MEMBRANE PROTEIN TRANSPORT A Multi-Volume Volumes
•
Treatise 1996
This Page Intentionally Left Blank
MEMBRANE PROTEIN TRANSPORT A Multi-Volume Treatise Editor:
STEPHEN S. ROTHMAN University of California San Francisco, California
VOLUMES
•
1996
i^n) Greenwich, Connecticut
\M PRESS INC. London, England
Copyright © 7 996 by JAI PRESS INC. 55 Old Post Road, No. 2 Greenwich, Connecticut 06836 JAI PRESS LTD. 38 Tavistock Street Covent Garden London WC2E 7PB England All rights reserved. No part of this publication may be reproduced, stored on a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, filming, recording, or otherwise, without prior permission in writing from the publisher. ISBN: 1-55938-989-3 Manufactured in the United States of America
CONTENTS
LIST OF CONTRIBUTORS THE NUCLEAR PORE COMPLEX: TOWARD ITS MOLECULAR ARCHITECTURE, STRUCTURE, A N D FUNCTION Nelly Pante and Ueli Aebi
vii
1
STRUCTURE A N D FUNCTION OF MITOCHONDRIAL PRESEQUENCES Merritt Maduke and Da vid Roise
49
BACTERIAL TOXIN TRANSPORT: THE HEMOLYSIN SYSTEM Jonathan A. Sheps, Fang Zhang, and Victor Ling
81
PROTEIN SORTING TO THE YEAST VACUOLE Bruce F. Horazdovsky, Jeffrey H. Stack, and Scoff D. Emr
119
BACTERIAL EXTRACELLULAR SECRETION: TRANSPORT OF a-LYTIC PROTEASE ACROSS THE OUTER MEMBRANE OF ESCHERICHIA COLI Amy Fujishige Boggs and David A. Agard
165
MECHANISMS OF PEROXISOME BIOGENESIS: REGULATION OF PEROXISOMAL ENZYMES, A N D THEIR SUBSEQUENT SORTING TO PEROXISOMES Gillian M. Small
181
PEROXISOMAL TOPOGENIC SIGNALS A N D THE ETIOLOGY OF PEROXISOME-DEFICIENT DISEASE Yukio Fujiki
213
vi ATP BINDING CASSETTE PROTEINS IN YEAST Carol Berkower and Susan Michaelis MEMBRANE PROTEIN TRANSPORT IN EUKARYOTIC SECRETION CELLS Kaarin K. Concz and Stephen S. Rothman INDEX
CONTENTS 2 31
279 295
LIST OF CONTRIBUTORS
Ueli Aebi
M.E. Muller Institute for Microscopy, Biozentrum University of Basel Basel, Switzerland
David A. Agard
Howard Hughes Medical Institute and the Departments of Pharmaceutical Chemistry and Biochemistry University of California, San Francisco San Francisco, California
Carol Berkower
Department of Cell Biology and Anatomy The Johns Hopkins University School of Medicine Baltimore, Maryland
Amy Fujishige Boggs
Howard Hughes Medical Institute and the Departments of Pharmaceutical Chemistry and Biochemistry University of California, San Francisco San Francisco, California
Scott D. Emr
Division of Cellular and Molecular Medicine University of California, San Diego School of Medicine and Howard Hughes Medical Institute La Jolla, California
Yukio Fujiki
Department of Biology Faculty of Science Kyushu University Fukuoka, Japan
Kaarin K, Goncz
Cardiovascular Research Institute School of Medicine and Dentistry University of California, San Francisco San Francisco, California VII
vm
LIST OF CONTRIBUTORS
Bruce Horazdovsky
Department of Biochemistry University of Texas Southwestern Medical Center Dallas, Texas
Victor Ling
Division of Molecular and Structural Biology The Ontario Center Institute Departmentof Medical Biophysics University of Toronto Toronto, Ontario, Canada
Susan Michaelis
Department of Cell Biology and Anatomy The Johns Hopkins University School of Medicine Baltimore, Maryland
Merritt Maduke
Department of Chemistry and Biochemistry University of California, San Diego La Jolla, California
Nelly Pante
Department of Cell Biology and Anatomy The Johns Hopkins University School of Medicine Baltimore, Maryland
David Roise
Department of Chemistry and Biochemistry University of California, San Diego La Jolla, California
Stephen S. Rothman
Departments of Physiology and Stomatology Schools of Medicine and Dentistry University of California, San Francisco San Francisco, California
Gillian M. Small
Department of Cell Biology and Anatomy Mount Sinai School of Medicine New York, New York
Jonathan A. Sheps
Division of Molecular and Structural Biology The Ontario Center Institute Department of Medical Biophysics University of Toronto Toronto, Ontario, Canada
Jeffrey H. Stack
Division of Cellular and Molecular Medicine University of California, San Diego School of Medicine and Howard Hughes Medical Institute La Jolla, California
L ist of Contributors Fang Zhang
I Division of Molecular and Structural Biology The Ontario Cancer Institute Department of Medical Biophysics University of Toronto Toronto, Ontario, Canada
This Page Intentionally Left Blank
THE NUCLEAR PORE COMPLEX: TOWARD ITS MOLECULAR ARCHITECTURE, STRUCTURE, AND FUNCTION
Nelly Pante and Ueli Aebi
Abstract I. Introduction II. The Nuclear Envelope III. Structure of the Nuclear Pore Complex A. STEM Mass Analysis B. Structure of the Basic Framework of the NPC C. Cytoplasmic and Nuclear Ring D. Cytoplasmic Filaments and Nuclear Basket E. The Central Plug or Channel Complex F. A Consensus Model for the NPC IV. Toward a Molecular Architecture of the NPC A. Integral Membrane Proteins of the NPC B. Peripheral Membrane Proteins of the NPC C. Yeast NPC Proteins D. IsolationandCharacterizationof Distinct NPC Components V. Molecular Trafficking Through the NPC A. Passive Diffusion
Membrane Protein Ti-ansport Volume 3, pages 1-47. Copyright © 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 1-55938-989-3 1
2 2 3 5 8 9 11 12 15 18 18 19 24 31 35 36 36
NELLY PANTE and UELI AEBI
B. Import of Nuclear Proteins C. Export/Import ofRNAs and RNP Particles V. Conclusions and Future Prospects Acknowledgments References
36 38 40 42 42
ABSTRACT The nuclear pore complex (NPC) is a -120 MDa macromolecular assembly embedded in the double-membraned nuclear envelope (NE) that mediates bidirectional molecular trafficking between the cytoplasm and the nucleus of interphase cells. Electron microscopy (EM) of negatively stained and frozen-hydrated specimens combined with three-dimensional (3-D) reconstruction has yielded the architecture of the basic framework of the NPC, and a number of specimen preparation methods and microscopy techniques have revealed the structure of distinct peripheral NPC components such as the cytoplasmic and nuclear ring, the cytoplasmic filaments, and the nuclear basket. Recently, significant progress has been made toward identifying, characterizing, and cloning and sequencing NPC proteins. However, to date only about two dozen NPC proteins, called nucleoporins, have been identified, a number accounting for less than 15% of the NPC mass. Some of these nucleoporins have been localized within the 3-D structure of the NPC; thus the molecular architecture of the NPC is starting to emerge. Nevertheless, the 3-D structure and functional significance of the different structural components and proteins of the NPC remain to be established. Relatively little is known about the molecular mechanisms underlying nucleocytoplasmic transport through the NPC. However, over the past few years there have been some advances in establishing the signals, receptors, and factors mediating nuclear import of proteins and snRNPparticles> and nuclear export ofRNAs and RNP particles. These functional studies have indicated that there may exist multiple signaling pathways for these different processes and substrates. Here we review recent advances toward the 3-D structure and molecular architecture of the NPC, and the molecular basis of nucleocytoplasmic transport of proteins, RNAs, and RNP particles through the NPC.
I. INTRODUCTION In eukaryotic cells the nucleus is separated from the cytoplasm by the nuclear envelope (NE), which enables each compartment to have its own distinctive composition and function. Beside separating the genetic machinery from protein synthesis, the NE allows nucleocytoplasmic trafficking; most of the RNA synthesized in the nucleus is transported to the cytoplasm where it is used for protein synthesis, while proteins required for nuclear function are synthesized in the cytoplasm and imported into the nucleus. This bidirectional exchange of material between the nucleus and the cytoplasm occurs through the nuclear pore complexes
Nuclear Pore Complex
3
(NPCs), large macromolecular assemblies that span the NE. The NPC allows passive diffusion of ions and small molecules, but molecules larger than 9 nm in diameter do not traverse the NPC freely. These are selectively imported into the nucleus by a signal-requiring and ATP-dependent mechanism. Macromolecular assemblies such as ribonucleoprotein (RNP) particles containing mRNA are also selectively exported from the nucleus by active transport. Despite significant progress in understanding the structure and function of the NPC, the molecular mechanism of nucleocytoplasmic transport through the NPC has remained elusive. Here we review recent advances made toward the elucidation of the architecture and biochemical composition of the NPC, and recent progress made toward the understanding of the mechanism of mediated transport through the NPC.
II. THE NUCLEAR ENVELOPE As illustrated schematically in Figure 1, the nuclear envelope (NE) is made up of a double membrane enclosing a lumen, the "perinuclear space." The outer nuclear membrane faces the cytoplasm and is continuous with the endoplasmic reticulum (ER). Thus the perinuclear space is contiguous with the lumen of the ER. On its cytoplasmic surface the outer nuclear membrane is often studded with ribosomes, as is the rough ER. The inner nuclear membrane faces the nucleoplasm and in its nucleoplasmic surface is lined by the nuclear lamina, a polymer of intermediate filament-like proteins, the nuclear lamins (Aebi et al., 1986; Gerace and Burke, 1988). The nuclear lamina provides a general framework for NE structure and is an anchoring site for interphase chromatin. The NPCs are interposed at irregular intervals between the inner and outer nuclear membranes, where the two membranes are fused to form the "pore membrane." The density of NPCs within the NE varies among different cell types, but it is roughly proportional to the metabolic activity of the cell (Maul, 1977): it ranges between 2-4 pores/|im^ for lymphocytes and 50-60 pores/|Lim^ for maXurQ Xenopus laevis oocytes. Because of the high density of NPCs and the ease of manually isolating its NE, amphibian oocytes have been extensively used in structural studies of the NPC (see below). However, due to the lack of a reproducible bulk isolation procedure, this preparation is not of practical use for biochemical studies: about 100 nuclei have to be isolated to obtain enough material to reveal most of the NPC proteins on a SDS-polyacrylamide gel. Therefore, most of the biochemical information about the NE has been obtained from rat liver preparations. These preparations yield a fraction enriched in NPCs attached to the nuclear lamina that has been successfully used for the identification and isolation of several NPC proteins (see Section IV) and for raising antibodies directed against a number of NPC proteins (Davis and Blobel, 1986; Snow et al., 1987). During cell division the NE disassembles and reassembles in a strictly coordinated manner (reviewed by Wiese and Wilson, 1993). Membrane fragmentation of the NE occurs after chromatin condensation at prometaphase when the NPCs
NELLY PANTE and UELI AEBI Outer nuclear membrane Perinuclear space Inner nuclear membrane
Nuclear pore complex Endoplj retlj
Cytoplasm Ions, small molecules
Passive
Proteins, snRNPs RNAs/RNPs
diffusion
ATP-dependent mediated transport
Figure 1. Schematic representation of the different components of the NE together with the molecular trafficking occurring across the nuclear envelope (NE) through the nuclear pore complexes (NPCs). The NE consists of an inner and an outer nuclear membrane enclosing the perinuclear space. The outer nuclear membrane is continuous with the endoplasmic reticulum (ER), so that the perinuclear space of the NE is contiguous with the lumen of the ER. The inner nuclear membrane is lined by the nuclear lamina (NL), a near-tetragonal meshwork made of intermediate filament-like proteins called nuclear lamins. The NPCs are interposed at irregular intervals between the inner and outer nuclear membrane. Bidirectional molecular trafficking between the nucleus and the cytoplasm occurs through the NPCs, which allow passive diffusion of ions and small molecules, and ATP-mediated transport of proteins, RNAs, and RNPs.
completely disappear and the nuclear lamina depolymerizes. Reassembly of the NE occurs during telophase when the membrane fragments aggregate at the chromosome surfaces and intact NPCs start to reappear, even on small pieces of NEs (Roos, 1973). Little is known about the mechanism of fragmentation and reformation of the NE, and the disassembly and reassembly of the NPCs during mitosis. Immunofluorescence microscopy with antibodies against NPC proteins has shown that some NPC proteins disperse throughout the cytoplasm during mitosis and become progressively concentrated around the periphery of the chromosomes in late anaphase and early telophase (Snow et al., 1987). The incorporation of NPC proteins at the reforming NE seems to be a stepwise process in which assembly of
Nuclear Pore Complex
5
protein constituents of the NPC proper precedes assembly of the peripheral components of the NPC (Byrd et al., 1994). Some aspects of nuclear lamina depolymerization and repolymerization during cell division have been well established (reviewed by Gerace and Burke, 1988). This process is regulated by specific phosphorylation-dephosphorylation of the nuclear lamins: during mitosis when the nuclear lamina depolymerizes, the nuclear lamins are phosphorylated, and during telophase when the nuclear lamina reassembles, they are dephosphorylated (Gerace and Blobel, 1980; Ottaviano and Gerace, 1985). It has been shown that phosphorylation of the lamins by the cell-cycle-specific cdc2 kinase induces lamina disassembly in vitro (Peter et al., 1990; Dessev et al., 1991). The mitotic phosphorylation sites have been mapped at either end of the lamin molecule (Ward and Kirschner, 1990; Peter et al., 1991), and phosphorylation of these sites results in the disassembly of in vitro formed lamin head-to-tail polymers (Peter et al., 1991).
III. STRUCTURE OF THE NUCLEAR PORE COMPLEX The structure of the NPC has been extensively investigated by different electron microscopy (EM) specimen preparation methods and imaging techniques (reviewed by Pante and Aebi, 1993, 1994), including scanning force microscopy (SFM) in a physiological buffered environment (Pante and Aebi, 1993; Goldie et al., 1994). As documented in Figure 2, a and b, when intact Xenopus oocyte NEs are spread on an EM grid and viewed in a transmission EM after negative staining, extensive arrays of NPCs are revealed. In these images the NPCs appear as round particles with a diameter of-125 nm (Unwin and Milligan, 1982; Reichelt et al., 1990; Jamik and Aebi, 1991). Depending on the face of the NE that is adsorbed to the EM support film, the NPCs reveal different morphologies: if the NE adsorbs to the EM film with its cytoplasmic face, so that its nuclear face is exposed, the NPCs appear irregularly stained and poorly preserved (see Fig. 2a); whereas when the cytoplasmic face is exposed, the NPCs appear well preserved and exhibit a distinct eightfold rotational symmetry (Fig. 2b). In cross sections of Epon-embedded intact Xenopus oocytes or Xenopus oocyte nuclei (Fig. 2c), the NPCs appear to span the NE (i.e., '^SO nm), and they reveal filamentous structures associated with both its cytoplasmic and nuclear periphery (see Fig. 2c, arrowheads). Thus both negatively stained and embedded/thin-sectioned NPCs clearly reveal that the NPC is a structure with eightfold rotational symmetry, but its cytoplasmic and nuclear periphery exhibit a high degree of asymmetry. EM images of negatively stained and embedded/thin-sectioned NEs (i.e., Fig. 2, a-c) reveal the morphology of the NPC in the presence of the nuclear double membrane, which is probably essential for stabilization of the structure of the NPC. However, information on the different structural components of the NPC comes from studies where the nuclear membranes have been solubilized. As documented in Figure 2d, when NEs are exposed to treatment with nondenaturing detergents.
Figure 2. Xenopus oocyte nuclear envelope (NE) after different preparations for EM and different chemical treatments, (a) Nuclear and (b) cytoplasmic face of negatively stained intact NEs. Depending on which face of the NE adsorbed to the EM grid, the NPCs revealed a different morphology, thus revealing the highly asymmetric architecture of the NPC with regard to its nuclear (a) and cytoplasmic (b) periphery, (c) Cross section of Epon-embedded NE revealing both cytoplasmic (large arrowheads) and nuclear (small arrowheads) filamentous structures associated with the NPC periphery. (d) Negatively stained NPCs after detergent treatment of a spread NE. Whereas intact NPCs (a-c) are embedded in the NE, the detergent-treated NPCs in (d) have detached from the nuclear envelope, revealing the basic framework of the NPC consisting of the plug-spoke complex. Scale bar, 100 nm (a-d).
Nuclear Pore Complex
7
the basic framework of the NPC is yielded more clearly. It consists of eight "spokes" embracing a "central pore," which is sometimes "plugged" with a "central channel complex" (also called "central plug" or "transporter"). Recently, the architecture of the basic framework of the NPC has been determined by 3-D reconstruction of both negatively stained (Hinshaw et al., 1992) and frozen hydrated (Akey and Radermacher, 1993) detergent-treated NPCs. These studies are discussed in more detail in Section IIIB. As documented in Figure 3, besides revealing the basic framework (i.e., the spoke complex), detergent treatment also releases two types of rings: (i) cytoplasmic rings, which are predominantly positively stained and are yielded by rolling intact nuclei on an EM grid (Jamik and Aebi, 1991), and (ii) nuclear rings, which are negatively stained and appear less massive than the cytoplasmic rings, both by comparison of their radial mass density profiles (Fig. 3, c and d) and from direct mass analysis (discussed in Section IIIA). Both the cytoplasmic and nuclear rings
Figure 3. Negatively stained intact NPCs and distinct NPC components. Electron micrographs, correlation averages (always including 20 eight-fold rotationally symmetrized NPCs; upper insets), and radial mass density profiles (lower insets) of intact NPCs (a), and distinct NPC components yielded after treatment of spread NEs with 0.1% Triton X-100 (b-d). (a) Membrane-bound, intact NPCs; (b) plug-spoke complexes; (c) cytoplasmic rings, which are also left behind after rolling intact nuclei on a carbon film-coated EM grid; and (d) nuclear rings. Scale bar, 100 nm (a-d). Adapted from Jarnik and Aebi (1991).
8
NELLY PANTE and UELI AEBI
contain associated filamentous structures that were first observed in embedded/thin-sectioned NEs but were believed to represent a specimen preparation artifact since they were not depicted by other EM preparation techniques (Franke and Scheer, 1970a,b; Franke, 1974). Recently, the use of more elaborate EM specimen preparations and imaging techniques has further documented the existence of these peripheral filamentous components of the NPC (Jamik and Aebi, 1991; Ris, 1991; Goldberg and Allen, 1992; Goldie et al, 1994). As will be discussed in more detail in Section HID, these studies have revealed that the cytoplasmic ring is decorated with eight short cytoplasmic filaments, whereas the nuclear ring is topped by a mechanically fragile "baskef formed by eight thin filaments joined distally by a terminal ring. In summary, the NPC is composed of (i) a basic framework made of eight spokes embracing a central pore, (ii) a central plug or channel complex, (iii) a cytoplasmic ring from which eight short, kinky filaments emanate, and (iv) a nuclear ring with a basket-like assembly attached. In this NPC architecture the basic framework is sandwiched between the cytoplasmic and nuclear ring. In the following sections, we first present mass measurements of the entire NPC and its different components, before we discuss various structural aspects of the different NPC components in greater detail. A. STEM Mass Analysis
Not only does the NPC have large dimensions (^125 nm in diameter and -70 nm high), but it also harbors an enormous mass. Quantitative scanning transmission EM (STEM) has revealed for the intact NPC a total mass of 124 MDa (i.e., 124 x 10^ Da; Reichelt et al., 1990). The membrane-bound NPC without the central channel complex has a mass of 112 MDa; thus the central plug is -^12 MDa. These authors also measured the mass of the different NPC components obtained after detergent treatment of NEs. Accordingly, as summarized in Table 1, the basic framework of the NPC as reconstructed in 3-D by Hinshaw et al. (1992) and Akey
Table 1. Masses of the Intact NPC and Its Major Structural Components Mass (MDaf Nuclear pore complex Basic framework Central channel complex Cytoplasmic ring Nuclear ring
124.0 ±11.0 51.7 ± 5.3 12.0 ± 1.1 32.0 ± 5.5 21.1 ± 3.7
Note: ^Determined by quantitative scanning transmission EM (STEM) and adapted from Reichelt et al. (1990).
Nuclear Pore Complex
9
and Radermacher (1993) (see Section IIIB) has a mass of 52 MDa without the central channel complex. Two types of rings were also measured in detergenttreated NEs: heavy (32 MDa) and light (21 MDa). Since the 32-MDa rings were also observed as "footprints" when nuclei were rolled back and forth on charged EM grids, they must represent the cytoplasmic rings with the remnants of one to several collapsed cytoplasmic filaments attached (see Section HID). The 21-MDa nuclear rings frequently revealed mass in the center (see Section HID), which due to its variation among rings was excluded from the mass measurements. Thus the sum of the mass of its principal components (i.e., the basic framework, the central plug, and the cytoplasmic and nuclear ring) yields a total mass of 117 MDa (i.e., 52+12 + 32 + 21 MDa) for the intact NPC. The small difference (-6%) between the measured mass of the intact membrane-bound NPC and the sum of its components may be due to loss of some of its peripheral components (i.e., the cytoplasmic filaments and nuclear baskets) during detergent treatment and/or specimen preparation. B. Structure of the Basic Framework of the NPC
The architecture of the basic framework of the NPC (i.e., the spoke complex) has recently been revealed in 3-D reconstructions of both negatively stained (Hinshaw et al., 1992) and frozen hydrated (Akey and Radermacher, 1993) NPCs obtained after detergent treatment. As documented in Figure 4a, the 3-D mass map of negatively stained, detergent-treated NPCs is built of eight multidomain spokes, with each spoke consisting of two identical halves. Hence the entire spoke complex yields 8-2-2 symmetry with one half-spoke representing the asymmetric unit. Each half-spoke is built from four distinct morphological domains termed the "annular," "column," "ring," and "lumenal" domains (see Fig. 4b). The lumenal domain extends into the lumen of the NE and is believed to contain as one of its constituents the glycosylated NPC protein gp210 (see Section IVA). Since the basic framework of the NPC as shown in Figure 4a has a mass of 52 MDa (Reichelt et al., 1990; see also Table 1), the mass of one half-spoke is ~3.3 MDa, that is, it is on the order of a ribosome. When the pore membrane is positioned in the negatively stained detergent-released NPC map, eight ~10-nm diameter "peripheral channels" are created between two adjacent spokes and the pore membrane border at a radius of -40 nm from the NPC center (see Fig. 4a). As speculated by Hinshaw et al. (1992), these peripheral channels may represent sites for passive diffusion of ions and small molecules, and they may also facilitate import of inner nuclear membrane proteins (SouUan and Worman, 1993; reviewed by Wiese and Wilson, 1993). The 3-D reconstruction of frozen hydrated detergent-treated NPCs (Akey and Radermacher, 1993) produced a similar 3-D map of the basic framework of the NPC. However, these authors included the central channel complex in their reconstruction. Thus in their 3-D map the spoke complex embraces an elaborate, barrel-like central channel complex. Also by including the central channel complex,
10
NELLY PANTE and UELIAEBI
this reconstruction yields eight internal channels between two adjacent spokes and the central channel complex. However, since this tomographic reconstruction is based on a relatively large missing tilt cone of information that affects the final 3-D map predominantly at low radii, the central mass of their 3-D model may have been peripheral nel
leral
channel
Figure 4. Surface renderings of the 52-MDa basic framework of negatively stained NPCs released from the NE upon detergent treatment, (a) Slightly tilted view of the basic framework of the NPC, which consists of eight multidomain spokes and exhibits strong 8-2-2 symmetry thus indicating that it is built of two identical halves relative to the central plane of the nuclear envelope. Note that because of its irreproducible appearance the central plug or channel complex has been omitted in this reconstruction. When the pore membrane is positioned in the map of negatively stained detergent-released NPC, eight ~10-nm diameter "peripheral channels" are created between two adjacent spokes and the pore membrane border at a radius of - 4 0 nm. (b) Three different side views of one multidomain spoke cut out from the basic framework of the NPC as shown in (a). Each half-spoke is built from four distinct morphological domains, annular (a), column (c), ring (r), and lumenal (I). Adapted from H i n s h a w e t a l . (1992).
Nuclear Pore Complex
11
overestimated despite application of a solvent flattening procedure to the reconstruction (Akey and Radermacher, 1993). In additipn, as we will discuss in Section HIE, other NPC components (i.e., the nuclear basket) may contribute to what in ice-embedded NPC images appears as the central plug. Moreover, since the abundance and morphology of the central channel complex depend on the isolation and preparation conditions employed (see Section IIIE), it is necessary to develop a procedure to control the reproducible appearance of the central plug among NPCs before computing a 3-D map of an irreproducible structure. C. Cytoplasmic and Nuclear Ring
The basic framework of the NPC (Fig. 4a) is sandwiched between a cytoplasmic and a nuclear ring. Both cytoplasmic (Fig. 2b) and nuclear (Fig. 2a) views of negatively stained intact NPCs revealed a ring at the periphery of the NPC. These rings readily detach from the NPC proper upon chemical or physical manipulation of the NE: (i) Xenopus oocyte NEs extracted with Triton X-100 revealed distinct rings lying next to spoke complexes (Unwin and Milligan, 1982; Jamik and Aebi, 1991; Hinshaw et al., 1992); (ii) thin rings are observed in osmotically shocked NEs (Akey, 1989); and (iii) rolling of isolated Xenopus oocyte nuclei on an EM grid (without spreading the NE on the grid) leaves cytoplasmic rings (so-called footprints) behind (Jamik and Aebi, 1991). Projection maps of correlation averaged NPCs and their corresponding radial density profiles clearly demonstrated that after such treatments the remaining NPCs had lost material at their periphery (compare Fig. 3a with Fig. 3b; see also Jamik and Aebi, 1991). Moreover, STEM mass measurements of the rings released after detergent treatment revealed the existence of two types: 32-MDa and 21-MDa rings (Reichelt et al., 1990; see Section IIIA and Table 1). Because the footprints left behind after rolling isolated nuclei on EM grids (see above) also revealed a mass of 32 MDa, it was concluded that the 32-MDa rings represented cytoplasmic rings, whereas the 21 -MDa rings represented nuclear rings. Thus although the cytoplasmic and nuclear rings have a similar appearance, they are different. The difference between these two rings has also been demonstrated by limited proteolysis (Goldberg and Allen, 1993). Accordingly, tryptic digestion of NPCs sequentially removes subunits of both rings, but the nuclear rings are more sensitive to proteolysis than are the cytoplasmic rings. The above described studies have clearly documented the existence of distinct cytoplasmic and nuclear rings. In contrast, it has been argued that these rings are an integral part of the basic framework of the NPC (Hinshaw et al., 1992; Akey and Radermacher, 1993). In the 3-D map of the basic framework the ring domain of the multidomain spokes defines two tenuous rings at the cytoplasmic and nuclear faces of the NPC (Fig. 4). Since according to these authors the basic framework of the NPC exhibits good 8-2-2 symmetry, these two tenuous rings should be identical or at least very similar. In contrast, the cytoplasmic and nuclear rings released upon detergent treatment of Xenopus oocyte NE preparations yield distinct masses (see
12
NELLY PANTE and UELI AEBI
above and Table 1; Reichelt et aL, 1990). Moreover, the basic framework of the NPC reconstructed from negatively stained NPCs after detergent treatment (Hinshaw et al., 1992) has a mass of 52 MDa without the central plug. Hence there must be additional components associated with this basic framework to account for the --110-MDa mass of the intact, unplugged NPC (Reichelt et al., 1990). Interestingly, the masses of the 32-MDa cytoplasmic ring and 21-MDa nuclear ring amount to just about what has to be added to the 52-MDa mass of the basic framework (i.e., 105 MDa) to arrive at a mass of approximately 110 MDa for the intact, unplugged NPC. Therefore, it is conceivable that the cytoplasmic and nuclear rings are not (or only partially) represented in the 3-D mass density maps of the basic framework of detergent-released NPCs (see Fig. 4a). D. Cytoplasmic Filaments and Nuclear Basket Recently, several independent investigations have confirmed the existence of filamentous structures associated with both the cytoplasmic ring and the nuclear ring of the NPC (Jamik and Aebi, 1991; Ris, 1991; Goldberg and Allen, 1992; Goldie et al., 1994). As illustrated in Figure 5, a and b, v^h^n Xenopus oocyte NEs are visualized in the EM after quick freezing/freeze drying/rotary metal shadowing, the cytoplasmic face of the NPC looks distinct from the nuclear face (see also Jamik and Aebi, 1991). Accordingly, the cytoplasmic face of the NPC is topped with a 100-110-nm outer diameter ring, from which eight kinky filaments protrude (see Fig. 5a, arrowheads), which have a tendency to collapse into themselves and thus often appear as short cylinders or "cigars." The presence of these cytoplasmic filaments is best documented in situations where they have bent to the side and adhered to filaments of adjacent NPCs, thus appearing as NPC connecting fibrils (see Fig. 5a, small arrows). As shown in Figure 5b, the nuclear face of the NPC accommodates a more tenuous, 90-100-nm outer diameter ring from which eight thin, 50-100-nm long filaments emanate and are joined distally by a 30-50-nm diameter terminal ring, thus forming a "basket" or "fishtrap." These cytoplasmic filaments and nuclear baskets make the NPC distinctly asymmetric relative to the plane oftheNE. High-resolution scanning EM of critical point-dried/metal-sputtered isolated NEs have revealed similar structures (Ris, 1991; Goldberg and Allen, 1992). In addition, in Triturus cristatus this technique has depicted the existence of an ordered fibrous nuclear lattice, termed the "NE lattice" or NEL, that is connected to the nuclear baskets via their terminal rings (Goldberg and Allen, 1992). The chemical composition and function of this NEL remain to be established. Remnants of such a lattice or filament system connecting adjacent baskets has also been observed in Xenopus oocyte NEs in the form of basket connecting filaments (Jamik and Aebi, 1991; Ris, 1991). More recently, the native cytoplasmic and nuclear NPC topography has been visualized by scanning force microscopy (SFM) of spread Xenopus oocyte NEs
Nuclear Pore Complex
13 ■^JTT
b # :
Figure 5. The NPC Is highly asymmetric with regard to its cytoplasmic and nuclear periphery. The cytoplasmic (a and c) and nuclear (b and d) face of NPCs revealed by transmission EM of quick-frozen/freeze-dried/rotary metal-shadowed intact Xenopus oocyte NEs (a and b) and by scanning force microscopy (SFM) of intact Xenopus oocyte NEs kept In physiological buffer (c and d). Relatively short cytoplasmic filaments (a; arrowheads), and nuclear baskets (b) are revealed by quick freezing/freeze drying/rotary metal shadowing. The resolution of the SFM images (c and d) is insufficient to resolve individual NPC-associated filaments; thus the cytoplasmic face of the NPC appears "donut-IIke" (c), whereas the nuclear face exhibits a "dome-like" appearance (d). The arrowheads In (a) point to cytoplasmic filaments protruding from the cytoplasmic ring of the NPC, whereas the small arrows in (a) mark filaments that have bent to the side and thereby adhered to filaments of adjacent NPCs, thus appearing as "NPC connecting fibrils." Scale bar, 100 nm (a-d).
kept in physiological buffer (Pante and Aebi, 1993; Goldie et al., 1994). In agreement with the results of dehydrated specimens (Jamik and Aebi 1991; Ris, 1991; Goldberg and Allen, 1992), corresponding SFM topographs revealed a high degree of asymmetry betw^een the nuclear and cytoplasmic periphery of the NPC. As documented in Figure 5, c and d, by SFM the cytoplasmic face of the NPC appears "donut-like," whereas the nuclear face exhibits a "dome-like" appearance. However, since the resolution in these SFM images is insufficient to resolve individual NPC-associated filaments, the in vivo conformation of these cytoplasmic
14
NELLY PANTE and UELI AEBI
and nuclear filaments has remained uncertain. For example, the cytoplasmic filaments have been described as "granules" (Stewart and Whytock, 1988), "short cylinders" (Jamik and Aebi, 1991), and "T-shaped" particles (Goldberg and Allen, 1992). At this stage it is difficult to determine which of these conformations, if any, represents the native one or to which extent these different conformations may merely represent preparation artifacts. The integrity of the nuclear baskets critically depends on divalent cations. In the presence of 0.5 mM MgCl2 or CaCl2, well-formed baskets are observed. In contrast, as documented in Figure 6a (see also Jamik and Aebi, 1991), if the divalent cations
Figure 6. The structural integrity of the nuclear baskets depends on divalent cations. (a) Nuclear face of a quick-frozen/freeze-dried/rotary metal-shadowed Xenopus oocyte NE depleted of divalent cations with 2 mM EDTA. (b) Nuclear face of a Xenopus oocyte NE after treatment with 2 mM EDTA prior to quick freezing/freeze drying/rotary metal shadowing. Depletion of divalent cations with 2 mM EDTA destabilizes the terminal rings and thereby causes disassembly of the nuclear baskets (a), whereas the addition of divalent cations after destabilization with EDTA causes reformation of the nuclear baskets (b). (c, d) Electron micrographs, correlation averages (always including 20 eight-fold rotationally symmetrized NPCs; upper insets), and radial mass density profiles (lower insets) of negatively stained intact NPCs after treatment with 2 mM EDTA (c) or in the presence of 0.5 mM MgCl2 (d). The mass density profile about the center of the NPC is significantly attenuated in the presence of 2 mM EDTA, a condition that causes disassembly of the nuclear baskets (see a). Scale bar, 100 nm (a-d). Adapted from Jamik and Aebi (1991).
Nuclear Pore Complex
15
are chelated by 2 mM EDTA or EGTA, the nuclear baskets become destabilized and are disrupted. If exogenous ATP is present during isolation and preparation of NEs, a similar effect is observed (our own unpublished results). Surprisingly, if after destabilization by EDTA or EGTA divalent cations are reintroduced, the nuclear baskets reform (see Fig. 6b). These results indicate that the nuclear baskets are dynamic structures in that they disassemble upon removal of divalent cations and reassemble upon their addition (Jamik and Aebi, 1991) and therefore could be directly involved in the active transport of proteins, RN As, or RNP particles through the NFC. As to their possible functional role, the cytoplasmic filaments have been implicated (i) as docking sites for import into the nucleus through the NPC and (ii) in delivering the docked material to the central channel complex for active translocation (Gerace, 1992). Some indirect evidence that is consistent with such a scenario is the observation that the cytoplasmic filaments sometimes bend around so that their distal end reaches down into the central pore (see Fig. 12, a and c, and Fig. 13, arrowheads). E. The Central Plug or Channel Complex
Often a massive, ~12-MDa particle, which has been termed the "central plug," "central channel complex," or "transporter," is depicted in the central pore of the NPC (see Fig. 2, b and d). The frequent lack of this central structure and its highly polymorphic appearance within a given NPC population has given rise to suggestions that at least in part it may represent particles (i.e., RNP particles) in transit rather than an integral component of the NPC (e.g., Jamik and Aebi, 1991; Gerace, 1992). Nevertheless, based on a computer analysis of several thousand NPCs from frozen-hydrated NEs, Akey (1990) has classified this central structure into four different groups that are related to different transport states of the NPC and termed "closed," "docked," "open/in transit," and "open." Based on this computer classification, Akey (1990) has proposed this central structure to represent the actual "transporter" and modeled it as a double-iris arrangement that can assume several distinct configurations as it actively transports molecules and particles through the NPC. Since, depending on the isolation and/or specimen preparation of the NEs, both the abundance and appearance of this central plug are highly variable (Unwin andMilligan, 1982;Reicheltetal., 1990; Jamik and Aebi, 1991), it is far from clear whether the different morphologies presented by Akey (1990) do indeed represent distinct transport-related states. Thus the extent to which this central structure is an integral component of the NPC, or whether it represents at least in part material in transit remains to be determined. Moreover, to establish the functional significance of the different morphologies of this central structure identified by Akey (1990), it is necessary to directly correlate these with corresponding transport assays. In view of the recently described peripheral components of the NPC (see Section HID and Fig. 5), it is also conceivable that a substantial fraction of what in proj ection
16
NELLY PANTE and UELI AEBI
appears as the central plug in fact represents remnants of the nuclear basket (including the terminal ring), which may have been squashed into the pore upon embedding the NPC in a thin layer of negative stain or a thin ice film, or material associated with it. In support of this notion, the mass density profile of negatively stained NPCs in the presence of EDTA (Fig. 6c), a condition that causes disassembly of the nuclear baskets (see above; Jamik and Aebi, 1991), is significantly attenuated when compared with that of NPCs isolated in the presence of millimolar amounts of divalent cations (Fig. 6d). Thus to more systematically investigate the nature and 3-D structure of the central plug or channel complex, it is first necessary to develop a method that yields a reproducible presence and appearance of this NPC component, and at the same time control the structural integrity of the nuclear baskets. Toward achieving this goal, we have been exploring a number of different buffers, incubation conditions, and chemical fixation protocols. As illustrated in Figure 7, we have found that 95% of the NPCs harbor a massive central plug when during isolation the NEs are stabilized with Cu-orthophenanthroline, an oxidizing agent causing S-S bridge formation.
Figure 7. Chemical manipulation of isolated NEs changes both the abundance and appearance of the central plug or channel complex, (a, b) Negatively stained Xenopus oocyte NEs that have been isolated in the presence of 0.1 mM Cu-orthophenanthroline, an oxidizing agent causing intra- and intermolecular S-S bridge formation, (c) Control, prepared as in (a) but in the absence of Cu-orthophenanthroline. Accordingly, --95% of the NPCs yield a massive central plug when they are stabilized with Cu-orthophenanthroline during isolation. Scale bar, 200 nm (a) and 100 nm (b-c).
Nuclear Pore Complex
17
Cytoplasmic fliumcnts
r
Outer membrane l.umenal domain
Inner membrane \
Nuclear ring
Nuclear basket
^
Terminal ring
Figure 8. Schematic representation of the consensus model of the membrane-bound NPC. Its major structural components include the basic framework (i.e., the spoke complex as shown in Figure 4a), the central plug or channel complex, the cytoplasmic and nuclear rings, and the cytoplasmic filaments and nuclear basket. The 52-MDa basic framework of the NPC has been adapted from the 3-D reconstruction of negatively stained detergent-released NPCs (Hinshaw et al., 1992). The cytoplasmic filaments and nuclear basket have been modeled based on EM data obtained by Ris (1991), Jarnik and Aebi (1991), and Goldberg and Allen (1992) (see also Figure 5, a and b). In this consensus model of the NPC, we have also pictured a cytoplasmic and nuclear ring in addition to the two tenuous rings defined by the ring domain of the spoke complex (see Figure 4a). The central plug or channel complex has been modeled as a transparent ellipsoidal particle to indicate the fact that its definite structure remains to be determined.
Recently, the central plug as it appears in frozen-hydrated NPCs after detergent treatment has been reconstructed in 3-D (Akey and Radermacher, 1993). Accordingly, it has a rather complex, hourglass-like structure '-62-nm long with distal and central diameters of-42 and ~32 nm, respectively. However, as we have discussed in Section IIIB, reconstruction errors due to sampling and limited tilt range accumulate predominantly at low radii (i.e., in the center of the 3-D mass map); thus the central mass in this 3-D reconstruction may be overestimated. On the other
18
NELLY PANTE and UELI AEBI
hand, this representation of the central plug is not entirely consistent with the double-iris-like model previously proposed by Akey (1990). Based on the similarity of its overall size and shape, octagonal symmetry, and mass, it has been proposed that the central plug may represent a "vault" ribonucleoprotein particle (Kedersha et al., 1991), with the vaults acting as transport vehicles, that is, carrying nuclear proteins as their cargo. Recent evidence for vault immunoreactivity at the NPC (Chugani et al., 1993) supports this idea. However, a more specific and comprehensive characterization of the central plug in terms of its molecular structure and, association with the basic framework of the NPC still has to be achieved before it can be identified with any known particle such as vaults. Similarly, any such candidate particles must be subjected to a more stringent analysis regarding their possible interaction or association with the NPC. F. A Consensus Model for the NPC The large amount of structural studies (reviewed above) have elucidated the structure of some of the components of the NPC. Thus a consensus model for the architecture of the NPC is starting to emerge. In the model presented in Figure 8 we have identified a number of distinct structural components. Accordingly, intact membrane-embedded NPC consists of a basic framework (i.e., the spoke complex shown in Fig. 4a) sandwiched between a cytoplasmic and a nuclear ring. These are two rings in addition to the two tenuous rings defined by the ring subunits of the spokes. The cytoplasmic ring is decorated with eight cytoplasmic filaments, and the nuclear ring is topped with a basket-like filamentous assembly. The central pore often harbors a central plug or channel complex whose definite structure and functional role remain to be established.
IV. TOWARD A MOLECULAR ARCHITECTURE OF THE NPC In contrast to the relatively large number of structural studies (see above), less is known about the chemical composition and molecular architecture of the NPC. Based on its molecular mass of about 120 MDa (Reichelt et al., 1990), it is believed that the NPC is composed of multiple copies (i.e., 8 or 16) of on the order of-100 different proteins. For several years gp210 and p62 have been the only two well-characterized NPC polypeptides. However, the production of specific antibodies, the use of molecular genetics, and the development of an isolation procedure for NPCs from yeast (Rout and Blobel, 1993) have recently enabled the identification, characterization, and cloning and sequencing of an increasing number of NPC proteins. Moreover, the combination of well-characterized antibodies with different EM specimen preparation methods has allowed localization of several of these proteins to distinct structural components of the NPC. Thus elucidation of the molecular architecture of the NPC has now definitely started.
