out rotor (Beaufay 1966). Yeast microbodies are poorly separated from mitochondria on isopycnic banding, but may be separated by rate sedimentation (Cartledge et al. 1971). Yeast ‘lysosomes’ are very heterogeneous in density and sedimentation coefficient. Clearly there are several different types of structure containing acid hydrolases, but it has not yet been possible to identify these with intracellular structures visible in the electron microscope. Subcellular fractionation of algae give results similar to those obtained with higher plants (Lloyd and Cartledge 1974). Little work has been carried out on the systematic subcellular fractionation of fungal cells although Neurospora crassa has been a favoured species for the study of mitochondria1 DNA. References to fractionation methods will be found in Lloyd and Cartledge (1974). Up till now, studies on eucaryotic microorganisms have concentrated on establishing the distribution of enzymes between the larger Stihli < I wI
232
DENSITY GRADIENT CENTRIFUGATION
subcellular particles which may be identified by electron microscopy. Isopycnic banding in zonal-type rotors has been the favoured technique. It is clear, however, that the fractions obtained are nowhere near pure. Further studies of the subcellular organelles of eucaryotic micro-organisms will require pure preparations and it is likely that the technique of choice will be rate sedimentation followed by isopycnic banding, the S-p technique first described by Anderson (1966a).
7.5. Fractionation of macromolecules 7.5.1. Nucleic acids MA. The most usual method for fractionating RNA is by ratezonal centrifugation on sucrose gradients (McConkey,1967)although, as discussed earlier (Ch. 1) better analytical separations are achieved by gel electrophoresis (see Fig. 1.8). Nevertheless density gradient centrifugation is still a convenient method for analysing radioactively-labelled RNA, as it is much easier to fractionate liquid gradients than polyacrylamide gels. Problems do, however, arise in the counting of 3H in the presence of appreciable amounts of sucrose (see 9 8.2.3). For special purposes, media other than sucrose may be used. A complex density gradient in which a gradient of phenol overlies a sucrose gradient has been proposed by Hastings (see Parish 1972) for the direct analysis of very small amounts of RNA (and DNA). Thecellsare lysed and the nucleic acids are deproteinised in the phenolcontaining layer. It should be noted that some nucleic acids have a tendency to aggregate on density gradient centrifugation. This may sometimes be overcome by centrifuging at a higher temperature than normal (Parish 1972) or by using conditions in which the RNA is totally denatured. Gradients of dimethyl sulfoxide and hexadeuterodimethylsulfoxide (Sedat and Sinsheimer 1970) and of methanol/ methoxyethanol (Hastings and Brown 1971) have been used for this purpose. The rate of sedimentation in such gradients is directly proportional to molecular weight, but very odd patterns may be obtained due to the revelation of ‘hidden breaks’ which exist naturally
Ch. 7
APPLICATIONS
233
in the chains of high molecular weight RNA (Kokileva et al. 1971). Polyacrylamide gel electrophoresis is replacing gradient centrifugation as the method of choice for analysing RNA (The techniques are given in Gould and Matthews, this series Vol. 4) but density gradient centrifugation is still useful for preparing pure RNA fractions. Preparative gel electrophoresis can also be used, so can column chromatography (see Gould and Matthews), but zonal rotors are useful when reasonable amounts of even minor components are to be separated (Williamson et al. 1971)*. RNA is not normally fractionated by isopycnic banding, although separation on the basis of base composition can be obtained in the same way as with DNA. Cs,SO, must be used as the density gradient solute since RNA pellets through saturated CsCl. The use of Cs,SO, gradients has been reviewed by Szybalski (1968). D N A . DNA is normally separated on the basis of its density, which reflects its base composition, but rate separation may be useful in separating viral DNA (Jaenisch et al. 1969) or, with more difficulty, bacterial DNA (Kavenoff 1972). In the latter case great care must be exercised if the DNA is not to be degraded; the bacteria must be lysed directly on to the surface of the gradient, for example by preparing protoplasts and treating them in situ with sodium dodecylsulphate. This detergent should be included in the gradient to dissociate DNA and protein. When the gradient is fractionated one must avoid the use of narrow flow tubes and the displacement should be carried out slowly to minimise shearing (Kavenoff 1972). It is not possible to maintain mammalian DNA molecules intact, but rate sedimentation in alkaline sucrose gradients has been used for the isolation of fragments of sheared DNA of homogeneous size for hybridisation experiments (Hell et al. 1972; Birnie et al. 1973~).An interesting feature of these experiments was the use ofvery shallow (5-1 1% w/v) sucrose gradients and of long centrifugation (up to 52 hr) to separate small molecules.
*
It cannot be emphasised too strongly that commercially available sucrose, unless it is of a special ‘enzyme-free’ grade, is contaminated with ribonuclease. This must be removed by treatment of solutions with activated charcoal or with diethyl pyrocarbonate (see 4 3.3.1.2). S i i b / c i ~iiidu p 287
234
DENSITY GRADIENT CENTRIFUGATION
Practically no change was found in the gradients at the end of centrifugatio n . The fractionation of DNA by isopycnic banding on CsCl gradients was one of the earlier applications of density gradient centrifugation. Cheaper solute materials such as sodium iodide (NB, high U.V. absorbance) may be useful for large-scale separations, but CsCl is likely to remain the solute of choice, especially if methods for its recovery can be made cheaper (Wright et al. 1966). Some care should be exercised in choosing the source of CsCl for DNA fractionation as some commercial preparations contain sufficient heavy metal ions to affect the density of the DNA (Flamm et al. 1972). The fractionation of DNA may be carried out using either equilibrium or preformed gradients (see Ch. 3). Either angle-head or swing-out rotors can be used, but the former appear to have significant advantages (Flamm et al. 1966, 1972). Titanium zonal rotors may be used, but the large volume of these rotors means that large amounts of expensive CsCl would be required. Aluminium zonal rotors should not be used (see 8 1.7.2.2.). Aluminium swing-out rotors may be used, but great care should be taken to avoid prolonged contact between the metal and CsCl solutions. When fractionating DNA gradients very strict precautions should be exercised to prevent mixing, for CsCl solutions have only a low viscosity. Thus after fractionation the rotor should always be decelerated without the brake, and great care should be taken to avoid density inversions (see Q 3.2) in the line leading from the displacement head to the spectrophotometer. DNA for analysis by density-gradient centrifugation may either be purified by extraction with phenol or by a preliminary centrifugation in CsCl, when DNA-protein bonds are disrupted by the high ionic strength of concentrated CsCl solutions (Flamm et al. 1972). Density gradient centrifugation has largely been employed as a method for fractionating DNA on the basis of its base composition and especially is purifying regions of unusual base composition (satellite DNA). It has also been used extensively as a method for determining the base composition of bacterial DNA as a taxonomic tool as with streptococci and micrococci where convergent evolution has led to great
Ch. I
235
APLICATIONS
resemblances in phenotype (Mandel et al. 1970). If the pH of CsCl gradients is increased to 12.5 then the two strands of DNA will separate (Szybalski et al. 1971). If the two complementary strands differ considerably in base composition, they may be separated directly by banding in CsCl gradients. However, even where the overall base composition of the two strands and their banding densities are similar, separation may be achieved by use of agents, such as natural o r synthetic ribopolymers which complex specifically with one strand and so alter its density. 7.5.2. Fractionatiori ofproteins In contrast to the almost universal use of density gradient centrifugation in the fractionation of DNA, it is only recently that this technique has been used for the purification of larger proteins. Almost all separations of protein have been carried out on the basis of sedimentation rate, and zonal rotors have been widely used to allow the purification of appreciable amounts of material in a single centrifugation. Thus Edelman et al. (1958) showed that serum macroglobulins could be separated by density gradient centrifugation in the tubes of swing-out rotors, but only very small amounts of material could be separated so that the method was not widely used. With the introduction of B-IV and then B-XIV and B-XV zonal rotors it became possible to scale-up the procedure sufficiently to allow the isolation of useful amounts of macroglobulins (Fisher and Canning 1966; Cooke and Apsey 1971, see Fig. 7.11). More recently, even greater yields have been obtained by the use of K-type zonal rotors (Cline et al. 1973). A related application has been the isolation of tumor transplantation antigen from mouse spleen (Popp et al. 1968). More interesting, because of the much wider applications, are the separations of solubilised mouse plasma membrane proteins by Evans and Gurd (1973a, b). The membranes were solubilised by treatment with the detergent sarcosyl (sodium dodecyl sarcosinate) and fractionated by centrifugation on a B-XIV zonal rotor using sucrose gradients containing 0.25 sarcosyl. 5'-Nucleotidase (Evans and Gurd 1973a) and alkaline phosphodiesterase (Evans et al. 1973b) Sirhi<
1
p 287
236
DENSITY GRADIENT CENTRIFUGATION
Fig. 7.11. Separation of serum macroglobulins. 3 ml of serum from a patient with hyperglobdinaemia was centrifuged for 6 hr at 47,000 revs/min in a B-XIV zonal rotor. The 19 S macroglobding are responsible for the peak at a volume of 350 ml. 27 S macroglobulins are indicated by the shoulder at 420 ml (from Cooke and Apsey 1971).
have been purified to electrophoretic homogeneity. The technique would seem readily extensible to components of other types of membranes and to membranes of other tissues so long as the material which is to be separated does not depend on lipid for its function. Isopycnic gradient centrifugation cannot be used for the separation of simple proteins, as all band at about the same density. Proteins with lipid or carbohydrate prosthetic groups will, however, vary in density depending on the proportion of non-polypeptide groups. Isopycnic banding has been used to fractionate mucoproteins in CsCl gradients (Starkey et al. 1972). However, proteins are likely to be denatured and their subunit structure, if any, may be disrupted by
Ch. 7
APPLICATIONS
231
the high concentrations of salt needed for isopycnic banding. Hence isopycnic banding is unlikely to be of major importance as a fractionation method even for mucoproteins and glycoproteins.
7.6. Other applications of’density gradient centrijugation in biochemistry 7.6.1. Separation of serum lipoproteins Lipids are carried in the blood in the form of lipoprotein particles. Fasting serum contains 3 distinct types, high density lipoprotein (HDL), low density lipoprotein (LDL) and very low density lipoprotein (VLDL). In addition the sera of fed subjects contain chylomicra, particles carrying lipid from the intestine which resemble VLDL in size, but are, on average, rather larger. The three classes of endogenous lipoprotein differ considerably from each other in sedimentation rate and in density. The original method for their analysis was by the analytical ultracentrifuge (Frederickson et al. 1968) although electrophoretic methods were also introduced (Hatch and Lees 1968; Lindgren et al. 1972). The normal procedure for the preparative fractionation of serum lipoproteins is a stepwise flotation (Lindgren et al. 1951, 1972) which is an exceptionally long and tedious procedure needing over 60 hr of centrifugation for the separation. Better resolution has been achieved by floating through a density gradient (Lindgren et al. 1962, 1972), but early experimentersused an essentially stepwiseprocedure in which the specimen was centrifuged repeatedly, lipoproteins which had reached the surface of the tube being removed after each step. Viikari et al. (1968) and Wilcox et al. (1971) obtained excellent separations by flotation in a B-XIV zonal rotor, with resolution essentially similar to that achieved in the analytical ultracentrifuge. Rather large amounts of sample were required if the separated bands were to be detectable in a U.V. monitor. However, it has since been found that equally good separations can be achieved by use of a swing-out rotor (Mallinson and Hinton 1973; Hinton et al. 1973b) and that sufficient material can be separated in this way to allow analysis Suhptl ,,u/c\ p 287
238
DENSITY GRADIENT CENTRIFUGATION
of the lipid composition of the major fractions (Hinton et al. 1974a). NaBr is the gradient material of choice for the gradient flotation of serum lipoproteins. There are some advantages in using mixed NaBr-sucrose gradients (Mallinson and Hinton 1973; Hinton et al. 1973b) but there are also disadvantages so that we would now recommend the use of simple NaBr gradients. The degree of separation can be influenced to some extent by the slope of the gradient but we have not carried out systematic experiments to determine the optimum shape. for usewithswing-out rotors, we use a linear gradient extending from a density of 1.006 to 1.2, with zonal rotors a linear gradient extending from a density of 1.006 to 1.2, with zonal rotors a linear gradient extending from 1.006to 1.4as used by Wilcox and colleagues; the gradient solutions should contain 0.195 M NaCl and 0.1 mg/ml EDTA (Lindgren et al. 1972). The sample is adjusted to a density greater than the dense end of the gradient by the addition of solid NaBr and layered under the gradient as described in the appropriate operations section (0 4.1.2 and 0 5.1.3.3). An underlay of very dense NaBr is introduced to move the sample away from the base of the tube or the wall of the zonal rotor. An excellent separation of the larger lipoprotein particles from a small volume of serum is achieved by centrifugation for about 1 hr at 94,000 g, although slightly longer may be required if a very large sample is to be fractionated (Fig. 7.13 and Hinton et al. 1974a). Separation of HDL from albumin and other proteins requires that centrifugation should be continued for about 4 hr, with a gradient extending to a density of 1.4. In particular the rate of acceleration and deceleration mukt be strictly controlled in view of the mobility of the NaBr gradient. Isopycnic banding of separated lipoprotein fractions may be most useful in characterising the ‘abnormal’ lipoproteins separated from some pathological specimens. Ifthelipoproteins have been fractionated by rate flotation as described above, it is usually sufficient to simply overlay the fraction with 0.195 M NaCl and centrifuge. An approximately linear gradient will form in the course of the prolonged centrifugation (about 18 hr at 205,000 g) needed for the isopycnic banding of the smaller lipoproteins (Hinton et al. 1974a).
Ch. 7
239
APPLICATIONS
I+ BDL
w
I so
Fig. 7.12. Separation of serum lipoproteins. Flotation was for 2 hr at 30,000 revs/min through a linear p = 1.2 - 1.0 NaBr gradient in the tubes of an MSE 3 x 23 ml swing-out rotor. The sample was a) 0.5 ml of serum underlaid with 2 ml of NaBr p = 1.5, h) 2.0 ml of serum underlaid with 0.5 ml of NaBr p = 1.5 (from Hinton et al. 1974).
7.6.2. Separation of viruses The large-scale separation of virus particles from fragments of the host cells is usually performed using continuous-flow rotors and so lies outside the scope of this article. The construction of these rotors, their operation and their uses are described in articles by Anderson et al. (1969), Cline et al. (1971), Cline (1971), Cline and Dagg (1973) and Birnie et al. (1973b). However, while continuous-flow rotors must be used for the separation of large volumes of liquid, batch rotors may be very useful for trial experiments involving only a few litres or less of suspension. A great deal of attention was given to the problem of separating viruses by the group at the Oak Ridge National Laboratory under Anderson. As is pointed out by Anderson et al. (1966), a large number ofviruses have a combination of size and density which is distinct from that of any subcellular particles, hence they fall within what they call the virus window of an S-p diagram (Fig. 7.14). Thus it should be possible to separate viruses by density gradient centrifugation. In their early experiments Anderson et al. (1966) used rate sedimentation S u h / r ~iwdcx ~ p , 287
240
DENSITY GRADIENT CENTRIFUGATION
Fig. 7.13. The ‘virus window’. An S-p diagram which shows that it may be difficult to separate many viruses by a single rate zonal run or directly by isopycnic banding, but that many viruses possess a combination of size and density not found with normal cell constituants and so can be separated by a combination of the two techniques. Note that the banding densities on this figure are the banding densities in CsCl gradients not, as in earlier S-p diagrams, the densities in sucrose gradients (from Anderson 1966).
followed by isopycnic banding. Most later workers have found it convenient to either reverse these procedures or, indeed, to dispense with the rate separation step altogether. In addition, Anderson et al. (1966) and other early workers used principally CsCl gradients. Since then it has been found that many viruses can be banded on sucrose gradients (Birnie et al. 1973b). Potassium tartrate has been proposed as an alternative to CsCl for the fractionation of denser
Ch. 7
APPLICATIONS
241
viruses (Reeve and Alexander 1970), but the pH of tartrate solutions changes markedly with the concentration. It should be possible to use the much cheaper NaBr in place of CsCl, but the authors do not know whether this has been done. All these ionic gradients may, however, damage viruses during separation and hence sucrose, when it can be used, is probably the safest as well as the cheapest solute. When considering which gradient material to use, one should bear in mind that viruses, like other nucleoprotein particles (Q 7.2.5), band at lower densities in gradients of non-ionic solutes such as sucrose than in CsCl. For example a virus that bands in CsCl at a density of 1.38 bands on sucrose gradients at a density of 1.20 (Nash et al. 1973). As the virus is separated from cell debris on the basis of banding density, the size of the sample does not affect resolution. Thus Leach (1971) separated influenza virus from 1 1 of culture medium in a single centrifugation step using a B-XV zonal rotor. The suspension containing the virus was pumped into the rotor and followed by a buffer layer of 20% (w/v) sucrose and a steep gradient extending from 20% to 60% (w/v) sucrose. After centrifugation the virus was found in a band only 2 M O ml wide. When the sample region contained a large concentration of protein, this tended to contaminate the virus band due to sedimentation in droplets (see Ch. 2). This contaminating protein could be removed by taking the virus-rich region, increasing its density by the addition of further sucrose and layering it under a second density gradient ; upon centrifugation the virus particles will float up and so separate from the protein. Alternatively, the sucrose may be dialysed out and the virus purified by rate sedimentation. The major problem in separating live viruses probably lies not in the separation techniques themselves, but in the precautions which must be taken to prevent the escape of potentially dangerous material. It should be realised that aerosols tend to form at the seals of zonal rotors (see 0 1.6.2.2) and that particles in such aerosols may travel a considerable distance. The precautions which should be taken in handling live virus are discussed briefly in 0 1.6.2.2 and by Webb et al. (1975). It is impossible for us to do justice to all the work which has been Sl,h/rcr eid<,* p 287
242
DENSITY GRADIENT CENTRlFUGATlON
carried out on the separation of viruses by density-gradient centrifugation, or even to the separations carried out on batch rotors. We have endeavoured to outline the general principles ; readers who are interested in further details should consult articles by Anderson and Cline (1967) and Fox et al. (1968). Recent references, concentrating on the use of continuous flow rotors are given in articles by Nash et al. (1975), van Frank and Kleinschmidt (1973) and Sotton (1974).
7.7. Other applications of density gradient centrifugation Although density gradient centrifugation can, in theory, be applied to the separation of any particles with diameters less than about 20 ,urn, in practice the technique has been mostly restricted to the separation ofparticles fromlivingcells. Density gradient centrifugation in B-XV zonal rotor has been used to characterise the size of china clay particles (J.B. Rowse, personal communication), and the analytical ultracentrifuge has long been used by polymer chemists for the analysis of molecular weight distributions (e.g. McCormick 1967). Surprisingly,there has been little use of density gradient centrifugation for theseparationofthelargerpolymers. Methods have been developed for the measurement of diffusion coefficients in zonal rotors (Halsall and Schumaker 1970, 1971) but though these experiments have been very useful in examining the mechanism of centrifugal artefacts such as sedimentation in droplets, the authors know of no practical applications. Thus up to now, deqsity gradient centrifugation has been a tool used almost entirely by biochemists. This may not always be the case.
CHAPTER 8
Artefacts arising during centrifugal separations
As we discussed earlier, density gradient centrifugation may be either used as a preparative technique to separate one or two particles from a mixture, or it may be used analytically, to determine the composition of a mixture. When centrifugation is used as a preparative technique, one is really only concerned that the isolated particles should be undamaged and should be reproducibly separated from the other particles in the mixture. When centrifugation is used analytically, if the results are to be useful in elucidating the structure of the living cell, one must avoid damaging any of the particles in the mixture. In addition the assay of marker enzymes and of other components should not be affected by the density gradient medium. Finally, one frequently wishes to obtain accurate information on the density and the sedimentation coefficient of the various particles in the mixture. As happens with all techniques, centrifugation causes a number of artefactsand we will now consider how theseaffect the results obtained from preparative and analytical centrifugal separations.
8.1. Damage to particles during Centrifugation The attitude of scientists to centrifugal separations has tended to be either an intense scepticism about the validity of the methods (Hillman 1972) or, more usually, an uncritical acceptance of the results. While thereisno evidencetojustifytherootand branch rejectionofcentrifugal techniques one must realise that there is a finite risk that particles will be damaged by exposure to high centrifugal fields. The least damage occurs in rate centrifugation. Particles moving through a medium at 243
Sshiu I
UI&\
p 287
244
DENSITY GRADIENT CENTRIFUGATION
a constant speed have no net force acting on them, the weight of the particles being exactly balanced by the viscous drag. The rate of sedimentation is usually so slow (1 cm/min) as to be negligible compared with the Brownian motion for all except the largest particles. Thus sedimentation in itself is unlikely to cause any damage. The situation on isopycnic banding can be more serious. While a uniform particle suspended in a liquid of its own density is subject to no force, only the smallest subcellular particles such as ribosomes are even approximately uniform. A non,-uniform particle such as a mitochondrion suspended in a liquid of its own mean density is subject to very considerable forces which tend to pull it apart. This can be seen dramatically in the banding of whole cells which are stratified at quite low centrifugal fields and are actually pulled apart under higher fields (Harvey 1932). Surprisingly, however, stratified cells can recover and divide normally. Thus, while one must be conscious of the problems caused by banding non-uniform particles, these problems do not seem to be very important in practice. Much more serious are three other effects, the crushing of pelleted particles, the damaging effects of the high hydrostaticpressures generated in the centrifugetube and the damaging effects of high concentrations of the gradient material. 8.1. I . Damage caused by pelleting While the effects of pelleting particles are not strictly relevant to to an article on density gradient centrifugation, pelleting is such a useful method of recovering particles after separation on density gradientsthat it must be mentioned. A particle sedimentingat constant velocity in a liquid medium is subject to no net force. This is not so of a particle in a pellet which is crushed by its own weight (up to 400,000 times its mass) and by the weight of the particles above it. This can result in extensive damage. One should, therefore, pellet particles at the lowest possible speed. Fortunately, the resistance of particles to crushing depends on the surface tension in their membranes and hence rises with decreasing size. As a general guide, one should not use much greater centrifugal fields on pelleting particles
Ch. 8
ARTEFACTS ARISING DURING SEPARATIONS
245
from density gradient fractions than would be used to separate similar particles during differential pelleting. 8.1.2. Damage due to high hydrostatic pressures The first evidence that particles could be damaged by high hydrostatic pressures came from observations that mitochondria banded at higher densities when centrifuged at 165,000 g than when centrifuged at 84,000 g and that the higher banding density was due to breakage of the outer membrane (Wattiaux et al. 1971; Wattiaux 1974). The damage was found to be related to the length of the liquid column overlying the mitochondria1 band and was not, therefore, due simply to the high centrifugal field. More recently the damaging effects of high hydrostatic pressures have been demonstrated directly (Bronfman and Beaufay 1973). Other workers showed that some unusual results obtained during the sedimentation of ribosomes were due to the dissociation of the two subunits being facilitated by high hydrostatic pressure (Infante and Krauss 1971). From these observations, it is clear that one should exercise some caution when interpreting separations carried out in very high centrifugal fields. As the hydrostatic pressure at any point in a centrifuge tube is related to the height of the liquid column above that point, it is clear that the hydrostatic pressure increases along the length of the tube and that especially high hydrostatic pressures will be found with the long thin tubes of many of the more modern swing-out rotors. Inspection of the papers quoted in the previous paragraph suggests that one should be cautious about speeds in excess of 30,000 revs/min when using swing-out rotors. The length of the liquid column in zonal rotors is less than the length of the tubes of most swing-out rotors, and calculations suggest that none of the current zonal rotors should be much affected by the above-mentioned problems.
8.1.3. Damage due to high concentration of the gradient solute As discussed in 8 3.3, density gradient materials can be divided into three major groups salts of the alkali metals, low molecular weight Sirhierr iiidcxp. 287
246
DENSITY GRADIENT CENTRIFUGATION
organic molecules, especially carbohydrates, and, finally, hydrophilic macromolecules. The first group are very destructive of biological structures and are only used when separating macromolecules such as DNA or when no other gradient material could be employed. The second group, of which sucrose is commonest, forms solutions of high osmotic strength and tend to penetrate biological membranes. This makes them lethal to living cells* (Mateyko and Kopac 1963) and may cause damage to the. most highly organised subcellular structures. Thus respiratory control in mitochondria is adversely affected by exposure tb high concentrations of sucrose (Zimmer et al. 1972). Otherwise there is remarkably little evidence of damage to biological structures by solutions of high osmotic strength. However, it is clear that the least risk of damige to biological structures will come from the high molecular weight solutes in the third group of Table 3.1 and from Urografin, Ludox and Metrizamide in the fourth group. These components do not penetrate biological membranes and in solution exert very small osmotic pressure. Indeed, such gradient materials are the only ones which can be used for the separation of living cells (see Ch. 7) and it is reasonable to assume that if a gradient material does not kill a living cell, it is likely to have only minimal effects on its component parts.
8.2. Factors affecting the accuracy of assays performed on fractions from density gradients When the distribution of particles through a density gradient is to be analysed, it is important to know whether and how the density gradient solute will affect the assays being performed on the separated fractions. Ignorance of such interference may lead to misinterpretation of the patterns. As was described in Ch. 6, an important check on such interference is the calculation of the ‘recovery’ of the component being assayed in the fractions obtained from the gradient.
*
References to ‘living cells’ in this section refer to isolated mammalian cells. Cells possessing a cell wall are, of course, more resistant to density gradient solutes.
Ch. 8
ARTEFACTS ARISING DURING SEPARATIONS
241
However, a good recovery is not necessarily evidence of the absence of interference as, for example, the reaction of contaminants with the reagents used in the assay may balance out an inhibition of the reaction of the reagents with the particles themselves. Therefore it is sensible, if one is using a novel density gradient material to test its influence on the assays which one will perform on the separated fractions. Four distinct ways in which a density gradient solute might produce erroneous results were listed by Hartman et al. (1974). 1) The solute may actually damage the particles, so affecting the activity of component enzymes or other molecules. 2) The solute may react itself with the reagents used in the assay. 3) The solute may interfere with the performance of equipment used in the assay. 4) The solute may interfere with the sensitivity of the assay without itself reacting with the reagents. The interaction of density gradient solutes with cells and subcellular structures has been discussed above. The other three effects of density gradient solutes will now be considered in turn. 8.2.1. Reaction of gradient solutes with reagents used in the assay of separated constituents It is obviously impossible to list all possible causes of interference under this head. Only the most potentially troublingcan be mentioned. Among these are the interference by ultraviolet absorbing solutes (see Table 3.1) in the assay of nucleic acids by procedures based on precipitationand extractions followed by reading the extinction at 260 nm. Even though the nucleic acids can usually be separated from the b d k of the density gradient material by precipitation with perchloric acid (an exception is Urografin which is precipitated by acid), the pellet must still be washed several times to ensure removal of all density gradient solute. A similar problem is encountered when estimating RNA ribose by the orcinol procedure after separation in gradients of carbohydrates such as sucrose or Ficoll or of Metrizamide or Urografin. Suh,r
248
DENSITY GRADIENT CENTRIFUGATION
Another estimation which frequently causes trouble is the estimation of protein. Sucrose does not give a blue colour with the reagents used in the procedure of Lowry et al. (1951), but reacts with copper reagents to give an ultra-violet absorbing complex (Hinton et al. 1969) which interferes severely with the estimation by a microbiuret procedure (Itzhaki and Gill 1964). Ficoll (Lo and Stelsen 1972) and Urografin (Hinton et al. 1973)react strongly with Lowry reagents, Metrizamide reacts more weakly, a 36% solution giving the same reaction as a 1 mg/ml solution of bovine serum albumin (Hinton et al. 1974b). Tris buffer also reacts slightly with the Lowry reagents (Rej and Richards 1974).
8.2.2. Interference with the performance of analytical equipment We have shown that sucrose affects the performance of AutoAnalysers quite markedly (Hinton and Norris 1972). The trailing edge of each peak is broadened leading to increased cross-contamination (Fig. 8.1)
100
I
L
0
1
2
3
4
5
Time (min)
Fig. 8.1. Effect of sucrose on the shape ofAutoAnalyserpeaks, tested on the manifold for the estimation of protein described by Hinton and Norris (1973). The shape of or in 2 M a 1 mg/ml bovine serum albumin standard dissolved in water (-) sucrose (-----). Samples were aspirated for 48 sec.
Ch. 8
ARTEFACTS ARISING DURING SEPARATIONS
249
but the leading edge is actually slightly sharpened. We suggested that the effects may be explained by a density inversion in the sample probe as the aqueous wash succeeds the dense sample. 8.2.3. Interference ofgradient solutes in the sensitivity of assays Both the assay of protein by the Lowry procedure and measurements of a wide variety of enzymes are inhibited by sucrose (Hinton et al. 1969;Hartman et al. 1974), probably due to interaction with proteins. This inhibition is linear with sucrose concentration (Fig. 8.2) and is readily reversible on dilution. A similar inhibition has been observed with Metrizamide (Hinton et al. 1974)with glycerol, a variety of low molecular weight carbohydrates and with urea at low concentrations
2
SOLUTE CONCENTRATION X w/v
Fig. 8.2. The effect of sucrose and a number of other potential density gradient solutes on the activity of alcohol dehydrogenase. Activities are presented as a percentage of the activity in the absence of density gradient solute. -- 0--. Dextran Ficoll (mean mol. wt. 400,000); -A-, sorbitol; (mean mol. wt. 115.000); -m-, glycerol; -.-n-.-. fructose; -0-. sucrose; -- x--. glucose; . . . x . . .. . . V..* urea (From Hartman et al. 1974).
+.
.
Sirhlcrr rr~dcr11. 287
250
DENSITY GRADIENT CENTRIFUGATION
(Hartman et al. 1974). No inhibition of enzymes was observed with Ficoll or a high molecular weight dextran (Hartman et al. 1974). Gradient materials may also interfere with the measurement of radioactivity by liquid scintillation counting. We have found that the counting of 3H-labelled RNA and ribosomes is quenched in the presence of sucrose or CsCl in the counting medium. The counting of 14C-labelled RNA or ribosomes is only slightly affected and the counting of low molecular weight compounds containing 3H or 14C is not affected (Dobrota and Hinton 1973). We have suggested (Hartman et al. 1974) that both the inhibition of enzyme activities and the quenching of the counting of 3H-radioactivity may be explained by the density gradient material binding into the hydration sphere of macromolecules. All the low molecular weight solutes which we tested have shown this effect. At high concentrations, the inhibition of enzyme activities and the quenching of radioactivity may be quite considerable and we would strongly advise people to check their own systems to see whether this problem is having a significant effect on their results.
8.3. Uncertainties in estimates of particle density and sedimentation coefflc ient 8.3.1. Particle density Although the density ofparticles varies with the density of the medium (0 2.2.4), the isopycnic banding density of particles on any particular gradient may be determined accurately. One must, of course, ensure that the particles have reached the equilibrium densities, but otherwisetheaccuracyoftheestimatewilldependonlyon the sizeof fractions which have been collected, the steepness of the gradients and the precision with which the concentration of gradient material was determined. The problems occur in interpreting the densities obtained, for, as discussed in Ch. 2, there may be very considerable variations in the banding density of a particular particle between different gradient solutes. This variation, in turn, depends on the degree to which the gradient solute can penetrate the particle.
Ch. 8
ARTEFACTS ARISING DURING SEPARATIONS
251
8.3.2. Sedimentation coeflicient In the authors' opinion it is not possible to calculate accurate sedimentation coefficients from the results of rate-zonal separations either in swing-out or zonal rotors unless the sedimentation coefficient of one of the components of the mixture is known. This is because the information about the conditions inside the rotor during centrifugation is insufficiently precise. There are four major uncertainties which will be considered in turn. Uncertainty in the density of sedimenting particles. As mentioned above and discussed in Ch. 2, the density of particles may vary considerably depending on the medium in which the particles are suspended. This variation is most dramatic with membrane-bound particles and failure to allow for this can have an appreciable effect on the calculated sedimentation coefficient. The uncertainty in the density of small 'uniform' particles such as ribosomes is much less. The real density of ribosomes in dilute sucrose solutions probably lies between 1.26, as measured by banding in Metrizamide gradients (Hinton et al. 1974), and 1.4; the value of 1.4 is obtained by extrapolating from the variation of sedimentation coefficients (measured in the analytical ultracentrifuge) with variations in the concentration of sucrose in the medium (Petermann 1964). This uncertainty would produce an uncertainty of about 10% in the calculation of sedimentation coefficient. Uncertainty in theposition of the starting zone. Two problems affect calculation ofthe position of the starting zone. Firstly, there is a slight uncertainty about the exact distance from the centre of rotation, which will effect calculation of the gravitational field. This will, however, have little effect on the calculation of sedimentation coefficients. Much more serious are effects arising from the tumbling or 'sedimentation in droplets' in a heavily loaded sample zone (see Q 2.1.2). The effect of such artefacts is to broaden the centrifugal side of the sample zone. The centre of mass of the sample zone will therefore be displaced from its expected position. It may be difficult to detect such broadening, for diffusion will tend to restore sedimenting peaks to a Gaussian shape. However, the centres of the peaks will SL,h,'.'f !,,,I',\ ,' 2x7
252
DENSITY GRADIENT CENTRIFUGATION
be at the position expected if they had started from the centre of mass of the broadened zone, and there is no simple way of determining this. The effects of such broadening on the calculation of sedimentation coefficient are especially serious, for the sedimentation coefficients calculated for the various peaks will not be related to the true sedimentation coefficients by a simple ratio as is the case with the problems described in the other paragraphs. In this case, sedimentation coefficients cannot be determined accurately even if the sedimentation coefficient of one of the components of the mixture is known. If sedimentation coefficients are to be calculated, special care must be taken, therefore, to avoid overloading the gradient. Uncertainty in the shape of the gradient. Diffusion of solute from the sample zone and from the cushion may cause considerable changes at the extreme ends of the gradient during centrifugation, especially with zonal rotors. Fig. 8.3 shows a gradient in a B-XV zonal rotor a) as formed and b) as recovered 18 hr later. However, such changes have little effect on the values calculated for the sed-
Fraction no.
Fig. 8.3. The effect of diffusion o n the shape of a gradient in a B-XV zonal rotor. the gradient as loaded; -----, the gradient as recovered after centrifugation for 18 hr at 21,000 revs/min. The dotted area indicates the sample zone.
-,
Ch. 8
ARTEFACTS ARISING DURING SEPARATIONS
253
imentation coefficient. We would, therefore, conclude that changes in the gradient shape during centrifugation are not likely to have much effect on the positions of the particles at the end of centrifugation. Uncertainty in the temperature inside the rotor. Computer simulation of the sedimentation of particles in sucrose gradients shows that the temperature of centrifugation will have an extremely marked effect on the sedimentation of particles. An error of 1" in the measurement will cause an error of over 2.5::: in the estimation of the value of the sedimentation coefficient of a particle sedimenting in that gradient. This problem will arise with all density gradient materials which give solutions with an appreciable viscosity but will not occur with salt gradients. At present, most preparative centrifuges have rather poor temperature control systems which are especially prone to error when zonal rotors are used (see Ch. 5). Therefore, errors in the estimation of the temperature of the contents of the rotor during centrifugation are likely to be the major cause of error in the calculation of sedimentation coefficients.
CHAPTER 9
Future prospects for density gradient centrifugation The technical development of centrifuges has been rapid, following The increasing interest in subcellular biochemistry. The most fundamental developments have been the introduction first of dynamically loaded zonal rotors and then of zonal rotors capable of successful reorientation of gradients. In recent years, the advertising of centrifuge manufacturers has been concentrated on publicising the maximum speed of their latest centrifuge. Concurrently, however, there has been an increasing awareness that biological structures may be damaged by the high hydrostatic pressures which are produced at such speed. We may hope to see less emphasis on ultimate speed in the future and more emphasis on the price and convenience of rotors. Among the uses of the ultracentrifuge in biochemistry, the most important is the production of purified preparations of subcellular structures. Up to this time, biochemists have worked predominantly with liver and other tissues available in large quantities. In the future, we may expect more emphasis on the biochemistry of less bulky, but equally important tissues. This will demand the development of micromethods for analysis and may well require the introduction of a new generation of centrifuge rotors. In addition, it is likely that biochemists will become increasingly unwilling to use the impure fractions prepared by differential pelleting, with the result that density-gradient centrifugation in zonal rotors will become more of a routine preparative method rather than a sophisticated research tool. If this is the case, then there will be an increasing emphasis on the use of the low-speed zonal rotors which are most suitable for separating the larger organelles of the cell. Finally, biochemists 254
Ch. 9
FUTURE PROSPECTS
255
are now realising that it is possible to separate the different cell types that make up a tissue and those of us who have had a guilty feeling for some years about the dangers of working with true ‘homogenates’ will have to do something about it. The result again will be a reemphasis on low-speed methods. With these ideas in mind, we will survey possible developments in the design of centrifuges, of their rotors, and of ancillary equipment. However, we believe that these developments will be of little significance compared with the improvements in manipulative technique which are introduced as biochemists become less willing to accept poor results as inevitable.
9.1. Centrifuge design Centrifuges have not really been invented, but rather have evolved over the years. The basic design has changed little, although gradually new features have been added with new demands from the research worker. The mechanical requirements even to this day are rather simple. Because of the heavy work load put on to preparative ultracentrifuges by most users, the most important requirement is reliability. Zonal rotors have introduced particular problems here, notably the contamination of the bearings, lubricants and the vacuum system with material picked up either by condensation or by spillage during the loading and unloading procedures. Although zonal rotors have been available for more than ten years, manufacturers have been slow and reluctant to redesign their machines to cope with these problems. The changes which we would most like to see in present day centrifuges would be derived from greater consideration to the convenience of the user. Temperature control is undoubtedly the most important single problem when zonal rotors are used (see $5.1.3.4).The main difficulty is to measure and control the temperature of a rotor spinning in a high vacuum. The temperature control systems fitted to present day centrifuges work very well with angle-head and swing-out rotors, but Siihipcr i n i k p ~ 287
256
DENSITY GRADIENT CENTRIFUGATION
break down with zonal rotors. This is due to the inevitable condensation of water on the bowl of the centrifuge during loading of the gradient. In consequence, the vacuum is poor at the beginning of the run and affects the temperature control. So far only one manufacturer has taken steps to overcome this serious problem. We feel that insufficient thought has been given, when designing present day centrifuges, to the convenience of the user. An example is the ease with which liquid spilt on top of a centrifuge can flood in to the bowl. Centrifuges, like cars, should be designed as far as possible to cope with accidents. Other points to which manufacturers could usefully pay attention are the design of the oil recirculation system, prevention of contaminants accumulating in the vaccuum pump, and access to components (e.g. motor brushes) which need periodic renewal. In these, and many other cases, if centrifuge designers were to take greater note of the comments of their own service departments, the product would be considerably improved. It has been assumed that centrifuges of the future would continue to be driven by electric motors sited outside the vacuum bowl and driving the rotor through a vacuum tight seal. There are many advantages in placing a synchronous motor inside the vacuum bowl; analytical ultracentrifuges, which are driven by this system, are almost completely silent. The low cost of electronic systems should now make this a practical proposition for preparative ultracentrifuges, but we are likely to see a greater use of turbine drives which are very compact and give excellent control of acceleration and deceleration. The air turbine drive used by Electro-Nucleonics on their K and RK machines works very well but is noisy. The oil turbine drive, which incidentally was the system used by many of the early workers (Svedberg and Pederson, 1940), is probably the drive system of the future. The turbine can be very small, and the pump driving the system can easily be housed inside a modern centrifuge. The electronics in such a machine should be simpler than in the present motor-driven models. (While this article was being prepared, Sorval introduced an Ultracentrifuge with an oil turbine drive (Sorval OTD 2.)
Ch. 9
257
FUTURE PROSPECTS
9.2. Developments in centrifuge rotors As we have suggested, centrifuge manufacturers have so far concentrated on producing higher and higher speed rotors with equally rapidly escalating costs. While this process seems likely to continue under its own momentum for a while, it has become of less interest to most biochemists, firstly because of the risk of damage to subcellular structures by the high hydrostatic pressures which may be generated, and secondly because the small particles which could be separated in very high speed rotors can already be fractionated efficiently by gel electrophoresis. Of much more interest to biochemists than the '500,000 g plus' rotors, would be cheaper and more durable rotors and an increased capacity in medium speed rotors capable of generating centrifugal fields of between 100,000 and 200,000 g. It is unlikely that these objectives could be met with conventional materials like aluminium and titanium. New composite materials, such as those containing carbon fibres would be well suited to the manufacture of centrifuge rotors in view of the extremely favourable strength to weight ratios and the appafent lack of fatigue. A further advantage of composite materials is that much less damage would be caused by rotor failure, consequently the thickness of the armour around the centrifuge bowl could be reduced. However, a major problem will be standardising the quality of rotors made from such materials. The use of composite materials would almost certainly neccessitate a radical redesign of centrifuge rotors and this will inevitably be slow and expensive. Such a redesign could have many advantages to the user. Sufficient is known about the properties of subcellular structures and macromolecules for rotors to be designed with more attention to their structure than heretofore. Thus the highest speed rotors, those generating more than about 300,OOOg are used mainly for the isopycnic banding of subcellular structures and macromolecules. A major problem with such high-speed rotors is the damage which may be caused by high hydrostatic pressures (Ch. 8). Rotors with shorter sedimentation path could be used at much higher speeds than present S l , l ~ / ' ' II l r h l p
147
258
DENSITY GRADIENT CENTRIFUGATION
day rotors. Thus a rotor with a 1 cm sedimentation path could be spun more than three times as fast as a rotor with a 10 cm path with no greater risk of damage to the material being separated. In addition, the shallower gradient space would in itself give more rapid movement of particles to their equilibrium positions. The experiments of Beaufay and his colleagues (Leighton et al. 1968; Beaufay et al. 1974) using their own E 40 rotor have shown that excellent results can be achieved with a rotor with a gradient space only 1.O cm deep. We seriously feel that this type of rotor, with suitable modifications to its feed head, could have a bright future. Whatever design is adopted for very high-speed rotors (>60,000 revs/min), they are likely to be expensive. However, new materials could reduce the cost of rotors capable of producing up to,300,000 g. There seems no reason, if composite rotors were to be manufactured, why the interior of angle head rotors should be made of the same material as the load-bearing outer shell. A possible design might be rotors consisting of an outer shell made from a strong material such as carbon fibres, provided with a series of interchangable cores made from a light weight, incompressible plastic. By ‘plugging in’ the right core one could have effectively a wide range of angle-head and zonal rotors. The latter could, with the right core, be either of the dynamically or the statically unloaded type. This system is perfectly feasible as is shown by the RK 11 tube core which can be inserted in to the titanium core of the RK zonal rotor manufactured by Electro-Nucleonicsand which allows the latter to be used as an angle head rotor. A problem encountered when using zonal rotors is that only a single sample can be separated at any one time. It is true that pilot work may be carried out on swing bucket rotors, but these are mechanically inefficient, in that the whole rotor must be built to take a strain which is in fact only exerted over a small portion of the circumference. In addition, it is unlikely that swing-out rotors could ever be built out of composite materials as the shape is too elaborate. A development which would be very useful is therefore the introduction of zonal rotors which could separate more than
Ch. 9
FUTURE PROSPECTS
259
one sample at a time. This would allow the separation of quantities ofmaterial intermediate between those separated on present day zonal and swing-out rotors, together with the side by side comparison of samples from different sources. Zonal-type rotors capable of separating more than one sample at a time could be designed in two ways depending on whether the sample and gradient were to be dynamically loaded and unloaded or loaded and unloaded statically and reoriented at the start and finish of the centrifuge run. The complicated coaxial seal system required for dynamic loading of a multi-compartment rotor renders reorientation a much more attractive solution. As with the other developments which we have discussed, this development has been shown to be perfectly feasible, in this case by the successful use of reorienting gradient zonal rotors. The separate compartments required for a multi-sample reorienting zonal rotor could be made as separate plastic inserts into which gradient, and even sample, could be loaded on the work bench and which could then simply be slotted in to the rotor. It should be relatively easy to make these inserts and they could even be designed so that they could be used in the present series of B-XIV and B-XV rotors. Provided that the centrifuge manufacturers were prepared to incorporate a speed control which would give the very slow and even acceleration required for reorientation, one could see rotors of this type eventually replacing conventional swing-out rotors. (See Note added in proof, p. 261.)
9.3. Developments in ancillary systems When the transparent A-XII, HS and Z-15 rotors are used, much information can be gained simply by watching the movement of the bands during sedimentation and problems with uneven unloading of the different segments (see $5.1.3.5) can be detected by observing the bands during unloading. Although a fluorescent light, such as is used with the A-XI1 rotor is extremely useful, we find that a stroboscopic lamp is more useful for some purposes, especially for observing aggregation in the rotor or uneven unloading of different Suh/w! side1 p 287
260
DENSITY GRADIENT CENTRIFUGATION
segments. A problem, however, is that slight changes in the speed of the rotor make it difficult to ‘freeze’ the bands. This problem would be avoided if centrifuge manufacturers fitted an external ‘trigger’ to flash a stroboscopic lamp once in each revolution. We have discussed the equipment needed for monitoring density gradients in $5.1.4. The greatest needs at the moment are for compact and reasonably priced spectrophotometers and densitometers suitable for use in monitoring density gradients, However if the monitoring of homogenates from human biopsy specimens (for example from lysosome storage diseases) became more widespread (Seymour et al. 1974), it would be more efficient to monitor the enzyme activities during displacement of the gradient rather than collect fractions for later, off line, analysis.
9.4. Uses of centrifugal methods We would expect to see a greater division, in the coming decade, between biochemists who use density-gradient centrifugation simply as a tool for the preparation of large amounts of material of subcellular fractions for further study and cell biologists who are interested in studying the properties of the subcellular fractions. Paradoxically, it is the former group who are likely to make more use of the zonal rotors which have always been thought of as the most complex equipment used in centrifugation. The great advantage of zonal rotors is the large amount of material which can be separated at one time and, as we hope we have made clear in this article, zonal rotors are not really difficult to operate while the use of density-gradient centrifugation, as opposed to differential pelleting, will provide the biochemist with much more homogeneous fractions. It is more difficult to predict how the cell biologist will use centrifugation The subcellular components of major tissues such as liver have been separated and their enzymology and functions largely explained. These studies have, however, all been carried out on homogenates of whole tissues. We would expect a greater emphasis in the future on the study of the individual types of cell that make up a
Ch. 9
261
FUTURE PROSPECTS
tissue. Even when simple and reproducible techniques have been developed for the separation of these different cells, it will not be possible to apply present methods for separating subcellular fractions directly, as the actual amounts ofcells separated will, generally be very small. A similar problem will be faced by those who move on from studying liver to studying the smaller organs of the body. There will, therefore, have to be a renewed emphasis on micro methods for subcellular fractionation and for analysis of the separated subcellular fractions. Initially it is likely that swing-out rotors will be used for such studies but later we may expect to see the introduction of commercial rotors resembling the E 40 rotor designed by Beaufay (see 55.1.2.9). For rate-zonal separations, swing-out tubes are likely to be used but will increasingly be supplemented by multicompartment reorienting zonal rotors such as are discussed above. Whatever the future developments in hardware, it is clear that density-gradient centrifugation will be more widely used in the coming years. It is difficult to remember that the technique was introduced only twenty years ago, and that much of the time since then has been spent in learning the potentialities and limitations of the technique and in finding out just what there may be in cells which can be separated. These points are now generally understood and we hope that our article will help others to judge where density gradient centrifugation may be helpful in their work.
Note added in proof Recently rotors in which the tubes are held exactly vertical have been introduced by IEC and by Beckman-Spinco. These rotors are designed for density gradient centrifugation and may be used either for rate zonal or for isopycnic separations. The rotors must be used with centrifuges giving very slow and controlled acceleration and deceleration to permit reorientation of the gradient and the sample band. The advantages of these rotors over swing-out rotors are the more rapid separation which is achieved and the increased number and volume of samples which may be handled. The major disadvantage is the somewhat poorer resolution. Suh,r c I
U ~ P p. \
287
Acknowledgements
We would like to thank Dr. E. Reid, the Director of the Wolfson Bioanalytical Centre, for his support and encouragement. We would also like to thank our past and present colleagues with whom we have studied density gradient centrifugation and especially Dr. B.M. Mullock for reading and commenting on a large part of the manuscript. We would also like to thank workers in other laboratories, especially Dr. R.G. Wallis and his colleagues of the Microbiological Research Station, Porton Down; Dr. K. Zechel of the Max Planck Institute for Biochemistry, Munich, for discussions on centrifuge systems with which we were unfamiliar; and Mr. J.B. Rowse of English China Clays Ltd for permission to quote his unpublished results. We have made free use of results published by other laboratories and would like to thank the authors and their publishers for permission to reproduce Copyright material. Individual references are made in Figure legends; detailed references are found at the end of the article. We would also like to thank manufacturers, especially MSE Scientific Instruments and Beckman-RIIC Ltd. for supplying photographs and diagrams of propriatory equipment. Finally we would like to thank all our friends from evening symposia, and especially Dr. G. Cline, Dr. J.A.T.P. Meuwissen and Dr. J. Steensgard, and hope we have not forgotten too much of what they taught us.
262
Appendices
APPENDIX I Manufacturers and suppliers of centrifuges, ancillary equipment and special chemicals.
Manufacturers mentioned in 5.1.4 and in Tables 3.1 and 4.2 are listed. Anachem Ltd., 20a North Street, Luton, Beds, U.K. Atago Opt. Works Co. Ltd., 2-16-1 1 Yushima, Bunkyo-ku, Tokyo, Japan. Baird & Tatlock Ltd., Freshwater Rd., Chadwell Heath, Romford, Essex, U.K. Beckman-RIIC, Eastfield Industrial Estate, Glenrothes, Fife KY7 4NG, U.K. Beckman Spinco, 11 17 California Av., Palo Alto,’ CA 94304, U.S.A. Braun Apparatebau, 3508 Melsungen, West Germany. (British agent : Shandon-Southern Inst. Ltd.) Buchler Instruments, 1327, 16th. Street, Fort Lee, New Jersey, U S A . (British agent: Baird & Tatlock) Cecil Instruments, Trinity Hall Industrial Estate, Green End Road, Cambridge CB4 ITG, U.K. ChemLab Instruments Ltd., Hornminster House, 129 Upminster Road, Hornchurch, Essex, U.K. Christ - see Hereus-Christ. F. Copley & Sons, Colnwick Industrial Estate, Nottingham NG4 2ER, U.K. Damon/IEC, 300,2nd Av., Needham Heights, MA 02194, U.S.A. (British agent Searle Instruments) Dawe Instruments Ltd., Concord Road, Western Avenue, Acton, London, W3 OSD, U.K. DuPont-Sorvall, Peck’s Lkne, Newtown, CT 06470, U.S.A. DuPont-SorvaU (U.K.)Ltd, 64 Wilbury Way, Hitchin, Herts, U.K. DuPont, Ind. & Biocbem. Dept., Wilmington, Delaware, 19898, U.S.A. Electro Nucleonics Inc., 368, Passaic Av., Fairfield, N.Y. 07006, U.S.A. Electronic Applications (Comm.) Ltd., 98 Ickleford Road, Hitchin, Herts, U.K. Electroplan Ltd., P.O. Box 19, Orchard Road, Royston, Herts. SG8 SHH, U.K. Ekco (Rubber) Ltd., 14-16 Great Portland Street, London, WIN SAB, U.K.
263
264
DENSITY GRADIENT CENTRIFUGATION
Gallenkamp & Co. Ltd., P.O. Box 290, Technic0 House, Christopher Street, London, EC2P 2ER. U.K. General Radio, 300 Baker Av.. Concord. MA 01742, U.S.A. Gilford Inst. Labs., Oberlin. OH 44074, U.S.A. Gilford Inst. Ltd., 188 Martin Way, Morden, Surrey, U.K. Gilson Med. Electronic Inc., P.O. Box Box 27, Middleton, Wisconsin, 53562, U.S.A. (British agent Anachem Ltd.) Grant Instruments Ltd., Barringtoti, Cambridge, CB2 5QZ, U.K. Hellma (England) Ltd., 370A London Road, Westcliff-on-Sea, Essex, U.K. Henley Medical Supplies Ltd., Clarendon Road, Hornsey, London N8 ODL, U.K. Hereus-Christ, 3360 Osterode am Marz, Postfach 1220, West Germany. (British agent V.A. Howe). A.R. Horwell Ltd., 2 Grangeway, Kilburn High Road, London, NW6 NB, U.K. V.A. Howe & Co. Ltd., 88, Peterborough Road, London SW6, U.K. I.E.C. (International Equipment Co.) - see Damon/l.E.C. ISCO (Instrument Specialities), Box 5347, Lincoln, NE 68505, U.S.A. (British agent MSE Scientific Instruments Ltd.) Heinz Janetzki, Leipzig, G.D.R. (British agent F. Copley and Son). Lab. Data Control, Box 10235, Riviera Beach, Florida 33404, U.S.A. LKB Instr. Ltd., LKB House, 232 Addington Road, Selsdon, Croydon CR2 8VD, U.K. May & Baker, Dagenham, Essex, RMlO 7XS, U.K. Mallinckrodt Chemical, St. Louis, Missouri, U.S.A. Metering Pumps Ltd.,49-51, Uxbridge Road. Ealing Broadway, London W5 5SD, U.K. MSE Scientific Instruments Ltd., Manor Royal, Crawley. Sussex, U.K. Nyegaard & Co. A/S, Postbox 4220. Oslo 4, Norway. Oxford Instrument Co. Ltd., Osney Mead, Osford, OX2 ODX, U.K. Perkin Elmer, 800 Main Ave. Norwal CT 06856, U S A . Perkin Elmer Ltd., Post Office Lane, Beaconsfield, Beds, U.K. Pharmacia (Gt. Britain) Ltd., Paramount House, 75, Uxbridge Rd., London, W5 5SS, U.K. Phoenix Precision Inst. Co., RT 208, Gardiner, NY 12525, U.S.A. (British agent Techmation Ltd.) Paar KG, Anton, Graz, Austria. (British agent Staton-Redcroft Ltd.) Portex Ltd., Hythe, Kent, U.K. (Deals only via agents all lab. equipment suppliers.) Pye-Unicam Ltd., York Road, Cambridge, CBI 2PX, U.K. Rikadenki Kogyo Co. Ltd. 9-1 Chome Chine-cha, Meyuro, Tokyo, Japan, (British agent TEM Sales Ltd.) Sage Instruments, 11 Blackstone St. Cambridge, MA 021 39, U.S.A. (British agent A.R. Honvell)
APPENDICES
265
Schering Chemicals Ltd., Burgess Hill. Sussex, U.K. Schering, A.G., 1000 Berlin 65, Mullerstrasse 170-172, Western Germany. Scientific Supplies Co. Ltd., Scientific House, Vine Hill, London ECI 5EB, U.K. Searle Instrument Division, West Rd. Temple Fields Industrial Estate, Harlow, Essex, U.K. Shandon Southern Ltd. Frimley Road, Camberley. Surrey, U. K. Stanton-Redcroft, Copper Mill Lane. London S.W. 17 OBN, U.K. Smiths Industries Ltd. (Instruments) Waterloo Road, London W3 05D.U.K. Sorvall see DuPont-Sorvall. E.R. Squibb & Sons, Regal House, Twickenham, Middlesex, U.K. Techmation Ltd., 58 Edgeware Way, Edgeware. Middx. HA8 87P. U.K. TEM Sales Ltd., Gatwick Road, Crawley, Sussex, RHlO 2RG, U.K. Uniclip Ltd., 100 Royston Road, Byfleet, Surrey, U.K. Varian Ass. Instruments, 61 1 Hansen Way, Palo Alto, Ca. 94303. U.S.A. Varian Ass. Ltd., Russell House, Molesey, Walton-on-Thames, Surrey, U.K. Waters Associates, 165 Mapel Street, Milford, MA 01757, U.K. Waters Associates (Instruments) Ltd., Vauxhall Works, Greg Street, Saint Reddish, Stockport. Cheshire. SK5 7BR. U.K. Winopal Forshung, 3004. Isernhagen, NB Sud/Hanover W. Germany. Winthrop Laboratories, Winthrop House, Surbiton. Surrey. U.K.
-
APPENDIX I1 Glossary of terms used in density gradient centrifugation Analytical ultracentrifuge A centrifuge equipped with rotors holding small cells with glass o r quartz windows which permit photometric monitoring of the sedimenting particles (see $1.2.1.) Artery,forcepps (also known as Haemostats or Spencer-Wells forceps) Used in surgery for clamping arteries, the self locking action being very useful for clamping plastic tubing. Band (or zone) A discrete region (annular in zonal rotors) containing particles sedimenting together. Batch zonal rotor A rotor which processes one sample at a time. As opposed to continuous flow rotors (q.v.) Bulk jlow see Hydrodynamic instability. Capacity q/’ a hand The maximum concentration of particles which will sediment through a particular gradient without tumbling (q.v.). Coaxial seal Seal which gives access to a spinning rotor and yet ensures that the fluid in the lines leading to the centre and edge of a zonal rotor are kept separate. Composed of a stationary seal and a rotating seal which are kept in firm contact by spring tension (see Fig. 5.4). Sirhiecr itidcx p . 2x7
266
DENSITY GRADIENT CENTRIFUGATION
Continuous-flow rotor A category of rotors, including Flo-band rotors (q.v.) into which a sample may be continuously fed while the rotor is spinning at full speed. Core Central part of a zonal rotor which contains the channels connecting the seal to the centre and, via the septa, to the edge of the rotor. (see $5.1.3.3 and Fig. 5,15). Cross leak A leak in which the two solutions which are normally kept separate by the coaxial seal leak into one another without entering the rotor (Fig. 5.15). Cushion (also called underlay) Solution denser than the densest part of the gradient whose purpose is to prevent any particles from pelleting on the outer wall of a zonal rotor. Density gradient A solution whose density progressively increases with volume. Dynamic loading/unloading Loading or unloading of a spinning rotor. Made possible by the coaxial seal. Displacing solution (or piston) A solution as dense as, or denser than, the cushion (9.v.) which is used to displace the density gradient. Droplet sedimentation see Sedimentation in droplets. Equilibrium banding see Isopycnic banding. Edge of rotor The wall of the rotor chamber which is analogous to the bottom of a conventional centrifuge tube. The periphery of the zonal chamber. Edge unloading Unloading of the gradient and the separated bands from the edge of the rotor and not from the centre, which is the more normal method (see B-XXIX rotor, $5.1.2.6). Feed head This contains both elements of the rotating seal plus the locating bearing (see Figs. 5.3 and 5.5). Flo-band rotors A zonal rotor, containing a density gradient, across which a sample is continuously pumped. As the sample passes over the surface of the gradient some particles enter and sediment to their isopycnic banding positions. The remainder of the sample exits as effluent. Flotation Movement of particles against the centrifugal field into regions of progressively decreasing density. The opposite of sedimentation. Flotation coefficient Analogous to sedimentation coefficient (q.v.). The sample to which the flotation coefficient is referred must be carefully specified. Gradienfmaker Machine or instrument which can make a density gradient of-a predetermined shape or profile. Gradient profire The shape of the density gradient expressed as the change of density against volume. Guard tray (Rotor shield) This tray, usually made of perspex (lucite) prevents the zonal rotor being accidently touched while it is spinning. It also locates the feed head on to the zonal rotor (Fig. 5.7). Inversion When a liquid of higher density lies above (or nearer to the centre of rotation than) a liquid of lower overall density. This condition can only exist momentarily.
APPENDICES
261
Integrator An electronic instrument which automatically integrates the angular velocity of the rotor and the time of centrifugation. Isokinetic gradient A gradient through which a particle moves at constant speed until it reaches the bottom of the tube. In such a gradient, in the tube of a swing out rotor, the sedimentation coefficients of particles of a given density are accurately proportional to their positions on the pump-out trace. Isovolumetric gradient A gradient in which a zone of particles passes through the same volume of gradient in any given interval of time. Equivalent to an isokinetic gradient (q.v.) in the case of a swing-out rotor. In the case of a zonal rotor the isovolumetric gradient differs from the isokinetic gradient due to radial dilution. On displacement of an isovolumetric gradient, the sedimentation coefficients of particles of a given density are accurately proportional to the volume of gradient through which they have passed. fsopycnic banding Separation on the basis of particle density. Centrifugation is continued until all particles have reached a liquid of their own density and hence have no weight and move no further. Hydrodynamic instability (or bulk flow) Broadening of zones by the influx of water (see 52.1.2). Occurs when large amounts of particles are loaded on to a gradient. Jetting Over-rapid injection of sample o r overlay in to the rotor, resulting in a jet of solution emerging from the feed channel and mixing with the light end of the gradient. Loading- Filling the zonal rotor with gradient and sample. Loading speed The ideal speed for loading and unloading a particular zonal rotor (see 551.3.I). Monitoring Continual measurement and recording of some parameter (e.g. absorbance or density) during loading or unloading. Overlay A solution of lower density than the sample which is used to push the sample away from the core of a zonal rotor. Overloading The result of too much, or too concentrated a sample. The sedimenting zones are broader than would be expected from diffusion due to hydrodynamic instability (q.v.) or sedimentation in droplets (q.v.). Partial specific volume For any compound the volume of water (in ml) displaced when 1 g of the compound is dissolved in a large volume of water. Peak Form in which a band is detected on a recorder trace. Perfusor A powered syringe which permits a constant and pulse-free injection of the sample in to a zonal rotor. Rate sedimentation Separation of particles on the basis of their sedimentation rates (normally proportional to particle size). Reorientation (Reograd) A gradient loaded with the rotor at rest has the layers of gradient arranged horizontally. When the rotor is slowly accelerated the layers move without mixing into a vertical position aligned along the centrifugal field - reorient. S t r h p r mdc.x p. 2U7
268
DENSITY GRADIENT CENTRIFUGATION
Recovery For an enzyme or chemical constituent the sum of the activities (or amounts) in all the fractions recovered from the gradient as a percentage of the material loaded. Reverse Wedge A sample region formed in a continuous density gradient in which the sample concentration is also a gradient, but in the reverse direction (see 85.1.3.3). Resolution The effective separation of two particles. Run A single spin with a rotor, up to operating speed and back to rest. Rotating seal Part of the coaxial seal (q.v.) which spins with the rotor. It is made of Rulon, a filled polytetrafluoroethylene material and is sometimes referred to as the Rulon. Seal see coaxial seal. Sector Part (normally a quadrant) of the chamber of a zonal rotor between the septa. Sectorial unloading Uneven unloading of the sectors resulting in loss of resolution (see Fig. 5.19). Sedimentation coefficient A measure of the speed of sedimentation of a particle in a standard medium, normally taken as water at 20 "C. The sedimentation coefficient is normally measured in Svedbergs (S) which are defined as IOl3 x the velocity of sedimentation (in m/sec) in a gravitational (or centrifugal field) producing an acceleration of 1 m/secZ. It is principally determined by the particle size and shape. Sedimentation in droplets Broadening of a sample zone due to differences in the diffusion coefficients of the sample material and the gradient solute (see 92.1.2). Septa Vertical divisions in the annular chamber of a zonal rotor which are designed to stop the gradient from swirling. Usually 4 are present. They also have channels which carry gradient solutions to the edge of the rotor. Stationary seal Part of the coaxial seal with the feed tubes which remains stationary (see Fig. 5.5). Streaming see Sedimentation in droplets. Svedberg (S) Unit of sedimentation coefficient. Swing-our rotor A rotor in which the tubes are carried in hanging buckets which swing up from a vertical to a horizontal position as the rotor accelerated. Temperature gradient Differences in temperature across a density gradient. Tumbling The bulk movement of particles and gradient solute together. Occurs when theconcentrationofparticles becomes so great that there is a density inversion (q.v.). Underlay see Cushion. Unloading Undisturbed removal of a density gradient from a zonal rotor or a swingout tube. Usually achieved by displacing the gradient with a dense displacing solution. Vacuum cap A cap for sealing a zonal rotor after the feed head is removed. The top part of the cap is free to rotate so that it can be gripped safely while the zonal rotor is still spinning. Vanes see Septa. Wall, rotor see Edge of rotor.