Nuclear Pore Complex
19
As summarized in Table 2, three major groups of NPC proteins have so far been identified and characterized. These are (i) integral membrane proteins with only part of their mass residing in the NPC, which have been proposed to anchor the NPC to the nuclear membrane; (ii) peripheral membrane proteins that are not associated with or anchored in the nuclear membrane; members of this group have been called "nucleoporins" (Davis and Blobel, 1986) and denoted by NUPx, where X indicates the molecular mass in kilodaltons (Sukegawa and Blobel, 1993); and (iii) related yeast NPC proteins whose exact relationship to the vertebrate NPC proteins is not yet known. Following is a description of the currently identified members of these three groups.. A. Integral Membrane Proteins of the NPC
The first NPC protein identified in rat liver NEs was a 190-kDa glycoprotein, originally termed gpl90 (Gerace et al., 1982) but renamed gp210 on the basis of the molecular mass of 210 kDa deduced from its amino acid sequence (Wozniak et al., 1989). gp210 was classified as an integral membrane protein since it remained associated with the nuclear membrane even after extraction with alkaline pH or chaotropic agents (Gerace et al., 1982). gp210 contains asparagine-linked, high mannose-type oligosaccharides and therefore binds concanavalin A (ConA), a lectin specific for a-D-mannopyranose and related sugars (Gerace et al., 1982; Wozniak et al., 1989). The topology of gp210 has been determined by the use of site-specific antibodies and proteolytic digestions (Greber et al., 1990). Accordingly, gp210 consists of a large (95% of its total mass) NH2-terminal domain residing in the lumen of the NE, a single, 21-residue-long transmembrane segment that has been shown to be sufficient for sorting gp2l0 to the nuclear membrane (Wozniak and Blobel, 1992), and a short, 58-residue long COOH-terminal domain associated with the NPC (see Fig. 9). Unexpectedly, an antibody directed against the lumenal domain of gp210 inhibits both passive diffusion of small molecules and mediated nuclear import of proteins (Greber and Gerace, 1992). Based on its topology and abundance in the NPC, there have been speculations that the lumenal domain of gp210 forms part of the "knobs" (Jarnik and Aebi, 1991) or lumenal subunits (see Fig. 4) that have been shown to extend from the spokes radially into the lumen of the NE (Hinshaw et al., 1992; Akey and Radermacher, 1993). As a possible functional role, gp210 has been proposed to act as a membrane anchor for the NPC and/or to have a topogenic role in membrane folding during nuclear pore formation (Greber et al., 1990; Jarnik and Aebi, 1991; Gerace, 1992; Hinshaw et al., 1992). Recently, another transmembrane NPC protein has been identified, cloned and sequenced, and characterized (Hallberg et al., 1993). Based on its predicted amino acid sequence, it has a molecular mass of 121 kDa and thus has been termed POM121 (for pore membrane protein of 121 kDa). This protein was not extracted from rat liver NEs by 7.0 M urea, and therefore it was classified as an integral
Table 2. Classification of Cloned and Sequenced NPC Proteins
VP~ Integral membrane proteins of the NPC
Name
Molecular Characteristics of primary and Other properties and possible massa (kDa) predicted secondary structure functions
gp2 10
210
POM121
121
POM152 (yeast)
152
Location
References
Most of its mass (N- Gerace et al., 1982; 2 1-residue long transmembrane Bears N-linked (via Asp) domain between a 58-residue high mannose domain) resides in Wozniak et al., long C-domainb and a 1783oligosacharides. Reacts the lumen of the 1989; Greber et with Con A. Antibodies NE. al.. 1990. residue long N - d ~ m a i n . ~ against lumenal domain inhibits NPC function. Possibly anchors the NPC to the pore membrane. Most of its mass (C- Hallberg et al., 1993. 44-residue long transmembrane Binds WGA. Possibly domain) resides in anchors the NPC to the domain between a 28-residue the NPC proper. long N-domain and a 1127pore membrane. residue long C-domain. Repetitive XFXFG motifs at C-terminal third. 20-residue long transmembrane Reacts with ConA. Possibly Most of its mass (C- Wozniak et al., 1994. domain between a 175anchors the NPC to the domain) resides in pore membrane. the lumen of the residue long N-domain and a NE. 1142-residue long C-domain.
Peripheral membrane p62 proteins of the NPC
62d
0-linked glycoprote- NUP153 ins
153
NUP107 NUPI55
107 155
Cytoplasmic and nu- Starr et al., 1990; clear periphery of Carmo-Fonseca et the central plug or al., 1991; Cordes channel complex et al., 1991; Finlay et al., 1991; Guan et al., 1995. Four zinc finger motifs. Binds WGA. Exist as a homo- Terminal ring of the Sukegawa and Repetitive XFXFG motifs at nuclear basket. Blobel, 1993; oligomer of 2 MDa. Binds C-terminal 213. DNA in vitro. McMorrow et al., 1994; Cordes et al., 1993; Pante et al., 1994. Repetitive XFXFG, SVFG, Binds WGA. Exist as a Cytoplasmic Kraemer et al., 1994; filaments. FGG, and FGG motifs. complex with p75. Pante et al., 1994. Leucine zipper motif. Putative oncogene product associated with myeloid leukemogenesis. Leucine zipper motif. Does not bind WGA. Unknown Radu et al., 1994. Nonrepetitive motifs. Does not bind WGA. Unknown Radu et al., 1993.
Tprlp265 p180?
265
- 1600-residue long a-helical
NSPl
87
Non-0-linked proteins
Yeast NPC proteins
a-helical coiled-coil C-domain. Binds WGA. Exists as a Repetitive XFXFG motifs at complex (p62-p58-p54N-domain. p45). Required for NPC function.
coiled-coil region. Acidic C-domain. Repetitive XFXFG motifs.
Very prone to proteolysis with a major proteolytic product of 180kDa. Essential for cell growth.
-
Cytoplasmic filaments.
Byrd et al., 1994; Wilken et al., 1993
Unknown.
Hurt, 1988. (continued)
Table 2. (Continued)
QP~ XFXFG family
Name NUPl NUP2
Molecular Characteristics of primary and Olher properlies and possible massa (kDa) predicted secondary structure functions 114 95
NUP/NSP49 49 NbPMSP100 100
GLFG family
Loca tion
Essential for cell growth. Unknown. Not required for cell growth. Unknown. Forms a complex with NUPI. Essential for cell growth. Unknown.
Repetitive GLFG motifs at Ndomain. Repetitive GLFG motifs at N- Binds RNA in vitro. domain. RNA binding motifs.
Unknown.
NUP/NSP116 116
Repetitive GLFG motifs at N- Deletions in the nupll6 gene Unknown domain. RNA binding motifs. yields sealed NPCs. Binds RNA in vitro.
NUPINSPI45 145
Repetitive GLFG motifs at N- Deletionsldisruptions in the Unknown domain. RNA binding motifs. nupl45 gene yield clusters of sealed NPCs. Binds RNA in vitro. Nonrepetitive motifs. Forms a complex with NSPI Unknown and NUP49.
N1C96 Notes:
Repetitive XFXFG motifs. Repetitive XFXFG motifs.
96
aCalculated from the amino acid sequence. b ~ ~ ~ ~ - t e r mdomain. inal CNH,-terminal domain. d~ependingon species.
References Davis and Fink, 1990. Loeb et al., 1993; Belanger et al., 1994. Wente et al., 1992; Wimmer et al., 1992. Wente et al., 1992; Wimmer et al., 1992; Fabre et al., 1994. Wente et al., 1992; Wimmer et al., 1992; Wente and Blobel, 1993; Fabre et al., !994. Fabre et al., 199-4; Wente and Blobel, 1994. Grandi et al., 1993.
NPC
perinuclear space
9"^^°
T T ^ c
NPC perinuclear space ' \ ^
?\
perinuclear space * " N
CI 1
^
ox^
C P0M121
NPC
- ^ ^ POM152 (yeast) glycosylatjon sites XePXePXePXeP XFXFG motifs C-G
V—L-G-PF-.- Y motif
Figure 9. Schematic diagram of the domain architecture and membrane topology of the cloned and sequenced integral membrane proteins of the NPC deduced from their amino acid sequences. Three transmembrane NPC glycoproteins, gp210, P O M 1 2 1 , and the yeast POM152, have been thus far cloned and sequenced. They all reveal a distinct stretch of hydrophobic residues that is predicted to be a transmembrane segment transversing the pore membrane. gp210 contains a large NH2-terminal domain residing in the lumen of the NE, and a small COOH-terminal tail associated with the NPC. In contrast, P O M ! 21 consists of a short NH2-terminal tail residing in the lumen of the NE, and a long COOH-terminal domain (with a number of repetitive pentapeptide motifs XFXFG at its COOH-terminal end) associated with the NPC. Very much like gp210, most of the mass of yeast POM152 is predicted to reside in the lumen of the NE. In addition, the amino acid sequence of POM152 contains eight repetitive segments each 24 residues long, with the consensus sequence C-G V— L-G-PF—Y. The N-linked glycosylated residues of gp210 occur atthe lumenal domain close to the nuclear membrane (Greber et al., 1990). By analogy, the N-linked glycosylated sites of POM152 are speculated to be located in the lumenal domain (Wozniak et al., 1994). For more information about these proteins, see Table 2 and references therein. 23
24
NELLY PANTE and UELI AEBI
membrane protein. However, similar to some of the peripheral membrane proteins of the NPC (see below and Table 2), POM 121 binds wheat germ agglutinin (WG A), a lectin that recognizes O-linked A^-acetylglucosamine (GlcNac) residues. Moreover, the cDNA-deduced amino acid sequence of P0M121 has revealed the presence of a repetitive pentapeptide motif XFXFG (where X indicates any amino acid), which is also present in the members of the O-linked NPC glycoprotein family as well as some yeast NPC proteins (see below and Table 2). Although antibodies directed against POM 121 clearly labeled the NPC, the exact location of this protein within the NPC remains to be determined. Since the primary structure of POM 121 has revealed a 44-residue-long transmembrane domain sandwiched between a short NH2-terminal tail (28 residues long) and a long COOH-terminal domain containing the repetitive XFXFG motifs, it has been predicted that the small NHj-terminal tail resides in the lumenal domain (see Fig. 9). Thus, in contrast to gp210, most of the mass of P0M121 is predicted to reside within the NPC proper. Very much like gp210, POM 121 has been proposed to function as a membrane anchor of the NPC. The recent development of a procedure to isolate milligram quantities of NPCs from yeast (Rout and Blobel, 1993) has opened the possibility of more systematically identifying novel yeast NPC proteins. Using this approach, an integral membrane glycoprotein that reacts with ConA has been identified, and cloned and sequenced (Wozniak et al., 1994). The deduced amino acid sequence revealed a 152-kDa protein, termed POM 152, that does not share any similarity with the two vertebrate transmembrane NPC proteins except for a small region (19 residues long), similar to P0M121. Analysis of the amino acid sequence of POM 152 indicated that this protein contains a 20-residue long transmembrane domain between residues 175 and 196. Since the COOH-terminal domain (residues 196— 1337) of POM 152 contains three putative sites for A^-linked glycosylation, at least one of which is glycosylated, in analogy to gp210 (see above), this domain has been proposed to reside in the lumen of the NE (see Fig. 9). However, the exact topology of this protein remains to be established. B. Peripheral Membrane Proteins of the NPC O-Linked Glycoproteins
A group of at least eight peripheral membrane proteins of the NPC has been identified by immunological approaches in rat liver NEs (Davis and Blobel, 1987; Holt et al., 1987; Snow et al., 1987). These proteins, which have molecular masses ranging fi-om 35 to 250 kDa, contain up to 10-20 O-linked A^-acetylglucosamine (GlcNac) residues and therefore bind the lectin WGA. These O-linked glycoproteins appear to be involved in mediated nuclear import, which is inhibited by both monoclonal antibodies to these proteins (Dabauvalle et al., 1988a; Featherstone et al., 1988) and WGA(Finlay et al., 1987; Dabauvalle et al., 1988b). As summarized in Table 2, three of these O-linked glycoproteins have now been cloned and
Nuclear Pore Complex
25
sequenced: p62 (Starr et al, 1990;Carmo-Fonsecaetal., 1991;Cordesetal., 1991), NUP153 (Sukegawa and Blobel, 1993; McMorrow et al., 1994), and rat NUP214, the ~210-kDa glycoprotein originally identified by Snow et al. (1987) and recently demonstrated to be a homologue of human CAN (Kraemer et al., 1994), a putative oncogene product associated with myeloid leukemogenesis (Von Lindern et al., 1992). All three of these proteins contain multiple copies of a more or less degenerate pentapeptide (XFXFG) motif (see Fig. 10), which is considered to be a diagnostic feature for the O-linked NPC glycoprotein family. As illustrated in Figure 10, these repeats are clustered within each protein: within the NH2-terminal half of p62, and within the COOH-terminal domain of NUP153 and CAN. In addition, CAN contains multiple copies of the tripeptide motif FGQ that is also present in two yeast NPC proteins, NUPlOO and NUP116 (see Fig. 14; Wente et al., 1992; Wimmer et al., 1992), and of the degenerate tetrapeptide motif SVFG and the tripeptide motif FGG that have so far not been found in other NPC proteins. CAN also contains a leucine zipper motif that may represent a protein-protein dimerization domain (Von Lindern etal., 1992). As indicated in Figure 10, in addition to the NH2-terminal domain containing multiple copies of the XFXFG motif, the COOH-terminal half of p62 contains heptapeptide repeats characteristic of a-helical coiled-coil conformations. Recently, p62 has been expressed in Escherichia coli and the recombinant protein visualized in the EM after glycerol spraying/rotary metal shadowing (Buss et al., 1994). Accordingly, recombinant p62 appears as a 3 5-nm rod-shaped molecule with a slight protuberance at the NH2-terminal end, thus confirming the a-helical coiled-coil conformation of the COOH-terminal domain. Circular dichroism of recombinant p62 has indicated that the repetitive NH2-terminal domain may have a cross-P conformation (Buss et al., 1994). As we will discuss in Section IVD, within the NPC p62 exists as a complex with at least two other NPC proteins, p58 and p54 (Finlay et al., 1991; Guan et al, 1995). This complex is required for protein import into the nucleus (Finlay et al., 1991). NUP153 is unique among the 0-linked NPC glycoproteins identified and characterized to date in that its primary sequence harbors four zinc finger motifs, each containing two pairs of cysteine residues (Cys2-Cys2) (Sukegawa and Blobel, 1993; McMorrow et al., 1994). Since this type of zinc finger motif is found in DNA-binding proteins (reviewed by Coleman, 1992), a fragment of NUP153 containing these four motifs was expressed in E. coli and demonstrated to bind DNA in a zincdependent manner (Sukegawa and Blobel, 1993). This result has given rise to speculations about a possible role of NUP 153 in gating transcribable genes to the NPC, thus facilitating export of the transcribed RNA (Sukegawa and Blobel, 1993). 3'D Localization of Some of the O-Linked Glycoproteins Since WGA specifically binds to the O-linked NPC glycoproteins, several attempts have been made to localize these polypeptides by labeling NPCs with
a) 0-linked glycoproteins
N
b) Non-0-linkedproteins
~c
NUP107
H
NI I . . . 1 C Tprtp265 ,
CS/irwW
coiled-mila-helix
v
XFXFG motifs
q
XFXFG motifs alemating with
R SVFG, FGQand FGG motifs Leucine zipper motif
Figure 10. Schematic diagram of the domain architecture of the cloned and sequenced peripheral membrane proteins of the NPC deduced from their amino acid sequence. Depending on the content of Olinked N-acetylglucosamine (GlcNac) residues, two families of peripheral membrane proteins of the NPC have been distinguished: (a) 0-linked glycoproteins that contain several copies of a more or less degenerate pentapeptide motif XFXFG, and (b) non-@linked proteins that do not contain any repetitive sequence motifs. In addition, p62 ,contains a COOH-terminal a-helical coiled-coil domain, and NUP153 harbors four zinc finger motifs. In the case of CAN/NUP214/p250, the repetitive XFXFG motif alternates with repetitive SVFG, FGQ, and FGG motifs. The amino acid sequences of the three members of the non-Olinked protein family are unique. In the case of Tprlp265, it contains a very long (1600 residues) a-helical coiled-coil domain near its NH2-terminal end. For more information about these proteins, see Table 2 and references therein.
Nuclear Pore Complex
27
ferritin-tagged (Finlay et al., 1987; Hanover et al., 1987) or colloidal gold-tagged WGA(Akey and Goldfarb, 1989; Pante and Aebi, 1993; Pante et al., 1994). These labeling studies have revealed that WGA specifically binds to the NPC, and they have localized at least one of its binding sites to the terminal ring of the nuclear basket (Pante and Aebi, 1993; Pante et al., 1994). In projection, additional WGA binding sites have been identified at two distinct radial locations: (i) at low radii between 3.6 and 12.7 nm, and (ii) at higher radii between 18.2 and 36.2 nm (Pante e t a l , 1994). Using antibodies to individual NPC proteins, immunofluorescence microscopy and immuno-EM studies have localized these to the NPC. However, the relatively strong cross-reactivity of these antibodies has made the localization of individual proteins within the NPC difficult. For example, this protein was symmetrically located at both the nuclear and cytoplasmic faces of mouse liver NPCs by using a polyclonal mouse anti-p62 antibody (Cordes et al., 1991), whereas the same antibody labeled only the nucleoplasmic face of Xenopus oocyte NPCs (Cordes et al., 1991; Pante and Aebi, 1993). Thus the localization of p62 has remained ambiguous. To resolve this ambiguity, we have recently produced a monoclonal antibody, RL31, which reacts specifically with rat p62 (Guan et al., 1995). As illustrated in Figure 11, RL31 predominantly labels the central plug or channel complex of rat liver NPCs, both its nuclear and cytoplasmic periphery, although the labeling at its nuclear face is more frequent. By the use of a polyclonal antibody raised against a fusion protein expressed from a NUP153 cDNA construct, NUP153 has been unequivocally localized to the nuclear periphery of the NPC (Sukegawa and Blobel, 1993). By the same approach, CAN/NUP214 has been localized to the cytoplasmic periphery of the NPC (Kraemer et al., 1994). However, in these studies there was no clear identification of a particular NPC component(s) labeled by these antibodies. Recently, more specific localization of these two proteins has been reported. Using an antibody raised against an extract of nuclear matrix proteins that recognizes NUP153, Cordes et al. (1993) have localized this protein to intranuclear NPC-attached filaments, which among other structures may represent nuclear baskets that have been disrupted during sample preparation. More specifically, Pante et al. (1994) have identified this protein as a constituent of the nuclear basket with at least one of its epitopes residing in the terminal ring (see below). Accordingly, as documented in Figure 12, a monoclonal antibody, termed QE5, that by immunoblotting recognized three O-linked glycoproteins (p62, NUP 153, and p250), labeled both the cytoplasmic and the nuclear peripheries of Xenopus oocyte NPCs. On the cytoplasmic face of the NPC gold-conjugated QE5 labeled (i) the cytoplasmic filaments and (ii) sites down in the pore disposed toward the nuclear opening (Fig. 12, a and c). As can be seen in Figure 12, b and c, at the nuclear periphery, gold-conjugated QE5 labeled predominantly the nuclear baskets. To identify these distinct labeling sites with the different NPC proteins recognized by QE5, Pante et al. (1994) have used an anti-peptide antibody against human NUP 153 and a monospecific anti-p250 poly-
28
NELLY PANTE and UELI AEBI
100 nm Figure 11. Localization of the rat NPC protein p62 by immunoelectron microscopy. Isolated rat liver NEs were incubated with the monoclonal antibody RL31 (which recognizes rat p62 on Western blots), conjugated to 8-nm colloidal gold, and prepared for EM by embedding and thin sectioning. Shown are a view along a single cross-sectioned NE stretch together with a gallery of labeled NPCs cross sections. These cross sections revealed that RL31 predominantly labels the central plug or channel complex of rat liver NPCs, both its nuclear and cytoplasmic periphery, although the labeling at the nuclear side appears to be more frequent, c, cytoplasmic; n, nuclear side of the NE. Scale bar, 100 nm from Guan et al. (1995).
clonal antibody. As illustrated in Figure 12, c-e, labeling with these two antibodies revealed that NUP153 is a constituent of the terminal ring of the nuclear basket, whereas p250 is a constituent of the cytoplasmic filaments. Localization of NUP153 at the nuclear basket together with its four zinc finger motifs is consistent with the stabilizing effect of Zn^"^ on the nuclear baskets (see Fig. 6; Jamik and Aebi, 1991). The '-250-kDa glycoprotein recognized by QE5 also reacts with the RLl monoclonal antibody used by Snow et al. (1987) (B. Burke and R. Bastos, personal Figure 12. Localization of CAN/NUP214/p250 and NUP153 by immunoelectron microscopy, (a) Cytoplasmic and (b) nuclear faces of quick-frozen/freeze-dried/rotary metal-shadowed spread Xenopus oocyte NEs labeled with the monoclonal QE5 antibody conjugated to 8-nm colloidal gold. QE5, which recognizes p250, NLJP153, and p62 on Western blots, specifically labeled the cytoplasmic periphery of the NPC at (i) the cytoplasmic filaments (a, arrowheads) and (ii) sites down in the pore disposed toward the nuclear opening, (continued)
Nuclear Pore Complex
29
Figure 12: (continued) At the nuclear periphery of the NPC, the gold-conjugated QE5 labeled predominantly the nuclear baskets (b, arrowheads), (c) Gallery of selected examples of quick-frozen/freeze-dried/rotary metal-shadowed NPCs labeled with the monoclonal QE5 antibody, a polyclonal antibody against p250, and an anti-peptide antibody against human NUP153. The anti-p250 antibody exclusively labeled the cytoplasmic filaments, whereas the anti-NUP153 anti-peptide antibody exclusively labeled the terminal ring of the nuclear baskets, (d, e) cross sections and selected examples of Epon-embedded, Triton X-100-treated Xenopus oocyte nuclei labeled with antl-p250 antibody (d) and anti-NUP153 anti-peptide antibody (e). In agreement with the quick freezing/freeze drying/rotary metal shadowing results shown In c, these cross sections documented that the anti-p250 antibody exclusively labeled the cytoplasmic filaments, whereas the antl-NUP153 anti-peptide antibody exclusively labeled the terminal ring of the nuclear baskets, c, cytoplasmic; n, nuclear side of the NE. Scale bar, 100 nm (a-e). Adapted from Pante et al. (1994).
30
NELLY PANTE and UEUAEBI
100 nm Figure 13. Localization of Tpr/p265 by immunoeiectron microscopy. Isolated rat liver NEs were incubated with the monoclonal RL30 antibody (which specifically reacts with rat p265) conjugated to 8-nm colloidal gold and salt-washed with 0.5 M NaCI prior to embedding and thin sectioning. Shown are a view along a single cross-sectioned NE stretch together with a gallery of labeled NPCs cross sections. These cross sections revealed that RL30 exclusively labeled the cytoplasmic side of the NE. Gold particles were found to be associated predominantly with the cytoplasmic filaments of the NPC, which in some cases bent around and reached down into the central pore (see arrowhead), c, cytoplasmic; n, nuclear side of the NE. Scale bar, 100 nm. Adapted from Byrd et al. (1994).
communication). In addition, p250 is recognized by polyclonal antibodies raised against the cloned NH2- and COOH-terminal domain of human CAN (B. Burke and R. Bastos, personal communication). Therefore, p250 corresponds to the (9-linked glycoprotein CAN/NUP214. Non-O'Linked Proteins A group of at least 30 proteins that do not contain GlcNac have recently been identified in rat liver NEs (Radu et al., 1993). This group of proteins has been separated from the O-linked NPC glycoproteins by WGA-Sepharose chromatography, and they have been further isolated by SDS-4iydroxylapatite chromatography (Radu et al., 1993). However, so far only two of these polypeptides, NUP155 (Radu et al., 1993) and NUP107 (Radu et al., 1994), have been demonstrated to be bona fide NPC proteins. Both proteins have been cloned and sequenced, and their
Nuclear Pore Complex
31
deduced amino acid sequences do not reveal any of the repetitive sequence motifs (i.e., the XFXFG motif), which seem to be a diagnostic feature for the 0-linked NPC glycoproteins (see Table 2). As illustrated schematically in Figure 10, NUP107 contains a leucine zipper motif at its COOH-terminal end, which has been suggested to cause dimerization with a second leucine-zipper-containing polypeptide (Radu et al, 1994). Anti-peptide antibodies against both NUP 155 (Radu et al., 1993) and NUP 107 (Radu et al., 1994) were raised and used to label several types of cultured cells by immunofluorescence microscopy and immuno-EM. Both antibodies labeled the NPCs of these cells, thus confining that NUP 155 and NUP 107 are bona fide NPC proteins. However, localization of these two proteins to distinct NPC components remains to be determined. By using an autoimmune serum from a patient with overlap connective tissue disease, a NPC polypeptide of 180 kDa that does not bind WGA has recently been identified (Wilken et al., 1993). Affinity-purified antibodies from this serum were used to \3bQ\Xen0pus oocyte NEs (Wilken et al., 1993). Accordingly, this antibody bound to the cytoplasmic ring and associated fibers of the NPC, Using both monoclonal and autoimmune antibodies, Byrd et al. (1994) have identified a 265-kDa non-0-linked NPC protein in rat liver NEs. As documented in Figure 13, by immuno-EM this protein localizes to the cytoplasmic filaments of the NPC. It has further been shown that p265 is very prone to proteolysis with a major 175-kDa proteolytic product (Byrd et al., 1994), which could be identical to the 180-kDa polypeptide identified by Wilken et al. (1993). Supporting this notion, the autoimmune antibody used by Wilken et al. (1993) also reacted with a '-260-kDa protein. Most interestingly, amino acid sequences of peptides from p265 and its proteolytic fragment pi 75 revealed 80-90% identity with the primary sequence of human Tpr (translocated promoter region), a '-265-kDa protein whose NH2-terminal domain appears in oncogenic fusions with the met, trk, and ra/protooncogenes (Mitchell and Cooper, 1992). Therefore, p265 is the rat homologue of human Tpr (Byrd et al., 1994). Consistent with being a non-O-linked NPC protein, the amino acid sequence of Tpr lacks the repetitive XFXFG pentapeptide motif diagnostic for the O-linked NPC glycoproteins (see Table 2). In addition, Tpr contains a predicted a-helical coiled-coil region over 1600 residues long (see Fig. 11; Mitchell and Cooper, 1992). Based on this secondary structure, it is conceivable that Tpr, together with p250 (see Fig. 12; Pante et al., 1994), forms the backbone of the cytoplasmic filaments. However, because of its size Tpr may also extend into the cytoplasmic ring. C. Yeast NPC Proteins
The use of antibodies against vertebrate NPC proteins in concert with the design of genetic screens has allowed identification of at least nine yeast NPC proteins (see Table 2; reviewed by Fabre and Hurt, 1994). The number of these are expected to rapidly increase, now that a procedure to bulk isolate NPCs from yeast has been developed (Rout and Blobel, 1993). As mentioned above, one of these yeast NPC
32
NELLY PANTE and UELI AEBI
a) XFXFG family ls\Jiodlpo€i€ta€iaacifSj/^^
Nl
NSP1
)DOOOOOCX30000C3|C»Q|
Nl
C
loooooooooc^
NUP1
NUP2
b) GLFG family Ntocwaadl
^c
NUP/NSP49
Ic
NUP/NSP100
N[
NUP/NSP116
NkH^ms^i^mm^imim
N)^ssmmii
JC Nl
1
c) Nucleoporin interacting component N f ^ ^:
\mmm\
coiled-coila-helix
jpocHBocj XFXFG motifs
n C
>^S=i«»ia
NIC96
GLFG motifs
J^WHSIWSBi I
I
related domains
RNA-binding motif
Figure 14, Schematic diagram of the domain architecture of the cloned and sequenced yeast NPC proteins deduced from their amino acid sequence. Depending on the occurrence of highly repeated motifs in their amino acid sequence, three families of yeast NPC proteins have been distinguished: (a) the XFXFG family, which contains several copies of a more or less degenerate pentapeptide motif XFXFG clustered in the central part of each protein; (b) the GLFG family, which contains several copies of a degenerate tetrapeptide GLFG within their NH2-terminai domain; and (c) the third family includes the yeast nucleoporin interacting component NIC96, which does not contain any repetitive sequence motifs and forms a complex with NSP1 and NUP49 (Grandi et al., 1993). Three members of the second family, NUP100, NUP116, and NUP145, contain related domains, including an RNA-binding motif. For more information about these proteins, see Table 2 and references therein.
Nuclear Pore Complex
33
proteins, POM 152, is an integral membrane protein. The rest of them can be classified into three groups based on the occurrence of highly repeated sequence motifs (see Table 2 and Fig. 14). First, the XFXFG family contains several copies of a more or less degenerate pentapeptide motif (XFXFG) clustered in the central part of each protein (see Fig. 14a). This XFXFG motif is characteristic of the vertebrate O-linked NPC glycoproteins (see above and Fig. 10), but it is not clear whether any of these yeast NPC proteins are in fact glycosylated. Members of this group include NSPl (Hurt, 1988), NUPl (Davis and Fink, 1990), and NUP2(Loeb et al., 1993). A second group represents the GLFG family, which contains multiple copies of a degenerate tetrapeptide motif (GLFG) within their NH2-terminal domain (see Fig. 14b). Members of this group include NUP/NSP49, NUP/NSP100, NUP/NSPl 16 (Wente et al., 1992; Wimmer et al., 1992), as well as NUP/NSP 145 (Fabre et al., 1994; Wente and Blobel, 1994). Finally, the third group is defined by the yeast nucleoporin interacting component NIC96 (see Fig. 14c), which does not exhibit any of these repetitive sequence motifs (Grandi et al., 1993). Although most of the vertebrate and yeast NPC proteins contain short repetitive sequence motifs, in particular the XFXFG motif, no obvious relationship has yet been depicted between these vertebrate and yeast proteins. The only exception is NSP1, which is considered to be the yeast homologue of vertebrate p62: both proteins have the same domain structure (see Figs. 10 and 14), with 50% similarity in their COOHterminal domains (Carmo-Fonseca et al, 1991; Fabre and Hurt, 1994). Three members of the GLFG family, three proteins, NUP/NSP 100, NUP/NSP 116, and NUP/NSP 145, contain highly homologous sequence regions (see Fig. 14b) including an RNA-binding motif (Fabre et al., 1994). Fragments of NUP/NSPl 16 and NUP/NSP 145 including the RNA-binding motif were expressed as fusion proteins in E. coli, and these were demonstrated to bind RNA in vitro (Fabre et al., 1994). Based on these results, it has been suggested that NUP/NSP 116 and NUP/NSP 145 may play a role in RNA recognition and/or transport through the NPC (Fabre and Hurt, 1994; Fabre et al., 1994). Whereas antibodies directed against some of these yeast NPC proteins do label the NPC by immuno-EM, the exact localization of their epitopes has not yet been established. One of the difficulties in determining their localization is that the 3-D structure of the NPC in embedded/thin-sectioned yeast NEs or nuclei is poorly preserved. On the other hand, the yeast system has the advantage that the phenotype of a mutation in a particular NPC protein can be directly characterized. As a consequence, the genes for several yeast NPC proteins have been shown to be essential for cell growth (reviewed by Fabre and Hurt, 1994). Yeast mutants can also be examined in the EM to determine whether they perturb the structure of the NPC and/or the NE. For example, Wente and Blobel (1993,1994) have investigated the phenotype after deletion/disruption of the yeast NPC proteins NUP/NSPl 16 and NUP/NSP 145. Accordingly, a membrane seal was formed over the cytoplasmic face of the NUP/NSPl 16-deficient NPCs, which did not block nuclear export but caused the export substrate to accumulate within the cytoplasmic membrane
NELLY PANTE and UELI AEBI
34 Tpr/NUP180
CAN / p250
POM121
p62 Complex
NUP153
Figure 15. Schematic diagram summarizing the immunolocalization of characterized NPC protein epitopes within the 3-D architecture of the consensus model of the NPC. Tpr/p265 and CAN/NUP214/p250 exhibit epitopes at the cytoplasmic filaments (see Figures 12 and 13), whereas NUP153 exhibits an epitope at the terminal ring of the nuclear basket (see Figure 12). p62 epitopes are exposed at both the cytoplasmic and nuclear peripheries of the central plug or channel complex (see Figure 11). The transmembrane glycoprotein gp210 exhibits several epitopes in the lumen of the NE (Greber et al., 1990) where, based on its topology, most of its mass resides (see Figure 9; Greber et al., 1990). Epitopes for the transmembrane glycoproteins POM121 and POM152 are also shown, but they should not be taken literally because their exact localization remains to be determined. In contrast to gp210, most of the mass of POM121 is predicted to reside within the NPC proper. Very much like gp210, most of the mass of yeast POM152 is predicted to reside In the lumen of the NE.
herniations covering the NPCs (Wente and Blobel, 1993). Furthermore, deletion/disruption of the NH2-terminal end of NUP/NSP145 yielded yeast nuclei with clusters of numerous NPCs interconnected by a network of NE herniations (Wente and Blobel, 1994). Based on these results, it has been proposed that NUP/NSPl 16 and NUP/NSP145 are possibly involved in establishing specific NPC—NE interactions and/or mediating NPC biogenesis (Wente and Blobel, 1993; 1994). In summary, only 17 (8 vertebrate and 9 yeast) of the approximately 100 polypeptides of the NPC have thus far been identified, characterized, and cloned and sequenced (see Table 2). As illustrated in Figure 15, epitopes of 7 of these NPC
Nuclear Pore Complex
35
proteins have been localized within the 3-D NPC architecture. Since the native conformation of none of these proteins is currently known, it is difficult to map the entire protein within the NPC. However, identification of some of their epitopes with distinct structural components of the NPC has been the first step toward elucidation of the molecular architecture of the NPC (see Fig. 15). D. Isolation and Characterization of Distinct NPC Components
When assembled in the NPC, different NPC proteins may mutually interact, thus forming distinct subcomplexes within the NPC. In vertebrate species, it was first reported that some of the soluble NPC proteins contained in in vitro nuclear reconstitution extracts from Xenopus oocytes form a large supramolecular complex with a molecular mass of 254 kDa, which contains p68, the Xenopus homologue of rat p62, interacting with other NPC proteins (Dabauvalle et al., 1990). This complex has also been isolated and characterized at the molecular level from rat liver NEs (Finlay et al., 1991; Kita et al., 1993). Accordingly, it consists of p62 interacting with two other proteins of molecular mass 58 (p58) and 54 (p54) kDa. Finlay et al. (1991) estimated the mass of the p62 complex to be 550-600 kDa with a molar stoichiometry of 4:1:1 (p62:p58:p54). However, Kita et al. (1993) have reported a mass of 231 kDa and a molar stoichiometry of 1:1:2 (p62:p58:p54). Thus the molecular mass and molar stoichiometry of the p62 complex remain controversial. More recently, Guan et al. (1995) have developed a modified procedure to isolate the p62 complex from rat liver NEs. Accordingly, in addition to p58 and p54, p62 is associated with an additional NPC protein, p45. Mass analysis by quantitative STEM has revealed that the native p62 complex is a ~200-kDa particle containing one copy each of p62, p58, p54, and p45 (Guan et al., 1995). This "minimal" p62 complex is able to self-associate into higher order oligomers. Using the monoclonaaibody, Pante et al. (1994) have identified, in addition to the p62 complex, two distinct NPC complexes in extracts of BHK cells. Although this antibody recognizes p62, NUP153, and p250 on Western blots, it immunoprecipitates three additional polypeptides (p54, p58 and p75) that were found to be associated with p62 (p54 and p58) and with p250 (p75). In addition, NUP153 was released as a homo-oligomer of >1 MDa, most likely representing an octamer. Because several of its epitopes are located on the terminal ring of the nuclear baskets (see Fig. 12), it is conceivable that the NUP153 octamer defines the basic framework of the octameric terminal ring. Since the long a-helical COOH-terminal domain of yeast NSPl directs cytosolic proteins to the NPC, it was proposed that this protein might interact with other yeast NPC proteins (Hurt, 1990). Indeed, NSPl has now been shown to form a complex with three other proteins, NIC96, NSP49, and a novel yeast NPC protein of 54 kDa that has not yet been cloned and sequenced (Grandi et al., 1993). A second NPC complex has recently been identified in yeast (Belanger et al, 1994). Accordingly, it consists of the yeast proteins NUP1 and NUP2 interacting with Srpl, the product
36
NELLY PANTE and UELI AEBI
of a gene previously identified as a suppressor of mutants defective in RNA polymerase I (Yano et al., 1992).
V. MOLECULAR TRAFFICKING THROUGH THE NPC As illustrated schematically in Figure 1, two different types of nucleocytoplasmic transport across the NPC occur: (i) passive diffusion of ions and small molecules through an aqueous channel with a physical diameter of--9 nm (Paine et al., 1975): and (ii) mediated transport of proteins, RNAs, and RNP particles through a gated channel with a functional diameter of up to 26 nm (Feldherr et al., 1984). Although the mediated nucleocytoplasmic transport of different substrates uses the same machine, depending on the substrate it appears to occur via different signal pathways. Some of the signals, receptors, and factors mediating nuclear import of proteins or RNPs and nuclear export of RNAs have begun to be identified and isolated. In this section we focus on some recent advances made toward identification and characterization of different signals and factors required for molecular trafficking trough the NPC. For background information and further details, the reader is referred to some recent reviews covering this topic (Forbes, 1992; Gerace, 1992; Izaurralde and Mattaj, 1992; Mattaj et al., 1993; Newmeyer, 1993). A. Passive Diffusion
Early microinjection experiments of dextrans have demonstrated that the NPC has the properties of a molecular sieve (Paine et al., 1975). Accordingly, molecules larger than ^9 nm are excluded from the nucleus, while smaller molecules can passively diffuse across the NPC with a rate inversely proportional to their size. The recent 3-D reconstruction of the NPC (see Fig. 4; Hinshaw et al., 1992) from negatively stained preparations of detergent-treated Xe«o/7M5 oocyte NEs (see Figs. 2d and 3b) has indicated that there may exist eight ^ 10-nm diameter, slightly kinked channels at a radius of-40 nm that are located between two adjacent spokes and the pore membrane (see Section IIIB and Fig. 4a). These peripheral channels have been proposed to be sites for passive diffusion of ions and small molecules through the NPC (Hinshaw et al., 1992). In contrast, the 3-D reconstruction of ice-embedded NPCs (Akey and Radermacher, 1993) has revealed eight ~ 10-nm diameter channels at a radius of-32 nm located between two adjacent spokes and the central channel complex; these channels have also been speculated to represent sites for passive diffusion (Akey and Radermacher, 1993). Thus the location of these diffusional channels remains controversial. B. Import of Nuclear Proteins
Nuclear import of proteins has been the most extensively studied nuclear transport process (review by Forbes, 1992; Gerace, 1992; Newmeyer, 1993). In vitro and in vivo studies have demonstrated that nuclear protein import is highly selective, and requires ATP and cytosolic factors (Feldherr et al., 1984; Newmeyer et al..
Nuclear Pore Complex
37
1986; Richardson et al., 1988; Adam et al., 1990; Newmeyer and Forbes, 1990). Targeting of nuclear proteins to the NPC is specified by short amino acid sequences, called nuclear localization signals (NLSs), on the protein to be transported (reviewed by Dingwall and Laskey, 1991; Garcia-Bustos et al., 1991). Many studies have focused on the characterization of NLSs and identification of receptors that interact with NLSs to mediate import of nuclear proteins. As a consequence, two types of NLSs have been identified: (i) the simian virus 40 (SV40) large T-antigen type of NLS, which consists of a single contiguous stretch of basic amino acid residues (Chelsky et al., 1989); and (ii) the Xenopus nucleoplasmin type of bipartite NLS, which contains two interdependent basic domains separated by 10 intervening "spacer" residues (Robbing et al., 1991). A number of NLS-binding proteins have also been identified (reviewed by Gerace, 1992). However, in addition to a NLS-binding protein, cytosolic factors are required in order to stimulate mediated nuclear import of a protein in vitro (Adam et al, 1990; Newmeyer and Forbes, 1990). Recently, considerable effort has been made to identify, isolate, and characterize the cytosolic factors essential for nuclear import of proteins. Using a cell-free in vitro assay for nuclear protein import (i.e., Xenopus egg extracts, nuclei from any source, and a fluorescently labeled nuclear protein; Newmeyer et al., 1986), Newmeyer and Forbes (1990) have identified two cytosolic factors, one required for ATP-independent binding of proteins to the NPC, and the second one for translocation of the substrate through the NPC. The development of another in vitro transport assay consisting of digitonin-permeabilized cells (digitonin permeabilizes only the plasma membrane while leaving the NE intact) supplemented with exogenous cytosol and ATP (Adam et al., 1990) has demonstrated the requirement of multiple cytosolic factors for nuclear import of proteins (Adam et al., 1990; Moore and Blobel, 1992). Four cytosolic factors essential for import of nuclear proteins have recently been identified with this assay: (1) two NLS binding proteins of 54 and 56 kDa that have the properties of a functional import receptor (Adam et al., 1989; Adam and Gerace, 1991); (2) a component that interacts with O-linked glycoproteins of the NPC (Steme-Marr et al., 1992); (3) the ubiquitous cellular protein hsc70 (Imamoto et al., 1992; Shi and Thomas, 1992); and (4) the small GTPase Ran/TC4 (Melchior et al., 1993; Moore and Blobel, 1993). The latter has to be in an "active," GTP-bound state to mediate nuclear import, which is inhibited by GTP-y-S and other nonhydrolyzable GTP analogues (Melchior et al., 1993; Moore and Blobel, 1993). In addition to these four cytosolic factors, Adam and Adam (1994) have recently identified another cytosolic factor that binds to the NLS receptor, thereby mediating the binding of the protein-receptor complex to the NPC, but does not mediate the translocation step. The relationship of all these cytosolic factors, as well as their site(s) and mechanism of action, remains to be established. Although these factors are predominantly cytosolic proteins, some of them (i.e., the 54/56 NLS receptor, the hsc70 protein, and Ran/TC4) are also found in the nucleus, suggesting that they may function as shuttling carriers and thus may be recycled for several rounds of transport (Adam et al., 1989; Adam and Gerace, 1991; Gerace, 1992).
38
NELLY PANTE and UELI AEBI
Originally, nuclear import of proteins has been described as a two-step process: (1) binding of the nuclear protein to the NPC and (2) translocation of the NPCbound protein through the NPC (Newmeyer and Forbes, 1988; Richardson et al., 1988). The first step does not require ATP and is temperature independent, whereas the second step requires energy via ATP hydrolysis. However, as illustrated in Figure 16, the recent advances on identifying and characterizing NLSs, NLS receptors, and cytosolic factors suggest that the NPC-mediated transport pathway of proteins into the nucleus occurs by at least five distinct steps: (1) While in the cytoplasm, the protein to be imported is complexed to a cytosolic NLS-binding receptor via its specific NLS (Newmeyer and Forbes, 1988; Adam and Gerace, 1991), and this interaction may be stabilized by some cytosolic factor(s) (Adam and Adam, 1994). (2) Depending on the action of additional cytosolic factors, this protein-receptor complex then docks to an NPC by specific binding to some "peripheral" NPC component such as the cytoplasmic ring or the cytoplasmic filaments (Richardson et al., 1988; Steme-Marr et al., 1992). (3) From this peripheral docking site the protein-receptor complex is next delivered to the central channel complex, which harbors the actual transport machine. (4) Active translocation of the protein-receptor complex through the central channel complex occurs after channel gating to accommodate the particular size and shape of the proteinreceptor complex. (5) After release into the nucleus, the protein-^*eceptor complex dissociates, and the receptor may be recycled for further rounds of transport (Adam et al., 1989; Adam and Gerace, 1991). Some of these nuclear protein import steps are still hypothetical, and several issues remain elusive. For example, the site(s) and mechanism of ATP utilization, the site(s) and mechanism of action of the different cytosolic factors, the NPC ligands and the role of specific NPC components involved in the different transport steps, and last but not least, the nature and molecular mechanism of the gated channel. Thus although a number of functional aspects of the mediated import of proteins into the nucleus have now been established, there remain many questions and ambiguities to be resolved before the detailed molecular mechanism implicated in this process will be understood. C. Export/Import of RNAs and RNP Particles
In addition to nuclear proteins, the NPC also imports ribonucleoprotein (RNP) complexes into the nucleus. The import of U-rich small nuclear ribonucleoprotein (U snRNP) particles which contain a 5' trimethyl G cap and are complexed with proteins termed Sm has been studied in some detail (reviewed by Izaurralde and Mattaj, 1992; Mattaj et al., 1993). For example, nuclear import of U6 snRNP particles is inhibited by proteins bearing the SV40 T antigen NLS (Michaud and Goldfard, 1991); thus nuclear import of this particle seems to occur by a mechanism identical (or at least very similar) to that of nuclear import of proteins (see above). However, in the case of Ul, U2, U3, U4, and U5 snRNP particles, their nuclear import is not inhibited by proteins bearing an NLS, but it requires both the trimethyl
• : protein ;
^
%-
M: receptor
1. formation of proteinreceptor complex
Blj! ^
ilMii ^
2. docking of protein-receptor complex to NPC periphery (cytoplasmic filaments and/or cytoplasmic ring)
3. delivery of protein-receptor complex to central channel complex
liiii J Jio4 ^ * '"^^
^* ^*^^'^'^®' 9^^^^9 ^^^ translocation of protein-receptor complex
my ~^ n^l
i'
7
5. release and dissociation of protein-receptor complex
Figure 16. Schematic diagram of the nuclear import pathway of proteins through the NPC. In the first step, the protein to be transported associates with an NLS receptor this complex then docks to the cytoplasmic periphery (i.e., to the cytoplasmic filaments or the cytoplasmic ring) of an NPC, from where it is delivered to the central plug or channel complex for translocation. Each one of these steps appears to be mediated by the action of one or several cytosolic factors. Once in the nucleus, the protein-receptor complex dissociates, and the receptor may be recycled for another round of transport. Adapted from Gerace (1992). 39
40
NELLY PANTE and UELI AEBI
G cap and binding of Sm proteins (Fischer and Luhrmann, 1990; Hamm et al., 1990; Michaud and Goldfard, 1992; Fischer et al., 1993). Moreover, nuclear import of the U3 snRNP particle is not inhibited by an excess of free trimethyl G cap or by proteins bearing an NLS (Michaud and Goldfard, 1992). Thus there appear to exist at least three distinct signaling pathways for import of snRNP particles. Recently, Marshallsay and Luhrmann (1994), using the digitonin-permeabilized cell system developed by Adam et al. (1990), have demonstrated that nuclear import of RNP particles requires cytosolic factors, as does nuclear import of proteins (see Section VB). This approach now opens the possibility of identifying and characterizing both the specific signals and factors mediating the nuclear import of RNP particles. Compared with the nuclear import of proteins and RNP particles, the export of RNAs and RNP particles is still rather poorly understood. The NPC exports several classes of RNAs, including snRNAs, mRNAs, and tRNAs. Since they are packaged with proteins into RNP complexes, these different RNAs are probably exported in the form of RNP particles. Indeed, export of RNP particles through the NPC has been visualized by EM (Stevens and Swift, 1966; Mehlin et al., 1992). For example, export of Balbiani ring granules (premessenger RNP particles in the salivary glands of Chironomus) seems to be a polar process in that the 5' end of its RNA exits the nucleus first (Mehlin et al., 1992). Similar to nuclear protein import, export of RNAs from the nucleus is a signal-dependent, receptor-mediated process that requires energy in the form of ATP hydrolysis (Zasloff, 1983; Bataille et al., 1990; Dargemont and Kuhn, 1992). However, since the export of RNAs may be controlled at multiple levels within the nucleus, it has been difficult to determine the signals involved in this process and the nuclear site(s) where these signals exert their effect. Nevertheless, some of the signals and factors mediating nuclear export of RNAs and RNP particles are starting to emerge. In the case of mRNAs and snRNAs, it has been shown that the monomethylated RNA cap structures facilitate their nuclear export; thus the signal(s) may reside on the primary structure of the RNA (Hamm and Mattaj, 1990). Moreover, a nuclear cap-binding protein that might play a role similar to the NLS receptors in nuclear protein import (see Section VB) has been identified (Izaurralde et al., 1992). More recently, Jarmolowski et al. (1994) have demonstrated that the transport of various classes of RNAs is mediated by distinct rather than common essential factors. Thus export of RNAs and RNP particles may occur by different signaling pathways, as does the nuclear import of proteins and RNP particles (see above).
V. CONCLUSIONS AND FUTURE PROSPECTS Considering the size and complexity of the NPC, significant progress has been made over the past few years in determining its 3-D architecture and molecular composition, and in describing the mechanisms of mediated transport through the NPC. As a consequence, the 3-D structure of its basic framework has now been
Nuclear Pore Complex
41
determined to a resolution of just under 10 nm (see Fig. 4). Several distinct peripheral NPC components such as the cytoplasmic and nuclear ring, the cytoplasmic filaments and the nuclear basket have been identified (see Fig. 8); and about two dozen of its protein constituents have been characterized and are starting to be localized within the 3-D structure of the NPC (see Fig. 15). Some of the signals, receptors, and factors mediating nuclear import of proteins and RNP particles and nuclear export of RNAs and RNP particles have also been identified. Nevertheless, there remain a number of questions concerning the 3-D structure, chemical composition, and functional role(s) of the different structural components of the NPC. Recently there has been some progress in localizing the epitopes of several NPC proteins with distinct structural components of the NPC by immuno-EM (see Figs. 11—13 and 15). Thus the molecular architecture of the NPC is slowly but definitely emerging. However, even if there are several copies (i.e., 8 or 16 because of the 8-2-2 symmetry of the basic framework of the NPC) of the NPC proteins thus far identified, they represent only --15% of the NPC mass. Therefore, we have a long way to go before the complete architecture of the NPC will be unveiled at the molecular level. Toward this goal, the recent success of isolating NPCs in bulk from yeast (Rout and Blobel, 1993) has opened the possibility of more systematically identifying the protein constituents of yeast NPCs. Most importantly, this system offers the advantage of combining molecular genetics approaches with biochemical, structural, and functional analyses of the NPC. The next step toward a more complete molecular architecture of the NPC will be not only to identify NPC proteins with distinct NPC components, but to map individual proteins within the 3-D structure of the NPC and to determine their conformation and specific interactions with other proteins residing within distinct NPC components. Moreover, to eventually reconstitute functional NPCs in vitro, we also have to isolate and molecularly characterize distinct NPC components (e.g., the spoke complex, the cytoplasmic and nuclear rings, the cytoplasmic filaments, the nuclear basket, and the central plug or channel complex), determine their 3-D molecular architecture, and decipher their functional roles in terms of distinct steps implicated in mediated nucleocytoplasmic transport. Toward this goal, several NPC subcomplexes have recently been identified: i.e., the p62 complex, the p250-p75 complex, the NUP153 homo-oligomer, the yeast NSPl complex, and the yeast NUPl—NUP2-Srpl complex (see Section IVD). However, these complexes remain to be further characterized at the structural and molecular levels. The recent chemical characterization of some of the peripheral components of the NPC (i.e., the cytoplasmic filaments and the nuclear basket; see Figs. 11—13 and 15) opens the possibility of testing their functional properties and determining the functional roles of distinct NPC components. For example, the hypothesis that the cytoplasmic filaments might be initial docking sites for nuclear import of NLS-bearing proteins (see Fig. 16) can now be tested by the use of antibodies directed against the constituent proteins of these filaments (i.e., Tpr/p265 and CAN/NUP214/p250). Similarly, the role of the nuclear basket in the nuclear export
42
NELLY PANTE and UELI AEBI
of RNAs and RNP particles can be tested by using antibodies directed against NUP153, a constituent of the nuclear basket. In the effort to understand the mechanism of mediated nuclear import of proteins and RNP particles, the development of a digitonin-permeabilized cell system to study these processes in vitro (Adam et al., 1990) has made possible the identification of a number of signals, receptors, and factors mediating nuclear import of proteins and RNP particles. However, the NPC site(s) where they exert their mechanism of action remain to be established. Similarly, the signals and factors that mediate nuclear export of RNAs have begun to be elucidated, but the nuclear site(s) where they exert their effect is unclear. Nevertheless, these results have revealed the existence of multiple nuclear import-export pathways that might be specific for certain cell types or stages of differentiation. Finally, the recent success in imaging and at the same time manipulating NPCs in their native buffer environment by scanning force microscopy (see Fig. 5, c and d) provides us with the exciting possibility of directly correlating NPC structure with function.
ACKNOWLEDGMENTS The authors are indebted to C. Henn for designing and preparing Figures 1,4, 8, and 15. We thank Dr. R. Milligan (Scripps Research Institute, La Jolla) who provided the data of the 3-D reconstruction of negatively stained detergent-released NPCs that enabled us to produce Figures 4 , 8 , and 15. We are grateful to Dr. B. Burke (Harvard Medical School, Boston) and Dr. L. Gerace (Scripps Research Institute, La Jolla) for providing us with several anti-NPC antibodies. We thank Ms. U. Sauder for her help with embedding and thin sectioning; R. Wyss for help with Figures 9,10,14, and 16; and K. N. Goldie for providing the micrographs for Figure 5, c and d. We thank Ms. H. Frefel and Ms. M. Zoller for their expert photographic work. This work was supported by the M. E. Muller Foundation of Switzerland, by grants from the Swiss National Science Foundation and the Human Frontier Science Program (HFSP).
REFERENCES Adam, J. H. & Adam, S. A. (1994). Identification of cytosolic factors required for nuclear localization sequence-mediated binding to the nuclear envelope. J. Cell Biol. 125, 547-555. Adam, S. A. & Gerace, L. (1991). Cytosolic proteins that specifically bind nuclear localization signals are receptors for nuclear import. Cell 66, 837-847. Adam, S. A., Lobl, T. J., Mitchell, M. A., & Gerace, L. (1989). Identification of specifically binding proteins for a nuclear location sequence. Nature (London) 337, 276—279. Adam, S. A., Steme-Marr, R., & Gerace, L. (1990). Nuclear protein import in permeabilized mammalian cells requires soluble cytoplasmic factors. J. Cell Biol. 111, 807-816. Aebi, U., Cohn, J., Buhle, L., & Gerace, L. (1986). The nuclear lamina is a meshwork of intermediatetype filaments. Nature (London) 323, 560-564. Akey, C. W. (1989). Interactions and structure of the nuclear pore complex revealed by cryo-electron microscopy. J. Cell Biol. 109, 955-970. Akey, C. W. (1990). Visualization of transport-related configurations of the nuclear pore transporter. Biophys. J. 58, 341-355.
Nuclear Pore Complex
43
Akey, C. W. & Goldfarb, D. S. (1989). Protein import through the nuclear pore complex is a multistep process. J. Cell Biol. 109, 971-982. Akey, C. W. & Radermacher, M. (1993). Architecture of the Xenopus nuclear pore complex revealed by three-dimensional cryo-electron microscopy. J. Cell Biol. 122, 1—19. Bataille, N., Helser, T., & Fried, H. M. (1990). Cytoplasmic transport of ribosomal subunits microinjected into the Xenopus laevis oocyte nucleus: A generalized, facilitated process. J. Cell Biol. Ill, 1571-1582. Belanger, K. D., Kenna, M. A., Wei, S., & Davis, L. (1994). Genetic and physical interactions between Srplp and nuclear pore complex proteins NUPlp and NUP2p. J. Cell Biol. 126, 619-630. Buss, F., Kent, H., Stewart, M., Bailer, S. M., & Hanover, J. A. (1994). Role of different domains in the self-association of rat nucleoporin p62. J. Cell Sci. 107, 631-638. Byrd, D., Sweet, D. J., Pante, N., Konstantinov, K. N., Guan, T., Saphire, A. C. S., Mitchell, R J., Cooper, C. S., Aebi, U., & Gerace, L. (1994). Tpr, a large coiled coil protein whose amino terminus is involved in activation of oncogenic kineses, is localized to the cytoplasmic surface of the nuclear pore complex. J. Cell Biol. 127, 1515-1526. Carmo-Fonseca, M., Kern, H., & Hurt, E. C. (1991). Human nucleoporin p62 and the essential yeast nuclear pore protein NSPl show sequence homology and a similar domain organization. Eur. J. Cell Biol. 55, 17-30. Chelsky, D., Ralph, R., & Jonak, G. (1989). Sequence requirements for synthetic peptide-mediated translocation to the nucleus. Mol. Cell Biol. 9, 2487-2492. Chugani, D. C, Rome, L. H., & Kedersha, N. L. (1993). Localization of vault particles to the nuclear pore complex. J. Cell Sci. 106, 23-29. Coleman, J. E. (1992). Zinc proteins: Enzymes, storage proteins, transcription factors, and replication proteins. Annul Rev. Biochem. 61, 897-946. Cordes, V., Waizenegger, I., & Krohne, G. (1991). Nuclear pore complex glycoprotein p62 oiXenopus laevis and mouse: cDNA cloning and identification of its glycosylation region. Eur. J. Cell Biol. 55,31^7. Cordes, V., Reidenbach, S., Kohler, A., Stuurman, N., van Driel, R., & Franke, W. W. (1993). Intranuclear filaments containing a nuclear pore complex protein. J. Cell Biol. 123, 1333-1344. Dabauvalle, M.-C, Benevente, R., & Chaly, N. (1988a). Monoclonal antibodies to a M^ 68,000 pore complex protein interfere with nuclear protein uptake in Xenopus oocytes. Chromosoma 97, 193-197. Dabauvalle, M.-C, Schultz, B., Scheer, U., & Peters, R. (1988b). Inhibition of nuclear accumulation of karyophilic proteins by microinjection of the lectin WGA. Exp. Cell Res. 174, 291—296. Dabauvalle, M.- C, Loos, K., & Scheer, U. (1990). Identification of a soluble precursor complex essential for nuclear pore assemble in vitro. Chromosoma 100, 56-66. Dargemont, C. & Kiihn, L. C. (1992). Export of mRNA from microinjected nuclei oi Xenopus laevis oocytes. J. Cell Biol. 118, 1-9. Davis, L. I. & Blobel, G. (1986). Identification and characterization of a nuclear pore complex protein. Cell 45, 699-709. Davis, L. I. & Blobel, G. (1987). The nuclear pore complex contains a family of glycoproteins that includes p62: Glycosylation through a previously unidentified cellular pathway. Proc. Natl. Acad. Sci. USA 84, 7552-7556. Davis, L. I. & Fink, G. R. (1990). The NUPl gene encodes an essential component of the yeast nuclear pore complex. Cell 61,965-978. Dessev, G., lovcheva-Dessev, C, Bischoff, J. R., Beach, D., & Goldman, R. (1991). A complex containing p34^ ^ and cyclin B phosphorylates the nuclear lamin and disassembles nuclei of clam oocytes in vitro. J. Cell Biol. 112, 523-533. Dingwall, C. & Laskey, R. A. (1991). Nuclear targeting sequences—^A consensus? Trends Biochem. Sci. 16,478-481. Fabre, E. & Hurt, E. C. (1994). Nuclear transport. Curr. Opin. Cell Biol. 6, 335-342.
44
NELLY PANTE and UELI AEBI
Fabre, E., Boelens, W. C, Wimmer, C, Mattaj, I W., & Hurt, E. C. (1994). Nupl45p is required for nuclear export of mRNA and binds homopolymeric RNA in vitro via a novel conserved motif. Cell 78,275-289. Featherstone, C, Darby, M. K., & Gerace, L. (1988). A monoclonal antibody against the nuclear pore complex inhibits nucleocytoplasmic transport of protein and RNA in vivo. J. Cell Biol. 107, 1289-1297. Feldherr, C. M., Kallenbach, E., & Schultz, N. (1984). Movement of a karyophilic protein through the nuclear pores of oocytes. J. Cell Biol. 99, 2216-2222. Finlay, D. R., Newmeyer, D. D., Price, T. M., & Forbes, D. J. (1987). Inhibition of in vitro nuclear transport by a lectin that binds to nuclear pores. J. Cell Biol. 104, 189-200. Finlay, D. R., Meier, E., Bradley, P., Horecka, J., & Forbes, D. J. (1991). A complex of nuclear pore proteins required for pore function. J. Cell Biol. 114, 169-183. Fischer, U. & Luhrmann, R. (1990). An essential signaling role for the m G cap in the transport of Ul snRNPs to the nucleus. Science 249, 786-790. Fischer, U., Sumpter, V., Sekine, M., Satoh, T., & Luhrmann, R. (1993). Nucleocytoplasmic transport of U snRNPs: Definition of a nuclear localization signal in the Sm core domain that binds a transport receptor independently of the m G cap. EMBO J. 12, 573-583. Forbes, D. J. (1992). Structure and function of the nuclear pore complex. Annul Rev. Cell Biol. 8, 495-527. Franke, W. W. (1974). Structure, biochemistry and functions of the nuclear envelope. Int. Rev. Cytol. Suppl. 4, 71-236. Franke, W. W. & Scheer, U. (1970a). The ultrastructure of the nuclear envelope of amphibian oocytes: A reinvestigation. I. The mature oocyte. J. Ultrastruct. Res. 30, 288-316. Franke, W. W. & Scheer, U. (1970b). The ultrastructure of the nuclear envelope of amphibian oocytes: A reinvestigation. II. The immature oocyte and dynamic aspects. J. Ultrastruct. Res. 30, 317-327. Garcia-Bustos, J., Heitman, J., & Hall, M. N. (1991). Nuclear protein localization. Biochim. Biophys. Acta 1071, 83-101. Gerace, L. (1992). Molecular trafficking across the nuclear pore complex. Curr. Opin. Cell Biol. 4, 637-645. Gerace, L. & Blobel, G. (1980). The nuclear envelope lamina is reversibly depolymerized during mitosis. Cell 19,277-287. Gerace, L. & Burke, B. (1988). Functional organization of the nuclear envelope. Annu. Rev. Cell Biol. 4, 335-374. Gerace, L., Ottaviano, Y., & Kondor-Koch, C. (1982). Identification of a major polypeptide of the nuclear pore complex. J. Cell Biol. 95, 826-837. Goldberg, M. W. & Allen, T. D. (1992). High resolution scanning electron microscopy of the nuclear envelope: Demonstration of a new, regular, fibrous lattice attached to the baskets of the nucleoplasmic face of the nuclear pores. J. Cell Biol. 119, 1429-1440. Goldberg, M. W. & Allen, T. D. (1993). The nuclear pore complex: Three-dimensional surface structure revealed by field emission, in-lens scanning electron microscopy, with underlaying structure uncovered by proteolysis. J. Cell Sci. 106, 261-274. Goldie, K. N., Pante, N., Engel, A., & Aebi, U. (1994). Exploring native nuclear pore complex structure and conformation by scanning force microscopy in physiological buffers. J. Vac. Sci. Technol. B 12, 1482-1485. Grandi, P., Doye, V., & Hurt, E. C. (1993). Purification of NSPl reveals complex formation with "GLFG" nucleoporins and a novel nuclear pore protein NIC96. EMBO J. 12, 3061-3071. Greber, U. F. & Gerace, L. (1992). Nuclear protein import is inhibited by an antibody to a lumenal epitope of a nuclear pore complex glicoprotein. J. Cell Biol. 116, 15—30. Greber, U. F., Senior, A., & Gerace, L. (1990). A major glycoprotein of the nuclear pore complex is a membrane-spanning polypeptide with a large lumenal domain and a small cytoplasmic tail. EMBO J. 9, 1495-1502.
Nuclear Pore Complex
45
Guan, T., Muller, S., Klier, G., Pante, N., Blevitt, J. M., Haner, M., Paschal, B., Aebi, U., & Gerace, L. (1995). Structural analysis of the p62 complex, an assembly of 0-linked glycoproteins that localizes near the central gated channel of the nuclear pore complex. Mol. Biol. Cell 6,1591-1603. Hallberg, E., Wozniak, R. W., & Blobel, G. (1993). An integral membrane protein of the pore membrane domain of the nuclear envelope contains a nucleoporin-like region. J. Cell Biol. 122, 513-521. Hamm, J. & Mattaj, I. W. (1990). Monomethylated cap strucmres facilitate RNA export from the nucleus. Cell 63, 109-118. Hamm, J., Darzynkiewicz, E., Tahara, S. M., & Mattaj, I. W. (1990). The trimethyguanosine cap structure of Ul snRNA is a component of a bipartite nuclear targeting signal. Cell 62, 569-577. Hanover, J. A., Cohen, C. K., Willingham, M. C, & Park, M. K. (1987). O-linked N-acetylglucosamine is attached to proteins of the nuclear pore. J. Biol. Chem. 262, 9887-9894. Hinshaw, J. E., Carragher, B. O., & Milligan, R. A. (1992). Architecture and design of the nuclear pore complex. Cell 69, 1133-1141. Holt, G. D., Snow, C. M., Senior, A., Haltiwanger, R. S., Gerace, L., & Hart, G. W. (1987). Nuclear pore complex glycoproteins contain cytoplasmically disposed O-linked N-acetylglucosamine. J. Cell Biol. 104, 1157-1164. Hurt, E. C. (1988). A novel nucleoskeletal-like protein located at the nuclear periphery is required for the life cycle oiSaccharomyces cerevisiae. EMBO J. 7, 4323—4334. Hurt, E. C. (1990). Targeting of a cytosolic protein to the nuclear periphery. J. Cell Biol. Ill, 2829-2837. Imamoto, N., Matsuoka, Y., Kurihara, T, Kohno, K., Miyagi, M., Sakiyama, F., Okada, Y., Tsunasawa, S., & Yoneda, Y. (1992). Antibodies against 70-kD heat shock cognate protein inhibit mediated nuclear import of karyophilic proteins. J. Cell Biol. 119, 1047—1061. Izaurralde, E. & Mattaj, I. W. (1992). Transport of RNA between nucleus and cytoplasm. Semin. Cell Biol. 3, 279-288. Izaurralde, E., Stepinski, J., Darzynkiewicz, E., & Mattaj, I. W. (1992). A cap binding protein that may mediate nuclear export of RNA polymerase Il-transcribed RNAs. J. Cell Biol. 118, 1287—1295. Jarmolowski, A., Boelens, W. C, Izaurralde, E., & Mattaj, I. W. (1994). Nuclear export of different classes of RNA is mediated by specific factors. J. Cell. Bio. 124, 627-635. Jamik, M. & Aebi, U. (1991). Toward a more complete 3-D structure of the nuclear pore complex. J. Struct. Biol. 107,291-308. Kedersha, N. L., Heuser, J. E., Chugani, D. C, & Rome, L. H. (1991). Vaults. III. Vault Ribonucleoprotein particles open into flower-like structures with octagonal symmetry. J. Cell Biol. 112,225-235. Kita, K., Omata, S., & Horigome, T. (1993). Purification and characterization of a nuclear pore glycoprotein complex containing p62. J. Biochem. (Tokyo) 113, 377-382. Kraemer, D., Wozniak, R. W., Blobel, G., & Radu, A. (1994). The human CAN protein, a putative oncogene product associated with myeloid leukemogenesis, is a nuclear pore complex protein that faces the cytoplasm. Proc. Natl. Acad. Sci. USA 91, 1519-1523. Loeb, J. D. J., Davis, L., & Fink, G. F. (1993). NUP2, a novel yeast nucleoporin, has functional overlap with others proteins of the nuclear pore complex. Mol. Biol. Cell. 4, 209-222. Marshallsay, C. & Liihrmann, R. (1994). In vitro nuclear import of snRNPs: Cytosolic factors mediate m^G-cap dependence of Ul and U2 snRNP transport. EMBO J. 13, 222-231. Mattaj, I. W., Boelens, W., Izaurralde, E., Jarmolowski, A., & Kambach, C. (1993). Nucleocytoplasmic transport and snRNP assembly. Mol. Biol. Rep. 18, 79-83. Maul, G. G. (1977). The nuclear and cytoplasmic pore complex. Structure, dynamics, distribution and evolution. Int. Rev. Cytol. Suppl. 6, 75-186. McMorrow, I. M., Bastos, R., Horton, R., & Burke, B. (1994). Sequence analysis of a cDNA encoding a human nuclear pore complex protein, hnupl 53. Biochim. Biophys. Acta 1217, 219-223. Mehlin, H., Daneholt, B., & Skoglund, U. (1992). Translocation of a specific premessenger ribonucleoprotein particle through the nuclear pore studied with electron microscope tomography. Cell 69, 605-613.
46
NELLY PANTE and UELI AEBI
Melchior, F., Paschal, B., Evans, J., & Gerace, L. (1993). Inhibition of nuclear protein import by nonhydrolyzable analogues of GTP and identification of the small GTPase Ran/TC4 as an essential transport factor. J. Cell Biol. 123, 1649-1659. Michaud, N. & Goldfard, D. (1991). Multiple pathways in nuclear transport: the import of U2 snRNP occur by a novel kinetic pathway. J. Cell Biol. 112, 215-223. Michaud, N. & Goldfard, D. (1992). Microinjected U snRNAs are imported to oocyte nuclei via the nuclear pore complex by three distinguishable targeting pathways. J. Cell Biol. 116, 851-861. Mitchell, P. J. & Cooper, C. S. (1992). Nucleotide sequence analysis of human tpr cDNA clones. Oncogene 7, 383-388. Moore, M. S. & Blobel, G. (1992). The two steps of nuclear import, targeting to the nuclear envelope and translocation through the nuclear pore, require different cytosolic factors. Cell 68, 939-950. Moore, M. S. & Blobel, G. (1993). The GTP-binding protein Ran/TC4 is required for protein import into the nucleus. Nature (London) 365, 661-663. Newmeyer, D. D. (1993). The nuclear pore complex and nucleocytoplasmic transport. Curr. Opin. Cell Biol. 5, 395-407. Newmeyer, D. D. & Forbes, D. J. (1988). Nuclear import can be separated into distinct steps in vitro: Nuclear pore binding and translocation. Cell 52, 641-653. Newmeyer, D. D. & Forbes, D. J. (1990). An N-ethylmaleimide-sensitive cytosolic factor necessary for nuclear protein import: Requirement in signal-mediated binding to the nuclear pore. J. Cell Biol. 110,547-557. Newmeyer, D. D., Finlay, D. R., & Forbes, D. J. (1986). In vitro transport of a fluorescent nuclear protein and exclusion of non-nuclear proteins. J. Cell Biol. 103, 2091—2102. Ottaviano, Y. & Gerace, L. (1985). Phosphorylation of the nuclear lamins during interphase and mitosis. J. Biol. Chem. 260, 624^32. Paine, P. L., Moore, L. C, & Horowitz, S. B. (1975). Nuclear envelope permeability. Nature (London) 254 10^114. Pante, N. & Aebi, U. (1993). The nuclear pore complex. J. Cell Biol. 122, 977-984. Pante, N. & Aebi, U. (1994). Towards understanding the 3-D structure of the nuclear pore complex at the molecular level. Curr. Opin. Struct. Biol. 4, 187-196. Pante, N., Bastos, R., McMorrow, I., Burke, B., & Aebi, U. (1994). Interactions and three-dimensional localization of a group of nuclear complex proteins. J. Cell Biol. 126, 603-617. Peter, M., Nakagawa, J., Doree, M., Labbe, J. C, & Nigg, E. A. (1990). In vitro disassembly of the nuclear lamina and M phase-specific phosphorylation of lamins by cdc2 kinase. Cell 61,591-602. Peter, M., Heitliger, E., Haner, M., Aebi, U., & Nigg, E. A. (1991). Disassembly of in vitro formed lamin head-to-tail polymers by CDC2 kinase. EMBO J. 10, 1535-1544. Radu, A., Blobel, G., & Wozniak, R. W. (1993). Nupl55 is a novel nuclear pore complex protein that contains neither repetitive sequence motifs nor reacts with WGA. J. Cell Biol. 121, 1—9. Radu, A., Blobel, G., & Wozniak, R. W. (1994). Nupl07 is a novel nuclear pore complex protein that contains a leucine zipper. J. Biol. Chem. 269, 17600-17605. Reichelt, R., Holzenburg, A., Buhle, E. L., Jamik, M., Engel, A., & Aebi, U. (1990). Correlation between structure and mass distribution of the nuclear pore complex, and of distinct pore complex components. J. Cell Biol. 110, 883-894. Richardson, W. D., Mills, A. D., Dilworth, S. M., Laskey, R. A., & Dingwall, C. (1988). Nuclear protein migration involves two steps: Rapid binding at the nuclear envelope followed by slower translocation through the nuclear pores. Cell 52, 655-664. Ris, H. (1991). The 3-D structure of the nuclear pore complex as seen by high voltage electron microscopy and high resolution low voltage scanning electron microscopy. EMSA Bull. 21,54-56. Robbins, J., Dilworth, S. M., Laskey, R. A., & Dingwall, C. (1991). Two interdependent basic domains in nucleoplasmin nuclear targeting sequence: Identification of a class of bipartite nuclear targeting sequence. Cell 64, 615-623.
Nuclear Pore Complex
47
Roos, U.-P. (1973). Light and electron microscopy of rat kangaroo cells in mitosis 1: Formation and breakdown of the mitotic apparatus. Chromosoma 40, 43-82. Rout, M. R & Blobel, G. (1993). Isolation of the yeast nuclear pore complex. J. Cell Biol. 123, 771-783. Shi, Y. & Thomas, J. O. (1992). The transport of proteins into the nucleus requires the 70-kilodalton head shock protein or its cytoplasmic cognate. Mol. Cell Biol. 12, 2186-2192. Snow, C. M., Senior, A., & Gerace, L. (1987). Monoclonal antibodies identify a group of nuclear pore complex glycoproteins. J. Cell Biol. 104, 1143-1156. Soullan, B. & Worman, H. J. (1993). The amino-terminal domain of the lamin B receptor is a nuclear envelope targeting signal. J. Cell Biol. 120, 1093-1100. Starr, C. M., D'Onofrio, M., Park, M. K., & Hanover, J. A. (1990). Primary sequence and heterologous expression of nuclear pore glycoprotein p62. J. Cell Biol. 110, 1861—1871. Steme-Marr, R., Blevitt, J. M., & Gerace, L. (1992). 0-linked glycoproteins of the nuclear pore complex interact with a cytosolic factor required for nuclear protein import. J. Cell Biol. 116, 271—280. Stevens, B. J. & Swift, H. (1966). RNA transport from nucleus to cytoplasm in Chironomus salivary glands. J. Cell Biol. 31, 55-77. Stewart, M. & Whytock, S. (1988). The structure and interactions of components of nuclear envelopes from Xenopus oocyte germinal vesicles observed by heavy metal shadowing. J. Cell Sci. 90,409-423. Sukegawa, J. & Blobel, G. (1993). A nuclear pore complex protein that contains zinc finger motifs, binds DNA, and faces the nucleoplasm. Cell 72, 29-38. Unwin, P. N. T & Milligan, R. A. (1982). A large particle associated with the perimeter of the nuclear pore complex. J. Cell Biol. 93, 63—75. Von Lindem, M., Fomerod, M., van Baal, S., Jaegle, M., de Wit, T, Buijs, A., & Groveld, G. (1992). The translocation (6;9), associated with a specific subtype of acute myeloid leukemia, results in the fusion of two genes, dek and can, and the expression of a chimeric, leukemia-specific dek-can mRNA. Mol. Cell Biol. 12, 1687-1697. Ward, G. E. & Kirschner, M. W (1990). Identification of cell cycle-regulated phosphorylation sites on nuclear lamin C. Cell 61, 561-577. Wente, S. R. & Blobel, G. (1993). A temperature-sensitive NUP116 null mutant forms a nuclear envelope seal over the yeast nuclear pore complex thereby blocking nucleocytoplasmic traffic. J. Cell Biol. 123,275-284. Wente, S. R. & Blobel, G. (1994). NUP 145 encodes a novel yeast glycine-leucine-phenylalanine-glycine (GLFG) nucleoporin required for nuclear envelope structure. J. Biol. 125, 955—969. Wente, S. R., Rout, M. P., & Blobel, G. (1992). A new family of yeast nuclear pore complex proteins. J. Cell Biol. 119,705-723. Wiese, C. & Wilson, K. L. (1993). Nuclear membrane dynamics. Curr. Opin. Cell Biol. 5, 387-394. Wilken, N., Kossner, U., Senecal, J.-L., Scheer, U., & Dabauvalle, M.-C. (1993). Nupl80, a novel nuclear pore complex protein localizing to the cytoplasmic ring and associatedfibrils.J. Cell Biol. 123,1345-1354. Wimmer, C , Doye, V., Grandi, P., Nehrbass, U., & Hurt, E. C. (1992). A new subclass of nucleoporins that functionally interact with nuclear pore protein NSPl. EMBO J. 11, 5051—5061. Wozniak, R. W & Blobel, G. (1992). The single transmembrane segment of gp210 is sufficient for sorting to the pore membrane domain of the nuclear envelope. J. Cell Biol. 119, 1441—1449. Wozniak, R. W., Bartnik, E., & Blobel, G. (1989). Primary structure analysis of an integral membrane glycoprotein of the nuclear pore. J. Cell Biol. 108, 2083-2092. Wozniak, R. W, Blobel, G., & Rout, M. R (1994). POM 152 is an integral protein of the pore membrane domain of the yeast nuclear envelope. J. Cell Biol. 125, 31—42. Yano, R., Oakes, M., Yamaghishi, M., Dodd, J. A., & Nomura, M. (1992). Cloning and characterization of SRPl, a suppressor of temperature-sensitive RNA polymerase I mutations, in Saccharomyces cerevisiae. Mol. Cell. Biol. 12, 5640-5641. Zasloff, M. (1983). tRNA transport form the nucleus in a eukaryotic cell: Carrier-mediated translocation process. Proc. Natl. Acad. Sci. USA 80, 6436-6440.
This Page Intentionally Left Blank
STRUCTURE AND FUNCTION OF MITOCHONDRIAL PRESEQUENCES
Merritt Maduke and David Roise
Abstract I. Introduction A. Protein Import into the Mitochondrial Matrix B. Mitochondrial Presequences II. Structure of Presequences III. Surface Activity A. Insertionof Presequences into Monolayers B. Disruptionof Phospholipid Vesicles C. Effectof Length on the Properties of Presequences IV. Binding to Membranes A. Binding to Phospholipid Vesicles B. Binding to Mitochondrial Membranes C. Effects of Surface Potential on Binding to Membranes V. Import into Mitochondria VI. Import Into Phospholipid Vesicles VII. Model for Recognition of Mitochondrial Presequences A. Bindingof Precursors to Membranes/« Vivo B. TheMembranePotentialMay Determine Specificity of Import
Membrane Protein TVansport Volume 3, pages 49-79. Copyright © 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 1-55938-989-3 49
50 50 50 52 55 56 56 56 57 58 58 59 62 65 68 70 70 71
50
MERRITT MADUKE and DAVID ROISE
C. Role ofMitochondrial Membrane Proteins VIII. Summary Acknowledgments References
72 74 74 75
ABSTRACT Mitochondrial presequences are responsible for the specific targeting of proteins to the matrix of the mitochondria. The presequences lack sequence homology but share common physical characteristics—positive charge and amphiphilicity—which make them ideally suited for interacting with the mitochondrial outer membranes. Structural studies have demonstrated that presequences adopt a-helical structure in the presence of negatively charged surfaces and that the a-helices are amphiphilic. Quantitative studies using synthetic presequences have shown that the presequences bind to lipid vesicles and to the lipids of mitochondrial membranes, and that the binding to these membranes is similar. Other studies have shown that the import of mitochondrial presequences into isolated yeast mitochondria and their import into protein-free phospholipid vesicles are also similar. These results are discussed, and a model for the recognition of mitochondrial presequences within cells is presented.
1. INTRODUCTION The goal of this chapter is to summarize the results of research performed in our laboratory on mitochondrial presequences and to describe how these results are relevant to the mechanism of transport of proteins into mitochondria. Our work has been directed at elucidating the mechanism of import of presequences into mitochondria. Understanding, in detail, how presequences interact with and are imported into mitochondria should serve as a first step toward understanding the mechanism of import of full-length precursor proteins into the organelle. Our results demonstrate that mitochondrial presequences, because of their physical characteristics, are ideally suited for their role as targeting signals for proteins destined for the mitochondria. A. Protein Import into the Mitochondrial Matrix
Most mitochondrial proteins are synthesized in the cytoplasm and must be imported into the organelle. Proteins that are destined for the interior of the mitochondria are usually synthesized as longer precursor forms with amino-terminal extensions. These extensions, called presequences, are responsible for the specific targeting of the protein to the mitochondria and are proteolysed once inside the matrix. When expressed on hybrid proteins through the use of gene fusions, mitochondrial presequences are capable of directing the import of even nonmito-
Mitochondrial Presequences
51
chondrial proteins into the mitochondria (Hurt et al., 1984; Horwich et al., 1985; Hurt et al., 1985b; Hurt and van Loon, 1986). A diagram representing the import of a precursor protein into the mitochondrial matrix space is shown in Figure 1. Listed in the figure are components that are involved in the protein import process. Cytosolic ATP and heat shock proteins are required for the import of many proteins (Deshaies et al., 1988; Murakami et al.. 1988; Gething and Sambrook, 1992; Stuart et al., 1994), presumably to maintain the precursors in a conformation that is competent for being imported (Eilers and Schatz, 1988). Other cytosolic factors may be involved in protein import (Firgaira et al., 1984; Ohta and Schatz, 1984; Ono and Tuboi, 1988, 1990a), although it has been observed that chemically pure precursors can be efficiently imported into purified mitochondria in the absence of added factors (Eilers and Schatz, 1986; Becker et al., 1992). The cytosolic factors may act to stabilize the precursor and prevent aggregation, and they thus may not be necessary for protein translocation under all conditions. Several outer and inner mitochondrial membrane proteins have been implicated in the import process and are listed in the figure. The proteins identified in yeast have been named Mas (for mitochondrial assembly) and ISP (for import site protein) (Schatz, 1993); the proteins from Neurospora crassa have been named either MOM (for mitochondrial outer membrane) or MIM (for mitochondrial inner
BINDING
\
CYTOSOL ATP Hsp70 other soluble factors?
OUTER MEMBRANE Mas20p/MOM19 Mas70p/MOM72 ISP42/MOM38 MOM30, MOMS. MOM7. MOM22 INNER MEMBRANE ISP45/MIM44 Mas6p/MIM23
MATRIX Hsp60 mHsp70 MPPa.p ATP
Figure 1. Protein import into the mitochondrial matrix. See text for explanation of steps and components.
52
MERRITT MADUKE and DAVID ROISE
membrane) (Kiebler et al., 1993); the numbers in each case refer to the apparent molecular mass (kDa) of each protein. These proteins have been identified by biochemical and/or genetic approaches. In contrast to the protein translocation process in the endoplasmic reticulum, which has recently been reconstituted from a limited number of purified components (Gorlich and Rapoport, 1993), none of the mitochondrial membrane proteins has yet been studied in a purified and reconstituted system. The exact functions of the components are, therefore, still debatable. The mitochondrial membrane potential (A^) also plays an important role in the translocation of precursors across the inner membrane. It is required for the initiation of transport, presumably because it induces electrophoresis of the positively charged presequences across the inner membrane (Pfanner and Neupert, 1985; Martin et al., 1991). Inside the matrix, ATP and heat shock proteins are required to facilitate protein folding and to limit aggregation (Stuart et al., 1994). The subunits of the mitochondrial processing peptidase (MPP-a, MPP-P) (Kalousek et al., 1993) are required for cleavage of the presequence. Cleavage of the presequence is not necessary for some proteins and in some cases may occur prior to the complete translocation of the precursor into the matrix space (Schleyer and Neupert, 1985). As shown in the model in Figure 1, an early step in the import of a precursor is likely to be the insertion of the presequence, as an a-helix, into the lipids of the mitochondrial outer membrane. This suggestion is consistent with many studies on mitochondrial presequences, which will be discussed. A precursor protein, bound to the membrane as shown, would diffuse laterally in the outer membrane until it reaches a site of translocation. For proteins destined for the matrix space, translocation occurs at contact sites, where the outer and inner mitochondrial membranes are closely associated (Schleyer and Neupert, 1985). It should be noted that the detailed structure of the membrane at the contact sites is unknown, and it is not clear whether these contact sites represent stable structures, or whether they form transiently as precursors are in the process of being transported (Schwaiger et al., 1987; Pon et al., 1989; Pfanner et al., 1990). Indeed, under some conditions the two membranes can become uncoupled as a precursor is being imported, and the protein may only reach the intermembrane space (Hwang et al., 1991; Wachter et al., 1994). It is not yet known whether transport across the two membranes at the contact site occurs through a translocation pore, directly through the lipid bilayer, or through some combination of protein and lipid; hence, a question mark is shown in the model. B. Mitochondrial Presequences Mitochondrial presequences are rich in positively charged residues and hydroxylated residues, and they generally lack negatively charged residues. They do not, however, contain any obvious consensus sequence. The lack of a consensus
Mitochondrial Presequences
53
sequence makes it puzzling how mitochondria specifically recognize the targeting information contained within the hundreds of matrix-targeted proteins. Some clues to this enigma come from studies on the physical properties of presequences. Roughly 10 years ago, it was noted that the presequence of yeast cytochrome oxidase subunit IV (CoxIV) could form an amphiphilic a-helix (Roise et al., 1986). This type of structure is common among polypeptides that interact with surfaces (Kaiser and Kezdy, 1983; Roise, 1993). Although the primary sequence of the CoxIV presequence lacks any clustering of hydrophobic or basic residues, a plot of the sequence on a helical wheel diagram shows that clustering of hydrophobic and basic residues occurs when the presequence forms an a-helix (Fig. 2). Experiments with synthetic peptides corresponding to segments of the CoxIV presequence confirmed that this sequence forms an amphiphilic a-helix (Roise et al., 1986). A peptide corresponding to the complete, 25-residue presequence was unstructured in aqueous solution but formed an a-helical structure in the presence of negatively charged detergent micelles. The peptide was able to insert into phospholipid monolayers and to disrupt phospholipid bilayers. These results verified the prediction that the peptide would form an amphiphilic a-helix. To establish that amphiphilicity is a general property of mitochondrial targeting sequences, 23 presequences were compared, and the hydrophobic moment of each was predicted (von Heijne, 1986). The hydrophobic moment is a numerical evaluation of the amphiphilicity of an a-helix (Eisenberg, 1984). It is calculated as a
HaN-MetLeuSerLeuAi^GlnSerMArgPhePheLysProAlaThrArgT^
ile "let [ys-*"8 _ 1 _ is .
Figure 2. Presequence of the yeast cytochrome oxidase subunit IV (CoxIV). The primary structure is shown with charges indicated and hydrophobic residues underlined. In an a-helical conformation, the side chains of the presequence of CoxIV are arranged so that the positively charged residues lie on one face of the helix and the hydrophobic residues lie on the opposite face.
54
MERRITT MADUKE and DAVID ROISE
vector sum of the hydrophobicities of the amino acids in a sequence. The hydrophobic moment is used to provide a more quantitative prediction of amphilicity than does simple inspection of the sequence on a helical wheel diagram. The analysis by von Heijne revealed that all the mitochondrial presequences examined have the potential to form amphiphilic a-helices (helices of high hydrophobic moment). This observation suggested that formation of a surface-active, positively charged amphiphilic helix may be a universal property of mitochondrial presequences. Experiments with precursor proteins have confirmed that amphiphilicity is a general feature of functional matrix-targeting presequences. With genetic techniques the presequence of any precursor protein can be manipulated as desired. The effect of changes in the presequence on the ability of a precursor to be imported into mitochondria either in vivo or in vitro can then be determined. Numerous site-directed and random mutagenesis studies have been performed on various precursor proteins, and the results generally indicate that positive charge and amphiphilicity are important for presequence function (reviewed in Roise, 1993). Mutations that decrease positive charge or hydrophobicity, or that introduce residues that disrupt the helix (glycine or proline), decrease import efficiency. In an extremely thorough study, a strong correlation between the hydrophobic moment of a predicted a-helix and the efficiency of import of the attached protein was observed (Bedwell et al., 1989). To test the hypothesis that mitochondrial presequences are recognized because of their general physical properties and not because of a particular amino acid sequence, artificial presequences were fused to a CoxIV protein lacking its own presequence (Allison and Schatz, 1986). These presequences were designed to avoid resemblance to any particular mitochondrial presequence and were composed solely of arginine, leucine, and serine, residues that occur frequently in mitochonddrial presequences. Several of the artificial presequences were capable of directing attached proteins to the mitochondrial matrix. The functional presequences were later found to be amphiphilic, whereas the artificial sequences that lacked hydrophobic residues and were nonfunctional were not amphiphilic (Roise et al., 1988). In a complementary approach, random DNA fragments were fused to a truncated CoxIV gene (Lemire et al., 1989). The precursor proteins generated by these gene fusions were analyzed for their ability to be imported into mitochondria in yeast cells. Approximately 25% of the random presequences were found to be capable of directing the protein into the mitochondrial matrix. Examination of the primary sequences revealed that the functional presequences were all positively charged and had the potential to form amphiphilic a-helices. The predicted hydrophobic moment of a given sequence was strongly correlated with the efficiency of the sequence as a targeting signal. Taken together, the results demonstrate that mitochondria recognize positively charged, amphiphilic a-helices at the amino-termini of precursor proteins.