269
APPENDICES
Zonal rotor A rotor consisting of an annular chamber with its surrounding walls which spins on its vertical axis. A density gradient can be pumped in and out of the rotor while it spins. The sample is separated into a series of annular zones, hence the name (see Fig. 5.1). Zone see Band.
APPENDIX 111 Density and sedimentation coeficients of rat liver cell organelles Organelle
Nuclei
Sedimentation coefficient, S (referred to water at 20" unless otherwise noted) -
Mitochondria
10,0004~s
Lysosomes
4,0004,5
Micro bodies (Peroxysomes) Golgi fragments Plasma membrane
4,0OO4~'
Monomeric ribosomes
79'3 83 3"
Polysomes dimers trimers tetramers pentamers hexamers heptamers octamers nonamers decamers Small ribosome subunit EDTA-derived Active in protein synthesis
7 -7
+-
+
123 4'' 154 k 4 " 183 f 5" 211 f 4 237 f 3 257 _ +6 282 299 & 5 316
Density, p
In 0.25 M Sucrose 1 .067Is2 Isopycnic 1.352,3 In 0.25 M Sucrose 1 .1256 I sopycnic 1.196 In0.25MSucrose 1.136 Isopycnic 1.206 In 0.25 M Sucrose 1. I l6 Isopycnic 1.2S6 Isopycnic 1.138 I n 0.25 M Sucrose 1.05' Isopycnic 1.14- 1.18'" In Metrizamide 1 .2712 In sucrose 1 .4313 In CsCl 1.5513
as monomers as monomers as monomers as monomers as monomers as monomers as monomers as monomers as monomers
28.614
InCsCl
40.914
I n CsCl
1.59314 1.551'4 Seb,rrr erder p 287
270
DENSITY GRADIENT CENTRIFUGATION
Organelle
Sedimentation coefficient, S (referred to water at 20 unless otherwise noted)
Large ribosome subunit EDTA-derived Active in protein synthesis Glycogen particles Ferritin granules Smooth endoplasmic reticulum Rough endoplasmic reticulum 1 2
3
4
5
8
9
49.914
In CsCl
1.60214
59.1 I4 40- 10,000' 6316
In CsCl In CsCl InCsCl
1.61414 1.6515 1.6516
-
In sucrose I . 1616
1000
500-3000
In sucrose 1.1% I . 2 P
Approximate value. Johnston et al. (1968) Nuclei are freely permeable to sucrose and, with rat liver nuclei, the density pp in a sucrose medium of density p,,, is given by pp = (0.898 x pm + 0.138). Considerable variation in density of nuclei from different tissues and between nuclei of the same tissue of different animals (Hilderson 1974). Referred to 0.25 M sucrose at 0 "C. For approximate sedimentation coefficients in water multiply by 2.5. Such figures are meaningless as particles are damaged by hypotonic solutions. Beaufay et al. (1959). Beaufayet al. (1964).The density of membrane-bounded particles is greatly affected by the surrounding medium. The equation relating particle density to the density of the surrounding medium is given in $2.3 equat. 2.25. The constants for mitochondria, lysosomes (as indicated by acid phosphatase) and microbodies (as indicated by acid phosphatase) and microbodies (as indicated by uric acid oxidase are
Mitochondria Lysosomes Microbodies 7
Density, p
U
P
0.110 0.102 0
0.76 1.05 2.58
Pd
.2 ,226 ,248
d
0.36 0.25 0.22
Very considerable variation of size with the method used for homogenisation. Fleischer and Fleischer (1970). Isopycnic banding density of plasma membrane fragments from Ehrlich ascites cells in isotonic Ficoll gradients (Wallach 1967). Experiments of House and Weidemann (1970) suggest that liver plasma membrane fragments have a similar density. The
APPENDICES
271
isopycnic banding densities of particles in isotonic Ficoll gradients may generally be taken as being the same as the effective density of the particles in 0.25 M sucrose. 10 Plasma membrane fragments of microsomal size are heterogeneous in density (Norris et al. 1974). Large sheets of plasma membrane band at density 1.18 (see Hinton 1972). 11 Pfuderer et al. (1965). 12 Hinton et al. (1974b). 13 Petermann (1964). 14 Hamilton et al. (1971). 15 Barber et al. (1966). 16 Approximate data calculated from our experiments. Note that the density of rough endoplasmic reticulum fragments varies with the number of ribosomes and that of ferritin with the iron content.
APPENDIX IV Theory of preparation of density gradients
For the purposes of this discussion, gradient makers may be divided into two classes. In method A, a liquid of concentration Cadd is pumped from a reservoir into a mixing vessel at a rate rl and liquid is extracted from the mixing vessel at a rate r2. In method B, the two liquids are introduced into a mixing area at rates rl and r 2 . The gradient is formed by varying the rates rl and r2, usually in such a way that r l + r2, the total rate of delivery of the gradient, is constant. Method B is theoretically very simple, but elaborate electronic or mechanical linkages are required to control the pump rates. However, once these engineering problems have been overcome, there is no limitation to the shape of the gradient which may be prepared, so that method B is the one normally chosen for expensive propriatory gradient makers (Table 4.2 and 45.1.4.1. It is easy to build density gradient makers working on principle A in the laboratory, but difficult to predict and control the shape of the gradient which is formed. It is therefore necessary to look at the theory of such gradient makers in some detail. If the initial volume and, concentration of the liquid in the mixing chamber are M , and C, and liquid of a concentration Cadd is added from the reservoir, then at any time t the total colume of gradient pumped is V
=
r2t
(A4.1)
and the volume of the liquid in the mixing chamber is M
= M,
+ (rl - rz)/
(A4.2) Svh/<wind'a p. 287
272
DENSITY GRADIENT CENTRIFUGATION
By equating the change in mass of liquid in the mixing chamber to the difference in mass between the liquid pumped into the mixing chamber and that pumped out in time dt if dC and dM are the change in concentration and volume of the liquid in the mixing vessel then
I f f , dt, M and dM are expressed in terms of V and dV by equat. 1 and 2, equat. 3 can be expanded and limits taken, to give
If new variables r written as
=
rl/r2 and v
=
V / M , are defined, the last equation may be re-
At the start of the gradient Y = v = 0 and C = C, where C, is the initial concentration of liquid in the mixing vessel. Thus, on integrating eq. 4,
which may be rewritten
or defining a new variable c
=
(C - Co)/(Cadd- C,)
c=l -[I + ( r - l)v]-r'(r-I) If r
=
(A4.5)
1 then eq. 4 integrates to give the equation for an exponential gradient maker
(A4.6)
i.e.
APPENDICES
213
In other words, when the rate of flow into the mixing vessel is half the rate of flow from the mixing vessel. then a linear gradient will be formed. This last situation occurs in a gradient maker of the type shown in Fig. 4.1. which consists of two cylinders of equal diameter interconnected at their base. If a volume d V is removed from one cylinder, the height of liquid in that cylinder will be reduced. The difference in hydrostatic pressure between the two cylinders will cause liquid to flow from the other cylinder. As the two cylinders have equal cross-sectional areas, when the liquid is a t equilibrium (i.e. when the height in the two cylinders is the same) the volume in the two cylinders will be the same. Thus if dV is removed from one cylinder, then 4dVwill flow from the other cylinder to restore equilibrium. Thus, in terms of eq. A4.1, r I = 4 r l ; or in eq. A4.5 r = f. With a closed mixing vessel of the type described in Fig. 4.3, r l must equal r z , a s by definition the volume of liquid in the mixing vessel is constant and hence the rate of flow from the mixing vessel will equal the rate of flow into the mixing vessel. The gradient produced by such a gradient maker is described by eq. A4.6. The major point to note about equations A4.5 and A4.6 is that the concentration depends o n v, the volume of gradient formed as a proportion of the volume of liquid initially placed in the mixing vessel. Thus even when r is fixed, as in an ‘exponential gradient maker*’ quite marked variations in shape may be obtained by varying the volume of liquid in the mixing chamber. In fact. quite adequate variation can be achieved to enable isokinetic gradients to be prepared from the tubes of swing-out rotors or isovolumetric gradients for zonal rotors. Details of the volumes of liquid to be placed in the mixing vessel and of the densities of the solutions have been published for isokinetic gradients in the tubes of IEC and Beckman swing-out rotors (Noll 1970).
*
An ‘exponential’ gradient maker is the name often given to a gradient maker with a closed mixing vessel which produces gradients described by eq. (A4.6). Siih,wI w d m 11 287
References
AAS,M. (1973) Abstracts 9th Internat. Congr. Biochem., Stockholm p. 31. ALLFREY, V. (1959) In: J. Brachet and A.E. Mirsky (eds.) The Cell Vol. 1 (Academic Press, New York) pp. 193-290. M.. A. WIBO,D. THINES-SEMPOUX, H. BEAUFAY and J. BERTHET AMAR-COSTESEC, (1974) J. Cell Biol. 62, 717-745. ANDERSON, N.G. (1956) In: G. Oster and A.W. PaIater (eds) Physical Techniques in Biological Research, Vol. 111. Cells and Tissues (Academic Press, New York) pp. 299-352. ANDERSON, N.G. (1962) J. Phys. Chem. 66, 1984-1989. ANDERSON, N.G. (1966a) Nat. Cancer Inst. Mono. 21 (editor and article pp. 9-39). ANDERSON, N.G. (1966b) Science (Wash.) IJ4, 103-1 12. ANDERSON, N.G. (1968a) Quart. Rev. Biophys. 1,218-263. ANDERSON, N.G. (1968b) Anal. Biochem. 23, 72-83. N.G.( 1969) Anal. Biochem. 28, 545-562. ANDERSON, ANDERSON, N.G. and G.B. CLINE(1967) In: K. Maramorosch and H. Koprowski (eds.) Methods in Virology, Vol. 2 (Academic Press, New York) pp. 137-178. ANDERSON, N.G. and E. RUTENBERG (1967) Anal. Biochem. 21. 259-265. ANDERSON, N.G., C.A. PRICE,W.D. FISHER, R.E. C A N N I NandC.L. G BURGER (1964) Anal. Biochem. 7, 1-9. ANDERSON, N.G., H.P. BARRINGER, N. CHO.C.E. NUNLEY,E.F. BABELAY, R.E. (1966a) Nat. Cancer Inst. Mono. 21, 113-136. CANNING and C.T. RANKIN and P. MAZUR(1966b) Nat. Cancer Inst. Mono. 21, ANDERSON, N.G., J.G. GREEN 415-430. ANDERSON, N.G., D.A. WATERS,W.D. FISHER,G.B. CLINE,C.E. NUNLEY, L.H. ELRODand C.T. R A N K IJr. N (1967) Anal. Biochem. 21, 235-252. ANDERSON, N.G.. D.A. WATERS, C.E. NUNLEY, R.F GIBSON, R.M. SHILLING, E.C. DENNY, G.B. CLINE,E.F. BABELAY and T.E. PERARDI (1969a) Anal. Biochem. 32,460-194. Jr. (1969b) Anal. Biochem. 31, ANDERSON, N.G.. C.E. NUNLEY and C.T. RANKIN 255-271. (1969a) Biochem. Biophys. Res. Comm. 34,328-334. ANET,R. and D.R. STRAYER 274
REFERENCES
275
ANET,R. and D.R. STRAYER (1969b) Biochem. Biophys. Res. Comm. 37,52-58. ARBORCH. B., J.L.E. ERICSSON and H. G L A U M A(1973) N FEBS Letters 32, 190-194. ASPRAY. D.W. and T.G. PRETLOW (1975) Anal. Biochem. 66, 353-364. ATHERTON, R.S.. I.S. BOYCE,G.C. C L A Y - I Oand N A.R. THOMSON (1973) I n : E. Reid (ed.) Methodological Developments in Biochemistry, Vol. 3 (Longman. London) pp. 71-76. AUSTOKER. J., D. COXac-l A.P. MA.IHIAS (1972) Biochem. J. 129, 1139-1 155. AYUSO-PARILLA. M., E.C. HENSHAW and C A H I R S C H(1973) J. Biol. Chem. 248, 4386-439 3. BACH,M.K. and BRASHLER. J.R. (1970). Exp. Cell. Res. 61. 387-396. BACHOFEN, R. (1974) I n : E. Reid (ed.) Methodological Developments in Biochemistry, Vol. 4 (Longman, London) pp. 31-38. BAGGIOLINI, M.. J.G. HIRSCHand C. de DUVE(1969) J. Cell Biol. 40. 529-541. BAGGIOLINI, M..J.G. HIRSCHand C. de DUVE (1970) J. Cell Biol. 45, 586-597. BARBER. A.A., W.W. H A R R Iand S N.G. ANDERSON (1966) Nat. Cancer Inst. Mono. 21, 285-296. BARRINCER. H.P. (1966) Nat. Cancer Inst. Mono. 21. 77-11], BAUDHUIN, P., P. EVRARD and J. BERTHET(1967) J. Cell Biol. 32. 181-191. BAUER,J.H. and E.G. PICKELS(1940) I n : T. Svedberg and K.O. Pederson The Ultracentrifuge (Clarendon Press, Oxford) pp. 191-212. BEAMS.J.W., E.G. PICKLES and A.J. WEED(1934) J. Chem. Phys. 2, 143. BEAUFAY. H. (1966) La Centrifugation en Gradient de Densite (Thesis, Universitk Catholique de Louvain). BEAUFAY. H. (1969) I n : J.T. Dingle and H. Fell (eds.) The Lysosomes in Biology and Pathology (North Holland. Amsterdam) pp. 515-546. BEAUFAY, H.. D.S. BENDALL, P. B A U D H U I R. N , WATTIAUX and C. de D U V E(1959) Biochem. J. 73. 628-637. BEAUFAY, H., P. JACQUES,P. B A U D H U I N 0.2. , SELLINGER. J. BERTHET and C. de DUVE(1964) Biochem. J. Y2, 184205. BEAUFAY, H., A. AMAR-COSTESEC, D. THINES-SEMPOUX, M. WIBO,M. ROBBIand J. BERTHET (1974) J. Cell Biol. 61. 213-231. BEEVERS, L. and R. POULSON (1972) Plant. Physiol. 4Y, 476481. BEHRENS. M.(1932) Hoppe-Seylers 2. Physiol. Chem. 2OY. 59-74. BEHRENS, M . (1938) In: E. Abderhalden (ed.) Handbuch der biologischen Arbeitsmethoden, Vol. 5 part 10 I1 (Urban and Schwarzenburg, Berlin) pp. 1363-1392. BENEDICT, R.C., V.N. SCHUMAKER and R.E. DAVIES(1967) J. Reproduct. Fert. 13, 237-249. BENSLEY. R.R. and N.L. HOERR(1934) Anat. Rec. 60,449455. BERMAN A,S . (1966) Nat. Cancer Inst. Mono. 21.41-76. BHORJEE. J.S. and T. PEDERSON (1973) Biochemistry 12, 2766.2773. BIRNIE, G.D. (1967) Lab. Practice 16, 705-724. B I R N I EG.D. . (1969) Lab. Equipment Digest 7 ( 5 ) . 59-70.
276
DENSITY t i R A D l E N T CENTRIFUGATION
BIRNIE.G.D. (1 972) (ed.) Subcellular Components, Preparation and Fractionation (Butterworths, London). BIRNIE. G.D. (1973) In: E. Reid (ed.) Methodological Developments in Biochemistry, Vol. 3 (Longman. London) pp. 17-28. BIRNIE. G.D., D. RICKWOOD and A. HELL(1973a) Biochim. Biophys. Acta 331, 283-294. B I R N I FG.D.. . D.R. HARVEY, J. G R A H A M A.. COOPER and J. MOLLOY (1973b) In: E. Reid (ed.) Methodological Developments in Biochemistry. Vol. 3 (Longman, London) pp. 51-60. B I R N I G.D.. E. A. HELL.T.K. SLIMMING and J. PAUL(1973~)op. cit. pp. 127-134. BLYTH.C.A.. R.P. C L A R KR.B. , FREEDMAN, J. HAMMOND, D.W. JAMES, B.R. RABIN. (1973) Eur. J. Biochem. 32. 57-62. D. RIDGE, J. VINTERand D. WILLIAMS BOONE. C.W., G.S. H A R E Land L H.E. BOND(1968) J. Cell Biol. 36, 369-378. BOWEN. T.J. (1970) An Introduction to Centrifugation (Wiley Interscience, New York). BRAKKE. M.K. (1951) J. Amer. Chem. SOC.73, 1847-1848. B R A K K EM.K. . (1953) Arch. Biochem. Biophys.45. 275-290. BRAKKE. M.K. (1960) Adv. Virus Res. 7. 193-224. B R A K K EM.K. . (1964) Arch. Biochem. Biophys. 107. 388403. BRONFMAN, M. and H. BEAUFAY (1973) FEBS Letters 36. 163-168. BROWN,D.H.. E. CARLTON, B. BYRD,B. H A R R E Land L R.L. HAYES(1973) Arch. Biochem. Biophys. 15.5. 9-18. BU(.hlN(oHAM.M.E. and F. G ~ o s ( l 9 7 5 FEBS ) Letters 53, 255-259. BURGE. M.L.E. (1973) PhD Thesis (University of Surrey. Guildford). B u R ~ ~ M.L.E. E. and HINTON,R.H. (1971) In: E. Reid (ed.) Separations in Zonal Rotors (Wolfson Bioanalytical Centre, University of Surrey, Guildford) pp. S-5.1S.5.10. BIIRNS.J.B. (1972) Lab. Practice 21, supplement A 9. CANONICO. P.G. and J.W.C. BIRD(1970) J. Cell Biol. 4.5, 321-333. CARTLEDGE. T.G. and D. LLOYD (1972) Biochem. J. 126, 381-393. CARTLEDGE. T.G., R.A. COOPER and D. LLOYD (1971) I n : E. Reid (ed.) Separations in Zonal Rotors (Wolfson Bioanalytical Centre, University of Surrey, Guildford) pp. V-4.1-V-4.16. CHAPPELL. J.B. and R.G. HANSFORD (1972) In: G.D. Birnie (ed.) Subcellular Components Preparation and Fractionation 2nd edn (Butterworths, London) pp. 77-91. CHAUVEAU. J., Y.MOULE and Ch. R O U I L L E (1956) R Exptl. Cell Res. 11, 317-321. C H E V R E N KC.H. A . and L.H. ELROD(1972) Manual of Methods for Large Scale Zonal Centrifugation (Beckman-Spinco, Palo Alto, California). CHO. N., H.P. BARRINGER, J.W. AMBURGEY, G.B. CLINE,N.G. ANDERSON, L.L. MCCAULEY, R.H. STEVENS and W.M. STARTOUT (1966) Nat. Cancer Inst. Mono. 21.485-502. CLINE, G.B. (1971) Progress in Separation and Purification 3. 297-324. CLINE.G.B. and M.K. DAGG(1973) In: E. Reid (ed.) Methodological Develop-
REFERENCES
277
ments in Biochemistry, Vol. 3 (Longman, London) pp. 61-70. CLINE, G.B. and RYEL,R.B. (1971) Methods in Enzymology XXII, 168-204. CLINE, G.B., ANDERSON,N.G. and FENNELL, R. (1971) In: E. Reid (ed.) Separations with Zonal Rotors (Wolfson Bioanalytical Centre, University of Surrey, Guildford) pp. A-3.1-A-3.2. CLINE, G.B., M.K. DAGG, J. SROUR,M. WICKENHAUSER and H. JAMES(1973) in: E. Reid (el/.) Methodological Developments in Biochemistry. Vol. 3 (Longman. London) pp. 105-1 12. COHEN.E.. Y . SHAIN.Y. BEN-SHAUL and A.M. M A Y E R(1971) Can. J. Bot. 4, 2053-2057. CONTRACTOR. S.F. (1973) In: E. Reid (ed.) Methodological Developments in Biochemistry, Vol. 3 (Longman, London) pp. 187-189. COOKE,K.B. and APSEY.M. (1971) In: E. Reid (ed.) Separations with Zonal Rotors (Wolfson Bioanalyticdl Centre, University of Surrey, Guildford) pp. M.2.1-M-9. COOPER, R.A. and DOBROTA, M . (1974) In : E. Reid (ed.) Methodological Developments in Biochemistry, Vol. 4 (Longman, London) pp. 429-432. COPE,J. and H.R. MATTHEWS (1973) In: E. Reid (ed.) Methodological Developments in Biochemistry, Vol. 3 (Longman, London) pp. 77-84. C R A N EC.J.. , and MILLER, D.L. (1974) Biochem. Biophys. Res. Comm. 60. 1269-1277. DAHLBERG. A.E.. C.W. D I N G M Aand N A.C. PmcOrK (1969) J . Mol. Biol. 4 / . 139-147. DALLNER . H . SVFNSSON G.. and L. ERNSTER (1972) In: A. Akeson and A. Ehrenburg (uds.) Oxidation-Reduction Enzymes (Pergamon Press, Oxford) p p . 567-574. D E DUVE,C. (1966) In: D.B. Roodyn (ed.) Enzyme Cytology (Academic Press. New York) pp. 1-26. D E DUVE.C. (1969) Ann. N.Y. Acad. Sci. 168. 369-381. D E DUVE. C. and J. BERTHET (1954) Int. Rev. Cytol. 3,225-275. DE DUVE. C. (1971) J . Cell Biol. SO. 20D-55D. DE D U V EC.. , B.C. PRESSMAN, R . G I A N ~ ~ R. T OWATTIAUX , and F.APPELMANS (1955) Biochem. J. 60. 604-617. DE DUVE. C.. J. BERTHET and H. BEAUFAY (1959) Prog. Biophys. Biophys. Chem. Y, 325-369. D E N M A NA.M. . and B.K. PELTON (1973) In: E. Reid (ed.) Methodological Developments in Biochemistry, Vol. 2 (Longman, London) pp. 185-199. DE P I E R R E J.W. , and M.L. KARNOVSKY (1973) J. Cell Biol. 56, 275-303. DICK€, K. (1973) In: E. Reid (ed.) Methodological Developments in Biochemistry, Vol. 2 (Longman. London) pp. 175-184. DOBROTA.M. (1971) In: E. Reid (ed.) Separations with Zonal Rotors (Wolfson Bioanalyticdl Centre. University of Surrey, Guildford) pp. 2.2.1-2.2.6. DOBROTA. M. and R.H. HINTON (1973) Anal. Biochem. 51.270-274. DOLIROTA, M. and R.H. HINTON(1974) In: E. Reid (ed.) Methodological Developments in Biochemistry, Vol. 4 (Longman. London) pp. 177-186. DOHROTA. M. and E. R F I D(1971) In: E . Reid (ed.) Separations with Zonal Rotors
278
DENSITY GRADIENT CENTRIFUGATION
(Wolfson Bioanalytical Centre, University of Surrey, Guildford) pp. A-I .l-A-l.lO. DONALDSON, R.P., N.E. TOLBERT and C. SCHNARRENBERGER (1972) Arch. Biochem. Biophys. 152, 199-215. DUMANSKII, A., E. ZABOTINSKI and M. EVSEYEV (1913) Z. Chem. Ind. Kolloide 12, 6-11, EDELMAN, G.M., H.G. KUNKEL and E.L. FRANKLIN (1958) J. Exp. Med. 108, 105-120. EIKENBERRY, E.F., T.A. BICKLE, R.R. TRAUT and C.A. PRICE(1970) Eur. J. Biochem. 12, 113-1 16. ELROD.L.H., L.C. PATRICK and N.G. ANDERSON (1969) Anal. Biochem. 30,230-248. EL-AASER, A.A., E. REID,E. KLUCIS, P. ALEXANDER, J.T. LETTand J. SMITH(1966) Nat. Cancer Inst. Mono. 21, 323-332. EL-AASER,A.A., J.T.R. FITZSIMONS, R.H. HINTON, K.A. NORRIS, R.H. HINTON and E. REID(1973) Histochem. J. 5, 199-223. EMMELOT, P., C.J. Bos, E.L. BENEDETTI and Ph. R U M K (1964) E Biochim. Biophys. Acta 90, 126-145. EVANS, W.H. and J.W. GURD(1973a) Biochem. J. 133, 189-199. EVANS, W.H. and J.W. GURD(1973b) In: E. Reid (ed.) Methodological Developments in Biochemistry, Vol. 3 (Longman, London) pp. 121-126. EVANS, W.H., D.O. H o o ~ a n dJ.W. CURD (1973) Biochem. J. 135,819-826. FISCHER, L. (1969) In: T.S. Work and E. Work (eds.) Laboratory Techniques in Biochemistry and Molecular Biology, Vol. I . (North Holland, Amsterdam) pp. 151-346. FISHER, W.D. and R.E. CANNING (1966) Nat. Cancer lnst. Mono. 21,403-412. FISHER, W.D., G.B. CLINE and N.G. ANDERSON (1964) Anal. Biochem. 9,477-482. FLAMM, W.G., H.E. BOND and H.E. BURR(1966) Biochim. Biophys. Acta 129, 310-31 7. FLAMM, W.G.. M.L. BIRNSTIEL and P.M.B. W A L K E(1972) R In: G.D. Birnie (ed.) Subcellular Components 2nd edn. (Butterworths, London) pp. 279-310. (1970) Biochim. Biophys. Acta 219, 301-319. FLEISCHER, B. and S. FLEISCHER FLEISCHER, s.and L. PACKER( 1974)(eds.) Methods in Enzymology. XXXI and XXXII. Fox. S.M., G.D. BIRNIE, D.R. HARVEY, E.M. MARTIN and J.A. S O N N A B E(1968) ND J.Gen. Virol. 2, 45-59. FREIFELDER, D. (1973) Methods in Enzymology XXVII 140-150. FREDRICKSON, D.S., R.I. LEVYand F.T. LINDCREN (1968) J. Clin. Invest. 47, 24462457. FUNDING,L. and J. STEENSGARD (1973) MSE Applications Information A8/6/73, MSE Scientific Instruments, Crawley. GAHAN, P.B. (1973) In: J.B. Dingle (ed)The Lysosomes in Biology and Pathology, Vol. 3 (North Holland, Amsterdam) pp. 69-85. GALIBERT, F.. C. LARSEN, J.C. LELONG and M. BOIRRON (1965) Nature (Lond.) 207, 1039-1041. CLICK, M.R., BURNS,A.H. and REDDY,W.J. (1974), Anal. Biochem. 61. 32-42.
REFERENCES
279
GODSON,G.N. and R.L. S I N S H E I M(1967) E R Biochim. Biophys. Actd 149,489495, GRAHAM J.M. , (1972) Biochem. J. 131. I 1 13-1 124. GRAHAM J.M. , (1973) I n : E. Reid (ed.) Methodological Developments in Biochemistry, Vol. 3 (Longman, London) pp. 205-217. HADDEN, C. and NESTER, E. W. (1968). J. Bact. Y5, 86-885. HALSALL, H.B. and V.N. SCHUMAKER (1969) Anal. Biochem. 30,368-376. HALSALL.H.B. and V.N. SCHUMAKER (1970) Biochem. Biophys. Res. Comm. 39, 479-485. HALSALL. H.B. and V.N. SCHUMAKER (1971) Biochem. Biophys. Res. Comm. 43. , 601-606. HALLIWELL, B. (1974) In: E. Reid (ed.) Methodological Developments in Biochemistry, Vol. 4 (Longman, London) pp. 357-366. HAMILTON. M.G.. A. PAVLOVEC. and M.L. PETERMANN (1971) Biochemistry 10. 3424 -3427. HANCOCK,R. (1970) J. Mol. Biol. 48. 357-360. H A R T M A G.C., N. N. BLACK,R. S I N C L Aand I R R.H. HINTON (1974) I n : E. Reid (ed.) Methodological Developments in Biochemistry, Vol. 4 (Longman. London) pp. 93-102. HARVEY E.N. , (1931) Biol. Bull. 61, 273-279. HARVEY, E.B. (1932) Biol. Bull. 62, 155-167. HARWOOD, R. (1974) Int. Rev. Cytol. 38, 369403. HASTINCS, J.R.B., J.H. PARISH, K.S. KIRBYand E.A. COOK(1968) Biochim. Biophys. Aced 155. 603405. HATCH,F.T. and R.J. LEES(1968) Adv. Lip. Res. 6, 1-68. HELL,A., G.D. BIRNIE,T.K. S L I M M I Nand G J. PAUL(1972) Anal. Biochem. 48, 369-377. HENNING R., and H-G. HEIDRICH (1974) Biochim. Biophys. Acta 345, 326-335. HENNING R., and H. PLATTNER (1974) Biochim. Biophys, Acta 354, 114-120. HENSHAW, E.C. (1968) J. Mol. Biol. 36. 401411. HENSHAW. E.C., D.G. G U I N E Y and C.A. HIRSCH(1973) J. Biol. Chem. 248, 43674376. HILDERSON, H.J. (1974) I n : E. Reid (4.) Methodological Developments in Biochemistry, Vol. 4 (Longman, London) pp. 137-144. HINTON, R.H.( 1972)In:G.D. BIRNIE(ed.) Subcellular Fractions 2nd edn (Buttenvorths, London) pp. 119-156. HINTON, R.H. (1971) In: E. REID(ed.) Separations with Zonal Rotors (Wolfson Bioanalytical Centre, University of Surrey, Guildford) pp. Z-5.1-Z-5.10. HINTON, R.H. and M. DOBROTA, (1969) Anal. Biochem. 30, 99-110. HINTON. R.H.and K.A. NORRIS (1972) Anal. Biochem. 48.247-258. HINTON.R.H. and E. R E I D(1975) In: G.A. Jamieson and D.M. Robinson (eds.) Mammalian Cell Membranes (Butterworths, London) pp. 161-197. HINTON, R.H.. E. KLUCIS. A.A. EL-AASER, J.T.R. FII-ZSIMONS, P. A L E X A N Dand E R E.
280
DENSITY GRADIENT CENTRIFUGATION
R E I D(1967) Biochem., J. 105, 14P. H I N T O NR.H.. , M.L.E. BURGEand G.C. H A R T M A N (1969) Anal. Biochem. 29. 248-2 56. HINTON, R.H., J.T.R. FITZSIMONS, M. DOBROTA and E. REID(1970) Eur. J. Biochem. 12, 349-361. HINTON,R.H., K.A. NORRISand E. REID(1971) In: E. Reid (ed.) Separations with Zonal Rotors (Wolfson Bioanalytical Centre, University of Surrey, Guildford) pp. S-2.1-S-2.16. HINTON,R.H. and MULLOCK, B.M. (1976) I n : D. Rickwood (ed.) The use of iodinated density gradient media for biological separations. (Information Retrieval Ltd., Lond.) p. 1-14. HINTON,R.H., B.M. MULLOCK, E. REIDand C.C. GILHUUS-MOE (1973a) Abs. 9th Internat., Cong. Biochem. Stockholm p. 31. HINTON,R.H., A.J. KOWALSKI and A. MALLINSON (1973b) Clin. Chim. Acta 44, 267-270. HINTON,R.H., Y. AL-TAMER,A. MALLINSON and V. MARKS(1974a) Clin. Chim. Acta 53, 355-360. HINTON,R.H., B.M. MULLOCK and C.C. GILHUUS-MOE (1974)b) I n : E. Reid (ed.) Methodological Developments in Biochemistry, Vol. 4 (Longman, London) pp. 103-110. HOGEBOOM, G.H. and E.L. KUFF(1954) J. Biol. Chem. 210, 733-751. HOSSAINY, E., A. Z W E I N D L and E R D.P. BLACK (1973) J. Mol. Biol. 74, 283-289. HOUSE,P.D.R. and M.J. W E I D E M A N(1970) N Biochem. Biophys. Res. Comm. 41, 541 -548. INFANTE, A.A. and M. KRAUSS (1971) Biochim. Biophys. Acta 246, 81-99. ITZHAKI,R.F. and D.M. G I L L(1964) Anal. Biochem. 9, 401410. JAENISCH, R., P.H. HOFSCHEIDER and A. PREUSS (1969) Biochim. Biophys. Acta 190, 88-100. JOHNSTON, I.R. and A.P. MATHIAS (1972) Ill: G.D. Birnie (ed.) Subcellular Components (Butterworths, London) pp. 53-75. JOHNSTON, I.R., A.P. MATHIAS,F. P E N N I N C ~ Tand O ND. R I D G E(1968) Biochem. J. 109, 127-135. E.B. TUCKER and S. ZALIK (1973) Can. J. Biochem 51, JONES,B.L., N. NAGABHUSAN, 686-693. KAHLER. H. and B.J. LLOYDJr. (1951) J . Phys. Colloid Chem. 55, 13441350. KAVENOFF, R. (1972). J. Mol. Biol. 72, 801-806. KEMPF,J., J.M. EGLY,C. STRICKER,M. SCHMIDT and P. MANDEL(1972) FEBS Letters 26, 130-134. KLUCIS,E.S. and J.T. LETT(1970) Anal. Biochem. 35,480-488. KOKILEVA, L., I. MLADENOVA and R. TSANEV (1971) FEBS Letters 12, 313-316. LAGERCRANTZ, H . and H. PERTOFT (1972) J. Neurochem. 19,811-823. LEACH,J. (1971) I n : E. Reid (ed.) Separations with Zonal Rotors (Wolfson Bio-
REFERENCES
28 1
analytical Centre, University of Surrey, Guildford) pp. P-5.1-P-5.3. LEIF, R.C., W.C. KNEECE Jr.. R.L. WARTERS, H. GRINVALSKY and R.A. THOMAS (1972) Anal. Biochem. 45, 357-363. P. BAUDHUIN, J.W. COFFEY, S. FOWLER and LEIGHTON, F., B. POOLE,H. BEAUFAY, C. DE DUVE(1968) J. Cell Biol. 37,482-513. LEITIN, V.L. and M.1. LERMAN (1969) Biokhimiya 34, 839-848. LESKES, A., P. SIEKEVITZ and G.E. PALADE (1971) J. Cell Biol. 49, 288-302. LINDGREN, F.T., H.A. ELLIOTand J.W. GOFMAN(1951) J. Phys. Colloid. Chem. 55, 8C93. LINDGREN, F.T., A.V. NICHOLS,F.T. UPHAMand R.T. WILLS(1962) J. Phys. Chem. 66, 2007-201 1. LINDGREN, F.T., L.C. JENSEN and F.T. HATCH(1972) In: G.J. Nelson (ed.) Blood Lipids and Lipoproteins, Quantitation, Composition and Metabolism (Wiley Interscience, New York) pp. 181-274. LLOYD,D., and T.G. CARTLEDGE (1974) In: E. Reid (ed.) Methodological Developments in Biochemistry, Vol. 4 (Longman, London) pp. 339-346. S.E. VENABLES, G.I. ROACHand G. T U R N E (1971) R LLOYD,D., R. BRIGHTWELL, J. Gen. Microbiol. 65, 209-223. Lo, C-H. and H. STELSON (1972) Anal. Biochem. 45, 331-336. (1973) J. Cell Biol. 57,659-667. LORD,J.M., T. KAGAWA, T.S. MOO RE^^^ H. BEEVERS (1970) FEBS Letters 6, 1 1 4 116. LOWE,D., E. REIDand T. HALLINAN (1951) J. Biol. LOWRY,O.H., N.J. ROSEBROUGH, A.L. FARRand R.J. RANDALL Chem. 193, 265-275. LUKANIDIN, E.M., E.S. ZALMANZON, L. KOMAROMI, 0.P. S A M A R I N A and G.P. GEORGIEV (1972) Nature New Biol. 238, 193-197. MALLINSON, A. and R.H. HINTON(1973) In: E. Reid (ed.) Methodological Developments in Biochemistry, Vol. 3 (Longman, London) pp. 113-1 19. MANDEL, M., L. IGAMBI, J. BERGENDAHL, M.C. DODSON Jr. and E. SCHELTGEN (1970) J. Bact. 101, 333-338. MATEYKO, G.M. and M.J. KOPAC(1963) Ann. N.Y. Acad. Sci. 105, 183-286. MATHIAS, A.P. and C.V.A. WYNTER (1973) FEBS Letters 33, 18-22. (1969) Biochem. J. 111, 583-591. MATHIAS, A.P., D. RIDGE,and N. St.G. TREZONA MATILE,Ph. (1968) Planta 79, 181-196. MATILE,Ph. (1969) In: J.T. Dingle and H. Fell (eds.) The Lysosomes in Biology and Pathology, Vol. 1 (North Holland, Amsterdam) pp. 406-433. MCCONKEY, E.H. (1967) Methods in Enzymology 12A, 62C634. MCCONKEY, E.H. (1974) Proc. Nat. Acad. Sci. USA 71, 1379-1383. MCCORMICK, H.W. (1967) In: M.J.R. Cantor (ed.) Polymer Fractionation (Academic Press, New York) pp. 251-284. MCEWAN, C.R. (1967) Anal. Biochem. 20, 114149. MACGILLIVRAY, A.J., A. CAMERON, R.J. KRAUZE, D. RICKWOOD and J. PAUL(1972) Biochim. Biophys. Acta 277. 384402.
282
DENSITY G R A D I E N T CENTRIFUGATION
MCQUILLEN, K., R.B. ROBERTSand R.J. BRITTEN(1959) Proc. Nat. Acad. Sci. USA 45, 1437-1447. (1957) Proc. Nat. Acad. Sci. U.S. 43, MESELSON, M., F.W. STAHLand J. VINOGRAD 581-588. MEUWISSEN, J.A.T.P. (1971) In:E. Reid (ed.) Separations with Zonal Rotors (Wolfson Bioanalytical Centre, University of Surrey, Guildford) pp. B-2.1-B-2.8. MEUWISSEN, J.A.T.P. (1973) I n : E. Reid (ed.) Methodological Developments in Biochemistry, Vol. 3 (Longman, London) pp. 29-14. (1970) Biochem. Biophys. Res. Comm. MEUWISSEN, J.A.T.P. and K.P.M. HEIRWEGH 41, 675-68 I . MIESCHER, F. (1 871) In: Hoppe-Seylers Medizinisch-chemische Untersuchungen Vol. 4 (A. Hirschwald, Berlin) 441460. MILLER,R.G. and R.A. PHILLIPS (1969) J. Cell Physiol. 73, 191-202. MONAHAN, J.J. and R.H. HALL (1974) Anal. Biochem. 62, 217-239. MoRRE,D.J.,R.L. HAMILTON, H.H. MOLLENHAUER, R.W. MAHLEY, W.D. CUNNING(1970) J. Cell Biol. 4 4 , 4 8 4 4 9 1 . HAM, R.D. CHEETHAM and V.S. LEQUIRE MORRE, D.J., R.D. CHEETHAM, S.E. NYQUIST and L. OVTRACHT (1972) Prep. Biochem. 2. 61-69. MORRE, D.F., W.N. YUNGHANS, E.L. VIGILand T.W. KEENAN (1974) In: E. Reid (ed.) Methodological Developments in Biochemistry, Vol. 4 (Longman, London) pp. 195-236. MULLER, M. (1972) J. Cell Biol. 52, 478-487. MULLER, M. (1973) J. Cell Biol. 5 7 , 4 5 3 4 7 4 . MULLOCK, B.M. and R.H. HINTON (1973) Trans. Biochem. SOC.1,578-581. MULLOCK, B.M., R.H. HINTON, M. DOBROTA, D. FROOMBERG and E. Reid (1971) Eur. J. Biochem. 18,485495. MUNTHE-KAAS, A.C. and P. SEGLEN (1974) FEBS Letters 43,252-256. NASH,C.H., R.J. DOUTHART, L.F. ELLIS,R.M. V A N FRANK, J.P. BERNETT and P.A. LEMKE (1973) Can. J. Microbiol. 19, 97-103. B. HALSALL and J. SCHWEDES (1969) Biopolymers 7, NASON,P., V. SCHUMAKER, 241-249. NELSON, E.B., B.S.S. MASTERSand J.A. PETERSON (1971) Anal. Biochem. 39. 128-134. NEVILLE, D.M. (1960) J. Biophys. Biochem. Cytol. 8, 413422. NEAL,M.W. and J.R. FLORINE (1972) Anal. Biochem. 45,271-276. NEAL,W.K., H.P. HOFFMAN, C.J. AVERS and C.A. PRICE(1970) Biochem, Biophys. Res. Comm. 38,414422. NOLL,H. (1967) Nature (Lond.) 215, 360-363. NOLL,H. (1969) In: P.N. Campbell and S.R. Sargent (eds.) Techniques in Protein Biosynthesis Vol. 2 (Academic Press, New York) pp. 112-171. NOMURA, M., B.D. HALLand S. SPIEGELMAN (1960) J. Mol. Biol. 2, 306-326. NORMAN.M. (1971) In: E. Reid (ed.) Separations with Zonal Rotors (Wolfson Bioanalytical Centre, University of Surrey, Guildford) pp. P-6.12-6.4.
R E F E R E N CES
283
R.H. H I N T O N and E. R E I D(1974) Biochem. NORRIS.K.A.. M DOBROTA, F.S. ISSA, J. 142, 667-67 I . NOVIKOFF, P.M., A.B. NOVIKOFF, N. Q U I N T A Nand A J-J. H A U W(1971) J. Cell. Biol. 50. 859-886. PALADE, G . (1971) J. Cell. Biol. 50, 5D-19D. PARISH, J.P.(1972) In: G.D. B I R N I E(ed.) Subcellular Components 2nd edn (Butterworths, London) pp. 251-278. PERRY,R.P. and D.E. KELLEY (1966) J. Mol. Biol. 16,255-268. PERTOFT,H. and T.C. LAURENT (1969) In: T. Garritson (ed.) Modern Separation Methods of Macromolecules and Particles (John Wiley, New York), pp. 71-90. PETERMANN, M.L. (1964) The Structure and Function of Ribosomes (Elsevier. Amsterdam). PFUDERER. P., P. CAMMARANO, D.R. HOLLIDAY and G.D. NOVELLI (1965) Biochim. Biophys. Acta 109, 595-606. PICKELS. E.G. (1943) J . Gen. Physiol. 26. 341-360. POLLACK. M.S. and C.A. PRICE(1971) Anal. Biochem. 42, 38-47. POLLACK. J.K. and MU", E.A. (1970) Biochem. J. 117, 913-919. POOLE.R.K. and D.. LLOYD (1973) Biochem. J. 136, 195-207. POPP,R.A.. POPP.D.M. and ANDERSON, N.G. (1968) Transplantation, 6, 664-666. PRESTAYKO, A.W.. B. LEWISand H . BUSCH(1972) Biochim. Biophys. Acta 269, 90-103. PRETLOW, T.G.. C.W. BOONE.R.I. SCHRAGER and G . H . WEISS(1969) Anal. Biochem. 2Y, 230. PRICE, C.A. (1972) In: W.W. Umbreit, R.H. Burris and J.F. Staufer (eds) Manometric Techniques 5th edn, (Burgess, Minneapolis) pp. 216-243. PRICE,C.A. and R.J. CAsciA-ro (1974) Anal. Biochern. 57, 356-362. PRICE.C.A. and A. KOVACS (1969) Anal. Biochem. 28,460468. PRICE,L.W. (1972) Lab. Practice 2/(2), 93-96. PRICE,R.G., D.G. TAYLOR and D . ROBINSON (1972) Biochem. J. 129,919-928. PROSPERO, T.D. (1973) In: E. Reid (ed.) Methodological Developments in Biochemistry. Vol. 3 (Longman, London) pp. 1-16. PROSPERO, T.D. and R.H. HINTON(1973) op cir pp. 171-186. R A H M A NY.E.. , J.F. HOWE.S.L. N A N C E and J.F. THOMSON (1967) Biochim. Biophys. Acta 146, 484492. RAYNEAUD, A. and H.H. OHLENBUSCH (1972) J. Mol. Biol. 63, 523-537. REEVE,P.and D.J. ALEXANDL-R (1970) Microbios 2, 53-58. REID, E. (1971) (ed) Separations with Zonal Rotors (Wolfson Bioanalytical Centre, University of Surrey. Guildford). REID;E. (1972a) I n : G.D. B I R N I (ed.) E Subcellular Components 2nd edn (Butterworths. London) pp. 93-1 18. REID, E. (1972b) Sub-cell. Biochem. I , 217. ) Equip. Digest. 10, 52-54. REID,E. ( 1 9 7 2 ~ Lab.
284
DENSITY GRADIENT CENTRIFUGATION
REID, E. (1973) (ed.) Methodological Developments in Biochemistry, Vol. 3. Advances with Zonal Rotors (Longman, London). REID,E. (1974) (ed.) Methodological Developments in Biochemistry, Vol. 4. Subcellular Studies (Longman, London). REJ,R. and A.H. RICHARDS (1974) Anal. Biochem. 62,24&247. RICKWOOD, D. (1976) (Ed). The use of iodinated density gradient media for biological separations (Information Retrieval Ltd.. Lond.). RICKWOOD, D. and G.D. B l R N l E (1975) FEBS Letters 50. 102-110. RICKWOOD, D., A. HELL,and G.D. B l R N l E (1973) FEBS Letters 33- 221-224. RICKWOOD, D., A. HELL,S. MALCOLM, A.S. MACGlLLlVRAY and J. PAUL(1974) Biochim. Biophys. Acta 353. 353-361. ROCHA, V. and I.P. TIN(;(1970) Archives Biochem. Biophys. 140, 398407. Ro-CHOI,T.S., Y. MORIYAMA, Y.C. CHOIand H. BUSCH (1970) J. Biol. Chem. 245, 197&1977. ROMRELL, L.J., COPPE,M.R.. MUNRO.D.R. and GOTO,S. (1975). J. Cell Biol. 65. 428438. ROODYN, D.B. (1970) In: T. Work and E. Work (eds.) Laboratory Techniques in Biochemistry and Molecular Biology, Vol. 2 (North Holland, Amsterdam) pp. 1-231. RUTENBERG, E. (1966) Oak Ridge National Laboratory Report ORNL 3978 pp. 19-25. SARTORY, W.K. (1969) Biopolymers 7,251-263. SCHACHMAN, H.K. (1959) Ultracentrifugation in Biochemistry (Academic Press, New York). SCHREIER, M . H . and T. STAEHELIN (1973) Nature New. Biol. 242, 35-38. SEDAT,J.W. and R.L. SINSHEIMER (1970) Cold Spring Harb. Symp. Quant. Biol. 25, 163. SHAKOORI, A.R., W. ROMEN.W. OELSCHLAGER, B. SCHATTERER and G . SIEBERT (1972) Hoppe-Seylers 2. Physiol. Chem. 353, 1735-1748. C.A.. V.G. E R W Iand ~ J.D. GREENAwAlT(~967)J. Cell Biol. 32.719-735. SCHAITMAK. SCHNARRENBERGER, C., A. OESER and N.E. ToLew.r (1972a) Plant Physiol. 50, 55-59. SCHNARRENBERGER. c..A . OESER and N.E. TOLBERT (1972b) Planta 104. 185-194. SCHNEIDER, W.C. and G.H. HOGEBOOM (1951) Cancer Research 11, 1-22. SCHUMAKER. V.N. (1967) Adv. Biol. Med. Phys. 11, 245-339. SEYMOUR, C.A., G . NEALEand T.J. PETERS (1974) Biochem. SOC.Trans. 2, 1101-1 104. SHEELER. P. and J.R. WEI.LS(1971) Sorvall Bulletin No. 3-71. SOTTOCASA, G . , B. KUYLENSTIERNA, L. ERNSTER and A. BEROS-IRAND (1967) J. Cell Biol. 32,415-438. SOTTONG, P.R. (1974) Process Biochem. 9(3), 30-31. SPIRIN. AS., N.V. BELITSINA and M.I. LERMAN (1965) J. Mol. Biol. 14,611-615. SPRAGG, S.P. and C.T. RANKIN (1967) Biochim. Biophys. Acta 141, 164-173. STAHN, R., K-P. MAIER and K. HANNIG (1970) J. Cell Biol. 46, 576-591. STARKEY. B.J., D. SNARY and A. ALLEN (1972) Biochem. J. 128, 123P. STEELE. W.J. and H. BUSCH(1967) In: H. Busch (ed.) Methods in Cancer Research.
REFERENCES
28 5
Vol. 3 (Academic Press, New York) pp. 61-152 vide p. 104. STEENSGARD, J. (1970) Eur. J. Biochem. 16, 66-70. STEENSGARD. J., L. F U N D I Nand G J.A.T.P. MEUWISSEN (1973) Eur. J. Biochem. 39, 481491. STEENSGARD. J.. L . F U N D I Nand G J.A.T.P. M E C ~ W I S S(1974) E N In: E. Reid (ed.) Methodological Developments in Biochemistry, Vol. 4 (Longman, London) pp. 67-80. STILL. C.C. and C.A. PRICE (1967) Biochim. Biophys. Acta 141, 176-178. STORRIE, B. and G. ATTARDI (1973) J . Biol. Chem. 248. 58165834. SVEDBERG, T. and R. FAHRAEUS (1926) J. Amer. Chem. Soc. 48,430-438. SVEDBERG, T. and K. PEDERSON (1940) The Ultracentrifuge (Clarendon Press. Oxford). (1934) Nature 134, 98. SVEDBERG, T.. G. BOES-TAD and I-B. ERIKSSON-QUENSEL SwIcK, R.W. J.L. STRANGE. S.L. NANCFand J.F. THOMSON (1967) Biochemistry 6, 737-744. SYPHERD, P.S. and J.W. WIREMAN (1974) Methods in Enzymology X X V , 349-354. SZABO,A.S. and C.J. AVERS(1968) J. Cell Biol. 3Y. 132A. SZYBALSKI, W. (1968) Methods in Enzymology XIIB, 330-360. SZYBALSKI. W., H. K U B I N S KZ. I , HRADECNA and W.C. S U M M E R(1971) S Methods in Enzymology X X I , 383-413. TAMIR, H. and G I L V A R G C.. (1966) J. Biol. Chem. 241, 1085-1090. TATA,J.R. (1972) In: G.D. B I R N I E (ed.) Subcellular Components 2nd edn (Butterworth, London) pp. 185-213. THINES-SEMPOUX. D. (1973) In: J.T. Dingle (ed.) The Lysosornes in Biology and Pathology, Vol. 3 (North Holland. Amsterdam pp. 278-299. THOMPSON, J.F. (1963) The Biological Effects of Deuterium (Macmillan. New York). THOMSON, J.F. and F.J. KLIPFEL (1958) Exp. Cell Res. 14, 612-614. THOMSON, J.F. and E.T. M I K U T A (1954) Archives Biochem. Biophys. 51. 487498. T R A U T M AR. N ,(1964) In: D.W. N E W M A(ed.) N Instrumental Methods in Experimental Biology (MacMillan. New York) pp. 21 1-297. T K A U T M AR. N . and K.M. C O W A N (1968) I n : C.A. Williams and M.W. Chase (eds) Methods in lmmunologyand Immunochemistry. Vol. 2 (Academic Press, New York) pp. 81-118. TKINICK J.A. , and A.J. ROSE(1973) MSE Applications Information A9/11/73 (MSE Scientific Instruments. Crawley, Sussex). V A N F R A N KR.M. , and W.J. K L E I N S C H M (1973) I D T Anal. Biochern. 55,601-608. VANH O I D F K . . R . a n d R.L. B,AID W I N (1958) J. Phys. Chem. 62. 7 3 4 7 4 3 . V I I K A RJ.E., I . E. HAAHT-I. T-T. HELELA. K . J U V Aand T. N I K K A R(1969) I Scand. J. Clin. Lab. Invest. 23. 85-88. WALLACE, H. (1969) Anal. Biochem. 32, 334. WALLACH. D.F.H. (1967) In: B.D. Davies and L. Warren (eds.) The Specificity of Cell Surfaces (Prentice Hall, Englewood Cliff, New Jersey) pp. 129. WATTIAUX, R. (1974) Mol. Cell Biochem. 4 , 21-29.
286
DENSITY GRADIENT CENTRIFUGATION
WATTIAUX, R., M. WIBOand P. BAUDHUIN (1963) In: A.V.S. de Reuck and M.P. Cameron (eds.) CIBA Foundation Symposium on Lysosomes (Churchill, Lond.) pp. 176196. de CONINCK and M.F. RONVEAUX-DUPAL (1971) Eur. WATTIAUX, R., S. WATTIAUX J. Biochem. 22, 31-39. (1975) Biotech. Bioeng. 17, WEBB, N.L., B.M. RICHARDS and A.P. COODERS 1313-1322. WELLS,J.R., P. S H E E L Fand R D.M. GROSS(1972) Anal. Biochem. #6,7-18. S. and W. NEUPERT (1972) Eur. J. Biochem. 25, 379-396. WERNER, WHITTAKER, V.P., I.A. MICHAELSON and R.J.A. K I R K L A N(1964) D Biochem. J. 90, 243-303. WIENER, E. and J.M. ASHWORTH (1970) Biochem. J. 118, 505-512. WILCOX.H.G.. D.C. DAVIS and M. HEIMBERC (1971) J. Lipid. Res. 12. 160-172. WILLIAMSON, R. (1971) In: E. Reid (ed) Separations with Zonal Rotors (Wolfson Bioanalytical Centre. University of Surrey Guildford) p. 2.2-6. R., M. MORRISON. G. LANYON. R. EASONand J. P A U L(1971) BioWILLIAMSON, chemistry 10, 3014- 3021. WILSON, M.A. and J. CASCARA NO(^^^^) Biochem. J. 129. 209-218. WOLF,A.V. and M.G. BROW\(1967) In, R.C. Weast (ed.) Handbook of Chemistry and Physics, 48th. edn. (The Chemical Rubber Co., Cleveland. Ohio) pp. D 1 4 4 D183. WOLFF,D.A. and H. PERTOFT (1972a) J. Cell Biol. 55, 579-58 I . WOLFF,D.A. and H. PERTOFT (1972b) Biochim. Biophys. Acta 286. 197-204. J.A. CARTER and I.W. WEBER(1966) Nat. Cancer Inst. WRIGHT,R.R., W.S. PAPPAS, Mono. 21,241-244. YOUNG,B.D. (1972) MSE Applications Information A6/6/72 (MSE Scientific Instruments, Crawley, Sussex). (1972) Archives Biochem. Biophys. 152, ZIMMER, G., A.D. KEITHand L. PACKER 105-1 13. ZOLOTOR, L. and R. ENGLER (1967) Biochim. Biophys. Acta 145,52-59.
Subject index
Albumin (as gradient material) 266 aluminium alloys (in centrifuge rotors) 18. 19. 32, 36. 37. 234 algae 231 analytical rotors 9ff. 17. 18. 204. 242 angle head rotors 14. 19, 20, 64. 234 autoAnalysers 197, 248
cold trap 167 colloidal silica 87, 92, 206 colorimeter (seespectropho tometer) computers 96 continuous flow rotors 14G142, 239240, 242 J, K and RK types 23, 34, 42, 128, 14C141, 192, 235 cooling coil 188 cooling rotors 147 Coriolis forces 105, 118 cross leakage 158-161, 172-173 cytochemical stains 199, 221
Beaufay rotors (E-40 and S-25) 79. 143, 218, 231,258. 261 bovine serum albumin (see albumin) bulk flow (see hydrodynamic instability) C series rotors 142 caesium acetate 80 caesium chloride 33, 37, 39, 55, 66, 67, 80, 115. 1 18. 222, 225, 234, 235, 240, 241,250 caesium formate 80 caesium oxalate 66 caesium sulphate 66,225. 233 capacity of gradients 49-50, 58, 59 carbon fibres 257 cells, centrifugation of whole, 15, 24, 88, 205ff, 246 china clay particles 202 chloral hydrate 67, 82. 225 chloroplasts 226227 chromatin 67, 81,225 cleaning flow lines 171 cleaning rotors 36-38
D-series rotors 142 deoxyribonucleic acid 27, 64, 66, 80, 225, 229, 253 deoxynucleoprotein (see chromatin) density of particles 221. 241 changes with medium concentration 64-69,25 1 density meters 183-184 deuterium oxide 82 dextrans (as gradient solutes) 86,207, 221 diethylpyrocarbonate 82 differential pelleting 9, 1I f f diffusion coefficient 61, 242 drop counters 180 droplet sedimentation (spe sedimentation in droplets)
287
288
DFNSIlY CrRADIENl CE"rTRIFU<,ATION
E-40 rotor (see Beaufay rotors) electron microscopy 198-1 99 equilibrium gradients 60-63, 91-93. 114I I6 endoplasmic reticulum 93, 215. 21 7-222. 227.229.230 erythrocytes (see red blood cells) enzyme assays 197. 200 F series rotors 142 feed head 42, 126ff. 137. 151-153, 156. 160, 171 feed lines 148. 168ff ferritin particles 217 Ficoll 86, 206, 220, 247, 248, 249 flotation 73. 105, 218, 237-238 flow cells (for spectrophotometers) 1 I2113. 177-178, 135 fractionation of gradient 108ff fraction collectors 114. 178-180, 188 free flow electrophoresis 215 Gel electrophoresis 25-27. 232 GeMSAEC analyser 198 glucose (in density gradients) 67, 82, 225 glycerol 82. 92, 249 glycogen (as gradient material) 86 glycogen particles 217, 218 glyoxysomes 226 Golgi apparatus 90, 2 13. 21 7. 226. 227 gradient loading 148. 15311' gradient preparation 97ff, 115-1 16. 145. 149, 180-182 gradient shape 90-103 guard tray 42. 130. 152-1 53 guarantee 43ff Haemoglobin 17 honiogenate, separation of whole 93,201 hydrodynamic instability 54-57 hydrostatic pressure in rotors 143. 245. 257-258
Integrator 189 inverse gradients (for sample layer) 53, 104. 158. 161 -162 iodinated aromatic compounds (see X-ray contrast media) lsokinetic gradients 91. 101. 204 isovolumetric gradients 91. 204 Laboratory design 27ff layering sample 104. 187 lead ions 221 lipoproteins (serum) 73, 81. 95. 237-239 loading speed 145-146 Lucite (see Perspex) Ludox (see colloidal silica) lysosoines Y, 15, 24. 68. Y4. 215-217. 227, 230, 231 lysosome membranes 217 Macroglobulins 235 magnesium ions 221 messenger RNA 222 methanol (as gradient material) 89, 232 methoxyethanol (as gradient material) 89. 232 Metrkamide66,67,88,207,223.225,248, 249 microbodies 16.69.215.226.227,230.231 microsomes 73-75. 217ff microscopy 198 (see crlso electron microscopy) mitochondria 8. 15. 24. 68. 75, 94, 212214. 215. 227. 230,231, 246 mitochondrial fraction 212-214 mitochondrial membranes 214 monitoring 112-1 14. 155. 174ff mucoproteins 236 Noryl 38. 137 nuclear fraction 210-21 1 nuclei 8. 15, 89. 21 1-212 nucleic acids and nucleoproteins (see
SUBJECT I N D E X
ribonucleic acid, deoxyribonucleic acid etc.) Organic solvents. use in centrifugation 15. 36, 89 overlay 157 Partial specific volume 54 pathogenic organisms 34 pellets. damage to particles in 244 pellets. striation of 198 peroxysomes (see microbodies) Perspex 36. 38-39.40 plasma membrane fragments 75.93. 210211, 217, 220. 235 polysomes 14. 25. 51. 217. 222. 229 potassium tartrate 37. 81. 240-241 presentation of results 200-202, 218 proplastids 227 protein. estimation of 248 proteins. separation of 88.235-240 protozoa 230 pumps 114. 146. 173. 179. 182-183. 186 Quality control 200 quartz centrifuge tubes I12 Radioactive sample 34 red blood cells 209. 210 recorders I86 refractometers 176, 183 relaxation technique 63. I18 Renografin 207 reorientating zonal rotors 23-24. 189ff JCF-2 126, 140 SZ-14 192. 213 resolution. limits on 27. 76. 238 reversed wedge (sample zone) (see inverse gradient) ribonuclease (in sucrose) 81-82 ribonucleic acid 14, 25-27, 80, 81. 225, 233, 247. 250
289
ribonucleoprotein particles 25-27. 66, 81, 88. 218. 22-225 ribosomes and ribosomal subunits 66, 226. 229. 250, 251 Rulon seal 20. 22. 39. 42. 126ff. 152, 160 S-l) diagram 71. 93 S-25 rotor (see Beaufay rotors) Safety 29ff. 241 sample loading (zonal) 1 5 6 1 58 (swing-out) see layering sample sarcosyl (sodium dodecyl sarcosinate) 225 Schlieren optics I0 scintillation counting 232. 250 sectorial unloading 173 sedimentation indroplets 52-57. 105,241, 251 sedimentation coefficient 48.49,202-204. 251-253 silica (see colloidal silica) sodium bromide 33, 81. 237. 238. 241 sodium citrate 37 specific activity (of enzymes) 201 spectrophotometers 176. 185 step gradients 59 sterilisation 37 storage granules 229 streaming (see sedimentation in droplets) stroboscopic light 186 sucrose 15, 55. 66, 81. 92, 225. 233. 240, 246250. 251 swing-out rotors 19-20. 63, 77-79, 210. 213, 218.232 syphons 179
Temperature control (in Centrifugation) 164ff. 253. 255 temperature control (in loading and unloading 148-149, 154155. 178. 187 thermocoupleand thermocouplecell 149150. 170. 186
290
DENSITY ( J R A D I E N l CENTRIFU(oA fION
titanium alloys (in centrifuge rotors) 18-19. 32. 37-38 transplantation antigen 235 tubing 147, 171. 187 tubing clamps 147. 170. 188 tubing connectors 147. 170. 187 tumbling 50 turbine drive air 17, 18, 192, 256 oil 18, 192. 256 Unloading 168ff Urografin 247, 248 Vacuole (in plant cells) 228 vertical tube rotors 245, 261 viruses, separation of 24, 34, 8 1, 239-242 viscosity of gradient solutes 51 Width of starting zone 51-57. 76 of bands in rate sedimentation 57-59
of bands in isopycnic separations 60 X-ray
media 87-89. 207 Metrizamide. Kenografin. Urografin) COlltrdSI
(see t h o
Yeast 23 I Zonal rotors 20ff. 4&41. 76-79. 12Off. 254.260-26 I A-XI1 28.40,4142, 130-134. 162-164, 176. 209-213. 226, 259 B-IV 22Z23. 42. 135--136. 135 B-XIV and B-XV 22-23. 40, 42. 50. 136 i39.166.’IX.’31.235.237.142
€3-XXIX and B-XXX 139. 162. 173.228 HS and Z-15 23, 40. 42, 134-135, 163. 231. 259 (sec a/so Brduray rotors. Continuous flow rotors and Reorienting rotors) Zymogen granules ( s w storage granules).