Mitochondrial Presequences
55
The rest of the chapter will summarize recent experiments that confirm the role of mitochondrial presequences as amphiphilic a-helices and that suggest that direct interactions between presequences and the lipid bilayer are involved in the import of precursors into mitochondria. A model will be presented that describes how the physical properties of presequences are important for the association of precursors with the mitochondrial surface and how the transmembrane potential across the mitochondrial inner membrane may be responsible for the specific uptake of precursor proteins by mitochondria in vivo.
II. STRUCTURE OF PRESEQUENCES To determine how presequences function, it is important to understand the structural and physical properties of the presequences. By using synthetic peptides corresponding to mitochondrial presequences, it has been possible to measure these properties directly. Circular dichroism (CD) measurements have been used to determine the a-helical content of several synthetic presequences. All of the presequences are random coils in solution but adopt varying degrees of a-helical structure in the presence of detergent micelles, small vesicles, or trifluoroethanol (TFE, a helix-stabilizing solvent) (Epand et al., 1986; Roise et al., 1986; Aoyagi et al., 1987; Tamm and Bartoldus, 1990; Hoyt et al., 1991). The surfaces (detergent or lipid) generally must be negatively charged to induce the formation of helical structure. Nuclear magnetic resonance (NMR) studies have supplemented the CD studies by identifying the specific residues that compose the helices of several presequences. The CoxIV presequence was shown by CD to be 40-50% a-helical in the presence of sodium dodecylsulfate micelles (Roise et al., 1986). Nuclear Overhauser effects (NOEs) indicated that residues 3—11 of the CoxIV presequence form interresidue a-helical contacts in the presence of dodecylphosphocholine micelles (Endo et al., 1989). In a similar environment, the presequence of rat liver aldehyde dehydrogenase displayed two helical regions (Karslake et al., 1990). The NOEs, together with the rates of amide proton exchange, indicated that the COOH-terminal helix is more stable than the NH2-terminal helix. In TFE, the presequence of the P-subunit of yeast Fj-ATPase also adopted a-helical structure, between residues 4 and 10 and between residues 14 and 19 (Bruch and Hoyt, 1992). With this presequence, the NH2-terminal helix is the more stable of the two. Finally, two synthetic peptides corresponding to the targeting segments of two precursors that are not proteolytically processed in the mitochondrial matrix have been studied using NMR (Hammen et al., 1994). Residues 4-21 from the targeting sequence of rhodanese and residues 4—14 from thiolase were found to be helical in the presence of TFE. The results for the thiolase presequence in TFE and in micelles were similar. The general conclusion from these studies is that mitochondrial presequences tend to form a-helices in the presence of an appropriate surface. This conclusion is
56
MERRITT MADUKE and DAVID ROISE
consistent with the predictions from helical wheel projections of functional presequences.
III. SURFACE ACTIVITY A. Insertion of Presequences into Monolayers
The ability to insert into membranes is thought to be an important feature of mitochondrial targeting signals. This behavior has been confirmed by studies measuring the surface activity of synthetic presequence peptides. One measure of surface activity is the insertion of synthetic presequences into phospholipid monolayers. Monolayers are formed at the air—water interface by spreading phospholipids on the surface of an aqueous buffered solution. When a presequence is injected into the solution below the monolayer, an increase in surface pressure is observed if the presequence inserts into the monolayer. The initial surface pressure of the monolayer can be controlled by adjusting the surface area. If the increase in surface pressure upon addition of a presequence is measured at various initial surface pressures, the data can be linearly extrapolated to determine a limiting pressure at which the presequence will no longer be able to insert into the monolayer. For CoxIV presequences of 25 or 33 residues, the limiting pressure was 40-50 mN m"^ (Roise et al., 1986; Tamm, 1986). Although a monolayer is not a perfect model for a bilayer, some studies have suggested that phospholipid packing in a monolayer is most similar to that in a bilayer at a surface pressure of roughly 33 mN m~^ (Demel et al., 1975; Seelig, 1987). The ability of the wild-type CoxIV presequence to insert into monolayers at pressures higher than 40 mN m~' is consistent with the ability of this presequence to insert into membrane bilayers. In contrast, a mutant form of the CoxIV presequence that contained a two-residue deletion designed to disrupt the amphiphilic helix (Al 1,12-CoxIV) inserted into monolayers only up to 30 mN m~^ (Roise et al., 1988). This presequence also had less detergent-induced a-helical content than the wild-type presequence, and it directed the import of attached proteins less efficiently. The relatively low limiting pressure for insertion of the mutant presequence into monolayers suggests that the decreased function of the mutant presequence may be due to a decreased ability to insert into membranes. B. Disruption of Phospholipid Vesicles
Another assay used to measure surface activity is leakage of the dye 6-carboxyfluorescein from phospholipid vesicles. Carboxyfluorescein trapped within vesicles at a high concentration has a quenched fluorescence. Disruption of the vesicles by surface-active agents induces a leakage of the dye that can be measured as an increase in fluorescence as the dye dilutes into the solution. Peptides that correspond to functional presequences can disrupt vesicles in these assays (Roise et al.,
M itochondria I Presequences
57
1986, 1988; Skerjanc et al., 1987; Zardeneta and Horowitz, 1992), whereas the mutant version of the CoxIV presequence, Al 1,12-CoxIV (Section III A), is greatly decreased in its ability to disrupt fluorescein-containing vesicles (Roise et al., 1988). This result is consistent with the low limiting pressure that was observed for insertion of this presequence into monolayers. The disruptive effects of presequences on vesicles also appear to depend on the length of the synthetic sequence; short versions of functional presequences do not disrupt vesicles, whereas all synthetic presequences consisting of 22 or more residues and corresponding to functional presequences have been found to disrupt membranes (Roise et al., 1986, 1988; Hoyt et al., 1991; Zardeneta and Horowitz, 1992; Nicolay et al., 1994; see also Section IIIC). These results suggest that the ability to disrupt membranes is a common feature of functional presequences. C. Effect of Length on the Properties of Presequences
Comparison of synthetic presequences of different lengths supports the hypothesis that insertion into the membrane is required for function. The length of a synthetic presequence is an important determinant of the ability of the presequence both to insert into membranes and to inhibit protein import. It should be noted that import inhibition studies must be interpreted with caution — mitochondrial presequences are able to dissipate the membrane potential and can thus cause nonspecific inhibition of protein import. Nevertheless, protein import can be inhibited at concentrations of presequence that do not dissipate the membrane potential (Gillespie et al., 1985; Glaser and Cumsky, 1990b; Cyr and Douglas, 1991). This observation, along with the fact that presequences themselves can be efficiently imported into mitochondria (Glaser and Cumsky, 1990a; Pak and Weiner, 1990; Cyr and Douglas, 1991; Roise, 1992), suggests that the presequences are competing for a step in the normal import process and that inhibition studies can be a useful measure of presequence function. The ability of various CoxIV presequences to inhibit protein import and to disrupt vesicles is correlated with the length of the presequences. A peptide corresponding to the first 22 residues of the CoxIV presequence was found to inhibit reversibly the import of precursors into mitochondria (Glaser and Cumsky, 1990b). A shorter peptide, corresponding to the first 16 residues, was a much weaker inhibitor of import, even though it contained the 12 residues that are sufficient to import an attached protein (Hurt et al., 1985a,b). Furthermore, the disruptive effects on membranes of 25-residue and 33-residue versions of the CoxIV presequence were much greater than those of shorter 15-residue and 17-residue versions (Roise et al., 1986; Nicolay et al., 1994). Thus, even though the targeting information appears to be contained in the extreme NH2-terminal part of the presequence, the context of a longer sequence is required for efficient insertion into the membrane and for efficient inhibition of protein import.
58
MERRITT MADUKE and DAVID ROISE
Studies on the presequence of rat liver pre-omithine carbamyltransferase (pOCT) also show a correlation between the ability to form a surface-active helix and the ability to act as an efficient inhibitor of protein import. A synthetic peptide corresponding to residues 1—27 of pOCT increased in a-helical content in the presence of lipids and was able to disrupt large lipid aggregates (Epand et al., 1986). A shorter peptide (residues 16-27), which had five of the eight charges and was predicted to be amphiphilic, did not exhibit this behavior. The longer, surface-active peptide inhibited the import of precursor proteins into mitochondria; the shorter peptide did not (Gillespie et al., 1985). A peptide corresponding to 19 amino acids of the presequence of the Fj-ATPase P-subunit did not induce carboxyfluorescein leakage from vesicles, even at high ratios of peptide to lipid (Hoyt et al., 1991). Nor did this peptide efficiently compete with precursor protein for mitochondrial import. Peptides corresponding to the amino-terminal 32 or 51 residues of the precursor, however, were efficient inhibitors of protein import (Cyr and Douglas, 1991). Unfortunately, the effects of these longer peptides on lipid vesicles were not measured, but the longer peptides presumably would have induced carboxyfluorescein leakage. The 19-residue segment from the presequence of the FpATPase P-subunit was also found to insert into monolayers (Hoyt et al., 1991). A titration of the peptide indicated that it inserted efficiently at an initial surface pressure of 20 mN m~^ The ability of the peptide to insert at higher initial surface pressures was not tested, however. The relatively low maximal surface pressure increase observed suggests that the limiting pressure may be lower than 33 mN m~^ and that this peptide may not insert into bilayers. The low limiting pressure is consistent with the inability of the peptide to induce fluorescein leakage from vesicles and the inefficiency of the peptide as an inhibitor of protein import. In conclusion: several studies have shown that the length of a synthetic presequence correlates with the ability of the presequence both to insert into membranes and to inhibit mitochondrial protein import. Even though the information necessary for targeting proteins to mitochondria may be contained in a short segment, the context of a longer peptide or protein appears to be necessary for the presequence to function as an inhibitor of protein import. The correlation between the ability to inhibit protein import and the ability to insert into membranes, although not proof, is consistent with the hypothesis that presequence binding to lipids is the initial step in mitochondrial protein import.
IV. BINDING TO MEMBRANES A. Binding to Phospholipid Vesicles
A few studies have measured directly the binding of presequences to phospholipid vesicles. A fluorescently labeled CoxIV presequence was shown to bind with moderately high affinity (10^—10^ M~^) to sonicated unilamellar vesicles that
Mitochondrial Presequences
59
contained negatively charged lipids (Frey and Tamm, 1990). Measurements of the lateral diffusion of the presequence suggested that the presequence inserted either parallel to the surface of the membrane as a monomer or perpendicular to the surface as an oligomer. The former possibility is supported by measurements with phospholipid monolayers using a radioactively labeled CoxIV presequence (Tamm, 1986). An area of 560 ± 170 A^ per bound presequence was calculated by measuring changes in the surface area of the monolayer as a function of the amount of presequence bound. This value is consistent with an orientation of the a-helix parallel to the surface of the membrane. Molecular modeling suggests that, as a monomer, the CoxIV presequence would insert to a depth of 5-8 A into a bilayer (Tamm, 1991). Calorimetric techniques were used to measure the association of a synthetic human ornithine transcarbamylase (pOCT) presequence with sonicated unilamellar vesicles (Myers et al., 1987). The peptide associated strongly (--10^ M~^) with vesicles that contained negatively charged phospholipids. The emission maximum of the tryptophan fluorescence of the presequence was shifted 4 nm toward the blue upon binding of the presequence to the vesicles. This shift in the emission is an indication that the tryptophan became exposed to a more hydrophobic environment, although the shift is not as great as those observed for tryptophans contained in peptides that insert deeply into bilayers (Surewicz and Epand, 1984; McKnight et al., 1991). Since the tryptophan in the pOCT sequence is predicted to be on the hydrophilic face of the a-helix, the small blue shift in the fluorescence is consistent with insertion of the presequence parallel to the surface of the membrane, so that the hydrophilic face remains partially exposed to the solvent. A similar blue shift was observed for the rhodanese presequence, whose tryptophan is also predicted to reside on the hydrophilic face of the helix (Zardeneta and Horowitz, 1992). The binding of the presequence of rat pOCT to lipid vesicles was quantitated by using assays to measure vesicle disruption (Skerjanc et al., 1987). The affmity of this presequence for sonicated unilamellar vesicles was similar to that observed for the CoxIV presequence. Fluorescence energy transfer measurements showed that, although the presequence oligomerizes in solution, it probably associates with the lipid as a monomer. The binding of the presequence of pOCT was not dependent on the application of a transbilayer potential, with the negative charge inside. This result is in contrast to measurements of the binding of the CoxIV presequence to extruded large unilamellar vesicles, which showed that the binding was increased in the presence of a potassium difflision potential (de Kroon et al., 1991). B. Binding to Mitochondrial Membranes
The experiments with model membranes demonstrated clearly that presequences can insert directly into lipid monolayers and bilayers. The interactions between presequences and mitochondria have also been measured (Roise, 1992; Swanson and Roise, 1992). To monitor binding, the CoxIV presequence was labeled at its
60
MERRITT MADUKE and DAVID ROISE
sole cysteine with iodoacetamidofluorescein. The fluorescein label permitted detection of the presequence at nanomolar concentrations with minimal interference from mitochondrial chromophores. The fluorescent label does not appear to affect the physical properties of the peptide significantly. The ability of the fluorescent presequence to bind to isolated yeast mitochondria was confirmed by experiments in which the fluorescence was observed to cosediment with the mitochondria. In a more convenient and quantitative approach, binding of the fluorescein-labeled presequence to mitochondria was observed directly by a decrease in fluorescence that occurred upon binding. Atypical binding curve based upon this approach is shown in Figure 3A. The data fit well to a two-state model of binding in which the free presequence has a relatively high fluorescence and the bound form has a quenched fluorescence. Ofparticular interest was the observation that the binding was not saturable up to the highest concentration of presequence measured (0.1 juM); regardless of the total concentration of presequence, the same fraction of presequence was bound to a given amount of mitochondria (the same amount of quenching was observed). This result showed that the presequence was binding directly to the lipid bilayer of the mitochondrial outer membrane. If the presequence had bound to a protein on the surface of the mitochondria, the fraction of total presequence bound should have decreased at increasing concentrations of presequence as the binding protein became saturated. Since binding of the CoxIV presequence to mitochondria does not occur at discrete sites, the binding is best described as a partitioning equilibrium (Figure 3B). The partitioning is analogous to the distribution of a solute molecule in an organic extraction—for example, between water and hexane. In the case of presequence binding to mitochondria, the organic phase corresponds to the membrane surface. The partition coefficient can be obtained directly from an experimental value, A/5Q, which is the concentration of membrane required to bind half the presequence. This value can be used to calculate the amount of presequence bound to a given concentration of membrane. In the experiment shown in Figure 3 A, 50 pmol of presequence was bound to 0.024 mg of total mitochondrial protein at a concentration of mitochondria equal to the value of M5Q. The amount of even the most abundant mitochondrial outer membrane protein, porin, is only 2.9 pmol in this amount of mitochondrial protein (Freitag et al., 1982; Riezman et al., 1983); therefore, this calculation confirms that the presequence must have bound to the lipids of the outer membrane rather than to a protein. Similar observations were made for the binding of a radioactively labeled CoxIV presequence to rat liver mitochondria (Nicolay et al., 1994). The binding of the presequence to mitochondria is rapid, and it is reversible. Although the rate of binding was not measured directly, the quenching of presequence fluorescence occurs within the mixing time of an experiment (<10 seconds). Release of the bound presequence also appears to occur rapidly and completely. Addition of trypsin to a solution of membranes containing the bound presequence results in a rapid increase in the fluorescence. The digestion generates a fluorescent
Mitochondrial A
Presequences
61
0.60
Figure 3. (A) Binding of fluorescein-labeled CoxlV presequence to mitochondrial membranes. Fluorescence quenching was measured as a function of the concentration of mitochondrial membranes ([M]) with 50 n M (o) and 100 n M (•) presequence. The data were fit to the equation, Q = QmaxI^KMso + [M])"^ where Q is the relative quenching and [M] is the concentration of mitochondrial protein. The value of Q is given by {Fj - Fobs)^T^ / where Fy is the total maximal fluorescence, observed when the presequence is diluted into a solution lacking mitochondria, and F^bs is the fluorescence observed when the presequence is diluted into a solution containing mitochondria. Q^ax 'S a constant equal to the maximal quenching, (Fj - F^JFj^ and is obtained by curve fitting. F^^^^ is the fluorescence of the bound presequence. M^Q is the amount of membrane at which half-maximal quenching is observed and is obtained by curve fitting. Adapted from Roise (1992). (B) Partitioning model for presequence binding to membranes. The fluorescently labeled peptide (R = fluorescein) partitions between the aqueous phase and the membrane phase (M). The free form of the peptide (Pp) has a relatively high fluorescence; the bound form (Pg) has a quenched fluorescence. The partition coefficient (/Cp) corresponds to the ratio between the mole fraction of presequence in the membrane phase ( [ P B ] [ M ] " ^ ) and the mole fraction of presequence in the aqueous phase ([PpKHjO]"^) and is related to the experimentally determined M^Q.
peptide fragment that has the same fluorescence as the intact, unbound presequence but that does not bind appreciably to mitochondrial membranes. Since trypsin can digest only the free presequence, the rapid increase in fluorescence demonstrates that the bound presequence is in rapid equilibrium with the free presequence. That the fluorescence returns to the level of the unbound presequence indicates that the dissociation can occur completely.
MERRITT MADUKE and DAVID ROISE
62 B Binding Measurements H
Binding Equilibrium Pp + M
Kp
^^
[PB][M]-'
[H,0]
[P,]lH,Or
M^
Figure 3. Continued
C. Effects of Surface Potential on Binding to Membranes The results from the binding studies with mitochondria clearly show that the CoxIV presequence is able to bind to lipids of the mitochondrial membrane. The mechanism of binding was analyzed in further detail, using both model membranes (phospholipid vesicles) and isolated yeast mitochondria (Swanson and Roise, 1992). The binding of the fluorescein-labeled CoxIV presequence to extruded large unilamellar vesicles was similar to that observed with mitochondria; the data could be fit to a two-state partitioning equilibrium, and the binding was rapidly reversible by trypsin. The model membranes, like the mitochondrial membranes, contained negatively charged phospholipids and a net negative charge. The negative surface potential resulting from this charged surface causes an accumulation of positively charged ions in the solution immediately adjacent to the membrane (Fig. 4A). Since the presequence is positively charged, it too will accumulate in the solution near the membrane. The concentration of the charged presequence at the surface is described by a Boltzmann distribution of the form [PM] = [PF]exp(-z/4'o(/?7)-')
(1)
where [Pj^] is the concentration of presequence at the surface, [Pp] is the bulk concentration of presequence, z^ is the effective charge of the presequence, VJ/Q is the surface potential of the membrane, F is the Faraday constant, R is the gas
Mitochondrial Presequences
63
Fixe)6 Charges
1
1
© © -
®
0 © © ® ©
-
Bilayer
o
®
© ©
©
o
0
t Ions adjacent to the membrane
Ions in bulk solution
Figure 4. (A) Representation of the effect of a negative surface potential on the concentrations of ions adjacent to the surface. (B) Effects of the membrane surface potential on the binding of the fluorescein-labeled CoxlV presequence to negatively charged lipid vesicles and to mitochondria. For vesicles containing phosphatidylcholine:phosphatidylglycerol (4:1) and 7% cholesterol, the best fit of the data to the combined Boltzmann and Gouy-Chapman equations with a = -0.040 C m"^ gave Kp = 145 M"^ and Zp = +1.9. For binding to yeast mitochondria, the best fit to the data with Zp = +1.9 gave K^ = 6.3 L g"^ and a = -0.021 C m"^. Data were taken from Swanson and Roise (1992), but the parameters obtained are slightly different because of the use of a refined fitting procedure. Note that the partition coefficient for vesicles and mitochondria cannot be directly compared because of the different units of concentration. The lines shown were generated from the combined equations by using the values obtained by curve fitting.
constant, and Tis the absolute temperature. The surface potential can be calculated using the Gouy-Chapman equation, Vo = 2/?r(z/)-'sinh-'[a(8000C^8T^8o/^7)-"-']
(2)
where z^ is the charge on the salt, a is the surface charge density of the membrane, Cg is the concentration of salt, 8j is the dielectric constant of the solution, and SQ is the permittivity of vacuum. Note that the magnitude of the surface potential depends on the salt concentration. Therefore, the surface potential can be modulated experimentally by varying the salt concentration in the solution. By measuring partition coefficients at various salt concentrations, it was possible to determine the effect of the surface potential on the binding of the presequence. Since the presequence partitions into the membrane from the pool adjacent to the membrane, the partition coefficient measured at a given concentration of salt is
MERRITT MADUKE and DAVID ROISE
64
H
\-
Yeast Mitochondria 0.03
0.02
o S 0.01 0.00 [ 0.00
0.04
0.08
0.12
0.16
0.20
[KCI] (M) Figure 4. Continued
actually an apparent partition coefficient, K^ = [PB][Mr'([Pp][H20r^r^ The absolute partition coefficient, K^ = [PB][M]~^([PJ^] [H20]~^)~^ is independent of the concentration of salt. Combining these equations with Equation 1 gives K;=Kexp[-z^y^^(RTr']
(3)
From the experiments with phospholipid vesicles, the absolute partition coefficient, ATp, and the effective charge of the presequence, z , could be estimated by curve fitting Equations 2 and 3 using a charge density calculated from the composition of the vesicles and the surface area of the lipids. The effective charge of the presequence obtained from curve fitting, +1.9, is considerably lower than the fixed charges of the presequence. This is a common observation with large molecules, probably because these molecules do not behave as point charges (Camie and McLaughlin, 1983; Langner et al., 1990). Although the charge density of the mitochondrial surface was not known, it could be obtained from the experimental binding data by curve fitting, this time using the effective charge of the presequence that was obtained from the experiments with vesicles. Curves generated from the values obtained show excellent agreement with the experimental data (Fig. 4B).
Mitochondrial Presequences
65
The results from this study demonstrate the remarkable similarity between the binding of the CoxIV presequence to phospholipid vesicles and to isolated yeast mitochondria. In both cases, the binding can be described by simple physical principles; the presequence is accumulated near the membrane because of electrostatic effects of the negatively charged surface and, from this local pool, partitions directly into the lipid phase of the membrane. The partitioning presumably depends on hydrophobic interactions between the core of the membrane and the hydrophobic residues of the presequence. These results provide further evidence that mitochondrial lipids are important for the binding of mitochondrial presequences.
V. IMPORT INTO MITOCHONDRIA Although much effort has been invested in the identification of components required for mitochondrial protein import, very little is understood about the molecular mechanism of protein translocation across the membranes. Part of the difficulty lies in the nature of the assays that have been used to follow protein import. Typically, precursor proteins, which are often synthesized in crude cell lysates, are incubated with mitochondria, and the mitochondria are then reisolated and analyzed by gel electrophoresis and fluorography or immunoblotting. These steps make the assays tedious, subject to error, and inappropriate for obtaining quantitative information. Since mitochondrial presequences can be readily synthesized in quantifiable amounts and can be purified to a high degree, they are ideally suited for mechanistic studies. Determination of the mechanism of interaction of presequences with mitochondria is an important step toward the goal of understanding the mechanism of import of full-length proteins into mitochondria. Several studies have shown that synthetic presequences can be imported rapidly and efficiently into mitochondria (Glaser and Cumsky, 1990a; Pak and Weiner, 1990; Furuya et al., 1991). To measure this import quantitatively, a fluorescencebased assay was developed (Roise, 1992). When mitochondria express a membrane potential across the inner membrane, bound presequences are imported into the organelle. Presequences that have been imported into mitochondria are protected from digestion by externally added trypsin, while those bound to the outer mitochondrial membrane are digested (Section IVB). Since digestion by trypsin is rapid, compared to import, changes in fluorescence that occur upon addition of trypsin can be used to measure the amount of presequence that has been imported (Fig. 5 A). Data from an experiment using this assay are shown in Figure 5B. The kinetics of presequence import into mitochondria have been discussed thoroughly (Roise, 1992). Various models, indistinguishable by the kinetic data, were presented. The most striking observation was that the data could be fit to the simple model shown in Figure 6, where the presequence translocates across the mitochondrial membranes directly from the membrane-bound pool. In this simple scenario, there would be no need for a translocation catalyst, although models that include a translocator cannot be excluded. It is important to note that the rate of
A
Protectfon from protease digestion cIs
, trans
cis
Protease
I N)bs ~ ^Pp ■•■
^PB
"*" ^P,
[PB] = 45 nM
B
N)bs ~ n^o ■*" ^p,
35 nM
22 nM
/
/
12 nM
/
0.2 O CO
/ ^Trypsin |
0}
^ 0.1
/ Presequence
0.0
|<Mitbchon(Jria
0
2
^
4
6
8
10
12
Time (min) Figure 5. (A) Assay to measure import of the fluorescently labeled presequence into mitochondria. Binding and import of the presequence are shown schematically. Initially, the presequence is in equilibrium between free (Pp) and membrane-bound (PB) forms. Presequences that have been imported are represented as P,. The observed fluorescence (Fobs) 's equal to the sum of the fluorescence of all three forms. Upon addition of protease (usually trypsin), Pp and PB (since PB is in equilibrium with Pp) are digested. The digested fragment, PQ, does not bind to the membrane and thus has fluorescence identical to that of Pp. The amount of presequence imported can be calculated from the difference in observed fluorescence before and after protease treatment. It is not necessary, therefore, to know the ratio of free and bound peptide on the trans side of the membrane. (B) Import of the fluorescein-labeled CoxlV presequence into mitochondria. The fluorescence traces from three separate experiments are superimposed. The small initial signal was due to light scattering by the membranes (0.20 ml"^ mitochondrial protein). At 1 minute, the presequence was added, to 50 n M . The fluorescence was relatively low because, under these conditions, most of the presequence was bound to the membranes. Trypsin was added to each experiment, as indicated, after 1, 4, or 8 minutes of import. Presequences that had not been imported were rapidly digested and released from the membrane. The addition of trypsin, therefore, produced an immediate increase in fluorescence. The fluorescence of the imported presequence remained quenched, presumably because it remains bound to the inner membrane. The initial amount of presequence bound, [PBIO = [PjJQoQmax = 45 n M , was calculated by using parameters obtained from the binding curve (Fig. 3A); the amount of presequence remaining bound to the outer membrane at each time point was calculated by using the change in fluorescence that occurred upon the addition of trypsin: [Pg] = [Pj](^Fi^ps\nFj^)Q-^^^ . The amount bound at 1, 4, or 8 minutes is indicated above the figure. Adapted from Roise (1992).
66
Mitochondrial
67
Presequences
[PpIM] [PJ OM
„
.
-^'(PFI+IPBI)
Rate law: v = —
\^ at
, ^ ,
..^,„,
MSO|PB]
= k^lPs] with [PF] =
integrated rate law: In
[PBJO
/fit[M]
[PB]t
(/If50 + (M])
,^„ [MJ
Figures. Kinetic mechanism of presequence translocation. The equilibrium between free and bound presequence Is described by the constant M^Q. The rate of Import from the membrane-bound pool, Pg, can be fit using a first-order rate constant, k^. The derivation of the rate law can be found In Rolse (1992). The membrane potential across the inner mitochondrial membrane, which is required for presequence Import, is Indicated. O M , outer mitochondrial membrane; IM, Inner mitochondrial membrane.
translocation, with a half-time of 5.3 minutes at 20-C, is fast enough that the presequence could be following the same pathway as that used by full-length precursors. It is also significant that a large fraction of the added presequence is imported into the mitochondria. Thus, the import of presequence by this pathway is rapid and efficient, and should not be considered a bypass pathway (Pfaller et al., 1989). Regardless of whether the translocation of the presequence into mitochondria requires a translocator, the translocation does require a membrane potential. In this sense, the translocation of the CoxIV presequence can be compared to the behavior of the positively charged dye rhodamine 123. When applied to living cells, rhodamine 123 is accumulated in mitochondria with high specificity, and the accumulation is strictly dependent on the membrane potential (Wu, 1987; Chen, 1988). Because of the possibility that the CoxIV presequence, like rhodamine 123, could be acting as a membrane-permeant cation, and that this behavior may be responsible for the specific targeting of proteins into mitochondria, the ability of the CoxIV presequence to be imported into lipid vesicles was tested.
68
MERRITT MADUKE and DAVID ROISE
VI. IMPORT INTO PHOSPHOLIPID VESICLES If presequences are able to translocate across the mitochondrial membranes directly, then they might be able to translocate across a pure lipid bilayer. To test this hypothesis, vesicles with a transmembrane potential similar to that of mitochondria were examined in import assays. Phospholipid suspensions were extruded through Nucleopore filters to generate uniform, unilamellar vesicles (Hope et al., 1985), and valinomycin was added to the vesicles to induce a diffusion potential (Figure 7) (Maduke and Roise, 1993). For the import experiments, the CoxIV presequence was labeled with 7V,7V-dimethyl-A^-(iodoacetyl)-A^-(7-nitrobenz-2-oxa-l,3-diazol4-yl)ethylenediamine (NBD iodoacetamide), and import was measured as the time-dependent protection of the presequence from added, membrane-impermeant reagents. One assay was similar to that used to measure import into mitochondria. Presequences that have been imported into the vesicles should become protected from digestion by added trypsin. The other assay was based on a method developed by Mclntyre and Sleight to measure the asymmetry of phospholipids in bilayers (Mclntyre and Sleight, 1991). Sodium dithionite (Na2S204) can rapidly reduce NBD to a nonfluorescent compound. Since dithionite is negatively charged and passes through membranes very slowly, NBD-labeled presequences that have been translocated into the vesicles should be protected from reduction by added dithionite (Fig. 8). Binding of the NBD-labeled presequence to the vesicles was analyzed in the same manner as described for the fluorescein-labeled presequence. With the NBDlabeled presequence, however, the bound form has higher fluorescence than the free, and so an enhancement of fluorescence upon binding was observed. The affinity of the NBD-labeled presequence for both mitochondria and vesicles was slightly higher than that of the fluorescein-labeled presequence, as expected from the difference in charge between the two fluorescent labels. When the presequence was bound to vesicles that expressed an internal negative transmembrane potential, there was a protection of the presequence from digestion by trypsin or reduction by dithionite. The rates of protection from trypsin and dithionite were similar. This result indicated that the two assays were monitoring the same step, the disappearance of the presequence from the vesicle surface. The protection did not occur in vesicles lacking the potential, and it was eliminated by dissolving the vesicles with detergent. That the presequence had indeed been translocated completely across the vesicle membrane was confirmed by experiments using vesicles that contained trypsin. Together, these experiments showed clearly that a mitochondrial presequence can pass through even protein-free lipid bilayers. There are several reasons why the ability of the CoxIV presequence to be imported into phospholipid vesicles is relevant to the mechanism of import of the presequence into mitochondria. First, the import depends on the membrane potential, as does the translocation of presequences into mitochondria. Second, a mutation in the CoxIV presequence that decreases the ability of the presequence to be
Mitochondrial Presequences
69
1)Upid=P0PC:P0PG4:1 2) Suspend at 50 mM in 10 mM NaHEPES, pH 7.0 lOOmMKCI 3) Extrude through 0.1 |xm Nudeopore filters
lOOmM KCI
lOOmM KCI
4) Add valinomycin (K*-specific ionophore)
lOOmM KCI
5) Dilute into assay solution (1:100)
ImM KCI 99 mM NaCI
lOOmM KCI
A \|/
=0
A v
A \|/ =-120mV =-60mVlog
[Kg
Figure 7. Preparation of vesicles used in import assays. POPC, 1 -palmitoyl-2-oieylsn-glycero-3-phosphocholine; POPG, 1 -palmitoyl-2-oieyl-sn-glycero-3-[phosphorac-(1-glycerol)]. The transmembrane potential was calculated from the Nernst equation as shown.
transported into mitochondria decreased the ability of the presequence to be imported into vesicles. Third, the first-order rate constant for the import of the NBD-labeled presequence into vesicles, 0.09 min~^ is comparable to that obtained with mitochondria, 0.6 min~^ Since the mitochondrial membrane potential is probably greater than the membrane potential generated in the vesicle experiments
MERRITT MADUKE and DAVID ROISE
70 Protection from chemical reduction cis
trans
CIS
trans
NagSp^ P *-I^ P
^obs "" '^P.
'^obs = ' ^ P , + ^ P e + ^ P .
NagSgO^
7-Nitro-2,1,3-ben20xadiazolyl (NBD) Fluorescent
7-Amino-2,1,3-t)enzoxadiazolyl (ABD) Non-fluorescent
Figure 8. Protection of imported presequences from sodium dithionite, Na2S204. The diagram is similar to Figure 5A. Pp and PB represent free and bound presequence in which the NBD label has been reduced and is no longer fluorescent. In the chemical reaction shown, R represents the presequence attached through the cysteine side chain.
(Lards et al., 1975; Kamo et al., 1979), and since we have observed that the rate of import of the presequence into vesicles varies as a ftmction of the magnitude of the potential, the observed difference in rates may be negligible. For these reasons, the import of the presequence into protein-free phospholipid vesicles displays properties similar to those observed for import of the presequence into mitochondria. The results suggest that the presequence may inherently be able to translocate directly across the mitochondrial lipids and that this ability may be important for the function of the presequence as a targeting signal.
Vll. MODEL FOR RECOGNITION OF MITOCHONDRIAL PRESEQUENCES A. Binding of Precursors to Membranes In Vivo
If presequences attached to proteins have affinities for lipid bilayers similar to those of the free presequences, most precursors inside a cell would be bound to
Mitochondrial Presequences
71
membranes. The bound state is favored because of the high effective concentration of membranes inside a cell. The dissociation constants measured for the binding of presequences to bilayers range from to 10~^ to 10"^ M (Myers et al., 1987; Frey and Tamm, 1990; Swanson and Roise, 1992; Wang and Weiner, 1994). These values correspond to the concentration of phospholipid necessary to bind half of the presequences in a solution. Because the aqueous volume inside a cell is small, the effective concentration of membranes is extremely large, and the membrane-bound form of the precursors would predominate. To illustrate this effect, assume that a cell is an empty sphere with a diameter of 10 ^im. The internal volume of the cell is 5.2 X 10"^^ liters, and the surface area is 3.1x10^^ A^. Assuming that the sphere is composed entirely of lipid and that two lipids in a bilayer occupy 70 A^ (Evans et al., 1987), the surface of the sphere would contain 1.5 x 10"^^ moles of lipid and the effective concentration of lipid inside the sphere would be 2.9 x 10"^ M. Thus, even an empty cell would have an internal lipid concentration sufficient to bind most of the presequences. In a real cell, the presence of internal organelles would result in an even higher effective concentration of lipid, and the majority of precursors would be bound to internal membranes. These interactions are lost when cells are broken, because the contents of the cytoplasm are diluted into the surrounding solution, and the concentration of membranes, from the perspective of the cytoplasm, decreases dramatically. Although the surface of the mitochondria represents only a small part of the total membrane that is exposed to the cytosol, small changes in the surface charges of the membranes could alter the relative affinities of presequences for mitochondria and minimize the amount of mitochondrial precursors bound inappropriately to nonmitochondrial membranes. In addition, because the nonspecific partitioning into bilayers is rapidly reversible, precursors bound to nonmitochondrial membranes would be free to dissociate and could eventually reach the mitochondria and be imported. A nonspecific binding step would be a reasonable way for precursors to associate with membranes within a cell. B. The Membrane Potential May Determine Specificity of Import
As described above, mitochondrial precursors are likely to insert nonspecifically into any intracellular membrane. Although this insertion is nonspecific, it is rapidly reversible so that any precursors bound inappropriately to nonmitochondrial membranes can return to the cytoplasmic pool. Precursors bound to the surface of a mitochondrion, however, are able to continue on the productive pathway by translocation into the matrix space. Bound precursors should be able to diffuse rapidly over the surface of the mitochondrion (Frey and Tamm, 1990). Precursors that diffuse over contact sites would be initiated into the translocation process because of the strong attraction of the electrical potential for the positive charges of the presequence. The ability of the presequence to create a local disruption in the bilayer may allow it to forge a pathway through the membrane for its passenger
72
MERRITT MADUKE and DAVID ROISE
protein. In addition, at any point during its association with the mitochondrial surface, a precursor could encounter components of the mitochondrial membranes that may help to facilitate its translocation across the membranes. The possibility that the mitochondrial membrane potential plays a role in determining the specificity of uptake of precursors by mitochondria is consistent with the physical properties of presequences and the unique nature of the mitochondrial potential. From the perspective of the cytoplasm, the mitochondrial inner membrane represents the only membrane system with a potential of negative internal polarity (Chen, 1988), and the potential could, therefore, be responsible for the specificity of uptake of mitochondrial precursors. As long as the nonspecific steps leading up to translocation are reversible (Fig. 1), the potential-dependent translocation of a positively charged presequence across the inner membrane could represent the only truly specific step in the entire import process. This model is consistent with the results of experiments described previously showing that presequences behave as membrane-permeant cations. C. Role of Mitochondrial Membrane Proteins
Although their role has not been elaborated in this discussion, the mitochondrial membrane proteins described in the introduction must play a role in the import process. Some of these proteins have been termed receptors, although the mechanism of association of these proteins with their ligands, the precursor proteins, is not yet established. Indeed, there are several reasons why the exact role of the putative outer membrane receptor proteins is uncertain. First, these proteins have never been shown experimentally to interact directly with presequences—the sole determinant of uptake specificity for mitochondrial precursors. In fact, the proteins that have been used most often to measure binding, the ADP/ATP carrier and the outer membrane porin, are both intrinsic membrane proteins that lack cleavable presequences and carry cryptic targeting information. Clearly, they are poor models for typical matrix-targeted precursors, which are soluble proteins with aminoterminal, cleavable presequences. Second, the activities of the receptors are typically tested by the correlation between the effects of treatment of the mitochondrial surface with proteases or specific antibodies and the loss of protein import activity. In general, loss-of-function assays provide only negative results, and in particular, the loss of translocation activity is not a direct measurement of effects on the binding function of a receptor. Third, the binding studies that have been reported were not performed under equilibrium conditions; the mitochondria were washed following the binding experiments in order to remove proteins bound nonspecifically (Pfaller andNeupert, 1987; Sollneretal., 1989,1990; Stegeretal., 1990; Hines and Schatz, 1993), and the observed interaction therefore is not likely to reflect a classic receptor/ligand complex. Rather, the binding may correspond to the type of interactions that occur between unfolded proteins and their molecular chaperones (Landry and Gierasch, 1991), a process not likely to involve specific recognition
Mitochondrial Presequences
73
sites within a protein. Finally, even after inactivation of the mitochondrial outer membrane receptors by treatment with either trypsin or antibodies against the outer membrane proteins, the mitochondria still retain a significant amount of residual translocation activity (Pfaller et al, 1989; Sollner et al., 1990; Hines and Schatz, 1993). This activity, which can be as much as 40% of the translocation observed in untreated mitochondria, still displays the characteristics of the normal translocation process (Pfaller et al., 1989). In one case, the ability of a precursor to be imported into mitochondria that had been treated with protease was almost completely recovered when the precursor was denatured with urea (Glaser and Cumsky, 1990b). The ability of precursor proteins to be imported into protease- and antibody-treated mitochondria demonstrates clearly that the import of proteins into mitochondria can occur independently of at least the MOM 19 and MOM72 surface proteins. Gene disruption experiments also cast doubt upon the exact fiinction of the outer membrane proteins. In yeast, ISP42 (MOM38 in Neurospora) is an outer membrane protein proposed to be involved in the transfer of precursors across the outer mitochondrial membrane (Vestweber et al., 1989). As expected for a protein of such vital ftinction, disruption of the gene coding for ISP42 is lethal in haploid cells (Baker et al., 1990). In contrast, deletion of the gene coding for either of the so-called receptor proteins, Mas20p or MasTOp, had little effect on cells that were given time to adapt to the growth conditions. Mitochondria purified from the adapted cells lacked both Mas20p and MasTOp but were able to import precursor proteins into the matrix (Lithgow et al., 1994). Although it was proposed that there must be a third receptor that can share duties with the other two, it is possible that precursor proteins are able to make their way to the import site without the aid of receptor proteins and that the outer membrane proteins provide some other function for import. Other proteins have also been proposed to be receptors for mitochondrial precursors. In one case, anti-idiotype antibodies raised to mimic a synthetic mitochondrial presequence were found to inhibit the import of precursors into isolated yeast mitochondria (Pain et al., 1990). These antibodies were used to identify a 32-kDa protein that was proposed to be a receptor for presequences. Proteins with similar properties had been identified previously in other systems by chemical crosslinking (Gillespie, 1987) and affinity chromatography (Ono and Tuboi, 1990b) with synthetic presequences. The yeast gene coding for the protein identified by antiidiotype antibodies was cloned (Murakami et al., 1990) and later found to be identical to the gene coding for the phosphate transport protein (Phelps et al., 1991), an integral protein of the mitochondrial inner membrane. It is not yet clear what role, if any, this protein plays in the protein import process, although it has recently been shown to interact directly with presequences (Murakami et al., 1993). The location of this protein in the inner membrane, however, makes a role for the protein as a surface receptor unlikely. Perhaps this protein is simply susceptible to interactions with amphiphilic molecules. Indeed, the ADP/ATP carrier, an inner membrane
74
MERRITT MADUKE and DAVID ROISE
protein belonging to the same family as the phosphate transport protein (Klingenberg, 1990), was recently identified as a specific receptor for the NH2-terminal, amphiphilic segment of pp60^'^'^, the transforming protein of Rous sarcoma virus (Resh and Ling, 1990; Sigal and Resh, 1993). This interaction is unlikely to have any physiological significance—the pp60^'^'^ protein normally associates with the plasma membrane—and the ADP/ATP carrier resides in the mitochondrial inner membrane—but the result shows that proteins in this family are susceptible to interactions with amphiphiles. Further experiments should determine whether the phosphate transporter is actually involved in the import of proteins into mitochondria.
Vm. SUMMARY The results of many studies suggest that precursors destined for the matrix of the mitochondria initially bind to the lipids on the mitochondrial surface. From a physical perspective, mitochondrial presequences are ideally suited to interacting with membrane lipids; presequences lack a consensus sequence, which is characteristic of ligands recognized by protein receptors, but share common physical properties, which impart an affinity for negatively charged surfaces. The development of fluorescence-based assays allowed a quantitative demonstration that a presequence binds directly to mitochondrial lipids. This result, along with the lack of direct evidence for interactions between presequences and receptors, strongly suggests that precursor proteins initially bind to the lipids on the mitochondrial surface. The mechanism of the protein translocation step remains unknown. The CoxIV presequence can translocate into protein-free phospholipid vesicles with the same characteristics observed for translocation into mitochondria. This result suggests that protein import may occur directly through lipids because of the disruptive effects of the presequence. On the other hand, several mitochondrial membrane proteins have been implicated as components of an import machinery. The existence of these proteins suggests that protein import may occur through an aqueous pore created by protein translocators. Direct experimental evidence concerning the mechanism of protein import across the mitochondrial membranes has not yet been obtained, however, and the mechanism could be either of those described above, or some combination of these possibilities. Future mechanistic studies on the transport of full-length precursor proteins into mitochondria should shed light on this problem.
ACKNOWLEDGMENTS We thank Eric Lanzendorf, Osama Khouri, and Chris Sherrill for comments on the manuscript.
Mitochondrial Presequences
75
REFERENCES Allison, D. S. & Schatz, G. (1986). Artificial mitochondrial presequences. Proc. Natl. Acad. Sci. USA, 83,9011-9015. Aoyagi, H., Lee, S., Kanmera, T., Mihara, H., & Kato, T. (1987). Interaction of synthetic fragments of the extension peptide of cytochrome P-450 (SCC) precursor with phospholipid bilayer. J. Biochem. (Tokyo) 102, 813-820. Baker, K. P., Schaniel, A., Vestweber, D., & Schatz, G. (1990). A yeast mitochondrial outer membrane protein essential for protein import and cell viability. Nature 348, 605-609. Becker, K., Guiard, B., Rassow, J., SoUner, T., & Pfanner, N. (1992). Targeting of a chemically pure preprotein to mitochondria does not require the addition of a cytosolic signal recognition factor. J. Biol. Chem. 267, 5637-5643. Bedwell, D. M., Strobel, S. A., Yun, K., Jongeward, G. D., & Emr, S. D. (1989). Sequence and structural requirements of a mitochondrial protein import signal defined by saturation cassette mutagenesis. Mol. Cell. Bipl. 9, 1014^1025. Bruch, M. D., & Hoyt, D. W. (1992). Conformational analysis of a mitochondrial presequence derived from the FpATPase p-subunit by CD and NMR spectroscopy. Biochim. Biophys. Acta 1159, 81-93. Camie, S. & McLaughlin, S. (1983). Large divalent cations and electrostatic potentials adjacent to membranes. A theoretical calculation. Biophys. J. 44, 325-332. Chen, L. B. (1988). Mitochondrial membrane potential in living cells. Annu. Rev. Cell Biol. 4,155—181. Cyr, D. M. & Douglas, M. G. (1991). Early events in the transport of proteins into mitochondria. Import competition by a mitochondrial presequence. J. Biol. Chem. 266, 21700-21708. de Kroon, A. I., de Gier, J., & de Kruijfif, B. (1991). The effect of a membrane potential on the interaction of mastoparan X, a mitochondrial presequence, and several regulatory peptides with phospholipid vesicles. Biochim. Biophys. Acta 1068,111-124. Demel, R. A., Guerts van Kessel, W. S. M., Zwaal, R. F. A., Roelofsen, B., & van Deenen, L. L. M. (1975). Relation between various phospholipase actions on human red cell membranes and the interfacial phospholipid pressure in monolayers. Biochim. Biophys. Acta 406,97-107. Deshaies, R. J., Koch, B. D., Werner-Washbume, M., Craig, E. A., & Schekman, R. (1988). A subfamily of stress proteins facilitates translocation of secretory and mitochondrial precursor polypeptides. Nature 332, 800-805. Eilers, M. & Schatz, G. (1986). Binding of a specific ligand inhibits import of a purified precursor protein into mitochondria. Nature 322, 228-232. Eilers, M. & Schatz, G. (1988). Protein unfolding and the energetics of protein translocation across biological membranes. Cell 52, 481-483. Eisenberg, D. (1984). Three-dimensional structure of membrane and surface proteins. Annu. Rev. Biochem. 53, 595-623. Endo, T., Shimada, I., Roise, D., & Inagaki, F. (1989). N-terminal half of a mitochondrial presequence peptide takes a helical conformation when bound to dodecylphosphocholine micelles: a proton nuclear magnetic resonance study. J. Biochem. (Tokyo) 106, 396-400. Epand, R. M., Hui, W.-H., Argan, C, Gillespie, L. L., & Shore, G. C. (1986). Structural analysis and amphiphilic properties of a chemically synthesized mitochondrial signal peptide. J. Biol. Chem. 261, 10017-10020. Evans, R. W., Williams, M. A., & Tinoco, J. (1987). Surface areas of 1-palmitoyl phosphatidylcholines and their interactions with cholesterol. Biochem. J. 245, 455-462. Firgaira, F. A., Hendrick, J. R, Kalousek, F., Kraus, J. R, & Rosenberg, L. E. (1984). RNA required for import of precursor proteins into mitochondria. Science 226, 1319-1322, Freitag, H., Neupert, W, & Benz, R. (1982). Purification and characterisation of a pore protein of the outer mitochondrial membrane from Neurospora crassa. Eur. J. Biochem. 123, 629-636.
76
MERRITT MADUKE and DAVID ROISE
Frey, S. & Tamm, L. K. (1990). Membrane insertion and lateral diffusion of fluorescence-labelled cytochrome c oxidase subunit IV signal peptide in charged and uncharged phospholipid bilayers. Biochem. J. 272, 713-719. Furuya, S., Mihara, K., Aimoto, S., & Omura, T. (1991). Cytosolic and mitochondrial surface factorindependent import of a synthetic peptide into mitochondria. EMBO J. 10, 1759-1766. Gething, M. J. & Sambrook, J. (1992). Protein folding in the cell. Nature 355, 33-45. Gillespie, L. L. (1987). Identification of an outer mitochondrial membrane protein that interacts with a synthetic signal peptide. J. Biol. Chem. 262, 7939-7942. Gillespie, L. L., Argan, C, Taneja, A. T., Hodges, R. S., Freeman, K. B., & Shore, G. C. (1985). A synthetic signal peptide blocks import of precursor proteins destined for the mitochondrial inner membrane or matrix. J. Biol. Chem. 260, 16045-16048. Glaser. S. M. & Cumsky, M. G. (1990a). Localization of a synthetic presequence that blocks protein import into mitochondria. J. Biol. Chem. 265, 8817-8822. Glaser, S. M. & Cumsky, M. G. (1990b). A synthetic presequence reversibly inhibits protein import into yeast mitochondria. J. Biol. Chem. 265, 8808-8816. Gorlich, D. & Rapoport, T. A. (1993). Protein translocation into proteoliposomes reconstituted from purified components of the endoplasmic reticulum membrane. Cell 75, 615-630. Hammen, P. K., Gorenstein, D. G., & Weiner, H. (1994). Structure of the signal sequences for two mitochondrial matrix proteins that are not proteolytically processed upon import. Biochemistry 33,8610-8617. Hines, V. & Schatz, G. (1993). Precursor binding to yeast mitochondria. A general role for the outer membrane protein Mas70p. J. Biol. Chem. 268, 449-454. Hope, M. J., Bally, M. B., Webb, G., & CuUis, R R. (1985). Production of large unilamellar vesicles by a rapid extrusion procedure. Characterization of size distribution, trapped volume and ability to maintain a membrane potential. Biochim. Biophys. Acta 812, 55-65. Horwich, A. L., Kalousek, F., Mellman, I., & Rosenberg, L. E. (1985). A leader peptide is sufficient to direct mitochondrial import of a chimeric protein. EMBO J. 4, 1129-1135. Hoyt, D. W., Cyr, D. M., Gierasch, L. M., & Douglas, M. G. (1991). Interaction of peptides corresponding to mitochondrial presequences with membranes. J. Biol. Chem. 266, 21693—21699. Hurt, E. C. & van Loon, A. P. G. M. (1986). How proteins find mitochondria and intramitochondrial compartments. Trends. Biochem. Sci. 11, 204-207. Hurt, E. C , Pesold-Hurt, B., & Schatz, G. (1984). The cleavable prepiece of an imported mitochondrial protein is sufficient to direct cytosolic dihydrofolate reductase into the mitochondrial matrix. FEBSLett. 178,306-310. Hurt, E. C, Muller, U., & Schatz, G. (1985a). The first twelve amino acids of a yeast mitochondrial outer membrane protein can direct a nuclear-encoded cytochrome oxidase subunit to the mitochondrial inner membrane. EMBO J. 4, 3509-3518. Hurt, E. C , Pesold-Hurt, B., Suda, K., Oppliger, W., & Schatz, G. (1985b). The first twelve amino acids (less than half of the pre-sequence) of an imported mitochondrial protein can direct mouse cytosolic dihydrofolate reductase into the yeast mitochondrial matrix. EMBO J. 4, 2061—2068. Hwang, S. T, Wachter, C, & Schatz, G. (1991). Protein import into the yeast mitochondrial matrix. A new translocation intermediate between the two mitochondrial membranes. J. Biol. Chem. 266, 21083-21089. Kaiser, E. T. & Kezdy, F. J. (1983). Secondary structures of proteins and peptides in amphiphilic environments (A review). Proc. Natl. Acad. Sci. USA, 80, 1137-1143. Kalousek, E, Neupert, W., Omura, T, Schatz, G., & Schmitz, U. K. (1993). Uniform nomenclature for the mitochondrial peptidases cleaving precursors of mitochondrial proteins. Trends. Biochem. Sci. 18,249-249. Kamo, N., Muratsugu, M., Hongoh, R., & Kobatake, Y. (1979). Membrane potential of mitochondria measured with an electrode sensitive to tetraphenyl phosphonium and relationship between proton
Mitochondrial Presequences
77
electrochemical potential and phosphorylation potential in steady state. J. Membr. Biol. 49, 105-121. Karslake, C, Piotto, M. E., Pak, Y. K., Weiner, H., & Gorenstein, D. G. (1990). 2D NMR and structural model for a mitochondrial signal peptide bound to a micelle. Biochemistry 29, 9872-9878. Kiebler, M., Becker, K., Pfanner, N., & Neupert, W. (1993). Mitochondrial protein import: Specific recognition and membrane translocation of preproteins. J. Membr. Biol. 135, 191—207. Klingenberg, M. (1990). Mechanism and evolution of the uncoupling protein of brown adipose tissue. Trends. Biochem. Sci. 15, 108-112. Landry, S. J. & Gierasch, L. M. (1991). Recognition of nascent polypeptides for targeting and folding. Trends. Biochem. Sci. 16, 159-163. Langner, M., Cafiso, D., Marcelja, S., & McLaughlin, S. (1990). Electrostatics of phosphoinositide bilayer membranes. Theoretical and experimental results. Biophys. J. 57, 335-349. Laris, P. C, Bahr, D. P., & Chaffee, R. R. J. (1975). Membrane potentials in mitochondrial preparations as measured by means of a cyanine dye. Biochim. Biophys. Acta 376,415-425. Lemire, B. D., Fankhauser, C, Baker, A., & Schatz, G. (1989). The mitochondrial targeting function of randomly generated peptide sequences correlates with predicted helical amphiphilicity. J. Biol. Chem. 264, 20206-20215. Lithgow, T., Junne, T., Wachter, C, & Schatz. G. (1994). Yeast mitochondria lacking the two import receptors Mas20p and Mas70p can efficiently and specifically import precursor proteins. J. Biol. Chem. 269, 15325-15330. Maduke, M. & Roise, D. (1993). Import of a mitochondrial presequence into protein-free phospholipid vesicles. Science 260, 364-367. Martin, J., Mahlke, K., & Pfanner, N. (1991). Role of an energized inner membrane in mitochondrial protein import. A^' drives the movement of presequences. J. Biol. Chem. 266, 18051-18057. Mclntyre, J. C. & Sleight, R. G. (1991). Fluorescence assay for phospholipid membrane asymmetry. Biochemistry 30, 11819-11827. McKnight, C. J., Rafalski, M., & Gierasch, L. M. (1991). Fluorescence analysis of tryptophan-containing variants of the LamB signal sequence upon insertion into a lipid bilayer. Biochemistry 30, 6241-6246. Murakami, H., Pain, D., & Blobel, G. (1988). 70-kD heat shock-related protein is one of at least two distinct cytosolic factors stimulating protein import into mitochondria. J. Cell Biol. 107, 2051— 2057. Murakami, H., Blobel, G., & Pain, D. (1990). Isolation and characterization of the gene for a yeast mitochondrial import receptor. Nature 347, 488-491. Murakami, H., Blobel, G., & Pain, D. (1993). Signal sequence region of mitochondrial precursor proteins binds to mitochondrial import receptor. Proc. Natl. Acad. Sci. USA 90, 3358-3362. Myers, M., Mayorga, O. L., Emtage, J., & Freire, E. (1987). Thermodynamic characterization of interactions between ornithine transcarbamylase leader peptide and phospholipid bilayer membranes. Biochemistry 26, 4309-4315. Nicolay, K., Laterveer, F. D., & van Heerde, W. L. (1994). Effects of amphipathic peptides, including presequences, on the functional integrity of rat liver mitochondrial membranes. J. Bioenerg. Biomembr. 26, 327-334.Ohta, S. & Schatz, G. (1984). A purified precursor polypeptide requires a cytosolic proteinfractionfor import into mitochondria. EMBO J. 3, 651-657. Ono, H. & Tuboi, S. (1988). The cytosolic factor required for import of precursors of mitochondrial proteins into mitochondria. J. Biol. Chem. 263, 3188-3193. Ono, H. & Tuboi, S. (1990a). Presence of the cytosolic factor stimulating the import of precursor of mitochondrial proteins in rabbit reticulocytes and rat liver cells. Arch. Biochem. Biophys. 277, 368-373. Ono, H. & Tuboi, S. (1990b). Purification of the putative import-receptor for the precursor of the mitochondrial protein. J. Biochem. (Tokyo) 107, 840-845.
78
MERRITT MADUKE and DAVID ROISE
Pain, D., Murakami, H., & Blobel, G. (1990). Identification of a receptor for protein import into mitochondria. Nature 347, 414-449. Pak, Y. K. & Weiner, H. (1990). Import of chemically synthesized signal peptides into rat liver mitochondria. J. Biol. Chem. 265, 14298-14307. Pfaller, R. & Neupert, W. (1987). High-affinity binding sites involved in the import of porin into mitochondria. EMBO J. 6, 2635-2642. Pfaller, R., Pfanner. N., & Neupert, W. (1989). Mitochondrial protein import. Bypass of proteinaceous surface receptors can occur with low specificity and efficiency. J. Biol. Chem. 264, 34—39. Pfanner. N. & Neupert, W. (1985). Transport of proteins into mitochondria: a potassium diffusion potential is able to drive the import of ADP/ATP carrier. EMBO J. 4, 2819-2825. Pfanner, N., Rassow, J., Wienhues, U., Hergersberg, C, Sollner, T., Becker, K., & Neupert, W. (1990). Contact sites between inner and outer membranes: Structure and role in protein translocation into the mitochondria. Biochim. Biophys. Acta 1018, 239-242. Phelps, A., Schobert, C. T., & Wohlrab, H. (1991). Cloning and characterization of the mitochondrial phosphate transport protein gene from the yeast Saccharomyces cerevisiae. Biochemistry 30, 248-252. Pon, L., Moll, T., Vestweber, D., Marshallsay, B., & Schatz, G. (1989). Protein import into mitochondria: ATP-dependent protein translocation activity in a submitochondrial fraction enriched in membrane contact sites and specific proteins. J. Cell Biol. 109, 2603-2616. Resh, M. D. & Ling, H. P. (1990). Identification of a 32K plasma membrane protein that binds to the myristylated amino-terminal sequence of p60^"^^ Nature 346, 84—86. Riezman, H., Hay, R., Gasser, S., Daum, G., Schneider, G., Witte, C , & Schatz, G. (1983). The outer membrane of yeast mitochondria: Isolation of outside-out sealed vesicles. EMBO J. 2,1105-1 111. Roise, D. (1992). Interaction of a synthetic mitochondrial presequence with isolated yeast mitochondria: mechanism of binding and kinetics of import. Proc. Natl. Acad. Sci. USA 89, 608-612. Roise, D. (1993). The amphipathic helix in mitochondrial targeting sequences. In: The Amphipathic Helix (Epand, R. M., Ed.), pp. 257-283. CRC Press, Boca Raton. Roise, D., Horvath, S. J., Tomich, J. M., Richards, J. H., & Schatz, G. (1986). A chemically synthesized pre-sequence of an imported mitochondrial protein can form an amphiphilic helix and perturb natural and artificial phospholipid bilayers. EMBO J. 5, 1327-1334. Roise, D., Theiler, R, Horvath, S. J., Tomich, J. M., Richards, J. H., Allison, D. S., & Schatz, G. (1988). Amphiphilicity is essential for mitochondrial presequence function. EMBO J. 7, 649-653. Schatz, G. (1993). The protein import machinery of mitochondria. Protein Sci. 2, 141-146. Schleyer, M. & Neupert, W. (1985). Transport of proteins into mitochondria: translocational intermediates spanning contact sites between outer and inner membranes. Cell 43, 339-350. Schwaiger, M., Herzog, V., & Neupert, W. (1987). Characterization of translocation contact sites involved in the import of mitochondrial proteins. J. Cell Biol. 105, 235-246. Seelig, A. (1987). Local anesthetics and pressure: A comparison ofdibucaine binding to lipid monolayers and bilayers. Biochim. Biophys. Acta 899, 196-204. Sigal, C. T. & Resh, M. D. (1993). The ADP/ATP carrier is the 32-kilodalton receptor for an NH2-terminally myristylated src peptide but not for pp60^'^ polypeptide. Mol. Cell. Biol. 13, 3084-3092. Skerjanc, I. S., Shore, G. C, & Silvius, J. R. (1987). The interaction of a synthetic mitochondrial signal peptide with lipid membranes is independent of transbilayer potential. EMBO J. 6, 3117—3123. Sollner, T., Griffiths, G., Pfaller, R., Pfanner, N., & Neupert, W. (1989). MOM19, an import receptor for mitochondrial precursor proteins. Cell 59, 1061—1070. Sollner, T., Pfaller, R., Griffiths, G., Pfanner, N., & Neupert, W. (1990). A mitochondrial import receptor for the ADP/ATP carrier. Cell 62, 107-115. Steger, H. R, Sollner, T., Kiebler, M., Dietmeier, K. A., Pfaller, R., Trulzsch, K. S., Tropschug, M., Neupert, W., & Pfanner, N. (1990). Import of ADP/ATP carrier into mitochondria: Two receptors act in parallel. J. Cell Biol. HI, 2353-2363.
Mitochondrial Presequences
79
Stuart, R. A., Cyr, D. M., Craig, E. A., & Neupert, W. (1994). Mitochondrial molecular chaperones: Their role in protein translocation. Trends. Biochem. Sci. 19, 87-92. Surewicz, W. K. & Epand, R. M. (1984). Role of peptide structure in lipid-peptide interactions: A fluorescence study of the binding of pentagastrin-related pentapeptides to phospholipid vesicles. Biochemistry 23, 6072-6077. Swanson, S. T. & Roise, D. (1992). Binding of a mitochondrial presequence to natural and artificial membranes: Role of surface potential. Biochemistry 31, 5746-5751. Tamm, L. K. (1986). Incorporation of a synthetic mitochondrial signal peptide into charged and uncharged phospholipid monolayers. Biochemistry 25, 7470-7476. Tamm, L. K. (1991). Membrane insertion and lateral mobility of synthetic amphiphilic signal peptides in lipid model membranes. Biochim. Biophys. Acta 1071, 123-148. Tamm, L. K. & Bartoldus, I. (1990). Secondary structure of a mitochondrial signal peptide in lipid bilayer membranes. FEBS Lett. 272, 29-33. Vestweber, D., Brunner, J., Baker, A., & Schatz, G. (1989). A 42K outer-membrane protein is a component of the yeast mitochondrial protein import site. Nature 341, 205—209. von Heijne, G. (1986). Mitochondrial targeting sequences may form amphiphilic helices. EMBO J. 5, 1335-1342. Wachter, C, Schatz, G., & Glick, B. S. (1994). Protein import into mitochondria: The requirement for external ATP is precursor-specific whereas intramitochondrial ATP is universally needed for translocation into the matrix. Mol. Biol. Cell. 5, 465—474. Wang, Y. & Weiner, H. (1994). Evaluation of electrostatic and hydrophobic effects on the interaction of mitochondrial signal sequences with phospholipid bilayers. Biochemistry 33, 12860-12867. Wu, F.-S. (1987). Localization of mitochondria in plant cells by vital staining with rhodamine 123. Planta 171, 346-357. Zardeneta, G. & Horowitz, P. M. (1992). Analysis of the perturbation of phospholipid model membranes by rhodanese and its presequence. J. Biol. Chem. 267, 24193-24198.
This Page Intentionally Left Blank
BACTERIAL TOXIN TRANSPORT: THE HEMOLYSIN SYSTEM
Jonathan A. Sheps, Fang Zhang, and Victor Ling
Abstract 82 I. Introduction 82 II. ThePlaceofHlyB in the Phylogeny of ABC Transporters 85 III. The Hemolysin Transporter 89 A. Cellular Localization of the Hemolysin Transporter 90 B. Transmembrane Topology of HlyB 91 C. Energeticsof Hemolysin Transport: Proton Motive Force and ATP 92 D. HlyD: Structure and Function 94 E. TolC: The Universal Component 94 F. The Transport Complex 95 IV. The Hemolysin A Signal Sequence 96 A. Localization of the Transport Signal to the COOH Terminus of HlyA . . . 96 B. Cross-Species Comparison of Signal Sequences 98 C. Features ofthe HlyA COOH-Terminal Signal. 99 V. The Mechanism of Transport 104 A. Hemolysin A—Hemolysin B interactions 104 B. Hemolysin A-Membrane Interactions 106 C. The Role of Ca ^-Binding Repeats in Transport 107
Membrane Protein Transport Volume 3, pages 81-118.
Copyright © 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 1-55938-989-3 81
82
JONATHAN A. SHEPS, FANG ZHANG, and VICTOR LING
D. Comparison to the General Secretory Pathway VI. Conclusion Acknowledgments References
109 110 Ill Ill
ABSTRACT Uropathogenic Escherichia coli strains often secrete a hemolytic toxin, a-hemolysin (HlyA), a 107-kDa protein, into the culture medium. Transport of this protein across the cell envelope is dependent on a 50-amino acid signal at the COOH terminus of HlyA and on the expression of three envelope proteins: HlyB, HlyD, and TolC. Of these the hemolysin B protein (HlyB) bears structural and sequence homology to the ATP-binding cassette (ABC) family of ATP-dependent transport proteins. ABC transporters have been found to be involved in the transport of a wide range of substances across biological membranes in both eukaryotes and eubacteria. Certain members of the ABC transporter family have broad substrate specificities, and this family includes at least two distinct mammalian multidrug transporters. Biochemically, the hemolysin transporter is being characterized in terms of its energy requirements and the organization of the transmembrane complex as well as the biophysical behavior of the signal sequence. Genetic methods are being used to define the functional interactions between the signal sequence and the transporter. This chapter focuses on the complex nature of the hemolysin signal and its relationship to signals from toxin secretion systems found in other species. The extent of multivalent recognition as a feature of substrate specificity in ABC transporters, even those with unique natural substrates, is surprising. The hemolysin system can be used as a model for genetic and biochemical studies of the molecular mechanisms that underlie multidrug resistance and the evolution of different functions in a versatile family of transport proteins.
I. INTRODUCTION Many biological systems depend on the specific translocation of proteins across cellular membranes by dedicated transport complexes of membrane-associated proteins. The Escherichia coli hemolysin exporter catalyzes the translocation of the a-hemolysin (HlyA) of £". coli (a protein toxin of 107 kDa) across both membranes of the Gram-negative cell envelope to the extracellular medium in a single step (Gray et al., 1989). This transport system is easily manipulated since, unlike the general secretory pathway, it is not essential for cell survival and is composed of only three membrane proteins. Transport of HlyA is dependent on a COOH-terminal 50-amino acid signal sequence (Koronakis et al., 1989; Hess et al., 1990) and is specifically recognized and translocated by the action of inner membrane proteins HlyB and HlyD (Mackman et al., 1985a; Wang et al., 1991; Juranka et al., 1992)
Hemolysin
Transport
83
and the outer membrane protein TolC (Wandersman and Delepelaire, 1990). The latter is a host protein, whereas the inner membrane proteins, along with the toxin itself, are encoded by a single operon (Fig. 1), present on a plasmid or inserted into the chromosome (sometimes between flanking insertion elements) (Zabala et al., 1984) in uropathogenic strains ofE. coli (Hacker and Hughes, 1985). HlyB is a member of the ATP-binding cassette (ABC) transporter superfamily (Juranka et al., 1989) and is thought to be responsible for specific recognition of the HlyA signal (Zhang et al., 1993b) and transduction of energy from ATP hydrolysis to transport or pore opening (Koronakis et al., 1993). HlyD is an inner membrane-anchored protein, the bulk of which is located in the periplasm (Wang et al., 1991; Schiilein et al., 1992) and may serve to couple HlyB to TolC via a path free of interference by other periplasmic constituents (Fig. 2). TolC is an outer membrane porin that probably comprises the outer section of the transport channel (Benzetal., 1993). The hemolysin translocator of E. coli is the best studied of a family of related protein toxin secretion systems distributed through numerous Gram-negative bacterial species. Proteins of this kind are named the RTX (for repeat toxin) family (Felmlee and Welch, 1988) since their only conserved primary structure feature is a series of glycine and aspartate-rich nonapeptide repeats that are involved in binding Ca^"*" and perhaps in targeting the toxins to their target cells (Ludwig et al., 1988; Boehm et al., 1990a,b). The repeat domain also influences transport of some fusion proteins (Kenny et al., 1991) but is not essential for this. HlyB and other RTX transporters are a small part of the increasingly well-studied ABC transporter superfamily of ATP-dependent membrane-bound transport systems (for more detailed reviews see Higgins, 1992; Fath and Kolter, 1993; and
Mor
^
Structural —- ^
,
———^
HlyC
HiyA
HlyB
20 kOa
107 kDa
66 kDa
vl 1
ExFK>rt
HlyD 53 kDa
RN>
Figure 1. The hemolysin determinant. The functions for each protein product are shown above and the molecular weights are shown below the gene arrangement. Two mRNA transcripts are initiated from the hlyC promoter, the longer one dependent on readthrough of the rho-independent terminator sequence upstream of h/yB (Koronakis et a!., 1988a). A putative promoter sequence has been identified upstream of hlyB (Blight et al., 1995), but whether an mRNA is produced from this in the intact operon is not known.
84
JONATHAN A. SHEPS, FANG Z H A N G , and VICTOR LING
Childs and Ling, 1994). These transporters consist of a conserved ATP-binding cassette (from which the name ABC derives) linked (covalently or noncovalently) to a block of five to eight transmembrane segments (Fig. 2). The ABC transporters are distributed throughout eukaryotic and eubacterial organisms. The ABC family as a whole is responsible for transporting a diverse array of substrates that ranges from ions to proteins. Individual ABC transporters commonly either have multiple natural substrates (such as the mammalian multidrug transporter, P-glycoprotein (Pgp)) or can accommodate substrates normally associated with other ABC transporters. The hemolysin transporter, for example, can transport a number of other RTX toxins, despite a large degree of sequence divergence.
Figure 2. A diagram of the hemolysin transport system in the E. coli membrane. Two HlyB (cylinders and ribbons) and two HlyD (golf-tee shape) molecules are located in the inner membrane. Each HlyB molecule contains eight transmembrane helices (cylinders) and a cytoplasmic domain with the ATP binding motifs (the fattened region of the ribbons). Most of HlyD is located in the periplasmic space with one transmembrane helix to anchor the molecule in the inner membrane. The TolC molecule is in the outer membrane. The stoichiometry of the complex is hypothetical, HlyB (and by extension HlyD) being represented as a dimer by analogy to other ABC transporters.
Hemolysin Transport
85
The most captivating problem in the ABC transporter family is that of substrate specificity. The challenge is to understand how a single family of proteins can accommodate such a wide range of substrates within what is likely a conserved transport mechanism. The hemolysin system allows genetic manipulation of both transporter and substrate to reveal details of their interaction. Extensive mutational analysis has revealed that the COOH-terminal signal of HlyA is itself a redundant, overlapping set of distinct recognition motifs, making the transport signal very resistant to mutation. Mutational studies of a number of ABC transporters, including HlyB (Zhang et al., 1993b), has shown that the specificity determinants generally reside in the transmembrane domains and in the cytoplasmic loops connecting them. Analysis of interacting pairs of HlyA and HlyB mutants is beginning to unravel the determinants of substrate specificity in at least one ABC transporter. The ability of HlyB to recognize a variety of RTX toxins (or severely mutated versions of its natural substrate) may point to a conserved mechanism for recognition of multiple substrates that could underlie the diversity seen in the phenomenon of multidrug resistance, as well as the adaptability of the ABC transporter family as a whole.
IL THE PLACE OF HlyB IN THE PHYLOGENY OF ABC TRANSPORTERS The ABC superfamily extends throughout prokaryotic and eukaryotic organisms, with multiple transporters performing numerous functions in each cell (Higgins, 1992; Path and Kolter, 1993; Childs and Ling, 1994) (Table 1). At least 80 ABC transporters have been cloned from a wide variety of organisms. We have performed a phylogenetic analysis of selected members of the family based on amino acid sequences. The resulting phylogenetic tree reveals a complex pattern, with no clear division emerging between pro- and eukaryotic parts of the family (Fig. 3). Near the base of the tree are bacterial periplasmic binding protein-dependent permeases (for a review see Doige and Ames, 1993) in which ABC domains are associated noncovalently with multiple membrane spanning (TM) domains. These generally import small molecules into the cell; and although the family of permeases is large and diverse each individual member is usually restricted in its substrate specificity. The middle and upper branches of our tree (Fig. 3) show a mix of bacterial and eukaryotic exporters in which the ABC domain has become frised to the COOH terminus of a multiple TM domain. Certain of the eukaryotic members of the family possess a tandem duplication of this motif Even though the duplicated arrangement (such as that of Pgp) is restricted to eukaryotes, it appears to have arisen at least twice independently in two separate parts of the ABC transporter family, which must themselves have diverged prior to the division between eukaryotes and bacteria. Fath and Kolter (1993) presented a phylogeny for bacterial ABC transporters that differs in some respects from our tree. The inclusion of eukaryotic sequences in our analysis is largely responsible for the differences. They divided the bacterial ABC
86
JONATHAN A. SHEPS, FANG ZHANG, and VICTOR LING Table 1. ABC Transporters Included in the Phylogenetic Tree
Gene name AprD HasD PrtD Tapl CvaB NdvA ChvA HepA LktB HlyB CyaB Pgpl AtPgpl PfMDR LcnC Ste6 MRP yCFl CFTR SpaB Pmp70 Brown White Pdr5 McbF Nodi MalK HisP OppD UvrA Note:
Substrates Metalloprotease Heme binding protein Metalloprotease Peptides Colicin V p-(l-2)Glucan p-(l-2)Glucan
Species
Pseudomonas aeruginosa Serratia marcesens Erwinia chrysanthemi Mammals Escherichia coli Rhizobium meliloti Agrobacterium tumefaciens Polysaccharides? Anabaena Leukotoxin A Pasteurella haemolytica Hemolysin A Escherichia coli Cyclolysin Bordetella pertussis Drugs, steroids, C\~ Mammals ? Arabidopsis thaliana Chloroquine, mefloquine?' Plasmodium falciparum Lactococcin A Lactococcus lactis Saccharomyces cerevisiae a-factor Drugs Cadmium
cr Subtilin Peroxisomal proteins? Pteridine precursors Pteridine precursors Drugs Microcin B17 Lipooligosaccharides? Maltose Histidine Oligopeptides NA
References Duong etal., 1992 Letoffe et al., 1993, 1994a Letoffe et al., 1990 Trowsdale et al., 1990 Gilsonetal., 1990 Stanfield et al., 1988 Cangelosietal., 1989
Holland and Wolk, 1990 Strathdee and Lo, 1989 Juranka et al., 1992 Glaseretal, 1988 Gerlach et al., 1986 Dudler and Hertig, 1992 Footeetal., 1989 Stoddard et al., 1992 McGrath & Varshavsky, 1989 Cole etal., 1992 Mammals Saccharomyces cerevisiae Szczypka et al., 1994 Riordan et al., 1989 Mammals Chung etal., 1992 Bacillus subtilus Kamijoetal., 1990 Mammals Drosophila melanogaster Dreesen et al., 1988 Drosophila melanogaster O'Hareetal., 1984 Saccharomyces cerevisiae Balzietal., 1994 Escherichia coli Garrido et al., 1988 Rhizobium leguminosarum ! Evans & Downie, 1986 Escherichia coli Gilsonetal., 1982 Salmonella typhimurium Higgins et al., 1982 Salmonella typhimurium Higgins et al., 1985 Escherichia coli Husainetal., 1986
Genes are listed in the order in which they appear on the phylogenetic tree in Figure 3. Substrates listed are those naturally associated with the transporter, as far as is known. In many cases cross-complementation is possible among ABC transporters.
figures. Phylogenetic tree of ABC transporters. Thirty representative members of the ABC transporter family were aligned and their relationships assessed by comparison of amino acid sequences. A region of approximately 200 amino acids from the conserved ATP-binding cassette of each protein v^as used. Where tv^o ABC domains are present in a single polypeptide only the COOH-terminal one was considered. Sequences were aligned using the PILEUP program in the University of Wisconsin GCG package of computer programs. The alignment was edited to remove uninformative sites using MacClade (Maddlson and Maddison, 1992). Poorly conserved amino acid positions at which ten or more different amino acids could be found among the 30 sequences (counting gaps as a single "kind" of amino acid) were excluded from the analysis.
Hemolysin Transport
87 lUvrA lOppD ]Hi8P iMalK ]Whtte iBrown iPdrS iMcbF iNocll |Pmp70 iSpaB |Ste6 ICFTR lyCFI IMRP llxnC iCyaB iHlyB ILMB iPfMDR lAtPgpl iChPgpl iHepA iChvA iNdvA iCvaB iTapl iPrtD iHasD lAprD
psj
m B ■1
s
B ■■
gggaj
^M
Figure 3. (Continued) This left 134 reasonably conservative positions at which amino acid identities were phylogenetically infornnative. This data set was subjected to cladistic analysis using Hennig86 (Farris, 1988), with UvrA (the only nontransporter in the sample) as the outgroup. All characters were treated as unordered. This produced four equally parsimonious trees (EPTs) of length 1485 with a consistency index of 0.56 and a retention index of 0.45. Use of the successive approximations weighting option in Hennig86 produced a single tree with CI = 0.65 and Rl = 0.57. The tree in the figure is the strict consensus tree of the four EPTs from the unweighted data set plus the successive approximations weighting tree. Superimposed on the amino acid phylogeny is shading Indicating the overall domain organization of the proteins. Boxes with vertical lines indicate blocks of multiple transmembrane (TM) domains. ABC, domains containing conserved ATP-binding motifs, shaded or patterned to match the tree. The first structural class (in which ABC and TM blocks are not contiguous) includes proteins (such as Brown, White and Pdr5) whose domain organization is more complex (not shown), but which differs from the two other patterns figured here and which presumably evolved Independently. See Table 1 for information on the individual proteins.
Table 2. Ability of RTX Toxins to Be Secreted by Heterologous Transportersa Secreted proteins
Transporters Species
E. coli I? haemoly~ica E. chrysanthemi S. marcescens S. marcescens I? aeruginosa B. pertussism E. coli 03 03
IM
OM
HlyB HlyD TolC LktB LktD ? PrtD PrtE PrtF ?
HasD HasE AprD AprE CyaB CyaD CvaB CvaA
HlyA
LktA
PrtB
HlyA
100%~
1%'
+g
nt
d
PrtSM
HasA j
AprA
CyaA
ColV
-
14%~
+e
20%~
+IY
+'
nt
nt
r, L
nt
-j
120%~
-
20%f
PrtSM
nt
+k
HasA
nt nt
nt nt nt
nt +I nt nt nt nt nt
nt nt
? ?
nt
4
LktA nt nt nt
AprF CyaE TolC
nt nt
nt nt
nt nt
nt nt
nt nt
nt
CyaA
nt nt nt nt
-
nt
nt
nt
nt
nt
nt
ColV
-c.f
f
PrtB nt
Nod0
AprA
h
Notes: 'All of the experiments were done in E. coli. Except for (b), the percentage was calculated by comparison to the protein secreted by its authentic transporter in E. coli. b ~ kcan t ~ be secreted by HlyBID (Highlander et al. 1989; Chang et al., 1989). When COOH-terminal70 amino acids of LktA replaced the HlyA COOH-terminal sequence, the HlyA-LktA fusion protein can be secreted with same efficiency as HlyA (Zhang et al.. 1993a). 'Delepelaire and Wandersman (1990). d~uzm et a1. (1991). 'Masure et al. (1990); ~ e b and o Ladant (1993). '~athet al. (1991). BStrathdee and Lo, unpublished result. h ~ i t o f fand i Wandersman (1992) 'Letoffe et al. (1991). 'Letoffe et al. (1994a). In these transport experiments, the HasD, E proteins were co-expressed with either TolC or PrtF. HasA could inhibit secretion of PrtB and C, although it could not be secreted by the PrtD,E,F transport system in E. coli. k~etoffe et al. (1993). In this transport experiment, the HasD, E proteins were co-expressed with TolC. ' ~ c h e uet al. (1992). NodO has no linked transporter but can be secreted by unknown factors in many strains of Rhizobia in addition to Hly and Prt systems. "'The transporter proteins cannot be expressed in E. coli nt, not tested The ability to be transported, if not quantified. is indicated by +and inability to be transported by -.
Hemolysin Transport
89
transporters into two groups, those in which the ABC domains and the transmembrane domains are carried on separate polypeptides and those in which they are fused, as in HlyB. Our analysis indicates that the latter structure is derived from the former and that additional duplications of HlyB-like proteins have occurred to yield the Pgp- and CFTR-like protein families in eukaryotes. The RTX toxin secretion proteins occupy branches of the tree that are positioned close to the entirely eukaryotic Pgp/MDR branch. The RTX toxin transporters fall into at least two groups: the pore-forming toxin transporters (hemolysin and leukotoxin) and the metalloprotease transporters. The cyclolysin of Bordetella pertussis is an adenylate cyclase protein fused to the NH2 terminus of a hemolysinlike protein (Glaser et al., 1988). The colicin V secretion protein of £. coli, although not strictly an RTX transporter, seems to be related to the metalloprotease secretion genes. HasD is capable of secreting an RTX toxin, even though its natural substrate is a non-RTX heme-binding protein (Letoffe et al., 1994b). TAPl, a transporter of antigenic peptides (Trowsdale et al., 1990) in the ER of mammalian cells, appears also to be related to the metalloprotease transporters. Among the RTX family of bacterial protein exporters it is quite common for individual transporters to have some ability to secrete the proteins normally associated with other members of the family (Table 2). Of all the RTX transporters it appears that HlyB,D is the most flexible as regards substrate specificity. The E. coli hemolysin operon is often plasmid borne and is flexible enough to rely on host factors like TolC for fiinction and to sometimes exchange substrates with leukotoxin-like systems in nature (Chang et al., 1991). An extreme case of the flexibility often found in plasmid-based genetic systems is the NodO gene of Rhizobium leguminosarum. This is an RTX protein found in some strains oiR. leguminosarum, which forms ion channels in target membranes (Sutton et al., 1994), but rather than being a toxin it is a nonessential signaling molecule that promotes root nodulation in legumes (Economou et al., 1990). The NodO-containing plasmid encodes no transport ftmction, and NodO secretion relies on unknown host transporters (Scheu et al., 1992). NodO is able to use RTX transporters of both subfamilies (Table 2) and can be secreted by many strains of rhizobia, so the potential host range of this protein is much wider than its present distribution.
III. THE HEMOLYSIN TRANSPORTER The hemolysin transporter is generally assumed to function as a multi-subunit complex that spans the two membranes of the E. coli envelope and is composed of the proteins HlyB, HlyD, and TolC. HlyB is usually thought of as a dimer by analogy to other ABC transporters (Kerppola et al., 1991). HlyD is generally drawn as spanning the periplasmic space and thereby connecting HlyB to TolC via protein-^jrotein interactions (Fig. 2). Energy for transport of the presumably unfolded HlyA is assumed to come from ATP hydrolysis by virtue of the conserved ATP-binding domain that HlyB shares with other ABC transporters.
90
JONATHAN A. SHEPS, FANG ZHANG, and VICTOR LING A. Cellular Localization of the Hemolysin Transporter
Early work centered on the identification of polypeptides produced from the cloned hemolysin determinants. Expression of plasmid-encoded genes was carried out in mini- and maxicell systems (Mackman et al., 1985b). The products of the hlyB gene were identified in minicells and an in vitro translation system as proteins of 66 kDa and 46 kDa, whereas clyD generated a 53-kDa product. Transposon insertion mutagenesis verified the genetic identity of the observed polypeptides. Fractionation of minicells localized HlyB and HlyD to the inner membrane (Wang et al., 1991), but some of each protein was seen in outer membrane fractions. It is not clear, however, whether this represents a tight association between HlyD and TolC or just inadequate separation of the two membrane fractions. The low level of expression of HlyB prompted the search for an antibody that would permit the visualization of HlyB in vegetative cells. Linker insertion mutagenesis demonstrated that only the extreme COOH-terminal end of HlyB was able to tolerate insertion of a short peptide encoding the epitope of the anti-P-glycoprotein (Pgp) monoclonal antibody C494 (Juranka et al. 1992). This allowed cell fractionation studies to be carried out without the extreme overexpression of mini- or maxicells. By using this monoclonal antibody it was determined that the membrane-bound pool of HlyB was localized entirely to the inner membrane ofE. coli. It is interesting that a substantial pool of cytoplasmic HlyB was also observed, the functional significance of which is unknown. Similar membrane localization has been observed for transporters homologous to HlyB,D. The PrtD and PrtE proteins of Erwinia chrysathemi (analogs of HlyB and HlyD, respectively) have been shown to fractionate with inner membrane proteins in sucrose gradients (Delepelaire and Wandersman, 1991). Under the same conditions prtF (a TolC analog) fractionates with outer membrane fractions. Studies in which hlyB was expressed in minicell or in vitro systems consistently found a 46 kDa protein derived from the hlyB gene (Hartlein et al., 1983; Felmlee et al., 1985b; Mackman et al., 1986). Analysis of transposon insertion mutants in hlyB showed that this protein included the C-terminal of HlyB, and an internal methionine (Met286) in HlyB at an appropriate position was postulated to act as a start codon for this protein (Felmlee et al., 1985a). The 46-kDa protein was not detected in Western blots of cells expressing epitope-tagged HlyB, however (Juranka et al., 1992). A mutant HlyB in which the putative initiator met286 was replaced by valine was still functional, although expression was reduced somewhat, with a concomitant reduction in hemolytic activity. Mutations at Met286 of HlyB (Gentschev and Goebel, 1992), when analyzed in minicells, showed that the 46-kDa band was skill present, indicating that even when it exists it is not initiated from this methionine and likely represents an aberrantly migrating degradation product specific to minicells.