A N INTRODUCTION TO RADIOIMMUNOASSAY A N D RELATED TECHNIQUES T. Chard Departments of Obstetrics, Gynaecology and Reproductive Physiology, St. Bartholomew's Hospital, Medical College. and the London Hospital Medical College, London, U.K.
This Page Intentionally Left Blank
Contents
List of’ abbreviations . . . . . . . . . . . . . . . . .
299
Chapter I . The background to radioimmunoassay
301
1.1. Introduction . . . . . . . . . . . . . . . . 1.2. Terminology . . . . . . . . . . . . . . . . 1.3. Early development of radioimmunoassay . . . 1.4. Basic principles of binding assays . . . . . . 1.5. Binder dilution curves and standard curves . . 1.6. Methods of plotting the standard curve . . . 1.7. The importance of K value . . . . . . . . . . 1.8. The measurement of K value . . . . . . . . 1.9. A model system for binding assays . . . . . . 1.10. Some implications of the model system . . . .
.
.
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . .
. . . . . .
. . . . . .
301 302 . . . . . . . 304 . . . . . . . 306 . . . . . . . 311 . . . . . . . 314 . . . . . . 321 . . . . . . . 322 . . . . . . . 324 . . . . . . . 325
Chapter 2. Requirements.for a binding assay - purified ligand
329
2.1. Requirements for a binding assay . . . . . . . . . . . . . . . . . . 329 2.2. The need for purified ligand . . . . . . . . . . . . . . . . . . . . 329 2.3. Availability of pure ligand . . . . . . . . . . . . . . . . . . . . . 333 2.3. I . Large protein hormones . . . . . . . . . . . . . . . . . . . . 333 2.3.2. Carcinofetal antigens . . . . . . . . . . . . . . . . . . . . . 334 2.3.3. Steroid hormones and drugs . . . . . . . . . . . . . . . . . . 334 2.3.4. Small peptide hormones . . . . . . . . . . . . . . . . . . . . 335 2.4. Dissimilarity between purified ligand and endogenous ligand . . . . . . 335 2.5. Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337 341 2.6. Storage of materials . . . . . . . . . . . . . . . . . . . . . . . .
293
294
RADlOlM M UNOASSAY A N D RELATED TECHNIQUES
Chapter 3. Requirements for binding assays .tracer ligand . . . 343 3.1. Radioactive isotopes . . . . . . . . . . . . . . . . . . . . . . . 343 3.2. Counting of radioactive isotopes . . . . . . . . . . . . . . . . . . 345 3.3. Choice of a counter . . . . . . . . . . . . . . . . . . . . . . . . 347 3.4. Some practical aspects of isotope counting . . . . . . . . . . . . . . 349 3.5. Essential characteristics of a tracer . . . . . . . . . . . . . . . . . . 351 3.6. Preparation of tracers . . . . . . . . . . . . . . . . . . . . . . . 351 3.7. Iodinated tracers . . . . . . . . . . . . . . . . . . . . . . . . . 353 3.7.1. Iodination methods . . . . . . . . . . . . . . . . . . . . . . 353 3.7.2. Practical aspects of iodination . . . . . . . . . . . . . . . . . 356 3.7.3. Iodination damage . . . . . . . . . . . . . . . . . . . . . . 360 3.7.4. Purification of iodinated tracer . . . . . . . . . . . . . . . . . 365 3.7.5. Chemical assessment of the tracer . . . . . . . . . . . . . . . . 370 3.8. Alternative labels for tracers . . . . . . . . . . . . . . . . . . . . 374 3.8.1. Fluorescent labels . . . . . . . . . . . . . . . . . . . . . . . 374 3.8.2. Enzyme labels . . . . . . . . . . . . . . . . . . . . . . . . 374 3.8.3. Free radical labels . . . . . . . . . . . . . . . . . . . . . . . 376 3.8.4. Bacteriophage labels . . . . . . . . . . . . . . . . . . . . . 376
Chapter 4 . Requirementsfor a binding assay . the binder . . . . . 377 4.1. Characteristics required of a binder . . . . . . . . . . . . . . . . . 4.2. Antibodies . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.1. Chemistry of antibodies . . . . . . . . . . . . . . . . . . . . 4.2.2. Chemistry of antigens . . . . . . . . . . . . . . . . . . . . . 4.2.3. Cellular basis of immune response . . . . . . . . . . . . . . . 4.2.4. Physiology of immune response . . . . . . . . . . . . . . . . 4.2.5. Characteristics of antibodies with respect to binding assays . . . . . 4.2.6. Production of antibodies . . . . . . . . . . . . . . . . . . . 4.2.7. The nature and dose of the immunogen . . . . . . . . . . . . . 4.2.8. Preparation of haptens as immunogens . . . . . . . . . . . . . 4.2.9. The use of adjuvant . . . . . . . . . . . . . . . . . . . . . . 4.2.10. The animal species . . . . . . . . . . . . . . . . . . . . . . 4.2.11. The route of immunisation . . . . . . . . . . . . . . . . . . 4.2.12. The timing of injections and collection of antisera . . . . . . . . . 4.2.13. Selection of antisera for use in radioimmunoassay . . . . . . . . . 4.2.14. Storage of antisera . . . . . . . . . . . . . . . . . . . . . . 4.3. Cell receptors . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4. Circulating binding proteins . . . . . . . . . . . . . . . . . . . . 4.5. Radioassay for the detection of endogenous antibodies and circulating binding proteins . . . . . . . . . . . . . . . . . . . . . . . . .
377 379 379 381 382 383 384 385 386 388 390 391 392 393 393 396 396 397 398
CONTENTS
295
Chapter 5 . Rtquirementsfor a binding assay -separation of bound 401 and.frer ligand . . . . . . . . . . . . . . . . . . . . . 5.1. Efficiency of separation methods . . . . . . . . . . . . . . . . . . 5.2. Practicality of separation methods . . . . . . . . . . . . . . . . . 5.3. Methods for the separation of bound and free ligand . . . . . . . . . 5.3. I . Electrophoresis . . . . . . . . . . . . . . . . . . . . . . . . 5.3.2. Gcl filtration . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.3. Adsorption methods . . . . . . . . . . . . . . . . . . . . . . 5.3.3.1. Charcoal . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.3.2. Silicates . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.3.3. Hydroxyapatite . . . . . . . . . . . . . . . . . . . . . . 5.3.4. Fractional precipitation . . . . . . . . . . . . . . . . . . . . 5.3.5. ‘Double’ antibody methods . . . . . . . . . . . . . . . . . . . 5.3.5.1. ‘Double‘ or ‘second’ antibody . . . . . . . . . . . . . . . . 5.3.5.2. Double-antibody solid phase . . . . . . . . . . . . . . . . 5.3.6. Solid-phase systems . . . . . . . . . . . . . . . . . . . . . . 5.3.6.1. Binder attached to discs and tubes . . . . . . . . . . . . . . 5.3.6.2. Binder attached to particulate solid phase . . . . . . . . . . 5.3.7. Conclusions: the choice of a separation procedure . . . . . . . . . 5.4. Immunoradiometric techniques . . . . . . . . . . . . . . . . . . . 5.4.1. Advantages and disadvantages of the immunoradiometric assay . . .
401 403 405 405 406 407 408 409 410 410 413 413 417 418 418 419 421 423 425
Chapter 6 . Requirementsfor a binding assay . extraction of ligand .from biological.fluids . . . . . . . . . . . . . . . . . 427 6. I . Extraction for concentration of ligand . . . . . . . . . . . . . . . . 427 6.1. I . Extraction and concentration procedures using particulate adsorbents . 428 6.2. Extraction for purification of ligand . . . . . . . . . . . . . . . . . 433 6.2.1. Extraction to improve specificity . . . . . . . . . . . . . . . . 434 6.2.2. Extraction to free the ligand from conjugates or complexes . . . . . 436 6.3. General aspects of extraction procedures . . . . . . . . . . . . . . . 439
Chapter 7. Requirementsfor binding assays . calculation of results 440 7.1. 7.2. 7.3. 7.4.
Calculation of results by simple manual extrapolation . . . . . . . . . 440 Linearisation of the standard curve . . . . . . . . . . . . . . . . . 441 Electronic aids to calculation of results . . . . . . . . . . . . . . . 443 Estimation of confidence limits to the results . . . . . . . . . . . . . 445
296
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
Chapter 8 . Characteristics of binding assays .sensitivity . . . . . 446 8.1. Definition of sensitivity . . . . . . . . . . . . . . . . . . . . . . 8.2. Methods of increasing the sensitivity of a binding assay . . . . . . . . 8.2.1, Reducing the amount of tracer . . . . . . . . . . . . . . . . . 8.2.2. Reducing the amount of binder . . . . . . . . . . . . . . . . 8.2.3. Increasing the incubation time . . . . . . . . . . . . . . . . . 8.2.4. Reducing the incubation time - disequilibrium assays . . . . . . . 8.2.5. Order of addition of reagents . . . . . . . . . . . . . . . . . 8.2.6. Purification of the binder . . . . . . . . . . . . . . . . . . . 8.2.7. Increasing the sample volume . . . . . . . . . . . . . . . . . 8.2.8. Temperature of incubation . . . . . . . . . . . . . . . . . . . 8.2.9. Increasing the number of replicates . . . . . . . . . . . . . . . 8.2.10. Extraction and concentration . . . . . . . . . . . . . . . . . . 8.3. Methods of decreasing the sensitivity of an assay . . . . . . . . . . . 8.4. Targeting of binding assay -the importance of ranges . . . . . . . . . 8.5. Optimisation of an assay by theoretical analysis . . . . . . . . . . . . 8.6. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . .
Chapter 9 . Characteristics of binding assays
446 448 449 450 452 453 454 454 456 456 457 457 457 460 461 462
specificity . . . . 463
.
9.1. Definition of specificity . . . . . . . . . . . . . . . . . . . . . . 463 9.2. Specific non-specificity . . . . . . . . . . . . . . . . . . . . . . 464 9.2.1. The basis of specific non-specificity . . . . . . . . . . . . . . . 464 9.2.2. Assessment of specific non-specificity . . . . . . . . . . . . . . . 466 9.2.3. Methods for improving specificity . . . . . . . . . . . . . . . . 469 9.3. Non-specific non-specificity . . . . . . . . . . . . . . . . . . . . 472 9.3. I . Presence of materials which interfere with the binder-ligand reaction . 473 9.3.2. Variations of blank values in samples . . . . . . . . . . . . . . . 473 9.3.3. Destruction or sequestration of binder or tracer . . . . . . . . . . 474 9.3.4. Destruction or sequestration of unlabelled ligand . . . . . . . . . . 475 9.3.5. The detection and elimination of non-specific non-specificity . . . . . 476
Chapter 10. Characteristics of binding assays -precision . 10.1, Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2. Factors affecting precision . . . . . . . . . . . . . . . . . . . . 10.2.1. Errors in the reagents and design of the assay . . . . . . . . . . 10.2.1.1. Errors in the primary reagents . . . . . . . . . . . . . . . 10.2.1.2. Errors due to the separation procedure . . . . . . . . . . . .
479 479 480 480 480 480
297
CONTENTS
10.2.1.3. Errors due to disequilibrium . . . . . . . . . . . . . . . 10.2.1.4. Errors due to standards . . . . . . . . . . . . . . . . . 10.2.1.5. Errors at different points on the standard curve . . . . . . . 10.2.1.6. Counting errors . . . . . . . . . . . . . . . . . . . . . 10.2.2. Errors in the technical operation of the assay . . . . . . . . . . 10.3. Methods for monitoring the precision of a binding assay . . . . . . . 10.3.1. Preparation of quality-control materials for monitoring precision . 10.3.2. Methods for examining precision using quality-control materials . 10.3.3. Other methods of monitoring precision . . . . . . . . . . . . 10.4. Methods for optimising the precision of a binding assay . . . . . . .
. 481 . 481 . 484 485
. 487 . 488
. . . .
488 490 493 494
Chapter I 1 . Characteristics of' binding assays .relation to other 495 types o f assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495 Receptor assays . . . . . . . . . . . . . . . . . . . . . . . . . 496 Assays using circulating binding proteins . . . . . . . . . . . . . . 496 lmmunoassays . . . . . . . . . . . . . . . . . . . . . . . . . 497 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . 503
11.1. Definition
I 1.2. 11.3. 11.4. 11.5.
Chapter 12. Automation o f binding assays . . . . . . . . . . . . 504 12.1. 12.2. 12.3. 12.4. 12.5. 12.6. 12.7. 12.8.
General . . . . . . . . . . . . . . . . . . . . Identification and dispensing of the sample . . . . Addition of reagents . . . . . . . . . . . . . . . Incubation . . . . . . . . . . . . . . . . . . . Separation of bound and free ligand . . . . . . . Counting of radioactivity . . . . . . . . . . . . . Calculation of results . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . .
. . . . . . . . 504 . . . . . . . . . 505 . . . . . . . . 506 . . . . . . . . 506 . . . . . . . . . 507 . . . . . . . . 508 . . . . . . . . 508 . . . . . . . . 509
Chapter 13. Organisation o f assay services . . . . . . . . . . . . 510 13.1. Who should perform radioimmunoassay? . . . . . . . . . . . . . . 510 13.2. Organisation of an assay laboratory . . . . . . . . . . . . . . . . 511 13.3. Organisation of assay services . . . . . . . . . . . . . . . . . . . 515
Appendix I. Manufacturers and suppliers of equipment . . . . . 518
298
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
Appendix II . Suppliers of special reagents and chemicals . . . . . 520 Appendix III. Suppliers of general reagents and materials . . . . 521 Appendix I V. Manufacturers oj'reagent kitsjor radioimmunoassay and related techniques . . . . . . . . . . . . . . . 522 Appendix V. Safety precautions in the handling of radioactive 523 isotopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
527
Subject index . . . . . . . . . . . . . . . . . . . . . . . . . . .
531
References
List of abbreviations
RIA EIA FIA CPB RRA Ag Ab AgAb
= radioimmunoassay = enzymoimmunoassay
= fluoroimmunoassay = competitive protein binding = radioreceptor assay
=antigen =antibody = antigen-antibody complex k, = forward association constant k, = reverse association constant K = affinity constant B = bound fraction F = free fraction = proportion of tracer bound as "/, b of that in zero standard pg = microgram ng =nanogram pg = picogram fg = femtogram pl = microlitre pm =micron mol = mole M =molar mM = millimolar Ci = curie mCi = millicurie pCi = microcurie 'H =tritium I4C = carbon 14 lzsl =iodine 125 "'1 =iodine 131
cpm =counts per minute PPO = 2,5-diphenyloxazole POPOP = 1,4-di-2(5-phenyloxazolyl)benzene v =volts MW = molecular weight Log e =natural logarithm v/v = volume by volume w/v = weight by volume hPL =human placental lactogen hGH =human growth hormone ACTH = adrenocorticotrophic hormone LH = luteinizing hormone FSH = follicle-stimulating hormone TSH = thyroid-stimulating hormone AVP = arginine-vasopressin LVP = lysine-vasopressin hCG =human chorionic gonadotrophin AFP = a-fetoprotein CEA = carcinoembryonic antigen T, = triiodothyronine T4 =thyroxine CBG =cortisol-binding globulin TBG = thyroxine-binding globulin SHBG =sex hormone-binding globulin LATS = long-acting thyroid stimulator AMP = adenosine monophosphate PGF,, = prostaglandin F,, = fragment D of fibrinogen FgD = fragment E of fibrinogen FgE 299
300
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
I gG = immunoglobulin G DNP = bovine serum albumen WHO BSA 2nd IRP-HMG = second international reference preparation RSO of human menopausal gonadotrophins
= dinitrophenol
=World Health Organisation = radiation safety officer
CHAPTER 1
The background to radioimmunoassay
I . I . Introduction The introduction of radioimmunoassay (Yalow and Berson 1960) represents probably the single most important advance in biological measurement ofthe past two decades. Together with related techniques, similar in principle but using binders other than antibodies, it has revolutionised one major discipline - endocrinology - and is now exerting a similar influence in other fields, notably haematology, pharmacology and cancer detection. This success can be attributed to 3 factors. First, that it has many advantages when compared with previous methods; for instance, in the case of a hormone, the gain in sensitivity, specificity, and ease of performance over a classical biological assay. Second, that it can be applied to substances for which there was no previous assay method. Finally, and perhaps most important, is that radioimmunoassay and related techniques offer a general system for the measurement of an immensely wide range of materials. Thus, a technician skilled in the radioimmunoassay of a protein such as growth hormone would find little difficulty, either conceptual or in terms of equipment required, in setting up a comparable assay for a steroid hormone such as cortisol, a drug such as digoxin, or a cancer product such as carcino-embryonic antigen. The aim of this book is to set out the basic concepts of radioimmunoassay and related techniques in such a way as to assist those who are embarking on the subject for the first time, or who already have some practical experience but are seeking to broaden their knowledge of the subject as a whole. It does not set out to provide an exhaustive 301
Subpcl m d n p 531
302
RADlOlM MUNOASSAY AND RELATED TECHNIQUES
review, which would now be almost impossible with the size and rapid growth ofthe subject, or to provide detailed recipes for individual assays which would again, and for the same reasons, be impossible. Instead, an attempt will be made to underline the common ground in all assays of this type, whether it be in the preparation of materials, the use of these materials or the interpretation of results. Specific examples, where they are used, are always intended to illustrate the general rather than the particular.
I .2. Terminology It is necessary at an early stage to define some terms which will recur throughout the book. The techniques discussed all involve the combination of two materials, one of which is referred to as the ‘binder’ and the other, the substance being measured, as the ‘ligand’. These non-specific terms are valuable because radioimmunoassay is only one, albeit the most widely applied and known, of a series of related techniques. The term which best embraces the whole field is ‘binding assay’; this is any procedure in which quantitation of a material depends on the progressive saturation of a specific binder by that material, and the subsequent determination of its distribution between ‘bound’ and ‘free’ phases. Thus binding assays represent a general principle which can include many related techniques, but which do not specify the nature of the binder, or the methods used for determining the distribution between the bound and free phases (Table 1.1). The most familiar subdivisions of binding assays are based on the nature of the binder employed: ‘immunoassay’ where this is an antibody; ‘competitive protein binding assay’ where this is a naturally occurring binding protein in the circulation ; and ‘receptor assay’ where the binder is a naturally occurring cell receptor. Determination of the distribution between bound and free phases usually depends on physicochemical separation of these phases, the distribution being followed by the incorporation of a ‘tracer’ consisting of a small amount of the ligand labelled with a radioactive isotope. However, neither of these features are implicit in the definition of a binding
Ch. 1
303
BACKGROUND
TABLE 1.1 The place of radioimmunoassay and related techniques among assays for biological materials The 3 basic procedures are biological, physicochemical, and binding assays. ‘Binding’ assays may be directed at the measurement of binder or ligand. In assays for the ligand. 3 different types of binder are available, which in principle may be used with any of the 6 types of tracer shown. With any combination of these, the final step may separate bound and free ligand, or bound and free binder, or may not require any separation Assays for biological materials
i
Biological assays
Binding assays
Physicochemical assays
r
Assays for the binder
Assays for ligand
Antibody
None
Isotope
Separation of bound and free ligand
Circulating binding protein
Enzyme
Fluorescent
No separation
Cell receptors
Phage
Particle
Separation of bound and free binder
} }
]
types O f
binder
types
of traeer
types
of separrrt ion
304
R A D I O I M M U N O A S S A Y A N D RE1.ATE.D TECHNIQUES
assay. For instance, it is possible to use labelled binder as tracer rather than labelled ligand ; this is the ‘immunoradiometric’ assay. The label does not have to be a radioisotope: it can equally be any material which can be measured precisely in very small quantities, such as a fluorescent compound (fluoroimmunoassay) or an enzyme (enzymoimmunoassay). Finally, the separation of the phases is not always essential: if the characteristics of the tracer in the bound and free phases are sufficiently different then no separation is necessary.
1.3. Early development of radioimrnunoassay Recent discussion of the history of binding assays has been marred by claims and counter-claims for priority in the development of the technique. This may be expected in any subject which expands far beyond the expectations of its originators. The argument will not be described in detail here but a simple and doubtless equally arguable conclusion will be put forward: that a technique using an antibody as the binder (i.e. radioimmunoassay) was first described by Yalow and Berson in the U.S.A. (Yalow and Berson 1960); and that a technique usinga naturally occurring binding protein (i.e. competitive protein binding assay) was described by Ekins in the U.K. (Ekins 1960). For a full historical perspective one must look back further than the original description of individual methods. It is noteworthy that the first techniques were developed in laboratories of nuclear medicine. The common thread was the ability to handle radioactive isotopes, which in turn had become available for routine biomedical use as a byproduct of the development of nuclear weapons. The striking feature of many radioactive isotopes is that the energy of their emission is so great that it becomes possible to detect even a few atoms with relatively simple equipment. When such atoms are attached to another molecule, then the latter can be detected in equally small numbers. In practice, it was probably the ability to measure very small quantities of a tracer compound, rather than an appreciation of the potential use of antibodies and other binders of high affinity,
Ch. 1
305
BACK Ci KOUN D
which led to the original development of radioimmunoassay. Claims for the originality of the concept of the technique must go back much further since it is, in essence, a combination of Archimedes' principle and the law of mass action. The work of Yalow and Berson arose from their earlier studies on the behaviour of "'I-labelled proteins in vivo; at the time (the early 1950's) this was a new and highly effective approach to the study of protein metabolism. One of the materials studied was "'1-labelled insulin, and led to the demonstration that insulin-requiring diabetics almost always have a circulating insulin-binding protein (Berson et al. 1956). Yalow and Berson encountered considerable difficulty in convincing the scientific world that this binding protein was an antibody, a conclusion in which they have by now been vindicated many thousands of times. At the same time as demonstrating binding of ['"I]insulin, they also showed that the tracer could be displaced from the binder by the addition of large quantities of unlabelled insulin, and recognised that the degree of binding of the tracer was quantitatively related to the total amount of insulin present. These observations formed the basis of the first radioimmunoassay for insulin which was applied to the measurement of exogenously administered beef insulin in rabbits (Yalow and Berson 1971). However, it soon became apparent that the sensitivity of the assay using antibodies from the sera of insulin-treated diabetics was inadequate for the measurement of the relatively low circulating levels of endogenous human insulin. One reason for this was that the antisera were directed to beef insulin; human insulin, though it did crossreact in the system, did so only at a much lower energy. The next step, therefore, was to prepare specific antisera to human insulin. which was eventually achieved by the immunisation of guinea-pigs and resulted in the first radioimmunoassay capable ofdetectingendogenous insulin in human blood (Yalow and Berson 1960). While this work was in progress, Ekins in the United Kingdom was examining the use of thyroxine-binding globulin for the measurement of thyroxine in human plasma. Together with techniques for vitamin B,? using intrinsic factor as the binding substance, these Strhlrcr
i i i d [ ~ \p
531
306
RADlOlM MUNOASSAY A N D KELA'TED TECHNIQUES
studies underlie all subsequent work on circulating binding protein assays. Their application to steroid hormones, pioneered by Beverly Murphy, has proved particularly successful. Of the many later developments which have advanced the field to the stage at which we know it today, one in particular should be singled out - the description of the chloramine T iodination procedure for peptides by Greenwood et al. (1963). This represented a massive practical advance over most of the earlier methods, which involved the handling of large quantities of radioactive isotopes and were thus limited to those laboratories with the most elaborate and expensive facilities. The method is now in almost universal use, and its simplicity and efficiency must have greatly enhanced the rate of development of new assays.
1.4. Basic principles ojbinding assays The basic principles of binding assays are simple once grasped. However, there are eminent practitioners of the subject who confess to having taken their first technical steps without proper appreciation of the underlying mechanisms. The newcomer is not helped by the fact that almost every writer on the subject has their own way of explaining the principles and that in the course of the explanation they are likely to denigrate the views of others as oversimplified or even positively in error. Much of the argument is semantic, since all are looking at the same basic system with the same basic endpoints, albeit reached by different paths. For the sake of simplicity and familiarity, it will be assumed here that the technique under discussion is a radioimmunoassay, i.e. using an antibody as the binder and an antigen as the ligand. The same principles, however, apply to any of the other systems which come under the heading of binding assays. A simple illustration of the mechanism of a radioimmunoassay is presented in Fig. 1 . 1 which shows binding of antigen to antibody in the presence of different total masses of antigen. The distribution of the antigen between the bound and free phases is directly related to the total amount of antigen present and thus provides a means for quantitating the latter.
Ch. 1
307
BACKGROUND
0 0O Antibody
Antigen
+
Anti body
..x+>k.:.* *.Gg.::*.*. 5**:*.:$:k.: ............... :.:.:.: . + ............. y . : . ; . : . : ........... .:gk.:.>:.:.:.:.:.fi ............. ...........*.p.. ................ ................... ..;.:.:.:.: ..:.:.:.:.:... ............. ~~~~~~
Antigen
Q
............ ............. ........... ................. ................... .........: .......... .................... ........................ iiiiiiiiBfi fiF% ..................... ....................... ........... .:::::::::: ........... .................. ............ ............. ............... ............ .:::::::::::: ::::::::::
a.::.
Fig. 1. I . The basic principle of a binding assay. using immunoassay as an example. If given amounts of antigen and antibody are allowed to react together (above) then a t equilibrium they will form an antigen-antibody complex (the overlapping area B) together with a proportion of both the antibody and the antigen (F) which remain free. If the amount of antibody is held constant while the total amount of antigen is increased (below) then at equilibrium the amount of antigen-antibody complex (B) is increased; however, the increase i n the free fraction (F) is relatively greater and thus yields a lower bound to free ratio.
A more productive approach to understanding the principle is through consideration of reversible reactions such as the reaction between an antigen and an antibody to form the antigen-antibody complex : Ag + Ab
k, =$
AgAb
(1.1)
k2
In this equation, the rate constant of the forward reaction (association of antigen and antibody to form a complex) is denoted by k , , and the rate constant of the reverse reaction (dissociation of the antigenantibody complex) is denoted by k2. It should be noted that k , and k, are constants, in other words they describe the @action of the available molecules which will react within unit time. The absolute rate - the number of molecules which react in unit time - is obviously Subjccr mC.r p . 531
308
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
dependent on the concentration of molecules. Thus, when the reaction begins with the addition of antigen and antibody, the forward reaction rate is high and the reverse reaction rate correspondingly low.
As the reaction proceeds, the concentrations of free Ag and Ab decrease, and so too will the forward reaction rate. At the same time the concentration of the antigen-antibody complex increases, and with it the rate of the reverse reaction. Eventually the stage is reached where the number of free Ag and Ab molecules reacting in unit time to form AgAb is identical with the number of AgAb molecules which dissociate ir this time: Ag +- Ab
e AgAb
(1.3)
At this stage of equilibrium there will be no further net change in the concentrations of the molecules on the two sides of the equation. In most, but not all, radioimmunoassay systems the reagents are permitted to react for long enough for equilibrium to be reached, and eq. (1.3) can therefore be used as the basis for further discussion. The exact concentrations which are reached at equilibrium will depend on the energy with which the binder and ligand react. In thermodynamic terminology, this can be described as A S , or the change in free energy within the system as a whole. The situation can be described by the law of mass action which states that, at equilibrium, the ratio of the products of the concentrations on the two sides of the equation will be a constant, usually designated as K :
where [Ag], [Ab] and [AgAb] are the concentrations of free antigen, free antibody and antigen-antibody complex respectively, and K is
Ch. 1
309
BACKGROUND
the affinity constant*. Since the final concentrations are determined by the rate constants of the forward and reverse reactions it is also apparent that:
The affinity constant, K, for any system provides a measure of the energy of reaction between the original reagents, in this case Ag and Ab. Given fixed starting concentrations, a high K value would imply that at equilibrium the concentration of the Ag-Ab complex will be much greater than that of free Ag and Ab, while a low K value would imply the inverse situation. Assume a system of given K value, that the initial molar concentration of Ag and Ab are equal, and that these concentrations are chosen so that at equilibrium both Agand Ab will be equally distributed between the bound and free phases. Let us also assume that the starting concentration of each reagent is 2 (which could be in any concentration units). At equilibrium: Ag + Ab 1 1
+ AgAb 1
Looking at the antigen alone, 1 unit is bound, and 1 unit is free. Now take the same system but with 4 units of Ag rather than 2. It follows from the law of mass action that an increase in the concentration of reactants on one side of the equation will produce a corresponding increase on the other side, i.e. an increase in Ag or Ab, or both will produce an increase in the concentration of AgAb at equilibrium. But in the case described, an increase in AgAb cannot be equivalent to that in Ag, because Ab has remained unchanged.
* The units of K are litres per mole (l/mol) which can be confusing to the uninitiated. It arises from the fact that K is derived by dividing a concentration (molil) by the product of two other concentrations. i.e. mol/l - I/mol. mol/l x mol/l mol/l
'
S u h p r mde/rl p. 531
310
KADIOIMMUNOASSAY A N D RELATED TECHNIQUES
Thus, although the concentration of AgAb will increase, that of free Ag will increase relatively more. For the sake of illustration, some entirely hypothetical figures may be given to this situation: Ag + Ab 2.8 0.8
+ AgAb 1.2
Looking again at the antigen alone, 1.2 units are bound, but 2.8 units are free. In the situation of eq. (1.6), the ratio of bound to free antigen at equilibrium was 1. The addition of more antigen has decreased this ratio to 0.43. A conclusion may now be drawn :given an unvarying quantity of’ binder offi’xed K value, the ratio ojbound to jree ligand at equilibrium will be quantitatively related to the total amount of ligand present. This is the basic principle of all binding assays. The concepts of tracer, of standards, and of the separation of bound and free phases have not yet been introduced because all of these are secondary to the underlying principle, and not defining features. For example, in most if not all cases the labelled ligand should be identical with the unlabelled ligand and should behave identically in combination with the binder. Under these circumstances the eventual distribution between bound and free phases will be determined by the total amount of ligand present and not independently by the amount of labelled or unlabelled ligand. The tracer is incorporated simply because it provides a technically convenient means for measuring the distribution of bound and free; the irrelevance of tracer ligand to the basic principle is illustrated by the fact that it is equally possible to use the binder as the tracer (see 9 5.4). Following on these arguments, the use of the term ‘competition’ which is frequently used to describe the relationship between labelled and unlabelled ligand should be discouraged. In most binding assays both have equal opportunities for combination with the binder, and the final result depends on the total number of ligand molecules present. Standards - a series of different concentrations of purified hormone against which the results of unknowns can be judged - are again a technical convenience. It is quite possible to conceive a situation in
Ch. I
BACKGKOUND
31 1
which, knowing the exact concentrations of tracer and binder, and the affinity constant of the latter, the concentration of an unknown could be determined by simple calculation without the need for reference to standards. In practice, however, the vagaries of reagents are such that this hypothetical situation cannot be achieved. Finally, the separation of the bound and free phases is yet another technical manoeuvre unrelated to the basic principle ; unrelated because, as will be seen later, systems have been described in which separation is unnecessary and the distribution can be followed through alterations in the characteristics of the tracer ligand when combined with the binder.
I .5. Binder dilution curves and standard curves The discussion of principles will now be extended to the two basic experimental procedures of saturation analysis - the setting up of binder dilution curves and of standard curves. The binder dilution curve (or antibody dilution curve in the context of a radioimmunoassay) involves the incubation of a fixed amount of tracer ligand with different concentrations of the binder; the latter might, for example, consist of serial doubling dilutions of an antiserum. Following incubation, the distribution of the tracer in the bound and free fractions is ascertained. The general appearance of this type of curve is shown in Fig. 1.2. Plotted as percentage of tracer bound against serial dilutions of the binder on a logarithmic scale, the curve is sigmoidal. Given a suitable set of reagents, the construction of this curve is the first step in setting up a binding assay system, since the results determine the amount of binder optimal for use in a standard curve. As a rule of thumb, the concentration of binder chosen for use in a standard curve will be that which is sufficient to bind approximately 50'j/, of the tracer (Fig. 1.2). Exceptions to this rule will be discussed in later sections. At this concentration, at which the amount of binder is sufficient to fix only half of the tracer, it is apparent that the addition of further ligand must lead to a substantially greater increase in the Subircl i n d r x p . 531
312
RADIOIMMUNOASSAY A N D R E L A l F D T E C H N I Q U F S
loo
-
binder concentration
80-
60-
4020-
Fig. 1.2. A binder dilution curve (e.g. an antibody dilution curve). Serial dilutions of the binder (horizontal axis) are incubated with a fixed amount of tracer ligand, and the percentage ofthc latter bound plotted on the vertical axis. In a typical binding assay the amount of binder chosen is that which will bind 50?,<,of the tracer; however, there are many exceptions to this. The example shown is derived from a theoretical model of a binding assay (see # 1.9).
free fraction than the bound fraction. By contrast, if a much higher concentration of binder is chosen (e.g. 2.5 in Fig. 1.2) the amount of ligand required to produce a significant shift in the bound and free fractions will be much greater, and the eventual assay less sensitive. The standard curve involves the incubation of fixed amounts of tracer ligand and binder with different concentrations of purified unlabelled ligand. Plotted as the percentage of tracer bound against serial dilutions of the ligand on a logarithmic scale this again gives a sigmoid curve (Fig. 1.3). At the upper end of this curve, the small quantities of unlabelled ligand present produce relatively minor shifts in the distribution between the bound and free phases. At the lower end of the curve, the total concentrations of ligand relative to
Cll. I
313
BACKGROUND
40.
%TRACER BOUND
3. 20. 10
. 1' .008
.032
.128
.5
2
CONCENTRATION OF STANDARD
Fig. 1.3. A standard curve. A fixed amount of tracer ligand and binder (the latter at the concentration shown in Fig. 1.2) is incubated with varying concentrations of standard unlabelled ligand (horizontal axis). The percentage of the tracer bound (vertical axis) is progressively reduced with increasing concentrations of standard. Note, however, from Fig. 1.1 that the fotul amount of ligand bound (tracer plus standard) is actually increased as the propor/ion of ljgand bound is decreasing.
that of the binder is such that the majority is in the free form. Between these two extremes is a range of ligand concentrations at which relatively small changes produce a significant alteration in the distribution of bound and free. For the situation shown in Fig. 1.3 this range would be from 0.032 to 2. In practical terms this, the steepest part of the slope, represents the effective range of the assay. The standard curve is the basic requirement for quantitation of the ligand in unknown samples. When the sample is substituted for the standard, and using the same fixed concentrations of binder and tracer, the value determined for the distribution of the bound and free phases will be equivalent to some value on the horizontal scale of the standard curve. This value can be read by simple extrapolation (Fig. 1.4). S d j e c r index p. 531
314
RADIOIMMUNOASSAY A N D KELAI € D 1ECHNIQUES
.oos
.(n2
,128
.5
2
Fig. 1.4. Estimation of the amount of ligand in an unknown sample using a standard curve. The conditions are identical with those shown in Fig. 1.3. The unknown sample is incubated with the same fixed concentrations of binder and tracer. and the percentage of tracer bound is approximatcly 34';b; this corresponds to a value of 0.128 for the standards. and this i s therefore the concentration in the unknown.
Practical techniques for setting up binder dilution curves and standard curves are shown in Tables 1.2 and 1.3 and Figs. 1.5-1.7. It must be emphasised that these, together with all other detailed technical descriptions in the book, are intended as examples. The potential range of variations, including buffers used, volumes, time of incubation and temperature, is enormous. Some of these variations will have little effect on the results obtained; others, for example order of addition of reagents, may have dramatic effects which are considered in later chapters.
1.6. Methods ofplotting the standard curve The convention followed here (Fig. 1.3) is a semilogarithmic plot of percentage tracer bound against the log of concentration of unlabelled ligand. This is a popular type of presentation, but there are many
Ch. I
BACKGROUND
315
T A B L E1.2 Construction of a binder dilution curve (Radioimmunoassay of human placental lactogen, hPL) Diluent huifbrt 0.05 M phosphate, pH 7.5, containing 2 mg/ml bovine serum albumin (see Table I .4for buffer preparations). 1. Prepare serial doubling dilutions of an antiserum to hPL in 1 ml of diluent buffer
(see Figs. 1.5 and 1.6).The range to be covered will vary between different antisera, but typically would be I : 100- 1 : 100.000. 2. Dispense 0.25 ml of each dilution into duplicate tubes; include one pair containing diluent buffer with no antiserum to act as an assay blank. 3. To each tube add 0.2 ml of a solution of ['251]hPL, diluted such that the total counts per tube are approximately 10,000- 15.000 in 10 sec. Include one pair of tubes containing tracer alone for measurement of total counts. 4. To each tube add 0.05 ml of serum or plasma from a non-pregnant adult or a non-human species such a s the cow or horse. 5. lncubate for I h a t room temperature. 6. Add I ml of a 20'>(, solution of polyethylene glycol and proceed as shown in Table 5.3. 7. Place the tube in the well crystal of a gamma counter and count for 10 sec. 8. Plot the results a s of tracer bound on the vertical axis against the dilution of antiserum on a log scale on the horizontal axis (see Fig. 1.2). I;;,
Comments. The procedure outlined has been chosen arbitrarily from among an immense range of potential variations, all of which are likely to make little difference to the end result. However, some general comments can be made about the individual steps. (a) As a point of good practice, a series of 'master' dilutions are made (volume I ml) from which aliquots are then dispensed into the incubation tubes; this is much preferable to making 2 sets of doubling dilutions in 0.25 ml. (b) The volume chosen (0.25 ml) is based on two factors: achieving a final incubation volume of 0.5 ml; and doing this with volumes of the individual reagents which can be accurately and precisely dispensed from widely available equipment. Any other convenient volumes could be used (e.g. 0.05 ml antibody and 0.4 ml tracer) provided that the final concentrations are correctly recorded. (c) The same observations on volume apply also to the tracer. The choice of the total amount of tracer added. and the counting time required, will vary extremely widely according to the substance involved. The principles for selection of this quantity are discussed in 9 8.2.1 and 9 8.3. (d) The volume of serum chosen is based on the decision that this shall make up I0):b of the total incubation volume. Smaller volumes may be difficult to dispense Suhjiw inhlcr ii.53/
316
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
with a high degree of precision; large volumes may yield non-specific effects (5 9.3). (e) As with other variations, the possible range of times and temperature is immense. In the early development of a new assay these factors should be investigated in detail, beginning with curves incubated for I , 4, 24, and 48 hr, at 4 C , room temperature, and 37°C. Longer incubation times should be particularly examined in those systems in which high sensitivity is a requirement. (f) Almost all separation procedures are applicable to hPL (see Table 5.1). The polyethylene glycol method has been chosen for its efficiency and simplicity. (g) It is not recommended that total counts are estimated for every tube. This will onty show pipetting errors with the tracer; similar but sometimes more serious errors with the other reagents are not revealed. Note: Because most commercial RIA kits for hPL contain an excess of antibody (in order to de-sensitise the system) they can be used as part of a class exercise in setting up binder dilution curves.
TABLE1.3 Construction of a standard curve (Radioimmunoassay of human placental lactogen, hPL) Diluenr bu//er: As for Table 1.2. I , Prepare standards of hPL in hPL-free serum or plasma. using a master standard containing 1 mg/ml (Table 2.1). 2. Dispense 0.05 ml of each dilution into duplicate tubes; include two pairs containing hPL-free serum alone to act as assay blank and '0 standard (Fig. 1.7). and further duplicate tubes containing serum or plasma from subjects in late pregnancy. 3. To each tube add 0.2 ml of a solution of [ '2SI]hPL, diluted such that the total counts per tube are approximately 10.000- 15.000 in 10 sec. Include one pair of tubes containing tracer alone for measurement of total counts. 4. To each tube add 0.25 ml of a 1 : 500 dilution of antiserum to hPL; in the case of the assay blank tubes add 0.25 ml diluent buffer in place of this. 5. Incubate for 1 hr at room temperature. 6 . Add 1 ml of a solution of 20% polyethylene glycol and proceed as shown in Table 5.3. 7. Count radioactivity of each precipitate. 8. Plot the results as tracer bound on the vertical axis against the concentration of standard on a log scale on the horizontal axis (see Fig. 1.3). 9. Calculate the results of unknowns as shown in Fig. 1.4. Comments. The procedure outlined has been chosen arbitrarily from among an immense range of potential variations, many of which have already been discussed
Ch. I
317
BAC K<JIIOUND
(see general comments to Table 1.2). Note that the choice of a I : 500 dilution of antiserum is arbitrary and would obviously have to be tested and varied with each different antiserum used. The dilution of antiserum appropriate to this type of desensitised assay represents a relative excess of antiserum and cannot. therefore. be extrapolated from the binder dilution curve (see 9 8.3).
Fig. 1.5. Preparation of doubling dilutions of antiserum (see Table 1.2).
Fig. 1.6. Technical procedure for an antiserum dilution curve (see Table 1.2). Strhpn rinb\ p 53)
318
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
Fig. 1.7. Setting up a standard curve and appropriate controls (see Table 1.3).
other ways of showing the same data. For the independent variable, there is only one commonly used alternative, an arithmetic scale of standard concentrations. Substitution of the latter in the percentage bound plot yields the standard curve shown in Fig. 1.8. For the dependent variable, by contrast, there are many possible alternatives. The most popular is to use the ratio of bound to free ligand ; Figs. 1.9 and 1.10 show this in arithmetic and logarithmic form. The inverse ratio, free to bound, can also be used (Figs. 1.11 and 1.12). A common variation on the percentage bound plot is to express the results as a percentage of the distribution observed in the zero standard (i.e. a tube containing fixed amounts of tracer and binder, but no
Fig. 1.8
conwnlration of siandard
Fig.
1.1)
concentration 01 standard
Ch. 1
freulbound ratio
boundlfree rdio
Fig. 1.10
319
BACKGROUND
concentration d stanbrd
Fig. 1.1 I
concentralion af standard
Fig. ' . I 3
concentration af standard
frwlbound ralio
Fig. 1.12
concentration of standard
%tracer bound
Lcmlb
0.1
Fig. I.14 concmtralion d
standard
Fig. 1.15
0.2
0.3
0.4
0.5
comntratlm of standard
Figs. 1.8-1.15. The standard curve of Fig. 1.3 is plotted in various ways. Superficially the appearances are very different although based on identical data. Subject tiidcx p. 531
3 20
IlADlOlM MUNOASSAY A N D KELA.rED TECHNIQUES
unlabelled ligand; this is also sometimes referred to as the ‘antibody blank’); this is shown in Fig. 1.13. Finally, the response variable may be described by several complex transformations of which the best known is the logit transformation (Rodbard and Lewald 1970): logit b
=
log e (
b ,
where b is the proportion of tracer bound expressed as a percentage of that in the zero standard. This type of plot has the advantage that in some, but not all situations, it gives a straight line rather than a curve (Fig. 1.14), thus simplifying calculation. The choice of plot is very much a matter of the personal taste or past experience of the individual worker. Good arguments can be put forward for and against all the types described. A graph of percentage bound against log dose (Fig. 1.3) has the advantage that it directly reflects the output from the tracer detection system. Thus, in the case of a radioimmunoassay, if counts/min is substituted for percentage bound, the shape of the curve would be identical. It is possible to see, at a glance, whether an apparent difference between two standard points is likely to be experimentally significant. This is not the case in the type of plot shown in Fig. 1.13. By describing all points in relation to the zero standard, it may conceal problems from the uninitiated; for instance that the true percentage bound in the zero standard is less than 20”/,, and thus that the precision of all subsequent points must be highly suspect. The plot of bound/free ratio against log dose (Fig. l.lO>, which is very popular in the United States, yields a curve very similar to that of percentage tracer bound. By contrast, the plot of free/bound ratio against log dose (Fig. 1.12) has a completely different appearance, and has the important disadvantage that the slope at various parts of thecurve is quite unrelated to the experimental findings. It is accepted that it presents data identical to that of other plots; nonetheless, to the unwary who may associate steepness of slope with precision, it would seem that the effective range of the assay lies between
Ch. 1
HACKGROUND
321
standard concentrations of 0.5 and 2 . A glance at the other plots clearly shows that this is not the case. Plots using an arithmetic scale of standard concentrations again have a quite different appearance (Figs. 1.8, 1.9 and 1.1 l), although based on the same data. Two criticisms may be levelled against this type of presentation. The first is similar to that already raised in the case of the free/bound ratio: that the apparent slope, if equated with precision, is unrelated to the experimental difference between successive standard points. This is clearly shown in Fig. 1.8. The second objection is that calculation of small ligand concentrations by extrapolation may be awkward, since the part of the scale covering these concentrations is relatively cramped. However, an arithmetic scale may offer practical advantages if the assay is directed to determination in a narrow physiological range, with a correspondingly narrow range of standards (Fig. 1.15, an enlarged portion of Fig. 1A), because manual extrapolation may be possible. It also permits plotting of the zero standard, which is not possible with a logarithmic plot. Finally, it is worth emphasising that the standards should always be expressed as a concentration and not as an absolute amount per tube. The latter can be very misleading as, for instance, when the minimum detection limit of an assay is described as 1 pg. This might represent the amount in a final volume of 0.1 ml, in which case the effective sensitivity is 10 pg/ml, or in 2 ml, in which case the effective sensitivity is 0.5 pg/ml. It would be a considerable service to the literature of binding assays if the use of concentration were made a rule.
1.7. The impqrtance oj’K value Returning to basic principles, the importance of the affinity constant (K value) between binder and ligand should now be emphasised. The end results of a binding assay system will be influenced by a variety of technical factors, such as the availability of pure ligand the ability to prepare tracer ligand of high specific activity, Suh,c
322
KADIOIMMUNOASSAY A N D R E L A I E D rECHNlQUES
and the elimination of non-specific interference by biological fluids. But these are secondary factors, and the fundamental characteristics of a binding assay are determined above ail by the energy of reaction as reflected by the K value. Since this theme will occur throughout the text, it is worthwhile to consider how the K value is measured, and some of the theoretical implications with respect to the characteristics of any assay system.
1.8. The measurement of K value The K value for any system can be calculated from the data contained in a standard curve, redrawn according to the method of Scatchard. Assuming that the binder is homogeneous and has a single K value, the ‘Scatchard plot’, with the bound/free ratio on the vertical axis and the bound fraction on the horizontal axis, yields a straight line, the slope of which is equal to the K value, and the intercept of which on the horizontal axis gives the total concentration of binder (Fig. 1.16). The basis of this relationship can be explained by further consideration of equation (1.4) :
I
\
Total concentration of binder
Concentration of bound fraction
Fig. 1.16. Diagram of a Scatchard plot used to estimate K . For a system with a single affinity constant (K value) this yields a straight line of slope-K, which intercepts the horizontal axis at a value equivalent to the total concentration of the binder.
Ch. 1
BACKGROUND
323
For simplicity, substitute X for [AgAb], the concentration of the bound complex at equilibrium; use Ag to describe the total concentrations of antigen in the system, and Ab to describe the total concentration of antibody. Equation (1.3) then becomes: X =K (Ag - X)(Ab - X)
(1.9)
Multiplying both sides by (Ab - X) gives us: (1.10) or : (1.11) Examining the equation in this form it can be seen that on the left hand side X is equivalent to the bound fraction, and (Ag - X) to the free. Therefore, X/(Ag - X) is the bound/free ratio. On the right hand side both K and Ab are constants; so too is K.Ab. The variable is -KX. Thus, a plot of X/(Ag - X), the bound/free ratio, against X, the bound fraction, gives a straight line with slope -K. This is the Scatchard plot. The amount of antibody Ab, is given by the intercept on the horizontal axis, in other words, where the bound/free ratio is 0. Substituting in equation ( 1 . 1 l), 0 = K . Ab - KX
(1.12)
so :
K - A b = KX
(1.13)
and :
Ab=X
(1.14)
Although simple in principle, in practice the measurement of K value can present many problems. This is particularly the case with Subpcr rnderp 531
324
RADIOIMMUNOASSAY AND RELATED TECHNIQUES
antibodies. With few exceptions, most antisera to a given antigen will contain several populations of antibodies, each one with a different K value. Under these circumstances the Scatchard plot is not a straight line, but instead a curve which represents a combination of the lines from each specific antibody population. The fact that antibody molecules are divalent, and that binding at one site may influence binding at the other, should also be taken into account, as well as the possibility that the separation system disrupts the equilibrium. Another type of problem is seen with the naturally occurring binding proteins, such as cortisol-binding globulin ; the apparent K value varies according to the temperature at which it is measured, being greater at 4 C thanat 37'C. A similar but less striking variation may be seen with antibodies to peptides, whereas antibodies to steroids or other small haptens show little or no variation (Keane et al. 1976). '
I .9. A model system for binding assays For those systems in which it is possible to make a reasonable estimate of a single K value, it would be very helpful to use this estimate to predict the characteristics of the assay at various concentrations of reagents. A good example is the choice of tracer concentrations which will yield maximal sensitivity (see 9; 8.2.1); this involves a lengthy series of experiments, many of which could be eliminated if a good theoretical prediction were possible. To use an estimated K value in this way requires a model system. Several have been described, but most demand sophisticated computing facilities which are often either unavailable, or in themselves take longer to set up than would the empirical exploration of many different assays. A simple model system will be described here, derived from the law of mass action, which can readily be used by anyone with access to a programmable calculator. The principle is to take equation (1.9) : X (1.9) (Ag -X) (Ab - X) =
Ch. 1
325
BACKCIKOUND
and to turn it into quadratic form so that it can be rapidly solved for any set of values entered : [Ag] [Ab] - X[Ab]
+ [Ab]
+ ( l / K ) + X2 = 0
(1.15)
A solution in terms of a percentage is provided by the equation: percent bound
=
(a - ,/-)c
(1.16)
where a = [Ab] + [Ag] + 1/K, b = 4 [Ag] [Ab], and c = 5O/[Ag]. Simple additions to the calculator programme permit the evaluation of series of doubling dilutions of the binder or of the standard.
1.10. Some implications oftlie model system As an example of the application of the model system, the question of the choice of the optimal concentration of tracer for use in a binding assay will be considered. This is a question which almost invariably arises in the initial development of an assay, and will be discussed in some detail in 5 8.2.1. If a series of binder dilution curves are constructed, using successively smaller concentrations of tracer, the pattern observed will be that shown in Fig. 1.17 for both a model system and a real system. With the larger amounts of tracer, reduction in tracer concentrations leads to a shift of the curve to the right; the concentration of binder required to bind 50% of the tracer progressively decreases. However, beyond a certain point further reduction in tracer concentration produces no significant shift of the curve or of apparent titre. The point a t which this occurs is determined entirely by the K value of the binder with respect to the ligand. It is not, as occasionally suggested. a result of the failure of small concentrations of reactants to reach equilibrium. The importance of these observations is that the limiting level of tracer in a series of binder dilution curves is also the limiting level for use in a standard curve. This is illustrated in Fig. 1.18. Reduction of tracer concentration in a series of standard curves results in a progressive shift to the left and an increase in the sensitivity of the system Suhleci index p . 531
326
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
im
r
k BOUND
-
'
01' 32
'
'
8 4 2 1 0.5 A M O U M OF A M l 0 O D Y
16
'
a25
25
m 50 im 200 400 8a) id00 32m RECIPROCAL OF AMIBODY DILUTION
Fig. 1.17. Antibody dilution curves derived from a theoretical model of a radioimmunoassay (left; see $ 1.9) and from a radioimmunoassay for oxytocin (right). Ag* represents labelled antigen. Note that in both systems there is a limiting concentration of tracer antigen below which there is no further shift in the position of the dilution curve.
'
1
2
4 8 1 6 3 2 6 4 UNIABELLED ANTIGEN
B 125 250 Mo 1wO 2MwI 4w0 pg UNlABELLED OXYTOCIN
Fig. 1.18. Standard curves derived from a theoretical model of a radioimmunoassay (left; see $ 1.9) and from a radioimmunoassay for oxytocin (right). Ag* represents labelled antigen. Note that in both systems there is a limiting concentration of tracer antigen below which there is no further shift in the position of the standard curve.
Ch. 1
327
BACKGROUND
(defined as the minimum detection limit, see 9 8.1). But a t the same limiting concentration of tracer seen in the dilution curves, there is no further change in the standard curves. Clearly, therefore, it would be superfluous to try to achieve additional sensitivity by the use of TABLE1.4 Buffers for use in radioimmunoassay
A wide range of buffers has been used in radioimmunoassay. With few exceptions the exact nature of the buffer is irrelevant provided it meets the following criteria: ( I ) p H within 1 unit of neutrality (i.e. 6-8); (2) molarity in the range 0.01-0.1; (3) freedom from organisms; (4) freedam from contamination with ligand*. Buffers are usually made up in distilled or de-ionised water; in some parts of the world, tap water would be equally satisfactory though a more stringent specification would suggest that conductance should be less than 1 ymho. The composition of 3 commonly used bufi'ers is given here. Some commonly used additives are shown in Table 1.5. In some commercial kits dyes are added to aid identification, e.g. blue for antibody and yellow for tracer yielding green in the final mixture. It is good practice to check the p H of a buffer after preparation, either using a p H meter or narrow range indicator sticks (Neutralit, Merck). Salt Phosphate buffer, p H 7.4 Na2HP04. 12H20 NaH,P04 2 H 2 0
MW
Amount in g/l
358.22 156.03
14.5
Tris buffer, pH 8.0 Tris (hydroxymethyl) 121.14 aminomethane Titrate t o pH 8.0 with 1 M HCI (approx. 29 ml) Barbiturate buffer, pH 7.6 Sodium diethylbarbiturate 206. I8 Titrate to pH 7.6 with 1 M HCl (approx. 33 ml)
I .48
6.07
10.3
~~
*
A routine and apparently simple assay occasionally goes dramatically wrong because of contamination of one of the basic reagents with ligand. A few crystals of purified material can turn a 20 litre container of distilled water into a potent standard solution. S u h p r mdev p 531
328
KADIOIMMUNOASSAY AND RELATLD TtCHNlQUES
TAHLE 1.5 Common additives to buffer used in radioimmunoassay Material
Source
Concentration
Proteins such as bovine serum albumin’ Sodium azide Merthiolate Trasylol
Armour Pharmaceutical Co.
1-10 rngiinl
-
Bayer Pliarniaccuticals
0.2-1 mg/ml 0.1 mglrnl 200 KIU/ml*
Calcium salts
-
0.005 M
I
2
E. Lilly & Co.
Purposc
To stabilise and prevent absorption losses o n to surfaces of tubes. etc. Bacteriostatic Bacteriostatic Inhibitor of trypsin-like enzymes Enhance activity of certain antiscra
Batches or purified proteins for use in radioimmunoassay should bc checked for enzyme activity (which may lead to protcolysis of a traccr with corresponding reduction in the ‘zero’ Standard) and for prescncc of endogenous ligand (particularly with materials which are not species specific such as thyroxine and cortisol). Set up ‘zero‘ standards and assay blanks containing the new batch of protein, compare rcsults from a prcvious satisfactory batch. Kallikrein inhibitor units.
tracer concentrations of less than 0.0125 (for the model system), or 8 pg (for the oxytocin radioimmunoassay). Fig. 1.18 demonstrates a practical application of the model system. Given a set of reagents whose characteristics are unknown (for example, oxytocin and an antibody to oxytocin), it would be possible from a single standard curve to estimate the K value, and then to use this figure in the model system to evaluate a range of different reagent concentrations. Obviously it would be necessary to check the result experimentally, but the procedure would have two advantages: first, as a short cut to eliminate a considerable amount of trial and error; second, to provide a better understanding of the underlying basis of the technique. Equally, there are many systems in which the heterogeneity of the binder and the difficulty of estimating a single K value, would vitiate any attempt at theoretical analysis.
CHAPTER 2
Requirements for a binding assay - purified ligand
2.I . Requirements for a binding assay The requirements for a binding assay are purified ligand, tracer ligand, a specific binder, a method for distinguishing bound and free ligand, and, in some cases, a procedure for extracting the ligand from biological fluids.
2.2. The need,for purified ligand A supply of highly purified ligand is an essential prerequisite to the development of any binding assay and the application of the technique is limited to the substances for which this criterion can be met. If highly purified material is freely available (as, for example, with many of the steroid hormones, and with the synthetic small peptide hormones), then it will be routinely used for standardisation, for the preparation of tracer, for immunisation in the case of a radioimmunoassay, and for monitoring recovery from biological fluids. If appropriate material is in very short supply (as for example, with some of the hormones which have to be purified from biological sources), relatively less pure material may be used at some stages of the system. But this is a compromise, and may be applied only insofar as at least one of the basic reagents is pure. To illustrate the last point, consider a theoretical example. Experimental work reveals the probable existence of a hitherto unknown hormone in the pituitary gland; a biological assay is available, but is very insensitive, giving a positive response with pituitary extracts 329
Ssb,ucr mder p 531
330
RADIOIM MUNOASSAY A N D RELATED TECHNIQUES
but not with samples of peripheral fluids such as blood. Chemical purification of the new hormone has proceeded to the stage at which material is available containing 10% hormone, and 90% of nonspecific contaminants. Is it now possible to develop a radioimmunoassay which will assist with further purification, and with assessment of its physiological role in vivo? The first step would be to immunise animals with the antigen for the production of an antiserum (see ch. 4 and Table 4.1). Inevitably, the latter will contain multiple populations of antibodies directed both to the hormone and to the contaminants. Furthermore, such are the vagaries of immunisation, that the proportion of specific antibodies is unlikely to be exactly lo%, but could equally be 1% or 30%. Thus the antiserum is of no help in the development of a specific assay. The next step would be to prepare tracer, which in this case would usually involve iodination of the antigen. But again, the isotopic label is likely to distribute among all the molecules; the proportion of the labelled preparation representing specific hormone will still only be 10%. A separation procedure would be no problem, since a universal method such as a second antibody could be used, but this in no sense confers specificity. As standard, the semi-purified antigen would obviously in no way help towards a specific assay. A possible way around this would be to use, as standard, dilutions of plasma containing high levels of the hormone. But this has two drawbacks: first, unless the physiology and pathophysiology of the hormone is already well understood, there is no guide to the appropriate conditions for the collection of such a sample; second, the hormone in the sample, even if present in massive excess, can only produce a small change in the bound/free ratio of the system described. The implication of the example given above is that nonavailability of purified antigen virtually excludes the development of a practical radioimmunoassay. However, the situation is quite different if a small amount of highly purified antigen were available. In the system under discussion, the pure antigen could be used for immunisation, yielding a highly specific antiserum. This would be of little value if used with the mixer tracer, since the total bound under any cir-
Ch. 2
P U R I F I E D LIGAND
33 1
cumstances could not exceed 10%. But combined with a tracer prepared from pure antigen, it would yield a specific system in which the nature of the standard is relatively unimportant. Thus, if the standard consists of semi-pure antigen, it will still yield a full doseresponse curveagainst which the results of unknowns can be compared. The results will not be quantitative in terms of weight, since the proportion of pure hormone in a partly purified preparation is often uncertain, but they can be related to some other property of the semipure antigen, such as its biological activity, or the amount in an ampoule distributed as an International Reference Preparation. The practical use of semi-pure antigen as standard, coupled with more specific assay reagents, is seen with the gonadotrophins, luteinking hormone (LH), and follicle-stimulating hormone (FSH). The model described above is hypothetical, but if the words 'human prolactin' are substituted for 'hormone' then the story would be instantly identified by workers in this particular field. Until quite recently the existence of a separate prolactin molecule in the human was seriously doubted and its physiology was, therefore, unknown. Earlier attempts at purification were vitiated by its close physicochemical similarity to growth hormone, and many supposedly good preparations proved in the end to be growth hormone with properties subtly modified during the purification procedure. Eventually small amounts of pure prolactin were produced, uncontaminated with growth hormone (Hwang et al. 1971). Used as a tracer together with antisera raised to impure growth hormone (i.e. containing antibodies to prolactin present as a contaminant in the growth hormone), this yielded the first specific human prolactin assay. The characteristics of this type of assay are shown in Fig. 2.1 and demonstrate the critical dependence on the nature of the tracer. Subsequently, specific antisera to human prolactin were produced, and now all components of the assay system (tracer, antibody and standard) are based on highly purified materials.
Suhjerr index p. 531
332
RADlOlM MUNOASSAY A N D R E L A T E D TECHNIQUES
0.001
0.01
1.0
0.1
10
100
HORMONE CONC. ( p g i m l )
B
50 r
hPL
40
030 z =3
-
0 m
e.20 10
1
'
0.001
"
'
0.01
'
0.1
'
~
1.0
'
~
10
'
l
100
HORMONE CONC. ( ~ l m l )
Fig. 2.1. Standard curves for human growth hormone (hGH), human prolactin. and human placental lactogen (hPL) using an antiserum t o a partially purified preparation of human prolactin. A : tracer is [1251]hGH; B: tracer is [1251]prolactin. Note that the specificity of this assay is almost entirely dictated by the nature of the tracer. With appropriate choice of the latter, a relatively non-specific antibody can yield a highly specific assay. (From data kindly supplied by Dr. A.S. McNeilly.)
Ch. 2
PURIFIED 1.IGAND
333
2.3. Availability ojpure ligand It is impossible to discuss here all the ligands currently in use in binding assays. Instead, examples will be considered which illustrate some of the problems which may arise. A single generalisation may, however, be made: that ligands prepared synthetically (e.g. steroids, small peptide hormones, drugs) present relatively few problems, while ligands which have to be prepared from natural sources (e.g. large protein hormones, carcinofetal antigens) can present considerable problems. As with all generalisations, exceptions will be encountered. 2.3.1. Large protein hormones The first radioimmunoassays, for insulin, suffered problems due to the difficulty in obtaining adequate supplies of pure hormone of human origin. These problems no longer exist, since supplies of both synthetic and highly purified natural material are available. Whether or not the materials used are optimal with respect to the circulating forms of insulin, or its biological activity, will be discussed in later sections. Today, the most important problemsarise with thevarious hormones of the anterior pituitary gland. This is particularly the case with the glycoproteins, LH, FSH, and TSH (thyroid-stimulating hormone). Preparations of these are available from international agencies but their purity, as judged by biological potency per unit weight, is invariably less than that of the most highly purified preparations reported by individual workers. Furthermore, some of the most widely used preparations of the gonadotrophins are derived from post-menopausal urine ; from the immunochemical point of view they differ significantly from the corresponding hormone in both the pituitary and the circulation. In attempts to overcome this problem standards are now available, though not yet fully agreed, which consist of either purified pituitary material (e.g. 1st International Reference Preparation of Human Pituitary LH for Immunoassay, MRC 68/40), or of aliquots of a plasma pool collected from postmenopausal women (MRC 69/175). Suh,e'l rii[Ct p 531
334
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
Details of procedures for the purification of the larger proteins can be found in other volumes in this series. Such procedures will not concern the majority of those who work with binding assays because they usually demand facilities and skilled personnel available at relatively few centres. However, it is worthwhile to state a general principle: the success of a purification procedure is related both to the amount of starting material available, and to the abundance of the desired substance in the starting material. This follows from the fact that, in any sequence of purification procedures (such as gel filtration, ion exchange chromatography, electrophoresis) approximately 30-60% of the desired material may be lost at each step. For a 3-step sequence this might imply a final recovery of around lo%, a serious drawback if the starting material contained 4 mg of hormone, but no problem if it contained 5 kg. 2.3.2. Carcinofetal antigens The best known of these are a-fetoprotein (A.FP) and carcinoembryonic antigen (CEA). Highly purified AFP is now fairly easily obtained; it is abundant in the tissue of origin (fetal serum, fetal liver), and fetal material is, in many countries, readily available. Although CEA is also abundant in the tissue of origin (colorectal carcinomas), the purified materials are less satisfactory because they exhibit the heterogeneity which seems to be characteristic of all glycoproteins prepared from natural sources. This heterogeneity, and the resulting possibility of cross-reactions with normally occurring antigens, may explain why the measurement of circulating CEA as an index of malignancy is less specific than was originally thought to be the case. 2.3.3. Steroid hormones and drugs Since the majority of these compounds are prepared chemically, material ofvirtually 100%purity is usually available in good quantities from commercial sources. However, purity should not always be assumed, and it is incumbent on any laboratory setting up a new assay to check their basic materials - for example, by thin-layer or
Ch. 2
PURIFIED LIGAND
335
gas-liquid chromatography, and mass spectroscopy. A notable exception to the usual purity of this type of material is seen with the antibiotic gentamicin. It is clinically important to know the circulating levels of this antibiotic during treatment, yet the basic material contains several components. This fact calls for critical evaluation of a new binding assay in the light of clinical findings. 2.3.4. Small peptide hormones The most familiar of these (e.g. oxytocin, arginine vasopressin, angiotensin, LH/FSH-releasing hormone, TSH-releasing hormone) are preparedsynthetically to a high degree of purity. For the laboratory which cannot do this for itself (i.e. the majority) availability usually depends on pharmaceutical interest in the material. Thus, synthetic oxytocin, which is widely used in the induction of labour, is freely available in milligram quantities. By contrast, arginine-vasopressin (AVP), which is the naturally occurring antidiuretic hormone of most mammalian species, is relatively difficult to obtain. The analogue which is prepared for clinical use is lysine-vasopressin (LVP) ; though readily available, the immunochemical properties of LVP can differ considerably from those of AVP, thus rendering it unsuitable for use in an assay for the endogenous human hormone. As with steroids and drugs, it should not be assumed that synthetic peptides are invariably 100%pure. In the course of preparation ‘error’ peptides are very likely to be produced and free amino acids to remain. The purity of the end-product depends on the care with which these are removed. Furthermore, alterations may occur with storage, in particular oxidation and the formation of dimers. It is desirable that all such preparations should be examined both physicochemically (e.g. by thin-layer chromatography) and by biological assay before they are used in any other assay system.