Hemolysin Transport
91
B. Transmembrane Topology of HlyB
Another feature of HlyB predicted from the primary sequence is that Uke other ABC transporters it has a number of transmembrane helices in a large domain in the NH2-terminal two-thirds of the protein. Computer modeling by the method of Eisenberg et al. (1984) predicts a series of six transmembrane helices for HlyB, and in this respect it resembles similar predictions for the homologous halves of P-glycoprotein (Pgp) and a number of other ABC transporters (Gerlach et al., 1986; Wang et al., 1991; Gentschev & Goebel, 1992). Wang et al. (1991) used P-lactamase fusions with HlyB as a topology probe and discovered that there were two transmembrane segments in the extreme NH2 terminus not predicted by any computer program. Gentschev and Goebel (1992) performed a topology study using fusions of HlyB to both P-galactosidase and to alkaline phosphatase, the first being a probe for cytoplasmic localization, and the latter for periplasmic localization. These results also demonstrated the presence of an additional two transmembrane domains within HlyB. The P-lactamase model also predicts substantially larger periplasmic domains than the original computer models (Gerlach et al., 1986). However, it is known (Broome-Smith and Spratt, 1986) that P-lactamase fusions have a tendency to overpredict periplasmic domains. The results of Gentschev and Goebel (1992) fit well with the predictions made using three computer programs. The programs are also in agreement with von Heijne's (1986) "inside positive" rule concerning the arrangement of charges surrounding transmembrane segments in polytopic membrane proteins (even though this rule is not part of the algorithm of any of the programs). Our Figure 4 shows Gentschev and Goebel's (1992) topology scheme, except in the location of the first transmembrane segment, where the fusion proteins of Wang et al. (1991) strongly contradict them. Our eighth transmembrane (Fig. 4) is placed two amino acids further toward the NH2 terminus than in Gentschev and Goebel's (1992) model (their figure 3). However, this is in accordance with the predictions shown in their own figure 4. The function of the NH2-terminal two transmembrane segments of HlyB is uncertain. At least the first 25 amino acids of HlyB can be replaced with no loss of function (Blight and Holland, 1990). However, the insertion of a short sequence at the extreme NH2 terminus of HlyB (Juranka et al., 1992) resulted in a loss of function. It has been proposed that this region might, be "flexible" in terms of its transmembrane topology and that this flexibility might be involved in transport of HlyA across the cytoplasmic membrane. This would imply a mechanism quite distinct from that employed by other ABC transporters that lack this NH2-terminal extension, although it is interesting that a similar hydrophobic extension at the NH2 terminus is seen in the TAP genes, which are peptide transporters in the endoplasmic reticulum (ER) of mammals. TAP and HlyB may also share the property of being part of larger macromolecular complexes with transmembrane proteins—presumably HlyD in the case of HlyB and MHC class I glycoproteins in the case of TAP (Suhetal, 1994).
JONATHAN A. SHEPS, FANG ZHANG, and VICTOR LING
92
f\
27 I
\ 67
178 ^^^
|_| 16/ Cy
|_|
191
289
295
I I
408
II
410
Periplasm
inmiMwi
177 L158F \
A146V
Cytoplasm
Figure 4. Topology of the HlyB molecule. The transmembrane helices are depicted as rectangles. Numbers marked are those of the first amino acid outside of the inner membrane on either side. Walker A and B motifs for ATP binding are the ovals. Point mutations that complement the transport-deficient HlyA signal mutants are marked. References: Wang et al., 1991; Gentschev and Goebel, 1992; Zhang et al., 1993b; Sheps, unpublished observations.
C. Energetics of Hemolysin Transport: Proton Motive Force and ATP
Being an ABC transporter, HlyB is possessed of a conserved putative ATP-binding domain, and it was expected that ATP would provide the energy for transport. Nonconservative mutations within the ATP-binding domain cause severe reductions in secreted hemolytic activity (Koronakis et al., 1988b). By using inhibitors specific for both membrane potential (valinomycin/KCl) and pH gradient (nigericin) and by using an imposed pH gradient to compensate for a discharged membrane potential, Koronakis et al. (1991) showed that proton motive force (AP) is required for transport. This would seem to make HlyA transport resemble NH2-terminal signal-directed transport, which requires both membrane potential and ATP (Arkowitz et al., 1993). They also used pulse labeling to define the energy requirements of intracellular HlyA, which had been synthesized prior to addition of the inhibitors. Unlike HlyA synthesized in the presence of the membrane poisons, this population of HlyA did not require proton motive force for secretion. Protease
Hemolysin Transport
93
treatment of cells after labeling showed that the AP-independent population was not accessible to protease, either on the exterior of the cells or within the periplasm, and it appears that this population of HlyAis membrane associated. It was suggested that, as in the NH2-terminal signal pathway, AP might be required for the association of the HlyA signal with HlyB (Koronakis et al., 1991), with transport of the remainder of HlyA being driven by ATP. A GST-HlyB fusion protein was constructed (Koronakis et al., 1993) in which the ATPase, domain of HlyB was attached to the COOH-terminus of glutathiones-transferase (GST). This construct was shown to be capable of both ATP binding and hydrolysis. The kinetic parameters are comparable to those reported for purified bacterial permeases such as MalK and HisP and their eukaryotic homolog P-glycoprotein, with AT^^ ranges of 0.07-0.94 mM and a V^^^ of about 1 jumol min~^ mg~^ (Ames et al., 1989; Morbach et al., 1993; Shapiro and Ling, 1994). Recently the PrtD gene, the HlyB-like component of the E. chrysanthemi metalloprotease secretion system, was partially purified. It displays an ATPase activity with a K^ of 12 jiM and a V^^^of 1.5 fimol ATPhydrolyzedmin"' mg~' (Delepelaire, 1994). The PrtD ATPase is strongly inhibited by peptides corresponding to the COOHterminal signals from E. chrysanthemi metalloproteases, but not by a signal sequence from another ABC transporter substrate (HasA) that PrD,E,F cannot secrete (Delepelaire, 1994). These results were interpreted to mean that signal peptide binding by PrtD strongly couples ATP hydrolysis to transport, and that in the absence of transport (because of the purification of PrtD in the absence of PrtE) this coupling prevents ATP hydrolysis. However, the GST-HlyB ATPase is not inhibited by HlyA signal peptide (Koronakis et al., 1993). The GST-HlyB fusion protein lacks the transmembrane domains of HlyB, implying that the transmembrane domains are required for the HlyA signal sequence to interact with HlyB. Hydrolysis of ATP by P-glycoprotein is also coupled to drug transport but with effects opposite of those seen with PrtD, an increase in ATPase activity being observed in the presence of some substrate drugs as well as inhibitors of drug transport (Shapiro and Ling, 1994, and references therein). This stimulation by substrate is much greater when Pgp is reconstituted into lipid vesicles rather than solubilized. Interestingly, HasA protein inhibits secretion of co-expressed PrtB and PrtC (normal substrates of PrtD) (Letoffe et al., 1994a) but is unable to inhibit ATP hydrolysis by PrtD (Delepelaire, 1994). This suggests that there are two distinguishable steps in RTX toxin transport by PrtD: binding of the COOH-terminal signal to the transporter, and coupling of ATP hydrolysis to transport of bound substrate. HasA appears to be competent for the first step and not the second, and so might prove a useful reagent for future biochemical investigation of this system, with significant implications for the mechanism of hemolysin transport.
94
JONATHAN A. SHEPS, FANG ZHANG, and VICTOR LING D. HlyD: Structure and Function
HlyD is a component of the hemolysin transporter system that is little studied. Its topology has been probed using fusions to P-lactamase (Wang et al., 1991), P-galactosidase, and alkaline phosphatase (Schiilein et al., 1992). The results from fusion proteins as well as computer predictions of membrane-spanning regions are consistent with a model in which the NH2-terminal 60 amino acids of HlyD are located in the cytoplasm, with the bulk of the protein being in the periplasm. Mutagenesis of HlyD indicates that residues near the COOH terminus are required for hemolysin transport and release from cells (Schiilein et al., 1994). Among the ABC transporters only the RTX transporters are associated with HlyD-like n^olecules. No eukaryotic homologs of HlyD are known. However, among Gram-negative bacteria there are HlyD-like genes associated with non-ABC transporters, in systems where secretion occurs across both inner and outer membranes. These proteins may function as a bridge between inner membrane translocators and outer membrane porins. Since these transport complexes are presumed to exist at (or be responsible for) sites of fusion between inner and outer membranes the family of HlyD-like proteins has recently been dubbed the membrane fusion protein (MFP) family. In most cases of ABC-MFP collaboration the involvement of a TolC analog has not been demonstrated but is commonly assumed (Lewis, 1994). There are some examples of ABC transporters involved in protein secretion from Gram-positive bacteria in which HlyD-like proteins are present in the same operon, even though there is no periplasm to traverse or outer membrane component to connect with (Chung et al., 1992; Stoddard et al., 1992). It may well be that HlyD acts more as a periplasmic chaperone rather than a transport channel, and that in Gram-positive cells such a function may still be needed to get secreted protein safely beyond the cell wall and capsular polysaccharides. E. TolC: The Universal Component
The role of TolC in hemolysin transport was discovered by comparison of cloned RTX toxin secretion determinants across species (Wandersman and Delepelaire, 1990). The observation was made that encoded in the metalloprotease and cyclolysin secretion determinants, in addition to HlyB,D-like molecules, was a third protein involved in secretion. In searching for the E. coli counterpart for this third protein they found sequence similarity to the outer membrane protein TolC. Expression of the hemolysin operon in tolC mutant cells showed that in the absence of TolC protein no hemolysin was secreted. It has been widely accepted from this time that TolC forms the outer membrane component of a protein-conducting channel that leads HlyA from the cytoplasm, via HlyB and HlyD, across the inner membrane; keeps it sequestered from the periplasm; and finally leads to the external medium.
Hemolysin Transport
95
TolC itself was previously known as a minor outer membrane protein, mutations in which have pleiotropic effects on the expression of other porins, DNA supercoiling, and cell division (Dorman et al., 1989; Hiraga et al., 1989). The best evidence of its direct involvement in transport comes from work by Benz et al. (1993) in which it is shown that TolC can form an ion conducting channel in an artificial lipid membrane and that this channel can be blocked by peptides. Thus, TolC could be part of a channel for conducting HlyA across the outer membrane. Intriguingly it has been suggested that there is a region of sequence identity shared by TolC and HlyD, which may represent a common motif for interaction with substrates moving through a channel formed by these two proteins (Schiilein et al., 1994). F. The Transport Complex
In Figure 2 we show HlyB as a dimer by analogy to other ABC transporters that are known to be dimers (in the case of at least some bacterial permeases) or whose monomeric structure is that of a pair of HlyB-like domains (as in the case of P-glycoprotein). There is a small amount of direct biochemical evidence that the three proteins required for HlyA secretion associate with each other in the cell envelope, and for what the stoichiometry of transporter components might be. The cell fractionation experiments of Delepelaire and Wandersman (1991) showed that PrtF (the E. chysanthemi TolC homolog) is not strongly associated with PrtD/E but migrates with the outer membrane in sucrose gradients. However, competition between mutant and wild-type HlyD molecules indicates that HlyD interacts with at least one other molecule as part of a functional transport complex (be it another HlyD molecule or HlyB) (Schiilein et al., 1994). A couple of laboratories have observed bands in gels that they interpret as HlyB or PrtD dimers (Thomas et al., 1992; Delepelaire, 1994), whose existence appears to depend on electrophoresis conditions but could well reflect physiological reality. It has often been speculated that the HlyB,D, TolC complex could exist at "Bayer bridges," which connect inner and outer membranes (Holland et al., 1990; Hirst and Welch, 1988). As originally proposed by Bayer (1968), these structures are large (up to 35 nm) lipid tubules that might be permanent features of the Gramnegative envelope. Recently this picture has been challenged by cryofixation methods in electron microscopy (Kellenberger, 1990), which fmd no evidence for permanent lipid connections between the inner and outer membranes of £. coli. The new evidence cannot rule out small proteinaceous connections between the membranes, and Kellenberger (1990) suggests that such connections might be the nuclei around which the classic Bayer bridges assemble as an artifact of plasmolysis and fixation methods. It is, therefore possible that a trans-periplasmic complex of HlyB,D,TolC could exist without being part of any special membrane structure.
96
JONATHAN A. SHEPS, FANG ZHANG, and VICTOR LING
IV. THE HEMOLYSIN A SIGNAL SEQUENCE Many studies have been done to define the features of the GOGH terminus of HlyA that are responsible for its ability to be transported from the cell. The secretion signal of HlyA has been convincingly mapped to the COOH-terminal 50 amino acids (Koronakis et al., 1989) (Fig. 5), and this region of HlyA alone is sufficient to target fusion proteins for secretion via HlyB,D (Hess et al., 1990). However, the COOH-terminal signal of HlyA has proved to be complex. Multiple rounds of mutagenesis, structural modeling, and comparison to similar signal sequences from a range of organisms have failed to generate a consensus motif but have led to a model in which multiple features may explain the activity of the hemolysin signal (Fig, 6) (Stanley etal., 1991; Kenny etal., 1992; Zhang etal., 1993a). In this section we focus on the proposed models of the HlyA COOH-terminal signal and briefly compare these to what is known of the transport signals from other RTX toxins. A. Localization of the Transport Signal to the C O O H Terminus of HlyA
Hemolysin A does not have a classic NH2-terminal signal, and the NH2 terminus of the protein is not cleaved upon secretion (Felmlee et al., 1985a). Unlike export of protein to the periplasm via the general secretory pathway, transport of hemolysin to the medium is independent of the sec A and secY gene products (Gray et al., 1989; Gentschev et al., 1990) and no periplasmic intermediate is observed (Mackman et al., 1986). A deletion construct consisting of only the COOH-terminal 23 kDa of HlyA was still secreted from HlyB,D-expressing cells with high efficiency (Nicaud et al., 1986). This demonstrated that no NH2-terminal signal was needed for HlyA
1
■J
2
'
3
4
' Hydrophobic formation
5
'
6
'
7
' ^.g ^„ ®"*
8
9
HIHIIH^ C a - Binding Site
10
++ S^jretjon Signal
LXGGXGNDX Repeats Figure 5. The functional domains of the HlyA molecule. The hemolysin A protein is represented at a scale of 100 amino acids per demarcation. The thick vertical bars represent the RTX repeats. The other functional domains are as indicated. References: Koronakis et al., 1989; Hess et al., 1990; Pellett et al., 1990; Ludwig et al., 1991; Menestrina et al., 1994.
Hemolysin
97
Transport
Mackmanetal.,1987 ^ Hess etal.,1990 Predictd Helicas
Koronakis et at.. 1989 Stanley etal. 1991 Zhang etal.. 1993a
Measured Helices
Yinet al.. 1995
-60 LE2001 pHly152 pSF4000
-S
-50
^
-40 -40
Charged ■ cluster | B Aspartic acid box 1 ■
..1-30 0»J,-3O
100
8. «« 90
1P
1
70
AN
8 ^ so
1 50 \
40
8
30
1
B
-10
^1
1
lA D K A Q 1 I N
1
1
S
F 1
M
S
T^
Q
c 1 IIAA L A
K
1
1 K
1
y
1
K
1II
B| l v Q I T | K PDS
Jlu^11tr
Q I
80
"5 /
^
Hydroxyl cluster ciusier
STYGSQDNLNPLIWBISKIISAAGNgtD^n^BRt3AASLLQLSGNASDPSYGRN!5ITLTASA| LA GD IT 1 1 ^1 AT | LA G £1 A
§ S »
-20
1 D| I
i I>
B
w
1 Q
1
1
R
L
20 10
Figure 6. The HlyA signal sequences. The COOH-terminal 60 amino acids of HlyA (strain LE2001) are presented. The polymorphisms of the COOH-terminal of HlyA proteins from two other £ coli isolates are listed below. The predicted helices and measured helices are listed above the sequence. The open rectangles and filled rectangles are the predicted turns and predicted p-strands, respectively. Point mutations and their effects on transport are listed. In each percentage level, the upper line contains the results of Kenny et al. (1992) and the bottom line is from Stanley et al. (1991). Dots above amino acids in the HlyA sequence represent the seven "critical" amino acids discussed by Kenny et al. (1994).
secretion and that the transport signal was contained within a small COOH-terminal fragment. Koronakis et al. (1989) constructed internal deletions in HlyA which demonstrated that the COOH-terminal 53 amino acids were sufficient for fiill transport activity. Hess et al. (1990) fused the last 60 amino acids of HlyA to an alkaline phosphatase reporter gene and showed that this was sufficient, in the absence of any HlyA-derived upstream sequence, to give full transport function. Deletions internal to the COOH-terminal 60 amino acids of HlyA (Mackman et al., 1987;
98
JONATHAN A. SHEPS, FANG ZHANG, and VICTOR LING
Koronakis et al., 1989; Hess et al., 1990; Zhang et al., 1993a) or COOH-terminal truncations (Gray et al., 1986; Koronkis et al., 1989) resulted in substantial transport defects. Together these results indicate the presence of a signal contained within the COOH-terminal 53 amino acids. B. Cross-Species Comparison of Signal Sequences
Generally, transport of RTX toxins is mediated by their extreme COOH termini. Zhang et al. (1993a) showed that the COOH-terminal 70 amino acids of the Pastevrella haemolytica leukotoxin (LktA) could replace the COOH-terminal 58 amino acids of HlyA with equal transport activity. Direct assay of the activity of this signal in the leukotoxin transport system is difficult since expression in E. coli has been reported to be poor (Strathdee and Lo, 1989). The cyclolysin transporter of ^. pertussis (CyaB,D,E) is also poorly expressed in E. coli, but Sebo and Ladant (1993) have shown that the COOH-terminal 74 amino acids of cyclolysin (CyaA) function as a signal sequence that is recognized by the hemolysin transporter. The metalloprotease transport apparatus ofE. chrysanthemi (PrtD,E,F), when expressed in E. coli, recognizes the COOH-terminal 56 amino acids of PrtG (Ghigo and Wandersman, 1992) and the last 39 amino acids of PrtB (two of the four native substrates of PrtD,E,F) (Delepelaire and Wandersman, 1990) and the last 80 amino acids of PrtSM from Serratia marcescens (Letoffe et al., 1991). Alignment of the several COOH-terminal signals that are recognized by HlyB,D or related transporters reveals that the primary sequence is not well conserved. Transporters from different systems may of course differ in their substrate recognition requirements and so there is no need for strict conservation among signal sequences. Cross-complementation experiments (Table 2) reveal that recognition of heterologous substrates is often inefficient relative to transport by native systems, but it is often much better than would be expected on the basis of primary sequence conservation alone, particularly in the case of HlyA and LktA, where transport efficiency is close to 100% and sequence conservation is minimal (Zhang et al., 1993a). One exception to the rule of COOH-terminal signals in RTX toxins is the colicin V secretion system ofE. coli (consisting of the genes CvaA, CvaB, and TolC), in which the signal sequence that can be used by HlyB,D is in the NH2-terminal 57 amino acids (Path et al., 1991). This is only a partial exception since colicin V (CvaC) is not an RTX toxin but a small bacteriocin with no structural similarity to RTX toxins. However, CvaA,B are homologous to HlyB,D (Fig. 3). The signal recognized by the CvaA,B transporter resides within the NH2-terminal 39 amino acids (Gilson et al., 1990), but there may be some differences in how the two transporters recognize colicin V since mutations in CvaC affect transport by the two systems differently (Path et al., 1991).
Hemolysin Transport
99
C. Features of the HlyA COOH-Terminal Signal.
The most extensively analyzed of the COOH-terminal signal sequences is that of E. coli hemolysin. It should be noted that three different E. coli hemolysin isolates, which differ by only a few amino acids (Fig. 6), have been studied intensively, and these are considered by us to be functionally identical. The hemolysin signal sequence of £". coli and its analogs in other RTX transport systems are unusually complex. We may eventually be able to discern a single structure or sequence motif recognizable as a unique functional domain, but it seems more likely that no simple explanation exists for the activity of the hemolysin signal. The hemolysin signal might be a collection of motifs, all of which can be recognized by some, if not all, RTX transporters. There is mutational evidence to indicate that the disruption of any one of the proposed motifs can still be tolerated with only incremental loss of function. The hemolysin signal is highly resistant to point mutations, with only a handful ever being known to decrease transport by more than 50% (Kenny et al., 1992; Stanley et al., 1991) (Fig. 6). It seems as if the transporter recognizes each component of the transport signal in an additive fashion, with each intact structure contributing to the overall efficiency of the process. Each of the RTX signals may have only some of the total set of motifs that can be recognized by one transporter or another (Table 3). A transporter may have the capacity to recognize motifs not present in its natural substrate, as in the case of PrtSM secretion by HasD (Tables 2 and 3). The ability of HlyB to recognize the LktA signal with high efficiency despite the many differences between the sequences points to the conclusion that HlyB has a certain "reserve" capacity to recognize motifs that are novel to it, to compensate for those that LktA is missing (such as the acidic repeats; see below). The question remains of how the recognition elements are arranged within the transporter molecule, and what geometric constraints this places on the arrangement of signal motifs in the substrates. The following subsections will deal with individual features that have been shown to be important in transport function (Fig. 6). The extent to which these features are conserved among transport signals of other RTX toxins will be discussed (Table 3). Helices
Helical regions can be predicted within most of the COOH-terminal signals, and the helical conformation itself may be a distinct transport motif independent of its amino acid composition. Mackman et al. (1987) showed that the ability of the COOH-terminal 27 amino acids of HlyA to promote transport, when fused by internal deletion to upstream HlyA sequences, depended on the presence of helical structure upstream. They pointed out that the sequences normally found upstream of the COOH-terminal 27 amino acids were predicted to contain a strong a-helix
100
JONATHAN A. SHEPS, FANG ZHANG, and VICTOR LING Table 3. Features of Bacterial Protein Toxin Signal Sequences
Transport signal
Predicted helices^
Measured helices
Aspartic box
Acidic repeats
Hydroxylated Hyd 'rophobic tail^ taif
HlyA
+
7^
+
+
+
LktA
+
+^
+
-
CyaA
+
nt
+
+
Prt (A,B,C,SM,G)
+
+'
+
-
^'^ —
AprA
+
nt
+
HasA
-
nt
+
NodO
+
nt
+
ColV'
-
nt
+
-
-
+ + + + +•
Notes: ^Stanley et al. (1991); Kenny et al. (1992); Zhang et al. (1993a); and predicted in this study using the secondary prediction program by Hamodrakas (1988). ''Zhang etal. (1995). 'Wolffetal.(1994). ''Kenny etal. (1992). *Sebo and Ladant (1993). ^Koronakisetal. (1989). ^ h i g o and Wandersman (1994). ''The hydroxylated cluster in LktA is not directly at the COOH terminus. 'The transport signal sequence of Col V is at the NH2 terminus of the molecule, so the significance of this motif is unclear, nt not tested. The presence of a motif is indicated by + and its absence by-. Absences are either recorded in the references given or represent our own observations.
(residues —35 to —23). Subsequently this helical region was shown to be needed for full transport activity (Koronakis et al, 1989; Hess et al., 1990). Koronakis et al. (1989) pointed out a possible amphiphilic helix even further upstream corresponding to residues—52 to—35 from the COOH terminus. A deletion of most of this region (from-52 to—39) caused a major defect in transport activity. Stanley et al. (1991) proposed that if the residues between the two helices were assumed to be helical then the entire region could be thought of as a single amphiphilic helix from ^ 8 to —23. This extended helix is characterized by a separation of charges, with positively charged residues exposed on the hydrophilic face and a patch of negatively charged residues interrupting the hydrophobic face of the helix. Mutations introduced by Stanley et al. (1991) reinforced this picture in that decreases in amphiphilicity or helix propensity were correlated with greater loss of function (but see Kenny et al., 1992). This model was then generalized to a number of other RTX toxin secretion signals. Except for the metalloprotease signals, all of the non-jE". coli hemolysins, leukotoxins, and cyclolysin share some detectable primary sequence similarity or could form amphiphilic helices with
Hemolysin Transport
101
negative charges interrupting the nonpolar face of the helix (Stanley et al., 1991; see their figure 5). Another theory has sprung mainly from a comparison of the predicted secondary structures of the HlyA and LktA signal sequences (Zhang et al., 1993a). Both contain "helix-tum-helix, strand-loop-strand" motifs despite the large difference in sequence, and the LktA sequence can fully replace that of HlyA in terms of recognition by the HlyB,D transporter. Zhang et al. (1993a) presented mutational evidence, mostly deletions and their partial functional replacements with random sequences, of a correlation between the presence of predicted structure and function. Similar structural predictions can be made for the cyclolysin and metalloprotease B signal sequences. To test whether the predicted structures actually exist in the HlyA and LktA signals, purified peptides were prepared to correspond to the signal sequences of the two toxins and subjected to analysis by circular dichroism (CD) spectroscopy (Zhang et al., 1995) and two-dimensional nuclear magnetic resonance (NMR) spectroscopy (Yin et al., 1995). CD analysis showed that the signal peptides were in an unstructured conformation in aqueous solution but that some helical conformation was induced in membrane-mimetic environments, but only if negatively charged lipids were used. Two-dimensional NMR in deuterated SDS micelles provided evidence of a-helical regions in both signal sequences (Fig. 6). A similar study was performed on the signal peptide of metalloprotease G from E. chysanthemi (Wolff et al., 1994) with similar results. In trifluoroethanol (TFE) solutions the NMR structure of the PrtG signal contains a pair of hydrophobic a-helices followed by an unstructured COOH terminus. Charged Cluster The charged cluster lying between the two observed helices has attracted some attention. Koronakis et al. (1989) deleted a small region that included it (—37 to -27) and destroyed almost all transport activity. However, a deletion of only the charged residues -34 to -29 (DVKEER) still retained 70% of wild-type activity, whereas deletion of residues—35 to—39 (a potential turn-forming region) caused a much larger defect (Zhang et al. 1993a). The charged cluster is predicted to place its negative charges in the middle of the hydrophobic face of the extended amphiphilic helix while keeping its positive charges on the polar face. The positive charges on the polar face of the amphiphilic helix might interact with the phospholipid head groups consonant with helix formation being dependent on negativity charged lipid as indicated by the CD data above (Zhang et al., 1995). Point mutations of the charged residues have had ambiguous results (Stanley et al., 1991; Kenny et al., 1992; Zhang et al., 1993b). Many of the observed effects can be explained in terms of other hypothesized functional motifs (see below), and it appears that a charged cluster jper se is not a significant feature of the HlyA signal. A charged cluster is not apparent in LktA or other COOH-terminal signals.
102
JONATHAN A. SHEPS, FANG ZHANG, and VICTOR LING
Hydroxylated Tail and Other Tails
The extreme COOH-terminal tail of the hemolysin signal is uncharged and rich in hydroxylated amino acids. This region has been predicted to form an amphiphilic P-sheet (Stanley et al, 1991). Deletions of the COOH-terminal eight amino acids can reduce secretion by 80% (Stanley et al., 1991), and removing the last 27 amino acids can cause an almost complete loss of transport activity (Stanley et al., 1991; Zhang et al., 1993a). Replacing the deleted COOH-terminal tail with random sequences predicted to have a P-sheet conformation restores some transport activity (Zhang etal., 1993a). In contrast, the COOH-terminal five amino acids of PrtG contain an acidic residue followed by four hydrophobic residues (Ghigo and Wandersman, 1994). Progressive deletions from the COOH terminus of PrtG caused defects in transport until a similar internal hydrophobic motif was fortuitously exposed, at which point much of the original transport activity was restored. A similar motif exists at the COOH terminus of all metalloproteases that can be secreted by the PrtD,E,F system and in the NodO gene of Rhizobium leguminosarum. No such motif is present in the pore-forming RTX toxins, however, and in general these cannot be secreted by PrtD,E,F (Table 2). This motif must be exposed at the COOH-terminal end of the protein since addition of even two amino acids to this COOH-terminal motif abrogates transport (Ghigo and Wandersman, 1994). NMR measurements show that both the hydroxylated COOH termini of HlyA and LktA as well as the hydrophobic COOH terminus of PrtG are unstructured under conditions where helix formation is possible (Wolff et al., 1994; Yin et al., 1995). The Aspartate Box
A motif dubbed the "aspartate box" has been proposed to be a well-conserved feature of proteins that are translocated across membranes (Gray et al., 1986). The aspartate box is defined as a region of 12 to 16 small, uncharged, or hydrophobic residues, flanked by charged residues. The HlyA aspartate box (located at positions -29 to —15) occupies the space between the charged cluster and the hydroxylated tail. This corresponds to the second half of the extended amphiphilic helix and contains the second helix measured by NMR. The secondary structure motif predicted to be associated with the aspartate box is a rapid transition from helix to sheet to turn with short overlaps (Fig. 6). The aspartate box motif has been observed in all RTX toxins. A similar feature is present in the COOH-terminal extensions possessed by the glycosomal isoforms of phosphoglycerate kinase in trypanosome parasites (Gray et al.,1986). These glycosomal proteins are imported post-translationally into glycosomes (trypanasome homologs of peroxisomes) in a process that depends on the their unique COOH-terminal sequences (Swinkels et al., 1988). The only candidate protein importers in peroxisomes that have been cloned so far are a pair of ABC transporters
Hemolysin Transport
103
(PMP70 and ALDP) (Kamijo et al., 1990; Mosser et al., 1993). An aspartate box-like feature is also present in the NH2-temiinal peroxisomal targeting signal of mammalian ketoacyl CoA-thiolase (Swinkels et al., 1991; J. Sheps, unpublished observation). The aspartate box is even more resistant to point mutational damage than most other regions of the HlyA transport signal (Stanley et al., 1991). Kenny et al. (1992) showed that the charges at the NH2-terminal end of the aspartate box could be either negative or positive, but that the COOH terminus is required to be negatively charged. The aspartate box has been deleted only in the context of COOH-terminal truncations (Stanley et al., 1991) that are already severely defective. Nevertheless, what activity exists when only the aspartate box remains (approximately 8%, even with the COOH-terminal negative charge removed) is neatly eliminated with its removal. It may be concluded that an uncharged region of a certain length is required for transport by HlyB,D. Critical Amino Acids
Kenny et al. (1992) generated an extensive set of point mutations in the transport signal of HlyA. Many of these were not consistent with the previously proposed theories concerning the role of the amphiphilic helix in that they were predicted to reduce the extent of amphiphilicity or disrupt the predicted helices. Single point mutants in the transport signal typically did not have large effects. However, they identified a number of single residues at which mutations produced unusually large effects (see Fig. 6). A combination of mutations at the "critical" residues —46, —35, and -15 reduces transport to 0.6% (Kenny et al., 1994). Kenny et al. (1992) postulate that signal recognition is dependent on this set of dispersed residues, which form critical contacts with the hemolysin transporter. Some generality is emerging from this model, in as much as there are hints that some of the "critical" residues may form part of a larger pattern of conservation that extends throughout the RTX toxin family (Blight et al., 1994a). The two "critical" acidic residues are part of the "acidic repeat" motif (see below) (Sebo and Ladant, 1993). Between the two acidic residues is a "critical" aromatic residue. The presence of a similar group of three amino acids, with similar spacing, can be seen in the signals of both the pore-forming toxins and the metalloproteases (Blight et al., 1994a). There is a disagreement between different laboratories on the effects of identical mutations at one position, however: the Phe to Leu mutation at—35 causes either a loss of 70% of transport activity (Kenny et al., 1992) or only 20% in our hands (Zhang etal., 1993b), Recently a nontransportable HlyA fusion protein was constructed that consisted of the COOH-terminal 81 kDa of HlyA and most of p-galactosidase (Kenny et al., 1994). This construct inhibits transport of co-expressed wild-type HlyA. Introduction of the "triple critical" mutation into the fusion protein resulted in a loss of the ability to inhibit HlyA secretion.
104
JONATHAN A. SHEPS, FANG ZHANG, and VICTOR LING
Acidic Repeats
The Ca^"*"-binding repeats typical of RTX toxins lie in a compact array NH2-terminal of the signal sequence (Fig. 5). Neither the RTX repeats nor the sequences that separate them from the signal sequence are necessary for transport of HlyA (Koronakis et al., 1989). When the COOH-terminal signal sequence of cyclolysin (CyaA) is deleted certain sequences NH2-terminal of the signal sequence (but still COOH-terminal to the repeat sequence array) are capable of being recognized by the hemolysin transporter (Sebo and Ladant, 1993). These were identified as consisting of degenerate forms of the typical RTX repeat motifs, which retained little except a regular spacing of acidic residues. This "acidic repeat" motif extended into the signal region as well, where it may contribute to normal recognition processes. Acidic repeats are also present in HlyA. Two of the residues identified as part of the "acidic repeat" are also "critical" residues, according to Kenny et al. (1994). The LktA signal, however, lacks this motif Two HlyB point mutants have been isolated that have specific defects in transport of the LktA but not the HlyA signals. These two HlyB mutants (D259N and D433N) (see Fig. 4) both represent the loss of negative charge at positions that influence signal recognition by HlyB. Differences in the distribution of charged residues in the LktA signal may be the cause of its differential sensitivity to the loss of charged residues in HlyB.
V. THE MECHANISM OF TRANSPORT In the hemolysin transport system issues exist of whether the signal is recognized from within the cytoplasmic membrane, whether membrane association is a necessary step in the secretion process and whether there is a relationship between membrane association motifs and signals recognized by the transport proteins. A. Hemolysin A-Hemolysin B interactions
Among some ABC transporters it has been observed that mutants that affect substrate specificity are found within the multiple transmembrane spanning domains. Mutations in the mammalian multidrug transporter P-glycoprotein that affect substrate specificity have been found on cytoplasmic loops near the third (Safa et al., 1990; Loo and Clarke, 1994), second, fifth, and ninth transmembrane segments (Loo and Clarke, 1994), as well as within the sixth (Devine et al., 1992; Loo and Clarke, 1993b), fourth, tenth, twelfth (Loo and Clarke, 1993a, b), and eleventh (Gros et al., 1991) transmembrane segments. In the cystic fibrosis-associated protein CFTR, mutations that alter anion selectivity were generally located within the transmembrane domains, and these were used to show that this protein was in fact a chloride channel (Anderson et al., 1992). These findings suggest that the multiple membrane spans of ABC transporters act to form a substrate-conducting channel. In the TAP system there are numerous polymorphisms that have been
Hemolysin Transport
105
shown to influence peptide selection by the class IMHC molecules that are fed by the TAP transporters (Powis et al., 1992). These polymorphisms are principally located within the NH2-terminal half of the molecule, within the multiple-membrane spanning domain, although not all are within the cytoplasmic or membraneassociated domains. It is not yet clear what subset of the observed polymorphisms are responsible for the differences in substrate specificity. Within the hemolysin system mutations have been found that can complement mutations in the COOH-terminal signal of HlyA (Zhang et al., 1993a; and our unpublished results; see Fig. 4). Bacteria having deletions within the COOH-terminal signal of HlyA are severely defective in transport and result in very small hemolytic zones on blood agar. This phenotype was complemented by cloning a library of chemically mutagenized HlyB-bearing plasmids into E. coli carrying the mutant HlyA. Colonies with larger hemolytic zones were selected, and the mutant HlyB genes were sequenced. In all cases we found single point mutations in HlyB to be responsible for the associated phenotypes. Of 12 mutations isolated, 11 were found to map within the multiple transmembrane domain of HlyB. Only one mutation was found in the ATP-binding cassette domain. No mutations were found within the region of the first two predicted transmembrane segments of HlyB. This NH2-terminal region is unique to HlyB-like molecules and is not analogous to any part of the basic permease structure (Doige and Ames, 1993). It therefore seems likely that this region is not involved in substrate recognition. Perhaps the extra two transmembrane segments form a larger pore, are flexible regions involved in transport, or may be needed for association with other components of the transmembrane complex. Complementation is a classic technique for discovering genes that lie along the same biochemical pathway. When the complementation between two proteins is allele-specific it is more parsimonious to interpret this as evidence of a direct physical association between the two proteins than to postulate third factors with which both mutated proteins might interact. The revertant alleles of HlyB we have isolated can partially restore transport efficiency to mutant alleles of HlyA. However, these HlyB mutations do not increase the efficiency of wild-type HlyA transport, nor do they revert all HlyA signal sequence mutations. In some cases HlyB mutations increase transport of some HlyA signal mutants, while decreasing transport of others or of a HlyA fusion that uses the LktA signal sequence (Zhang et al., 1993b; and our unpublished results). By the above criteria HlyB is almost certain to interact directly with the signal of HlyA, and these mutated sites in HlyB are likely to be involved in recognition of the HlyA signal, although not necessarily in direct contact themselves. It has been suggested by Doige and Ames (1993) that the region of the ATP-binding cassette between the two Walker motifs is involved with the transmembrane domains of the transporter. HisP, the ATP-binding component of the Salmonella typhimurium histidine permease, was shown to be accessible to proteases and biotinylating agents from the periplasmic side of the membrane (Baichwal et al.,
106
JONATHAN A. SHEPS, FANG ZHANG, and VICTOR LING
1993), but only in the presence of the transmembrane components HisQ and HisM. It was suggested that this region (called the "helical region" for its predicted secondary structure (Mimura et al., 1990) is involved in substrate binding or coupling energy released by ATP hydrolysis to the transmembrane segments in order to open and close the channel formed by them. Within the "helical" region of the ATP-binding region of HlyB a mutation affecting substrate specificity was found (Fig. 4). In three instances we found that mutations (D259N, D433N, and R212K; see Fig. 4) in HlyB that affected charged residues caused defects in secretion that were specific to HlyA signal mutants in which there are alterations in charged residues or in transport directed by the LktA signal (Zhang et al., 1993b), which is much more positively charged than HlyA (Strathdee and Lo, 1987). These results suggest a direct interaction between HlyA and HlyB in this region, involving at least these charged residues. It may be that electrostatic interactions are important in the recognition process by which HlyB selects HlyA for secretion. B. Hemolysin A-Membrane Interactions
The HlyA signal sequence, which has been shown to be sufficient for directing transport of fusion proteins through the HlyB,D,TolC transporter (Hess et al., 1990), can also target a nonsecretable fusion protein (such as (3-galactosidase) to the membrane (Gentschev and Goebel, 1992). This membrane association becomes salt resistant in the presence of HlyB (even in the absence of HlyD), which suggests a transmembrane intermediate held in place by interactions between the signal sequence and HlyB. Nevertheless, since the HlyA signal can associate with membranes by itself (Gentschev & Goebel, 1992), it is tempting to speculate whether this is the normal state in which the toxin is recognized by the transporter, and if so, what its conformation might be when membrane bound. Recent CD studies have shown that in a charged lipid environment the HlyA and LktA signals adopt a partially helical conformation, but that they are unstructured in aqueous solvents (Zhang et al., 1995). Similar results have been obtained for the metalloprotease G signal sequence of E. chrysanthemi in a membrane-mimetic triflouroethanol (TFE) solution (Wolff et al., 1994). The transport model of Stanley et al. (1991) involves the signal sequence first associating with the membrane and then entering into association with HlyB,D so as to form a membrane-spanning helix that parallels those of HlyB. The bulk of HlyA is then envisioned as sliding through the channel spanned by HlyB and the HlyA signal in a hairpin loop. The COOH-terminal signal would then be the last part of HlyA to be released from the transporter. Stanley et al. (1991) proposed that the increased hydrophobicity of the leukotoxin of Actinobacillus actinomycetemcomitans (AaltA) in the "extended amphiphilic" region might be responsible for the fact that this RTX toxin remains cell associated (Lally et al., 1989).
Hemolysin Transport
107
The translocation of the numerous charged residues within the signal sequence into the membrane bilayer implied by the model of Stanley et al. (1991) might require a mechanism for energy and charge balance (perhaps exploiting AP) that could be mechanistically distinct from the later steps in transport in which the polypeptide is nonspecifically moved through a transmembrane pore (driven by ATP hydrolysis by analogy to SecA (Arkowitz et al., 1993)). However, this model, as originally formulated, implied a restricted role for the "amphiphilic helix" region of the signal sequence in binding to the membrane, whereas the extreme COOH terminus of the signal (the predicted amphipathic p-sheet) would interact with the transporter. This model must take into account evidence that the NH2-terminal half of the signal interacts directly with HlyB, as demonstrated by the isolation of HlyB mutants that complement mutations in this part of the HlyA signal (Zhang et al., 1993b). Perhaps the amphiphilic helices within the HlyA signal bind first to the inner membrane of E. coli and then come to associate directly with HlyB along with the more COOH-terminal elements of the signal. An important question is whether HlyB,D and TolC form a continuous proteinaceous pore through which HlyA travels all the way from the cytoplasm to the extracellular medium. Functional expression of the hemolysin transporter confers on E. coli sensitivity to the antibiotic vancomycin (Wandersman and Letoffe, 1993). Since vancomycin is a fairly large hydrophilic molecule, normal resistance to it is thought to arise from its inability to penetrate through existing pores in the membrane. It has been proposed that vancomycin sensitivity results from the antibiotic being able to penetrate through the hemolysin export channel to the periplasm, where vancomycin acts to inhibit cell wall synthesis. Mutations in TolC or expression of HlyB,D results in a loss of hemolysin secretion and a restoration of vancomycin resistance (Wandersman and Letoffe, 1993). Mutations in the signal sequence of HlyA can also restore vancomycin resistance (Blight et al., 1994b), implying that the channel for vancomycin (presumably in TolC) is open only when TolC is coupled to an actively secreting HlyB,D. C, The Role of Ca^^-Binding Repeats in Transport
Mutational analysis has mapped the HlyA signal sequence to the COOH-terminal 50 amino acids (Koronakis et al., 1989; Hess et al., 1990). However, chimeras that fuse passenger proteins with COOH-terminal sequences of HlyA reveal a more complicated picture. Different passenger proteins require different lengths of HlyA COOH-terminal sequences (Hess et al., 1990; Kenny et al., 1991), and transport efficiency declines with larger passenger proteins. In the metalloprotease B transport system a 6-kDa peptide containing the COOH-terminal 39 amino acids can be secreted; however, longer sequences that include glycine-rich Ca^'^-binding repeats were required to drive secretion of a larger molecule (Letoffe and Wandersman, 1992).
108
JONATHAN A. SHEPS, FANG ZHANG, and VICTOR LING
The repeats are also required for Ca^"^ binding, stability of the secreted protein, and target cell binding and lysis by HlyA (Boehm et al., 1990a,b). The crystal structure of the alkaline protease (AprA) from/? aeruginosa reveals that each repeat unit forms part of a repeating array of p-strands and reverse turns in which Ca^"^ ions intercalate between the turns on both sides of the array (Baumann et al., 1993). The structure of the metalloprotease from S. marcescens (PrtSM) is essentially the same (Baumann, 1994). It has been suggested that the glycine-rich repeats remain unfolded in the calcium-poor cytoplasm (Baumann et al., 1993), and this may contribute to maintaining a transport-competent conformation. Folding in the extracellular medium would probably be irreversible. This function may not be required for secretion of small, normally secreted passenger proteins such as PhoA (Hess et al., 1990) but may explain the requirement for the repeat zone for transport of large, otherwise cytoplasmic, proteins such as P-galactosidase (Kenny et al., 1991), which cannot even be efficiently secreted to the periplasm by the classical signal pathway (Gentschev and Goebel, 1992). If the COOH-terminal region of HlyA is translocated out of the cytoplasm ahead of the rest of the toxin, Ca^"^-driven folding of the repeat domain could drive the irreversibility of the transport reaction. This might explain why the repeat domains are vital for secretion of large fusion proteins that might otherwise fold in the cytoplasm and pull back out of the transporter. This model might require that the GOGH terminus of HlyA leave the transport complex ahead of the NH2 terminus and so may be incompatible with the "GOGH terminus last" model of Stanley et al.(1991). The crystal structure of AprA reveals that Ga^"*" ions are bound not only by the classic RTX repeats but also at sites on AprA that depart substantially from the classic definition of RTX repeats, including one atom bound within the GOGH-terminal signal sequence. If calcium binding follows as loose a consensus in HlyA and LktA, then one or two Ga^"^ ions might be bound within the GOGH-terminal signal itself (Fig. 5) as well as the upstream repeat-like sequences identified by Sebo and Ladant (1993) in GyaA. Since the cytoplasm of cells is relatively calcium poor, this binding may not actually occur before substrate recognition, but this point serves to illustrate the possible connection between the acidic repeat motifs and the structure (and possibly evolutionary origin) of the signal sequence. Golicin V is predicted by the secondary structure program of Hamodrakas (1988) to assume a conformation that is almost entirely (3-strand with numerous reverse turns (our unpublished observation). Perhaps this resemblance to an RTX toxin (ahhough these turns never bind Ga^"*") is partly responsible for the otherwise anomalous example of the secretion of this peptide antibiotic by an RTX transporter (Table 2).