2.4. Dissimilarity between purified ligand and endogenous ligand In an ideal world the ligand used in an assay (as tracer and standard) should be identical with the endogenous ligand which that assay is Suhjecr index p. 531
336
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
intended to measure. It must already be apparent, from the preceding discussion, that this ideal is not always achieved. Dissimilarities may arise at either or both of two points: in the basic materials used, or in the handling of the basic materials in the assay itself. There are .many examples of such dissimilarities, including: (1) Species differences: the use of animal materials in radioimmunoassays for human hormones has been very common in the past; for instance, material of bovine origin in assays for insulin and for parathormone. (2) Strain differences: hepatitis B antigen occurs in various types; a common determinant designated ‘a’, and one from each of two pairs of mutually exclusive determinants ‘d’ and ‘y’, and ‘w’ and ‘r’ (Le Bouvier 1972). (3) Tissue differences: the exact tissue or fluid of origin can be very critical. Glucagon prepared from the gut differs immunochemically from pancreatic glucagon (Bloom 1974). The peptides released by ‘ACTH-secreting’ tumours may be very different from normal pituitary ACTH (Rees 1976). The immupochemical nature of the pituitary glycoproteins (LH, FSH and TSH) differs according to whether they are prepared from urine or pituitary extracts (see Franchimont 1971). (4) Multiple forms of the material: there are many instances when the endogenous material may occur in several different forms. Insulin is synthesised in the body as a single large precurqor molecule, ‘proinsulin’; a t the time of release this is broken down to yield the two peptide chains of authentic insulin, together with the connecting peptide chain or ‘C-peptide’ (Steiner et al. 1968). Under certain conditions pro-insulin is released into the circulation, but this will not be revealed by the majority of insulin assays. Metabolism of the endogenousmaterialmay lead to non-identity ;for example, fibrinogen cross-reacts only partly in a radioimmunoassay for the fibrinogen degradation product, FgD (Fig. 2.2) (Gordon et al. 1973). Metabolites of a hormone may give positive results in a radioimmunoassay for that hormone and thus provide a poor reflection of the biological activity (see 5 11.4). Conjugates of steroids react poorly or not a t all
Ch. 2
331
P U R I F I E D LlCiAND
60
15
1.2
320
20
5000
nghl
Fig. 2.2. Standard curves for fibrinogen and its degradation products fragments D (FgD) and E (FgE) using [12s1]FgD as tracer and an antiserum to FgD. Note that fibrinogen shows only partial cross-reaction in this assay although it must, by definition. contain the sequences which make up FgD. (From data kindly supplied by Dr. Y.B. Gordon.)
with antisera to the non-conjugated molecule and in most cases would not be detected in an assay directed to the latter.
2.5. Standards Differences between laboratories in binder, tracer and detailed technique for a given assay would almost inevitably yield a wide range of different results for the same material. In order to overcome this problem considerable efforts have been directed towards the use by all groups of a single material as standard. The preparation and distribution of common standard materials of this type have now become a function of the World Health Organisation, the principle agencies being the National Institute for Biological Standards and Control of the Medical Research Council in the United Kingdom, and the Natiodal Pituitary Agency of the National Institute of Arthritis, Metabolism and Digestive Diseases in the U.S.A. The purpose of standardisation by the use of common reference materials is to permit expression of results which are consistent S u h p I mde! p 531
338
RADlOlM MUNOASSAY A N D RELATED TECHNIQUES
between different laboratories and at different times in the same laboratory. A material intended as a standard should have certain characteristics (Bangham and Cotes 1974): (1) it should be available in quantities large enough to supply many laboratories for many years; (2) it should be stable, and the effects of storage for several months at different temperatures should be known; (3) it should not contain substances which can interfere with assays, and for this reason it must be tested in a wide variety of assay systems before it is distributed; (4) ideally it should contain highly purified substance (i.e. a single molecular species) but this is sometimes impossible for the reasons already discussed (8 2.3). The usual procedure for setting up a standard is to obtain a large quantity of the relevant material, and then to distribute identical volumes of a solution into several hundred neutral glass ampoules. These are then frozen and stored at low temperature. Alternatively, the ampoules are freeze-dried as a single batch and then sealed under nitrogen. Thisis particularly convenient because it permits distribution through ordinary postal services; by contrast, material in solution should be maintained in the frozen state until it is used. The disadvantage of freeze-drying is that it may produce considerable alterations in a molecule. As a general rule, when a widely available standard is used in an assay system the results should be expressed in terms of that standard, and not arbitrarily converted to other units. This rule is often broken. For example, the international reference preparations for human LH (e.g. MRC 68/40) are distributed in ampoules whose contents are defined in units; this is necessary because the contained material has not been fully characterised by physicochemical means. It is not uncommon for workers to convert results obtained using this standard to an absolute estimate of weight, using a conversion factor based on the biological activity per unit weight of highly purified LH. There are two drawbacks to this procedure: first, the biological activity of the standard is disputed, since different assay systems give different results; second, the biological activity of available preparations of highly purified LH is also disputed. Thus, conversion to weight gives
Ch. 2
PURIFIED LlCiAND
339
a spurious impression of accuracy which is not justified by the basic facts. The opposite situation is seen with small peptide hormones, such as oxytocin and vasopressin. Almost invariably, radioimmunoassays for these hormones have been developed and applied using commercial pure synthetic material. Very reasonably, these are always defined by weight, but it is common for authors to convert the results to units of biological activity. The purpose of this is to make newer findings easily comparable with earlier studies based on biological assay. Though not strictly logical, use of this procedure is mitigated by the fact that the conversion factor from weight to biological units is not in dispute. Many laboratories conducting binding assays on a large scale use their own materials as working standards, and calibrate these from time to time against the appropriate international standards. This is also a universal practice in the manufacture of radioimmunoassay kits, since materials from the World Health Organisation are not available for commercial distribution. No exception can be taken to the use of working standards, provided that calibration is adequately carried out, and that it is always appreciated that the material is not necessarily identical with the international standard. It may, for example, be inferior in terms of stability, a fact which can pass unrecognised unless stringent tests are carried out. On the other hand, it is quite possible for a working standard to be superior to the international standard, especially in terms of physicochemical homogeneity. Thus, it would be impossible to explore the specificity of radioimmunoassay for oxytocin and vasopressin using the international standard for those hormones which contain a mixture of the two. Problems may also arise when a working standard yields a doseresponse curve which is non-parallel to the international standard, thus making quantitative comparison difficult or impossible. This is the situation encountered when assays for circulating LH based on highly purified pituitary LH are compared with an international standard (the 2nd IRP/HMG) consisting of hormone extracted from urine. Some recommendations for the preparation of standards for use in S u b p i rndei p 531
340
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
a radioimmunoassay are shown in Table 2.1, and for the preparation of hormone-free serum as a solvent for standards in Table 2.2. The latter is essential in cases such as thyroxine, in which a natural
TABLE 2.1 Preparation of standards The preparation of standards is probably the single most important step in ensuring the on-going quality of results in a binding assay (see also 9: 10.2.1.4). It should invariably be carried out by skilled and reliable personnel, using the best available balances and volumetric glassware. Standards prepared by serial dilution are to be deprecated in the context of a clinically important assay. Rather, they should be prepared as individual dilutions and a scheme for this is set out below, taking human placental lactogen as an example: (1) Weigh out 10 mg highly purified hPL (ICN Pharmaceuticals). (2) Dissolve in 10 ml hPL-free serum (from a non-pregnant human or an animal species). (3) Dispense as lots of 0.5 ml and store deep-frozen in closed tubes. (4) Thaw and mix thoroughly; dilute an aliquot 1 : 10 (0.4 ml + 3.6 ml hPL-free serum) to give a solution of 100 pg/ml. (5) Set out 6 tubes each containing 10 ml of hPL-free serum. ( 6 ) Remove an appropriate volume from each tube and replace it with the solution from (4) above. Tube No.
Volume exchanged (mu
Standard concentration Olg/ml)
1.2 0.8 0.6 0.4 0.2 0.1 Close tubes and freeze assays. Store frozen
(9: 2.6) as aliquots (0.2-0.4 ml) each sufficient for
1 week of
Nore: (I) a relatively large amount of standard is weighed out, in order to ensure maximum accuracy and precision assuming use of a microgram balance; (2) aliquots
Ch. 2
PURIFIED L I G A N D
341
of the master standard are prepared sufficient for at least 1 year's work; (3) each dilution of standard is prepared separately in order to avoid cumulative errors. It is difficult to make general recommendations with respect to the 'spacing' of standards which is often a technically convenient mathematical progression but may be unrelated to the logic of the assay. However. 3 general comments can be made: ( I ) standards should not be chosen in such a manner that the experimental difference between successive points is very small (for most common assays the difference should be a t least 4-5YJ; (2) it is good practice t o choose as one standard a concentration value which is considered l o represent the cut-off between normality and abnormality (e.g. 2.5 ng/ml in a digoxin assay); (3) the standard defined in (2) above should never lie at either extreme of the standard curve.
TABLE 2.2 Preparation of ligand-free sera for standards For many naturally occurring materials which are not species-specific a source of ligand-free serum may not be readily available. In the case of small molecules such as cortisol and thyroxine it is possible to prepare ligand-free serum by treatment with a potent adsorbent such as charcoal. All steps should be carried out in the cold. ( I ) Add 10 g of dry, unwashed Norit OL charcoal to 100 ml plasma or serum. (2) Mix overnight on a magnetic stirrer or by vertical rotation. (3) Centrifuge for I h at 10,000 x g or higher to remove the bulk of the charcoal. (4) Pass through a sequence of Millipore filters (e.g. 7, 5. 2. and 0.22 pm) to remove remaining charcoal. The efficiency of the extraction can be monitored by the addition of a small amount of labelled material (e.g. ['251]thyroxine). It is important that all the charcoal should be removed since even trace amounts can interfere with the assay.
source of hormone-free serum is not readily available. Standards are further discussed in 5 10.2.I .4.
2.6. Storage of materials This is an appropriate place to consider some aspects of reagent storage, whether of standard, binder, tracer, or samples for assay. Suhiecr index p , 531
342
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
The number of possible materials is vast ranging from those, such as aqueous solutions of drugs, which are stable almost indefinitely at room temperature, to others, such as small peptides in plasma, which may be destroyed in a matter of hours. However, certain generalisations may be made. Most biological materials are more stable at low temperature, and it is usual to store them in the frozen state. Two specific points should be made about the freezing of solutions of biological materials. First, the process of freezing can be very damaging unless carried out rapidly; in normal domestic freezers the process is slow and can easily lead to damage due to formation of large ice-crystals ; initial freezing in liquid nitrogen, oramixtureofdry ice andethanol, is much more rapid and satisfactory. The rate of initial cooling is probably more important than the eventual temperature at which the material is kept, and money spent on the purchase of ultra-low temperature equipment (below -40°C) is often wasted. The second important point about low-temperature storage is that repeated cycles of freezing and thawing are highly detrimental to the majority of biological molecules. If a given material has to be removed frequently from a deep-freeze, then it should be stored in aliquots small enough to obviate the need for re-freezing. Another aspect of storage is the chemical environment of the material. Proteins should always be kept at a concentration of 1 mg/ml or greater; for example, it is well known that antisera lose activity much more rapidly if diluted. If the material is not available at sufficiently high concentration, then theaddition of a carrier protein should be considered (Table 1.5). Some biological molecules, such as ACTH, are particularly liable to damage by oxidation, and may be better stored in the presence of a reducing agent, or in an oxygenfree atmosphere (i.e. under NJ. Other factors include the pH and ionic strength of the environment.
CHAPTER 3
Requirements for binding assays - tracer ligand
Essential to any binding assay is a means for determining the distribution between the bound and free fractions. For this purpose, a small amount of highly purified labelled ligand (the ‘tracer’) is incorporated in the system. The label may be any substance having the primary characteristic that it can be measured accurately by direct and simple methods, and that the sensitivity with which it can be detected is greater than that of direct methods for the measurement of the ligand itself. In practice the label used is almost invariably a radioactive isotope; most of the discussion in this section will be concerned with such isotopes. However, in principle, the use of a radioactive tracer is neither a defining feature of binding assays, nor of the immunoassays which constitute so important a part of the subject. Other labels, such as enzymes, fluorescent agents, luminescent agents, and phages, which meet the primary criteria described above, can equally be used, subject only to practical limitationswhich their detection may impose on the eventual sensitivity and reproducibility of the system. This chapter will begin with a short account of radioactive isotopes and their detection, and continue with a description of the characteristics and preparation of tracers intended for use in a binding assay.
3.1. Radioactive isotopes An atom of any element consists of a central nucleus around which rotate small particles known as electrons. The nucleus contains 2 343
Brh,rcr rndvr 1’ 531
344
RADIOIMMUNOASSAY A N D KELATED TECHNIQUES
main types of particle, protons and neutrons; these are of approximately the same mass, but the protons carry a positive electrical charge, while the neutrons are uncharged. Electrons carry a negative electrical charge equal to that of the proton, but their mass is only 1/1840 of that of a proton; for this reason, the mass of an atom is roughly the same as the total mass of protons and neutrons in the nucleus. The number of protons, or ‘atomic number’, determines the chemical characteristics of an atom and defines its position as an element in the periodic table. With many elements, the number of neutrons in the nucleus is variable, leading to differences in atomic mass but not in chemical properties. The term ‘isotope’ is applied to these variants. Hydrogen, for example, can exist in three isotopic forms; in the simplest and commonest there is a single proton in the nucleus and one electron in an orbital ring. Less common are forms in which the mass is increased by the addition of one or two neutrons, usually referred to as ‘deuterium’ and ‘tritium’ respectively. Individual isotopes are commonly referred to by the symbol for the element with the atomic mass written as a superscript, e.g. ‘H, 2H, jH ; the atomic number (number of protons) is sometimes included as a subscript. e.g. I,H, ?,H, j,H. The nucleus of some isotopes is unstable, and these are known as ‘radioactive isotopes’. Under these circumstances the nucleus will undergo spontaneous transformation to a more stable state, and in the process will emit energy in the form of either particles or nonparticulate electromagnetic vibrations such as prays, or X-rays. The following processes may be involved : (1) Expulsion of an a-particle (2 protons and 2 neutrons, equivalent to a helium nucleus). This occurs only with heavy elements. (2)Conversion of a neutron into a proton, an electron, and a neutrino (a small uncharged particle); the last two are expelled ( 8 - emission). (3) Conversion of a proton into a neutron, a positron (a positively charged electron), and a neutrino; the last two are expelled (/?+emission). (4) Electron capture, in which a nuclear proton is converted to a
Ch. 3
345
TRACER LlCiAND
neutron by the capture of an orbital electron; the process leads to emission of weak X-rays. (5) Isomeric transition, in which an unstable nucleus changes to a more stable isomer, the excess energy being emitted as prays. ( 6 ) Internal conversion of prays, leading to emission of an orbital electron. As a result of the first 3 of these processes the nucleus will have a different atomic number, but the atomic mass may be the same or lower. If the residual conformation of the nucleus is still unstable, further transformations occur until it becomes stable. The rate of disintegration (or 'decay') of a given radioactive isotope is specific to that isotope; it is described by the 'half-life', the time required for 50% of the radioisotope to decay. The unit of radioactivity, originally defined as the radioactivity of 1 g radium is the 'curie' where : 1 curie (Ci) = 3.7 x 10") disint/sec, I millicurie (mCi) = 3.7 x lo7 disint/sec, 1 microcurie @Ci)= 3.7 x lo4 disint/sec. The radioactivity of a sample of an isotope will thus depend on the amount of isotope present, and on its half-life.
3.2. Counting of' radioactive isotopes Detection and quantitation of an isotope in a radioimmunoassay depend on the use of a scintillation counter; the essential parts of this system are a scintillator, a photomultiplier tube and electronic circuits (Fig. 3.1). A scintillator is a material which emits a light flash when struck by ionising radiation; the intensity of the flash varies with the energy of the radiation. The light flashes are detected by a ion is i ng radiation,
~1
scintillator
photons
n ~1 electric
cF2
pulser
photomultiplier tube
electronic circuits
Fig. 3.1. The principle of scintillation counting. Suhjecr index p. 531
346
RADIOIMMUNOASSAY AND RELATED TECHNIQUES
photomultiplier, which converts them into electrical pulses. The amplitude of these pulses is proportional to the intensity of the scintillation, and thus to the energy of the radiation which produced it. For any given isotope the radiation shows a continuous distribution of energies, or spectrum, with a maximum which is characteristic of that isotope. By means of pulse-height analysis a scintillation counter can be set to detect pulses within a narrow range of amplitude, and thus to count an isotope with minimal interference from other isotopes or background radiation (Fig. 3.2). The type of scintillator used depends on the nature of the radiation emitted, which in practical terms means either P-particles (from isotopes such as .'H or I4C) or prays (from isotopes such as "I1 or IzSI). Beta-particles have very low penetrating power in matter, and can only be detected when the isotope is in intimate contact with the scintillator. For this purpose, a solution is made in an aromatic solvent of the isotope together with an aromatic compound which has the property of fluorescing (i.e. emitting light) when excited by ionising radiation. The most widely used compound is 2,5-diphenyloxazole (PPO). The efficiency of a liquid scintillator can be much reduced by the presence of impurities such as proteins; this is the phenomenon known as 'quenching', and corrections may be necessary
X
Y
-
particle energy (pulse amplitude) Fig. 3.2. The energy spectra of 2 radioactive isotopes, A and B. By setting a 'window' on the counter which sees only energies lying between x and y. most of the radiation from isotope A will be counted, but little of that from isotope B.
Ch. 3
341
TRACER LlCiAND
if the solutions counted differ widely in composition (e.g. standards prepared in simple buffer while the unknowns are samples of whole plasma). Liquid scintillation counting is discussed in detail by Fox in Vol. 5 of this series. In contrast to P-particles, y-rays have high penetrating power. Intimate contact between the isotope and the scintillator is therefore unnecessary, and the scintillator consists of a crystal of sodium iodide coated with thallium, usually formed as a well; as the radiation strikes the molecules making up the crystal lattice, ionisation occurs and results in a light flash which is then detected by the photomultiplier. Because some y-rays may be partially quenched by glass, the use of plastic tubes is always recommended. Gamma-counting is much simpler and cheaper than liquid scintillation counting. No sample preparation is required, giving considerable savings in time and materials. The specific activities of y-emitting isotopes are much greater than those of P-emitters, so that counting times are reduced or alternatively, smaller amounts of tracer can be counted in the same time. These characteristics are summarised in Table 3.1.
3.3. Choice of a counter A wide variety of equipment is available for both liquid and crystal scintillation counting. For a given technical specification there is TABLE 3.1 Comparison of the Characteristics of a commonly used y-emitting isotope that of p-emitters (I4C and 'H) Isotope
Half-life
Atoms/Ci
Detection efficiency
(%I 1251
14c
3H
60 days 5730 years 12.26 years
2.77 9.65 2.08
x
x
1017
80
lo2' loiy
85
55
(I2q
with
No. of atoms equivalent to 1 detectable atom of 1251 1 37,010 51.6 Subpci index p , 53/
348
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
little to choose between the products of different manufacturers, and the final decision is likely to rest with factors such as cost and availability of service; the latter is particularly critical for users who work at any distance from major cities. The following specific factors should be taken into account: (1) Liquid or crystal scintillation counter : this is determined by the intended application. The much greater convenience of y-counting means that whenever possible a y-emitting isotope is used, and p-emitters are used only when there is no alternative. Until quite recently the latter applied to most small molecules (e.g. steroid hormones) ; increasingly, however, y-labelled tracers are becoming available for all materials of biomedical interest, and the day is foreseeable when liquid scintillation counting will be little used for routine assays. (2) Manual or automatic sample changing: this is determined by a balance between requirements and finance. If the total counting time required is less than 1 hr/day (e.g. 60 tubes counted for 1 min each) then an automatic changer is unnecessary. For larger throughputs an automatic changer is essential ; although the capital cost is at least 5 times that of a manual model, the savings in technician time and the ability to operate for 24 hr a day will more than repay the extra costs. (3) Sample capacity: for an automatic system, this should be in the range 100-500. Equipment of larger capacity is cumbersome, and only of value if counting times are very short. A point of considerable practical importance is the transit time between samples (i.e. the time taken between the ending of counts on one tube and the beginning of counts on the next). For efficient operation this should be as short as possible as it can make a large difference to total counting time (e.g. 10 sec counts of 500 tubes with a transit time of 30 sec will take 5.5 hr; with a transit time of 1Osecthetotal time is less than 3 hr). (4) Size ofcrystal: most y-counters have a well-type crystal 2 or 3 in. in diameter. Although the larger crystal may be more efficient for emitters of high penetration, it is expensive and liable to yield high background values. For most purposes, and certainly for Iz5Iwhich is the most commonly used isotope, the larger crystal is unnecessary. (5) Single or multiple detectors: y-counting equipment is now available with multiple
Ch. 3
TRACER LICAND
349
detector heads. Several samples can be counted simultaneously, representing a considerable saving in time, space, and money for users with a high throughput. (6) Electronic circuits and controls: in the context of assay work, these should be as simple as possible. The great complexity of some equipment, particularly that intended for liquid scintillation counting and designed to permit simultaneous measurement of a mixture of isotopes in several different channels, is usually quite unnecessary. For most purposes a single-channel machine is perfectly adequate and cheaper. Furthermore, several recent models are 'Iz5I dedicated', in other words, set to count only this isotope. The only control in which versatility is essential is the count timer, which should be adjustable in steps of 1 sec over a range from 1 to at least 1000 sec. (7) Output: all types of counter usually incorporate a decade scaler. For automatic changers, an output printer is also necessary. More sophisticated equipment sometimes incorporates a 'mini-computer', capable of data reduction and on-line analysis of results (see 5 7.3).The cost of this seems to be greatly in excess of the use of 'off-line' equipment, since the latter can also be used for other purposes. Purchase of on-line data reduction must at present be regarded as a luxury.
3.4. Some practical aspects of' isotope counting (a) Choice of counting time: the error of a count is approximately proportional to the square root of the total number of counts accumulated; for 100 counts the error is 10, or 10%; for 10,000 counts the error is 100 or ly<. All tubes of an assay should be counted for sufficient time to yield total counts with an error not greater than that introduced by other steps in the assay procedure. For practical purposes, the minimum number of counts accumulated for any tube should be 2000, with an error of approximately 2%,. Accumulation of more than 10,000 counts is usually unnecessary, and leads to an undesirable increase in counting time. One currently available y-counter (LKB Instruments) offers the option of a microprocessor which will adjust the counting time for each tube to the Sublrcr i,rde\ p 5.li
3 50
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
optimum for the dose measured - this can lead to considerable reductions in counting times. (b) Fixed time or fixed total counts: as an alternative to the number ofcounts in a fixed time, many counters have the facility to accumulate a fixed number of counts and to present the result as a variable of time. This has the advantage that the counting error for all tubes is identical. The disadvantage is that for tubes containing low levels of activity counting times may be very prolonged, and for this practical reason the approach is rarely used. (c) Background counts and background subtraction : a well-maintained y-counter with adequate shielding around the crystal should not yield more than 100 counts/min (cpm) as background. For most practical purposes this can be ignored, and background subtraction is unnecessary. High background counts (100 cpm or more) may arise for several reasons: (1) mis-setting of electronic controls; (2) contamination of the crystal with isotope; (3) contamination of a carrier or an automatic sample changer; and (4)presence of a powerful radioactive source in the near vicinity of the crystal. Means for the correction of these are obvious. The major problem arises with contamination of a crystal ;repeated washing with detergent and ethanol may be necessary, and unless carried out with great care the thallium coating can be damaged. With liquid scintillation counting, background subtraction is widely used; the low specific activity of p-emitters imposes long counting times, and under these circumstances background counts may form a substantial proportion of the total. (d) For y-counting of precipitates, liquids, sections of paper strips, etc., any type of plastic tube which will fit the well of the crystal can be used. Counting can be performed without any further preparation of the sample. If the volume or size of the sample is such that some extends above the opening of the well then some counting efficiency may be lost. Given a choice between counting the bound fraction or counting the free fraction (e.g. between the precipitate and the supernatant with most separation procedures), it is a general rule to count the smaller of the two fractions which is usually the bound. Manufacturers of counting equipment are shown in Appendix I.
Ch. 3
TRACER LlCiAND
351
3.5. Essential characteristics oj’a tracer Since the tracer is intended simply to provide a measure of the total ligand in the bound and free phases of the system, it is obvious that its behaviour must be as nearly identical as possible with that of the unlabelled ligand. ‘Behaviour’, in this context, refers to its property of combination with the binder. Other properties may be a useful practical guide, but do not necessarily reflect what will happen in an assay system. For example, the labelling of a small peptide hormone, such as vasopressin, with an isotope of iodine will substantially alter the physicochemical properties of the peptide : the molecular weight is increased by some 12%. Yet the properties of such a tracer in a radioimmunoassay system may be virtually identical with those of the unlabelled hormone. At the same time, small variations in a peptide molecule which would be barely detectable in most systems of physicochemical analysis may dramatically alter its properties in combination with an antibody: for example, oxidation of the methionine residue in ACTH. There is only one ultimate criterion for the properties of a tracer - that it should behave similarly to the unlabelled ligand in the ussuy system. By definition the tracer must be slightly different from the pure ligand. Differences may arise, as noted above, because of the sheer presence of the label on the molecule. More important in practical terms, differences may arise because of alterations in the molecule, or the introduction of impurities, in the course of preparation. The general term for variations in the tracer which alter its binding properties is ‘damage’. Any tracer or fraction of the tracer which reacts with the binder with a lower affinity than that of the unlabelled ligand can be described as damaged, and the causes of this are discussed in 5 3.7.3.
3.6. Preparation of tracers Tracers can be divided into 2 types: those with an internal label and those with an external label. With an internal label, an existing atom Suhjr.cf index p. 531
352
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
in the ligand molecule is replaced by a radioactive isotope of that atom (e.g. I4C for 12C, 'H for 'H). In principle, though not always in practice, a tracer with an internal label should be virtually identical with the unlabelled ligand. With an external label, an atom or atoms of a radioactive isotope (e.g. '"I, Iz5I) are substituted for an existing atom on the ligand molecule; to achieve stability a covalent link must be established between label and ligand. A tracer with an external label is, by definition, not identical with the unlabelled ligand, though in practice its behaviour may not be distinguishable from the latter. Tracers with an internal label are commonly used in the case of small molecules such as steroid hormones or drugs. They are usually prepared on a commercial scale, and the techniques will not be described in detail here. Essentially there are 4 methods: (1) Neutron bombardment of the compound in an atomic pile. This is not commonly used because it is unselective (i.e. may alter many atoms of different types of isotopic forms), and is likely to disrupt the molecule. (2) Chemical synthesis. The molecule is synthesised from simpler molecules or elements, one or more of which is in the form of a radioactive isotope. This approach has been used with small peptide hormones; for example, the synthesis of vasopressin from its component amino acids, using [3H]tyrosine. (3) Biological synthesis. The molecule is synthesised by a biological system in vivo or in vitro, using isotopically labelled precursors: for example, prostaglandin F,, from tritiated arachidonic acid. (4)Exchange reactions. The best known method of this type is the Wilzbach technique. Material for labelling is distributed over the walls of a tube, then exposed to tritium PH) gas at room temperature for 1 week; the process can be enhanced by the use of microwaves. The material is separated from free tritium and carefully re-purified by crystallisation, chromatography, etc. to remove damaged products. This procedure yields tracers of high specific activity (up to 100 mCi/g). However, the labelling is not necessarily uniform, nor is it specific to any particular site in the molecule.
Ch. 3
TRACER L I G A N D
353
3.7. Iodinated tracers Tracers with an external label are commonly used in the case of larger peptide and protein molecules, and increasingly with haptens such as steroids and drugs because of the high specific activities which can be achieved (e.g. 20004000 Ci/mmol for an iodinated steroid rather than 25-100 Ci/mmol with an internal )H label). Though alternative labels have been described (see below), isotopes of iodine are almost universally employed in present day work, and their use will be described in some detail. General methods of modification of protein side-chains are described by Glazer et al. in Vol. 4 of this series. Precautions to be observed in handling these isotopes are set out in Appendix 5 . 3.7.1. Iodination methods Iodine can be substituted into the aromatic side-chain of tyrosine residues with relative ease (Fig. 3.3) yielding a stable compound which, if the iodine is in the form of a radioactive isotope such as I3'Ior IrsI, forms a highly efficient tracer. Iodine may also substitute into other aminp acids, including histidine, although the rate of the
Fig. 3.3. Substitution of Izsl on the aromatic side-chain of tyrosine in the presence of the oxidising agent, chloramine T, Subjrrl index p . 53/
354
RADIOIMMUNOASSAY AND RELATED TECHNIQUES
latter reaction is some 30-80 times less than that for tyrosine. The exact nature and location of the substitution in tyrosine varies with both the amount of iodine used and the nature of the peptide molecule. At low levels of iodine (1 atom of iodine per molecule or less) the majority of substitutions are single (i.e. mono-iodotyrosine) ; only at much higher levels is diiodotyrosine formed; this should be avoidedas it leads to instability. In a given molecule, tyrosine residues may differ widely in their accessibility: for example, insulin has 4 tyrosines, in positions 14 and 19 on the A chain, and 16 and 26 on the B chain; the majority of the iodine substitutes at position A14, some at A19, and very little at B16 or B26 (Freedlender and Cathou 1971). Many procedures have been described for iodination. All have in common the conversion of iodide (I -), which is relatively unreactive, into a more reactive species such as free iodine (I2), or positively charged iodine radicals (I+). The basic chemistry of the reaction is poorly understood, the literature on the subject is conflicting, and will not be discussed further. The following methods of iodination have been described, the main variable being the agent used to convert iodide to a more reactive form; only the first, the chloramine T technique, is in widespread use. (1) Chloramine T (Greenwood et al. 1963), originally marketed as a disinfectant, is a potent oxidising agent capable of converting iodide to a more reactive form. The procedure is simple, since all that is required is mixing of solutions of the peptide, sodium iodide (1251or "'I), and chloramine T ; the reaction is terminated by the addition of a reducing agent, sodium metabisulphite. The simplicity explains the almost universal acceptance of the technique and, as already noted, has done much to popularise the general application of radioimmunoassay. Practical aspects of the procedure are described in 9 3.7.2. (2) Iodine monochloride (McFarlane 1958): a solution of iodine monochloride (IC1) is mixed with solutions of the isotope and the peptide.
Ch. 3
TRACER LIGAND
355
(3) Chlorine and hypochlorite (Redshaw and Lynch 1974). (4) Lactoperoxidase (Marchalonis I969 ; Thorell and Johansson 1971): enzymatic iodination using lactoperoxidase in the presence of a trace of hydrogen peroxide has the advantage that the peptide is not exposed to high concentrations of a chemical oxidising agent such as chloramine T. Furthermore, a reducing agent is not required since simple dilution will stop the reaction. Alternatively, the lactoperoxidasecan be attached to a solid phase and removed by centrifugation (Karonen et al. 1975). It has been claimed that tracers prepared by this technique suffer less ‘damage’ than those prepared by the chloramine T method (Thorell and Johansson 1971; Karonen et al. 1975). and the procedure is therefore often used where tracer with minimal alteration is essential: for instance, in radioreceptor assays. However, rigorous comparison with methods using small amounts ofchloramine T has not been attempted. The disadvantage of lactoperoxidase is that the preparation of the reagents and the conditions for the reaction itself are more technically demanding than those for the chloramine T procedure. (5) Electrolysis (Rosa et al. 1964): iodide in the reaction mixture is converted to reactive forms by the passage of an electric current. (6) Iodine vaporisation: a mixture of chloramine T and isotopic sodium iodide yields gaseous iodine. The reaction is carried out in a gas-tight outer vessel in which is a smaller inner vessel containing a solution of the peptide ; the reactive iodine vapour diffuses into this solution. In a modification of this procedure (Butt 1972), chloramine T and sodium chloride are mixed in the outer vessel, while the inner vessel contains the peptide solution and the isotopic sodium iodide. Diffusion of chlorine gas leads to release of iodine and thus to iodination. Both procedures have lower yields (percentage incorporation of isotope into peptide) than conventional methods, but the quality of the resulting tracer is claimed to be superior because of the mild nature of the reaction. (7) Conjugation labelling (Bolton and Hunter 1973): in this procedure the iodine is first coupled to an appropriate carrier ‘handle’ containing a phenol or imidazole group for iodination, and an amine Suhierf indui p 531
356
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
for coupling to the ligand or its derivative (Figs. 3.4 and 3.5, and Table 3.2). This technique has several advantages: it does not expose the peptide to the chemical damage associated with conventional iodination reaction ; it can be applied to peptides which do not have tyrosine residues; the final reaction, mixing of the peptide and the iodinated ester, is very simple and it can be applied to other materials (e.g. the steroid hormones) which cannot be iodinated directly. The disadvantages are: (1) the substituted label, being considerably larger than the iodine atom, may lead to physicochemical alteration of the tracer; (2) with haptens. the label may bind well to antibody but fail to be displaced by unlabelled material; this is because the antiserum contains populations of high affinity antibodies directed both towards the bridge between hapten and carrier protein and to the linkage between the hapten and the tag (Fig. 3.6). 3.7.2. Practical aspects of‘iodination A protocol for the iodination of a ‘typical’ peptide by the chloramine T technique is shown in Table 3.3. It should be emphasised that the .I,
315
/ CH3
F
HO
/O
CHz-F-NHz H
Tyrosine methyl ester
II,#
31.,..
Fig. 3.4. lodinated compounds which can be used as a ‘handle’ for the indirect attachment of lZ5I to the ligand.
Ch. 3
351
TRACER LIOAND
OH
,
"%Na II
CHI
' '
C - CH,
H,-k-C-O-N
It * chloramine 1
I
C-CH2
0
II
CH, C - CH, Ha- C - C- 0- N I II 0 'C-CH,
0 wdinated ester
3-(Phydroxyphenyll propionic acid N.hydroxysuccmimideester
il 0
4 OH
I
c=
0
I
NH CH,
H
NH I
H,-C - C - N - (CH,
0
la-
CH
c=o
I
NH
I
NH,-
(CH,
I*-
I
CH
I c= I
0
NH
I
c . amino group of lysine or N-terminusin protein to be labelled
Fig. 3.5. Scheme of the coiljugation babelling technique of Bolton and Hunter (1973).
TABLE 3.2 General method for conjugation lahelling of proteins (from Bolton and Hunter. 1973') 1. Open a vial containing 1 mCi N-succinidimyl-3-(4-hydroxy, 5-1'251]-iodophenyl)
propionate (Radiochemical Centre, Code: No I M 861) and evaporate solvent by directing a gentle stream of nitrogen on to the surface. 2. Add protein ( 5 p g ) in 0.01 ml of 0.1 M borate buffer, pH 8.5, and agitate for 15 inin at 0 C. 3. Add 0.5 ml or 0.2 M glycine in 0. I M borate buffer, pH 8.5. Keep for 5 min at 0 C (the glycinc reacts with the unchanged ester and prevents subsequent conjugation to carrier proteins). 4. Purify the reaction mixture on a column of Sephadex G-50 or G-75, and assess tracer (see Tables 3.3, 3.4 and 3.5, and Figs. 3.10, 3.11 and 3.12). 1
ThemethodwasoriginallydescribedforhGH butislikelyto besuitable for all proteins. Subject index p. S31
358
RADIOIMMUNOASSAY A N D R E L A T E D T E C H N t Q U E S
Bridge
Hapten
Carrier protein
IMMUNOGEN
Fig. 3.6. A problem with the conjugation labelling of small molecules (haptens). The antiserum will contain populations of antibodies directed to the carrier protein (A), the hapten (C), and the bridge between the two (B). If the tracer contains the same bridge then it will be firmly bound by antibodies to the bridge, and cannot be displaced by pure hapten (i.e. standard or endogenous material).
exact conditions for a given substance may vary from those shown, but certain general principles can nevertheless be stated : (1) Choice of isotope: for iodine, this lies between the sodium salts of I”I and IZSI.In current practice, the latter is recommended for the following reasons: (a) the half-life is greater (60 days as against 8 days) so that the shelf-life of the tracer is potentially greater; (b) the counting efficiency for lZsI in a typical 2 in. well-type crystal is greater than that for l7lI;(c) the isotopic abundance (number of radioactive atoms (I3lIor IZ5I) relative to non-radioactive atoms (I2’I) is greater for current preparations of lZSI(100% as against 20%). This and the greater counting efficiency has the implication that lZSIgives an effective count rate, atom for atom, which is twice that of l3II, despite the shorter half-life of the latter; (d) the radiation emitted by lZSI is less penetrating than that of 13’I.Thus it presents less radiation hazard, and gives fewer problems as a source of external background counts, but it should always be treated as if it is a source of hard radiation and presents a health hazard. (2) Concentration of reagents: for rapid and efficient iodination,
359
Ch. 3 T A H L3.3~ Preparation of an iodinated protein (hPL) for radioimmunoassay
Dihroii hu[/cv; phosphate 0.05 M. pH 7.4 with / l o added protein. I . Dissolve purified hPL (50 pg) in 0.02 nil buffer in a small conical vial. It is convenient to prepare aliquots of this type for iodination by freezing the appropriate volume of a solution of hPL in a series of such vials. 2 . Add 2 mCi carrier-lice sodium (volume approx. 0.02 mi) (obtained from the Radiochemical Centre. Amersham. code IM S30, or similar supplier). 3. Add chloramine T (10 pg) in 0.02 ml buffer. The solution should he freshly prepared immediately before the iodination. 4. Mix thoroughly but hrietly (10-15 sec) by llicking with a finger; avoid splushirig. 5. Add sodium metahisulphite (20pg) in 0.02 ml buffer. Mix. 6 . Add 0.5 ml diluent buffer containing 2 nig/ml bovine serum albumin (BSA). 7. Transfer careftilly to a 1 x I5 cm (approx.) column of Sephadex G-75, previously washed with diluent buffer containing BSA. 8. Eluate with diluent buffer containing 2 mg/ml BSA. Collect fractions of approx. 0.5 ml into small vials. 9. Assess tracer as shown in Table 3.4. 10. Store tracer as deep-frozen aliquots in closed vials. Commcwrs.The procedure outlined h wide range of possible variations. particularly in the relative amounts of label and protein. However. certain general points should be stated: (a) thc volumes should be as small as possible in order to maintain high concentrations; (h) the amount of chloramine T should be kept to a minimum; (c) for this type of iodination. which yiclds one main protein peak and an iodide peak on gel filtration. the Sephadex column can he relatively small. Totdl running time is around 30 min and the fractions can he collected manually. In cases where the protein fraction is very heterogcncous a larger column and longer running time are necessary.
the concentration of all reagents in the reaction mixture should be as high as possible. Since the absolute amounts are determined by other factors, this means using the smallest possible volumes ; the total volume of the initial reaction mixture (isotope, peptide, chloramine T) should not usually exceed 100 pl. (3) Amount of chloramine T: since this is potentially damaging, the smallest possible amount should be used. Although traditionally 50 pg is used, it is often found that amounts of 10 or even 2 pg are equally effective and that the resulting product shows less damage Su h lc ,~mdr\ 1) i t /
360
RADIO1 M M UNOASSA Y AND
K ELATED 7'ECI-IN l O U kS
and is more stable on storage. The minimum amount can only be established by trial and error. (4)pH ofreagents: the optimal pH for iodination of tyrosine residues is 7.5; above pH 8 there is a tendency for other groups to be substituted and above pH 9 the reaction becomes highly inefficient. Since iodine isotopes are usually supplied as a solution in 0.1 N NaOH, the composition of the other reagents must be such as to buffer this to pH 7.5. ( 5 ) Mixing of reagents: one of the commonest faults leading to poor yields is inadequate mixing of reagents. This is especially the case when small volumes are used: a small drop on the wall of the tube, which is not shaken down into the reaction mixture, may represent SOY<, or more of one of the reactants. (6) Speed of mixing: for the concentrations shown in Table 3.3, the iodination reaction is virtually instantaneous. Mixing should, therefore, only continue for long enough to ensure that mixing has indeed occurred. The whole process, from addition of primary reagents until the reaction is stopped by the addition of reducingagent, should not occupy more than 20-30 sec. (7)Temperature of reaction: some workers prefer to carry out iodination with reagents cooled in ice. No consistent advantage over operation at room temperature has been demonstrated. (8) Quality of the isotope: it is traditional when an iodination has failed to attribute this to poor 'quality' of the isotope. There is no doubt that when the iodine isotopes first became widely available there was considerable variation from batch to batch. In recent years, however, the situation has improved to the point where failure of an iodination 'is virtually never due to the isotope itself; almost invariably, the problem can be traced to one or more simple mistakes in the technical procedure. 3.7.3. Iodinat ion damage 'Damage' can be simply defined as that fraction of the tracer ligand which will not react with the binder. This is a practical and operational definition, which does not take into account any of a variety of
Ch. 3
361
I IIAC'FR LICnAhD
ph ysicochemical alterations which are not reflected by the performance of the tracer in the assay. There are several possible causes of damage as defined above. (1) Alteration of the molecule by the presence of an iodine atom or atoms: in principle, it might be expected that the addition of iodine would affect the reactivity of a smaller molecule more than that of a larger molecule. This is confirmed by the observation that oestrogens, iodinated through their phenolic A ring, will no longer react with specific antisera. Under other circumstances, the label may alter only part of the molecule, affecting the specificity of the system. This is illustrated by the hypothetical situation shown in Fig. 3.7; in a radioimmunoassay for oxytocin, the antiserum contains two populations of antibodies. one directed to the N-terminus of the molecule, the other to the C-terminus. Substitution of iodine on to the tyrosine residue eliminates reaction with the C-terminally directed antibodies, but has no effect on the reaction with the N-terminal antibodies. In practical terms, and assuming that both antibody populations were of equal affinity, the sensitivity of the resulting system would be unaffected. The specificity with respect to the intact molecule would also be unchanged; however, the assay as described would measure N-terminal fragments of the hormone, and would be unable to distinguish these from the intact molecule.
+
C - terminal antibody
-,'125 *.-- I.
N .terminal antibody
A
'//I,'
Cys-T:r-Ile-Gln- Asn-Cys-Pro- Lev-Gly(NH,)
I , -I , Fig. 3.7. How iodine substitution could affect the specificity of an assay, in this case that for the nonapeptide oxytocin, using antiserum containing two populations of antibodies. directed to the C- and N terminus of the molecule respectively. lf the iodine completely alters the antigenicity of the C-terminus by substituting in the tyrosine residue. then only the N-terminal antibodies will be effective in the assay. The assay would measure the intact molecule and N-terminal fragments of the molecule, but not the damaged C-terminal fragments. Suhjecr iirdc,.v p . 531
362
RADIOIMMUNOASSAY A N D RELAI ED TECHNIQUES
Three methods are available for the comparison of labelled and unlabelled ligand. The first and simplest is comparison of iodinated material with material carrying an internal label such as 'H; the practicality of this depends on the availability of the internally labelled tracer. The second and more widely applicable method is that described by Hunter (1971): tubes are prepared containing a fixed amount of antibody and tracer; to one set are added serial dilutions of unlabelled ligand; to the other are added identical concentrations of tracer. Each corresponding pair of tubes should then contain the same total amount of ligand, either tracer alone or tracer plus unlabelled ligand. If the tracer is 'undamaged', then tubes containing identical amounts of total ligand should yield the same result in terms of percentage bound, and the curves from the serial dilutions should be superimposable. The third method is to compare the physicochemical properties of tracer and unlabelled ligand though, as noted above, this will not necessarily reflect their behaviour in combination with the binder. (2) Chemical damage: the peptide may be directly damaged by chemicals in the reaction mixture, most notably by oxidation or reduction. The latter is well illustrated by vasopressin: reduction of the disulphide link, with consequent splitting of the ring structure will, with certain antisera, destroy immunoreactivity of the molecule. Other possible sources of chemical damage are less specific: for instance, that due to unidentified impurities present in the isotope, or to an inherent instability of the peptide molecule in dilute solution in the absence of a carrier protein. (3) Internal radiation : the disintegration of an iodine atom will disrupt the peptide molecule to which it is attached. Furthermore, the radiation emitted may lead to free radical formation during its passage through the aqueous solution and thus damage other peptide molecules in the same solutio6: The likelihood of this occurrence will obviously increase with the concentration of the peptide. For practical purposes, internal radiation is not thought to be a significant cause of damage in the relatively dilute solutions in which tracers are usually stored or used. It may, however, be important during the actual
Ch. 3
363
TRACER LIGAND
process of iodination when all reagents are at high concentration. (4)‘Decay catastrophe’ (Yalow and Berson 1968): this term is used to describe the situation in which the decay of an attached iodine atom disrupts a molecule bearing another and as yet undecayed iodine atom (Fig. 3.8). The latter will then be attached to a molecular fragment which may no longer react with the binder. Nevertheless, since it is still labelled it will appear as damaged ‘tracer’ in the assay. This type of damage can only occur with molecules containing at least two isotopic iodine atoms. Both decay catastrophe and internal radiation are responsible for the familiar observation that such tracers have the shortest shelf-life in terms of damage. As emphasised elsewhere, in actual practice it is exceptional under the conditions of most assays that tracer is required of specific activity greater than that which can be achieved by a substitution level of one atom of iodine per molecule. Higher substitution levels should, therefore. be avoided whenever possible. (5) Incubation damage: this describes the situation in which the tracer ligand is progressively destroyed during the assay procedure itself because of the presence in the incubation mixture of damaging agents. Destruction of the tracer can easily lead to artefactual results. If, for example, 50% of the tracer is rendered non-reactive by damage of this type, the final result will be identical to that obtained in the presence of a concentration of unlabelled ligand sufficient to produce
decay of 1251 atom disrupts peptide molecule
4
/ 1251
’P125R \
/-
labelled fragments which remain no longer react with antibody (i.e. damage1
Fig. 3.8. The mechiiiiism of ‘decay catastrophe’.
SUh,P
.i3/
364
RADIO1 M M U N O A S S A Y A N D R E L A T E D 1 ~ E C H N I Q U E S
a 50% reduction in total binding. To the unwary, therefore, this reduction may be interpreted as an actual level of the unlabelled ligand. Examples of this situation and means for its avoidance are given in q 9.3.3. (6) Impurity damage: so-called ‘pure’ antigen used for iodination may be contaminated with irrelevant materials which, since they take up label and do not react with the binder, must be described as ‘damage’. In many cases the presence of such contaminants will already be known from physicochemical studies on the ‘pure’ antigen. However, these studies can be misleading when the methods available for the detection of the contaminants (e.g. spectrophotometry) are considerably less sensitive than their detection as radioactive tracers. This situation is shown diagrammatically in Fig. 3.9, in which the apparent difference in the chromatographic pattern of the labelled and unlabelled peptide isentirely due to the sensitivity of the detection system. Nevertheless, it would be easy to attribute the additional peaks seen with the tracer to ligand molecules which had been altered during the process of labelling.
actual amount of peptide
detection limit of spect rophotometry
detection limit of chromatographic fractions Fig. 3.9. How a protein can seem to consist of a single component when examined by an insensitive technique (spectrophotometry) and multiple components when examined after iodination by a sensitive technique (isotope counting). The extra peaks might be attributed to ‘damage‘ when i n fact they were present in the original preparation.
Ch. 3
T K A C E R LlCiAND
365
3.7.4. Purificatiori of iodinated tracer When the iodination is complete the reaction mixture contains the following: unlabelled ligand (damaged and undamaged) ; labelled ligand (damaged and undamaged) ; free iodide; salts including the oxidising and reducing agent. Since the tracer used in the assay should consist of not less than 95‘l/, undamaged labelled ligand, and since the proportion in the reaction mixture is usually much less, a purification procedure is almost always essential. Virtually all of the ph ysicochemical separation procedures which are used in protein and peptide chemistry have been applied to the purification of iodinated tracer. The criteria for a purification procedure are that it should be efficient, simple and rapid; the latter is critical because of the relative instability of the small concentration of material handled and represents a limitation which would not apply in the bulk purification of unlabelled materials. Although each ofthepossibleprocedureswill be briefly noted. it should be emphasised that the only technique commonly used is gel-filtration chromatography. (1) Dilutioii: it is customary to dilute the iodination mixture as soon as the primary reactions are complete. The volume added will depend on subsequent purification steps. and may vary from 0.5 ml prior to gel filtration to as much as 25 ml when purification is by addition ofan adsorbent. Although dilution cannot be truly described as purification, it nevertheless serves specific purposes. First it provides sufficient volume to make subsequent handling relatively easy. Second, it reduces the possibility of damage due to internal radiation or the presence of high concentrations of chemical agents. Third, it is usual to include a protein such as albumin in the diluent which serves as a carrier and obviates the instability of many proteins in very dilute solution. (2) Clicwtictrl precipitation : this might, for example, be applied to the separation of iodinated IgG from free iodide, by precipitation of the former with half a volume of saturated ammonium sulphate. In practice it is little if ever used because it can easily lead to damaged products. St,hpo t d c \ 1’ i I 1
366
R A D I O I M M U N O A S S A Y A N D R E L A T E D TECHNIQUES
(3) Dialysis: although in principle this would serve to separate most peptides from iodide, it has several disadvantages which reduce its practical value. These include the relatively long time which is necessary, the absorption of many peptides to dialysis membranes, and the fact that it will not readily separate damaged from undamaged labelled peptides. (4)Adsorption: in its simplest form this consists of the batchwise addition of an ion-exchange resin to the diluted reaction mixture, thus removing free iodide and salts. This is of value only with those materials, such as the small synthetically prepared peptide hormones, where damaged labelled peptide is relatively insignificant. For larger peptides, such as ACTH and gonadotrophins, extensive use has been made of particulate adsorbents such as cellulose, employed either batchwise or in the form of a small column. The nature of the process is ill-understood and likely to be complex. Nevertheless, under appropriateconditions undamaged tracer can be rapidly and efficiently separated from other components. The disadvantage is that the success of the procedure may vary with different batches of cellulose, and that the optimal characteristics of the material can only be established by trial and error. ( 5 ) Zon-exchange chrornarograplzy:this is a sophisticated procedure which with the use of appropriate gradients would readily separate minor variants of a peptide, including damaged and undamaged material. In the present context it is little used because it can be time-consuming and laborious. (6) Gel$ltration chromatography: this is the most widely applied technique for the purification of iodinated tracers. The detailed procedure does not differ radically from other applications (see Fischer 1971), with the exception that speed becomes an important criterion. Since the aim must be to prepare and characterise a tracer within a single working day, high flow-rates are chosen which may lead to sub-optimal resolution of different peaks, but which for all practical purposes gives sufficient separation. The scope of the technique can be illustrated by reference to an example, shown in Fig. 3.10, in which iodinated human placental lactogen (hPL) is
Ch. 3
TRACER L l t i A N D
367
14
12 10
a
c: p. 5. (X103) 6
4
2
Fig. 3.10. Chromatography of iodinated human placental lactogen (hPL) on a 60 cm x 4 cm column of Sephadex G-75 eluted with 60 mM barbitone buffer containing 0.5 mg/ml human serum. The first peak to emerge contains iodinated protein, while the second is free iodide. The first part of the protein peak is asymmetrical probably due to the presence of aggregated material. Fractions in this part of the peak showed considerably less immunoreactivity than later fractions.
fractionated on a column of Sephadex (3-75. Two major peaks are apparent, the first containing labelled protein and the second free iodide. It is also apparent that the first peak is not homogeneous: in the earliest fractions the percentage of tracer which can be bound to the antibody is less than that in later fractions. This suggests that the earlier fractions contain 'damaged' material, presumably high molecular weight polymers. Selection of fractions yielding the greatest immunoreactivity provides an efficient tracer for use in an assay. An even more striking example is shown in Fig. 3.11. Fractionation of a relatively impure preparation of human prolactin after iodination yields 3 clearly defined peaks of iodinated protein, only one of which reacts significantly with antibody. The assessment of a tracer after purification by gel filtration is summarised in Table 3.4. (7) Thin-layer cliromatograpliy :this is widely used in the purification of tracer ligands of low molecular weight, such as steroids and certain drugs (Stahl 1969). Although it could also be applied to iodinated small peptides, it has never been used for this purpose. Si,hj<~
368
KAUIOIMMUNOASSAY A N D RELATED TECHNIQUES
4
8
12
16
20
24 28 32 36 FRACTION (Srnll
40
44 48 52
Fig. 3.1 1. Chromatography of iodinated human prolactin on a 40 cm x I cm column of Sephadex G-75 eluted with 60 mM barbitone buffcr containing 0.5 mg/ml human serum albumin. The first 3 peaks represent protein. and the fourth peak free iodide. Of the 3 protein peaks only the third showed significant reaction with a specific antibody to prolactin.
(8) Electropltoresis: though widely applied to the assessment of labelled peptides (see ij 3 . 7 . 9 , electrophoresis is little used for purification. The reason is that the simple systems (paper or cellulose acetate) can only accept very limited quantities of material without showing extensive overlap of succeeding fractions. Systems such as starch-gel or polyacrylamide-gel will accept larger loads, but are relatively laborious to perform. With starch-gel electrophoresis, separation and purification of [i311]insulinon the basis of the level of iodine substitution has been demonstrated (Berson and Yalow 1966). Although of considerable theoretical interest, this procedure has not found a practical application. (9) Immurzopurijcation : since the operational definition of an undamaged tracer is that it should react with the binder, it would seem logical to take advantage of this in purification. Thus, in principle, a tracer could be purified from the initial reaction mixture by exposure to antibody, coupled to a solid-phase, followed by appropriate washes
Ch. 3
369
.TRACER I I G A N D
T A B L E3.4 Assessment of an iodinated protein for radioimmunoassay ~~~
~~~
1. Collect an aliquot (5-IOpI) from each fraction of the protein peak from a Sephadex column (Figs. 3.10. 3.1 I , 3.12). 2. Dilute each aliquot with diluent buffer containing 2 mg/ml BSA such that 0.2 ml yields approx. 10.000 counts in 10 sec. 3. Incubate each aliquot with and without an excess of antiserum, using conditions identical with those of the '0 standard and assay blank (Table 1.3 and Fig. 1.7). 4. Combine those fractions containing the most immunoreactive material. Test an aliquot of this solution in a standard curve.
The following data should be recorded: I . A graph of the counts eluted from the Scphadex column. 2. Plotted on thesame graph, the percentage of each fraction bound in antibody excess. 3 . The yield (I;,,) calculated from the elution pattern ([total counts i n protein peak/total counts eluted] x 100). 4. The concentration of tracer ligand in thc combined fractions (total amount of ligand used/volume of protein fractions; if a substantial amount of the protein peak is rejected allowance must be made for this). Additional methods of assessment are discussed in the main text. These have. as their object. the detailed assessment of a tracer during the development stage of an assay.
and elution. This approach has been little used in practice. presumably because it would demand relatively large amounts of solid-phase antibody, and thus an abundant supply of antiserum. However, a comparable approach is used in the immunoradiometric assay (Miles and Hales 1968) in which purified antibody is iodinated while actually in combination with solid-phase antigen. Repeated washing of the solid-phase will then remove all unwanted components. The preparation and purification of an iodinated peptide should not, as already goted, occupy more than a working day, since tracer will deteriorate more rapidly at room temperature than when chilled. The conditions for the storage of a tracer are similar to those for storage of solutions of standard (4 2.6). In particular, frozen tracer should be stored in aliquots such that it is never necessary to freeze and thaw the solution more than once. Radioactive iodinated materials should be kept in a freezer or refrigerator which is designated SiihIm~w / e ! p V /
370
I
specifically for that purpose and placed as far as possible from working areas and counters. Because of the possibility of progressive damage to the labelled antigen some workers repurify each stored aliquot by chromatography after it is thawed for use. This is undoubtedly effective, but very timeconsuming in a routine assay; such time would be better devoted to the production of a more stable tracer. 3.7.5. Clwmical assessment ofthe tracer A freshly prepared tracer must be carefully assessed both by chemical assay, and by its behaviour in the assay system. Chemical assessment is designed to ascertain 3 characteristics of the iodination: the yield, the specific activity of the tracer, and the absolute concentration of the tracer. ‘Yield’ is defined as the percentage incorporation of the isotope into labelled peptide ; this calculation does not usually take into account damaged or undamaged forms of the peptide. ‘Specific activity of the tracer’ is defined as the radioactivity per unit mass or mole of ligand (e.g. mCi/pg); for practical reasons this calculation again does not usually distinguish between damaged and undamaged. ‘Concentration of the tracer’ is simply mass per unit volume. Despite the importance of this parameter to assay design, its calculation is all too frequently ignored and the tracer distributed as ‘counts’ rather than concentration. In those situations in which the proportion of damaged tracer is high, appropriate corrections should be applied in calculating concentration so that the value refers only to undamaged material. All 3 parameters described are calculated from the results of a physicochemical separation of the components of the primary reaction mixture. The procedures available are essentially identical with those already described under purification (9 3.7.4) and, indeed, the yield, specific activity and concentration of tracer can often be obtained from the detailed results of the purification procedure. However, it is customary after iodination to apply an additional analytical procedure (usually electrophoresis) for comparison of the tracer before and after purification.
Ch. 3
371
I KACFK LIGAND
Simple electrophoresis on cellulose acetate strips (Table 3.5) will suffice to separate most iodinated peptides from free iodide, and in addition may also separate damaged from undamaged components. Perhaps the most familiar technique in the field of radioimmunoassay was ‘chromatoelectrophoresis’, originally introduced by Berson and his colleagues (Berson et al. 1956), but now largely superceded by other techniques. Of the non-electrophoretic techniques ‘wick chromatography’ (Orskov 1967) is rapid, simple and can be used with most iodinated peptides. An aliquot of the iodination mixture is spotted on to a TABLE3.5 Assessment of a freshly prepared tracer by cellulose acetate electrophoresis
I . Set up a paper electrophoresis tank1 and till with 0.06 M barbitone buffer, pH 8.6. 2. Prepare cellulose acetate strips (Oxoid Ltd). approx. 1.5-2.5 cm wide and 12 cm long and mark an origin with a pencil 2-3 cm from the cathodal end. 3. Soak the strips in 0.07 M barbitone buffer. p H 8.6. lightly blot them between filter papers. and position the strips across the bridge gap in the tank (this gap should be approx. 10 cm and the strips should extend I cm on to each shoulder piece). 4. Clamp the strips at each end with strip holders lined with filter paper; the latter should protrude suflXently at each end to dip into the buffer solution. 5. Apply 5-lop1 of the iodination mixture to the origin using a finely drawn Pasteur pipette or a capillary tube. 6 . Using a stabilised power pack apply a constant current of 0.4 mA/cm width of strip for 1 hr. 7. Remove and dry the strips. 8 . Divide the strip into 0.5 cm segments from 1 cin before the origin. 9. Place each segment in a tube and count in the well crystal of a y-counter. 10. Plot the results as counts versus distance from origin. There will usually be 2 peaks: that nearest the origin represents iodinated protein; that furthest from the origin represents free iodide. Assess the yield as shown in Table 3.4 and Fig. 3.12. Nolet In actual practice the Figures given above for current and time may vary according to the equipment used. This can be assessed only by trial and error. 1 A wide variety or electrophoresis apparatus is available. The ‘Kohn’ tank is very suitable and can be obtained. together with appropriate power supplies. from Shandon Scientific. Suhlccc
f d i , . p~.
SJI
312
RADIOIMMUNOASSAY A N D R E L A T E D l E C H N l Q U E S
narrow strip of paper one end of which is placed in a 10% solution of trichloroacetic acid. As this solution migrates up the strip the iodinated protein is precipitated and remains at its origin. Free iodide and small damaged products such as iodotyrosine and iodopeptides are not precipitated and thus move with the solvent front, giving a clear separation from the tracer. Theyield(percentage incorporation of isotope into labelled protein, damaged or undamaged) can be simply calculated from any procedure which separates iodinated peptide from free iodide. An example is shown in Fig. 3.12. The specific activity can be calculated as a function of the yield, together with the initial amounts of isotope and unlabelled peptide. The concentration o f the tracer can be determined in several ways, of which the simplest is shown in Fig. 3.12. It can also be determined from the specific activity, providing that the count rate corresponding to a given amount of radioactivity is known ; this forms a useful check on the simpler procedure.
C.P.O. per
ml eluate
20
40
60
80
100
ml eluate
Fig. 3.12. Theoretical example to illustrate calculation of yield after chromatography on an iodinated protcin. There are 20 ‘counts’ in the protein peak and 5 in the iodide peak. Thus the yield is 100 x 20/25 or 80%. The protein peak has a volume of 40 ml; assuming that the amount of protein iodinated was 10 jig, the concentration of the pooled tracer is 0.25 jig/ml.
Ch. 3
373
TKACER ILIGAND
Although the calculation ofyield, specificactivity, and concentration of tracer is relatively simple, certain provisos should be noted. Considerable losses of radioactivity may occur at different steps of the procedure: for example, by absorption on to surfaces such as those of the pipettes, the reaction vessel, and the matrix of any material used for purification. These losses may approach SOl;l,. To some extent they can be corrected by counting the residual radioactivity of the containers, etc. after use. However, with the exception of the pipette used for delivery of isotope, the nature of the retained radioactivity cannot easily be specified, and might represent proportions of free iodide and labelled peptide which differ widely from those present in the final preparation. This may explain why calculation of the yield from the results of two different procedures (e.g. gel filtration and cellulose acetate electrophoresis) can give discrepant answers. A further problem may arise with procedures which separate damaged and undamaged tracer: calculation of specific activity and concentration of the undamaged form is based on the assumption that the level of iodine substitution is equivalent for both forms. This assumption may be quite false: a small peak of radioactivity might represent a very large amount of protein with relatively few iodine atoms per molecule. The final calculation which can be of great value in the initial setting-up and design of an assay is to estimate the number of iodine atoms per molecule of ligand. This can be worked out from the specific activity expressed in molar terms, where the specific activity of lZsI at 100% isotopic abundance is approximately 1.8 mCi/nmol. For example, assume that 20 pg of a peptide hormone of molecular weight 20,000, such as hPL, hGH, or prolactin, is iodinated with 2 mCi of IZsI,giving a yield of 90% (90% of 2 mCi is 1.8 mCi, equivalent to I nmol of iodine). Twenty microgrammes of the hormone is also equivalent to 1 nmol. The level of substitution is therefore one atom of iodine per molecule of hormone. A calculation of this type can be of great practical value. All other things being equal, substitution of one atom per molecule can be regarded as the optimum; higher levels may both alter the immunoreactivity SlhlP'l I,,'/<., p
5il
374
RADiOlMMUUOASSAY A N D RELATED TECHNIQUES
of a molecule and also increase the likelihood of decay catastrophe (see Q 3.7.3); lower levels are unnecessary since the tracer will simply be diluted by unlabelled ligand molecules.
3.8. Alternative 1abels.for tracers Any material which can be accurately determined at low levels, and which can be firmly attached to the ligand molecule without grossly altering its properties, may serve for a tracer. Three examples of alternatives to isotopic labels will be noted : 3.8.1. Fluorescent labels The low limit of detection of a fluorescent tracer should permit the estimation of biological substances which circulate at relatively high levels, such as thyroxine, placental lactogen, chorionic gonadotrophin and cortisol; these determinations at present depend on the use of isotopic tracers. The advantages of a fluorescent tracer are its long shelf-life and the absence of radiation damage or hazard. A further advantage is the possibility that the physical characteristics of the fluorescent label may alter when the tracer combines with the binder, thus permitting estimation of the distribution of bound and free ligand without the need for separation of the two phases. The disadvantage of fluorescent tracers is that most biological fluids show a degree of fluorescence, thus yielding a background 'noise' level which can reduce both sensitivity and precision. 3.8.2. Enzyme labels Because of the catalytic nature of enzyme activity, a single molecule of an enzyme is responsible for the conversion of many molecules of substrate. Thus, small quantities of enzyme can be quantitated by studying substrate conversion using simple and relatively insensitive techniques such as colorimetry. Since specific enzymes can be coupled to other molecules by covalent links, the possibility exists that they might be used as labels for the production of tracer material. This possibility has now been demonstrated in practice for a wide variety
Ch. 3
.FRACEK 1.1( ;AN D
315
of materials (van Weemen and Schuurs 1971 ; Scharpe et al. 1976; Wisdom 1976).The enzymes used must possess a high specific activity at a pH that does not disturb antigen-antibody binding, and include alkaline phosphatase, P-galactosidase, glucose-6-phosphate dehydrogenase, malate dehydrogenase and peroxidase. Methods of conjugation to the ligand include the use of carbodiimides, glutaraldehyde, mixed anhydrides, and toluene-2,4-diisocyanate. The two principle types of enzymoimmunoassay are the enzymelinked immunosorbent assay (ELISA), and the homogeneous enzyme immunoassay, also known as enzyme-multiplied immunoassay (EMIT, Syva Corporation). In the ELISA system separation is achieved by the use of antibodies coupled to a solid-phase. The EMIT system dependson deactivation of the enzyme in the tracer-antibody complex; the activity can thus be measured without the need for a separation step (Rubenstein et al. 1972). In one variant of this system (assay of thyroxine using a malate dehydrogenase label) the enzyme activity is actually increased in the bound fraction (Ullman et al. 1975), and this procedure lends itself well to automation (Galen and Forman 1977). The advantages of an enzyme label are similar to those of a fluorescent label : prolonged shelf-life, avoidance of radiation damage and hazard, and, in this case, the requirement for relatively simple equipment. An additional advantage of an enzyme label is the potential for performing multiple simultaneous assays of different compounds using a different enzyme with each tracer. The disadvantages of enzyme labels are similar to those of fluorescent labels - background noise due to endogenous enzymes in a biological fluid and some loss of reproducibility due to the imprecision of the end-point determination (Sun and Spiehler 1976; Galen and Forman 1977). For this reason the best current applications of the alternative labels are those situations in which the substance measured circulates a t relatively high levels, and in which the clinical problem requires a qualitative ('yes or no') rather than a quantitative answer (e.g. hepatitis B surface antigen (Wei et al. 1977)). Substantial applications are foreseen in screening for infectious diseases (Voller et al. 1976).
376
RADIOIMMUNOASSAY AND RELATED TECHNIQUES
3.8.3. Free radical labels A molecule containing a ‘free radical’ (an unpaired electron) has a magnetic moment which can be detected by an electron spin resonance (ESR) spectrometer. An immunoassay for morphine has been described in which the tracer is morphine ‘spin labelled’ by conjugation with a nitroxide radical (Leute et al. 1972). In combination with specific antibody the free radical is immobilised, leading to a gross alteration in the spectral pattern. Bound and free tracer can thus be distinguished without requiring a separation step. This type of label has the potential advantages of other non-isotopic, non-separation systems already described. The disadvantage at present is that it is relatively insensitive an’d could only be applied to materials circulating in high concentrations. 3.8.4. Bacteriophage labels The bacteriophage is attached to the antigen and is inactivated by combination with antibody (Andrieu et al. 1975). Bound and free tracer can then be distinguished by the infectivity of the free phase.
CHAPTER 4
Requirements for a binding assay the binder -
The characteristics of a radioimmunoassay or similar technique will depend more than anything else on the properties of the binder. Other factors, such as the tracer or the method for separation of bound and free, are, in a sense, secondary: they are susceptible to manipulation and optimisation whereas the nature of the binder is usually not. Of available binders, specific antibodies are much the most widely used, with an isotopic tracer, as a radioimmunoassay. Less widely used, though still of importance, are assays based on naturally occurring circulating binders (commonly referred to as ‘competitive protein binding assays’) or cellular receptors (‘receptor assays’). Finally, there is a small group of less familiar methods which employ the same basic principle - a binder-ligand system - but are not normally thought of as binder assays. An example of such a system is the use of an enzyme as ‘binder’ for its substrate, providing a means for quantitating that substrate.
4.1. Characteristics required oj’a binder The general characteristics required of a binder can be considered under 3 headings : affinity, specificity and availability. The importance of affinity has already been noted (5 1.7) and will be further discussed (9: 8.2.2). With few exceptions, the sensitivity (minimum detection limit) of an assay depends on the affinity or K value of the binder. This follows directly from consideration of the law of mass action: for the measurement of a given con311
~ U b J C OIn&r
p. 53/
378
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
centration of unlabelled ligand the concentration of tracer and binder should be comparable or smaller. If the target concentration is low, the concentration of tracer and binder must be correspondingly low, and for any reaction to take place at these concentrations the K value must be high. Specificity is discussed in detail in Ch. 9. It is obvious that the binder should, as far as possible, be directed only to the ligand which the assay is intended to measure, and not to a wide variety of similar materials. The importance of specificity can be illustrated by an example. There are several materials (such as charcoal, Fuller’s earth, and glass beads) which can adsorb biological substances, and are used in extraction and separation procedures. The affinity of the adsorption is high: at the appropriate dilution, and with the use of a tracer, an absorbent of this type might make a highly efficient binder for an assay. That this principle has been rarely, if ever used, is due to the fact that such an assay, while sensitive, would be totally non-specific when applied to biological fluids. Specificity must be considered not only in relation to different molecules, but also to different parts of the same molecule. For example, the binder in a radioimmunoassay may recognise a different site in the molecule to that recognised by the binder in a receptor assay. This should make little difference to an estimate of the intact molecule, but substantial discrepancies can arise when, as is likely to occur in biological systems in vivo, metabolism leads to the formation of molecular fragments with different clearance rates (see Q 1 1.4). Availability of the binder is a strictly practical problem which, while it does not influence the characteristics of an assay, may nevertheless determine the choice between alternative systems of similar characteristics. Under the heading of availability may be included the difficulty of the initial preparation of the binder, the reproducibility of different preparations, the stability on storage, the cost, and balanced against these factors, the amount required in terms of the number of assays likely to be performed.
Ch. 4
T H E BINDER
3 19
4.2. An tibodies These are both the best understood and the most widely used agents in binding assays. 4.2.1. Chemistry of antibodies Antibody molecules or ‘immunoglobulins’ are found in the slower running fractions of serum proteins on electrophoresis - among the B- and y-globulins. The ‘immunoglobulins’ comprise 5 distinct classes, referred to as IgG, IgM, IgA, IgD and IgE, all of which are based on a common underlying structure (see Fig. 4.1). Each class has a characteristic spectrum of activity: for example, the IgE class contains those antibodies, or ‘reagins’, which are responsible for allergic phenomena such as asthma. The IgM class is the major immunoglobulin of several primitive species; in higher animals, the earliest phase of the immune response includes IgM antibodies which subsequently disappear, an example of ontogeny repeating phylogeny. However, from the point of view of radioimmunoassay only one class is of any significance, IgG, since the latter includes the vast majority of the antibodies which arise as the result of specific artificial immunisation. The IgG class also includes most of the familiar antibodies to bacterial and viral antigens, and thus plays an important role in the body’s defence mechanisms. IgG consists of 4 peptide chains linked by disulphide bonds (Fig. 4.1) : 2 so-called ‘heavy’ or H-chains and 2 ‘light’ or L-chains.
Fig. 4. I . The structure of the IgG molecule, consisting of 2 heavy chains (H) and 2 light chains (L) linked by disulphide bridges (S). Suhjecr index p. 5331
380
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
The molecular weight of the whole molecule is approximately 160,000. Immunoglobulins of other classes have the same basic structure but differ: (1) in the amino acid sequence of the constant portion of the H-chain; and (2) in their tendency to form polymers. IgM, for example, consists of 5 basic units linked in a ring structure. The molecular weight differences, expressed as the sedimentation coefficient in an ultracentrifuge, formed the basis of an earlier terminology for immunoglobulins which divide them into two groups: 7s (mostly IgG) and 19s (IgM). The functional activity of the different parts of the IgG molecule was originally defined on the basis of experiments in which purified IgG was subjected to limited enzyme digestion (Porter 1959). Papain splits IgG into 3 fragments (Fig. 4.2). Two of these are identical and consist of a light chain and the adjacent part of the heavy chain; this is the Fab fragment, so called because it contains the combining site of the molecule. The third fragment consists of the remaining parts of the heavy chains; this is the Fc fragment, so called because it can be crystallised. The Fc fragment, which also includes the carbohydrate moiety of the IgG molecule, is responsible for secondary activities such as the fixation of complement. IgG can also be split by pepsin, yielding two fragments: one is Fc, the other is equivalent to two Fab fragments and is referred to as F(ab),. The detailed structure of the Fab piece has been examined by studies on the sequence of myeloma proteins which, for a given
Fig. 4.2. Cleavage of the IgG molecule by the enzyme papain yielding two Fab fragments, each containing one antibody combining site, and one Fc fragment.
Ch. 4
381
T H E BINDER
tumour, are homogeneous. The N-terminal half of Fab shows great variability and is thus referred to as the ‘variable’ or V region. By contrast, the C-terminal half is relatively constant and referred to as the ‘constant’ or C region. The combining site lies in the variable region and it is this variability, with the potential existence of many millions of different structures, which is responsible for the great specificity of antibodies. It is also, in part, responsible for the great heterogeneity of an antibody population directed to a single antigen. In the course of the immune response large numbers of antibody molecules are produced with slightly different variable regions. Some fit the antigen very closely and thus have a high affinity; others with a less close fit are of lower affinity, and the resulting spectrum ranges down to values which cannot be distinguished from non-specific binding. Heterogeneity of an antibody population, which is revealed by a non-linear response in the Scatchard plot (see 5 1.5), can stem from other causes. The larger antigen molecules, such as proteins, are likely to contain several antigenic sites, each of which will be associated with a different antibody population of different characteristics. Furthermore, when a large antigen binds to one of the two combining sites on an antibody its physical bulk may partly inhibit the binding of a second antigen molecule to the remaining site. This would be revealed as an apparent reduction in affinity at high concentrations of antigen. 4.2.2. Clzemistry of antigens There would seem to be few substances, of molecular weight 300 or greater, which under the appropriate circumstances cannot stimulate an immune response with the formation of specific antibodies. Up to a molecular weight of around 1000 the antigen must be coupled to a larger molecule in order to be immunogenic; above this molecular weight most materials are immunogenic on their own. The antibody-combining site corresponds in size to approximately 3-5 amino acids. In the case of proteins it would appear that the antigenic site is usually conformational rather than sequential (Sela 1969); in other words, that it is not a single sequence of Suhlro
in&\.
p 531
382
RADlOlM MUNOASSAY A N D RELATED TECHNIQUES
amino acids which makes up the antigenic site but rather groups of one or more amino acids in different parts of the peptide chain which happen to be juxtaposed because of the tertiary structure of the molecule. 4.2.3. Cellular basis of immune response The basic cells involved in the immune response are small (‘B’) lymphocytes derived initially from the bone marrow in mammals and from the bursa of Fabricius in birds. These cells carry on their surface IgG molecules similar to those which they secrete when stimulated, a type of biological ‘free sample’. The mechanisms by which individual cells secrete molecules of a single specificity are a t present not clear. Two theories have been put forward: the first, or ‘germ-line theory’, suggests that the genes for all types of variable region are present in every cell, but that in a given cell only one gene is expressed; the second, or ‘somatic mutation theory’, suggests that the variability is generated within the individual cell by a series of gene mutations. But whatever the underlying mechanism, it now seems certain that the surface receptors determine those cells which will be stimulated, and thus the specificity of the antibody response. Lymphocytes of the B series are normally found in a wide variety of sites, including the marrow, spleen, lymph glands and blood. With the exception of some large antigens of regular structure such as polysaccharides, most substances require some sort of processing before they can efficiently stimulate the B lymphocyte. One such system is provided by macrophages which take up the antigen at the site of injection; the antigen is then released, possibly as complex with RNA, in a form which can directly stimulate the B lymphocyte. Another system may be operative in the case of ‘haptens’: these are small molecular weight materials such as steroid hormones which can bind to antibody, but cannot themselves stimulate an immune response unless linked to a larger ‘carrier’ molecule such as a protein. The carrier is recognised by and stimulates another class of lymphocytes derived from the thymus
Ch. 4
1 H E BINDER
383
(‘T’ lymphocytes) and these, in turn, release factors which enable B lymphocytes to respond to the hapten. A further factor which may stimulate the B lymphocytes is complement, and this can be activated by some antigens. The bulk of antibody produced is secreted by plasma cells. These are derived from small lymphocytes as a result of antigenic stimulation, and differ from them in having more cytoplasm and an abundant endoplasmic reticulum (i.e. they have the appropriate machinery for protein synthesis and export). 4.2.4. Physiology of immune response Classically the immune response is divided into primary and secondary phases. The primary response follows the first administration of antigen; although only small amounts of antibody are produced, it is the phase during which the small population of specifically programmed lymphocytes differentiate and proliferate to form a much larger population. The secondary response follows the second or subsequent administration of antigen; acting on the now much larger population of programmed lymphocytes, it results in the production of large quantities of antibody. The number of cells capable of responding is probably the underlying basis of the ‘memory’ which characterises immune responses in general, i.e. the greatly enhanced reaction in a second or subsequent exposure to the antigen. In practice, the distinction between the primary and secondary responses may be blurred by the nature of the immunisation process. For the production of antisera, the antigen is usually injected with an adjuvant ($ 4.2.9), one action of which is to retard the release of antigen from the injection site. If this process is sufficiently prolonged, the response will pass from the primary to the secondary phase following a single administration. At one time it was widely believed that an animal could not mount an immune response to its own antigens: the phenomenon of ‘self-recognition’. However, there are now numerous instances of successful immunisation with material which occurs naturally in the animal in question. Examples include the steroid hormones Sehjrcr i n k x p . 53/
384
RADlOlM MUNOASSAY A N D RELATED TECHNIQUES
and the posterior pituitary peptides, which have the same chemical structure in a variety of species. The presence of circulating antibody to an endogenous hormone can have striking physiological effects: thus animals immunised against vasopressin (the antidiuretic hormone) will occasionally develop diabetes insipidus with severe polyuria and polydipsia ;animals immunised against sex steroids can exhibit alterations in gonadal function (see Nieschlag 1975).
4.2.5. Characteristics o f antibodies with respect to binding assays (1) Affinity: the K value of antisera may be as high as lo'* l/mol. Taking, as a very approximate rule of thumb, the sensitivity as equivalent to 1/K, the detection range will extend to 10-l2 mol/l, or 1 pmol/l. This potential sensitivity is more than adequate for the majority of substances of biological interest, and in practice the K values of antibodies in commonly used radioiminunoassays range from loy to lo'* l/mol. (2) Specificity: antibodies are, in general, highly specific and can distinguish minor differences in closely related molecules (see 0 9.2). Related to this is another feature - wide applicability. In contrast to naturally occurring binders, antibodies can be raised to an immense range of materials including those, such as drugs, for which specific endogenous binders do not exist. The disadvantage of antibodies is that their specificity may be directed to sites on the antigen molecule which are unrelated to the biological activity of that molecule ($ 11.4). (3) Availability: a specific antibody, once produced, is usually available in quantities sufficient for very large numbers of assays. For example, a single animal such as a goat or sheep may yield one litre or more of serum containing the specific antibody operative at a dilution of 1 : 100,000; depending on incubation volume this would be sufficient for 1-2 x 108 assay tubes, ahd this could represent the world supply for many years. Furthermore, with materials which are good immunogens the production of new antiserum presents little or no difficulty. However, problems may arise with materials which are poor immunogens. The situation has occurred
Ch. 4
THE BINDER
385
in which, despite intensive effort by many laboratories, only one or two antisera are produced with affinity and specificity appropriate for use in an assay. If these antisera represent a single bleed from a small animal. supplies will be very limited and an alternative binder should be sought. 4.2.6. Production of antibodies A typical immunisation schedule, shown in Table 4.1, is based on a procedure which has been used successfully on several occasions in the author’s laboratory. The details of this schedule have been TABLE 4.1 An immunisation schedule (production of antiserum t o human placental lactogen) 1. Dissolve 0.6 mg of purified hPL in 2 ml phosphate buffer (Table 1.4) containing
no protein (use a &lasstube). 2. Add 4.5 ml of complete Freund’s adjuvant (Difco) (see (j 4.2.9). 3. Homogenise thoroughly: a convenient method for this is repeated aspiration and
4.
5. 6. 7.
8. 9.
expulsion from an all-glass syringe fitted with an all-metal needle. Plastic should be avoided as some types are attacked by components of the adjuvant. An alternative method is the use of a Potter-Elvehjem homogeniser. Inject 1 ml of the homogenate into each of 6 adult female New Zealand White rabbits. The injection should be subcutaneous and divided among 6 or more sites around the neck and shoulders. There is n o need to shave the animal for this procedure. Wait 6 weeks and repeat the procedure, but using 25 p g immunogen per animal rather than 92 pg! After a further 2 weeks take a test bleed (2 ml) and repeat the booster immunisation when this has been examined. Repeat for 4 booster injections in total, and then repeat at 1-3 month intervals according to the results and the requirements for antiserum. If the test bleed repeals a useful antiserum, a larger bleed (50 ml) should be collected prior to the next 6ooster injection. For a larger animal (sheep, goat) a similar schedule may be followed but the amount of immunogen should be increased 2-3 times.
Note: if, with an ‘easy’ immunogen, an animal has shown either no response or a poor response by the second test bleed then it should be eliminated from the series. With some haptens, however. a satisfactory response may not be seen for many months. Siiblecr rndeup. 531
386
RADIOIM MUNOASSAY A N D RELATED TECHNIQUES
chosen arbitrarily, since there must be almost as many different approaches as there are workers in the field of radioimmunoassay. For further examples see Clausen, this series, Vol. 1. In principle, this reflects the fact that the same end can be achieved by a variety of means; in practice, it is very difficult to show that any one scheme is superior to any other, because the variable nature of the antibody response would demand large groups of animals and appropriate controls. The literature is replete with claims for a particular method, usually based on anecdotal evidence which cannot be reproduced in the hands of other workers. Despite the uncertainties, there are certain specific factors which can be discussed in relation to the success or otherwise of an immunisation procedure. These are: the nature of the immunogen and its dose; the adjuvant; the animal species used; the route of immunisation ; and the timing of injection and collection of antisera.
4.2.7. The nature and dose of the immunogen In general, the immunogenicity* of a material is directly related to its molecular weight (MW). Materials of MW greater than 5000 are usually good immunogens; for example, if hPL (MW 20,000) or a-fetoprotein (AFP, MW 60-70,000) are injected with adjuvant into a group of 6 rabbits, most will respond and at least two are likely to produce antisera appropriate for use in an assay. However, 'appropriate' must be related to the actual demands of the assay; if very high sensitivity is required, and thus antisera of exceptionally high affinity, the success rate may be considerably less. By contrast, the immune response to small peptides, such as ACTH (MW 4500) and the posterior pituitary hormones (MW 1000) is often poor or non-existent, and large numbers of animals may have to be injected and tested before an efficient antiserum is found. Gaterials of molecular weight less than 800 (e.g. steroid hormones) are non-
* Immunogen must be distinguished from 'antigen'. An immunogen is material which will stimulate an immune response; an antigen is a material which will react with an antibody, but is not necessarily immunogenic on its own (e.g. most haptens).
Ch. 4
THE BINDER
387
immunogenic, but may become so when conjugated, as ‘hapten’, to a larger molecule such as albumen (see 8 4.2.8). The resulting antisera have affinity equivalent to that achieved with molecules which are immunogenic in their own right. Factors which may enhance the response to a conjugate include a high density of hapten on the carrier, and the use of a carrier which is itself immunogenic. For reasms which are not understood, immunisation with the poorly immunogenic small peptide hormones conjugated to larger proteins is not notably more successful than immunisation with the peptide alone. Adsorption of angiotensin (Boyd et al. 1967) or oxytocin (Chard et al. 1970) to carbon particles has been said to enhance the immunogenicity of these peptides, but there is no controlled evidence for this. There is no simple dose-response relationship in the immune reaction, and the amount of immunogen given is non-critical over a very wide range. Most workers use doses of 50-100 pg, though much smaller amounts may be effective. This is illustrated by the common occurrence of ‘non-specific’ antibodies to contaminants which represent 10% or less of the preparation used for immunisation. At the other end of the scale, doses of 1 mg or more may produce the phenomenon of ‘tolerance’ in the recipient animal, with no response to either this or subsequent injections of the immunogen. It is often stated that less pure material is more immunogenic than highly purified material, presumably due to an adjuvant effect of the contaminants. But the evidence for this statement is anecdotal and of doubtful validity. Although specificity of the binder is not a sine qua non of a specific assay, it should nevertheless be considered a desirable aim. As a general rule, and where sufficient quantities are available, the most highly purified material should be used for immunisation. This is particularly the case with materials prepared synthetically, since these may include contaminants closely related to the pure antigen. If the immunogen is identical or closely related to endogenous material in the recipient animal, the antibodies which result may react with the material both at the site of origin and in the cirSubiecr rndrrp 531
388
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
culation. This phenomenon can lead to biological effects which have already been described (5 4.2.3). Of more practical importance. it may also alter the apparent characteristics of the antiserum. Endogenous antigen will bind to the antibodies of highest affinity; if the dissociation constant is low, the process is essentially irreversible and when the antiserum is tested it will appear to be of relatively low affinity. In situations where this may occur, it is worthwhile to test the antiserum before and after treatment designed to split the antigenantibody complex (e.g. with acid or chaotropic agents such as urea), and to separate free antigen and antibody (e.g. by dialysis or gel filtration). The disadvantage of this procedure is that the chemical agents may themselves damage the antibody. 4.2.8. Preparation of' haptens as immunogens The carriers most widely used for preparing small molecules as immunogens are proteins such as albumen. A covalent link has to be formed between the two, usually a peptide bond between a carboxyl on the small molecule and a free amino group on the protein molecule, principally on the side-chain of lysine. But if the small molecule does not have a carboxyl or an amino group then a suitable derivative must be formed with an active group which is present. Examples of such derivatives are shown in Table 4.2. There are a variety of methods for forming the peptide link between the hapten or its derivative and protein molecule (Erlanger 1973). These include the mixed anhydride reaction and the use of carbodiimides. The latter has the drawback that it may form large numbers of cross-links between the protein molecules themselves. The number of hapten molecules which can potentially be attached to a protein molecule will depend on the number of free amino groups in the latter. For albumen the number is 60, of which at least 15-30 must join to a hapten for the complex to be effective as an immunogen. The use of thyroglobulin, which presents at least 400 sites, has been advocated though there is no concrete evidence for its superiority. The site of linkage between the hapten and the protein molecule
Ch. 4
389
THE B l N D E K
TABLE 4.2 Examples of derivatives through which a small haptenic molecule which does not possess an amino or a carboxyl group may be conjugated to amino groups of proteins (see Erlanger 1973, for principles and methods of making hapten-protein conjugates). Active group on small molecule
Carboxyl-containing derivative
Ketone Hydroxyl
carboxymethyl oxime succinate glutarate chlorocarbonate p-aminobenzoate
Phenol
is obviously of critical importance for the specificity of the resulting antiserum. The aim should be to prepare an immunogen in which the principle functional groups of the hapten are remote from the linkage site and are thus presented to the immune system of the animal in unaltered form. For example, with an oestrogen the site should not be on the highly characteristic A ring (see Fig. 4.3); the most specific antisera to oestrogens have been those raised against
Fig. 4.3. The oestradiol molecule showing the sites (arrowed) through which it can be conjugated t o a protein molecule in the preparation of an immunogen. Conjugates through the 6 position are much favoured because they leave exposed most of the characteristic sites o f the molecule. Sublccr rnderp 531
390
RADlOlM MUNOASSAY A N D RELATED TECHNIQUES
conjugates through the C6 or C7 position. Furthermore, the best results in terms of affinity are obtained if the hapten is separated from the carrier by a spacer group of at least 4-6 carbon atoms. Oestrogen derivatives suitable for direct conjugation to proteins (e.g. oestriol 6-carboxymethyloxime) are available commercially, as are ready-made conjugates (Steraloids Inc.). An example of a method for linking a hapten to a protein is shown in Table 4.3. Not all hapten-protein conjugates are equally successful as immunogens. For instance, specific antisera to prostaglandins of the A and F groups have proved relatively easy to produce, while those for the E groups have given rise to considerable problems. 4.2.9. The use of adjuvant In the vast majority of immunisation schedules the antigen is injected as an emulsion in ‘complete Freund’s adjuvant’. This is a mixture of mineral oil, detergent, and killed mycobacteria (adjuvant without the latter is referred to as ‘incomplete’). The original purpose of Freund’s adjuvant was to delay the absorption of materials which might prove toxic in large single doses (e.g. insulin). In terms of immune response, the components have the following effect. The mineral oil, being hydrophobic, tends to remain at the site of TABLE 4.3 Procedure for linking a hapten to a protein Preparation of oestriol-6-BSA. (Modified from Dean et al. 1971.) Dissolve 50 mg of oestriol-6-carboxymethyloxime(Steraloids Inc.) in 10 ml dioxan. Add 0.1 ml tri-n-butylamine. Cool the solution to 10 C, add 0.02 ml isobutyl-chloroformate and stir for 30 min. Dissolve 100 mg crystalline grade bovine serum albumen (Sigma Chemical Co.) in 10 ml distilled water adjusted to pH 9 with 2 M NaOH. 5. Mix the two solutions and stir at 4 C for 24 hr. 6. Dialyse the reaction mixture against water for 36 hr. 7. Freeze-dry the solution (conjugate) remaining in the dialysis bag.
I. 2. 3. 4.
Note: There are many different approaches to the preparation of hapten-protein conjugates. For details see Erlanger 1973.
Ch. 4
39 1
T H E BINDER
injection and thus to delay absorption; the droplets are slowly removed by the macrophages of the reticulo-endothelial system, and the contained antigen is thus progressively released by a route which will provide maximal exposure to the immune system proper. The detergent serves as an emulsifying agent, so that the aqueous solution of antigen is contained within the droplets of mineral oil (Fig. 4.4). A 'water-in-oil' emulsion is more effective than an 'oil-in-water' emulsion ; to ensure this, the proportion of adjuvant to antigen solution should always be 2 : 1 or greater (by vol), and the resulting emulsion should be tested by dispersion on a water surface. The purpose of the killed mycobacteria is to provide a general and non-specific stimulus to the reticuloendothelkal and immune systems with proliferation of macrophages and lymphocytes both locally and systemically. Some workers have employed pretreatment with killed tubercle bacilli in order to achieve this effect. 4.2.10. Tlie animat species
Consistent differences between species in their response to a given immunogen have rarely been shown. Inbred strains may occasionally show such a difference; for example, one strain of guinea pig will not respond to dinitrophenol (DNP) (Green et al. 1969). However, this is an exception rather than the rule, and the choice of animal is usually determined by a balance between the facilities available, and the likely requirements. in terms of volume, of the resulting antiserum. Other things being equal, a large species such as the goat,
'Water-in-oil' emulsion
'Oil-in-water' emulsion
Fig. 4.4.The two types of emulsion which can he formed when a solution of an immunogen is homogenised in complete Freund's adjuvant. A water-in-oil emulsion is desirable and can be achieved by using a relative excess of the adjuvant by volume. Suhjerr bide.^ p. .531
392
IlADlOlMMUNOASSAY A N D R1YLA.I'F.D 1EC'HNIQUES
the donkey and the horse would be chosen; if an efficient antiserum is obtained, it is then available in substantial quantities (1 litre or more). But many laboratories do not have access to accommodation for larger animals, whereas most will be able to keep reasonable numbers of smaller species such as rabbits or guinea pigs. It is for this reason that the hatter two species have been so widely used in the preparation of antisera for radioimmunoassdy. Indeed, with some antisera of high titre and affinity small volumes may suffice for immense numbers of assays, and quantity is not invariably a major criterion. 4.2.1 1. The route of immunisalion The immune response is systemic rather than local and may explain why the results of immunisation by different routes tend to be rather similar. Nevertheless, a literature has accumulated advocating one or the other injection site ; the various possibilities will be considered in turn: (1) Intradermal injection: for a true intradermal injection, the volume at any one site must be small (about 2 5 ~ 1 )Vaitukaitis . et al. (1971) have described a procedure in which the immunogen emulsion is injected at 40 or more sites spread. over a wide area of the body surface. This leads to a rapid and near maximal antibody response after a single immunisation, and is economical for both time and immunogen. The speed of the response can probably be attributed to the more rapid release of antigen from multiple small sites rather than a single large site. (2) Subcutaneous injection: this is the most widely used method. Approximately 1 ml of emulsion is divided among 3 or 4 subcutaneous injections in the neck. The chief disadvantage (which, indeed, is common to all routes) is that large abcesses may form and ulcerate; though unsightly, they do not seem to cause the animals undue distress. (3) Intramuscular injection: this route is rarely used. (4) Intravenous injection: this is never used for immunogen with adjuvant, since the toxic effects of the latter in the circulation will cause instant death of the animal. ( 5 ) Intraperitoneal injection: this is rarely used and would be likely to produce an acute local reaction with peritonitis. ( 6 ) Intranodal injection : direct
Ch. 4
I t l t RINDtR
393
injection of the emulsion into lymph nodes (Boyd et al. 1967), while superficially attractive, has not been shown to be superior to conventional routes (Hurn and Landon 1971). The procedure is technically exacting and there is a substantial primary mortality. (7) Footpad injection : this offers no documented advantage over other intradermal or subcutaneous injections. Because the animal may be in great discomfort, with a swollen, painful foot, the method should be discarded. 4.2.12. The timing of injections and collection of antisera Because variation between animals is so great and specific information so lacking, it is impossible to do anything else than offer simple rules of thumb for the timing of injection. The protocol shown in Table 4.1 is known to work, but this does not guarantee that shorter or longer intervals might not be equally or more effective. To explore the question would require more work than is usually devoted to the development ofa complete radioimmunoassay system; at the end, it is likely that the results would apply only to the particular immunogen studied. The optimum timing for antiserum collection is slightly more clearcut, and a 'typical' response following a booster injection is shown in Fig. 4.5. There is some argument as to whether the antibody titre declines once it has reached a peak. Experience shows this to be highly variable; in some animals it will be maintained for months or years in the absence of further booster injections; in others it shows a progressive fall. 4.2.13. Selection of antisera,for use in radioimmunoassay An immunisation programme of the type described above will yield a number of antisera from which one or more must be selected for use in an assay. The criteria for this selection are specificity, affinity and titre. Specijicity. This must be the primary criterion; unless an antiserum has the appropriate specificity it will be of no value in the assay, regardless of its affinity and titre. Subleer index p. 531
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
Animal d i d
f
R1
Days after lmmunisation
Fig. 4.5. Immunisation of 5 rabbits (RI-5) with a conjugate of oestriol (E,) to bovine serum albumen. Each booster injection is shown by an arrow, and the titre is expressed as the reciprocal of the antiserum dilution required to bind 50% of a tracer of ["]E,. Note the striking increase in titre after the first booster injection, but the very variable response thereafter (data kindly supplied by Dr. S. Khoshroo).
Affinity. Sensitivity of an assay is closely related to the affinity or K value of the antibody (4 1.7-1.10), and in general the antiserum of highest K value, as indicated by a-Scatchard plot, will be selected for use. This applies even with those assays where maximal sensitivity is not a requirement (4 8.3) since a high K value implies (though does not guarantee) a fast association constant and thus rapid attainment of equilibrium. Heterogeneity of apparent K value, as reflected by a Scatchard plot which yields a curve rather than a straight line, may also be an important factor as it will tend to reduce the slope of the standard curve (Fig. 4.6). Tirre. Given antisera of equivalent specificity and affinity, the titre (expressed as the reciprocal of the dilution which will bind 50% of a tracer) may determine the final choice. Thus an antiserum with a
1.2
0 5
50
P
1.o
.8 40 % of
tracer
.6
bound
.4
30
.2
20
1
2
3
Concentration of standard
..
$ s
2
4
.2
.4
.6
.a
1.o
1.2
Concentration of bound fraction
Fig. 4.6. (a) theoretical standard curves for a ligand and 2 different antisera to that ligand. (b) Scatchard plots derived from the standard curves of (a). Antiserum B contains a single population of antibodies with a single K value and therefore yields a straight line on the Scatchard plot. Antiserum A contains multiple populations of antibodies of different K value and yields a curve on the Scatchard plot. If the aim of the assay is to yield the best possible dynamic range over concentrations of 0-I then clearly antibody B is much the best.
396
RADIOIMMUNOASSAY AND RELATED TECHNIQUES
titre of 1 in 50,000 would be preferred to one with a titre of 1 in 20,000 because it would yield more assay tubes/ml of the raw serum. It should be emphasised that titre does not directly reflect affinity since it is also a function of the total antibody concentration. Two different antisera would yield identical dilution curves if the one contained 10 times as much antibody of one-tenth the K value of the other. 4.2.14. Storage ofaarzriseru
Serum is preferable to plasma because it is less likely to form cryoprecipitates. The serum should be divided and stored in aliquots at -20°C in order to minimise the potential damaging effects of repeated freezing and thawing. The size of each aliquot will depend both on the total volume and on the rate at which the antiserum is used ;in general, a thawed aliquot stored at 2-4"C with a preservative such as 0.1 g% sodium azide will be stable for at least 3-6 months.
4.3. Cell receptors The first step in the action of any hormone involves binding to a specific macromolecular receptor which then activates the cellular machinery for the hormonal response. In the case of steroid hormones the receptor resides in the target cell cytoplasm, while for protein hormones they lie on the cell surface membrane. Thus the use of isolated cell receptors as the binding agent in an assay has one great advantage - that it should in principle provide a direct measure of the functional site of the molecule. Other binders, by contrast, may associate with parts of the molecule remote from the functional site and may, therefore, measure inactive fragments. Cell receptors also have high affinity constants and therefore the potential for yielding a very sensitive assay (e.g. 1-10 pg for corticotrophin). It is for these reasons that the concept of receptor assays has been enthusiastically received within recent years. The first such assay to be described was that for oestrogens using the uterine cytosol receptor (Korenman 1968). Subsequently, procedures have been described for cortico-
Ch. 4
397
T H E BINDER
trophin (Lefkowitz et al. 1970), and gonadotrophins (Catt et al. 1972) employing cell-surface receptors extracted from the appropriate target organs, and for cyclic AMP (Gilman 1970) using a binding protein from skeletal muscle. That receptor assays have not, as yet, been widely applied in practice can be attributed to a number of disadvantages. First, their functional specificity may embrace molecules which are chemically very different: for example, the long-acting thyroid stimulator (LATS) can crossreact in a radioreceptor assay for TSH (Mehdi 1975). Second, a substantial proportion of cell-surface receptors may not be involved in the biological response of the cell and their functional specificity is therefore in doubt (Birnbaumer and Pohl 1973). Third, receptor assays are only applicable to those materials for which recognisable receptors exist; they cannot be used for substances such as a-fetoprotein or coagulation factors which have no function in the hormonal sense. Fourth, and much more important, receptor assays are seriously limited in their praticality., The preparation of the receptor involves homogenisation of tissue and fractionation of the extract by ultracentrifugation ; with a complex procedure of this type considerable batch-to-batch variation may occur. Furthermore, because many receptor preparations are not stable on long-term storage, new batches have to be prepared at frequent intervals. The requirements for the purity of the tracer are also much more stringent than those which apply to a radioimmunoassay. These practical considerations explain why receptor assays have not yet found application outside research units, though a radioreceptor assay for chorionic gonadotrophin (hCG) has been claimed to be suitable for routine use (Landesman and Saxena 1976).
4.4. Circuluting binding proteins The steroid and thyroid hormones are associated in the circulation with specific binding proteins : cortisol binding globulin (CBG) for the corticosteroids and progesterone ; sex hormone binding globulin (SHBG) for oestrogens and androgens; and thyroxine binding Blh)<’CI Ill,/<.\
p 531
398
KADIOIMMUNOASSAY A N D RELATED TECHNIQUES
globulin (TBG) for thyroxine and triiodothyronine. Because of the presence of these proteins the bulk of the active hormone in blood is in the form of a bound complex, leaving a small free fraction which is responsible for biological activity. The circulating binding proteins have been widely used as the binder in assays for the relevant ligand, and have the advantage that the primary material is reproducible and widely available. The first such assays to be described were those for thyroxine (Ekins 1960)andcortisol (Murphy et al. 1963). A very commonly used source of primary material is serum from women in late pregnancy which contains high levels of all the binding proteins noted above. However, this type of assay has several important disadvantages and is now almost entirely replaced by radioimmunoassay : (1) the binding proteins have relatively low affinity constants ( 107-108 l/mol) when compared with antibodies and thus cannot yield a very sensitive assay; (2) their specificity is poor - for example SHBG will bind a wide range of oestrogens and 17-hydroxy-androgens and the measurement of any one of these requires preliminary extraction and purification ; (3) the affinity constant is very temperature-dependent and the assays have to be conducted under carefully controlled conditions of low temperature; (4)because of the low affinity constant, the binder has to be used at high concentration and therefore demands the collection of large pools of material; with an antiserum, by contrast, the product of one animal may be sufficient for many millions of assays.
4.5. Radioassay .for the detection of' endogenous antibodies and circulating binding proteins Radioimmunoassays and related techniques are principally used for the estimation of ligand in samples of biological fluids. However, similar principles can be applied to the measurement of endogenous binder in the sample - for instance in the direct measurement of the naturally occurring circulating binding proteins such as sex hormonebinding globulin (SHBG) and thyroxine-binding globulin (TBG); in the measurement of antibodies in patients treated with cortico-
Ch. 4
399
T H E BINDER
trophin, pitressin, growth hormone and insulin ; in the detection of specific IgE antibodies in patients with allergic disorders; and in the quantitation of steroid receptors in neoplastic breast tissue. The approach to this type of measurement can be illustrated by examples: (1) sex hormone-binding globulin: the SHBG binding sites are saturated with [3H]5a-dihydrotestosterone (DHT). SHBG is then selectively precipitated with ammonium sulphate, and the amount of [’HlDHT bound gives a direct measurement of SHBG concentration (Rosner 1972). DHT is used as ligand in preference to testosterone because it has a higher affinity for SHBG; (2) thyroxine-binding globulin: binding sites on TBG not occupied by thyroxine (T,) are saturated with [‘25I]triiodothyronine (TI). Bound and free T, are then separated by one of a variety of methods of which the most familiar depend on adsorption of the free T, to a resin (Mitchell et al. 1960). In this case the tracer (T,) has a lower affinity for the binding protein than the endogenous ligand (TJ. For this reason the result is determined by both the total TBG and the total T, present and the test is frequently combined with a direct assay of T, to yield a ‘free thyroxine index’. The procedure is widely used in the assessment of thyroid function and goes under a variety of names including ‘thyroid hormone uptake test’ or ‘THUT’, and ‘T,-resin uptake’; (3) antibodies to growth hormone (hGH): dwarf children on longterm treatment with purified growth hormone frequently develop antibodies to this material. These can be identified and quantitated by the performance of binding studies using a tracer of [1251]hGH, serial dilutions of the patient’s serum, and separation of bound and free hGH by second antibody specific to human immunoglobulins (see 9 5.3.5.1); (4)specific IgE antibodies: allergic disorders (e.g. hay-fever, asthma) are associated with the presence of endogenous antibodies of the IgE class (see 9: 4.2.1) specific to one or more of a variety of allergens. The commonest method for identification of the latter is the use of a ‘skin-test’ in vivo, in which the patient is directly challenged with the suspected allergen. However, increasing use is being made of an ‘in vitro’ method in which the patient’s serum is reacted with an excess of solid-phase immobilised allergen which Sublet/
I I I ~ ~531 Y ~
400
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
will bind allergen-specific IgE antibodies, The solid phase is washed and reacted with labelled anti-IgE antibodies; the amount of radioactivity bound is then directly related to the amount of allergenspecific IgE in the sample. This is the so-called ‘radioallergosorbent test’ or ‘RAST’ (Wide et al, 1967); ( 5 ) oestrogen receptors in breast cancer: normal breast tissue contains oestrogen receptors (ER). Neoplastic tissue may or may not contain ER and on this finding depends the use of therapeutic endocrine manipulations (McGuire 1973). Cell cytosol from tissue extracts is incubated with tritiated oestradiol (C3H]E,) in a series of tubes containing graded amounts of nonradioactive E,. Unbound [’HIE, is adsorbed to charcoal to determine the distribution of bound and free, and the results assessed by Scatchard plot analysis to give the concentration of binder present (see 0 1.8).
CHAPTER 5
Requirements for a binding assay separation of bound and free ligand -
Once the primary binder-ligand reaction is complete it is necessary to determine the distribution of the ligand between the free and the bound form. Usually, but not always, this requires that the bound fraction be physically separated from the free fraction, and a variety of techniques have been described for this purpose. All such techniques exploit physicochemical differences between the ligand in its free and bound forms: for example, the addition of an organic solvent at a concentration which will precipitate binder molecules and thus a binder-ligand complex, but which will not precipitate the unbound 1igand . The main operative criteria for a separation procedure are efficiency and practicality. These will be considered in turn, followed by a discussion of the currently available methods.
5.1. EfJi'ciency of'separationmethods The efficiency of a procedure of this type can be defined as the completeness with which the bound and free phases are separated (Fig. 5.1). It is worthwhile to examine the meaning of this definition and to consider the relative contributions to overall efficiency made by the separation procedure on the one hand and the primary reagents on the other. In theory, a perfect separation system would completely divide the two components of the assay. In practice this is never achieved. First, it is always found that a fraction of the free ligand behaves identically with the bound, and does so even in the absence of the 401
Suh,p<~ r n C p~ 5J/
402
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
I0,ooO
a,
. ~ 2 Non - reactive m A
m;
*000 :w.* .c.w,:
tracer ligand
6000 Counts 4000
.
Tracer ligand reacting in the assay
2000
Assay blank 0
Fig. 5.1. Diagrammatic illustration of the efficiency of a separation procedure. In a perfect assay all of the tracer ligand would distribute between the bound and free phases. In practice, some of the free will be classified as bound (assay blank) and some of the free will consist of non-reactive tracer ligand. The properties shown here would vary considerably between different ligands and different separation procedures.
binder: this is the so-called ‘diluent’ or ‘assay’ blank. For example, a procedure which uses a chemical precipitation of the bound complex can invariably be shown to precipitate a proportion of the free ligand in control tubes containing no binder and this holds even when the ligand should be completely soluble under the conditions chosen. There are several possible explanations for this phenomenon : (1) physical trapping of free ligand in the bound complex, for instance in the interstices of a precipitate; this phenomenon can be quantitated by observing the effect of repeated washing of the precipitate, or following the distribution of an isotope such as 22Nawhich appears solely in the liquid phase; (2) the presence in the tracer ligand of impurities with chemical properties similar to those of the bound complex; (3) absorption of free ligand, for instance to the walls of the assay tube; and (4) incomplete separation of the 2 phases because the relevant properties of the free ligand are similar to, though not identical with, those of the bound complex (see Fig. 5.6). In practical terms it is often difficult to identify the process responsible for
Ch. 5
S E P A R A l ION OF BOIJND A N D F R E E LIGAND
403
the assay blank. Furthermore, it may vary strikingly with the materials in the incubation medium: the presence or absence of serum can make a substantial difference, and one which is highly important to control in any assay system. The second problem is incomplete separation of the bound complex. In the presence of an excess of antibody it would be expected that 160% of the tracer ligand would be bound, but this optimum is rarely achieved and figures of 70';/, or less are frequently encountered. Possible explanations of this include: (1) that a proportion of the binder and hence the binder-ligand complex behaves similarly to free ligand; for example, with a second-antibody technique in a radio-immunoassay (see Q 5.3.5.1) the precipitation may often be incomplete; (2) that impurities may be present in the tracer ligand which do not react with the binder and thus emerge in the free fraction (e.g. free iodide); (3) that the process of separation may lead to dissociation of the bound complex; for example, separation systems depending on absorption of free ligand by charcoal, leaving the bound fraction in solution, may be vitiated by the fact that the charcoal 'competes' with the binder and thus strips ligand from the bound complex (see Q 5.3.3). The reason for failure to achieve optimal separation of the bound complex is likely to vary in different systems. It is, therefore, essential when setting up a new assay that several separation procedures be examined. Of the available separation procedures, the best in terms of overall efficiency (low assay blank and completeness of separation of bound fraction) are probably solid-phase and second-antibody systems. However, both of these have disadvantages of practicality which will be discussed.
5.2. Practicality ojseparation methods This may be considered under the headings of speed, simplicity, applicability and cost, though the four are to some extent interrelated. Speed is desirable because, for many purposes, a result reported to the clinician today is more valuable than the same result Subirrr rnder p . 531
404
RADlOlM MUNOASSAY A N D RELATED TECH NlQU ES
reported tomorrow. In practice, most separation procedures take considerably less time than the incubation of the primary reaction, and therefore add little to the duration of the assay as a whole. The major exception to this is the use of second antibody in a radioimmunoassay, a step which can take 24 hr or longer. Simplicity is an important factor, because on this may well depend the total number of samples which can be processed by a single operator. Thus, if the separation step requires detailed handling of each tube - for instance, application to a chromatographic column or to an electrophoretic strip - the total number of tubes which can be handled is limited and the sample throughput correspondingly reduced. Furthermore, there is a direct relationship between the complexity of a technique and its reproducibility. Fortunately, most currently available separation procedures involve simple addition of one reagent, mixing, centrifugation and counting of the precipitate or supernatant; on the grounds of convenience the former is usually preferred because it does not involve a quantitative transfer from the assay tube. Of the common techniques the only one to involve more than this is the use of solid-phase immunosorbents which may require several wash steps if optimum results are to be achieved. Applicability - in the sense of a single procedure being applied to a wide variety of different assays - is desirable but not essential. The best examples of techniques fulfilling this criterion are second antibody and the use of binder coupled to a solid phase: these are, in principle, universal procedures which could be used for the separation of any binder-ligand system. The costs of a radioimmunoassay feside chiefly in labour and only to a relatively small extent in reagents and capital equipment. Nevertheless. the costs of separation procedures vary widely and, all other things being equal, should be taken into account when setting up an assay. The cheapest systems are those involving the addition of a simple organic solvent such as ethanol; the most expensive are those systems which are either labour intensive (chromatography, electrophoresis), or require considerable effort in the initial preparation of reagents (solid-phase systems, second antibody).
Ch. 5
SEPARATION OP B O U N D A N D FREE LIGAND
405
5.3.Methodsfor the separation of bound and.free ligand The individual methods will be classified as described by Ratcliffe (1974) (Table 5.1) and discussed in the order shown. 5.3.1. Electrophoresis Paper chromatoelectrophoresis was the first method described for separation of the reactants in a radioimmunoassay (insulin ; Yalow and Berson 1960) and evolved directly from the techniques originally used for the identification of antibodies to insulin in patients’ serum (Berson et al. 1956). For the purposes of separation of bound and free hormone the technique depends on the fact that free hormone is adsorbed to the paper at the site of application, while the antibodybound fraction migrates towards the centre of the strip; after drying, sections of the strip can be counted to determine the distribution of radioactivity between the 2 phases. However, chromatoelectro-
TABLE 5.1 Methods for the separation of bound and free ligand in a binding assay
Method Electrophoresis (starch gel, cellulose acetate, polyacrylamide gel) Gel Altration (column or batch) Adsorption (charcoal, silicates, hydroxyapatite) Fractional precipitation (ethanol, dioxan, polyethylene glycol. sodium sulphite, ammonium sulphate, trichloroacetic acid) Second-antibody precipitation (soluble and solid phase) Solid-phase antibody (particles, discs, tubes, gel entrapment, polymerised antibody)
Reference Hunter and Greenwood 1964. Haber et al. 1965. Herbert et al. 1965; Rosselin et al. 1969; Trafford et al. 1974. Grodsky and Forsham 1960; Heding 1966; Thomas and Ferrin 1968; Chard et al. 1971 ; Desbuquois and Aurbach 1971; Mitchell and Byron 1971. Utiger et al. 1962; Hales and Randle 1963; Morgan and Lazarow 1963; den Hollander and Schuurs 1971. Wide and Porath 1966; Catt and Tregear 1967; Donini and Donini 1969; Updike et al. 1973.
406
RADIOIMMUNOASSAY A N D RELATED 1ECHNIQUES
phoresis has many practical disadvantages and is rarely used in current practice. Other electrophoretic techniques, including starch gel, polyacrylamide gel and cellulose acetate, might be used, but also have practical disadvantages. 5.3.2. Gelfiltration By definition the binder-ligand complex must be larger than the ligand. In radioimmunoassay, with antibodies of MW 160,000, and antigens which in general are of MW less than 50,000, the complex is very much larger, and separation of the 2 phases can be achieved by molecular sieve chromatography on materials such as Sephadex and Biogel. Two approaches have been employed. In the first, the gel matrix is used in the form of a column: this procedure is clearly much too complex and time-consuming for routine application, even using multiple columns formed from the barrels of syringes (Giese and Nielsen 1971), and is only used as a reference method. More practicable is the second approach in which the gel is actually incorporated in the incubation medium (Fig. 5.2); low molecular weight material (free ligand) can then distribute freely both inside and outside the gel, whereas higher molecular weight material (the bound
Binder and bound complex excluded from gel particles
Free ligand inside! gel particles
Fig. 5.2. Separation by the use of porous gel. Free ligand can enter the gel and will distribute between this and the liquid phase. Bound ligand is of higher molecular weight and is excluded from the gel.
Ch. 5
407
SEPARArlON OF BOUND A N D F R E E LIGAND
complex) cannot enter the gel and thus is segregated in a small part of the system. In reality the mechanism is probably more complex than this, because many free ligands will specifically bind to the gel matrix. Separation by batch addition of a gel or similar matrix is widely used in commercial kits for the measurement of thyroid hormones. 5.3.3.Adsorption methods The non-specific adsorption of biological molecules to particle surfaces is widely used as a method for the separation of bound and free ligand; most such procedures depend on the fact that only the ligand and not the binder or bound complex have this property (Fig. 5.3). The detailed mechanism of adsorption processes is considered in the next chapter. In terms of the criteria for a separation technique, adsorption methods score high on practicality (speed, simplicity and cost), but less high on efficiency. Conditions (e.g. pH, temperature, ionic strength, protein concentration) must be carefully selected if adsorption of the bound complex is to be avoided ; furthermore, if the binder-ligand complex has a high dissociation rate the adsorbent may compete for the ligand, thus effectively splitting the bound complex (Fig. 5.4). The best known adsorption procedures are those using charcoal or silicates. Q-Binder
- ligand complex
Adsorbent particle
Free ligand
Fig. 5.3. Separation by the use of an adsorbent. Free ligand can enter 'crevices' on the particle and is firmly adsorbed. Bound ligand cannot enter the crevices and thus remains in the liquid phase. Suhjc'cr ifide.r p . 531
408
K A D I O I M MUNOASSAY A N D RELATED TECHNIQUES
Fig. 5.4. The effect of different separation procedures on the apparent titre of an antiserum, in this case two different antisera to oxytocin, with a tracer of [1251]oxytocin, and separation by the addition of ammonium sulphate (A-A, A- - -A), or charcoal (-.,. 0 . ’ .‘ 0 ) .The apparent titre is much lower with charcoal, since the latter can compete with the antibody for binding and thus ‘strip’ the bound complex.
5.3.3.1. Charcoal The use of powdered charcoal, for separation in a radioimmunoassay for insulin, was first described by Herbert et al. (1965) and has since been used in a wide variety of assays including competitive protein-binding assays. In its original form it was recommended that the charcoal be pretreated with dextran of a given molecular weight. ‘Coating’ of this type was thought to block the larger pores which might accommodate the binder while not hindering the adsorption of the smaller molecular weight free ligand. It is likely that this concept is an over-simplification, since several workers have shown that the coating is unnecessary, particularly if serum is included in the incubation mixture (Ekins 1969; Binoux and Ode11 1973). The most commonly used of the available charcoals are the Norit range (Norit SXI) with a maximum particle size of 60 pm.
Ch. 5
SEPARATION OF HOUND A N D FREE LIGAND
409
Considerable batch-to-batch variation may be found with these, and it should never be assumed that the material in bottles with an apparently identical label will behave identically in an assay. Each batch should be carefully tested before it is put into routine use, and a high assay blank or low zero standard is an indication for testing another batch. Some of the detailed procedures described in the literature demand close attention to factors such as molarity, temperature (operation at 4 C) and timing (centrifugation at a closely defined interval after addition of charcoal). For systems in which this degree of care is necessary, it is probably better to choose one of the many other separation procedures since exiguous requirements in the handling of the assay will inevitably, sooner or later, lead to errors and a loss of precision. A typical separation procedure using charcoal is shown in Table 5.2. 5.3.3.2.Silicutes Particulate silicates have adsorptive properties of which advantage is taken in numerous industrial processes. They have also been used for separation in radioimmunoassay. Materials employed include talc, Quso G-32 (Calbiochem), Fuller's earth, and T A B L5.2~ A separation procedure using dextran-coated charcoal
(Radioimmunoassay of ACTH ; Rees et al. 1971) I . Prepare a suspension containing 3.0 g of activated charcoal (Norit SXI. Hopkin and Williams Ltd). 0.75 g Dextran T-70 (Pharmacia AB), 10 ml 0.5 M phosphate buffer, p H 7.4,60 ml horse serum (Wellcome Reagents Ltd), and 100 ml deionised water. 2. Stir the suspension by a magnetic stirrer. 3. After incubation of the primary reaction mixture (see Rees et al. 1971) add 0.2 ml of the charcoal suspension to each tube, mix, and centrifuge at 2000 x g for 15 min at 4 C or room temperature. 4. Aspirate the supernatant with a Pasteur pipette attached to a simple Venturi water Pump. 5. Count the precipitate. containing the free fraction, by placing the tube in the wellcrystal of a ;'-counter. Slrbjrcr index p . 531
410
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
Florisil. The principle of their operation is very similar to that of charcoal, as are their advantages and disadvantages. If they work well in a given system they are highly practical; if they do not, or demand close attention to detail, another procedure should be tried. 5.3.3.3. Hydroxyapatite With some systems, particularly steroid radioimmunoassays, hydroxyapatite (BDH) will adsorb the bound but not the free phase, thus providing a simple and rapid technique for separation (Trafford et al. 1974). 5.3.4. Fractional precipitation Precipitation with salts or organic solvents was among the earliest methods of fractionation of biological molecules. It is now widely applied for the separation of bound and free ligand in radioimmunoassays and related systems and, where it is appropriate, offers great practical advantages. The principal is simple. When the primary binder-ligand reaction is complete, the separation material is added at a concentration in which the binder and bound complex are insoluble and therefore precipitate, while the free fraction remains in solution. The precipitate is packed by centrifugation, and radioactivity determined either in the pellet (bound fraction) or the supernatant (free fraction). The general mechanism of fractional precipitation depends on the use of salts and solvents which reduce the amount of ‘free’ water in a system: in other words, that water which is available to form a shell around a dissolved molecule and thus to keep it in solution (Fig. 5.5). For example, ammonium sulphate is capable of ‘binding’ water, which is then no longer available to form a shell for other molecules. Whether a substance will remain soluble at a given concentration of the separating agent is then determined by its own ability to attract water molecules ; for most biological materials the latter isdetermined by electrostatic charge and this, in turn by the isoelectric point and the pH of the medium. The greater the gap between these, the greater is the net charge which the molecule carries and thus its ability to form water shells. Assays are conducted at or around neutral
Ch. 5
SEPARATION OF HOUND A N D F K E E LIGAND
411
Biological molecule
Water molecules n
Fig. 5.5. The mechanism of action of precipitation of an antigen-antibody complex with ammonium sulphate. The latter takes up water molecules which are then not available to form a hydration shell for the protein.
p H ; increasing the concentration of salt or organic solvent will first precipitate those substances with an isoelectric point near neutrality, and then other materials in the order of their isoelectric points. The result is a fractionation very similar to that achieved by electrophoresis at the same pH, the slowest running molecules (with little net charge) being precipitated first. Molecules which are relatively hydrophobic (e.g. unconjugated steroid hormones, certain drugs) remain soluble even at very high concentrations of these separating agents. Antibody molecules are y-globulins which carry little charge at neutral pH and are therefore precipitated at low concentrations of reagents such as sodium sulphate (the first to be used in a radioimmunoassay system; Grodsky and Forsham 1960), ammonium sulphate, ethanol, dioxane and polyethylene glycol. Many antigens are not precipitated at these low concentrations, thus forming the basis for a simple and effective separation procedure. Although the mechanism described above holds good for most systems there are, nevertheless, discrepancies, and whether or not a given reagent will produce an efficient separation is best determined experimentally. The type of experiment required is illustrated in Fig. 5.6. It is also found with some reagents (ammonium sulphate, Suh/P
412
RADIOIMMUNOASSAY A N D RELATED .TECHNIQUES
80 70
60 50
4b BOUND 40
30 20
10
OF AMMONIUM SULPHATE ( M )
Fig. 5.6. An experiment t o demonstrate the efficiency of a fractional precipitation procedure in assays for oxytocin (solid lines), ACTH (- - - -) and insulin (-----). With oxytocin there is a substantial gap between the concentration of ammonium sulphate required to precipitate antibody-bound material and that which precipitates freeantigen (blank). With insulin, by contrast. there is almost no gap and this procedure would not be suitable for an assay.
polyethylene glycol) but not others (ethanol) that the antigen-antibody reaction will continue in the presence of the reagent, i.e. with the antibody in precipitated form. Advantage can be taken of this in designing a system in which only a single addition of reagents is required - a mixture of tracer, binder and the separating agent. In terms of efficiency; the major problem of fractional precipitation methods is that they tend to yield a high assay blank, usually in the range of 5-20%. This is often attributed to physical trapping of the tracer in the interstices of the precipitate. However, some other mechanism is almost certainly involved since the blank value does not
Ch. 5
SEPARATION OF BOUND A N D F R E E LIGAND
413
fall pari passu with the bulk of the precipitate, even when this is reduced to microscopic proportions by decreasing the amount of protein in the system (serum or carrier globulins). The incorporation of a small amount of second antibody in the solution may permit the use of a lower concentration of the separating agent, reducing the blank value without affecting the precipitation of the bound fraction. In terms of practicality the fractional precipitation methods are superior to all others. They are simple, fast and cheap. For those assays to which they can be applied (almost all, with the notable exception of those for very large protein molecules) they are highly reproducible and batch-to-batch variation is virtually nil. A typical separation procedure using polyethylene glycol is shown in Table 5.3. 5.3.5. ‘Double’antibody methods 5.3.5.1. ‘Double’ or ‘second’ antibody
Precipitation of the bound complex with an antibody directed to the binder is widely used as a separation procedure in radioimmunoassay systems and was first introduced by Utiger et al. (1962) and Morgan and Lazarow (1963). TABLE5.3 Separation of bound and free ligand using polyethylene glycol ~~~
~~
~~
I . Prepare a 20‘!$ (w/v) solution of potycthylene glycol 6000 in phosphate buffer. (Solution is speeded if a magnetic stirrer is used.) 2. Add 2 vol of this solution to the incubation mixture using a repeating syringe ( i t . 1 ml for the conditions shown in Tables 1.2 and 1.3). 3. Mix carefully on a vortex mixer to yield a homogeneous opalescent solution. There is no need to allow the suspension time to coagulate. 4. Centrifuge for 30 min at 2000 x g or greater. Temperature control is not necessary though results may be marginally improved at 4 C. 5. Decant or aspirate the supernatant. Aspiration, using a Pasteur pipette attached to a siinplc Venturi water pump, is usually preferable unless very wide base tubes are used. 6 . Count the precipitate by placing the tube in the well-crystal of a g-counter. A very similar procedure would be followed using ethanol (2 vol) or ammonium sulphate ( I vol o f a saturated solution).
414
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
The ‘second’ antibody is specific to the y-globulin of the species in which the first antibody was raised - for example, if a guinea pig anti-insulin serum is used in the primary reaction of an assay for insulin, an antiserum to guinea pig y-globulins raised in a goat might be used for the separation step (Fig. 5.7). Although most commonly used in radioimmunoassays, this concept could be applied to any binder. Precipitation reactions occur only at high concentrations of antigen and antibody; this is the reason why precipitation does not occur in most of the primary antigen-antibody reactions used in radioimmunoassay. Separation by this technique requires a relatively large concentration of second antibody and a correspondingly large amount of the species of y-globulins of which the first antibody forms a part must be included; for this purpose, a second antibody system always involves addition of carrier protein, either whole serum or y-globulins, from the species in which the first antibody was raised. The initial design of a second-antibody separation technique follows principles similar to that of any immunological procedure used as a measurement system. A number of animals are immunised with y-globulins, and their sera examined in experiments similar to that shown in Fig. 5.6 (i.e. precipitation of tracer in the presence and absence of antibody). In this case, however, it is necessary to test not only the concentration of antibody but also the appropriate concentration of carrier y-globulins, since the optimal amounts will IgG molecule of a goat antiserum to guinea. pig IgG
IgG moleculesof guinea. pig anti - insulin serum
Insulin molecules
Fig. 5.7. The principle of ‘second-antibody’ separdtion.With the addition of normal guinea pig serum to act as a carrier, a precipitate is formed which includes the bound fraction.