Hemolysin Transport
109
D. Comparison to the General Secretory Pathway
The much-studied NH2-terminal signal sequence that directs protein traffic through the general secretory pathway across the cytoplasmic membrane (or endoplasmic reticulum membrane in eukaryotes) has been shown to adjust its conformation in response to a lipid environment, and this has been postulated to be a critical step in its recognition by the transport apparatus and its subsequent translocation (Briggs et al., 1985). Indeed, this feature is one of the most characteristic features of signal sequences of this kind, since primary structure is not at all conserved. In general, NH2-terminal signals consist of a degenerate sequence of 15-30 hydrophobic amino acids with a positively charged NH2 terminus (Gierasch, 1989). However, genetic and biochemical studies of the general secretory pathway seem to indicate that the signal is recognized by protein-protein interactions (such as the signal recognition particle and secB) (Hartl et al., 1990) at all steps and then pumped across the inner membrane by a "sewing machine-like" action on the part of the Sec A ATPase (Wickner and Leonard, 1994). In contrast to the general secretory pathway, no chaperonin proteins have been implicated in hemolysin secretion (Gentschev et al., 1990). It seems that the HlyA signal is able to associate directly with the cell membrane (Gentschev and Goebel, 1992), whereas NH2-terminal signal peptides are likely bound by protein almost from the moment of their synthesis. Nevertheless, the putative membrane-binding regions of the HlyA signal overlap with the regions defined as interacting with HlyB by complementation experiments (Zhang et al., 1993b). In both the general secretory pathway and the hemolysin secretion system it seems that the ability to interact with membranes (whether or not such interaction actually occurs in vivo) is an important feature of the transport signal. It may be that the very motifs needed for efficient membrane interaction are those recognized by the protein secretion apparatus and that membrane association is an optional intermediate in the pathway. HlyB might therefore bind to membrane association motifs in the signal of HlyA, in effect substituting itself for the membrane as a first step in the specific secretion process. Like the general secretory pathway, the hemolysin transport system requires both ATP and proton motive force for function (Arkowitz et al., 1993; Koronakis et al., 1993). This dual energy requirement may reflect the different mechanistic requirements of membrane association, or translocation of the signal sequence, versus the later nonspecific translocation of the passenger polypeptide chain. There are precedents within the ABC superfamily for recognition of substrate from within the plane of the membrane. The yeast a factor (pheromone) exporter STE6 will not recognize its substrate unless it is acylated (which is required for a factor to associate with the cell membrane) (Caldwell et al., 1994), and P-glycoproteins either secrete hydrophobic drugs, probably from within the membrane bilayer (Pgpl and Pgp2) (Gottesman and Pastan, 1993), or act as a flippase for phosphatidylcholine (Pgp3) (Ruetz and Gros, 1994). A potential kinetic advantage could be
110
JONATHAN A. SHEPS, FANG ZHANG, and VICTOR LING
gained by a signal sequence that directs a protein to the cytoplasmic membrane and could then search for a transporter in two dimensions rather than three in the cytoplasm.
VI. CONCLUSION The overall behavior of ABC transporters as a family is that they are tremendously adaptable and can accommodate a large number of substrates. The secretion of RTX toxins appears to make use of a diverse group of primary amino acid sequences whose recognition by the transporter seems to be explicable in terms of secondary structure, charge distribution, and ability to interact with membranes. It could be that each transporter has its own preferred set of critical contact residues and these are presented to the transporter within a conserved secondary structure framework. At the same time the transporter is capable of recognizing, albeit with loss of efficiency, signals in which several elements have been disabled. Perhaps this adaptability has allowed ABC transporters involved in secretion of RTX toxins to take as recognition elements numerous features present in the COOH-terminal region of the protein. These elements may have originated as degenerate forms of the Ca^'*"-binding repeat motif Other features might be present in order to permit membrane association by the COOH-terminal signal, and these same elements may have been seized upon by the transporter as recognition features that would allow the transport complex to bind the signal by simply substituting itself for the membrane. A model that might explain all of the disparate data is one in which there is a conserved secondary structure that is required for the transport signal to gain admission to the binding pocket of HlyB. Once this loose association is achieved there would be a stage of tighter binding, presumably dependent on specific interactions between particular amino acids in the transporter and the toxin. If the binding pocket is sufficiently large and adaptable in terms of being able to transmit a signal for opening the transport channel, then different primary sequences could be accommodated, as long as their secondary structural features remained consistent with the overall dimensions of the binding pocket. The "critical" contact residues could be different from what are used by the native substrate and may differ for each substrate that can be acconmiodated. Indeed, much of the recognition may occur via hydrophobic interactions, which will be more difficult to define rigorously by mutagenesis than interactions involving charges. Thus, the available evidence suggests that the hemolysin signal is composed of a number of overlapping and redundant motifs. These signal motifs appear to be recognized by an array of semi-independent recognition sites within the sequence of HlyB. Heterologous or mutated substrates of HlyB could be accommodated by a combination of conserved and novel, opportunistic interactions. Single point mutations in HlyB are capable of creating large changes in substrate specificity (Zhang et al., 1993b), and
Hemolysin Transport
111
this hints that the broad specificity seen in HlyB could readily be broadened further in the course of evolution. It has been suggested that the mammalian multidrug transporter Pgp recognizes multiple substrates by virtue of having multiple independent recognition sites dispersed among its transmembrane spans and cytoplasmic loops (Loo and Clarke, 1994). Pgp may have evolved from an RTX-like protein transporter (see Fig. 3), and it appears that a HlyB-like substrate recognition site could easily be converted into a mechanism for coupling a number of independent drug-binding sites to a common transport pathway. The evolutionary flexibility of the ABC transporter superfamily can be seen as a product of the flexibility inherent in an RTX transporter-like substrate recognition system. A major challenge for the future will be to determine not only the details of particular substrate-transporter interactions but to elucidate the general mechanism whereby contacts across a number of recognition sites on the transporter can be integrated to activate the transport process.
ACKNOWLEDGMENTS We would like to thank our colleagues at the Ontario Cancer Institute for their critical reading of the manuscript, in particular, Sarah Childs, Adam Shapiro, and Cheryl Arrowsmith. We are especially grateful to our colleagues in other RTX transporter laboratories who shared with us reprints and information before publication. The studies in our laboratory were supported by a grant from the Medical Research Council of Canada. J. A. Sheps was the recipient of a graduate studentship from the Natural Sciences and Engineering Research Council of Canada.
REFERENCES Ames, G. F. -L., Nikaido, K., Groarke, J., & Petithory. J. (1989). Reconstitution of periplasmic transport in inside-out membrane vesicles. Energization by ATP. J. Biol. Chem. 264, 3998—4002. Anderson, M. R, Gregory, R. J., Thompson, S., Sonza, D. W., Paul, S., Mulligan, R. C, Smith, A. £., & Welsh, M. J. (1992). Demonstration that CFTR is a chloride channel by alteration of its anion selectivity. Science 253, 202-205. Arkowitz, R. A., Joly, J. C, & Wickner, W. (1993). Translocation can drive unfolding of a preprotein domain. EMBO J. 12, 243-253. Baichwal, V., Liu, D., & Ames, G. F. (1993). The ATP-binding component of a prokaryotic traffic ATPase is exposed to the periplasmic (external) surface. Proc. Natl. Acad. Sci. USA 90, 620-624. Balzi, E., Wang, M., Leterme, S., Van Dyck, L., & Goffeau, A. (1994). PDR5, a novel yeast multidrug resistance conferring transporter controlled by the transcription regulator PDRl. J. Biol. Chem. 269,2206-2214. Baumann, U. (1994). Crystal structure of 50 kDa metallo protease from Serratia marcescens, J. Mol. Biol. 242, 244-251. Baumann, U., Wu, S., Flaherty, K. M., & McKay, D. B. (1993). Three-dimensional structure of the alkaline protease of Pseudomonas aeruginosa: a two-domain protein with a calcium binding parallel beta roll modf EMBO J. 12, 3357-3364.
112
JONATHAN A. SHEPS, FANG ZHANG, and VICTOR LING
Bayer, M. E. (1968). Areas of adhesion between wall and membrane of Escherichia coli. J. Gen. Microbiol. 53, 395-^04. Benz, R., Maier, E., & Gentschev, I. (1993). TolC of Escherichia coli functions as an outer membrane channel. Zentralbl. Bakteriol. 278, 187-196. Blight, M. A. & Holland, I. B. (1990). Structure and function of haemolysin B, P-glycoprotein and other members of a novel family of membrane translocators. Mol. Microbiol. 4, 873-880. Blight, M. A., Chervaux, C, & Holland, I. B. (1994a). Protein secretion pathways in E. coli. Curr. Opin. Biotechnol. 5, 46^-474. Blight, M. A., Pimenta, A. L., Lazzaroni, J.-C, Dando, C, Kotelevets, L., Seror, S. J., & Holland, I. B. (1994b). Identfication and preliminary characterization of temperature sensitive and other secmutations affecting HlyB, responsible for the secretion of haemolysin (HlyA) from E. coli. Mol. Gen. Genet. 245,431-440. Blight, M. A., Menichi, B., & Holland, I. B. (1995). Evidence for post-transcriptional regulation of the synthesis of the E. coli HlyB haemolysin translocator and over-production of the ATP-binding domain. Mol. Gen. Genet. 247, 73-^5. Boehm, D. F., Welch, R. A., & Snyder, I. S. (1990a). Calcium is required for binding of Escherichia coli hemolysin (HlyA) to erythrocyte membranes. Infect. Immun. 58, 1951—1958. Boehm, D. F., Welch, R. A., & Snyder, I. S. (1990b). Domains of Escherichia coli hemolysin (HlyA) involved in binding of calcium and erythrocyte membranes. Infect. Immun. 58, 1959-1964. Briggs, M. S., Gierasch, L. M., Zlotnick, A., Lear, J. D., & DeGrado, W. F. (1985). In vivo function and membrane binding properties are correlated for Escherichia coli LamB signal peptides. Science 228, 1096-1099. Broome-Smith, J. K. & Spratt, B. (1986). A vector for the construction of translational fusions to TEM p-lactamase and the analysis of protein export signals and membrane protein topology. Gene 49, 341-349. Caldwell, G. A., Wang, S. H., Naider, F., & Becker, J. M. (1994). Consequences of altered isoprenylation targets on a-factor export and bioactivity. Proc. Natl. Acad. Sci. USA 91, 1275—1279. Cangelosi, G. A., Martinetti, G., Leigh, J. A., Lee, C. C, Theines, C, & Nester, E. W. (1989). Role of Agrobacterium tumefaciens ChvA protein in export of p-1,2-glucan. J. Bacteriol .171,1609-1615. Chang, Y.-F., Young, R., Moulds, T. L., & Struck, D. K. (1989). Secretion of the Pasteurella leukotoxin by Escherichia coli FEMS Microbiol. Lett. 60, 169-174. Chang, Y.-F., Young, R., & Struck, D. K. (1991). The actinobacillus pleuropneumoniae hemolysin determinant: unlinked appCA and appBD loci flanked by pseudogenes. J. Bacteriol. 173, 5151— 5158. Childs, S. & Ling, V. (1994). The MDR superfamily of genes and its biological implications. In: Important Advances in Oncology 1994 (DeVita, V. T, Hellman, S., & Rosenberg, S. A., Eds.) Lippincott, Philadelphia. Chung, Y. J., Steen, M. T, & Hansen, J. N. (1992). The subtilin gene of Bacillus subtilis ATCC 6633 is encoded in an operon that contains a homolog of the hemolysin B transport protein. J. Bacteriol. 174, 1417-1422. Cole, S. R C, Bhardwaj, G., Gerlach, J. H., Mackie, J. E., Grant, C. E., Almquist, K. C , Stewart, A. J., Kurz, E. U., Duncan, A. M. V., & Deeley, R. G. (1992). Overexpression of a transporter gene in a multidrug-resistant human lung cancer cell line. Science 258, 1650-1654. Delepelaire, P. (1994). PrtD, the integral membrane ATP-binding-cassette component of the Erwinia chysanthemi metalloprotease secretion system, exhibits a secretion signal-regulated ATPase activity. J. Biol. Chem. 269, 27952-27957. Delepelaire, P. & Wandersman, C. (1990). Protein secretion in Gram-negative bacteria. The extracellular metal loprotease B from Erwinia chrysanthemi contains a C-terminal secretion signal analogous to that of Escherichia coli a-hemolysin. J. Biol. Chem. 265, 17118-17125.
Hemolysin Transport
113
Delepelaire, P. & Wandersman, C. (1991). Characterization, localization and transmembrane organization of the three proteins PrtD, PrtE and PrtF necessary for protease secretion by the Gram-negative bacterium Erwina chrysanthemi. Mol. Microbiol. 5, 2427—2434. Devine, S. E., Ling, V., & Melera, P. W. (1992). Amino acid substitution in the sixth transmembrane domain of p-glycoprotein alters multidrug resistance. Proc. Natl. Acad. Sci. USA 89,4564-4568. Doige, C. A. & Ames, G. F.-L. (1993). ATP-dependent transport systems in bacteria and humans: relevance to cystic fibrosis and multidrug resistance. Annu. Rev. Microbiol. 47, 291-319. Dorman, C.J., Lynch, A. S., Bhriain, N. N., & Higgins, C. F. (1989). DNA supercoiling in Escherichia coli: topA mutations can be suppressed by DNA amplifications involving the tolC locus. Mol. Microbiol. 3, 531-540. Dreesen, T. D., Johnson, D. H., & Henikoff, S. (1988). The brown protein of Drosophila melanogaster is similar to the white protein and to components of active transport complexes. Mol. Cell. Biol. 8,5206-5215. Dudler, R. & Hertig, C. (1992). Structure of an mdr-like gene from Arabidopsis thaliana. Evolutionary implications. J. Biol. Chem. 267, 5882-5888. Duong, F., Lazdunski, A., Cami, B., & Murgier, M. (1992). Sequence of a cluster of genes controlling synthesis and secretion of alkaline protease in Pseudomonas aeruginosa: relationships to other secretory pathways. Gene. 121, 47-54. Economou, A., Hamilton, W. D. O., Johnston, A. W. B., & Downie, J. A. (1990). The Rhizobium nodulation gene nodO encodes a Ca ^-binding protein that is exported without N-terminal cleavage and is homologous to haemolysin and related proteins. EMBO J. 9, 349-354. Eisenberg, D., Schwarz, E., Komaromy, M., & Wall, R. (1984). Analysis of membrane and surface protein sequences with the hydrophobic moment plot. J. Mol. Biol. 179, 125-142. Evans, L J. & Downie, J. A. (1986). The nodi gene product of Rhizabium leguminosarum is closely related to ATP-binding bacterial export proteins: nucleotide sequence analysis of the nodi and nodJ genes. Gene 43, 95-101. Farris, J. S. (1988). HENNIG86, version 1.5. Distributed by the author. Port Jefferson Station, NY. Fath, M. J. & Kolter, R. (1993). ABC transporters: bacterial exporters. Microbiol. Rev. 57, 995-1017. Fath, M. J., Skvirsky, R. C, & Kolter, R. (1991). Functional complementation between bacterial MDR-like export systems: colicin V, alpha-hemolysin, and Erwinia protease. J. Bacteriol. 173, 7549-7556. Felmlee, T. & Welch, R. A. (1988). Alterations of amino acid repeats in the Escherichia coli hemolysin affect cytolytic activity and secretion. Proc. Natl. Acad. Sci. USA 85, 5269-5273. Felmlee, T, Pellett, S., Lee, E. Y, & Welch, R. A. (1985a). Escherichia coli hemolysin is released extracellularly without cleavage of a signal peptide. J. Bacteriol. 163, 88—93. Felmlee, T., Pellett, S., & Welch, R. A. (1985b). Nucleotide sequence of an Escherichia coli chromosomal hemolysin. J. Bacteriol. 163, 94-105. Foote, S. J., Thompson, J. K., Cowman, A. F., & Kemp, D. J. (1989). Amplification of the multidrug resistance gene in some chloroquine-resistant isolates of P. falciparum. Cell. 57, 921—930. Garrido, M. C, Herrero, M., Kolter, R., & Moreno, F. (1988). The export of the DNA replication inhibitor microcin B17 provides immunity for the host cell. EMBO J. 7, 1853—1862. Gentschev, L, Hess, J., & Goebel, W. (1990). Change in the cellular localization of alkaline phosphatase by alteration of its carboxy-terminal sequence. Mol. Gen. Genet. 222, 211-216. Gentschev, L & Goebel, W. (1992). Topological and functional studies on HlyB of Escherichia coli. Mol. Gen. Genet. 232, 40-48. Gerlach, J. H., Endicott, J. A., Juranka, P. F., Henderson, G., Sarangi, F., Deuchars, K. L., & Ling, V. (1986). Homology between P-glycoprotein and a bacterial haemolysin transport protein suggests a model for multidrug resistance. Nature 324, 485-489. Ghigo, J. M. & Wandersman, C. (1992). A fourth metalloprotease gene in Erwinia chrysanthemi. Res. Microbiol. 143,857-867.
114
JONATHAN A. SHEPS, FANG ZHANG, and VICTOR LING
Ghigo, J. M. & Wandersman, C. (1994). A carboxyl-terminal four-amino acid motif is required for secretion of the metalloprotease PrtG through the Erwinia chrysanthemi protease secretion pathway. J. Biol. Chem. 269, 8979-8985. Gierasch, L. M. (1989). Signal sequences. Biochemistry 28, 923-930. Gilson, L., Higgins, C. F., Hofnung, M., & Ames, G. F.-L. (1982). Extensive homology between membrane associated components of histidine and maltose transport systems of Salmonella typhimurium and Escherichia coli. J. Biol. Chem. 257, 9915-9918. Gilson, L., Mahanty, H. K., & Kolter, R. (1990). Genetic analysis of an MDR-like export system: the secretion of colicin V. EMBO J. 9, 3875-3894. Glaser, R, Sakamoto, H., Bellalou, J., UUman, A., & Danchin, A. (1988). Secretion of cyclolysin, the calmodulin-sensitive adenylate cyclase-haemolysin bifunctional protein of Bordetella pertussis. EMBO J. 7, 3997-4004. Gottesman, M. M, Pastan, 1. (1993). Biochemistry of multidrug resistance by the multidrug transporter. Annu. Rev. Biochem. 62, 385-427. Gray, L., Mackman, N., Nicaud, J. M., & Holland, I. B. (1986). The carboxy-terminal region of haemolysin 2001 is required for secretion of the toxin from Escherichia coli. Mol. Gen. Genet. 205, 127-133. Gray, L., Baker, K., Kenny, B., Mackman, N., Haigh, R., & Holland, I. B. (1989). A novel C-terminal signal sequence targets Escherichia coli haemolysin directly to the medium. J. Cell Sci. Suppl. 11,45-57. Gros, P., Dhir, R., Croop, J., & Talbot, F. (1991). A single amino acid substitution strongly modulates the activity and substrate specificity of the mouse mdrl and mdr3 drug efflux pumps. Proc. Natl. Acad. Sci. USA 88, 7289-7293. Guzzo, J., Duang, F., Wandersman, C, Murgier, M., & Lazdunski, A. (1991). The secretion genes of Pseudomonas aeruginosa alkaline protease are functionally related to those of Erwina chrysanthemi proteases and Escherichia coli a-haemolysin. Mol. Microbiol. 5, 447-453. Hacker, J. & Hughes, C. (1985). Genetics of Escherichia coli hemolysin. Curr. Top. Microbiol. Immunol. 118, 139-162. Hamodrakas, S. J. (1988). A protein secondary structure prediction scheme for the IBM PC and compatibles. CABIOS 4, 423-477. Hartl, F.-U., Lecker, S., Schiebel, E., Hendrick, J. P, & Wickner, W. (1990). The binding cascade of SecB to Sec A to SecY/E mediates preprotein targeting to the E. coli plasma membrane. Cell 63, 269-279. Hartlein, M., Schiessl, S., Wagner, W, Rdest, U., Kreft, J., & Goebel, W. (1983). Transport of hemolysin by Escherichia coli. J. Cell. Biochem. 22, 87-97. Hess, J., Gentschev. I., Goebel, W., & Jarchau, T. (1990). Analysis of the haemolysin secretion system by PhoA-HlyA fusion proteins. Mol. Gen. Genet. 224, 201-208. Higgins, C. F. (1992). ABC transporters: from microorganisms to man. Annu. Rev. Cell Biol. 8,67—113. Higgins, C. F., Haag, R D., Nikaido, K., Ardeshir, G., Garcia, G., & Ames, G. F. -L. (1982). Complete nucleotide sequence and identification of membrane components of the histidine operon of S. typhimurium. Nature 298, 723-727. Higgins, C. F., Hiles, I. D., Whalley, K., & Jamieson, D. J. (1985). Nucleotide binding by membrane components of bacterial periplasmic binding protein-dependent transport systems. EMBO J. 4, 1033-1040. Highlander, S. K., Chidambaram, M., Engler, M. J., & Weinstock, G. M. (1989). DNA sequence of the Pasteurella haemolytica leukotoxin gene cluster. DNA 8, 15—28. Hiraga, S., Niki, H., Ogura, T, Ichinose, C , Mori, H., Ezaki, B., & Jaffe, A. (1989). Chromosome partitioning in Escherichia coli: novel mutants producing anucleate cells. J. Bacteriol. 171, 1496-1505. Hirst, T. R. & Welch, R. A. (1988). Mechanisms for secretion of extracellular proteins by Gram-negative bacteria. Trends Biochem. Sci. 13,265-269.
Hemolysin Transport
115
Holland, D. & Wolk, C. P. (1990). Identification and characterization of hetA, a gene that acts early in the process of morphological differentiation of heterocysts. J. Bacteriol. 172, 3131-3137. Holland, I. B., Kenny, B., & Blight, M. (1990). Haemolysin secretion from E. coli. Biochimie 72, 131-141. Husain, I., van Houten, B., Thomas, D. C, & Sancar, A. (1986). Sequences of Escherichia coliuwA gene and protein reveal two potential ATP binding sites. J. Biol. Chem. 261,4895-4901. Juranka, P. R, Zastawny, R. L., & Ling, V. (1989). P-glycoprotein: multidrug-resistance and a superfamily of membrane-associated transport proteins. FASEB J., 3, 2583—2592. Juranka, P, Zhang, F., Kulpa, J., Endicott, J., Blight, M., Holland, I. B., & Ling, V. (1992). Characterization of the hemolysin transporter, HlyB, using an epitope insertion. J. Biol. Chem. 267, 3764^3770. Kamijo, K., Taketani, S., Yokota, S., Osumi, T., & Hashimoto, T. (1990). The 70-kDa peroxisomal membrane protein is a member of the mdr (P-glycoprotein)-related ATP-binding protein superfamily. J. Biol. Chem. 265, 4534-4540. Kellenberger, E. (1990). The "Bayer bridges" confronted with results from improved electron microscopy methods. Mol. Microbiol. 4, 697-705. Kenny, B., Haigh, R., & Holland, L B. (1991). Analysis of the haemolysin transport process through the secretion from Escherichia coli of PCM, CAT or beta-galactosidase fused to the HlyA C-terminal signal domain. Mol. Microbiol. 5, 2557-2568. Kenny, B., Taylor, S., & Holland, LB. (1992). Identification of individual amino acids required for secretion within the haemolysin (HlyA) C-terminal targeting region. Mol. Microbiol. 6, 14771489. Kenny, B., Chervaux, C, & Holland, I. B. (1994). Evidence that residues-15 to-46 of the haemolysin secretion signal are involved in early steps in secretion, leading to recognition of the translocator. Mol. Microbiol. 11,99-109. Kerppola, R. E., Shyamala, V. K., Klebba, P., & Ames G. F. (1991). The membrane-bound proteins of periplasmic permeases form a complex. Identification of the histidine permease HisQMP complex. J. Biol. Chem. 266, 9857-9865. Koronakis, V., Cross, M., & Hughes, C. (1988a). Expression of the E. coli hemolysin secretion gene hlyB involves transcript anti-termination within the hly operon. Nucleic Acids Res. 16, 47894800. Koronakis, V., Koronakis, E., & Hughes, C. (1988b) Comparison of the haemolysin secretion protein HlyB from Proteus vulgaris and Escherichia coli: site-directed mutagenesis causing impairment of export function. Mol. Gen. Genet. 213, 551—555. Koronakis, V., Koronakis, E., & Hughes, C. (1989). Isolation and analysis of the C-terminal signal directing export of Escherichia coli hemolysin protein across both bacterial membranes. EMBO J. 8, 595-605. Koronakis, V., Hughes, C, & Koronakis, E. (1991). Energetically distinct early and late stages of HlyB/HlyD-dependent secretion across both Escherichia coli membranes. EMBO J. 10, 3263— 3272. Koronakis, V., Hughes, C, Koronakis, E. (1993). ATPase activity and ATP/ADP-induced conformational change in the soluble domain of the bacterial protein translocator HlyB. Mol. Microbiol. 8, 116S-1175. Lally, E. T., Golub, E. E., Kieba, I. R., Taichman, N. S., Rosenbloom, J., Rosenbloom, J. C, Gibson, C. W., & Demuth, D. R. (1989). Analysis of the Actinobacillus actinomycetemcomitans leukotoxin gene. Delineation of unique features and comparison to homologous toxins. J. Biol. Chem. 264, 15451-15456. Letoflfe, S. & Wandersman, C. (1992). Secretion of CyaA-PrtB and HlyA-PrtB fusion proteins in Escherichia coli: involvement of the gycine-rich repeat domain of Erwinia chrysanthemi protease B. J. Bacteriol. 174, 4920-4927.
116
JONATHAN A. SHEPS, FANG ZHANG, and VICTOR LING
Letoffe, S., Delepelaire, P., & Wandersman, C. (1990). Protease secretion by Erwinia chrysanthemi: the specific secretion functions are analogous to those of Escherichia coli alpha-haemolysin. EMBO J. 9, 1375-1382. Letoffe, S., Delepelaire, P., & Wandersman, C. (1991). Cloning and expression in Escherichia coli of the Serratia marcescens metalloprotease gene: secretion of the protease from E. coli in the presence of the Erwinia chrysanthemi protease secretion functions. J. Bacteriol. 173, 2160-2166. Letoffe, S., Ghigo, J. M., & Wandersman, C. (1993). Identification of two components of the Serratia marcescens metal loprotease transporter: protease SM secretion in Escherichia coli is TolC dependent. J. Bacteriol. 175, 7321-7328. Letoffe, S., Ghigo, J. M., & Wandersman, C. (1994a). Secretion of the Serratia marcescens HasA protein by an ABC transporter. J. Bacteriol. 176, 5372-5377. Letoffe, S., Ghigo, J. M., & Wandersman, C. (1994b). Iron acquisition from heme and hemeglobin by a Serratia marcescens extracellular protein. Proc. Natl. Acad. Sci. USA91, 9876-9880. Lewis, K. (1994). Multidrug resistance pumps in bacteria: variations on a theme. Trends Biochem. Sci. 19, 119^123. Loo, T. W. & Clarke, D. M. (1993a) Functional consequences of proline mutations in the predicted transmembrane domains of P-glycoprotein. J. Biol. Chem. 268, 3143-3149. Loo, T. W. & Clarke, D. M. (1993b) Functional consequences of phenylalanine mutations in the predicted transmembrane domains of P-glycoprotein. J. Biol. Chem. 268, 19965—19972. Loo, T. W. & Clarke, D. M. (1994) Functional consequences of glycine mutations in the predicted cytoplasmic loops of P-glycoprotein. J. Biol. Chem. 269, 7243-7248. Ludwig, A., Jarchau, T., Benz, R., & Goebel, W. (1988). The repeat domain oi Escherichia coli haemolysin (HlyA) is responsible for its Ca ^-dependent binding to erythrocytes. Mol. Gen. Genet. 214,553-561. Ludwig, A., Schmid, A., Benz, R., & Goebel W. (1991). Mutations affecting pore formation by haemolysin from Escherichia coli. Mol. Gen. Genet. 226, 198-208. Mackman, N., Nicaud, J. M., Gray, L., & Holland, I. B. (1985a). Genetical and functional organization of the Escherichia coli haemolysin determinant 2001. Mol. Gen. Genet. 201, 282-288. Mackman, N., Nicaud, J. M., Gray, L., & Holland, I. B. (1985b). Identification of polypeptides required for the export of haemolysin 2001 from E. coli. Mol. Gen. Genet. 201, 529-536. Mackman, N., Nicaud, J. M., Gray, L., & Holland, I. B. (1986) Secretion of haemolysin by Escherichia coli. Curr. Top. Microbiol. Immunol. 125, 159-181. Mackman, N., Baker, K., Gray, L., Haigh, R., Nicaud, J. M., & Holland, I. B. (1987). Release of a chimeric protein into the medium from Escherichia coli using the C-terminal secretion signal of haemolysin. EMBO J. 6, 2835-2841. Maddison, W. P. & Maddison, D. R. (1992). MacClade: Analysis of Phylogeny and Character Evolution. Version 3.0. Sinauer Associates, Sunderland, MA. Masure, H. R., Au, D. C, Gross, M. K, Donovan, M. G., & Storm, D. R. (1990). Secretion of the Bordetella pertussis adenylate cyclase from Escherichia coli containing the hemolysin operon. Biochemistry 29, 140-145. McGrath, J. P. & Varshavsky, A. (1989). The yeast STE6 gene encodes a homologue of the mammalian multidrug resistance P-glycoprotein. Nature 340, 400-404. Menestrina, G., Moser, C, Pellet, S., & Welch, R. (1994). Pore-formation by Escherichia coli hemolysin (HlyA) and other members of the RTX toxins family. Toxicology 87, 249-267. Mimura, C. S., Holbrook, S. R., & Ames G. F. -L. (1990). Structural model of the nucleotide-binding conserved component of periplasmic permeases. Proc. Natl. Acad. Sci. USA 88, 84—88. Morbach, S., Tebbe, S., & Schneider, E. (1993). The ATP-binding cassette (ABC) transporter for maltose/maltodextrins of Salmonella typhimurium. Characterization of the ATPase activity associated with the purified MalK subunit. J. Biol. Chem. 268, 18617-18621.
Hemolysin Transport
117
Mosser, J., Douar, A. M., Surde, C. O., Kioschis, P., Feil, R., Moser, H., Poustka, A. M., Mandel, J. L., & Aubourg, P. (1993). Putative X-linked adrenoleukodystrophy gene shows unexpected homology with ABC transporters. Nature 361, 726-730. Nicaud, J. M., Mackman, N., Gray, L., & Holland, I. B. (1986). The C-terminal, 23 kDa peptide of E. coli haemolysin 2001 contains all the information necessary for its secretion by the haemolysin (Hly) export machinery. FEBS Lett. 204, 331-335. O'Hare, K., Murphy, C, Levis, R., & Rubin, G. M. (1984). DNA sequence of the white locus of Dwsophila melanogaster. J. Mol. Biol. 180,437-455. Pellett, S., Boehm, D. F., Snyder, L S., Rowe, G., & Welch, R. A. (1990). Characterization of monoclonal antibodies against ihQ Escherichia coli hemolysin. Infect. Immun. 58, 822-827. Powis, S., Deverson, E. V., Coadwell, W. J., Ciruela, A., Huskisson, N. S., Smith, H., Butcher, G. W., & Howard, J. C. (1992). Effect of polymorphism of an MHC-linked transporter on the peptides assembled in a class I molecule. Nature 357, 211—215. Riordan, J. R., Rommens, J. M., Kerem, B.-s., Alon, N., Rozmahel, R., Grzelczak, Z., Zielinski, J., Lok, S., Plavsic, N., Chou, J -L., Drumm, M. L., lannuzzi, M. C , Collins, F. S., & Tsui, L.-C. (1989). Identification of the cystic fibrosis gene: cloning and characterization of complementary DNA. Science 245, 1066-1073. Ruetz, S. & Gros, P. (1994). Phosphatidylcholine translocase: a physiological role for the mdr2 gene. Cell 77, 1071-1081. Safa, A. R., Stem, R. K., Choi, K., Agresti, M., Tamai, I., Mehta, N. D., & Roninson, I. (1990). Molecular basis of preferential resistance to colchicine in multidrug-resistant human cells conferred by gly-185 -^ val-185 substitution in P-glycoprotein. Proc. Natl. Acad. Sci. USA 87, 7225-7229. Scheu, A. K., Economou, A., Hong, G. F., Ghelani, S., Johnston, A. W. B., & Downie, J. A. (1992). Secretion of the Rhizobium leguminosarum nodulation protein NodO by haemolysin-type systems. Mol. Microbiol. 6, 231-238. Schulein, R., Gentschev, I., Mollenkopf, H. J., & Goebel, W. (1992). A topological model for the haemolysin translocator protein HlyD. Mol. Gen. Genet. 234, 155-163. Schulein, R., Gentschev, 1., Schlor, S., Gross, R., & Goebel, W. (1994). Identification and characterization of two functional domains of the hemolysin translocator protein HlyD. Mol. Gen. Genet. 245,203-211. Sebo, P. & Ladant, D. (1993). Repeat sequences in the Bordetella pertussis adenylate cyclase toxin can be recognized as alternative carboxy-proximal secretion signals by the Escherichia coli ahaemolysin translocator. Mol. Microbiol. 9, 999-1009. Shapiro, A. B. & Ling, V. (1994). ATPase activity of purified and reconstituted P-glycoprotein from Chinese hamster ovary cells. J. Biol. Chem. 269, 3745-3754. Stanfield, S. W., lelpi, L., O'Brochta, D., Helinski, D. R., & Ditta, G. S. (1988). The ndvAgene product of Rhizobium meliloti is required for p-(l->2)glucan production and has homology to the ATP-binding export protein HlyB. J. Bacteriol. 170, 3523-3530. Stanley, P., Koronakis, V., & Hughes, C. (1991). Mutational analysis supports a role for multiple structural features in the C-terminal secretion signal of Escherichia coli haemolysin. Mol. Microbiol. 5,2391-2403. Stoddard, G. W., Petzel, J. R, van Belkum, M. J., Kok, J., & McKay, L. L. (1992). Molecular analyses of the lactococcin A gene cluster from Lactococcus lactis subsp. lactis biovar diacetylactis WM4. Appl. Environ. Microbiol. 58, 1952-1961. Strathdee, C. A. & Lo, R. Y. C. (1987). Extensive homology between the leukotoxin of Pasteurella haemolytica Al and the alpha-hemolysin of Escherichia coli. Infect. Immun. 55, 3233—3236. Strathdee, C. A. & Lo, R. Y. C. (1989). Cloning, nucleotide sequence, and characterization of genes encoding the secretion function of the Pasteurella haemolytica leukotoxin determinant. J. Bacteriol. 171,916-928.
118
JONATHAN A. SHEPS, FANG ZHANG, and VICTOR LING
Suh, W.-K., Cohen-Doyle, M. F., Fruh, K., Wang, K., Peterson, P. A., & Williams, D. B. (1994). Interaction of MHC class I molecules with the transporter associated with antigen processing. Science 264, 1322-1326. Sutton, J. M., Lea, E. J. A., & Downie, J. A. (1994). The nodulation-signaling protein NodO from Rhizobium leguminosarum biovar viciae forms ion channels in membranes. Proc. Natl. Acad. Sci. USA 91, 9990-9994. Swinkels, B. W., Evers, R., & Borst, P. (1988). The topogenic signal of glycosomal phosphoglycerate kinase of Crithidiafasciculata resides in a carboxy-terminal extension. EMBO J. 7, 1159-1165. Swinkels, B. W, Gould, S. J., Bodnar, A. G., Rachubinski, R. A., & Subramani, S. (1991). A novel, cleavable peroxisomal targeting signal at the amino-terminus of the rat 3-ketoacyl-CoA thiolase. EMBO J. 10,3255-3262. Szczypka, M. S., Wemmie, J. A., Moye-Rowley, W. S., & Thiele, D. J. (1994). A yeast metal resistance protein similar to human cystic fibrosis transmembrane conductance regulator (CFTR) and multidrug resistance-associated protein. J. Biol. Chem. 269, 22853-22857. Thomas, W D., Jr., Wagner, S. P., & Welch, R. A. (1992). A heterologous membrane protein domain fused to the C-terminal ATP-binding domain of HlyB can export Escherichia coli hemolysin. J. Bacteriol! 174,6771-6779. Trowsdale, J., Hanson, I., Mockridge, I., Beck, S., Townsend, A., & Kelly, A. (1990). Sequences encoded in the class II region of the MHC related to the "ABC" superfamily of transporters. Nature 348, 741-744. von Heijne, G. (1986). The distribution of positively charged residues in bacterial inner membrane proteins correlates with the transmembrane topology. EMBO J. 5, 3021—3027. Wandersman, C. & Delepelaire, P. (1990). TolC, an Escherichia coli outer membrane protein required for hemolysin secretion. Proc. Natl. Acad. Sci. USA 87,4776-4780. Wandersman, C. & Letoffe, S. (1993). Involvement of lipopolysaccharide in the secretion oiEscherichia coli a-hemolysin end Erwinia chrysanthemi proteases. Mol. Microbiol. 7, 141—150. Wang, R. C, Seror, S. J., Blight, M., Pratt, J. M., Broome-Smith, J. K., & Holland, I. B. (1991). Analysis of the membrane organization of an Escherichia coli protein translocator, HlyB, a member of a large family of prokaryote and eukaryote surface transport proteins. J. Mol. Biol. 217, 441-454. Wickner, W & Leonard, M. R. (1994). How do proteins cross a membrane? In: Biological Membranes: Structure, Biogenesis and Dynamics (Op den Kamp, J. A. R, Ed.), pp. 207-214, NATO ASI series H, Vol. 82. Springer-Verlag, Berlin. Wolff, N., Ghigo, J. M., Delepelaire, P., Wandersman, C, & Delepierre, M. (1994). C-terminal secretion signal of an Erwinia chrysanthemi protease secreted by a signal peptide-independent pathway: proton NMR and CD conformational studies in membrane-mimetic environments. Biochemistry 33,6792-6801. Yin, Y., Zhang, F., Ling, V., & Arrowsmith, C. (1995). Strucmral analysis and comparison of the transport signal domains hemolysin A and leukotoxin A. FEBS Lett. 366, 1-5. Zabala, J. C, Garcia-Lobo, J. M., Diaz-Aroca, E., de la Cruz, F., & Ortiz, J. M. (1984). Escherichia coli alpha-haemolysin synthesis and export genes are flanked by a direct repetition of IS91-like elements. Mol. Gen. Genet. 197, 90-97. Zhang, F., Greig, D. I., & Ling, V. (1993a). Functional replacement of the hemolysin A transport signal by a different primary sequence. Proc. Natl. Acad. Sci. USA 90,4211—4215. Zhang, F., Sheps, J. A., & Ling, V. (1993b). Complementation of transport-deficient mutants of Escherichia coli a-hemolysin by second-site mutations in the transporter hemolysin B. J. Biol. Chem. 268, 19889-19895. Zhang, F., Yin, Y, Arrowsmith, C, & Ling, V. (1995). Secretion and circular dichroism analysis of the C-terminal signal peptides of HlyA and LktA. Biochemistry 34, 4193-4201.
PROTEIN SORTING TO THE YEAST VACUOLE
Bruce F. Horazdovsky, Jeffrey H. Stack, and Scott D. Emr
I. Introduction II. Biosynthesis of Vacuolar Proteins A. Soluble Vacuolar Proteins B. Sorting Signals for Soluble Vacuolar Proteins C. Vacuolar Membrane Proteins III. Vacuolar Protein Sorting Mutants A. vpr Mutants B. vp/Mutants C. pep Mutants D. Other Mutants IV. Vacuole Morphology and Biogenesis A. Vacuole Morphology B. Vacuole Biogenesis V. Receptor-Mediated Vacuolar Protein Sorting A. The CPY Sorting Receptor VI. A Protein Kinase and a Phosphatidylinositol 3-Kinase Are Required for Vacuolar Protein Sorting A. The Vpsl5 Protein Kinase Membrane Protein Transport Volume 3, pages 119-163. Copyright © 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 1-55938-989-3 119
120 120 120 123 125 127 128 129 129 132 132 133 135 137 137 139 139
120
VII.
VIII. IX. X.
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR B. The Vps34 Phosphatidylinositol 3-Kinase C. Interactions Between Vpsl5p and Vps34p GTP-Binding Proteins in the Sorting of Vacuolar Proteins A. Vps21p B. Ypt7p C. Vpslp Vesicle Targeting Events in Vacuolar Protein Localization Clathrin and Vacuolar Protein Sorting Conclusions References
140 144 146 147 148 148 151 153 154 155
I. INTRODUCTION The function of many eukaryotic cell organelles is determined in large part by the unique set of proteins residing within them. The sorting and localization machinery used to correctly deliver specific protein constituents to these organelles is, therefore, of fundamental importance for the maintenance of their structural and functional integrity. Of the many protein localization pathways found in the cell, the movement of proteins within the secretory pathway is among the most complex. Not only is this pathway involved in the delivery of proteins to the cell surface; it is also responsible for the delivery and/or retention of resident endoplasmic reticulum (ER), Golgi complex, and vacuole/lysosome proteins. Delivery of protein and membrane to the vacuole/lysosome represents one of the major diversions of the secretory pathway. Through extensive studies using mammalian cell systems, the predominant mechanism of lysosomal protein recognition has been well defined. Less is known about the molecules and the biochemical mechanisms involved in regulating this delivery pathway. Genetic studies using the budding yeast Saccharomyces cerevisiae have uncovered a large number of genes required for the delivery of proteins to the lysosome-like vacuole. Analyses of the gene products affected in a number of the vacuolar protein-sorting {vps) mutants have uncovered proteins that are involved in the recognition of vacuolar proteins as well as molecules that serve important regulatory functions in the movement of proteins and membrane through the vacuolar protein-sorting pathway. In this chapter we will discuss the roles several of these molecules play to facilitate and regulate the delivery of proteins to the yeast vacuole.
II. BIOSYNTHESIS OF VACUOLAR PROTEINS A. Soluble Vacuolar Proteins
The yeast vacuole is very similar to the mammalian lysosome. Both are acidic compartments that contain a large number of hydrolases involved in macromolecular degradation. Extensive effort has been directed toward characterizing the biosynthesis of vacuolar hydrolases, and as a result these proteins serve as excellent
Protein Sorting to the Yeast Vacuole
121
markers for analyzing the yeast vacuolar protein localization pathway. Vacuolar proteins can be divided into two classes, based on whether they are soluble or membrane associated. The class of soluble or lumenal vacuolar hydrolases include carboxypeptidase Y (CPY), proteinase A (PrA), and proteinase B (PrB). These proteins transit through the early stages of the yeast secretory pathway en route to the vacuole (Fig. 1) (Stevens et al., 1982). Cloning and sequencing of the genes encoding CPY (PRCI) (Stevens et al., 1986; Vails et al., 1987), PrA (:PEP4) (Ammereretal., 1986; Woolfordetal., 1986),andPrB(P/?57)(Moehleetal., 1987) revealed that the predicted coding sequence of each contains an NH2-terminal hydrophobic region that serves as a signal sequence to direct translocation into the ER. Entry into and progression through the secretory pathway result in compartment-specific modifications of these vacuolar proteins that allow determination of their relative localization in the pathway. The post-translational modifications of CPY serve as particularly good indicators of transport through the secretory pathway and delivery to the vacuole; therefore, CPY maturation will be described
Core Figure 1. The yeast secretory and endosomal systems. The movement of proteins through the secretory and endosomal pathways are shown. The marker proteins include vacuolar CPY (C), the secreted protein invertase (I), and endocytosed a-factor (aF), as well as the a-factor receptor. The endoplasmic reticulum (ER), the compartments of the Golgi complex (Golgi) with its resident mannosyl transferases (al -6, a l -2, al-3), and the protease Kex2p, the early and late endosomal compartments, as well as the vacuole are depicted.