Ch. 5
SCPARA’TION OF BOUND A N D F R E E LIGAND
415
vary with each antiserum tested. Selection is critical, and the following factors should be investigated: ( 1 ) Completeness of precipitation of the bound complex. In the presence of an excess of the first antibody this should represent as near as possible 100% of the immunoreactive tracer. (2) The minimum quantity of ‘second’ antibody required to achieve complete precipitation. Excessive amounts are likely to be both expensive (see below) and to lead to the problem of the ‘prozone’ phenomenon - the fact that in the presence of an excess of antibody immunoprecipitation may not occur (Fig. 5.8). (3) The ‘assay blank’ - the amount of tracer precipitated by second antibody in the absence of the first antibody. This should be less than 5%. A high value can occasionally be due to the presence in the antiserum of antibodies directed to the ligand. (4) Finally, a second antibody system should be evaluated in the presence of the fluid (e.g. plasma, serum, urine) for which the assay is designed. A procedure which appears satisfactory in the presence of diluent buffers may nevertheless, in the presence of biological material, be subject to striking non-specific effects which are reflected as a reduction in precipitation of the bound complex or an increase in the assay blank or both. This is particularly likely to occur if the concentration of the sample exceeds 10% of the total volume in the assay tube. Interference in a second antibody system by plasma or serum has been attributed to the presence of high molecular weight globulins and complement (Morgan and Lazarow 1963; Morgan et al. 1974). Materials used in the preparation of the sample (e.g. heparin, EDTA) may also be involved. In terms of efficiency, a well-designed second antibody system is at least as good as and usually superior to most other separation procedures. It also has the advantage, in radioimmunoassays, of being a universal procedure. Thus, a satisfactory second antibody directed, for example, to rabbit globulins will be equally effective in all assays using a rabbit antiserum as the first antibody. However, the use of second antibody suffers from two important practical disadvantages. The first is that it requires an additional period of incubation which may range from 1 to 24 hr, and can, therefore, Suhlerr tiidea p 531
416
K A D I O I M M U N O A S S A Y AND R E L A T E D TECHNIQUES
I
1:200
1:400
I
1:800
I
1:1600
I
1:3200
I
1:6lOO
1 1:12800
ANT I SER UM 0 I LUTl ON
Fig. 5.8. The precipitation of ['25I]LH (0.1 ng) by a rabbit anti-LH serum (final concentration I :40,000) by a second antibody (goat anti-rabbit IgG). Normal rabbit serum (final concentration I :20) was added as carrier. If the concentration of second antibody is too great ( 1 : 200) little or no precipitation occurs due to the prozone phenomenon. At 1 : 400 precipitation is optimal and at lower concentrations there is a progressive decline. In this system the concentration of second antibody is therefore highly critical. (From data kindly supplied by Dr. A.S. McNeilly.)
considerably extend the time required to complete the assay. This applies with particular force to the increasingly wide range of clinical assays in which the first incubation period is 2 hr or less, and in which the practical value of the result is inversely related to the time in which it is obtained. Incorporation of the second antibody with primary reagents, to yield a pre-precipitated complex (Hales and Randle 1963), permits immediate separation but may lead to loss of sensitivity and has not been widely applied. The rate of immuno-
Ch. 5
411
SEPARATION OF ROUND A N D F R E E LIGAND
precipitation can be enhanced by incorporation of either ammonium sulphate or dextran in the second antibody (Martin and Landon 1975); this is an interesting approach which deserves further study. The second practical disadvantage is that of reagent supply. As already pointed out, a new second antibody requires careful evaluation; of those which are tested, few will turn out to be completely satisfactory. Relatively high concentrations are required, and the product of one animal is only sufficient for a limited number of assays. Second antibody systems are, therefore, expensive, regardless of whether the material is prepared locally or purchased from a commercial supplier (e.g. Wellcome Reagents Ltd.). A typical separation procedure using second antibody is shown in Table 5.4. 5.3.5.2.Double-antibody solid pliast Coupling of the second antibody to an insoluble matrix such as cellulose (den Hollander and Schuurs 1971) yields a system which is convenient and which does not require the use of carrier y-globulins. For the latter reason it is TABLF5.4 Separation of bound and free ligdnd using second antibody The following recommendations are modified from the instruction leaflet of a commercial reagent (Wellcome Reagents Anti-Rabbit Globulin. code No. R D 17). This is chosen as an example because of its wide availability and excellent properties. It can be used with any system in which thc antiserum to the ligand was raised in a rabbit.
1. Incorporate normal rabbi1 serum in the diluent buffer (0.1 m1/100 ml) to act as a carrier. 2. Following incubation of the primary reaction mixture add 0.1 ml of a dilution of the precipitating serum. and incubate for a further 18 hr at 4 C. 3. Centrifuge at 2000 x g o r greater for 3 0 min. 4. Aspirate or decarlt the supernatant (see Table 5.3). 5. Count the precipitate (Table 5.3). Notr: I n the case of commercial reagents the appropriate d-ilution is usually stated with each batch. However, each user should check the exact conditions for his own assay (see text). In many cases it is possible to reduce the second incubation t o as little as 2 hr by adjustment of conditions. Suhjwr hrd<,~ p . 531
418
RADIOIM MUNOASSAY A N D RELATED TECHNIQUES
more economical of the second antibody itself. However, the preparation and evaluation is time-consuming, and the method is not widely used except in the form of commercially available reagent (Organon). 5.3.6. Solid-phase systems If the binder is covalently coupled to an insoluble support then both it and the bound complex can readily be separated from the soluble free fraction (Fig. 5.9). A wide variety of solid-phase supports have been described, which include particles of dextran (Sephadex and Sepharose)and cellulose, and continuous surfaces such as polystyrene or polypropylene discs or the walls of plastic tubes. In terms of function, it is necessary to consider the two types separately. 5.3.6.1. Binder attached to discs and tubes This approach has only been used for antibodies. Plastic surfaces exhibit absorptive properties, and simply exposing such a surface to an appropriate dilution of an antiserum will lead to the attachment of a proportion of the antibody molecules (Catt and Tregear 1967). It is clearly unlikely that the fixation involves covalent bonds, but instead is probably due to other types of interaction (ionic, hydrophobic, etc.). The bond is, therefore, inherently less stable than the covalent links used with the particulate solid phases, and this may well explain the major drawback of disc and tube systems - lack of reproducibility, though this can be partly overcome by the incorporation of glutaral-
t
-
Free ligand
Fig. 5.9. Separation by the use of solid-phase coupled antibody
Ch. 5
SEPARATION OF ROUND A N D F K E E L I G A N D
419
dehyde and other agents in the antibody solution used for coating. Non-reproducibility may extend not only to the results in a single assay, but also to the initial choice of tube or disc which may vary considerably from batch to batch. A second drawback is that the efficiency is rather less than that of other methods, since the total amount of tracer which can be bound by antibody excess is relatively low and there may be an apparent loss of affinity. However, coated tube systems offer great convenience in the actual performance of assays and the technique is widely used in commercial kits in which the problems of reproducibility are avoided by the use of large and carefully screened batches of tubes. There is also considerable promise in new methods for covalent attachment of antibodies to tube surfaces, and in the use of specifically designed tube inserts which greatly simplify manufacturing problems. In one variant (Schwartz-Mann Ab-TRAC) the tracer is also incorporated in the tube so that only a single addition is necessary. 5.3.6.2. Binder attached to particulate solid phase Particulate solid phases are widely used in a variety of different assays. Gammaglobulins from an antiserum are attached to the particles by any one of a number of techniques designed to yield a covalent link between the protein and the particle, for example diazotisation or cyanogen bromide activation (Wide and Porath 1966). The resulting material is then extensively washed to ensure that no free y-globulin molecules remain (see Table 5.5). Interesting alternative approaches are the use of antibody entrapped in the interstices of a polyacrylamide gel (Updike et al. 1973) or covalently bound to magnetic particles (polymer-coated iron oxide) (Nye et al. 1976). With the latter system, mixing and separation can be simply achieved by the application of a magnetic field. The assay tube is set up with sample or standard, tracer, and an appropriate amount of solid-phase antibody, and a detergent (Tween-20) to prevent aggregation of the particles and non-specific adsorption of the tracer. After an incubation period during which the tubes are continuously mixed, the solid phase is sedimented by Sublecr index p 531
420
R A D I O I M M U N O A S S A Y A N D RELATED TECHNIQUES
centrifugation ; the supernatant is removed and the solid phase subjected to two or more washes with buffer in order to remove free tracer trapped within and between the particles. The counts on the solid phase (bound fraction) are then measured (Table 5.5). Solid-phase systems have several advantages: ( 1) they can be TABLE 5.5 Preparation and use of a solid-phase antibody (Radioimmunoassay of human placental lactogen) (Modified from Gardner et al. 1974) 1. Prepare a y-globulin fraction from the antiserum by adding 1.8 g of Na2S04 (anhydrous) to 10 ml of serum. mixing for 30 min, centrifuging at 2000 x g for 10 min,
2.
3.
4.
5.
6. 7. 8. 9
washing the precipitate twice with 10 ml of l8'>" (w/v) Na2S04, and redissolving in 10 mlO.9% NaCI. Swell 1 g cyanogen bromide (CNBr)-activated Sepharose 4B (Pharmacia AB) in 10 mM HCI. This should be carried out in a Buchner funnel with a grade 4 Scinta glass filter. A total of 200 ml of 10 mM HCI is added and removed in several aliquots. After a final wash with 0.1 M NaHCO, transfer the gel to a glass tube, add 50 ml of 0.1 M NaHCO, containing 1 ml of the y-globulin solution from (I), and mix by vertical rotation (25 revlrnin) for 24 hr at room temperature. Wash the immunosorbent once with 100 ml of 1.0 M ethanolamine. pH 8.0, with continuous mixing for 2 hr and then twice with 100 ml of 0.5 M NaHCO,, twice with 50 ml of 0.2 M sodium acetate buffer, pH 4.0, and twice with 100 ml of assay diluent (identical with that shown in Table 1.2, but containing in addition 0. I :{, v/v Tween-20 (Koch-Light Laboratories Ltd.), with 30 min continuous mixing for each wash. The standard curve is set up as shown in Table 1.3, but substituting a suspension of the solid phase for the soluble antibody. The appropriate dilution of the solid phase will vary between different antisera and must be established by experiment with different dilutions. Cap all tubes and mix by vertical rotation (approx. 25 revimin) for I hr at room temperature. Centrifuge at 2000 x g for 5 min, uncap the tubes, aspirate the supernatants leaving a constant volume (0.1 ml) in each tube, and wash once with 1 ml assay diluent. Estimate the counts in the solid phase by placing each tube in the wellcrystal of a y-counter. Calculate results (Table 1.2).
Ch. 5
SEPARATION OF B O U N D A N D FREE LlCiAND
421
applied to virtually any binder capable of covalent attachment to the particle; (2) they are highly efficient and produce virtually complete separation of the bound fraction; (3) they give excellent precision if carefully used; (4) they are not as liable as some other systems to non-specific effects introduced by plasma and serum. At the same time, solid-phase methods have certain disadvantages which explain why these sophisticated systems are not in universal use: (1) the primary reagent is tedious to prepare; (2) the recovery of antibody activity on the solid phase is only 30% or less of that in the original y-globulin preparation, probably due to the fact that many of the molecules attach to the particles through their combining sites; the waste of activity is only acceptable if the supply of antiserum is abundant; ( 3 ) in the case of antibodies to larger molecules, attachment to solid phase results in a loss of affinity and hence of sensitivity in the assay; this is only critical with that minority of assays in which extreme sensitivity is required; (4)finally, and most important, the actual assay procedure is more complex than with some other systems; for continuous mixing the tubes must be capped and uncapped; a t the end of the procedure the washing of the solid phase sometimes involves several steps of centrifugation and aspiration. Thus, for any assay with a large sample throughput, particle solid-phase systems are not technically convenient. 5.3.7. Conclusions : the choice of'u separation procedure There is no magic in any one separation procedure. For a given assay system the choice potentially embraces all the techniques described above, and there is no reason to follow slavishly an earlier published technique. Instances abound of the perpetuation of a complex and sometimes inefficient method simply because it was the first to be described. For example, charcoal separation continues to be used in assays for steroids and small peptides, systems in which dissociation of the bound complex is very likely to occur and thus impair efficiency. Other systems offering equal and probably greater convenience are available and should be used. There is much to be said for keeping the number of separation Suh,r
422
RADIOIMMUUOASSAY AND RELATED TECHNIQUES
systems in any one laboratory to a minimum. Two or 3 general techniques whose characteristics are well understood and for which the reagents are freely available are clearly preferable to a multitude of different methods each with their own particular faults. In terms of radioimmunoassay the two techniques which have best stood the test of time are chemical precipitation and the use of second antibody. Faced with setting up a new radioimmunoassay, and assuming that the primary reagents are known to be appropriate, the worker would be well advised to explore the use of a simple precipitation system - for example, polyethylene glycol (see Table 5.3). If it is not possible to optimise a method of this type, for the reasons set out above, then a second-antibody system should be examined (see Table 5.4). In this author’s experience there is no assay for which one or the other system has not proved highly satisfactory. Regardless of theoretical arguments it should never be assumed that any separation technique is perfect and produces total separation of the free and bound fraction. Furthermore, it should not be assumed that, in terms of function, separation techniques differ only in their efficiency. The observation that two different procedures yield identical results for assay blank and zero standard does not necessarily mean that the composition of these fractions is identical: for example, in the one case much of the assay blank might consist of damaged tracer, and in the other of intact tracer precipitated because of an overlap between the characteristics of the bound and free fractions. Such a difference could yield a striking discrepancy between the actual assay results yielded by the two methods. Finally, it should be emphasised here as elsewhere that the results obtained with any separation procedure are almost certain to vary with the actual medium used in the assay (e.g. serum, plasma, urine). Comparison of standards prepared in a diluent buffer with unknowns which are serum samples will usually reveal non-identity which is in reality an artefact of the separation technique. The only solution to this problem is to ensure that standards are prepared in media as near as possible identical to that of the sample: for instance, in hormone-free plasma where this is available (see 9; 9.3.5).
Ch. 5
423
SEPARATION OF B O U N D A N D FREE LIGAND
5.4. Immunoradiometric techniques The elegant technique of immunorddiometric assay (Miles and Hales 1968; Woodhead et al. 1974) is considered in this chapter because it represents an important variant in which bound and free binder are separated rather than bound and free ligand. Reflection on the basic equations for binding assay systems (see 9; 1.4) will reveal that there is no fundamental difference between the two approaches, since the distinctions between binder and ligand are operational rather than real; it does not affect the basic logic of the system if the terms binder and ligand are reversed. Nevertheless, the immunoradiometric technique has certain potential practical advantages which make the procedure of great interest, although it is rarely employed in practice. The technique of a typical immunoradiometric assay is summarised in Fig. 5.10. A preliminary step is to isolate specific antibodies to the ligand by extraction of an antiserum with an immunoadsorbent, and for this purpose highly purified antigen is coupled to a solid phase such as cellulose or Sepharose. The antiserum is mixed with the solid phase and after an appropriate period for equilibration to occur the serum is removed leaving only specific antibodies attached to the solid antigen. As with other solid-phase systems (see above) extensive washing is essential at every step to ensure that no soluble material remains. The antibody is iodinated in situ on the solid phase using a standard chloramine T procedure and sufficient Iz5I to yield a substitution rate of 1-2 atoms of iodine per antibody molecule. Residual iodide and damaged products are removed by further washing of the solid phase, followed by elution of low affinity antibodies at pH 3 and higher affinity antibodies at pH 2. The purified labelled antibodies are collected into a neutral buffer and then stored until used, either in this form, or after recombination with the immunoadsorbent to improve their stability. In the assay itself (Fig. 5.10) samples and standards are incubated with an appropriate concentration of the labelled antibody (determined by experiment). After equilibration, unreacted antibody is removed by addition of an excess of the immunoadsorbent which Siibpl
index p 531
424
RADlOlM MUNOASSAY A N D RELATED TECHNIQUES
\ 2 days
[ insulin 3
<
fZ5
AB
I
supernatant
F-
residue
Fig. 5.10. The immunoradiometric assay of Miles and Hales. '251-labelled purified antibody to insulin is incubated with sample or standard. Free antibody is then separated by the addition of insulin coupled to a solid phase. The greater the amount of insulin in sample or standard, the smaller will be the amount of antibody bound to the solid phase.
after further incubation is removed, washed and counted. The greater the amount of ligand in the assay tube, the greater will be the amount of tracer antibody which forms a soluble complex and is therefore not available to bind to the immunoadsorbent. The result is a standard curve (Fig. 5.11) in which the number of counts in the supernatant is directly proportional to the amount of ligand present. A variation on this procedure is the so-called 'two-site' assay (Addison and Hales 1971). Unlabelled antibody is coupled to a solid phase and used to extract the antigen from a biological fluid. The
Ch. 5
SEPARATION O F B O U N D A N D F R E E LlCiAND
insulin
in
incubation
425
( pg )
Fig. 5.1 1. A standard curve for an immunoradiometric assay. performed according to the principle shown in Fig. 5.10.
insoluble complex so produced is separated and reacted with purified labelled antibody which, providing that the ligand molecule possesses at least two antigenic sites, will react in proportion to the amount of ligand present. The potential advantages of this procedure are that it avoids some of the non-specific effects which may arise from factors in the biological fluid, and that it permits an increase in sensitivity since relatively large volumes of the fluid can be extracted. Another variation, which is of value when only small quantities of antibody are available, is the use of an indirect procedure in which the purified antibodies are labelled with a preparation of [ 12SI]anti-IgG(Hales et al. 1975). 5.4.1. Advantages and disadvantages of'the imrnunoradiomrtric assay The advantages of the immunoradiometric assay are as follows: (1) iodination of the antigen is avoided, and with it the problems of Suhlrcl indexp 531
426
KADIOIMMUNOASSAY A N D R E L A T E D TECHNIQUES
damage to the tracer arising either during preparation or incubation (see $ 3.7.3); antibodies are large and relatively stable molecules and therefore less liable to damage; (2) because the antibody is iodinated as a complex with the immunoadsorbent, the combining site is protected; ( 3 ) the procedure can be applied to materials which are difficult to iodinate - for example, peptides lacking a tyrosine residue; (4)the shelf-life of the labelled antibody molecules, particularly when stored as a complex with the immunoadsorbent, is relatively long (2 months); ( 5 ) because it is possible to select high affinity antibodies from a heterogeneous population in the antiserum the procedure may be more sensitive than conventional techniques; however, this is a potential rather than a proved advantage, since a real increase in sensitivity has yet to be demonstrated either by theoretical analysis or practical experiment; ( 6 ) in the ‘two-site’assay it is possible to select and estimate only those molecules which contain both antigenic sites. The disadvantages of immunoradiometric assays are: (1) preparation of the basic materials is time-consuming and requires a high level of technical expertise; (2) as with other solid-phase systems, the consumption of reagent is high because of the losses associated with coupling and purification; for this reason, the technique is not suitable for assays in which the supply of basic materials is limited; ( 3 ) it is not suitable for antigen-antibody systems with a high dissociation constant, since the addition of an excess of solid-phase antigen at the end of incubation might disturb the equilibrium of the primary antigen-antibody reaction. Because of their practical disadvantages immunoradiometric systems have not been widely applied. The only exception is a modified version of the two-site system which is used in a popular kit for the measurement of hepatitis B antigen (Ausria, Abbott Laboratories). In this case the system has the great advantage that it can be supplied without specifically prepared standard antigen, thus avoiding the hazards involved in handling this potentially infectious material.
CHAPTER 6
Requirements for a binding assay - extraction of ligand from biological fluids
In the majority of radioimmunoassays the procedure consists of simple addition and mixing of sample, tracer and binder. However, there are situations in which this does not suffice, and the ligand must be processed prior to assay: first, if the assay does not have sufficient sensitivity and endogenous ligand must therefore be concentrated; second, if the assay does not have sufficient specificity and endogenous ligand must therefore be separated from other materials. These demands can be met by a variety of extraction procedures. Extraction procedures can be divided into 2 common types: those in which the ligand in the sample is adsorbed to particulate material and is then easily separated from unadsorbed components ; and those in which the sample is treated with an immiscible organic solvent which extracts the ligand. Adsorbtion to particles has been most widely used in situations where the ligand must be concentrated prior to assay (e.g. small peptide hormones). Partition with organic solvents is commonly employed where the requirement is separation of ligand from other components (e.g. steroid hormones).
6.1. Extraction.for concentration oj'ligand There are several important biological materials whose basal circulating levels are in the low picogram or even femtogram range (for example, ACTH, oxytocin, vasopressin, and the angiotensins). These concentrations are below or at the extreme limits of sensitivity of most radioimmunoassays, even when fully optimised (9 8.2) and a procedure for concentration is thus essential. 427
Suh/e(r ride\ p J3I
428
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
There is no theoretical limit to the extent by which sensitivity may be increased with the use of an extraction and concentration procedure. Furthermore, the process may confer considerable advantages in specificity. In practice the fluids most commonly studied are plasma and urine which represent opposite ends of a spectrum; plasma containsa high concentration of proteins and low fixed concentrations of electrolytes and urea; urine contains virtually no protein, but high and variable concentrations of electrolytes and urea. Once adequate methods have been devised for plasma and urine they are usually applicable to all other biological fluids including tissue extracts. The advantages of an extraction and concentration step are as follows: (1) Concentration of the endogenous ligand: the extraction procedure can be applied to a relatively large volume of the fluid in question and yields a final volume of extract which is very much smaller. Because the sensitivity of an immunoassay depends entirely on the concentration of a ligand (9: 8.2) and not on its absolute quantity, this confers a considerable advantage in net sensitivity of the assay (Fig. 6.1). (2) Improvement of specificity : a procedure designed primarily for concentration may also serve to improve the specificity of a procedure. This is further discussed in Q 6.2. (3) Avoidance of damage to endogenous ligand or tracer: many materials, particularly small peptide hormones, are associated in plasma or serum with potentially damaging enzymes - for example, oxytocin, which in late pregnancy plasma is rapidly destroyed by a circulating placental enzyme, oxytocinase. An extraction procedure can often separate the ligand from interfering agents of this type. 6.1.I. Extraction and concentration procedures using particulate adsorbents Virtually the whole range of physicochemical procedures used in the purification of biological materials might be applicable as extraction procedures in an assay. However, techniques which may be appropriate when the aim is a once and for all batch purification are
Ch. 6
429
EXTRACTION OF LlCiAND FROM BIOLOGICAL FLUIDS
CONVENTIONAL ASSAY
ASSAY AFTEREXTRACTION
Fig. 6.1. How an extraction procedure can considerably increase the sensitivity of a binding assay.
often not suitable in the context of an assay which may be applied to tens or hundreds of samples. The practical criteria for an extraction procedure associated with an assay are: (1) speed and simplicity, to match the large number of determinations which may be required; (2) concentration of the ligand into a volume considerably smaller than that of the original sample, to improve sensitivity; (3) in concentrating the ligand, the procedure should not concentrate nonspecific interfering factors including damaging enzymes; (4) the recovery of ligand should preferably be 50% or greater of the original material and, more important, should be reproducible; (5) the procedure should not cause denaturation of the ligand, or alteration of its binding properties; (6) the adsorbents should be readily available and not vary from batch to batch. Extraction for the purpose of concentration has been applied most extensively to the radioimmunoassay of small peptide hormones. The techniques which have proved most satisfactory in terms of the criteria set out above are those based on the batchwise addition of S u h p r indele\ 11 U/
430
RADlOlM MUNOASSAY A N D RELATED TECH NlQUES
adsorbents though other procedures, particularly ion-exchange chromatography, have also been used (Chard 1971). The commonly used adsorbents are all particulate silicates (e.g. Fullers’ earth, glass beads, silicic acid). These materials have in common a large surface area in relation to their weight. The precise nature of the adsorption process is uncertain. Most adsorption systems will be based on a procedure already described in the literature (see Table 6.1). Even under these circumstances, however, it is essential that the procedure be thoroughly checked because striking variations may be encountered between different laboratories in the use of an apparently identical material. If a new procedure is required, for a ligand not hitherto studied, then it is customary to examine a range of adsorbents - for example, Florisil, Fullers’ earth, and various types of glass beads such as the Spherosils and Vycor (Corning). Conditions of time, temperature, pH, ionic strength and the amount of adsorbent should be investigated for the initial adsorbtion process, and a range of solvents (e.g. aqueous acetone or ethanol) for the elution process. The initial stages of the study are much simplified by the use of isotopically labelled TABLE 6.1 An extraction and concentration procedure using a particulate adsorbent (Radioimmunoassay of ACTH, modified from Rees et al. 1971.) 1. To 2.5 ml plasma in a plastic tube add 50 mg Vycor glass (Corning Glass). 2. Cap tubes and mix by rotation for 30 min. 3. Centrifuge (2000 x g for 5 min); aspirate and discard the supernatant. 4. Wash the precipitate with 3 ml deionised water, then 2 ml of 1 M HCI: discard washings. 5. Mix precipitate with 2 ml of 60% (v/v) acetone in deionised water to elute ACTH. 6 . Centrifuge, carefully remove supernatant with a Pasteur pipette and transfer to a plastic tube. N.B., a different pipette should be used for each sample. I . Place the tube in a sandbath or waterbath at 50 C and evaporate to dryness with a fine jet of oxygen-free nitrogen. 8. Dissolve the residue in 0.5 ml diluent buffer for rddioimmunoassay.
Note: the procedure described is specific to the ACTH assay but with small variations can be used for a variety of small peptide hormones.
Ch. 6
43 1
EXTRACTION OF L l t i A N D FROM BIOLOGICAL F L U I D S
ligand as a tracer, and aqueous buffer solutions rather than a biological fluid as the medium for extraction. It must be recognised, however, that the tracer may not behave identically with labelled ligand, and that the recovery may vary with the composition of the fluid. Consequently, an extraction procedure can only ultimately be assessed by the recovery of intact, unlabelled ligand from fresh, whole blood or urine. When a satisfactory procedure has been established by pilot studies, the characteristics of the extraction system must be examined in detail with special reference to recovery, reproducibility, identity of extracted and unextracted ligand, and the problem of the blank value. Recovery: this is presented as the percentage of a fixed amount of ligand recovered in a series of extractions. The target is loo%, but this figure should not be slavishly adhered to; a simple and reproducible technique with a recovery of 60‘x will often be preferable to a complex and variable procedure with a recovery of 90x. Recovery should always be studied for different concentrations of the ligand, chosen to cover the range which is likely to be encountered in biological fluids. A procedure which shows substantial variation over this range should probably be rejected. Reproducibility: as in other situations this should be reported as the coefficient of variation of repeated determinations on the same sample. At least 20 observations are required for the figure given to have any validity, and it should be based on the results of experiments performed on separate occasions by different operators. It is particularly important to note whether the recovery varies significantly from one run to the next, and ‘significantly’ in this context means as a result of the performance of a statistical test (Student’s ‘t ’), not simply glancing through the data. If the difference between occasions is not significant then there is no need to run recovery controls with every assay - representing a considerable saving of time and materials. If the difference is significant then a separate recovery experiment is always necessary. Identity of’ extracted and unextracted ligand: the manipulations Siih,e
I I I ~ Cp ~ 531
432
I
involved in the extraction process can introduce two types of nonidentity. Either the ligand itself is physicochemically altered, so that it no longer reacts in the same way with the binder; or the extract contains irrelevant materials which interfere with the binder-ligand reaction. The most obvious result of non-identity is non-parallelism between dilutions of the unextracted ligand (standard) and the extracted ligand. This might reflect damage to some but not all of the binding sites on the ligand. Alternatively, it might be due to a completely non-specific chemical effect on the reaction such as that illustrated in Fig. 6.2. But it is very important to recognise that parallelism does not guarantee identity (see also 9: 9.2.2). Thus, partial damage to a single binding site, with a reduction in affinity between binder and ligand, will decrease the apparent potency of the extracted ligand but will not lead to non-parallelism. Similarly, there is no a priori reason to suppose that interference due to completely nonspecific factors will always lead to non-parallelism of the type shown in Fig. 6.2. Given apparent parallelism between extract and standard, final proof of identity can only be obtained from an experiment in which each of the standards is also separately extracted. For the 50 40
30 70BOUND 20
10 I
w
I
I
125 500 Zoo0 STANDARD ARGININE-VASOPRESSIN
Fig. 6.2. Non-specific non-specificity resulting from an extraction procedure (absorption and elution of tirginine-vasopressin from glass beads (Table 6.1)).0-0, A , serial dilutions of a plasma extract; 00, standard curve; 6serial dilutions of the same plasma extract which had not been adequately neutralised by washing (i.e. the antigen-antibody reaction was inhibited by acid pH).
Ch. 6
EXTRACTION OF LlCiAND FKOM BlOLOCilCAl FLUIDS
433
reasons given below, the experiment has to be conducted using standard added to biological fluid (e.g. plasma or urine) which is known to be free of endogenous ligand. The problem of’ the blank value: it is often found that extracts of a biological fluid, known to be free of the ligand, will nevertheless give a positive value when read against a standard curve prepared in a simple aqueous buffer solution. In some instances it is possible to recognise this ‘blank’ value as an artefact without difficulty - for example, if the ligand under investigation is a drug then samples from untreated subjects should definitively give a zero reading. In other instances, the identification is less simple - for example, with the basal levels of an endogenous hormone. ‘Hormone free’ samples can be obtained by 2 means: either from patients known to lack the hormone because the organ of origin has been destroyed, or by appropriate treatment of the sample with adsorbents known to remove the hormone. Neither type of sample is completely satisfactory. To guarantee that a patient has none of a certain hormone is rarely absolute; for example, many subjects who have had what should be a complete ablation of the pituitary will nevertheless continue to have normal or elevated circulating levels of prolactin. Artificial preparation of the sample by absorption of the hormone (as with charcoal for the production of insulin-free plasma) may be highly efficient but could easily have secondary effects on other components of the fluid which would eliminate a non-specific blank at the same time. It can never therefore be certain whether the blank is a true basal level or an artefact. Other studies, including changes under well-defined physiological circumstances and a formal physicochemical identification of the extracted material, should thus be performed.
6.2. Extraction.for purification of ligand There are many situations in which, although sensitivity is not a problem, the available binding assays do not have sufficient specificity to provide a meaningful determination of endogenous ligand. This may arise for 2 reasons: either the ligand is part of a family of Suhlcrr ride\ p 5.11
434
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
closely related substances any one of which may react equally well in the assay; or the ligand in a biological fluid is associated with other materials, such as binding proteins, which render it inaccessible to the assay. The most familiar methods which come under this heading are those in which steroid hormones and other small molecules are extracted by partition against an organic solvent. This may or may not be followed by a further step of chromatography. The general criteria for this type of procedure are virtually identical to those already set out for the particulate adsorbents (9 6.1). 6.2.1. Extraction to improve specificity There are numerous instances in the field of binding assays in which ambiguity may arise because of the presence of closely related materials in biological fluids. Probably the best examples are seen with steroid hormone assays, and in particular those which use naturally occurring binding proteins as the binder. As already emphasised ($4.3)this type of binder is likely to show a broad range of specificity, and some examples will serve to show the use of extraction to overcome the problem. (1) Measurement of progesterone using transcortin as the binding agent: transcortin will bind a variety of corticosteroids and other steroids in addition to progesterone. In order to achieve specificity, prior extraction of the sample with petroleum+ther, which selectively takes up progesterone, permits unambiguous determination of this steroid (Johansson 1969). (2) Measurement of androgens using sex hormone-binding globulin (SHBG) as binding agent: SHBG will bind a variety of 17j?-hydroxy-C19-~teroids,including both testosterone and oestradiol. A simple solvent extraction permits the separation and determination of ‘testosterone-like’ substances (Anderson 1970) which, though still a mixture, nevertheless provide a practical reflection of androgenic activity. Determination of testosterone alone then requires a chromatographic step in order to provide a full separation of the different 17j?-hydroxysteroids.
Ch. 6
FX'I KACTION OF LlClAND FROM BIOLOGICAL FLUIDS
435
( 3 ) Measurement of oestrogens using the uterine cytosol binder: this binder has an affinity for all the classic oestrogens including oestrone, oestradiol and, to some extent, oestriol. Determination of a single oestrogen therefore requires prior extraction and chromatography (Fig. 6.3). The range of organic extraction procedures available for steroids, drugs and other low molecular weight materials is extremely wide. Any system chosen must be rigorously evaluated according to the criteria already set out; this is necessary even for a widely accepted and published method because of variations in the quality of reagents particularly organic solvents. Problems with the specificity of the primary reagents are not confined to the steroid hormones. Other instances arise with the proteins and peptides, particularly where these exist in different molecular forms in biological fluids. 15
x
0 Volume Of Eluate
(
ML )
Fig. 6.3. Separation of tritiated oestrone (0-O), oestradiol (AA), and (0.1 ml containing 50,000 cpm) on a 10 cm x 0.5 cm column oestriol (0-0) of Sephadex LH-20 eluted with chloroform-hexane-methanol (46: 46: 8. v/v/v). This type of separation. applied to a plasma extract, permits subsequent specific quantitation of any of the 3 oestrogens using a relatively non-specific binding assay. (Data kindly supplied by Dr. S. Khoshroo.) Subicc I wde\ p 531
436
R A D l O l M MUNOASSAY A N D RELATED TECHNIQUES
The relative non-specificity of the circulating binding proteins is clearly an important disadvantage in the application of assays based upon them. For this reason they have been rapidly replaced by immunoassays in which a considerable degree of specificity is conferred by the primary binding reagent. To take the examples given above, highly specific antisera are now available to progesterone, to testosterone and to the 3 classic oestrogens. All permit determination of the relevant steroid after simple organic extraction without any need for chromatographic steps. In some cases it has even been possible to assay the steroid directly in unextracted fluids. 6.2.2. Extraction to f r e e the 1igand.from conjugates or complexes
Endogenous ligand in a biological fluid may be in a form in which it is not directly accessible to the assay - for example, as a conjugate, or as a complex with a circulating binding protein. With most steroid hormones a significant fraction circulates in the form of sulphate or glucuronide conjugates, and these will not react in assays directed towards the unconjugated material. In the case of oestriol in late pregnancy plasma, 90% or more of the steroid may be in this form. Hydrolysis of steroid conjugates in preparation for binding assays follows exactly the same principles as those applied to the classical fluorimetric or gas-chromatographic procedure. The sample is treated with either strong acid* or a hydrolytic enzyme**, and the steroid thus freed is extracted for assay. The advantage of enzymic hydrolysis is that, if the levels of the hormone are sufficiently high, the final extraction step may be unnecessary. An alternative to hydrolysis is the use of radioimmunoassays which are specific to the conjugate as a whole (e.g. to oestriol-16-glucuronide).This approach is theoretically very attractive since it eliminates a timeconsuming step which is also a potential source of error (Collins and Hennam 1976). * Prccdy and Aitken
(1961); 5-15 ml plasma arc diluted to 100 ml with water, 17.5 ml conc. HCI added and rcfluxcd for 45 min.
** Helix pomuriu extract (containing glucuronidase and sulphatc) (available from Calbiochcm).
Ch. 6
EXTRACTION OF LICrAND FROM BIOLO(ilC'AL F L U I D S
437
Complexing with circulating binding proteins is well illustrated by the steroids, and also with the hormones of the thyroid gland. The endogenous binder, which is usually a t high concentrations, will compete with assay binder for both ligand and tracer ligand. Thus, not only is a large part of the endogenous ligand sequestered and inaccessible to the assay, but also the endogenous binder will compete for the tracer. Depending on the nature of the assay binder (i.e. naturally occurring binder or antibody) and on the separation procedure used, the eventual result will be either a very low apparent level or a very high level, either equally unrelated to the circulating level of the hormone (Fig. 6.4). In the case of steroid assays the problem of endogenous binding can be overcome by an initial extraction step for example, ether will selectively remove unconjugated oestrogens including those which are protein bound. A similar approach has been adopted with the thyroid hormones. Alternatively, a dissociating agent may be used whose activity is sufficient to inhibit any reaction between natural, binder and ligand, but insufficient to affect the antigen-antibody reaction. A good example of this is the use of the organic detergent, anilino-naphthalene sulphonic acid (ANS) at final concentrations of around 1 mg/ml in the radioimmunoassay for thyroxine (Chopra 1972). This material can be incorporated with
0 Endogenous ligand
*
Tracer ligand
Fig. 6.4. The need for an extraction procedure when the sample contains an endogenous binder. The latter will compete with the assay binder for both endogenous and tracer ligand, thus invalidating any results obtained. Siihlvcr inder p. 531
438
RADlOlM MUNOASSAY A N D RELATED TECHNIQUES
one of the other reagents in the assay, such as the tracer, thus yielding a system of great convenience since it calls for no prior manipulation of the sample. A rather specidlised example of separation of ligand from endogenous binding protein is the measurement of ‘free’ hormones. In this situation the aim is to measure only that fraction which is not protein bound in the circulation, i.e., the fraction which is usually considered to be responsible for biological activity. Such a measurement can, in principle, eliminate the ambiguities which may arise from the estimation of total concentration because of variations in the amount of binding protein present. For example, subjects receiving oestrogen-containing oral contraceptive agents have elevated levels of thyroxine-binding globulin (TBG) ; total thyroxine levels are correspondingly increased, but the free fraction remains unaltered. The measurement of ‘free’ biologically active levels of a hormone can be achieved by dialysis of the serum sample and estimation of the concentration of the hormone in the dialysate, or by measuring distribution of labelled hormone in the same system. The final situation in which an endogenous binder can affect assay results is in the study of samples containing actively produced antibodies to the ligand. This may occur as the result of a planned immunisation in animals, or as an unintended side-effect of therapy in the human. For example, most patients on long-term therapy with hormones from natural sources (growth hormone, insulin, posterior pituitary extracts) will develop antibodies to these materials. In terms of the assay of endogenous ligand this has 2 effects: first, binding of the tracer by high levels of antibody in the sample will yield either very low or zero results (see Fig. 6.4); second, the presence of a circulating antibody, which will often negate the biological effect of the hormone, may be associated with total hormone levels which are in reality exceptionally high. The measurement of total hormone under these circumstances demands a preliminary extraction step. In practice, and depending on the nature of the material studied, this might involve any of the types of procedure already discussed. A more general method, which can in principle be applied to any
Ch. 6
EXTRACTION OF LlCiAND FROM BIOI.OCICAL FLUIDS
439
material but which is rarely employed in practice, is to dissociate the endogenous antigen-antibody complex with acid, and then to submit the sample to gel chromatography which will separate the free antibody and the free antigen.
6.3. General aspects of’ extraction procedures Though widely used in a variety of common binding assays, extraction procedures must be regarded as inherently undesirable. First, they inevitably lead to a loss of precision due to the summation of errors with every additional step in the procedure. Second, they may lead to problems of non-specificity which are difficult to control. Third, and most important, they drastically limit the number of samples which can be processed by a single operator. The addition of 2 or 3 reagents to a measured volume of sample in a straightforward nonextraction radioimmunoassay takes 20 sec or less, whereas none of the procedures described in this chapter takes less than 3-4 min of direct operator time. Throughputs of 100 or more samples per day may be reduced to a dozen. For all these reasons, extraction procedures should be regarded as an expedient, to be replaced whenever possible by a non-extraction technique. In practical terms this means one thing - the development of an improved binder, which will almost invariably be an antibody. Where sensitivity of an endogenous binder is the problem, an antiserum with higher affinity is required ;where specificity is the problem, an antiserum of better specificity.
Subjrcr index p. 531
CHAPTER 7
Requirements for binding assays - calculation of results
To judge from the literature, more effort is often devoted to increasingly complex and sophisticated methods for the calculation of the results of a binding assay than to the rather more important activities of optimising assay design and achieving an adequate system of quality control. In this section are discussed the simplest forms of calculation, together with the more sophisticated data transformations to linearise the dose-response curve, and calculator/computer aids to calculation.
7.1. Calculation of results by simple manual extrapolation Throughout the world the vast majority of binding assay results are calculated by this means. The standard points are plotted in any of the formats described in Q 1.6, and the points are then joined either by straight lines from one to the next or, more commonly, as the freehand curve which appears to the operator to fit most satisfactorily all the points. For those with an unsteady hand this procedure is facilitated by the use of a flexible ruler. The value of the bound/free distribution for an unknown sample is then located on this curve, and the corresponding standard value read from the horizontal axis (see Fig. 1.4). Any corrections for recovery of an extraction technique are applied to this value. Common though it is, several criticisms can be levelled at this procedure. The first concerns the plotting of the standard curve itself. Simple joining of the points by straight lines may often yield an irregular 'curve' which clearly does not reflect the true situation, but rather is influenced by the errors to which each individual point is 440
Ch. 7
CALCULA’IION OF K E S U L l S
441
subject. This method is not recommended. Joining the points by a continuous ‘line of best fit’ as judged by eye is easily attacked because of the subjective nature of the operation. Nevertheless, the human eye and brain as an analog machine is greatly underrated, and the end result will often be superior to that obtained by all but the most elaborate of electronic equipment. In particular, the eye is capable of taking in and excluding error points which are obvious outliers and of forming curves which mathematically are highly complex. The second and much more important disadvantage of manual plotting and reading of results is the errors to which it may give rise. This includes the straightforward mistake, when one of the coordinates is incorrectly read - a surprisingly common fault, especially if large series of figures are examined a t one time. It also includes the tendency of many operators to round off figures to the nearest convenient integer - for example to the nearest ten. These problems in manual calculation are much the best argument for machine calculation of results.
7.2. Linearisation of the standard curve Many difficulties, including the subjectivity of drawing a curve, would be solved if the dose-response relationship in a binding assay were linear. By appropriate manipulation of the data it is, in fact, possible to achieve this in many systems (see Q 1.6). The most familiar of these techniques is the ‘logit transformation’ of the response variable first introduced by Rodbard in 1970 and since then very widely applied. The logit is calculated as: logit b = log e
(-3 loo - b
where b is the proportion of tracer bound expressed as a percentage of that in the zero standard. Plotted against log dose this yields a straight line for many if not all assays (see Fig. 1.14). This is easy to plot manually (logit-log graph paper is available for the purpose). Subjecr index p . 531
442
KADIOIMMUNOASSAY A N D RELA-rED TECHNIQUES
More important, it is readily adaptable to the linear least squares regression programmes of the larger desk calculators and so provides a means for complete or near-complete automatic calculation of results using widely available hardware. However, certain practical points should be noted about the use of logit transformation: (1) The assay blank value must be subtracted from the % bound before the logit is calculated. Failure to do this will yield a non-linear response. For the same reason, assays with a high and variable blank value are often poorly suited for logit transformation. (2) The upper and lower 10% of a standard curve are frequently non-linear in logit transformation and therefore should be eliminated when the line of best fit is drawn or calculated. Similarly, values recorded for unknowns above and below these limits should be rejected as inaccurate. Exclusion of values of this type is anyway good practice in a binding assay. (3) The ‘goodness of fit’ of the straight line resulting from logit transformation is often judged by the correlation coefficient or r value of a least squares regression. It should be noted that, for a set of figures which the eye would consider a good fit, the r value is always 0.95 or greater. Lower figures, which might be perfectly acceptable in a statistical exercise on biological data, usually imply a deviation from linearity or that one or more points are grossly in error. (4)In many systems the logit transformation does not yield a straight line, and is then of no more practical use than any other way of plotting the standard curve. This is often found to be the case with radioimmunoassays and is due to heterogeneity of the antiserum - the same reason which explains why many systems cannot be analysed as a simple algebraic expression of the law of mass action (see Ch. 1). ( 5 ) The variable precision at different parts of the response (referred to as ‘heteroscedasticity’) implies that the best linearisation can only be obtained after weighting of the individual points ; this procedure requires substantial computing facilities. A variety of other transformations have been described which
Ch. 7
CALCULATION OF RESULTS
443
will linearise some, but not all, standard curves. Ekins (1974) has suggested the expressions :
(; i:)
log - - and
(A it)
log - --
where B is the fraction of tracer bound, F the free fraction, and B,, and F, the fractions in the absence of unlabelled ligand. These are mathematically equivalent to the logit transformation and therefore will have the same advantages and disadvantages. Other transformations, such as arc-sin and logistic have found little practical application and again are very similar to the logit-log model. Another and more general approach to the mathematical analysis of a standard curve is to fit a polynomial expression, usually a cubic or quartic, or a spline function. This will invariably require some sort of computational facility but has the great advantage that it is completely empirical and makes no assumptions about the underlying nature of the curve.
7.3. Electronic aids to calculation of’ results Virtually no laboratory today lacks some form of computational aid, even if it is only a simple calculator to work out means of replicates and percentages. More advanced aids to calculation can be divided as follows: (1) the large general purpose calculator, such as the Hewlett-Packard Model 10 ; (2) the general purpose digital computer whichcan be ofany common size; and ( 3 ) the ’dedicated’ micropressor, designed exclusively for the analysis of binding assay data. In turn, there are 3 ways in which any of these may be fed with the primary data (i.e. the output of the counter): (a) manual introduction of figures read from the scaler or counter print-out ; (b) use of punched tape output from the counter, which is then introduced via a tapereader; and (c) a direct on-line connection to the counter. Suhpct rrrde/er11 531
444
RADIOIMMUNOASSAY AND RELATED 1EC H N I Q U E S
To some extent the type of machine and input will be determined by availability and practical necessity. The laboratory which conducts assays on only a few dozen samples weekly would be well advised to use simple manual plotting and interpolation. However, above a certain limit - probably around 100 samples in a single assay machinecalculation becomes both more efficient (because it eliminates some errors) and cost-effective (because it represents a substantial saving in operator time). The use of calculator or digital computers is largely determined by availability. The large computer has the advantage that with appropriate programming, it can produce a very detailed analysis of the assay and its results in a neatly typed format. But it will almost always be used on a time-sharing basis and this can lead to serious difficulty in obtaining access with consequent delays. The large calculator is more limited in its capabilities, particularly formatted output, but being relatively inexpensive is more likely to be available on site in the assay laboratory. In fidct, current developments in desk calculators and their peripherals put them nearly on a par with the full digital computer, the more so because this is an application which does not demand extensive memory stores. The type of input is again determined by availability. With a calculator the count data can be entered manually but this is liable to error and punched tape is preferable. The digital computer will almost always be fed with paper tape. Both can, in principle, operate directly on-line to the counter; however, this is extremely wasteful because the slow rate of data entry, given the counting times currently in use, means that the data processor will only be operating at 1% or less of its potential speed. Counters incorporating this type of facility, usually at great expense, are not recommended. Exceptions to this rule would be the use of a data processor on-line to a group of counters, or to the type of multidetector counter which is becoming increasingly popular. A new development is the use of data processors dedicated to the analysis of binding assay data (Bell and Howell, Ltd.). These are a spin-off of the ‘calculator revolution’. They can be small, cheap,
Ch. 7
CALCULATION
OF RESU1.TS
445
and yet retain the abilities (such as polynomial fitting with exclusion of error points) of the largest digital computer. It will not be long before such processors are in routine use in most laboratories.
7.4. Estimation of‘confidencelimits to the results Every estimation has an error, and the purist would demand that this be presented as part of the result. In other words, that a value be not given simplyas’lo’, but rather as ‘9-1 1’. A simple way to calculate this is to apply the within-assay error to the figure obtained. For example, if the coefficient of variation is known to be 5”/,, then the 952, limits to an estimate can be given as the observed value plus and minus 10%. But for a binding assay, estimation of confidence limits presents the problem of the variation of precision with dose (see 9; 10.2.1.5). Sophisticated approaches to the determination of error have been described (see Rodbard 1971) using weighted values to take into account this variation. However, the procedure demands fairly elaborate computational facilities. More important, the majority of binding assays are performed as a clinical service and the average physician will not take kindly to receiving results as a range rather than a single figure. Calculation of error, other than as part of the quality control procedures which should be a feature of every assay (see Ch. lo), is unnecessary for most practical purposes.
CHAPTER 8
Characteristics of binding assays - sensitivity
8.1. Definition of sensitivity Sensitivity is defined as ‘the minimal detection limit of an assay’. In practice, it is frequently and incorrectly used to describe other aspects to which it is only indirectly related, such as the slope of the standard curve. The minimal detection limit requires further and careful definition inasmuch as it is often judged from a cursory and over-optimistic inspection of the results of a single experiment. Rather, it should refer to the least concentration of unlabelled ligand which can be distinguished from a sample containing no unlabelled ligand, the distinction being based on the confidence limits of the estimate of the zero standard on the one hand and the standard on the other. This concept is illustrated in Fig. 8.1. Furthermore, confidence limits in this context should be described as a statistical function and not merely as the mean and range of a pair of replicates. For example, consider a pair of estimates of 100 and 105 (which could be any parameter reflecting the response in the standard curve); this would yield a mean of 102.5 with confidence limits of 97.6-107.4. If further replicate estimates are available the confidence limits are sharply reduced. Thus, the set of estimates 99, 100, 105 and 106, which has the same mean but an apparently broader range, gives confidence limits of 99.1-105.9. A practical point of vital importance becomes obvious from Fig. 8.1 and the discussion above. It is that the minimal detection limit of an assay is critically dependent on the precision of the assay. Great efforts to optimise assay design, on the lines set out below, will be 446
Ch. 8
447
SEN SIT1V I T Y
----------
confidence limits to zero standard estimate confidence limits to
40%TRACER BOUND
standard estimate
3020-
detection limit of assay
I 0.016
0.064
0.25
CONCENTRATION OF STANDARD Fig. 8.1. The definition of ‘sensitivity’ or ‘minimal detection limits’ of an assay. This is the least quantity of ligdnd which can be distinguished from a sample containing n o ligand (the zero standard). Note that this figure is critically dependent on the precision with which the estimates are made.
completely vitiated if the reproducibility of assays is poor. A further point which emerges, and which is rarely taken into account in the literature, is that the minimal detection limits are likely to vary from one assay to the next and from one operator to another. Thus, a lower limit stated as ‘ 1 0 on the basis set out above will not apply to an individual assay in which the replication of this point or the zero standard or both is poor. Under these circumstances the lower limit might become 20 or even 40. A final point in the definition of sensitivity is the units in which it is specified. There is no doubt that this should always be in terms ofconcentration (i.e. weight/volume) and should refer to the biological fluid for which the assay is intended. Yet this simple and obvious rule is often broken. A common fault is to specify the absolute amount of ligand in a given tube rather than its concentration: suppose an assay has a true sensitivity of 1 ng/ml; for an incubation volume Suhle
448
RADIOIMMUNOASSAY A N D RFLA I FDTECHNIQUES
of 1 in1 this gives an absolute quantity of 1 ng; for an incubation volume of 0.1 ml it gives 0.1 ng. To the unwary the latter would appear to be the more sensitive, but in reality they are identical. Another fault is to specify sensitivity with respect to standard in aqueous buffer solution. This might yield a figure of 1 ng/ml for the whole volume in an assay tube. But if, in order to avoid non-specific effects, thevolume of sample is very much less than the total incubation volume (e.g. 50 p1 in a total volume of 1 ml) then the effective sensitivity is much less (i.e. 20 ng/ml in this case). Thus, sensitivity should always be expressed as the concentratiori in the biological,fluid under investigation. Extreme sensitivity is still generally regarded as the hallmark of a good assay, and much effort is devoted to its achievement. This arises from the fact that when radioimmunoassay was first introduced it was the sensitivity of the procedure, relative to other techniques available at the time, which led to its rapid establishment in biology and medicine. Yet for many important compounds high sensitivity is not a requirement, since their levels in biological fluids are orders of magnitude above the minimal detection limits of the assay. Traditionally it has been usual to dilute the sample so that the levels correspond to readings on the standard curve. However, there is much to be said for adjusting the conditions of the assay in such a way that the standard curve corresponds to the range of biological interest, assuming no sample dilution. ‘Biological targeting’ of this type is at least as important and, in the field of clinical medicine, more important than is maximisation of sensitivity. Considerable attention will be given to targeting in the discussion which follows.
8.2. Methods ojincreasing the sensitivity oj’a binding assay The choice of reagent concentrations in a binding assay has already been described briefly (5 lS), and in this section the conditions for achieving maximum sensitivity will be discussed in more detail. The headings set out below are in no sense mutually exclusive and for any one assay several of the approaches are likely to be combined.
Ch. 8
449
SENSITIVll-Y
8.2.1. Reduciiig the amount oftracer This is the most familiar approach to increasing the sensitivity of an assay and is highly effective provided that its limitations are recognised. Unfortunately, there is a widespread assumption that the minimal detection limits of an assay are directly and completely related to the amount of tracer used; in other words, that the sensitivity of an assay can be increased indefinitely simply by reducing the amount of tracer. This assumption, in turn, has led to the pursuit of high specific activity tracers, with all their attendant problems, when in reality a tracer of much lower activity would be perfectly adequate. The limitations on the amount of tracer can be judged both from theoretical analysis (9 1.10) and by experiment. The conclusions of either approach can be summarised as follows: that for any given set of conditions (affinity and concentration of binder, time of incubation, etc.) there is a limiting concentration of tracer below which further reduction leads to no significant change in the position of the standard curve or the sensitivity of the assay. This concept is illustrated theoretically in Fig. 8.2. In practice, an approach to finding the appropriate concentration of tracer for use in an assay is as follows: (1) select, as a rule of thumb, a concentration of tracer approximately equivalent to the least amount of unlabelled ligand which the assay is intended to measure; (2) perform a binder dilution curve and select the concentration of binder which binds around 502, of the tracer; ( 3 ) set up standard curves covering a wide range of concentrations of unlabelled ligand; add the concentration of antibody defined by (2) above; add 3 different concentrations of tracer - the amount specified under (1) above, an amount one-tenth of this, and an amount 10 times this value. The result, if the antiserum is of adequate affinity to yield results in the chosen range, will be 3 standard curves which can be interpreted as shown in Fig. 8.2. Other experimental approaches to the choice of a lower limit of tracer concentration have been described: for example, preparing a set of binder dilution curves with different amounts of tracer or a mixture of tracer and unlabelled ligand, or selecting a fixed concentration of standard at the intended limit of the assay and then comparing this Sithi<,
53/
I I ~ F p~
450
R A D I O I M M U N O A S S A Y A N D R E L A T E D 'TECHNIQUES
Binder concentration
I
.008
.OD
.I25
.5
2.0
Concentration of standard Fig. 8.2. Increasing the sensitivity of a binding assay by decreasing the amount of tracer (theoretical system, see 9: 1.9 and 9: 1.10). Note that below a certain limiting level (0.01 in this case) further reduction in tracer does not alter the characteristics of the assay.
with the zero standard using different amounts of tracer. The exact approach used is less important than an understanding of the basic mechanisms involved. The choice of the optimal amount of tracer has implications for the initial preparation of labelled ligand. In operative terms the aim is a tracer yielding total counts of 10,000 in 1 min or less. If the leait amount of tracer dictated by the above experiments yields considerably more counts, then it should be possible to reduce the specific activity of the initial preparation with all the advantages (less damage, easier handling) attendant upon this. Equally, the experiments might dictate an increase in specific activity. 8.2.2. Reducing the amount ojbinder A reduction in the amount of binder will increase the sensitivity of an assay. This is illustrated in Fig. 8.3 which demonstrates the wide
Ch. 8
45 1
SENSITIVITY
%
BOUND
.M25 .25 1 Moles of unlabelled antigen
4
16
62
Fig. 8.3. Increasing the sensitivity of a radioimmunoassay by decreasing the amount of antibody (theoretical system, see $ 1.9 and 5 1.10).Note that reducing the amount of antibody progressively increases the sensitivity, though this is eventually limited by the precision of estimates at the individual points.
variation of results which can be achieved by this means. Inspection of a simple theoretical presentation of this type might suggest that there is almost no limit to the possible increase of sensitivity. In practice, however, it is limited by the precision of the assay. Take, for example, a curve with a zero standard of only 10%; when this is drawn out on the scale shown in Fig. 8.4 it might a t first sight seem to offer exceptional sensitivity. But when the confidence limits of the replicates are taken into account it can be seen that the first point which differs significantly from the zero standard is almost in the middle of the curve. Thus, increasing sensitivity by decreasing the binder must always be a balance between what can be achieved theoretically and the reproducibility of replicates within the assay. This important point has been frequently and elegantly put forward by Ekins (1974). As a rule of thumb, the use of conditions in which the zero standard is 20”/, or less pf the total counts will often be found to be counter-productive - any increase in apparent sensitivity being negated by the decrease in precision. SuhjecI indcr p 531
452
I I A D I O I M M U N O A S S A Y A N D IIEI.ATI
12
mean and conlidana limits of zero standard
10
a
%tracer bound
6
mean and conlidma hmiu of f m point vhich a n bs
4
dmnwrhed from the zero standard
2
I 0
. 1
2
4
a
16
Standard Concentration
Fig. 8.4.How an apparently very sensitive assay, based on the use of a low concentration of bindcr (lOy,r, bound in the zero standard), is in reality much less sensitive because of the wide confidence limits to individual points.
8.2.3. Increasing the incubation time Increasing the incubation time is not, in itself, a means of increasing sensitivity; indeed, it may have the contrary effect. However, it is often implicit in the use of very small concentrations of reagents inasmuch as these will require a considerable length of time to reach equilibrium. In exploring the lower limits of an assay total incubation times of up to 7 days may be necessary*.
* It might be assumed that antibodies of high K value (i.e. with a large k,) would come to equilibrium very rapidly. However, there is evidence with multivalent antigens and antibodies that some conformations affording thermodynamically very stable binding are only rarely obtained (Barisas et al. 1977). and attainment of true
Cli. 8
SENSIlIVITY
453
Apart from achieving equilibrium with low concentrations of reagents, lengthy incubation times have almost no advantages and several disadvantages. First, they may not actually yield the most sensitive assay, a point further discussed below under the heading of disequilibrium assays. Second, they expose the reagents, and in particular the tracer, to the risk of damage by other components of the assay system: for example, in the case of protein hormones, to attack by proteolytic enzymes. For this reason it is often found that binding in the zero standard shows a progressive decrease with increasing incubation time, a point which is clearly unfavourable in terms of sensitivity. Finally, long incubation periods are technically inconvenient: storage space is required for large numbers of tubes; errors due to tube misclassification become more likely; and there is a substantial delay in the reporting of results to the clinician. In the initial development of an assay every effort should be made to reduce the incubation time. 8.2.4. Reducing the incubation time - disequilibrium assays If an assay is interrupted by separation of bound and free ligand before equilibrium is reached then zero standard binding is reduced and the apparent sensitivity may be increased. The effect is not dissimilar to that obtained by reduction in the concentration of antibody, the main difference being that it is achieved more rapidly. However, the separation procedure takes a finite length of time. If this is a significant fraction of the total time taken for the assay then the possibility arises of a substantial difference between the first and last tubes of a run, the last tube having had rather more time to reach equilibrium. An approach of this type requires great attention to detail if it is to be successful. Dramatic reductions in incubation time are now becoming the hallmark of many commercial radioimmunoassay kits. Sometimes equilibrium may be very slow indeed. This is due to the fact that the most favourable free energy may only be achieved when the highly flexible antibody molecule passes through an infrequently attained conformation. Suh,cn ,ndelrrp 531
454
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
this is achieved at the expense of a loss of accuracy; for example, in a survey of serum gastrin kits substantial deviations from a reference method were noted with incubation periods of 1.5-3 hr, which could be corrected by extending the time to 24 hr (Gibson et al. 1977). Assays with unusually short incubation times should be carefully evaluated for this phenomenon. Another type of disequilibrium assay, with late addition of tracer, is discussed below.
8.2.5. Order of addition of reagents The usual order of addition of reagents in a binding assay is sampletracer-binder. Under these circumstances both unlabelled and labelled ligand have equal access to the binding sites, and the system can be analysed relatively simply on the basis of two reacting components. However, if the order of addition of tracer and binder is reversed (i.e. sample-binder-tracer) the end-result can be a totally different standard curve and one which has much improved sensitivity (Samols and Bilkus 1963). This is the so-called ‘late-addition’ assay. An example of the effect of reversing addition of binder and tracer is shown in Fig. 8.5. This is a particularly striking case but in many assays, for instance those of small peptide hormones, reversed addition has no significant effect. Whether it applies to any given system must be established by trial and error. 8.2.6. Purification ofthe binder The heterogeneity of the antibody population within a given antiserum has already been emphasised and superficially it might appear desirable to select only those of highest affinity for use in an assay. This could be achieved by prior treatment of the antiserum with an immunoadsorbent, low affinity antibody molecules being eluted using mild dissociating conditions, and higher affinity molecules with more drastic conditions. Purification of an antiserum in this manner has several potential technical disadvantages : it wogld require a substantial supply of pure antigen for the preparation of the immunosorbent; it would be wasteful of antiserum because of poor recoveries;
Ch. 8
455
SENSITIVITY
%
of tracer
1
2
4
6
12
Concentration of standard Fig. 8.5. Increasing the sensitivity of a radioimmunoassay by adding the tracer after, rather than before, the antibody (radioimrnunoassay for human placental lactogen).
it is time-consuming and technically demanding; and it could result in damage to the antibody under the conditions used for dissociation of the molecules of highest affinity. More important, there is considerable doubt as to whether the intended effect - an increase in sensitivity of the assay - is ever actually achieved. On theoretical grounds it can be argued that simple dilution of an antiserum constitutes an immunopurification, since at low concentrations only those molecules of high affinity will be significantly involved in the antibodyantigen reaction. A possible exception to this would be if the antibody population consisted of a very small number of high affinity molecules and a very large number of low affinity molecules. Finally, it has never been convincingly demonstrated in practice that antibody selection yields an increase in sensitivity; claims to the contrary are usually invalid because the basic system has not been tested in fully optimised conditions. Purification of antibody cannot be recommended as a routine procedure. Suhlrcr mderp 53/
456
RADLOIMMUNOASSAY A N D RELATED TECHNIQUES
8.2.7. Increasing the sample volume It has already been noted that the effective sensitivity of the assay is determined in part by the volume of sample in the whole incubation mixture (5 5.1). It follows from this that an increase in sensitivity can be achieved by an increase in sample volume relative to the other components. In many assays the sample represents only 10% of the total incubation mixture (e.g. 50 pl in 0.5 ml); yet since the other reagents could if necessary be added, without loss of precision, in as little as 50 pl there would a t first sight seem to be no objection to increasing the sample volume to 450 pl, thus enhancing sensitivity 9-fold. In practice, this approach is limited by the fact that biological fluids will often produce non-specific interference if incorporated a t high concentrations (see 9: 9.3). The degree of interference varies considerably with the separation system and is probably least with well-optimised second antibody and solid-phase systems. Nevertheless, it must always be recognised that an increase in sample volume may lead to an increase in non-specific effect; if this approach is used to improve sensitivity the possibility of this must be carefully evaluated. As a rule of thumb (but with several exceptions) non-specificity due to irrelevant materials in the sample is insignificant if the total concentrationofsample in the incubation mixturedoesnot exceed 10%. 8.2.8. Temperature of incubation The rate of most biological reactions is temperature-dependent and, all other things being equal, the rate of approach to equilibrium of a binding assay will double for every 10 C rise in ambient temperature. This principle may be used to enhance the speed of an assay; however, it will not usually affect sensitivity and will also increase the rate of undesirable reactions such as enzymic destruction of the tracer. The likely upper limit of temperature is 45 C. The major exception to this generalisation is the group of assays which use a circulating binding protein as the binder. The affinity constant of these is very substantially increased at lower temperatures, and as expected the sensitivity is also increased. Immunoassays for peptides may show a similar phenomenon, and if extreme sensitivity
Ch. 8
451
SENSITIVll Y
is required the use of low temperatures should be explored (Keane et al. 1976). An assay which is truly temperature-dependent can present serious technical problems. Maintenance of large numbers of tubes at a fixed and identical temperature is not simple; for example, if a rack of closely packed tubes is removed from a cold room or refrigerator and placed on the laboratory bench a significant temperature difference (5'C or more) can soon develop between tubes on the periphery of the rack and those in the centre. This also applies to assays using an enzyme as the label (see 9 3.8.2). Radioactivity, by contrast, is not temperature-dependent. 8.2.9. Increasing the number of'replicates It has already been emphasised that sensitivity is critically dependent on precision. In turn, the confidence limits to an estimate are dependent on the number of replicate determinations; increasing this number will narrow the confidence limits and thus enhance the sensitivity of the assay. This approach is little used in practice because of the extra work involved. Nevertheless, of all methods so far discussed, it is the only one which, for a given set of reagents and conditions, can produce an almost unlimited increase in sensitivity. 8.2.10. Extraction and concentration
The means by which sensitivity can be enhanced by extraction and concentration has already been described (9 6.1).
8.3. Methods of decreasing the sensitivity of an assay The importance of targeting the standard curve to a range of clinical and biological interest has already been described. With very few exceptions, binding assays and particularly radioimmunoassays have a potential sensitivity (using fully optimised conditions) which is considerably in excess of that required for practical day-to-day operation, and it is thus a positive advantage to decrease the sensitivity. Suhierr mdel p . 531
458
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
The primary method for decreasing sensitivity is to increase the concentration of the binder. This will yield a progressive shift of the standard curve towards higher concentrations of unlabelled ligand (Fig. 8,6). An increase in binder concentration has several implications for the operation of an assay which can be summarised as follows: (1) Concentration of tracer: when the binder concentration is increased the amount of tracer can also be increased (a mirror image of the situation in which a limiting level of tracer is selected to enhance sensitivity). An increase in the tracer has 2 important advantages. First, for a given specific activity, it leads to a corresponding decrease in the counting time required. Second, it permits a reduction in the specific activity of the tracer with all that this implies in terms of ‘damage’ and increased stability. In a grossly desensitised assay such as that shown in Fig. 8.6 both advantages are gained: a very low specific activity tracer can be used in sufficient quantity to yield adequate counts in 10 sec or less, with no detriment to the position or shape of the standard curve. 80
60
%lZ5 I-APL bovnd
40
20
n” 0
2
8
32
125
500
2.m
hP1 (ng/O.lml)
Fig. 8.6. De-sensitisation of a radioimmunoassay for human placental lactogen. Increasing concentration of antibody (shown as the reciprocal of the titre at which it is used) produces a progressive shift of the standard curve to the right.
Cli. 8
SENSITIVITY
459
(2) ‘Zero’ standard and shape of curve: the use of a large amount of binder implies a high level of binding in the ‘zero’ standard; following from this the curvewill be steeper, and estimates of unknowns are more precise. (3) Incubation time: the use of high concentrations of all reagents means that equilibrium is reached very rapidly. For an extreme example such as that of Fig. 8.6, equilibrium at higher concentrations of antiserum is virtually instantaneous, and the separating agent can be added immediately after the binder. (4)Sample volume: this can be reduced to a minimum compatible with accurate and reproducible pipetting (in practice, about 50 PI). A small sample volume confers 2 further advantages: first it implies a substantial dilution of the sample in the incubation medium and thus obviates non-specific effects; second, it means that only a small specimen of blood is required from the patient. This is of particular importance today when multiple tests on a single sample are becoming the rule rather than the exception. A further advantage of a desensitised assay is that it may eliminate the need for prior dilution of the sample; this step must inevitably introduce another source of error which is reflected in reduced precision of the final result. (5) Affinity of the binder: if the binder is used in relative excess in order to yield a de-sensitised assay, then clearly the binder need not have exceptionally high affinity. This greatly extends the range of different binders which can be used - an antiserum rejected because it will not yield a certain level of sensitivity may be perfectly satisfactory for an assay targeted to a lower sensitivity. The opposite side of this coin is that a de-sensitised assay using high concentrations of binder will place considerable demands on the supply of binder. One ml of an antiserum used at 1 : 100,000 will suffice for the assay of some 40,000 samples; the same amount of antiserum used at 1 : 1000 is only enough for 400 samples. (6) Convenience: most of the features set out above - short incubation times, short counting times, optimal shape of standard curve make for technical convenience in the performance of the assay. It cannot be reiterated often enough that the eventual clinical value of a Sirhlerl m d e r p 531
460
RADIO1 M M UNOASSAY A N D RELATED TECH NlQU ES
technique is directly and completely related to the ease, speed and reliability with which it is carried out.
8.4. Targeting of'binding assay
-
the importance of ranges
Given that an assay should be arranged to suit a clinical purpose, rather than vice versa, it is worthwhile to consider briefly the choice of ranges for the standard curve. The basic principle here is that the most precise part of the standard curve (usually the central portion) should be set so that it coincides with the range of practical interest: this may either be the whole of a normal range (for materials with which both high and low values may be of diagnostic significance) or one extreme of a normal range (for materials with which either low values or high values, but not both, are of diagnostic significance). This principle is illustrated in Figs. 8.7 and 8.8. loo
range d physiological or clinical interest
r
60 %TRACER BOUND
40
.032
.128
.5
2
a
CONCENTRATION OF STANDARD
Fig. 8.7. Targeting a binding assay by varying the concentration or binder. In the system shown a very sensitive assay is yielded by a binder concentration of 0.1. However, it embraces only part of the range of physiological and clinical interest. By increasing the binder concentration to I a less sensitive standard curve is produced which covers the range of interest and is also considerably steeper.
Ch. 8
461
SENSITIVITY
100
t
-
80
from abnormal population mean value for normal population
60 Itracer
bound
40 20 1
.5
1
,
8 32 hormone concentration 2
Fig. 8.8. Targeting of a binding assay. I f the aim is to study a normal population, then the mid-point of the standard curve should be chosen as the mean value of the normal population (curve on right). If the aim is separation of a normal from an abnormal population. then the mid-point should be the value which best separates these two populations (curve on left).
8.5. Optimisatioi7 of'an assay by tlieoretical analysis If a binder-ligand system can be fully characterised with respect to the absolute concentrations of reactants, the K value of the binder and the errors implicit in the determination of the distribution of bound and free ligand, then it is possible to substitute these factors in a theoretical model and thus, by a purely mathematical exercise, to explore the performance of the assay and to select optimal concentrations of reagents. The simplest available model is the algebraic expression of the mass action equation set out in 4 1.9 and eq. (1.6). However, this takes no account of critically important factors such as differences between tracer and unlabelled ligand, heterogeneity of the binder and errors in determination. Vastly more sophisticated techniques have been described which take these factors into account: all require the useofacomputer, togetherwitha degree of mathematical J l I h / < < I rlfd<,\ /I
(3/
462
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
understanding which is exceptional in most workers in the biomedical field. Excellent reviews of this area are available (e.g. Ekins 1974), and those wishing to adopt this approach would be well-advised to consult a recognised authority. They should also appreciate that, with certain exceptions, some simple experimental trial and error on the lines set out above may yield an equally satisfactory answer.
8.6. Conclusions Optimisation of assay design is at the very core of radioimmunoassay and related techniques. All other aspects - preparation and examination of reagents - are in reality only branches of chemistry and biochemistry. A full understanding of the principles of assay design is the major characteristic of the expert in this field. The points emphasised in this chapter are the correct definition of sensitivity, the means for achieving the highest possible level of sensitivity, and most important, the targeting of the assay with respect to its intended use - the only merit of a technique being in its practical application.
CHAPTER 9
Characteristics of binding assays - specificity
9.1. Definition of‘specijicity The specificity of an assay can be defined as ‘thg degree to which an assay responds to substances other than that for which the assay was designed’. ‘Degree’ in this case is a relative term since it will vary considerably with the conditions of the assay - concentration of tracer and binder, separation procedure, etc. For example, in a radioimmunoassay it is possible to design conditions in which at one concentration of antibody there is no significant interference by a cross-reacting material, while at another concentration the interference is such as to render the assay valueless for all practical purposes. The subject of specificity will be discussed under two separate headings, which for convenience and familiarity will be termed ‘specific non-specificity’ and ‘non-specific non-specificity’.The former refers to interference by identifiable materials which are physicochemically similar to the ligand and may thus react directly with the binder. The latter refers to interference by materials which do not directly react with the binder, but can nevertheless affect the primary binder-ligand reaction - for example, acid conditions which produce partial or total inhibition of any antigen-antibody reaction. The two types of non-specificity may be difficult to distinguish, yet in operative terms it is highly important to do so because the means for their elimination may be quite different.
463
464
RADIOIMMUNOASSAY AND
K E L A I F D TFCHNIQUES
9.2. Speci/i’c non-specificity There are many groups of biological materials which are physicochemically very similar - for example the various steroid hormones, or the glycoprotein hormones of the anterior pituitary gland (LH, FSH, TSH). The specificity of their physiological action depends on target organ receptors capable of distinguishing between them on the basis of small differences in the molecule such as the presence or absence of qn hydroxyl group. The use of these receptors in binding assays confers similar specificity on in vitro measurement. However, the use of receptor assays is limited to a relatively small number of substances -clearly they cannot be applied to non-hormonal materials with no target organ specificity, and equally they are of little value in determining important metabolites of biological compounds. For most purposes other types of binder have to be used - chiefly the naturally occurring circulating binding proteins and antibodies. Much of the discussion which follows will be devoted to the latter. 9.2.1. The basis of specific non-specificily
An antiserum to a given material will usually react, albeit less effectively, with closely related materials. There are 3 basic reasons why this may occur: first, a single homogeneous antibody population may react with a range of related ligands, each reaction having a different affinity constant; second, the antiserum may contain populations of antibody molecules one or more of which is directed to a site on the primary material which also occurs on a related material ; finally, the antiserum may contain antibodies directed to contaminants present in the ligand and which are also present in the related material (Fig. 9.1). Examples of these situations will be found in $9.2.2 and 9.2.3. These concepts are important to an understanding of the experimental finding of ‘parallelism’ or ‘non-parallelism’. If the difference between two related materials is solely in their K value with respect to a single homogeneous binder, then standard curves for these materials using a homologous tracer will be parallel. However,
Ch. 9
465
SPECIFICllY
A
B
C
Fig. 9.1. Different types of specific non-specificity. (A) A single population of antibody molecules reacts with two related materials. hut with a different affinity constant (i.e. one molecule is a good .fit’ to the combining site, the other is a less good fit). (B) A single population of antibody molecules reacts with an antigenic site which is common to two different molcculcs (e.g. the subunit of the glycoprotein hormones). (C) The antiserum contains populations of antihodies to different ligands; the non-specific ligand may then cross-react in the assay.
deviation from parallelism will be seen if the tracer is not homologous and if the mass of tracer represents a significant fraction of the total ligand in a given tube. This type of non-parallelism is found only at the lowest concentrations of unlabelled ligand and would be barely apparent from simple inspection of a standard curve. The second and third types of non-specificity - due to heterogeneity of binder or binding sites - can yield more dramatic non-parallelism. The principle by which this can influence the shape of a standard curve is best illustrated by an extreme case (Fig. 9.2). Clearly, the precise result obtained will vary with the set of reagents used and, in the case of radioimmunoassay, the distribution of antibody populations will vary considerably between different antisera. One might yield apparent parallelism between two cross-reacting materials, while another would yield non-parallelism. In a real system the 3 types of specific non-specificity may often co-exist. The analysis of the main factor involved - heterogeneity of K value at a single site or heterogeneity of sites - can then become very complex. However, gross non-parallelism, particularly at higher concentrations af unlabelled ligand. suggests site heterogeneity. This is of practical importance because the identification of this effect suggests that specificity may be improved by absorption of the antiserum. S,,/7,‘<1,,,
466
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
40
%TRACER BOUND
M
I
A-6
\
20
A B
Primary material W Tracer ,A.B
10
.cog
,032 .la .5 CONCEWRATION OF STANDARD
Cross-reading material Ant lbodles 2
B, A A
A
B
Fig. 9.2. Non-parallelism due to heterogeneity. The primary material and the tracer have two antigenic sites, A and B. The antiserum contains 2 corresponding populations of antibodies. Material containing site B only will produce the curve shown. It can never produce complete inhibition in the assay because antibodies of population A will always be unoccupied and can still bind the tracer.
9.2.2. Assessment o f spec fic non-specficity To test cross-reaction serial dilutions of the material in question are prepared and assayed: the resulting curve is then compared with that given by the standard material for which the assay was designed. The most common way of presenting the result is to compare the amount of the material under study which yields 50% inhibition of binding with the amount of standard giving the same inhibition, and then to express the potency of the material as a percentage of that of the standard (Fig. 9.3). For example, if the respective concentrations are 1 and 100, the potency is stated as 1%. A number of problems can arise with this approach: Concentration by weight versus molar concentration : very often potency is calculated on the basis of relative concentrations by weight. If the molecular weights of standard and cross-reacting material are identical this yields a satisfactory result. If they are not then the answer
Ch. 9
SPECIFICI IY
461
Cross - reaction 25%
100
80
% tracer bound
1
2
4 8 16 Concentrationof Standard
32
64
Fig. 9.3. Estimation of the percentage cross-reaction from the amounts of material required to produce inhibition in the assay. The curve on the left is for the standard. that on the right for the cross-reacting material. Note that this procedure cannot be applicd if the curves are non-parallel.
may be misleading. For example, if the molecular weight of the crossreacting material is 10 times that of the standard, and the potency by weight is given as lo%, then in reality the two materials are equal on a molar basis. To avoid this pitfall calculation as molar concentrations is always to be recommended. Variation with conditions of’ assay: the apparent potency of a cross-reacting material may vary if the conditions of the assay are changed. For example, if an assay is de-sensitised by increase of binder concentration, then a population of non-specific binding sites may become apparent which were of no significance at lower concentrations. More important, specificity can show great variation according to the nature of the tracer employed: an example of this is Sublrrr rndrx p 531
468
KADIOIMMUNOASSAY A N D K E L A l f D 1 ECHNIQUtS
given in 9: 2.2 and Fig. 2.1A, B. Similar though less obvious effects may result from the substitution of highly purified tracer ligand with less purified material. Calculation if' curves are non-parallel: if the curves yielded by standard and cross-reacting material are non-parallel, then clearly a calculation of potency made at 50% inhibition will be very different from that made at 10% (see Fig. 9.2). There is no simple answer to this problem, and non-parallel cross-reaction must be judged in relation to the intended biological or clinical use of the assay. Ariing from this is the question of how to judge parallelism. Though usually a subjective impression, this can be highly misleading: substantial differences can easily be missed, particularly at the extremes of the curve. There are two approaches to this problem. The first is to perform a logit transformation of the results (Q 7.2) and then to make a statistical comparison of the 2 linear regressions. The second is to calculate the apparent potency of each dilution, and then to examine the figures for any systematic deviation between standard and crossreacting material. In judging non-parallelism it is also important to exclude experimental artefact, such as that which can arise as a result of serial dilutions of the standard; indeed comparison of serial dilutions of a material with independently prepared dilutions of the same material is almost certain to yield some degree of nonparallelism (see Q 2.5). Relevance to physiological and clinical situations : the eventual significance of a cross-reacting material lies in the extent to which it is likely to interfere with the practical operation of an assay. This is best illustrated by taking 2 extreme examples. Radioimmunoassays for thyroxine (T,) often show a substantial cross-reaction (10% or more) with triiodothyronine (T,) ; however, as the circulating levels of T, are at least 2 orders of magnitude lower than those of T, this cross-reaction is unimportant in practice. At the other extreme, a radioimmunoassay for T, may show a cross-reaction of 1 with T,. For exactly the reason given above this could be highly critical; with such an assay, circulating T, would make a substantial contribution to the result obtained.
Ch. 9
SPECIFICITY
469
Choice of' materials to examine jor cross-reaction: it is clearly impossible to examine every conceivable biological material for crossreaction in an assay. In practice, a judicious selection has to be made from those substances whose physicochemical nature suggests that cross-reaction is likely. In the case of steroid hormones, the choice is relatively easy since the different structures and their relationship are well understood. Furthermore, a wide range of highly purified materials are available for study. A similar situation exists with respect to small peptides. However, with larger proteins the choice of materials and the interpretation of results may be considerably more difficult. Highly purified preparations are generally scarce, and relatively large amounts may be needed to exclude cross-reaction at the 1 and 0.1 'i:levels. The use of impure material can be misleading since this may contain related proteins which react in the assay, but do not feature in the definition of 'purity' derived from studies with another type of assay. A good example of this is seen with placental and pituitary glycoprotein hormones : striking discrepancies may occur between the results of biological assays and radioimmunoassays (see 8 11.4). Finally, with some substances which are biologically unique (or apparently so) there is no logical basis for selecting other materials for specificity studies: specificity must then depend on the physicochemical characterisation of the substance, together with its apparent behaviour under physiological conditions. 9.2.3. Methods for improving specificity With the naturally occurring binders (circulating proteins or cell receptors) there is little which can be done to alter the specificity of the assay itself, other than the use of the preliminary extraction procedures which have already been described ($6.2).With antibodies, by contrast, the variability and heterogeneity of the primary reagent offers many opportunities for the direct improvement of specificity. Potential approaches include the type of immunisation schedule used, manipulation of the conditions of the assay, and absorption to remove unwanted populations of antibody mofecules. Zrnmunisation scliedule: clearly the most important factor here is S u h p i l in&\ p 531
470
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
the nature of the immunogen. Non-specificity of type C (Fig. 9.1) (due to the presence of irrelevant materials in the ligand) can be obviated if it is possible to use pure ligand. This approach is limited by the availability of highly purified material and by the fact that low-level contaminants may nevertheless be highly immunogenic. Non-specificity of type B (Fig. 9. I ) (due to the presence of common antigenic sites on the ligand and related material) can be avoided if the immunogen used is a fragment of the ligand which does not contain the common antigenic site. A good example of this is seen with the glycoprotein hormones (LH, FSH, TSH and hCG) each of which consists of 2 subunits: the so-called a-subunit which is identical in all four; and the /%subunitwhich is different in all four and confers biological specificity. Because of the common a-subunit, assays based on intact hormone often show striking cross-reactions among all types. However, the use of P-subunit as immunogen (and often as tracer too) yields an assay which is specific for the individual hormone. The drawback to this approach is that the isolated subunit may have a tertiary structure different from that in the intact molecule, and therefore a different antigenic structure. This may explain why assays based on this principle are often less efficient, in terms of antibody affinity and sensitivity, than the less specific assays directed to the intact molecule. In contrast to types B and C, non-specificity of type A (Fig. 9.1) (due to reaction of a single homogeneous antibody population with a range of related antigenic sites) cannot be influenced by the choice of immunogen. Manipulation o f conditions ofassay :as already noted the specificity of an assay may vary with the conditions chosen. For example, nonspecificitydue to a population of low affinity antibody molecules may become insignificant if the antiserum is used at very high dilutions. Non-specificity may also result from the use of short incubation disequilibrium assays ; binding of a ligand and cross-reacting material may be virtually equivalent under these circumstances, and only as equilibrium is approached does the relatively greater dissociation rate of the cross-reacting material yield a relative excess of bound
Ch. 9
SPECIFICITY
47 1
primary ligand. It is essential for this reason to re-check the specificity of an antibody under the exact conditions used in the assay. The specificity can also be influenced by the nature of the tracer; if this consists of highly purified ligand then some of the deficiencies of the antiserum may be obviated. An example of the dramatic effect of using different tracers with the same antiserum is shown in Fig. 2.1A and B. Absorption to remove unwanted popidations of antibody: nonspecificity of type B and C can be eliminated by prior absorption of the antiserum - in the case of type B with a purified preparation of a fragment of the molecule containing the common antigenic sequence; in the case of type C with a preparation of the contaminant material. On occasion the 2 approaches are combined. For example, in the earlier days of gonadotrophin assays it was customary to absorb antisera to hFSH with a preparation of hCG. This had 2 effects: first, to remove populations of antibodies directed towards the common a-subunit ; second, because of the close similarity between LH and hCG, to remove populations of antibodies directed towards LH which arose as a result of the presence of LH as a contaminant in the preparations of FSH used as immunogen. Elimination of type B non-specificity by absorption can also be used to render an assay specific for a ‘neo-antigen’, that is to say, an antigen which is expressed in a fragment of the molecule but not, because of the differences in tertiary structure, in the intact molecule. This is well illustrated by the assay of fibrinogen degradation products (Fig. 9.4). An antiserum to the terminal fragment D contains antibodies specific to the fragment itself, and in addition antibodies which will react with intact fibrinogen. Absorption of the antiserum with fibrinogen removes the latter population and thus yields an assay specific for fragment D with no interference by the parent molecule. In practical terms there are 2 possible approaches to the absorption of an antiserum: (1) simple addition of the cross-reacting material, and removal of the resulting immuno-precipitate by centrifugation (this has the disadvantage that soluble complexes may remain due to the ‘prozone’ phenomenon) ; (2) addition of the cross-reacting Suhpir rnrlcr p 53/
412
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
n
45
z
3
s ap
1
t
Fibrinogen
30 15
1.2
20
320
5m
80,m
nglml
Fig. 9.4. Improving specificity by absorption of an antiserum. In this case the antiserum was raised to pure fragment D (FgD), one of the terminal degradation products of fibrinogen, and then absorbed with fibrinogen. Used with a tracer of '2sI-fragment D this antiserum shows no cross-reaction with intact fibrinogen at the concentrations tested and relatively little cross-reaction with fragment E (FgE). Since fragment D is part of the fibrinogen molecule, this assay is specific to a 'neo-antigen', i.e., one that is revealed during the course of proteolytic destruction of fibrinogen. (From data kindly supplied by Dr. Y . 9. Gordon.)
material in insoluble form, for example, in the case of a protein, after cross-linking with glutaraldehyde. A more general technique with wider applications is the use of antigen coupled to a solid phase such as cellulose or Sephadex. The solid phase can then be added batchwise to the antiserum or used in the form of a column to which the antiserum is applied and eluted. Separation technique ,for bound and ,free ligand: the nature of the separation procedure can affect the specificityof an assay. For example, it has been shown that the cross-reaction of 5a-dihydrotestosterone can be substantially reduced if ammonium sulphate is used for separation rather than dextran-coated charcoal (Tyler et al. 1973). The mechanism of this phenomenon is uncertain.
9.3. Non-spec f i c non-specficity This term refers to interference in an assay by factors other than those which can be clearly identified by their physicochemical
Ch. 9
413
SPECIFICITY
similarity to the ligand. In its most familiar form this is manifest in an assay of an unknown sample by the measurement of an apparent concentration of ligand which is elevated by comparison with the concentration measured by other techniques. On occasion, however, non-specific effects may lead to a reduction in the apparent concentration when compared with the results of other methods. There are 4 basic mechanisms which may lead to non-specific nonspecificity : the presence in the sample of material which interferes with the binder-ligand reaction ; variations of 'blank' values in samples ; destruction or sequestration of binder or tracer; and destruction or sequestration of the unlabelled ligand. 9.3.1. Presence qf'materia1.Y which interfere with the binder-ligand
reactioii There are a variety of materials which will interfere non-specifically with binder-ligand reactions. Examples include the presence of large amounts of highly charged molecules such as heparin; high concentrations of low molecular weight materials such as salt and urea; and conditions of acid or alkaline pH. An example of the effect of acidity is shown in Fig. 6.2. 9.3.2. Variations of'bluiik values in samples
The nature of the fluid used in the assay can have a striking effect on the blank values. For example, with separation by organic precipitation the assay blank may be considerably higher in tubes containing whole serum than in those containing diluent with addition of carrier protein (standards) ; under these circumstances the values given by unknowns will be lower than the true value (Fig. 9.5). Furthermore, the assay blank value may vary between different samples of the same fluid. This could be avoided by carrying out an assay blank determination on every sample (i.e. replicate tubes with sample and tracer but no antibody); such a determination would be very tedious to perform and in practice is rarely necessary. Subject Inder p 531
474
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
100
80
7
actual reading of sample (2) true reading of sample (4)
60
%tracer bound 40
20
1
2
4
8
16
32
Concentration of Standard Fig. 9.5. Model curves showing the effect of a variable blank value on the specificity of a binding assay. If the blank value for the sample is higher than that of the standard, then the result obtained will be artefactually low. This problem is largely avoided if the composition of the standards and unknowns is identical.
9.3.3.Destruction or sequestration of binder or tracer In most systems the binder is relatively stable and is not subject to irreversible damage. The same, however, does not apply to the tracer. At least 3 mechanisms exist which can grossly affect the performance of the latter - enzymic destruction, absorption to surfaces, and binding by endogenous antibodies. Enzymic destruction is an important factor in assays using iodinated protein hormones and in particular small peptides such as vasopressin and angiotensin. These are highly susceptible to hydrolysis by enzymes present in normal plasma or serum, a process which leads to a reduction in binding which may be difficult to distinguish from that produced by the presence of unlabelled ligand (Fig. 9.6). Absorption to surfaces, such as those of glass tubes, is an effect which is also commonly seen with small
Ch. 9
415
SPECIFICITY
80 70
8 60
4
40
-
I
$ 30
20 10
I
’8
0I ’
50
I
I
I
25 12.5 6.3 CONCENTRATION
I
1
I
3.1 1.5 0.8 X PLASMA (sb)
I
0.4
Fig. 9.6. The effect of enzyme destruction on a tracer. [‘251]oxytocin is incubated with an antiserum to oxytocin and varying concentrations of oxytocin-free human late pregnancy plasma (horizontal axis). After 24 hr there is virtually no bound hormone in those tubes containing the highest concentrations of plasma. The loss of immunoreactivity is due to destruction of the tracer by the placental enzyme, oxytocinase.
peptides and which can occur with steroid hormones. Under these circumstances the surface may ‘compete’ with the binder for tracer or unlabelled ligand and thus produces effects which vary widely with thenatureoftheother constituents in the assay tube. Finally, biological samplesmay contain antibodies to the ligand - for example, the serum from a diabetic patient treated with insulin. Depending on the system used for separation, this can lead to grossly discrepant results because of competition for the tracer between the endogenous and added antibody. 9.3.4. Destruction or sequestration of unlabelled ligand The effects described above may also apply to the unlabelled ligand. This applies particularly during the process of collection, preparation Suhjecl indexp. 5 J I
476
RADIOIMMUNOASSAY A N D HELATED TECHNIQUES
and storage of the sample. Small peptides, for example, may be rapidly destroyed by endogenous enzymes and are very liable to irreversible absorption to the surfaces of syringes and tubes, most notably when the latter are made of glass.
9.3.5. The detection and elimination of non-speciJi’crton-specficity The traditional method for detecting non-specificity of any type is the examination ofparallelism between dilutions of sample (unknown) and standard. But, as with specific non-specificity, this can be misleading and should not be relied upon to guarantee identity. Other approaches should also be examined. The most satisfactory approach to the elimination of non-specific non-specificity is to ensure that the overall composition of the standard is as near as possible identical to that of the unknown. For example, if the fluid examined is serum and standards are prepared in ligand-free serum then problems are unlikely to occur. However, difficulties arise with fluids of highly variable composition, such as urine, or, more important, when a source of ligand-free fluid is not readily available. There are two potential sources of ligdnd-free fluid. The first and best is from a subject in whom none of the ligand is present: for instance, in the case of a drug, from an untreated patient; in the case of a hormone, from a subject in whom the gland of origin has been removed (however, see 9: 6.1), or in whom production has been inhibited by an appropriate drug (e.g. dexamethasone suppression of ACTH); in the case of a placental product, from a non-pregnant subject. The second approach is to use fluid from which all endogenous ligand has been removed by absorption; a typical example is the removal of insulin or thyroxine from serum by pretreatment with charcoal. The problem with this approach is the possible absorption of other factors which can interfere non-specifically in the system. As a result, the so-called ‘hormone-free’ serum is not strictly comparable with fresh normal serum and can yield misleading results. If total comparability between sample and standard cannot be guaranteed, and in particular if a reliable source of ligand-free fluid
Ch. 9
477
SPECIFICITY
50
r
40
30 % BOUND
20 10
1
3 4 5 6 7 8 9 FRACTION NUMBER iThin-Layer Chromatography )
2
10
Fig. 9.7. Thin-layer chromatography of an extract of urine after adsorption to and elution from glass beads (see 6 6.1.2 and Table 6.1). Each segment from the plate was extracted with 1 ml60':; (w/v) aqueous acetone. The extract was dried, dissolved in an aqueous buffer solution and incubated with 0.05 ng [12sI]oxytocin and a rabbit antiserum to oxytocin at a final concentration of 1 in 4000. L O . urine with no added hormone; C - H , urine withadded syntheticoxytocin; - 0 , synthetic oxytocin alone run on the same plate. Note that the elution patterns of endogenous material and recovered exogenous material are virtually identical. Neither, however, is identical to synthetic oxytocin alone, indicating some alteration of the material after addition to and extraction from urine.
is not available, there are a number of additional approaches to the elimination or at least the identification of non-specificity. At the level of the assay itself, careful attention to technique can eliminate many potential sources of error. Factors such as absorption of tracer or unlabelled ligand to the surface of tubes can be identified by adding and removing tracer ligand from the tube, and estimating the proportion of counts which remain. It is usually possible, by trial and error, to select a tube which does not show this phenomenon. Enzymic destruction can be a t least partly eliminated by the incorporation in the incubation mixture of enzyme inhibitors such as Trasylol. Extraction and concentration of ligand (Ch. 6 ) can serve
478
RADlOlM M UNOASSAY A N D RELATED TECH NlQUES
to transfer material from an ill-defined environment such as urine or a tissue extract into a well-defined aqueous buffer system. As noted, however, extraction itself may yield non-specific effects. Physicochemical studies may also be of value to determine the precise nature of material measured in an assay. For example, the demonstration that the material inhibiting the binder-ligand reaction has the same properties as the authentic material in several systems of chromatography can greatly enhance the confidence in the results (Fig. 9.7). Finally, and perhaps most important, it is critical with any system to examine the biological specificity of the results. In other words, that the answers obtained with the assay are comparable to those which would be expected from other information on the ligand in question. For example, it is possible with many hormones to obtain an indirect estimate of their circulating levels from a knowledge of their half-life in the circulation and of the amount of administered exogenous material which will produce maximal endorgan response. If the apparent levels measured in an assay are greatly in excess of indirect estimates, then it is likely that non-specific factors are operative. Similarly, the estimated levels should show appropriate variation under physiological or pathological conditions; for instance, material recorded as ‘basal levels’ of ACTH should be reduced or disappear following administration of corticosteroids.
CHAPTER 10
Characteristics of binding assays - precision
10.1. Definitions The aim of a binding assay is to give the ‘true’ concentration of a ligand in a biological fluid. In practice, however, the actual result will diverge from the ‘true’ result because of imprecision on the one hand, and inaccuracy on the other. As far as this chapter is concerned, it is worthwhile to define these two terms, because they are often confused (see also Fig. 10.1). ‘Precision’, often referred to as ‘reproducibility’, is a
@$I
prmsian
,w preelllon
. . . --. . . a a
ESTIMATED LEVEL OF LIGAND
.a me.
a.
..a .a
.a a
+*-- .-+.+-
TRUE LEVEL OF LIGAND
Fig. 10.1. Model scatter diagrams illustrating the concepts of precision (variability of repeated determinations) and accuracy (degree to which a measured concentration corresponds to the true concentration of a substance). 479
Suhlect index p 531
480
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
measure of the variation observed between repeated determinations on the same sample. ‘Accuracy’ is the degree to which the estimate approximates the true value: this is related to the specificity of the assay. Thus, an assay in which the endogenous ligand reacts differently to the standard and tracer ligand will always (unless correction factors are incorporated) be inaccurate, however precise it may be. Accuracy can therefore be thought of in broadly the same terms as specificity (see Ch. 9). This section will be concerned primarily with precision the factors affecting precision, and the means for assessment and improvement.
10.2. Factors affecting precision These can be considered under 2 headings: errors due to the nature of the reagents and assay design, and errors arising in the actual operation of the assay. In practice, of course, the two are very closely related. 10.2.1. Errors in the reagents and design of the assay 10.2.1.1. Errors in the primary reagents Imprecision due to some basic fault in the reagent solutions (standards excepted) is very rare. For this reason the common practice of comparing the precision of different assays, and particularly assay kits, on the basis of the difference in absolute count rates or percentage bound between replicates is largely irrelevant. All other things being equal, the major factor affecting this difference is the technical skill of the operator (or machine) which is unrelated to the reagents themselves. However, a specific error which can, but should not, arise is inadequate mixing of the primary reagents; this can lead to heterogeneity of the solution and of aliquots subsequently dispensed from this solution. Heterogeneity of this type is often seen when frozen material is thawed, or when lyophilised material is redissolved. 10.2.1.2. Errors due to the separation procedure The nature of the procedure used for separation of bound and free ligand can have a
Ch. 10
PRECYSION
48 1
considerable effect on the precision of an assay. Certain types. notably the earlier used coated tube methods (9 5.3.6.1), had an inherent problem ofreproducibilityand demanded multiple replicates if the best results were to be obtained. Less well recognised, because it is rarely the subject of comment, is the influence of the assay blank value on precision, an effect which will apply with any separation technique. The assay blank value, as all other parameters, shows variation: some of this is due to technical errors, but an important part is due to variation between different unknown samples. Let us assume that the variation attributable to this factor is equivalent to 10% of the counts in the blank tubes. This will have almost no effect on the precision of an assay with an assay blank of 1% (variation 0.1%) and a zero standard of 5 0 x . However, it will have a very substantial effect on an assay with a blank of 157(, (variation 1.5%) and a zero standard of 30%. For this reason separation procedures characterised by high blank values (chiefly precipitation of the bound fraction by, e.g., ethanol or ammonium sulphate) should not be used in assays with low zero standard values (less than 40%of total counts). The basic design of a separation procedure can also have an important influence on precision. This should be so arranged that the exact amount of the separating agent is not critical: for example, with an organic solvent added in a nominal volume of 1 ml, the addition of 0.8 ml or 1.2 ml should make little or no difference to the results. If this volume is non-critical over a reasonably wide range, one potential source of error is eliminated (Fig. 10.2). 10.2.1.3. Errors due to diseguilihrium Traditionally most binding assays are incubated until equilibrium is obtained. However, there are circumstances in which it may be desirable to separate bound and free ligand before equilibrium is achieved ($ 8.2.4.). This is a potential source of imprecision: if the time taken for separation represents a substantial part of the total assay time, then there may be significant differences between the first and last tubes of an assay.
Probably the major error which can arise from the primary reagents in an assay is in the standards. 10.2.1.4. Error dire to starzdurds
Suhjrtr eidrr p 531
482
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
80
60 %tracer bound
40
20
1
2
3
4
5
Amount of separating agent added
Fig. 10.2. Diagrammatic illustration of how precision can be affected by a separation procedure. With system B small variations in the amount of separating agent added would have a substantial effect on the observed distribution of bound and free ligand. With system A there is a long plateau where the amount of separating agent is noncritical, and even substantial errors would not affect the result. Systems of type A are much to be preferred.
Some important factors in the preparation of these have already been described (9 2.5, Table 2.1) but certain general rules should be emphasised again : (1) Individual sets of standards should always be prepared as aliquots from a larger pool made at each concentration. Not only is this technically convenient, but also it avoids the errors which may arise if the standards have to be prepared separately for each assay. (2) Each standard concentration should be prepared independently using the solution of master standard. The use of doubling dilutions is much to be deprecated; first an error made at any one step of the procedure will affect all subsequent steps; second, absorption of the
Ch. 10
483
PRECISION
lower concentrations of standards to the walls of tubes can lead to a progressive reduction in the actual amount transferred - an example of this is shown in Fig. 10.3. (3) Standards should always be prepared in the largest possible batches, assuming that the material is stable under the storage conditions available. Because every step of sampling and dispensing has an error, no two sets of standards prepared on different occasions from the same primary materials can ever be identical. Frequent changes of standards will inevitably lead to an increase in the longterm assay coefficient of variation (see below) and may also be associated with a persistent 'drift' of quality control values. In practice, the aim should be to prepare sufficient sets of assay standards to last for at least 1 year and preferably longer. These batches should, in turn, be prepared from a 'master standard' - a single solution of
%'=I
-
AFP BOUND
18
78
312
12H)
Moo
STANDARD AFP ( ng I ml )
Fig. 10.3. Standard curves for a preparation of pure a-fetoprotein (AFP) serially diluted in the presence and absence of carrier protein (2 mg/ml bovine serum albumen). With no protein there is a progressive loss due to absorption on the tubes and consequent incomplete transfer leading to gross non-parallelism. (From data kindly supplied by Dr. M. Al-Awqati.) Subject indexp. 531
484
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
concentrated standardwhich is prepared with great care and aliquotted to yield sufficient material for several years’ operation of an assay. All too often, unfortunately, this counsel of perfection proves impractical. In some cases insufficient material is available at any one time to permit the dispensing of large batches. In others it is vitiated by the replacement of one standard preparation by another. This is particularly common with materials which are obtained from natural sources and which are relatively difficult to purify - a notorious example is provided by the human gonadotrophins. (4) Sets of standards prepared by freeze-drying (as in kits) have the disadvantage of 2 additional sources of error - at the initial dispensing and when they are reconstituted before use. Both steps must be quantitative. By contrast, deep-frozen standards avoid both these steps, the only error being in the final dispensing into the assay tube. (5) Protein standards should always be prepared in solutions containing a carrier protein (e.g. BSA, whole serum) because of the inherent instability of proteins in very dilute solutions. 10.2.I .5. Errors at differenlpoints 011 the standardcurve The precision of a binding assay varies according to the dose-level measured, being at its best in the central part of the standard curve and declining at the extremes (Fig. 10.4). The design of an assay to ensure that the values of greatest biological or clinical interest are measured with the greatest precision has already been described (9; 8.4). Here it should simply be repeated that precision is critically dependent on this design, and that this must be taken into account when precision is compared for assays of different designs. A related question is whether or not the absolute precision of determination varies at different parts of the standard curve - ‘absolute precision’ being defined here as the variation between replicates. There is, in fact, little evidence to suggest that it does, apart from that variation which can be attributed to the errors of different count rates.
Ch. 10
485
PKEClSlON
+3 +2 +1
LOGIT ( Y 1 0 -1 -2
-3 10
100
loo0
STANDARD CONCENTRAT I ON
Fig. 10.4. A standard curve linearised by logit transformation. showing the confidence limits to the curve. Note that these are much broader at the extremes and that precision is correspondingly reduced for these values.
10.2.1.6. Counting errors Counting errors are an important source of imprecision in a binding assay using a radioactive tracer. However, it is frequently and quite wrongly assumed that they are virtually the only source of variation when in fact the technical errors described in 4 10.2.2 are considerably more important in a well-designed assay. Counting errors are of 2 types - those due to the equipment, and those intrinsic to the counting process itself. Errors attributable to the equipment are very rare: failures in up-to-date electronics, when they occur, are usually total and very obvious. Large fluctuations in the power supply may, however, give rise to problems. Equipment errors may also arise, particularly with well-crystals, if the volume of fluid counted varies widely; but in most assays a fixed volume of fluid or precipitate is examined, and counting geometry is therefore equivalent for all tubes. The intrinsic counting error is much the most important source of variation and applies S~&YI
,ndc\ p 5.31
486
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
regardless of the nature and quality of the equipment. Radionuclide disintegrations are random events and therefore 'follow a Poisson distribution for which the standard deviation is proportional to the square root of the total number of counts accumulated. In practical terms this implies that if repeated counts are performed on the same radioactive sample, then 95'%, (i.e. f 2 S.D.) of the values will lie within the range given by the mean plus and minus twice the square root of the mean. Thus the error of any given number of counts can be calculated directly without recourse to experiment. The implications of this for a selection of different total counts are shown in Table 10.1. These figures demonstrate very clearly that the gain in precision between 10,000 counts and 100,000 counts is small, and that between 2000 and 10,000 is not large relative to other errors in the assay. Furthermore, it should be emphasised that the counting error is independent of other errors and is not, therefore, a simple addition to these. For example, if technical errors are responsible for a 2% variation and the number of counts accumulated is 10,000, the total error is only 2.2%;for 2000 counts it is 3%. On this basis the precision of most assays will be perfectly satisfactory if the range of counts over the standard curve lies between 2000 and 10,000. Only rarely will a system be of such basic technical excellence TABLE 10.1 The theoretical counting error for different numbers of total counts; the error is the square root of the total counts expressed as a percentage of the total counts
Total counts 100,000 50,000 20,000 10,000 5.000 2,000 1,000
500
Error (%) 0.3 0.4 0.7 1
1.4 2.2 3.2
I Subject index p. 531
Ch. 10
PRECISION
487
that anything is gained by the accumulation of larger totals. Long countitig times and high numbers of counts are no suhstitute,for correct perjormance o j o h e r steps of an assay. 10.2.2. Errors in the technical operation of'the assay For a well-designed system using adequate reagents, technical errors are much the most important source of variation. Technical errors can be divided into major and minor. Major errors include failure to add reagent or sample to a tube, misidentification of samples, or serious faults in the calculation of results. Such errors are usually, but not necessarily always, recognised by the operator. The relevant samples or complete assay is repeated, and the results areusually excluded from any calculation of 'between-assay' variation. More important, because they are less easily identified or corrected, are the minor errors which occur because of the inevitable variability in the repetitive sampling and dispensing operations of which a binding assay consists. The errors arise both in the equipment used and in the operator using it. (1) Equipment errors: equipment currently used for repetitive sampling and dispensing is much superior to that which was available in the early days of radioimmunoassay, which consisted chiefly of non-disposable glass pipettes. There are 2 main types: the handheld sampler/dispenser with a piston and spring return, and the motor driven syringe (see Appendix I). If well maintained and operated correctly, the best models of these are reproducible to within 0.5% or better; indeed, it is often difficult to measure the error because it is equivalent to or less than that of the eventual system of detection (e.g. an analytical balance). However, the excellence of the equipment should be regularly examined and not simply assumed. Maintenance and cleaning is also essential traces of dirt around plungers and pistons can lead to discontinuity of action and serious dispensing errors. (2) Operator errors: the vast majority of minor errors in a radioimmunoassay can be firmly and exclusively attributed to the operator. Two well-known facts make this quite clear. First, there is often a ~
Suhlerr rnder p 531
488
RADIOIMMUNOASSAY AND RELATED TECHNIQUES
notable difference in the precision obtained by two or more operators using the same reagents on the same day. The difference is at least partly related to experience: the novice can rarely equal the expert. Second, precision will usually decline as increasing numbers of tubes are set up (a fatigue factor). As a rule of thumb, the maximum number of tubes set up manually in a day should not exceed 500 per operator. The exact nature of the errors involved are usually difficult to identify but may include failure of adequate mixing, leaving drops of reagent on the sides of the tube or the tip of a pipette, and variability in the decanting or aspiration of supernatants. It is for these reasons that operator.error can apply equally to the use of both hand pipettes and manually operated motor-driven pumps.
10.3. Methods for monitoring the precision of’ a binding assay Precision has already been defined as ‘a measure of the variation observed between repeated determinations on the same sample’. This is an operative definition which clearly describes the actual procedure used. In this section some of the practical aspect of monitoring precision will be discussed, together with other approaches to the quality control of binding assays. 10.3.I . Prrparatiorz ofquality-control materials,formorzitoringprecision The sample or samples used for monitoring precision should be used as follows : (1) They should consist of pooled samples containing endogenous ligand in the biological fluid for which the assay is intended (e.g. serum). Samples which are made up by addition of purified ligand to ligand-free fluid or diluent are not strictly speaking quality controls - rather they are standards. (2) The pooled samples should be available in substantial quantities, and should be aliquotted in such a way that the same basic pool will suffice for many months or even years of operation. (3) The pooled samples should be stored under conditions in which
Ch. 10
PRECISION
489
they are known to be stable, usually at the lowest temperature available in the laboratory (e.g. - 70°C). Lyophilisdtion is not usually desirable for the reasons already given in 9 10.2.1.4. Furthermore, freeze-dried material stored for long out of refrigerators is very likely to take up moisture unless it is extremely carefully ampouled. For many materials the demand for proof of stability is, of course, a counsel of perfection: to determine whether a substance is stable under given conditions for a period of years requires an experiment lasting several years, and a reference system which is definitely stable. The latter purpose is served by the use, where available, of 'biological pools' (see below). (4) Optimally, the pools should be chosen in such a way that their concentration represents high, medium and low values in the assay. This provides an on-going check of precision at different parts of the standard curve (see Fig. 10.4). ( 5 ) Ifpossible, samples for quality-control pools should be collected from subjects under defined conditions: for example, a pool for plasma oestriol could be derived from normal subjects at the 36th40th weeks of pregnancy. The great advantage of this 'biological pool' is that it can always, if necessary, be repeated. Materials may easily alter with prolonged storage in vitro ; physiology, however, does not change. But there are many situations in which 'biological pools' are not readily available - for instance in an assay for the blood levels of a drug. Under these circumstances quality-control pools can be prepared from assayed samples. If the range of ligand concentrations measured is 20-80 (arbitrary units), then 3 appropriate pools can be made from the residue of samples reading 2 M 0 , 40-60 and 60-80. The problem with such pools is that the mean value may not easily be reproducible on another occasion. (6) All the concepts presented here can equally be applied to the preparation of quality-control pools for use in several laboratories, or as part of national or supranational or commercial quality-control schemes. Supply problems usually dictate that such material cannot be used in every assay; it can, however, be used for the calibration of locally prepared pools. The concept of targeting numerous different Bibjerr rnder p 531
490
RADIOIMMUNOASSAY AND RELA1 ED TECHNIQUES
laboratories to the achievement of identical or near-identical results is of prime importance. Since clinical interpretation must depend on previous experience with an assay, it is of immense value if the ‘experience’ of a small laboratory can effectively embrace that of a much larger group. 10.3.2. Methodsfor examining precision using quality-control materials When suitable pools have been prepared, aliquots are examined to ascertain ‘within-assay’ and ‘between-assay’ variation. For withinassay variation a series of determinations are made in a single run, treating the pool identically with samples. Thus, if the samples are normally estimated in duplicate a series of duplicate tubes are made up from the pool; the number of tubes should be at least 20 (i.e. equivalent to 10 samples in the assay) and they should be spaced out between other samples in the assay to take account of the probability of drift. The mean and standard deviation of all the estimates on the pool are calculated, and the result expressed as a coefficient of variation (the standard deviation expressed as a percentage of the mean). For most systems this figure will be found to lie between 3 and 8%. As a control parameter, within-assay variation is of relatively limited value. There are few practical applications in which all samples are examined in one assay; furthermore, the presence of grouped control samples is likely to put the assay and the operator on their ‘best behaviour’. Between-assay variation is much more important in practical terms. Estimations are made on aliquots from the quality-control pool in every assay run. For this purpose the pool is treated as a single sample, i.e. if samples are assayed in duplicate, the control should also be examined in two tubes. The practice of running larger numbers of replicates of the control will give a falsely optimistic impression of precision. When a series of observations have been accumulated in successive assays there are a number of ways in which the results can be presented. (1)As a coefficient of variation as described above. For most systems this will lie between 8 and 20%, assuming that the figure
Ch. 10
49 1
PRECISION
applies to a long series of assays, performed by different operators, and allowing for several batch changes in all the primary reagents (antibody, tracer, standards and separation system). Lower estimates are frequently given, but will almost invariably be found to refer to a study limited to a single batch of reagents and one operator: the real practical value of this estimate must take into account such changes. (2)As a graph showing the serial values (Fig. 10.5). This is of
'I 2
6
I
I
I
I
10
14
18
22
Assay numbm
Fig. 10.5. Serial assay values for high level and low level quality-control pools in routine radioimmunoassays for hPL. The means and confidence limits to these values are shown. With a poorly executed assay both pools gave results well below the cumulative range; this assay was rejected. After replacement of the low level pool. the values for this pool showed a marked decrease. However, because the results of the high level pool remained stable it was assumed that this change represented a true shift and was not due to poor control of the assay. Suhjrrr inricv y . 531
492
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
great value in the individual laboratory since it is simple to plot and easy to comprehend. The value is further enhanced if the limits to the control pool, expressed as the mean and coefficient of variation, are also included. It should be placed in a prominent position where it can be easily seen by all concerned with the assay. It serves to indicate not only when a given assay run is out of control but also, if one pool is changed, whether or not an observed shift is affecting the assay as a whole (Fig. 10.5). (3) As a ‘cusum’ plot (Fig. 10.6). The ’cusum’ is the cumulative sum of the differences of serial values from the progressive mean of that series. This is rather more difficult to construct than the simple
+2
+1
Cusum 0
2
a
10
14
18
Assay number
Fig. 10.6. Serial values for a quality control pool in routine radioimmunoassays for hPL (below) and the ‘cusum‘ of these values (cumulative sum of the difference from the mean) (above). Note that the cusum plot is a much more sensitive indicator of a change in conditions than is the plot of serial values.
Ch. 10
PRECISION
493
graph of serial results, but is in many ways more instructive since it reveals very clearly a systematic but minor deviation in the assay. 10.3.3. Other methods of' monitoring precision The between-assay variation assessed from quality-control pools is much the most important means for assessing the precision of a binding assay. However, there are many other parameters which reflect precision in varying degrees and which should be routinely examined in any assay: (1) Total counts: counts and counting errors are considered in 5 10.2.1.6. (2) Blank wlues: the zero standard and assay blank should always be noted and if they fall outside limits set by previous experience then the fault must be identified before further runs are performed. A common cause of deviation is the use of outdated tracer which in many systems leads to a fall in the zero standard, and in some systems leads to a simultaneous increase in the assay blank. (3) The effective standard dose at 50% response. (4)Slope of the standard curve: this can only be usefully estimated if the curve is linearised (e.g. by the logit transform, see 0 7.2). The information it gives is rather similar to that of the blank values, i.e. if the zero standard falls and/or the assay blank rises then the result is a decrease in slope. (5) Differences between replicates: it is good practice to place an arbitrary upper limit on the difference between replicates. above which the result is ignored and the sample repeated. A typical figure for this limit would be 5% of the total counts in the system. The major purpose of this control is to eliminate figures one of which is probably grossly in error due to a fault in pipetting or counting. The fact that such figuresoccur not infrequently in both manual and semi-automatic assays is an excellent argument for the routine performance of replicates on every sample, standard and quality control. (6) Assay mean: the mean of all results in a given assay can be used as a quality control parameter. It assumes that on any one occasion the composition of the samples examined will be similar,
494
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
i.e. will contain similar numbers of high, normal and low results. This will usually apply only if there are relatively large numbers in the assay run; with small numbers (50 or less) the parameter is likely to be unstable and therefore unsuitable as a quality control. An alternative is the use of a ‘restricted population mean’ - samples for the assay are chosen on the basis of criteria which will yield a homogeneous population. In practice this is likely to be timeconsuming and demands patient information which is often not available to the laboratory.
10.4. Methodsfor optimising the precision o j a binding assay The means for optimising the precision of a binding assay are very largely implicit in the factors which have already been described as affecting precision. However, it is worthwhile to re-emphasise certain points. First, the importance of monitoring precision by quality control - the principle feature of which should be the inclusion in every assay of aliquots from an appropriate pool. If there is no quality control, then there is no assay, because the operator has no way of judging whether his results are of value or not. Second, the importance of good assay design, without which the best technician will obtain poor results regardless of the number and range of quality control parameters. Third, and finally, the importance of adequate training of the operator to ensure that simple technical errors are reduced to an absolute minimum. Automation, as discussed in another chapter, is some substitute for the operator, but will never be a complete substitute.
CHAPTER 11
Characteristics of binding assays - relation to other types of assay
11.I. Dclft’nition The perfect assay for a material in a biological fluid would consist of the extraction of the material at 100% purity and with 100% recovery, followed by weighing on a perfect balance. The result would be a true value of concentration as weight by volume which, if the molecular weight were also known, could be expressed as molar concentration. No known assay even approaches this degree of perfection. In reality all assay work consists of a series of compromises and yields results which may approximate the true value but can never be guaranteed as so doing. In the final analysis the principle criterion is perhaps that of practicality - that the results should be useful in the biological or clinical context for which the assay is intended. This criterion serves to free the assayist from the concept that any one method is intrinsically superior to all others - for example, the inferiority complex which many ‘radioimmunoassayists’ develop with respect to biological assay of hormones, when in fact the latter may deviate drastically from a real value because of the existence in a biological fluid of related but often unidentified materials. This section will review the relationship between binding assays and other types of assay and will attempt to avoid any a priori assumptions as to which method has the greatest intrinsic merit. It is convenient to divide the discussion according to the nature of the binder employed.
495
Suhjwr irjdrr p . J3I
496
KADIOIMMUNOASSAY A N D RELATED TECHNIQUES
I I .2. Receptor assays Assays based on receptor-binding proteins (see 9: 4.3) were introduced with much eclat because of their supposed close relationship to biological assay. Intuitively this would seem to be the case though it has only rarely been demonstrated experimentally. However, certain reservations must be made about this relationship, leaving aside here the question of practicality which in general is poor though no more so than that of most true biological assays. The first reservation concerns the fact that some receptors may apparently be different from those responsible for biological activity (Birnbaumer and Pohl 1973). Second, the receptor may be irrelevant to any known biological activity - an example is that which has been prepared from fat cells and which is capable of binding oxytocin. Third, the possibility exists that a receptor might bind fragments of a molecule which are not biologically active - for instance, if part of the intact molecule binds to the receptor and function then depends on cooperative reactions with other parts of the intact molecule. Fourth, the receptor may not bind a large pro-hormone molecule the measurement of which, in terms of physiology, is likely to be just as important as the active molecule. Finally, a receptor assay is unlikely to measure metabolites of an intact molecule - metabolites which, as in the case of the steroid conjugates, may be of considerable practical significance. All these factors serve to emphasise that assays using cell receptors are not necessarily and intrinsically superior to assays based on other types of binder.
I I .3. Assays using circulating binding proteins To some degree the use of a circulating binding protein, which is a naturally occurring material, might be considered an extension of a receptor assay. Functionally, however, they are quite distinct. The biological activity of a molecule is unrelated and largely irrelevant to the means by which it is transported to the receptor site, and this is reflected in the fact that the specificity of circulating binders is very
Ch. 11
RELATION TO OTHEK TYPES OF ASSAY
491
different from that of cell receptors. Assays using circulating binders are not, therefore, alternative biological assays. This must not be confused with assays .for circulating binders which, if they actually use binding of ligand as the response parameter, are true biological assays. Given these observations. the relation of this type of assay to others can essentially be described in terms of specificity. This has already been discussed (4 4.4). In general, the circulating binders tend to bind groups of materials - for example, oestrogens and androgens by sex hormone-binding globulin. The broad specificity, when compared with that of a radioimmunoassay, almost always dictates the need for preliminary extraction and purification of biological fluids, a factor which, together with supply problems, tends to limit their practical application.
11.4. Immunoassays The immense practical success of immunoassays and in particular of radioimmunoassay has led to the frequent criticism that they do not measure the functional activity of a molecule. This is undoubtedly true. Because the antibody-combining site is usually considerably smaller than the molecule to which the immunoassay is directed, only a small section of the molecule is actually measured and this section may be remote from the part responsible for biological activity (Fig. 1 1.1). The possibility therefore arises that the immunoassay might measure inactive fragments and thereby overestimate the Functional site
Fig. 1 I . I . Dissociation of immunological and biological activity. Splitting of the molecule has destroyed the functional site while leaving the antigenic site intact.
498
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
concentration of intact, active molecules. This phenomenon is usually referred to as the 'dissociation of immunological and biological activity'. Several examples of this situation have been investigated in detail. Amongthe most familiar isadrenocorticotrophin (ACTH). The ACTH molecule contains 39 amino acids of which the first 24 are responsible for biological activity and are similar in all species, while the sequence 25-39 has no biological activity and differs between species. Antisera raised to the intact ACTH molecule are usually directed to a site in the inactive C-terminal portion of the molecule and will therefore potentially detect inactive fragments resulting from metabolic degradation of ACTH in vivo. In practice this problem is partly overcome by the use of antiserum produced using synthetic 1.-24 sequence (Synacthen, Ciba) as immunogen. With an antiserum of this type and a tracer of the same material, estimates of ACTH by radioimmunoassay are very comparable to those of biological assay (Rees et al. 1973). Another example is presented by the small peptide hormone oxytocin. The bulk of the molecule consists of a ring structure completed by a disulphide bridge; biological activity is critically dependent on the integrity of this ring and is totally lost if it is broken at any point. By contrast, immunological activity is not dependent on the ring structure, though it is progressively destroyed by sequential removal of amino acids. When oxytocin is incubated in the presence of human late-pregnancy plasma, which contains the enzyme oxytocinase, the rate of destruction as assessed by biological assay is greater than that by radioimmunoassay (James et al. 1971). A similar phenomenon is seen following an infusion of oxytocin in vivo, though here the major factor leading to destruction of the hormone is metabolism in the liver and kidneys (Fig. 11.2). The degree to which dissociation of biological and immunological activity is likely will be highly variable according to the nature of the system studied. Where the antigenic site is coincident with the functional site little or no dissociation occurs. Where the sites are not coincident the amount of dissociation will depend on the relative
Ch. I 1
499
RELATION TO OTHER 1 YI’ES OF ASSAY
2500
W/ml
0
“
“
5
10
‘
--o
I
15 20 MlNLllES
25
30
Fig. 11.2. The disappearance of circulating oxytocin, measured by immunoassay and bioassay. following an intravenous infusion of the hormone.
rates of degradation of the 2 sites. If the functional site is less stable, then an immunoassay will overestimate the total activity present. On the other hand, if the antigenic site is less stable the immunoassay will yield an underestimate, a situation which appears to be extremely rare in practice. Dissociation of immunological and biological activity cannot always be explained on the basis of differences in the rate of degradation of the two sites concerned. An example which seems almost to defy analysis is that illustrated in Fig. 11.3; serial blood samples collected throughout the menstrual cycle and submitted to bioassay and radioimmunoassay for LH show a totally different pattern for the 2 types of measurement. Possible explanations include the existence of two forms of LH, o r the presence a t mid-cycle of factors which inhibit biological activity but do not affect immunological activity. Though the dissociation of biological and immunological activity is often regarded as a disadvantage of the latter systems, it can also have positive advantages. In situations where a biological material is highly unstable or has a very short half-life in vivo, the specific measurement of its metabolic products (‘footprints’) may be as Suhlcir
indry
p 531
500
K A D I O I M M U N O A S S A Y A N D K E L A T E D TECHNIQUES
80 70
60
M rnU LHlml 40
30 20 10
Fig. 11.3. Circulating levels of luteinising hormone (LH) in normal menstrual cycles as measured by immunoassay).--.( and a cytochemical bioassay (0-0). Note the complete absence of any relationship between the two. (From data kindly supplied by Dr. R. M. Kramer.)
valuable or more so than that of the parent compound. A good example is the measurement of the prostaglandin metabolite 15-ketoPGF,, (Levine and Gutierrez-Cernosek 1972). Though dissociation of biological and immunological activity is usually discussed in the context of hormones it may also apply to other materials. For example, enzymes are classically assayed by their biological ‘activity’ - the rate of conversion of a substrate to a product. The factors contributing to this activity are often complex, depending not only on the precise conditions under which the determination is carried out, but also on the fact that several different enzymes may contribute to a given ‘activity’. Immunoassays for enzymes have been described (e.g. for placental alkaline phosphatase see Jacoby and Bagshawe 1972). Provided that the usual criteria of purity of antigen and specificity of the antiserum are met, these
Ch. 11
RtL.ArlON TO O T H t R TYPES O F ASSAY
501
should measure the absolute concentration of a given enzyme independently of any of the factors which may influence its activity. The concept of 'enzymes by mass' is very attractive in practical clinical terms: for diagnostic as opposed to research purposes, the functional activity is irrelevant except as a means to an end, which is to estimate the release of an enzyme into the blood as a result of cell damage (Landon et al. 1977). Immunoassay can do this directly. Furthermore, it raises the interesting possibility of distinguishing between intracellular and circulating enzymes which have the same activity but different structures; specific measurement of the former in blood would be a much more sensitive indicator of trouble because it should not normally appear in this site (Fig. 11.4). Another example of this approach is the measurement of circulating myoglobin in the diagnosis of myocardial infarction (Kagen et al. 1975). Many materials which are measured by radioimmunoassay have no biological activity in the sense in which this term is commonly used - examples include carcinoembryonic antigen, a-fetoprotein and the fibrin degradation products. Furthermore, for most compounds of this type there is no alternative system of assay in biological fluids, INTACT CELL
intra-cellular o circulating
DAMAGED CELL
I
enzymes with different structure but identical activity
Fig. I 1.4. Why immunological activity may be of more practical value than biological activity. An enzyme may exist in 2 forms, one intracellular and one extracellular. They may have different chemical structures but idcntical functional activity. A classical assay could not distinguish these. However, an immunoassay which recognises structure and not function could specifically measure thc intracellular enzyme and thus provide a very sensitive indicator for this enzyme when it appears in the circulation as a result of cell damage. SUhj'~Ct !,ld<,.lp . 531
502
KADIOIMMUNOASSAY A N D RELATED TECHNIQUES
and therefore no reference system against which to judge ‘dissociation’. Under these circumstances, proof that the immunoassay is measuring the intended material and not some variant depends on physicochemical procedures of the type outlined in 9 9.3.5 for the examination of specificity. The finding of non-identity does not necessarily invalidate the assay. For instance, the bulk of material measured in serum by an assay for the terminal degradation product of fibrin/ fibrinogen, fragment E, is in fact material of higher molecular weight which represents partially degradated fibrinogen (Fig. 113. But since the use of fragment E has several technicaladvantages (availability Void Volume IFlbr/nogen
FgD FgE
II
r
90
-
r)
8070 60
50 40
30
. . .
PULMONARY EMBOLUS
-
PRE-OPERAT IM
lo
L+-+?--
NORMAL 60
90
120
1%
180
Volume (mls)
Fig. 1 I . 5 . Results of a radioimmunoassay for the terminal degradation product of fibrinogen, fragment E (FgE), when applied to fractions of human sera obtained by gel-filtration chromatography. Note that in both normal and abnormal cases, the bulk of the material detected by this assay is eluted earlier than fibrinogen itself, and so has a higher molecular weight; it is therefore not fragment E.
Ch. I 1
RELATION TO OTHER TYPES OF ASSAY
503
and stability of reagents, specificity with respect to intact fibrinogen), the use of this system has many practical advantages.
I I .5. Conclusions No assay is perfect, and different types of assay will reflect different aspects of the same material. Given that any assay should, during its development, be compared with other methods the final criterion must be: is it useful? There is little point in deprecating a technique because it yields results which are twice those of a more sophisticated procedure, when the former is simple and capable of high throughput. This has always been the great strength of radioimmunoassays over biological assays. Even if the result is not exactly ‘right’ according to some absolute reference, it is still immensely valuable if it serves to distinguish, on a routine basis, between health and disease.
Subject index p . 531
CHAPTER 12
Automation of binding assays
12.I . General All binding assays consist of repetitive steps of reagent addition, mixing, separation and counting of tubes. At the basic technical level this has been described as ‘the scientific equivalent of sewing mailbags’ (Bagshawe), and there can be no doubt that for the actual worker at the laboratory bench the performance of these assays is exceptionally tedious. But the disadvantages extend beyond a simple dislike of repetitive manual tasks: first, endless repetition can very readily lead to loss of precision ; second, manual methods are labour intensive and therefore expensive. For all these reasons a degree of automation or total automation of binding assays is highly desirable. The eventual aim should be that of any automated procedure in clinical chemistry - a sample, with appropriate identification, is put in at one end of a machine, and after the shortest possible lapse of time a result is printed out on a suitable form at the other end. The only demand on the routine operator should be the introduction of bulk reagent and general maintenance of the equipment. At the present time there is no widely used and fully accepted equipment for the performance of fully automated binding assays. However, this area is the subject of intensive development, not least in the activities of several large commercial organisations, and by the time this book is in print it is very likely that one or more machines will be available which meet the criteria of complete automation. The general features of automated systems can best be described under the headings of the successive steps in a single binding assay: 504
Ch. 12
AUTOMATION OF 13INDING ASSAYS
505
identification and dispensing of the sample; addition of reagents; incubation; separation of bound and free ligand; quantitation of the tracer; and calculation of results. The same headings also form the basis for discussion of the ‘semi-automated’ systems which are already in widespread use.
12.2. Identification aiTd dispensing of’ the sample There are 2 means for identification of patient samples in any assay: indirect by numbering of the samples and reaction tubes in a fixed sequence; and direct by labelling the reaction tubes with the information (patient’s name and number) from the sample tube. Both types of procedure are used in manual assays and both can be automated by the use of a digital computer. Identification by numerical sequence has the advantages of simplicity and convenience ; the disadvantages are the possibility of serious mistakes due to an error in the sequencing, and that a list of all samples has to be compiled. In the case of a fully automated assay the list can be in the form of a punched tape prepared as the samples are introduced, and subsequently read out as part of the on- or off-line calculation of results. By contrast, direct labelling of the reaction tubes has the great advantage that it almost completely eliminates errors of identification. The disadvantage is that it requires, with a fully automated procedure, a very sophisticated system for labelling each reaction tube in a machine-readable code. In most automated systems the sample is dispensed from tubes or cups set out in a tray - which may be in the form of a turntable or a rectangular cassette. Each position on the tray has a number which provides sample identification. A fixed amount of sample is aspirated from each tube of the reaction tray and transferred to the reaction tubes for mixing with reagents. An important and serious difficulty of this step is that of carryover from one sample to the next. A small residue of specimen in the delivery nozzle may be dispensed with the following specimen; if the first sample contains a high level of the substance being measured Subjeer index p . 531
506
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
and the second a low level the contamination can seriously affect the results. This problem can be overcome to some extent by arranging that the delivery of a fixed volume of sample is followed by the delivery of a fixed and larger quantity of assay diluent through the same nozzle. In practice ‘wash-ratios’ of 3 : 1 or greater are required if significant carry-over is to be avoided between samples which differ in their concentration by an order of magnitude or greater. Another problem at the step of sample dispensing is blocking of the aspiration tube with particulate material in the sample - especially when the latter is plasma. The operator working with a manual pipette can observe and correct this fault; a machine cannot and may dispense a samplevolume which is considerably less than that intended.
12.3. Addition of’ reagents The manner of addition of reagents will be determined by which of the 2 fundamental approaches to automation is adopted: a discrete system or a continuous-flow system. In a discrete system the sample and reagents are delivered into an individual reaction tube (replicated if necessary) via a system of machine-driven syringes or roller pumps. Current equipment of this type, which is fairly widely used as part of a semi-automated system, varies chiefly in whether the tubes are moved mechanically past a fixed delivery nozzle (as in the LKB 2071 and Micromedic system) or whether the tubes remain static and the delivery nozzle is moved on a mechanical head; combinations are also possible (as in the Analmatic system). In continuous-flow systems (Technicon), the reaction mixture circulates in a length of tubing, successive samples or replicates being separated by a bubble. Flow is maintained by roller pumps, and reagents are introduced via side arms controlled by valves.
12.4. Incubation This can be either ‘off-line’ or ‘on-line’. For most current assays, in which incubation takes considerably longer than all other steps,
Ch. 12
AUTOMATION OF BINDING ASSAYS
507
an off-line approach is necessary. With discrete systems the reaction tubes are in a cassette which can be removed, stored under appropriate conditions and then returned to the machine when incubation is complete. With a continuous-flow system the flow can be stopped and the incubation coil removed and stored. Automation, and particularly that of the continuous-flow type, is well suited to the type of 'disequilibrium' assay described in 4 8.2.4. With careful design these can achieve exceptional sensitivity in very short incubation times, and the nature of a continuous-flow process is such that the reaction time for every sample is identical. Developments in this area are awaited with great interest.
12.5. Separation o j bound and free ligand This is one of the most time-consuming steps in a binding assay involving, as it almost invariably does, the addition of an agent to insolubilise bound or free ligand, centrifugation to separate the insoluble phase and removal of the soluble phase prior to counting (see Ch. 5). It is also the step which, more than any other, has proved difficult to automate, and current efforts in this direction can perhaps better be described as ingenious rather than perfect. As there is no general approach to the automation of separation, the present available systems will be briefly summarised : (1) Kemtek 3000 system: following addition of the precipitating agent the contents of the reaction tube are transferred to filter discs (cellulose acetate, polycarbonate, glass fibre, or paper) located on a flexible plastic tape supplied from a spool. The tape moves on to a station at which each filter is sealed on its lower surface to a vacuum chamber ; the soluble phase is thus removed leaving the insoluble phase on the filter which, after drying, is wound through a detector system for counting. (2) Centria system (Union Carbide): the contents of each reaction vessel are transferred to an individual pre-packed Sephadex column, the latter being mounted in groups in a centrifuge head. Centrifugation results in separation of the bound and free fractions by molecularSubiccr eidc,r p. 531
508
RADIOIMMUNOASSAY A N D RELATED 1ECHNIQUES
sieve chromatography (see $5.3.2): the first fraction to emerge from the column (antibody-bound ligand) is then transferred for counting. (3) Technicon system: this is based on the continuous-flow principle (see above), and on the use of insolubilised antibody, the solid phase being cellulose coated on to iron particles. After incubation the latter are held static by a magnetic field while the free phase is removed by washing. The particles are then released from the magnet and transferred (again by continuous flow) to a counting head. (4)ARIA I1 (Becton Dickinson): employs a reusable chamber containing solid-phase antibody. The sample and other reagents pass through this in a continuous-flow system, bound antigen being eluted to regenerate the antibody. ( 5 ) Gamma-flow system (Squibb) (Brooker et al. 1976): employs a column containing an anion-exchange resin, with or without added charcoal, which adsorbs free antigen. The sample and other reagents pass through this in a continuous-flow system. (6) Micromedic Systems ‘Concept 4 Automatic Radioassay’ : employs antibody coated tubes (Johnson et al. 1976).
12.6. Counting of radioactivity This does not differ in principle from the general methods already described in Ch. 3. In the Kemtek 3000, the paper spool is wound between opposed pairs of crystal detectors; a total of 5 heads is available permitting the simultaneous counting of 5 locations on the tape. In the Centria system, 3 classical detector heads count 3 tubes at a time. In the Technicon machine the tubing of the continuousflow system is formed as a coil in a well-type crystal; each segment of magnetic particles is counted as it passes through this coil.
12.7. Calculation of’ results Most currently developed or developing systems of automated radioimmunoassay have full on-line computational facilities of the type described in $ 7.3. The use of a fairly sophisticated computer is
Ch. 12
AUTOMATION OF BINDING ASSAYS
509
justified in this context because it can also undertake other functions including sequence timing in the machine itself and replacement of electronic components in the detector system.
12.8. Conclusiom Fully automated systems for binding assays are not at present widely available, and of the machines under development it is difficult to predict which will become the ‘market leader’. That such machines will be extensively used is, however, guaranteed by the exponential growth in the practical application of radioimmunoassays. The principle advantages of automation are: (1) high sample throughput (20-60/hr) for a minimum of operator time - implicit in which is a reduction in costs because machines are usually less expensive than people; (2) increased speed of sample throughput both because work does not need to be deferred, and the possibility of using fast disequilibrium assays ; (3) increased precision through elimination of operator variation (see Q 10.2.2). The disadvantages are: (1) unless the machine is fully used the cost benefits are negated, for example, none of the equipment described above would be cost-effective in a laboratory receiving less than 50-100 samples/ day; (2) if the machine breaks down (which it inevitably will from time to time), then the laboratory without sufficient technicians to process urgent samples manually will be in serious difficulties. Both these points should be examined with care before a decision, which will often be influenced by the current passion for hardware, is made.
Ssbkcr index p. S31
CHAPTER 13
Organisation of assay services
Binding assays, and in particular radioimmunoassays, have long since passed the stage at which they might be regarded as research procedures. There are now several important branches of clinical medicine which could not be actively pursued if these techniques were not available. This being the case an important consideration is the question of how to deliver the best possible service to the patient at the lowest possible price. The subject of this last chapter is, therefore, the organisation of assay services - who should do them and how they should be done. Although the discussion will centre around radioimmunoassay, it should be emphasised again that the isotopic label is not a defining feature of these techniques: however, the existence of alternative but equivalent systems would not alter the observations presented here.
13.1. Who should perform radioimmunoassay ? The practice of medicine and pathology is highly structured on a departmental basis and when a new subject emerges there may be a long period of years or even decades before it is formally recognised and established as a series of independent units. The specialised branches of surgery provide an excellent example of this. Until a new subject comes of age it will usually be practised under the aegis of an existing department and the final separation of the fledgling speciality is frequently a painful process at both the personal and administrative level. Radioimmunoassay presents an extreme example of the difficulties of this process. The sheer breadth of its application 510
Ch. 13
ORGANISATION OF ASSAY SERVICES
511
has led to its emergence from departments as disparate as endocrinology, obstetrics/gynaecology, clinical chemistry, nuclear medicine, anatomy and physiology. The only notable absence from this list is immunology - a subject which in its more traditional reaches has completely ignored one of its most important potential fields of activity. It may be most difficult to predict within any one hospital complex where one will find the radioimmunoassay laboratory; commonly there are several. This may be satisfactory, indeed inevitable, at the research level. It is highly unsatisfactory when the demand is for a routine service to patients. With only few exceptions, the best place for routine radioimmunoassays to be carried out is in a radioimmunoassay laboratory managed and staffL by people who are experts in radioimmunoassay. Whether this laboratory operates under the auspices of a department of clinical chemistry, or of nuclear medicine, or any other, is immaterial - the principle point to recognise is that no existing department has any prior claim to house it in the absence of experienced personnel. Radioimmunoassay is a sufficiently large subject and sufficiently different from any other to merit full-time staff who, whatever their previous background, are dedicated to this subject alone. A busy centre should have full-time personnel at all levels who are not expected to have a substantial and simultaneous commitment to routine clinical chemistry, to in vivo isotopic methods, or to a clinical discipline. Exceptions must, of course, be made for those situations in which the result is always required urgently, such as measurement of drug levels.
13.2. Organisation of an assay laboratory Although most present radioimmunoassay (RIA) laboratories have arisen on an ad hoc basis in existing departments, it is possible to make certain generalisations on the requirements of a new laboratory: (1) Stufr: the number of staff required can be judged from the total throughput of assays intended. As a rule of thumb it may be assumed that one technician can process 10,000 samples per year, or 250 per Subjecr i n d u p 531
512
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
working week. There are, of course, exceptions to this: for assays with tedious extraction steps the throughput will be very much reduced; for those which are automated it may be substantially greater. Similarly, the output of a technician who is responsible for several differenttypes of assay will be less than that of one responsible for only a single assay. For every 3-4 technicians actively engaged in RIA procedures, the unit will require one well-trained graduate. Regardless of the total number of staff involved, the unit will also require a full or part-time director who may be a medical or nonmedical graduate with an extensive background in radioimmunoassay. There are many situations in which a laboratory will be required to perform RIA procedures at throughputs considerably less than those suggested above (i.e. fewer than 250 samples per week). Examples would include the smaller district laboratory responsible for a small number of urgent determinations such as those required in the assessment of drug levels or of fetoplacental dysfunction. In these circumstances, 2 or 3 members of the general laboratory staff should be deputed to become familiar with the use of reagent kits (see below), so that any one of them is able to meet demand on an ad hoc basis. (2) Stajftraining: it is impossible at the present time to lay down any specific course of training for those involved in the management of a RIA laboratory other than that they should have a period of not less than 1 year in a recognised and active unit carrying out this type of work. The subject has not crystallised to the extent that any formal training scheme can substitute for experience, and the typical educational course now available can only provide a theoretical background to what is essentially a practical subject. Similarly, and for the reasons already noted above, it is impossible to be specific about the type of person who should be trained - a graduate in almost any scientific discipline would be appropriate, with perhaps a slight edge towards the biochemist. At the technical level the training problem is simpler. In essence, any reasonably intelligent person with a secondary education in
Ch. 13
OR(iANISA1 ION OF ASSAY S E R V I C E S
513
science subjects is suitable and can be introduced immediately to RIA techniques. No amount of previous experience and training in nonRIA subjects (e.g. medical physics, clinical chemistry) will qualify them any better. A young man or woman taken direct from school to work in a busy unit should be capable of independent running of an assay within 3 months and of several assays, if necessary, within a year. If well supervised and provided with a reasonable background of principle and practice such people will be as competent and effective - often more so - than those with a long list of irrelevant qualifications. (3) Space: the space requirements for a RIA laboratory do not differ significantly from those of any general analytical laboratory. A provision of 3-5 m of bench should be made for each person involved in the day-to-day handling of specimens together with additional areas for counters, centrifuges, automated equipment, sample reception and offices. A substantial unit with 12 staff and an annual throughput of 50,OOO-I 00,000 samples would be wellhoused in a total area of 300 m2. Currently, of course, many units fall substantially below this optimum. A specific requirement of the larger unit (i.e. those performing their own iodinations) is a completely separate area reserved exclusively for the handling of high levels of radioactivity; the design of this area should conform to national and international requirements for the amounts of activity likely to be handled. (4)Equipmerit: apart from the general equipment to be found in any biochemical laboratory, the only specific requirements of a RIA laboratory are for centrifuges capable of handling large numbers of tubes (100 plus) at the same time, and nuclear counters. The latter have already been described (# 3.3). In general, the laboratory processing less than 100 specimens per week for any one assay will be best served by a simple instrument with manual sample changing. If the numbers for any one assay are greater than this, or if the number of different assays performed is large, then a counter with automatic sample changing becomes necessary. One automatic machine with a single detector head should suffice for a throughput of up to 20,000 Suhlrcr im/c\ 11 531
514
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
samples per annum; above this number additional machines will be required pro rata. These figures are likely to vary with the introduction of multi-detector counters and fully automated systems with integral detectors (see 4 3.3 and Ch. 12). ( 5 ) Reagenfs: the source of the reagents used by any one laboratory will vary greatly with the size and type of unit concerned. At one extreme will be the very large group with international status who will probably prepare the majority of their own reagents and, indeed, be responsible for distributing these to other laboratories. At the other extreme will be the very small group who use nothing but ‘kits’ of reagents from outside suppliers. Between the two, every possible combination of local and outside supply will be encountered. For laboratories performing RIA (the majority of current techniques) a key point is whether or not they have facilities for radioiodination. In the absence of these, the unit is likely to be a predominantly kit user since there are few prime suppliers, either in the public or the private sector, who are prepared to deliver tracer on its own. By contrast, the other reagents for RIA can be prepared relatively easily from bulk materials with a long or almost indefinite half-life. Whether such materials will be subjected to adequate quality control by the small user is, of course, another matter. The primary aim of many laboratories, certainly in the public sector in the U.K., has always been to prepare as many reagents as possible on site. While admirable in principle this approach can suffer from 2 serious disadvantages. First, it has been repeatedly emphasised that no two sets of reagents, however apparently identical, will ever give exactly the same result. For this reason the onus of preparing both normal and clinical ranges, and the transmission of these to the physician, must rest on the individual unit. This type of discrepancy is minimised, and the value of the test in the smaller laboratory correspondingly enhanced, if completely common sets of reagents and a common methodology are distributed from a central source. Second, it is all too frequently overlooked that the preparation of primary reagents is time-consuming and expensive. Over-simplified costings are often presented which ignore labour and general over-
Ch. 13
ORCiANISAl ION OF ASSAY SERVICES
515
heads - yet for most materials these are by far the largest proportion of the real costs. As a rule of thumb it may be taken that a throughput of less than 50 samples per week for a given assay does not merit local preparation of reagents -a kit will be more cost-effective and probably give better results. Kits may be prepared either by public organisations or private firms, usually the latter. Commercial kits are often disparaged because of the high profit margins involved. This is a historic concept: over the last few years prices have not proved sensitive to inflation and margins are now comparable to those on most high-grade chemical reagents. However the increasing and highly desirable control over the quality of kits by Government bodies will inevitably, as it already has with drugs, curb competition and thus lead to an increase in prices in real terms.
13.3. Organisation o f assay services Development in the practical application of RIA demands that the services be organised to serve the best interests of the patient and that careful consideration should be given to the organisation of these services at the hospital, regional and national level. The primary criteria for such services are availability, efficiency, and cost, the 3 factors being closely related. Availability implies that an assay of clinical value, however recondite, should nevertheless be accessible to every clinician. Efficiency implies that the result should be available in the shortest possible time and should have accuracy and precision such that it confers maximum clinical benefit. Finally, the cost should be as low as is compatible with the maintenance of an effective service. The emergence of RIA techniques from the research field, with widely scattered units, many of which have little or no clinical commitment, has led to a situation which in most countries can only be described as chaotic. There can be little doubt that the best approach for future organisation will be on the basis of a national or even a supranational plan. An excellent example of what can or might be Suhjccr index p. 5 3 /
516
KADIOIMMUNOASSAY A N D KELA-TED TECHNIQUES
achieved is provided by current developments in the United Kingdom. Although these have been devised within the context of a socialised system of medical care, there is little reason why they should not be applicable within any other system. The basic design is that of a 3-tier system. At the district level (2-3 small hospitals or one large hospital) there will be a laboratory for assays which are either verycommon (e.g. thyroxine) or very urgent (e.g. digoxin). This laboratory will probably have part-time staff, and will not be responsible for the production of reagents. At the regional level there will be a larger laboratory responsible for a wide spectrum of the commoner assays, and for the production and distribution of reagents including radioactive tracer to the district laboratories. This laboratory will have full-time staff. At the national level there are 14 or more supraregional laboratories which have some functions comparable to those of the regional laboratories but which are also responsible for unusual and difficult assays which it would be uneconomic to set up everywhere (a good example is ACTH, for which there are 2 centres in the U.K.). A specific function of the supraregional centres is to institute and manage quality-control systems which apply to all assays wherever performed. The system described above has many advantages. Equally it has several disadvantages or at least drawbacks when it is run in practice. First, it is important that the organisation be established as a whole and not piecemeal. There is a strong tendency to set up the prestigious supraregional centres and then to ignore the remaining tiers. As a result the large centres are swamped with common assays to the detriment of their proper reference functions. Second, it has to be recognised that the distribution of reagents is often far more demanding of time, space, and personnel than is the preparation of these reagents. Simply because the large group has substantial stocks of materials it should not be assumed that these can be dispersed without some further investment. The apparent inability of many sophisticated assays to ‘travel’ serves to underline this point. Finally, it must be recognised that any degree of centralisation is bound to impose delays in results because of the logistic problems involved.
Ch. 13
ORGANISATION OF ASSAY SERVICES
517
For many assays already available such delays can negate the clinical value of any result. For many assays currently under development, such as those for enzymes and drugs, the same limitations will apply. Total centralisation, however superficially attractive and cost-effective, will not always provide the best service to the patient.
Appendices
APPENDIX I
Manufacturers and suppliers of equipment RADIATION COUNTERS (automatic and/or manual sample changers) AmesCo., Division of Miles Laboratories Inc., 1127 Myrtle Street, Elkhart, Ind. 46514, U.S.A. Baird Atomic Inc., 125 Middlesex Turnpike, Bedford, Mass. 01730, U.S.A. or Baird Atomic Ltd., Station Lane, Hornchurch, Essex, U.K. Beckman Instruments Inc., 2500 Harbor Boulevard, Fullterton, Calif. 92634, U.S.A. or Beckman RIIC Ltd., Eastfield Industrial Estate, Glenrothes, Fife KY7 4NG, Scotland. ICN Pharmaceutical N.V., 277 Antwerpsesteenweg. B2800 Mechelen, Belgium. Inotron Ltd., 37 Windermere, Liden, Swindon, Wilts, U.K. Intertechnique, Western House, 4a Hercies Road, Uxbridge UBlO 9NA, U.K. J. & P. Engineering Ltd., Cardiff Road, Reading, Berks, U.K. Kontron Technik, Ag., Bernerstrasse 169, CH 8048 Zurich, Switzerland. LKB Produkter Ab., PO Box 76, S-161 25 Stockholm Bromma, Sweden. Nuclear Enterprises Ltd., Sighthill, Edinburgh, Scotland. Packard Instruments Inc., 2200 Warrenville Road, Downer’s Grove, Ill. 60515, U.S.A. Panax Equipment Ltd., Willow Lane, Mitcham, Surrey, U.K. Picker International Operations GmbH, Seldbergstrasse 6, 6201 Auringen, Wiesbaden, G.F.R. Searle Analytic Inc., 2000 Nuclear Drive, Desplaines, I11 60018. U.S.A. Wilj Electronics Ltd., Briscall House, Wotton Road, Ashford, Kent, U.K. Laboratorium Prof. Dr. Berthold, Calmbacher Strasse 22, 7547 Wildbad 1, G.F.R. RADIATION MONITORS Mini-Instruments Ltd., 8 Station Industrial Estate, Burnham-on-Crouch, Essex, U.K.
518
519
APPENDICES
CENTRIFUGES MSE Ltd., Manor Royal, Crawley, Sussex. U.K. Baird & Tatlock Ltd., P O Box I. Romford. Essex. U.K. IEC, 115 Fourth Avenue, Boston, Mass. 02195, U.S.A. Sorvall Instruments, El Du Pont de Nemours Inc., Instrument Products, Biomedical Division. Newtown. Conn. 06470, U.S.A. AUTOMATIC PIPETTING APPARATUS
Hook & Tucker Ltd., Vulcan Way, New Addington, U.K. Micromedic Systems Inc., Rohm & Haas Building, Independence Mall West, Philadelphia, Pa. 19105, U.S.A. Vitatron NV, 23 Spoorstraat, Dieren, The Netherlands. LKB-Produkter AB, Box 76. S-161 25 Stockholm Bromma. Sweden. MANUALLY-OPERATED PIPETTES Eppendorf Geratehau, Postfach 324. 2 Hamburg 63. G.F.R. Oxford Laboratories, Foster City, Calif. 94404. U.S.A. Labsystems Oy, Pulttitie 9. 00810 Helsinki 81, Finland. Jencons Scientific Ltd., Mark Rd, Hemel Hampstead, Herts, U.K. Scientific Manufacturing Industries, 1399 64th Street, Emeryville. Calif. 94608. U.S.A. A note on pipettes A wide variety of pipettes are available for RIA techniques. The types employing a plunger and a spring-return, with disposable tips. are particularly convenient and popular. With any pipetting system, and regardless of the manufacturers’ specifications, it is wise from time to time t o check both accuracy and precision. This is especially important with older equipment to assess wear and tear. Accuracy is tested by dispensing a fixed volume of de-ionised water and weighing this on an analytical balance. Precision is tested by dispensing 20 aliquots of a fixed volume of radioactive tracer. counting each aliquot, and calculating the coefficient of variation of these counts. The figure obtained should be corrected for the intrinsic counting error. which is equivalent to the square root of the mean counts for all aliquots.
ELECTROPHORESIS EQUIPMENT Shandon Southern Ltd., Frimley Road, Camberley, Surrey, U.K. Buchler Instruments lnc., 1327 16th Street, Fort Lee, N.J. 07024. U.S.A. Canal Industrial Corporation, 5635 Fisher Lane, Rockville, Md. 20852. U.S.A. Subjrrr indm p . 531
520
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
AUTOMATED A N D SEMI-AUTOMATED SYSTEMS Micromedic Systems Inc., Rohm & Haas Building, Independence Mall West, Philadelphia, Pa. 19 105, U.S.A. LKB-Produkter, AB, S-161 25 Stockholm Bromma I , Sweden. Baird & Tatlock Ltd., Chadwell Heath, Essex, U.K. Packard Instruments Inc., 2200 Warrenville Road, Downer’s Grove, 111. 6051 5, U.S.A. Technicon Instruments Corporation, Tarrytown, N.Y. 1059I , U.S.A. Union Carbide Inc., Clinical Diagnostics, 401 Theodore Fremd Avenue, Rye, N.Y. 10580, U.S.A. Kemtek, Kemble Instruments Ltd., 80 Park Road, Burgess Hill, Sussex, FH15 8HG, U.K. Auto-Assay, 175 W-2590 South, Salt Lake City, Utah 841 15, U.S.A. CALCULATORS Hewlett-Packard Inc., Route 41, Avondale, Philadelphia, Pa. 193 1 I , U.S.A. Tektronix Inc., P O Box 500, Beaverton, Ore. 97005, U.S.A. TUBES, TUBE RACKS, ETC. Luckham Ltd., Labro Works, Victoria Gardens, Burgess Hill, Sussex. U.K. COLUMNS FOR CHROMATOGRAPHY Wright Scientific Ltd., Cardigan Road, London NW6, U.K Pharmacia AB., Box 604. S-751 25 Uppsala I , Sweden, MISCELLANEOUS Millipore Corporation, Bedford, Mass. 01730, U.S.A. Sartorius-Membranfllter GmbH, 3400 Gottingen, G.F.R.
11 APPENDIX
Suppliers of special reagents and chemicals GENERAL National Pituitary Agency, Suite 503-7.210 West Fayette Street. Baltimore, Md. 21201, U.S.A. Medical Research Council, Division of Biological Standards, Holly Hill Laboratories, Hampstead, London NW3, U.K.
52 1
APPENDICES
ISOTOPES AND ISOTOPICALLY LABELLED MATERIALS
The Radiochemical Centre, Amersham, Bucks, U.K. New England Nuclear, Atomlight Place, North Billerica, Mass. 01862, U.S.A. CEA/IRE/SORIN, 13040 Saluggia, Vercelli, Italy. ANTISERA Antibodies Inc., PO Box 442, Davis, Calif. 95696, U.S.A. Calbiochem Inc., PO Box 12087, San Diego. Calif. 921 12. U.S.A. ILS Ltd., 99 New Cavendish Street, London WI, U . K . lnolex Biomedical, 2600 Bond Street, Park Forest South, Ill. 60466, U.S.A. Miles Laboratories, Stoke Court, Stoke Poges, Bucks, U.K. Serono Biodata, Via Casilina 125, 00176 Rome, Italy. Wellcome Reagents Ltd., Beckenham, Kent, U.K. HORMONES ICN Pharmaceuticals h e . , Life Sciences Group, 26201 Miles Road, Cleveland, Ohio 44128, U.S.A. CIBA Laboratories, Horsham. Sussex, U.K. Steraloids, Inc., Wilton, N.H. 03086, U.S.A. DOUBLE-ANTIBODY SOLID PHASE Organon Scientific Development Group, PO Box 20, Kloosterstraat 6, Oss, The Nether-
lands. QUALITY CONTROLS Hyland, Costa Mesa. Calif. 92626. U.S.A. Ortho Diagnostics lnc., Raritan, N.J. 08869, U.S.A
APPENDIX III
Suppliers of general reagents and materials Armour Pharmaceutical Co. Ltd., International Division, Hampden Park, Eastbourne, Sussex BN22 9AA, U.K. Bayer AG, Beraus Pharma B2, 509 Leverkusen, Bayerwerk, G.F.R. Bio-Rad Laboratories, 32nd and Griffin Avenue, Richmond, Calif. 94804, U.S.A. British Drug Houses Ltd. (B.D.H.), Poole, Dorset, U.K. Subject index p 531
522
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
Calbiochem Inc., PO Box 12087, San Diego, Calif. 921 12, U.S.A. Corning Glass Works, Corning, N.Y., U.S.A. Difco Laboratories, Detroit, Mich. 48232, U.S.A. Eli Lilly International, PO Box 32, Indianapolis, Ind. 4206. U.S.A. Hopkin and Williams Ltd., PO Box 1, Romford. Essex, U.K. Jencons Ltd., Mark Road, Hemel Hampstead, Herts, U.K. Koch-Light Laboratories Ltd., Colnbrook. Bucks, U.K. E. Merck, D61 Darmstadt, G.F.R. Oxoid Ltd., Southwark Bridge Road, London, SEI. U.K. Pbarmacia AB, PO Box 604, S-751 25 Uppsala 1, Sweden. Sigma Chemical Co., PO Box 14508, St. Louis, Mo. 63178, U.S.A. Whatman Labsales Ltd., Springfield Mill, Maidstone, Kent, U.K.
APPENDIX IV
Manufacturers of reagent kits for radioimmunoassay and related techniques (note: a very detailed list of kits and manufacturers appears in Clinical Chemistry (1977) 23,403441). Abbott Laboratories, Diagnostic Division, Abbott Park, North Chicago, Ill. 60064, U.S.A. Radiochemical Centre, Amersham, Bucks, U.K. Ames Co., Division of Miles Laboratories Inc., 1127 Myrtle Street, Elkhart, Ind. 46514, U.S.A. Beckman Instruments Inc., Campus Drive at Jamboree Blvd, Irvine, Calif. 92664, U.S.A. Biolab, Avenue de Tervuren 142, 1040 Brussels. Belgium. Bio-RIA, Div. of the Institute of Bio-Endocrinology, Inc., 10850 Hamon Street, Montreal, Que., Canada. Calbiochem, 10933 No. Torrey Pines Road, San Diego, Calif. 92037, U.S.A. Cambridge Nuclear Radio-Pharmaceutical Corp., 575 Middlesex Turnpike. Billerica, Mass. 01865, U S A . CEA/IRE/SORIN, 13040 Saluggia, Vercelli, Italy. Clinical Assays Inc., 237 Binney Street, Cambridge, Mass. 02142, U.S.A. Corning Diagnostics, Medfield, Mass. 02052, U.S.A. Curtis Laboratories Inc., 1948 East 46th Street, Los Angeles, Calif. 90058, U.S.A. Data Diagnostics Corp., 3401 Main Street, Houston, Tex. 77002, U.S.A. Diagnostics Product Corp., 12306 Exposition Blvd, Los Angeles, Calif. 90064, U.S.A. Electro-Nucleonics Laboratories Inc., 4809 Auburn Avenue, Bethesda, Md. 20014, U.S.A. Endocrine Sciences, 18418 Oxnard Street, Tarzana, Calif. 91365, U.S.A. Fisher Scientific Co., 71 1 Forbes Avenue, Pittsburgh, Pa. 15219, U.S.A.
APPENDICES
523
General Medical Systems Inc., 3814 Cavalier Street, Garland, Tex. 75042, U.S.A. Interscihce Institute, 200 Cotner Avenue, Los Angeles, Calif. 90025, U.S.A. Kallestad Laboratories Inc., 100 Lake Hazeltine Drive, Chaska, Minn. 55318, U.S.A. Life Systems Inc., 505 Northern Blvd, Great Neck, N.Y. 11021, U.S.A. Mallinckrodt Inc., 675 Brown Road, Hazlewood, Mo. 63042, U.S.A. Meloy Laboratories he., 6715 Electronic Drive, Springfield, Va. 22151, U.S.A. New England Nuclear, 549 Albany Street. Boston, Mass. 021 18, U.S.A. Nichols Inst., 1300 Beacon Street, San Pedro, Calif. 90731, U.S.A. Nuclear Diagnostics Inc., 575 Robbins Drive, Troy, Mich. 48084, U.S.A. Nuclear International Inc., 215 Middlesex Turnpike. Burlington, Mass. 01803, U.S.A. Nuclear Medical Laboratories Inc., 8700 Stemmons Expressway, Dallas, Tex. 75247, U.S.A. Nuclear Medical Systems Inc., 51 5 Superior Avenue, Newport Beach, Calif. 92660. U.S.A. Pantex, PO Box 966, Malibu, Calif. 90265, U.S.A. Pharmacia Laboratories, PO Box 604, S-751 25 Uppsala 1. Sweden. Quantimetrix, PO Box 1693, Beverley Hills, Calif. 90213. U.S.A. Radioassay Systems Laboratories Inc., 151 1 E. Del Arno Blvd, Carson, Calif. 90746, U.S.A. RIA Diagnostics, 8226 Allport Avenue, Santa Fe Springs, Calif. 90670, U.S.A. RIA Products Inc., PO Box 914, 97 Beaver Street, Waltham, Mass. 02154, U.S.A. Roche Diagnostics, Kingsland Street, Nutley. N.J. 0701 I , U.S.A. Schwartz/Mann, Mountain View Avenue, Orangeburg, N.Y. 10962, U.S.A. Smith & Kline InstrumentsInc., 880 W. Maude Avenue, Sunnyville, Calif. 94086, U.S.A. E.R. Squibb & Sons lnc., PO Box 4000. Princeton, N.J. 08540, U.S.A. Syva Corp., 3181 Porter Drive, Palo Alto, Calif. 94304. U.S.A. Technia Diagnostics Ltd., 48-50 Bartholornew Close. London ECI, U.K. Union Carbide Clinical Diagnostics, 401 Theodore Freud Avenue, Rye, N.Y. 10580, U.S.A. Wellcome Research Laboratories, Beckenham, Kent, U.K. Wien Laboratories Inc., PO Box 227, Succasunwa, N.J. 07876, U.S.A.
.
APPENDIX V
Safety precautions in the handling of radioactive isotopes Contamination of personnel with radioactive isotopes is regarded as one of the greatest hazards of work in a biomedical laboratory. As a result there are stringent regulations and requirements far precautions in the handling of these materials. Although the amounts of radioactivity involved in radioimrnunoassay are relatively small. such precautions must be strictly observed. A set of general rules for this type of work is Subjerr index p 531
524
RADIO1 M M UNOASSAY A N D RELATED TECH NlQUES
presented here, which can perhaps be summarised as commonsense, care, and the realisation that any errors are easily detected. A more formal account may be found in the ‘Code of Practice for the Protection of Persons against Ionizing Radiations arising from Medical and Dental Use‘ (1972, London, Her MA.iesty’s Stationery Office) and in ‘Rules and Regulations’, Title 10 - Chapter I, Code of Federal Regulations. Codified and re-issued 1975. U.S. Nuclear Regulatory Commission, Washington, D.C. Radiation kazards in tlic R I A luhorarory There are 3 main hazards in the handling of unsealed sources of activity: ( I ) External 8- and y-irradiation: these are of little importance in the RIA laboratory because the total amounts of radioactivity are relatively small and the y-radiation of the principle isotope used (1251)is of low energy (0.027 MeV) and has little penetration. (2) Skin contamination: this is very common and at high levels might lead to skin necrosis. More important, unrecognised contamination of thc hands is likely to lead to ingestion of isotope and deposition in the body (see ( 3 ) below). ( 3 ) Ingestion and deposition of isotopes in the body: this is the major hazard, and usually arises as the result of preparation and consumption of food and beverages by personnel with unrecognised contamination of skin or clothing. It can also result from inhalation, since all 1251-labelledcompounds contain some free iodide which can, under some conditions, be converted to volatile free iodine. 12sI localises in the thyroid gland in which it may, at high levels, cause acute necrosis. In the long term, there is also a potential carcinogenic hazard. General admiriis/ration
Laboratory areas used for the handling of radioisotopes should be designated as such, and the use and storage of isotopes strictly confined to these areas. The areas should be clearly identitied by the display of internationally accepted warning signs (available in the U.K. from Jencons Ltd.). In the larger laboratory further areas should be designated for the handling of bulk quantities of radioisotopes ( > 1 mCi) and subjected to particularly rigorous inspection. These areas should be provided with fume cupboards for all high activity work, should be well-ventilated, and should have a shower facility against the possibility of severe skin contamination. No personnel should be permitted to work in any of these areas unless they have received instruction in the safe handling of isotopes. Every laboratory, however small, should appoint one member of the staff as the Radiation Safety Officer (RSO) who will be responsible for instructing and advising the staff in relation to radiation protection, for periodic monitoring of both staff and equipment for contamination, and for assisting in the event of an accidental spill. The RSO is also responsible for maintaining a log of all radioactivity entering the laboratory, and for its eventual route of disposal. I n most Western countries such records are a statutory requirement.
APPENDICES
525
Rules for hundling radioluhelled preparatioris The following rules apply to all levels of radioactivity: ( I ) Laboratory coats must always be worn in the laboratories. (2) No eating. drinking, or smoking should be permitted in the active laboratories. (3) Disposable gloves must be worn for all procedures involving solutions with radioactive concentrations greater than 0.1 pCi/ml and large volumes of lower concentration but total activity greater than 10 pCi. (4) Hands must be monitored with a simple end-window radiation monitor (MiniInstruments Ltd.) immediately after handling any activity greater than I mCi. both before and after the removal of gloves. (5) All radioactivc sources ofconcentration greater than 0.01 pCi/ml or total content greater than 1 pCi must be clearly labelled stating the isotope, activity, date, and volume. ( 6 ) Pipetting must never be performed by mouth. (7) Disposable containers and pipetting devices with disposable tips should be used whenever possible. (8) All bench topareas should have disposable coverings (e.g. 'Benchkote', Whatman Labsales) which are regularly monitored and replaced as necessary. (9) Procedures involving high levels of activity ( > 0.1 mCi) should be carried out in a tray which is monitored immediately after use. Where possible all glassware should be disposable. (10) All non-disposable equipment (columns, syringes, pipettes) used for high levels of activity should be thoroughly rinsed in running water after use and then allowed to soak in a detergent solution until required again. Such equipment should always be stored in the 'hot' laboratory and not taken for use outside. ( I 1) All tools (bottle openers. forceps) should be checked for contamination before being put away. (12) All sources of activity greater than 0. I mCi should be stored in a separate and designated area. preferably with lead shielding. Moriiioring o/ iIw luhoraiorjs All benches and equipment in low activity laboratories should be monitored once a week. and i n high activity laboratories after completion of every procedure. The permissible level of contamination is 10-'pCi/cmz averaged over an area not greater than 300 cmz. Monitoring o/ pi~rsoritrrl All personnel should have a general medical examination and a full blood exdmination at the time of first employment. The blood examination should be repeated at yearly intervals. All working staff should be provided with a film-badge which is examined and changed at monthly intervals. The maximum permissible dose level Suhjwr d i * rp. 5.11
526
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
is equivalent to 5 rem* per year to critical organs such as the gonads, bone marrow, and eye, or 3 rem in any one-quarter. In practice, such doses are very rare even in the busiest radioassay laboratory. Personnel working with iodine isotopes should have a thyroid count performed every month. Disposal of’ radioactive waste The rules for disposal of radioactive waste vary widely both between countries and within the same country. For this reason every laboratory should consult their appropriate local authority for guidance. Emergency procedure .for spills of’ radioactive materials The RSO and Head of Department should be informed at once. No personnel should be permitted into the affected area unless wearing gloves and overshoes, both of which must be disposed of on leaving the area. Any contaminated clothing must be removed and discarded. If the spill is on the skin the area should be flushed thoroughly with tap water taking great care not to spread the contamination. Decontamination should aim at getting activity down to pCi/cmz. Contaminated areas should be carefully mopped with disposable paper, then washed until activity is down to pCi/cm*. If this cannot be achieved the area must be temporarily covered with polythene sheeting, until dealt with by the RSO. Desigri of’the radioassay laboratory Radioassay kits, as opposed to bulk isotopes, can be handled in any general clinical laboratory (class 3). However, their use should be confined to a specified area which is provided with appropriate warning signs and is regularly monitored for contamination. Bulk isotopes ( 1 mCi or greater) should only be handled in a class 2 radioisotope laboratory which has the following general specifications: (1) walls and ceilings: washable, non-porous paint; ( 2 ) floor: washable linoleum, rubber, or vinyl; joins between floors and walls should be rounded; (3) sinks: connected to main drainage; stoppers and taps should be operated by foot-pedal; (4) ventilation: air-flow must be to the outside of the building or to an appropriate filter; ( 5 ) fume cupboards: must have an eddy-free airflow drawing at least 150 ft/min and vented to the outside of the building or to an appropriate filter. A self-contained charcoal-filter unit with 90% trapping efficiency is available from Interex Corp., 3 Strathmore Road, Natick, Mass. 01760, U.S.A.; ( 6 ) shower: a shower facility should be available immediately adjacent to the high-activity area. The amount of isotope (if any) which can be disposed of via mains drainage is determined by local regulations.
* The ‘rem’ is a dose equivalent derived from the absorbed dose measured in ‘rad’, and a quality factor and a distribution factor which depend on the type of radiation. One rad is the dose absorbed when 62.5 x lo6 MeV of energy is deposited in 1 g of matter.
References ADDISON. G.M. and C.N. HALES (1971) Horm. Res. 3, 59. ANDERSON, D.C. (1970) Clin. Chim. Acta 29, 513. ANDRIE.~. J., S. MANASand F. DRAY(1975) In: E. Cameron, S. Hillier and K. Griffiths (eds.) Steroid lmmunoassay (Alpha-Omega-Alpha, Cardiff) p. 189. BANCHAM, D.R. and P.M. COTES(1974) Brit. Med. Bull. 30, 12. BARISAS, B.G.. S.J. SINGER and J.M. STURTEVANT (1977) Immunochemistry 14. 247. BERSON, S.A. and R.S. YALOW (1966) Science 152, 205. BERSON, S.A., R.S. YALOW, A. BALJMAN. M.A. ROTHSCHILD and K. NEWERLY (1956) J. Clin. Invest. 35, 170. BINOUX,M.A. and S.E. ODELL (1973) J. Clin. Endocrinol. Metab. 36, 303. BIRNBAUMER. L. and S.L. POHL(1973) J. Biol. Chem. 248,2056. BLOOM, S. (1974) Brit. Med. Bull. 30, 62. BOLTON, A.E. and W.M. HUNTER (1973) Biochem. J. 133, 529. BOYD,G.W., J. LANDON and W.S. PEART (1967) Lancet ii, 1002. BROOKER,G.. W.L. TERASAKI and M.G. PRICE(1976) Science 194,270. BUTT,W.R. (1972) J. Endocrinol. 55, 453. CATT,K.J. and C.W. TREGEAR (1967) Science 158, 1570. CATT,K.J., M.L. DUFALJ and T. TSLJRUHARA (1972) J. Clin. Endocrinol. Metab. 34, 123. CHARD,T. (1971) In: K.E. Kirkham and W.M. Hunter (eds.) Radioimmunoassay Methods (Churchill-Livingstone, Edinburgh) p. 491. CHARD, T., M.J. KITAU and J. LANDON (1970) J. Endocrinol. 46, 269. CHARD, T., M.J. MARTIN and J. LANDON (1971) In: K.E. Kirkham and W.M. Hunter (eds.) Radioimmunoassay Methods (Churchill-Livingstone, Edinburgh) p. 257. CHOPRA, I.J. (1972) J. Clin. Endocrinol. 34, 938. COLLINS. W.P. and J.F. HENNAM (1976) Mol. Asp. Med. I, 1 . DEAN,P.D.G., D. EXLEY and M.W. JOHNSON (1971) Steroids 18, 543. DESBUQLJOIS, B. and G.D. AURBACH (1971) J. Clin. Endocrinol. Metab. 33, 732. DONINI, S . and P. DONINI (1969) Acta Endocrinol. (Copenhagen) Suppl. 142,257. EKINS, R.P. (1960) Clin. Chim. Acta 5,453. EKINS,R.P. (1969) In: M. Margoulies (ed.) Protein and Polypeptide Hormones (Excerpta Medica, Amsterdam) p. 633. EKINS, R.P. (1974) Brit. Med. Bull. 30, 3. ERLANGER, B.F. (1973) Pharmacol. Rev. 25, 271. FISCHER, L. (1971) An Introduction to Gel Chromatography (North-Holland, Amsterdam). FRANCHIMONT, P. (1971) In: K. Kirkham and W.M. Hunter (eds.) Radioimmunoassay Methods (Churchill-Livingstone, Edinburgh) p. 535. 527
Subjerr kdcr p . 531
528
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
FREEDLENDER, A. and R.E. CATHOU (1971) In: K.E. Kirkham and W.M. Hunter (eds.) Radioimmunoassay Methods (Churchill-Livingstone, Edinburgh) p. 94. FRENKEL; E.P.. S. KELLER and M.S. MCCALL (1966) J. Lab. Clin. Med. 68, 510. GALEN, R.S. and D. FORMAN (1977) Clin. Chem. 23, 119. GARDNER, J., G. BAILEY and T. CHARD (1974) Biochem. J. 137.469. GIBSON, R.G., B.I. HIRSCHOWITZ and A.A. MIHAS (1977) Clin. Chem. 23, 1046. GIESE,J. and M.D. NIELSEN (1971) In; K.E. Kirkham and W.M. Hunter (eds.) Radioimmunoassay Methods (Churchill-Livingstone, Edinburgh) p. 341. GILMAN A.G. . (1970) Proc. Natl. Acad. Sci. U.S.A. 67, 305. GORDON.Y.B., M.J. MARTIN, A.T. MCNEILE and T. C H A R D (1973) Lancet ii. 1168. GRANT, D.B. (1968) Acta Endocrinol. (Copenhagen) 5Y, 139. GREEN, I., W.E. PAULand B. BENACERRAF (1969) Proc. Natl. Acad. Sci. U.S.A. 64, 1095. GREENWOOD, F.C., W.M. HUNTER and J.S. GLOVER (1963) Biochem. J. 89, 114. GRODSKY. G.M. and P.H. FORSHAM (1960) J. Clin. Invest. 3Y, 1070. HABER, E., L.B. PACEand F.F. RICHARDS (1965) Anal. Biochem. f2, 163. HALES.C.N. and P.J. RANDLE (1963) Biochem. J. 88, 137. HALES,C.N.. P. BECK,M.J. EVANS and J.S. WOODHEAD (1975) In: C.A. Pasternak (ed.) Radioimmunoassay in Clinical Biochemistry (Heydcn, London) p. 283. HEDING, L.G. (1966) In: L. Donato, G . Milhaud and J. Sirchis (eds.) Labelled Proteins in Tracer Studies (EAEC, Brussels) p. 345. HERBERT, V., K.S. LAU,C.W. GOTTLIEB and S.J. BLEICHER (1965) J. Clin. Endocrinol. Metab. 25, 1375. HOLLANDER, F.C. den and A.H.W.M. SCHUURS (1971) In; K.E. Kirkham and W.M. Hunter (eds.) Radioimmunoassay Methods (Churchill-Livingstone, Edinburgh) p. 419. HUNTER, W. (1971) In: K.E. Kirkham and W.M. Hunter (eds.) Radioimmunoassay Methods (Churchill-Livingstone, Edinburgh) p. 3. HUNTER, W.M. and F.C. GREENWOOD (1964) Biochem. J. YI, 43. H U R N B.A.L. , and J. LANDON (1971) 61: K.E. Kirkham and W.M. Hunter (eds.) Radioimmunoassay Methods (Churchill-Livingstone. Edinburgh) p. 121. HWANC,P., H. GUYDAand H. FRIESEN (1971) Proc. Natl. Acad. Sci. U.S.A. 68, 1902. JACOBY, B. and K.D. BAGSHAWE (1972) Cancer Res. 32.2413. JAMES,M.A.R.. T. CHARD and M.L. FORSLING (1971) In: K.E. Kirkham and W.M. Hunter (eds.) Radioimmunoassay Methods (Churchill-Livingstone, Edinburgh) p. 545. JOHANSSON, E.D.B. (1969) Acta Endocrinol. (Copenhagen) 61, 607. JOHNSON, E.G., T.E. STONEYCYPHER and B.E. STURGIS (1976) Clin. Chem. 22, 1164. KAGEN, L., S. SCHEIDT and L. ROBERTS (1975) Amer. J. Med. 58. 177. M. SIRENand U. SEUDERLING (1975) Anal. Biochem. KARONEN, S.-L., P. MORSKY, 67. 1.
REF t R E NC ES
529
KEANE, P.M.. W.H.C. W A L K E R J.,GAULD1EandG.E. ABRAHAM (1976) Chn. Chem. 22, 70. K O R E N M AS.G. N . (1968) J. Clin. Endocrinol. Metab. 28. 127. LANDESMAN, R. and B. SAXENA (1976) Fertil. Steril. 27, 357. LANDON, J., J. C A R N E and Y D. LANGLEY (1977) Ann. Clin. Biochem. 14.90. G. (1972) In: G.N. Vyas, H.A. Perkins and R. Schinid (eds.) Hepatitis LE BOUVIER. and Blood Transfusion (Grune and Stratton. New York) p. 97. LEFKOWITZ. R.J.. J. ROTHand J. PASTAN (1970) Science 170. 622. LEUTE,R., E.F. ULLMANand A. GOLDSTEIN (1972) J. Amer. Med. Assoc. 221. 1231. LEVINE, L. and R.M. GUTIERREZ-CERNOSEK (1972) Prostaglandins 2,281. MARCHAL.ONIS. J.J. (1969) Biochem. J. 113, 299. MARTIN.M.J. and J. LANDON(1975) In: C.A. Pasternak (ed.) Radioimmunoassay in Clinical Biochemistry (Heyden, London) p. 269. MCFARLANE, A S . (1958) Nature (London) 182. 53. MCGUIRE, W.L. (1973) J. Clin. Invest. 52. 73. MEDHI.S.Q. (1975) In: C.A. Pasternak (ed.) Radioimmunoassay in Clinical Biochemistry (Heyden. London) p. 21 3. MILES,L.E.M. and C.N. HALES(1968) Nature (London) ZIY, 186. MITCHELL, M.L., A.B. H A R D E N and M.E. O'ROURKE(1960) J. Clin. Endocrinol. Metab. 20. 1474. MITCHELL, M.L. and J. BYRON(1967) Diabetes 16, 656. MORGAN,C.R. and W.A. LAZAROW (1963) Diabetes 12, 115. MORGAN. C.R., R.L. SORENSON and A. LAZAROW (1974) Diabetes 13, 1 and 579. MURPHY, B.E.P., W. ENGELBERG and C.J. PATTEE(1963) J. Clin. Endocrinol. Metab. 23, 293. NIESCHLAG, E. (1975) Immunization with Hormones in Reproduction Research (North-Holland, Amsterdam). NYE. L., G.C. FORREST, H. GREENWOOD. J.S. GARDNER. R. JAY,J.R. ROBERTS and J. LANDON (1976) Clin. Chim. Acta 6Y, 387. ORSKOV, H. (1967) Scand. J. Clin. Lab. Invest. 20, 297. PORT-ER, R.R. (1959) Biochem. J. 73. 119. PREEDY, J.R.K. and E.H. A I T K E (1961) N J . Biol. Chem. 236, 1297. RATCLIFFE, J.G. (1974) Brit. Med. Bull. 30, 32. REDSHAW. M.R. and S.S. LYNCH(1974) J. Endocrinol. 60. 527. REES,L.H. (1976) Clin. Endocrinol. (Oxford) 5,3635. REES.L.H..D.M. COOK. J.W. KENDAI-L. C.F. ALLEN, R.M. KRAMER.J.G. RATCLIFFE and R.A. KNIGHT (1971) Endocrinology 8Y, 254. REES,L.H.,J.G. RATCLIFFE,G.M. BESSER. R.M. K R A M E R J. , LAND ON^^^ J. C H A Y E N (1973) Nature New Biol. 241, 84. RODBARD,D. (1971) In: W.D. Odell and W.H. Daughaday (eds.) Principles of Competitive Protein Binding Assays (Lippincott, Philadelphia. Pa.) p. 204. Sshjwi i1vle.r p . 531
530
RADIOIMMUNOASSAY AND RELATED TECHNIQUES
RODBARD, D. and J.E. LEWALD (1970) Acta Endocrinol. (Copenhagen) Suppl. 147, 79. F. PENNISI, N. RICCIONI, P. GIANONI and R. GIORDANI ROSA,U., G.A. SCASSELLATI, (1964) Biochim. Biophys. Acta 86, 519. ROSNER, W. (1972) J. Clin. Endocrinol. Metab. 34, 983. G., R. ASSAN,R.S. YALOWand S.A. BERSON(1969) Nature (London) ROSSELIN, 212, 355. RUBENSTEIN, K., R. SCHNEIDER and E. ULLMAN (1972) Biochem. Biophys. Res. Commun. 47,846. SAMOLS, E.C. and D. BILKUS(1963) Proc. SOC.Exp. Biol. Med. 115, 89. SCHARPE, E.L., W.M. COOREMAN, W.J. BLOMME and G.M. LAEKEMAN (1976) Clin. Chem. 22, 733. SELA,M. (1969) Science 166, 1365. A.H. RUBENSTEIN, S. CHO and C. BAYLISS(1968) STEINER, D.F., 0.HALLUND, Diabetes 17, 725. STAHL,E. (ed.) (1969) Thin Layer Chromatography (English edition) (Springer Verlag, New York). SUN,L. and V. SPIEHLER (1976) Clin. Chem. 22, 2029. K. and J. FERRIN (1968) J. Clin. Endocrinol. Metab. 28, 1667. THOMAS, THORELL, J.J. and B.G. JOHANSSON (1971) Biochim. Biophys. Acta 251, 363. TRAFFORD, D.J.H., P.R. WARD,A.Y. Foo and H.L.J. MAKIN(1974) J. Endocrinol. 65.44.
TYLER, J.P.P.. J.F. HENNAM, J.R. NEWTON and W.P. COLLINS (1973) Steroids 22, 871. ULLMAN, E., J. BLAKEMORE and R. LEUTE (1975) Clin. Chem. 21, 1011. UPDIKE,S.J., J.D. SIMMONS, J.D. GRANT,J.A. MAGNUSON and T.L. GOODFRIEND (1973) Clin. Chem. 19, 1339. UTIGER,R.D., M.L. PARKER and W.H. DAUGHADAY (1962) J. Clin. Invest. 41,254. VAITUKAITIS, J., J.B. ROBBINS, E. NIESCHLAG andG.T. Ross (1971) J. Clin. Endocrinol. Metab. 33, 988. VOLLER,A.. D.E. BIDWELL and A. B A R T L E T T (Bull. ~~~ Wld ~ ) Hlth Org. 53,55. WEEMEN, B.K. van and A.H.W.M. SCHUURS (1971) FEBS Lett. 15,232. and H.E. BOND (1977) Clin. Chem. 23, WEI, R., G.J. KNIGHT,D.H. ZIMMERMAN 813. WIDE,L. and J. PORATH (1966) Biochim. Biophys. Acta 130, 257. (1967) Lancet ii, 1105. WIDE,L., J. BENNICHand S.G.O. JOHANSSON WISDOM, G.B. (1976) Clin. Chem. 22, 1243. WOODHEAD, J.W., G.M. ADDISON and C.N. HALES(1974) Brit. Med. Bull. 30,44. YALOW, R.S. and S.A. BERSON (1960) J. Clin. Invest. 39, 1157. YALOW,R.S. and S.A. BERSON(1968) In: R.L. Hayer, F.A. Goswitz and B.E.P. Murphy (eds.) Radioisotopes in Medicine. In Vitro Studies (A.E.C. Symposium Series No. 11) p. 7. ~ In: W. Odell and W.H. Daughaday (eds.) YALOW,R.S. and S.A. B E R S O(1971) Principles of Competitive-binding Protein Analysis (Lippincott, Philadelphia, Pa.).
Subject index
Accuracy 479480 ACTH 336,342, 351, 366, 386, 396, 427, 430, 476477, 498 adjuvant 390-391 adsorption 366, 407-410, 428433, 474475 affinity constant 308-310 importance 321-327 measurement 322-327, 384, 394, 46446 5 alpha-fetoprotein 334, 386 alternative label 303-304, 374 et seq. ammonium sulphate 41041 1,472 androgens 434 anilino-naphthalene sulphonic acid 437438 antibodies absorption 471472 allergic 399400 availability 384, 393 characteristics 384-385, 393-396 chemistry 379-381 combining site 381-382,497498 heterogeneity 465 production of 385 et seq. selection of 393-396 specificity 384, 393 storage of 396 titre 394 to growth hormone 399,438 antigens 381-382 531
immunogenicity 386-388 assay mean 493 assay services 510 et seq. automation 504 et seq. Background counts 350 bacteriophage labels 376 beta-particles 344, 346, 524 between-assay variation 490 binder and sensitivity 450452, 458460 availability 378 characteristics 377 et seq. detection 398-400 specificity 378 binding assay 302-304 biological assays 303,495 blank values 402403,433. 473, 480. 493 ‘bridge’ antibodies 358 buffers 327-328 Calculation of results 313, 440 et seq. electronic 443445, 508-509, 520 linearisation 441443 manual 440-441 carcinoembryonic antigen 334 carry over 505-506 centrifuges 519 charcoal 341,408409,472 chemical precipitation 365 chloramine T 306, 353-354,358-360,423
532
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
coated tubes 418419, 508 coefficient of variation 490 collection of samples 475476 competitive protein binding 302-304, 397-398,496497 computers 443445,461462 confidence limits 445 conjugation labelling 355, 357 counters 345-350, 508, 513-514, 518 automatic 348 manual 348 outputs 3 4 9 , 4 4 3 4 5 counting errors 485487 counting time 348-350 crystal size 348 curie 345 Cusum plot 492493 Damage 360-364 decay catastrophe 363 dialysis 366 dilution curves 31 I dinitrophenol391 disequilibrium assays 452454, 470-471, 481, 507 dissociation constant 307 double antibody 413418 prozone 415416 solid phase 417 4 18 Electrophoresis 368, 370-371, 405-406, 519 enzymes 474-476,477, 500-501 enzymoimmunoassay 374-375 equilibrium 308 equipment 513-514, 518-520 extraction methods 427 et seq. recovery 43 1 reproducibility 431 specificity... 431433,434-436 Fibrinogen 336,471472, 501-503
Florisil 410, 430 fluoroimmunoassay 374 follicle-stimulating hormone 331, 333, 470,47 1 fragments 497-503 free hormones 438 free radical labels 376 Fuller’s earth 409. 430 Gamma-rays 344-347, 524 gel filtration chromatography 366, 406407, 507-508 glass beads 430 glucagon 336 growth hormone 331-332,373,399,438 Half-life 345, 347 haptens 382,385, 387, 388-390 heparin 473 hepatitis antigen 336, 375 heteroscedasticity 442 history 301-302,304306 hydroxyapatite 410 Immune response 382-384 immunisation 385 et seq. animal species 391-392 route 392-393 specificity 330,469470 timing 393 immunoassay 302-304 immunoglobulin 379 immunoradiometric methods 422426 incubation time 452454 insulin 305, 336,476 iodination 353 et seq. conjugation labelling 355 damage 360-364 electrolytic 355 vaporisation 355 I2%odine352, 353 et seq., 358, 373 monochloride 354
SUBJECT INDEX
ion-exchange chromatography 366 isotopes (radioactive) 304, 485487 chemistry 343-345 isotopic abundance 358 Kits 514-515, 522-523 Laboratory space 513 lactoperoxidase 355 late addition 454 Law of Mass Action 308 ligdnd 302 ligand-free fluids 476478 logit transform 320, 4 4 1 4 2 , 468, 485, 493 luteinising hormone 331, 333, 338, 470, 499-500 lymphocytes 382-384 lyophilisation 484, 489 Magnetic particles 419, 508 model system 324327 multiple detectors 349 myoglobin 501 Neo-antigen 471472 non-parallelism 432, 464465, 476 Oestrogens 361, 389-390, 396, 400, 435, 436 oxytocin 339,427,478,498499 Parathormone 336 peptides 335 pipettes 519 placental lactogen 315-318, 332, 340, 366-367, 373. 385,386,458 polyacrylamide gel 41 9 polyethylene glycol 317,411412,413 precision 479 et seq. definition 479480 errors in assay 480-488
533
principles 306 progesterone 434 prolactin 331-332, 367, 373 prostaglandins 352, 390, 500 protein hormones 333-334 purification antibodies 454456 iodination mixtures 365 et seq. proteins 334 Quality control 488494 quenching 347 Quso 409 Radioactivity 344-345, 523-526 reagent supply 514-515, 520-522 receptor assay 302-304, 396-397, 400, 464,496 reproducibility 479480 restricted population mean 494 Safety precautions 523-525 sample identification 505 scales, arithmetic and logarithmic 321 Scatchard plot 322, 395 scintillators 345-348 second antibody 413418 prozone 41 5 4 1 6 solid phase 417418 sensitivity concentration of sample 427429 definition 446448 optimisation 325-327,448462 separation methods 401 et seq. adsorption 407410 and precision 480482 and specificity 472 automated 507-508 efficiency 4014 0 3 electrophoretic 405406 fractional precipitation 410413 gel filtration 406407
534
RADIOIMMUNOASSAY A N D RELATED TECHNIQUES
immunoradiometric 422426 practicality 403404 second antibody 413418 solid phase 418421 sex hormone binding globulin 397, 399, 434,497 silicates 409410 solid-phase systems 418421,472, 508 specific activity 347, 353, 370, 372-374, 449,458 specificity 463 et seq. assessment 466499 definition 466467, 473 non-specific non-specificity 472478 specific non-specificity 464-472 staff 511-513 standard curves 31 1-321,484-485 standard curve methods of plotting 3 1 4 321,440443 standards 310-311, 321, 337-341, 481484 steroid hormones 334-335, 356, 367, 386387,434436 storage 341-342,480 supraregional assay service 515-517 Talc 409 targeting of assays 457461 technical skill 480,487488, 504,512-513
temperature effects 456-457 terminology 302-304 thin-layer chromatography 367,478 thyroid-stimulating hormone 333 thyroxine 305,437,468.476 thyroxine binding globulin 305, 397-398, 399,437438 tracer 303, 343 and sensitivity 449450,458 assessment 369 et seq. characteristics 351 damage 35 I external 351-352 internal 351-352 purification 351 et seq. triiodothyronine 468 tritium 344 two-site assays 424-425 tyrosine 353 Vasopressin 339, 352, 362,427, 432 vitamin BIZ305 Wick chromatography 371-372 Wilzbach technique 352 within-assay variation 490 Yield 370