122
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
as a paradigm for the biosynthesis of soluble vacuolar proteins. CPY is synthesized as a prepro precursor molecule and translocated into the lumen of the ER, where its signal sequence is cleaved to form proCPY (Blachly-Dyson and Stevens, 1987; Johnson et al., 1987). Within the lumen of the ER, proCPY receives four A^-linked core glycosyl modifications, which results in a 67-kDa form known as pi CPY (Stevens et al., 1982; Hasilik and Tanner, 1987). Therefore, the presence of pi CPY is indicative of translocation into the ER and serves as a marker for ER localization, pi CPY is delivered to the Golgi complex via transport vesicles together with proteins destined for the cell surface. This intercompartmental transfer requires the action of SEC gene products acting in the ER-to-Golgi step of the secretory pathway (Stevens et al., 1982). After arrival in the Golgi, the core oligosaccharides of p 1 CPY are extended by the addition of a-1,6, a-1,2, and a-1,3 mannose-linked carbohydrates. This results in an increase in the apparent molecular mass of CPY to 69 kDa (Hemmings et al., 1981; Stevens et al., 1982); this form is known as p2CPY. The p2 form of CPY is recognized in a late Golgi compartment and then delivered to the vacuole. Use of a CPY-a factor—invertase fusion protein has indicated that p2CPY is sorted away from proteins destined for the cell surface in a late Golgi compartment that corresponds to or is distal to the compartment containing the Kex2 protease (Fig. 1) (Graham and Emr, 1991). p2CPY is proteolytically processed in the vacuole to generate the 61-kDa enzymatically active mature form, referred to as mCPY This final processing event is dependent on a functional PEP4 gene product, PrA (Hemmings et al., 1981). The movement of CPY through the secretory pathway and its delivery to the vacuole are fairly rapid; the conversion of CPY to the mature, vacuolar form occurs with a half-time of approximately 6 minutes (Stevens et al., 1982). The requirement for early SEC gene products, the presence of a signal sequence, and the A/^-linked modifications of CPY indicate that it transits the early stages of the secretory pathway. However, delivery of CPY to the vacuole is not affected by mutations in SEC genes whose products facilitate vesicular traffic between the Golgi and cell surface, indicating that CPY is not transported to the cell surface and subsequently endocytosed into the vacuole (Stevens et al., 1982). In addition, studies with a temperature-sensitive allele of SEC 18 have elucidated the requirement for Secl8p in the delivery of CPY to the vacuole. Sec 18p is the yeast homologue of A^-ethylmaleimide-sensitive factor (NSF), which is required for membrane fusion events in the mammalian secretory pathway (Rothman and Orci, 1992). While sec 18 mutants were initially characterized as blocking ER-to-Golgi transport, Graham and Emr demonstrated a requirement for Secl8p in ER-to-Golgi, intra-Golgi, and Golgi-to-cell surface steps (Graham and Emr, 1991). Surprisingly, sec 18 mutants did not completely block Golgi-to-vacuole delivery of CPY, suggesting the possibility that a Seel 8p homologue functions between the Golgi and the vacuole. Subcellular fractionation studies have demonstrated that, like mammalian lysosomal hydrolases, p2CPY is transported to the vacuole via an endosomal intermediate acting between the Golgi and vacuole. By using pulse-chase labeling
Protein Sorting to the Yeast Vacuole
123
of CPY combined with membrane fractionation analyses, it was found that a significant proportion of p2CPY fractionated with a post-Golgi, prevacuolar compartment (Vida et al., 1993). In addition, a mutant form of CPY that lacks a functional vacuolar sorting signal (see below) was found only in fractions corresponding to the Golgi but not the prevacuolar compartment, indicating that the prevacuolar compartment is distal to the compartment where vacuolar proteins are sorted. Finally, morphological analyses of mutants defective in the Golgi-to-vacuole transport of soluble hydrolases have identified a class of mutants that accumulate vacuolar protein precursors in a novel, prevacuolar compartment (see Section IV A) (Raymond et al., 1992). This compartment most likely represents the same prevacuolar endosomal compartment suggested by the biochemical studies of CPY maturation. Proteinase A and proteinase B are also synthesized as inactive precursors that are processed to active enzymes in the vacuole. Like CPY, PrA is initially synthesized as a precursor that accumulates in sec mutants that block translocation across the ER membrane (Klionsky et al., 1990). Upon translocation, the signal sequence of PrA is cleaved and two A^-linked core oligosaccharides are added, generating an ER modified pi precursor (Mechler et al., 1982). These core oligosaccharides are modified in the Golgi complex, resulting in a p2 precursor of 48—52 kDa. The propeptide of PrA is cleaved upon delivery to the vacuole, generating the 42-kDa mature vacuolar form of the enzyme (Ammerer et al., 1986; Rothman et al., 1986). Like CPY, the half-time of PrA maturation (delivery to the vacuole) is approximately 6 minutes. The maturation and activation of PrA likely result from an autocatalytic event followed by a subsequent processing event by PrB (Jones et al., 1989). The serine protease PrB is modified by both A^- and O-linked oligosaccharides (Moehle et al., 1989). PreproPrB is initially synthesized as a large precursor of approximately 70 kDa. Upon translocation into the ER the signal sequence is removed followed by the cleavage of approximately 260 NH2-terminal amino acids, resulting in a 40-kDa species. This processing event also takes place in the ER. Upon arrival at the vacuole a PrA-dependent cleavage event produces the 37-kDa form of PrB, which is followed by an apparent autocatalytic event that produces the mature 31-kDa form of the protein (Moehle et al., 1989). The generation of mature PrB occurs with approximately the same half-time as that seen for CPY and PrA, suggesting that all three of these soluble proteases are delivered to the vacuole via the same localization pathway. Although the prevacuolar precursors of PrA and PrB are not quite as discernible as CPY, they represent valuable additional markers for testing the overall fidelity of this localization pathway. B. Sorting Signals for Soluble Vacuolar Proteins As is the case for the well-characterized mannose-6-phosphate recognition system for mammalian lysosomal proteins, there is compelling evidence for the
124
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
existence of positive sorting signals directing soluble hydrolases to the vacuole in yeast. However, glycosyl modifications do not appear to be involved in the targeting of hydrolases to the yeast vacuole. CPY and PrA are efficiently delivered to the vacuole in the presence of tunicamycin, which blocks A^-linked carbohydrate addition (Schwaiger et al., 1982; Stevens et al., 1982; Klionsky et al., 1988). In addition, a mutant form of CPY in which all of its A^-linked glycosylation acceptor sites have been altered is also efficiently sorted to the vacuole (Winther et al., 1991). Rather than resulting from the post-translational modification of a carbohydrate moiety, the sorting signal of vacuolar hydrolases in yeast resides in their primary amino acid sequence. Studies of fusion proteins formed between CPY and the normally secreted enzyme invertase (Inv) have mapped residues in CPY that are sufficient to direct CPY-Inv fusion proteins to the vacuole (Bankaitis et al., 1986; Johnson et al., 1987). The SUC2 gene encoding secreted invertase was engineered by removing portions corresponding to its signal sequence and then fused to the PRCl gene encoding CPY. This produced a fusion protein that contains the NH2 terminus of CPY fused in-frame with invertase. This CPY-Inv molecule retains invertase enzyme activity and thus serves as an excellent marker for determining cellular localization of the fusion. Preliminary work found that the NH2-terminal 433 amino acids of CPY fused to invertase could target the CPY-Inv fusion to the vacuole (Bankaitis et al., 1986). Further mapping studies narrowed the CPY sorting signal to the NH2-terminal 50 amino acids. These 50 residues include a 20-amino acid signal sequence that is cleaved upon translocation into the ER to expose a vacuolar targeting signal within the remaining 30 residues of proCPY. Deletion of residues 21—50 in wild-type CPY caused the secretion of the Golgi-modified precursor form of CPY, confirming the presence of a vacuolar sorting signal in these residues (Johnson et al., 1987). Extensive deletion and site-directed mutagenesis studies of CPY also demonstrated the existence of a vacuolar sorting signal in the propeptide of CPY (Vails et al., 1987). It was found that a stretch of four amino acids, Q24RPL27, constitutes the core of the vacuolar targeting signal of CPY. Mutations altering this sorting signal resulted in the missorting and secretion of the Golgi-modified p2 precursor form of CPY, indicating that these residues are necessary for delivery of CPY to the vacuole (Vails et al., 1990) The secretion of p2CPY requires the action of SEC gene products functioning between the Golgi and cell surface (Stevens et al., 1982), indicating that secreted mutant precursor CPY has progressed through the ER and Golgi compartments in a manner similar to that of wild-type CPY. Yeast cells expressing both sorting-defective and wild-type forms of CPY effectively localize wild-type CPY to the vacuole and missort and secrete the mutant form of CPY (Vails et al., 1990). This suggests that there is no interaction between the various forms of CPY during transit through the secretory pathway and does not support a role for CPY oligomerization during the sorting process. Taken together, these studies have demonstrated that a positive sorting signal for delivery to the vacuole exists in the propeptide of CPY and alteration of this signal leads to the missorting
Protein Sorting to the Yeast Vacuole
125
and secretion of precursor CPY. Therefore, secretion represents the defauh pathway that CPY follows in the absence of cw-linked signals directing it to the vacuole. This is further supported by the observation that p2CPY is secreted from the cell in mutants lacking Golgi-to-vacuole transport function (see Section III). Similar invertase fusion protein studies have established that a vacuolar proteinsorting signal is also present in the pro region of PrA. PrA-Inv fusions containing the NH2-terminal 76 amino acids of PrA were efficiently delivered to the vacuole (Klionsky et al., 1988). In a manner identical to that of CPY-Inv fusion proteins, vacuolar localization of PrA-Inv required SEC gene function, and secretion of mutant PrA-Inv fusion proteins is blocked in sec mutants defective in Golgi-to-cell surface transport. Unfortunately, deletions in wild-type PrA corresponding to the sorting signal mapped by using PrA-Inv fusions were unstable. This may indicate an important function for the propeptide of PrA in the folding of the protein into a protease-resistant form upon translocation into the ER (van den Hazel et al., 1993). Small deletions in the propeptide of PrA, while stable, did not result in the secretion of precursor PrA. Despite these difficulties, the analyses of PrA-Inv fusion proteins have demonstrated that a region of the propeptide of PrA is sufficient to direct the PrA-Inv fusion to the vacuole. Sequence comparisons of the pro regions of CPY and PrA have not detected any significant similarities, which suggests that they may not be recognized by the same receptor (see Section V). C. Vacuolar Membrane Proteins
In addition to soluble hydrolases, the yeast vacuole contains a variety of membrane proteins that function in the macromolecular degradation and maintenance of an acidic environment (Klionsky et al., 1990). Therefore, the accurate and efficient delivery of soluble and membrane vacuolar proteins is essential for proper vacuolar function. One of these proteins, alkaline phosphatase (ALP), is a wellcharacterized vacuolar integral membrane protein. Like soluble vacuolar proteins, ALP is synthesized as an inactive precursor that transits the early stages of the secretory pathway (Klionsky and Emr, 1989). ALP contains an NH2-terminal hydrophobic domain of approximately 20 amino acids that appears to function as the ER translocation signal sequence as well as a transmembrane domain anchor. Protease protection experiments have confirmed this prediction, demonstrating that ALP is a type II transmembrane protein. Upon delivery to the vacuole, this precursor is cleaved in a PrA-dependent reaction to generate the mature form of ALP. Unlike the soluble vacuolar proteins described above, the propeptide of ALP is found at its COOH terminus. Initial studies of ALP using ALP-invertase fusion proteins indicated that a vacuolar sorting signal may exist in the transmembrane domain of the cytoplasmic tail of ALP. However, recent studies with another vacuolar membrane protein and several resident Golgi membrane proteins indicate that a vacuolar sorting signal may not be required for the delivery of membrane proteins to the vacuole (Roberts et al., 1992).
126
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
Another vacuolar protein, carboxypeptidase S (GPS), is unique in that it appears to be transiently associated with the vacuolar membrane. Like other vacuolar proteins, CPS contains an NH2-terminal signal sequence and transits the early stages of the secretory pathway (Spormann et al., 1991, 1992), where it acquires A^-linked oligosaccharides. Modification by either two or three oligosaccharides generates two precursor forms of CPS of 78 and 81 kDa, respectively (Spormann et al., 1992). Upon their arrival at the vacuole, the CPS precursors are processed to their mature forms of 74 and 77 kDa in a PrB-dependent manner (Spormann et al., 1992). Interestingly, if maturation is blocked (using a PrB- and PrA-deficient strain) the CPS precursor molecules are found to be associated with the vacuolar membrane. Only if processing occurs are the mature forms of the enzyme found in soluble form. It has been proposed that the NH2-terminal hydrophobic domain of CPS serves as a transmembrane domain and that CPS remains associated with the membrane until delivery to the vacuole, where cleavage of the prosegment (containing the transmembrane domain) releases mature CPS into the vacuole lumen. In addition to ALP, one of the two major organellar dipeptidyl aminopeptidase activities (DPAP B) is found associated with the vacuolar membrane. Another activity (DPAP A) is localized to the Golgi complex. The genes encoding DPAP A and B have been cloned and sequenced (Roberts et al., 1989; J. Thomer, personal communication). DPAP A and DPAP B are both predicted to be type II integral membrane proteins and share extensive sequence similarity in their catalytic lumenal domains, although their transmembrane and cytoplasmic domains are not significantly homologous. Attempts to map a vacuolar localization signal in DPAP B examined the lumenal, transmembrane, and cytoplasmic domains separately (Roberts et al., 1992). Fusion of the signal sequence of prepro-a-factor to the lumenal domain of DPAP B resulted in secretion of the protein. Replacement of the lumenal domain of DPAP B with the coding sequence of the normally secreted enzyme invertase resulted in a DPAP B-Inv fusion protein that was delivered to the vacuole. These data indicate that the lumenal domain of DPAP B is neither necessary nor sufficient to target the protein to the vacuole. Replacement of the transmembrane domain of DPAP B with the transmembrane domain of DPAP A resulted in a hybrid protein that was efficiently localized to the vacuole. Finally, deletion of essentially the entire cytoplasmic tail of DPAP B still led to delivery to the vacuole. Delivery of these chimeric and mutant proteins to the vacuole does not appear to require their appearance at the cell surface and subsequent endocytosis. Use of a seel mutant, which blocks Golgi to plasma membrane traffic, demonstrated that these forms of DPAP B are targeted directly from the Golgi to the vacuole (Roberts et al., 1992). These unexpected data led Stevens and co-workers to conclude that a positive vacuolar sorting signal is not present in the lumenal, transmembrane, or cytoplasmic domains of DPAP B and led to the proposal that delivery of membrane proteins to the vacuole does not require specific sorting information, i.e., that the default destination for membrane proteins in yeast is the vacuole (Nothwehr and Stevens, 1994).
Protein Sorting to the Yeast Vacuole
127
Another interesting aspect of the delivery of membrane proteins to the vacuole concerns the vacuolar membrane H"^-ATPase (V-ATPase). The V-ATPase is a multi-subunit enzyme responsible for acidification of the vacuolar compartment (Anraku et al., 1992b). Considerable effort has been directed at purification and identification of the individual subunits of the ATPase (Kane et al., 1992; Kane and Stevens, 1992). The V-ATPase is composed of at least eight different subunits ranging in size from 17 to 100 kDa. The genes encoding seven of these have been cloned and sequenced. The subunits of the V-ATPase can be grouped into peripheral membrane proteins (V, complex) and integral membrane proteins (VQ complex). As such, the V-ATPase represents an interesting problem from the perspective of macromolecular assembly. Presumably, there must be coordination of synthesis of integral and peripheral membrane components, and accessory factors have been described that appear to be necessary for the efficient assembly or full activity of the V-ATPase (Kane et al., 1992; Bauerle et al., 1993; Ho et al., 1993a,b). Determination of the spatial and temporal aspects of V-ATPase assembly and the factors regulating it, in addition to the targeting of the enzyme to the vacuolar membrane, will be important developments. Clearly, resolution of the issue of default vs. signal-mediated delivery of membrane proteins to the vacuole will have profound ramifications for targeting the V-ATPase to the vacuole membrane.
111. VACUOLAR PROTEIN SORTING MUTANTS Studies with CPY and PrA clearly demonstrate the existence of sequence information within the propeptide of each of these proteins that serves in targeting to the vacuole. Alteration of the vacuolar sorting signal in the wild-type proteins or in fusions with invertase results in their appearance at the cell surface (see Sorting Signals for Soluble Vacuolar Proteins). The mislocalization of vacuolar proteins to the cell surface is the basis of several genetic selections that have resulted in the isolation of a large number of mutants specifically defective in the delivery of proteins to the vacuole. Genetic selections using CPY-invertase fusions led to the isolation of vpt (vacuolar protein targeting defective) mutants. A selection scheme based on detection of secreted CPY enzyme activity identified numerous vpl (vacuolar protein localization defective) mutants. These mutants are collectively referred to as vacuolar protein sorting defective, or vps mutants. Screens for yeast mutants defective in protease activity in the vacuole (pep mutants) have also identified mutants in the vacuolar protein sorting pathway. Complementation analyses of the vpt, vpl, and pep mutant collections determined that there was extensive genetic overlap between these mutants; they collectively define >40 complementation groups (Rothman et al., 1989; Raymond et al., 1992). This large number of genes indicates that protein sorting to the vacuole is a complex process. Possible molecular activities involved in vacuolar protein sorting include 1) receptor(s) for vacuolar protein precursors, 2) protein complexes for the recognition/segregation of receptor-ligand complexes and packaging into vesicular carriers, 3)
128
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
factors responsible for vesicle budding and membrane fusion, 4) vesicle coat proteins (clathrin and non-clathrin coat proteins), 5) proteins involved in targeting to and recognition of target organelles such as the endosome and vacuole, 6) receptor recycling mechanisms, and 7) proteins involved in the biogenesis and maintenance of the vacuole. A. vpt Mutants Extracellular invertase activity is required for yeast to grow on media containing sucrose as their sole fermentable carbon source. Invertase cleaves sucrose to fructose and glucose, which can then be taken up and utilized by the cell. As described earlier, a fusion protein containing the NH2-terminal domain of CPY, including its vacuolar targeting signal, and normally secreted invertase is efficiently targeted to the vacuole. Yeast cells containing this CPY-Inv fusion as their only form of invertase are therefore unable to grow on sucrose because of the sequestration of invertase activity in the vacuole (Suc~). The fact that mutations in the sorting signal for wild-type CPY or the CPY-Inv fusion result in its appearance at the cell surface led to a positive selection for mutants defective in the delivery of the CPY-Inv fusion to the vacuole. Disruption of the Golgi-to-vacuole step in this sorting pathway results in the mislocalization of the CPY-Inv fusion to the cell surface, and this provides the extracellular invertase activity required for growth on sucrose-containing media (Suc"^). Selection for spontaneous Sue"*" mutants in strains containing the CPY-Inv fusion allowed the isolation of >600 mutants (Bankaitis et al., 1986; Robinson et al., 1988). Subsequent enzymatic assays confirmed that each of the mutants was indeed secreting invertase activity. In addition, assays for a cytoplasmic marker enzyme demonstrated that extracellular invertase activity in the mutants was not simply the result of cell lysis. The recessive vpt mutants were subjected to complementation analysis and were found to define 33 complementation groups (Robinson et al., 1988). Further analysis of representatives of each of the complementation groups confirmed that they were also defective in the delivery of wild-type CPY to the vacuole. Pulse-chase labeling experiments followed by immunoprecipitation of CPY demonstrated that the vpt mutants missorted and secreted CPY as the Golgi-modified p2 precursor form (Robinson et al., 1988). These data confirmed that the CPY-Inv fusion protein and wild-type CPY were missorted in an identical fashion in vpt mutant cells. Examination of two other soluble vacuolar hydrolases, PrA and PrB, showed that they were missorted and secreted in most vpt mutant strains in a similar manner to that of CPY. Analysis of vacuolar membrane proteins suggested that they may not be transported to the vacuole in a FPr-dependent manner. Mislocalization of the vacuolar membrane protein a-mannosidase to the cell surface would result in extracellular enzyme activity. The majority of vpt mutants do not exhibit a significant increase in external a-mannosidase activity compared to a wild-type strain. The vpt strains
Protein Sorting to the Yeast Vacuole
129
that do show an increase in external a-mannosidase activity include those mutants that exhibit severe morphological changes in vacuole structure (see below), which may suggest that an intact vacuole is required for vacuolar a-mannosidase localization. As it has also been recently suggested that a-mannosidase may enter the vacuole directly from the cytoplasm (Yoshihisa and Anraku, 1990), the lack of FPr-dependent delivery may not be surprising. Another vacuolar membrane protein, alkaline phosphatase (ALP), has been shown to follow the secretory pathway together with soluble vacuolar proteins (Klionsky and Emr, 1989). Analysis of the maturation of ALP in vpt mutants also indicated that mislocalization of ALP is much less sensitive to mutations in the VPT genes than are soluble vacuolar hydrolases (Robinson et al., 1988). Collectively, these data suggest that the mechanisms of delivery of soluble and membrane vacuolar proteins may be distinct and further implicate the majority of the VPS gene products specifically in the delivery of soluble vacuolar proteins. B. vpl Mutants
Overproduction of CPY results in secretion of a portion of p2CPY into the periplasm, where it is converted to an enzymatically active form by an undefined periplasmic protease(s). Selection for external CPY activity allowed the isolation of vpl mutants. CPY will cleave the peptide bond of the dipeptide A^-carbobenzoxyL-phenylalanine-L-leucine (CBZ-PheLeu) to release leucine. Strains auxotrophic for leucine are able to grow on media containing CBZ-PheLeu as the sole source of leucine in the presence of secreted CPY activity. This selection scheme led to the isolation of-600 vpl mutants that were categorized into 19 complementation groups (Rothman and Stevens, 1986). Pulse-chase immunoprecipitation experiments found that the vpl mutants mislocalized significant amounts of CPY to the cell surface. The mislocalized form of CPY was found to be the Golgi-modified p2 precursor. All vpl mutants secreted multiple soluble vacuolar hydrolases, demonstrating the pleiotropic nature of these mutants as well. Secretion of p2CPY in vpl mutants was also shown to require the action of SEC gene products, which act between the Golgi and cell surface (Rothman and Stevens, 1986). In vpl sec J double mutants incubated at the nonpermissive temperature, p2CPY accumulated inside the cell, presumably within secretory vesicles that are known to accumulate in sec I mutants. These data indicate that missorted CPY utilizes the secretory pathway in a manner identical to that of secreted proteins. In addition, it was shown that the secretion of invertase in vpl mutants was very similar to that of wild-type cells, indicating that the vpl mutants do not significantly affect portions of the secretory pathway prior to the diversion of vacuolar proteins. C. pep Mutants
Yeast mutants showing reduced amounts of carboxypeptidase Y enzyme activity were isolated and designated eispep mutants (Jones, 1977, 1984). Genetic analysis
Table 1. Yeast Genes Involved in Vacuolar Protein Sorting VPS Designation
Other Designation
Vacuole Predicted Morphology Molecular Mass Subcellular Location and (Daltons) Association (Class)
VPSl
Soluble and membrane
Cytoplasmic Soluble and peripheral membrane Membrane
Membrane Vacuole surface Peripheral membrane
93,000
Particulate
Comments GTP-binding protein, dynamin homologue
Acidic phosphoprotein, associates with Vpsl7p Synataxin homologue
Ras activating protein homologue Type I integral membrane CPY sorting receptor ATP binding domain, Zn finger motif Myristolated serlthr protein kinase, autophosphorylated, associates and activates vps34p Part of a protein complex
References Rothman et al., 1990; Yeh et al., 1991; Vater et al., 1992 Davis et al., 1993 Raymond et al., 1990 B. Horazdovsky and S. Emr, unpublished E. Jones, unpublished B. Horazdovsky and S. Emr, unpublished C. Burd and S. Emr, unpublished Marcusson et al., 1994 Dulic and Riezman, 1989; Woolford et a]., 1990 Herman et al., 1991a; Stack et al., 1993, 1995; Stack and Emr, 1994
Horazdovsky and Emr, 1993
PEP3
C
VACI, PEP7 -
D D
-
E E F
PEP8 -
Soluble and peripheral membrane Soluble and peripheral membrane Soluble and membrane
Phosphoprotein associated with Vps5p CAMPbinding domain, Zn finger motif Zn finger motif Small GTP-binding protein of the rab family -
Peripheral vacuole membrane Soluble
-
E A C
-
D
-
A
Soluble and peripheral membrane Peripheral membrane
SSTlO
D
Peripheral mem brane
Robinson et al., 1991 Weisman and Wickner, 1992 Horazdovsky et al., 1994 B. Wendland and S. Em, unpublished
B. Wendland and S. Em, unpublished Mouse protein homologue
Bachhawat et al., 1994 L. Banta and S. Emr, unpublished
-
Soluble
Kohrer and Emr, 1993
J. Cereghino and S. Emr, unpublished Member of the Secl protein family PI3-kinase associated with v p s l Sp Mutation results in a CPYspecific sorting defect Member of the Secl protein family
Banta et al., 1990 Herman and Emr, 1990; Schu et al., 1993; Stack et al., 1993 Paravicini et al., 1992 Cowles et al., 1994
132
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
of recessive/7^p mutants demonstrated that they defined 17 complementation groups. These mutants showed pleiotropic defects in several vacuolar proteolytic activities, including CPY, PrA, and PrB. There appear to be two distinct subclasses of pep mutants. One class is represented by the pep4 mutant, which contains mutations in the structural gene for PrA (Rothman et al., 1989). The fact that PrA is involved in the proteolytic maturation of several other vacuolar hydrolases, including CPY and PrB, accounts for the pleiotropic nature of pep4 mutants. Analysis of other/7e/7 mutants showed that they had characteristics very similar to those of vpl and vpt mutants; most pep mutants secrete p2CPY and proPrA from the cell. Complementation analysis between the pep, vpt, and vpl mutants revealed substantial genetic overlap (Rothman et al., 1989). Therefore, it is likely that a large number of pep mutants are defective in vacuolar hydrolase activity as the result of missorting and secretion of vacuolar protein precursors in a manner identical to that of the vps mutants. The genes and gene products affected in many of the vps and pep mutants have been characterized and are summarized in Table 1. D. Other Mutants
The yeast vacuole appears to play a cellular homeostatic role under a variety of growth conditions. The multiple functions of the vacuole are reflected in the pleiotropic defects observed in mutants lacking vacuolar function. Mutants defective in the delivery of proteins to the vacuole have been isolated in screens for maturation of yeast mating pheromone (lam; Wilsbach and Payne, 1993), yeast genes required for sporulation (spo; Yeh et al., 1991), endocytosis (end, ren; Chvatchko et al., 1986; Davis et al., 1993), normal vacuolar morphology (vam; Anraku et al., 1992a,b), and resistance to high concentrations of calcium (els; Kitamoto et al., 1988) or lysine (sip; Ohya et al., 1986). In addition, a number of vps mutants are unable to grow at elevated temperature. Such strains are temperature sensitive (ts) for growth at 37°C and presumably reflect a requirement for vacuolar functions at high temperature, as these strains missort vacuolar proteins at both the permissive and nonpermissive growth temperatures. Several vps strains also exhibit sensitivity to osmotic stress; they are unable to grow on 1.5 M NaCl plates (Banta et al., 1988). Osmotically sensitive vps strains are among the most defective mutants for vacuolar protein sorting and underscore the importance of the vacuole in responding to environmental stress.
IV. VACUOLE MORPHOLOGY AND BIOGENESIS The morphological aspects of the yeast vacuole have been the subject of a number of studies. These include analyses of vacuolar morphology under a variety of growth conditions and in a number of mutant strains at both the light and electron microscopic levels. Vacuole inheritance, the segregation of vacuolar material during cell division, has also received attention recently. Morphological issues are
Protein Sorting to the Yeast Vacuole
133
relevant to vacuolar protein sorting because efficient delivery to the vacuole requires the presence of an organelle that is competent to serve as a target for delivery. Indeed, a number of mutants isolated as being defective in vacuolar protein sorting have been found to be defective in aspects of vacuole morphology and biogenesis and vice versa. It is likely that factors regulating the formation and maintenance of a vacuolar structure may be required for efficient delivery of enzymes to the vacuole. A. Vacuole Morphology
A number of techniques, including the use of vital dyes that accumulate in the vacuole, have revealed that in wild-type yeast cells the vacuole is a large, prominent structure consisting of one to three organelles per cell (Roberts et al., 1991; Raymond et al., 1992). Time-lapse video microscopy using differential interference contrast optics to visualize organelles showed that there is interconversion between one large vacuole and several small vacuoles (Jones et al., 1993). Banta et al. (1988) performed an initial morphological characterization of the vps mutants using light and electron microscopy. It was found that the vps mutants could be categorized into three general classes based on vacuole morphology. Class A mutants contained a wild-type or nearly wild-type-appearing vacuole. Most of the vps mutants fell into this category. The vacuoles of most class A vps mutants could also be stained with quinacrine, which indicates that the vacuole has been acidified. The phenotype of class A vps mutants suggests that they are competent for the assembly and maintenance of a morphologically normal vacuole. The fact that many of the class A mutants are very defective for the sorting of multiple soluble vacuolar hydrolases further suggests that soluble and membrane vacuolar constituents may utilize different transport pathways to the vacuole. The class A mutants may therefore represent mutants defective in specific functions required for the delivery of soluble vacuolar proteins rather than components required for vacuole biogenesis. Mutants in the complementation groups comprising the class B category exhibited gross alterations in vacuolar structure. In contrast to the 1—3 large vacuoles found in wild-type cells and class A mutants, class B vps mutants contained highly fi'agmented vacuoles with each cell containing -35—40 small vacuoles. These fragmented vacuoles accumulated both vital dyes and quinacrine, suggesting that they may retain some vacuolar function (Banta et al., 1988; Raymond et al., 1992). The class B vacuoles are not recognized as completely fiinctional targets for the delivery of soluble vacuolar proteins, as class B mutants are very defective for the sorting of CPY (Robinson et al., 1988). The small vacuoles observed in class B mutants may represent intermediates in the assembly of a wild-type vacuole, possibly analogous to the proposed maturation of endosomes into lysosomes in mammalian cells. Alternatively, the class B mutants may lack a factor(s) required for maintaining vacuolar structure, and this results in fragmentation of the vacuole. The class C vps mutants lack any identifiable vacuolar structure. In addition to the inability
134
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
of class C mutants to accumulate dyes in a vacuole-like structure, electron microscopic analysis of class C mutants demonstrated the absence of a vacuole. These mutants also accumulated abnormal membrane-enclosed structures, including vesicles, multilamellar structures, and Berkeley body structures that have been observed to accumulate in certain sec mutants (Banta et al., 1988). These abnormal intracellular structures are most likely the consequence of abnormal membrane flow to the vacuole that has affected the secretory pathway combined with lack of ability to assemble a vacuole. Indeed, one of the class C mutants, vpsl8, is defective for late Golgi functions in addition to vacuolar functions (Robinson et al., 1991). Consistent with the complete absence of a vacuole, the class C mutants are among the most severely affected mutants in the vps collection. In addition to being extremely defective for sorting of vacuolar proteins, the class C mutants also exhibit growth defects under conditions of environmental stress; they are temperature-sensitive for growth, sensitive to osmotic stress, and grow much more slowly than wild-type strains under optimal nutrient conditions. Because these mutants almost certainly lack any vacuolar function, these phenotypes may reflect a need for a functional vacuole under conditions of environmental stress. The fact that no VPS gene yet characterized is essential when deleted is consistent with this notion and underscores the nonessential nature of the vacuole. The class C mutants are good candidates for factors involved in vacuolar biogenesis. As such, the vacuolar protein sorting defect observed in class C mutants may be the consequence of the lack of a functional acceptor organelle. Alternatively, the class C (and class B) mutants may indeed be defective in the sorting of a vacuolar component critical for organelle integrity. A more detailed analysis of the role of the gene products affected in these mutants is needed to differentiate between these general models. Raymond et al. (1992) have conducted a detailed morphological re-examination of the vps mutants using antibodies raised against the vacuolar integral membrane protein ALP and the peripheral membrane 60-kDa subunit of the vacuolar H"^-ATPase. They confirmed many of the classifications of Banta et al., and further subdivided the original class A mutants into several categories based on morphological criteria. The analyses using antisera to ALP and the H'^-ATPase revealed that many of the original class A mutants (characterized by Banta et al., as having wild-type appearing vacuoles) are actually defective to some degree in aspects of vacuolar acidification, ATPase assembly, vacuole segregation at mitosis, or some vacuolar morphological feature (Raymond et al., 1992). The revised classification scheme consists of six categories, classes A—F. The characteristics of mutants in classes A, B, and C are unchanged from those of Banta et al.: class A vps mutants possess a morphologically wild-type vacuole, class B mutants have fragmented vacuoles, and class C mutants have no readily identifiable vacuole. The class D mutants have a normal-appearing (although slightly enlarged) vacuole; however, they are defective in mother-to-daughter vacuole inheritance and do not properly assemble the vacuolar H'^-ATPase. Class F vps mutants have a large central vacuole surrounded by
Protein Sorting to the Yeast Vacuole
135
smaller class B-like fragmented vacuolar structures. The vacuole in class F mutants stains with ALP and ATPase antisera and accumulates quinacrine; however, mutants in this class are defective for the sorting of soluble vacuolar hydrolases. The most interesting observation by Raymond et al. is that class E vps mutants appear to accumulate a novel organelle distinct from the vacuole (Raymond et al., 1992). This structure is stained by antibodies specific for ATPase but not ALP and accumulates quinacrine. In addition, the novel class E compartment also contains significant amounts of soluble vacuolar hydrolases, as indicated by its staining with antibodies to CPY, PrA, and PrB. These characteristics suggest that the class E compartment may represent an exaggerated prevacuolar endosome-like organelle similar to that identified by Vida et al. (1993) using biochemical techniques. Interestingly, this novel compartment also stains with antibodies specific for a late Golgi membrane protein (Raymond et al., 1992). This suggests that resident late Golgi integral membrane proteins may normally traffic to this prevacuolar compartment and would presumably be recycled back to the Golgi in wild-type cells. Further evidence for an endosomal-like function for the class E compartment comes from the observation that the cell surface a-factor receptor, Ste3p, which is normally endocytosed and degraded in the vacuole, accumulates in this compartment in a class E vps mutant (Davis et al., 1993). This is consistent with the view of an endosomal pathway described for mammalian cells in which endocytosed material from the plasma membrane is initially delivered to an early endosome and then to a late endosome, where it meets traffic from the biosynthetic route (Fig. 1) (Vida et al., 1993). Proteins destined for the vacuole from both the biosynthetic and endocytic pathways then move from the late endosome to the vacuole. In such a pathway, the class E prevacuolar organelle would correspond to a late endosomal compartment. The class E category contains 13 vps complementation groups (Raymond et al., 1992), indicating that delivery from this putative prevacuolar intermediate to the vacuole may be a complicated process. B. Vacuole Biogenesis The organelles in a cell must be faithfully segregated between mother and daughter cells during cell division. The yeast vacuole is segregated in an active process that occurs very early following bud emergence. Vacuoles can be detected in buds that are as small as one-tenth the size of the mother cell (Weisman et al., 1987). Fluorescence microscopy using endogenous vacuolar dyes demonstrated that vacuolar material from the mother cell is transferred to the daughter cell in the form of small membranous vesicles or tubules (Weisman and Wickner, 1988). These "segregation structures" form prior to nuclear migration upon bud emergence; time lapse photomicrography determined that within 15 minutes of bud emergence, 80% of the cells contained vacuolar segregation structures (Gomes de Mesquita et al., 1991). The concept of mother-to-daughter transfer of vacuolar
136
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
material (vacuole inheritance) is well illustrated by the phenotypic lag observed when PEP4/pep4 heterozygotes are sporulated. Haploid progeny that are genotypic^\\ypep4 exhibit a wild-type phenotype with respect to activity of the PEPA gene product, PrA (Jones et al., 1982). The wild-type P^P^ phenotype persists for several generations before a combination of dilution and turnover results in loss of PrA activity and acquisition of the pep4 mutant phenotype. Mutants defective in vacuole inheritance have been isolated and described (Weisman et al., 1990). Despite a defect in mother-to-daughter vacuole inheritance in these vac mutants, most of the daughter cells do contain a vacuole by the time of cell division. This suggests that a mechanism for de novo biosynthesis of a vacuole must exist that functions in concert with or in the absence of vacuolar inheritance from the mother cell. The concept ofde novo vacuole biosynthesis was also demonstrated using class C vps mutants that contain no observable vacuolar structure. If two class C mutants from different complementation groups are mated, vacuoles are observed in each of the conjugating cells of the zygote (Banta et al., 1988). These data indicate that the formation of a vacuole can occur in the absence of a normal template organelle. Unlike other vac mutants, which missort vacuolar proteins and are allelic to known vps mutants, the vac2 mutant appears to be selectively defective in vacuole segregation (Shaw and Wickner, 1991). The vac2-J strain was isolated from a collection of yeast ts mutants and has a temperature-conditional vacuole segregation defect; vac2-l cells grown at the permissive temperature have wild-type vacuole morphology, whereas a significant fraction of vacl-l cells grown at the nonpermissive temperature for 3 hours do not contain a vacuole in unbudded cells or in large buds. CPY and PrA are matured normally in vac2-l cells at the permissive and nonpermissive growth temperatures. These data suggest the existence of a vacuole inheritance/segregation pathway that has components distinct from those involved in the proper localization of vacuolar proteins. The vacuolar segregation defect observed in many vps mutants may reflect the need for crucial component(s) to be delivered to the vacuole for efficient vacuole inheritance. Because many of the vps mutants affect vacuolar morphology, it is not surprising that some are also defective in vacuole segregation at mitosis. As indicated earlier, one of the criteria for characterization of the class D vps mutants is a defect in vacuolar inheritance. The eight complementation groups in this class suggest that the sorting of vacuolar proteins may affect this process. Indeed, the pleiotropic nature of vps mutants is emphasized by the finding that nearly all of the vac mutants are allelic to existing vps mutants and again suggests that the sorting of a critical component(s) is important for mother-to-daughter vacuolar inheritance. The recent development of an in vitro reconstitution of vacuole inheritance may allow for the biochemical characterization of factors required for this process (Conrad et al., 1992, 1994; Haas etal., 1994).
Protein Sorting to the Yeast Vacuole
137
V. RECEPTOR-MEDIATED VACUOLAR PROTEIN SORTING In contrast to the mannose-6-phosphate recognition system for soluble lysosomal proteins, the receptor-mediated sorting of soluble vacuolar proteins appears to be independent of carbohydrate modification and instead is based on sorting signals found in the polypeptide sequence of vacuolar proteins (see Sorting Signals for Soluble Vacuolar Proteins). Numerous early observations indicated that a sorting receptor is utilized in the delivery of soluble hydrolases to the vacuole in yeast. The existence of a sorting receptor for vacuolar proteins was initially suggested by the observation that overproduction of a CPY-invertase fusion protein or CPY itself resulted in the missorting and secretion of a substantial proportion of these proteins (Bankaitis et al., 1986; Stevens et al., 1986; Johnson et al., 1987). One explanation for these results is that overproduction of CPY leads to saturation of a limiting component, possibly a receptor protein, required for the recognition and/or sorting of CPY. A similar result was obtained when wild-type PrA was overexpressed (Rothman et al., 1986). Interestingly, overproduction of PrA does not lead to the missorting and secretion of CPY (Rothman et al., 1986), and overproduction of CPY does not result in secretion of PrA (Stevens et al., 1986). These data suggest that CPY and PrA may utilize different receptors, which is consistent with the lack of significant homology among their sorting signals. A. The CPY Sorting Receptor
The apparent difference in sorting mechanisms between CPY and PrA was utilized to isolate a receptor protein for CPY. Mutations in the VPS 10 gene result in a CPY-specific sorting defect; PrA, PrB, and the membrane protein ALP are delivered normally to the vacuole while CPY is missorted and secreted in a AvpsIO strain (Marcusson et al., 1994). Characterization of the VpslO protein revealed that it has the structural and functional features expected of a transmembrane receptor protein. It is a -180 kDa type I integral membrane protein that contains two hydrophobic regions, one at its NH2 terminus (21 amino acids) encoding a signal sequence and a longer COOH-terminal hydrophobic region (25 amino acids) that serves as a transmembrane domain (Marcusson et al., 1994). VpslOp primarily localizes to a late Golgi compartment (Marcusson et al., 1994), the site where sorting of vacuolar proteins appears to take place (Graham and Emr, 1991). In addition, immunoprecipitation with carbohydrate-specific antisera has shown that VpslOp is modified by mannosyl residues added by glycosyltransferases present in the most distal compartment of the Golgi complex, further indicating a Golgi localization for VpslOp. Using chemical cross-linking, VpslOp has been shown to specifically interact with the Golgi-modified p2 form of CPY; VpslOp could not be cross-linked to the ER (pi) or vacuolar (mCPY) forms of CPY. As an important control, VpslOp was unable to interact with a mutant form of CPY that is defective for sorting to the vacuole (Marcusson et al., 1994). This mutant CPY protein is
138
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
derived from a single point mutation in the gene that encodes CPY, which alters a single amino acid residue within the previously characterized sorting signal found in proCPY (Vails et al., 1990). Furthermore, VpslOp could also be cross-linked to a CPY-invertase fusion protein that contains the NH2-terminal 50 amino acids of CPY (the 20-residue signal sequence and 30 residues of the prosegment), which is correctly delivered to the vacuole (Marcusson et al, 1994). These data indicate that VpslOp specifically binds the Golgi-modified p2 form of CPY and this interaction is dependent on the presence of an intact CPY vacuolar protein sorting signal. It is likely that vacuolar protein receptors in yeast function in a manner analogous to that of the mannose-6-phosphate receptors in mammalian cells (Fig. 2). Vacuolar proteins are presumably bound to receptors in the late Golgi sorting compartment and delivered via vesicular carriers to a prevacuolar endosomal compartment. In the endosome, ligand would dissociate from the receptor because of the chemical characteristics (such as pH) of the compartment, then ligand would continue on to the vacuole and the receptor would recycle back to the Golgi for additional rounds of sorting. In such a scenario, multiple components would be required for the efficient function of the receptor. Interestingly, /^vps35 and tsyps29 strains exhibit a CPY-specific sorting defect similar to that of a t^vpslO strain and may represent
(4)
Late Endosome
Late Golgi (Kex2p)
Vacuole
¥\gure 2. A model of CPY sorting to the yeast vacuole. In a late Golgi (Kex2p) compartment, p2CPY binds to the Vpsl Op receptor. (1) A signal is transduced to the Vps15p-Vps34p complex, resulting in the packaging of p2CPY-Vps10p complexes into transport vesicles. (2) These vesicles are then packaged and targeted to an endosomal compartment. This targeting event appears to require Vps21 p and Vps45p function. (3) At the endosome the receptor releases p2CPY and recycles back to the Golgi for another round of sorting. (4) p2CPY continues on to the vacuole, where it is matured to its active form.
Protein Sorting to the Yeast Vacuole
139
components that specifically interact with Vps 1 Op to facilitate CP Y sorting (Paravicini et al., 1992; E. Marcusson, unpublished observations). Other components involved in the trafficking of receptor molecules may interact with the cytoplasmic tail of receptors such as Vps 1 Op. Indeed, deletion of the cytoplasmic tail of Vps 1 Op results in the missorting and secretion of CPY (E. Marcusson, unpublished observations). Candidates for factors binding to the tail of receptors include coat proteins such as clathrin and clathrin-associated adaptor proteins Clathrin has recently been shown to be required for the sorting of soluble vacuolar proteins, including CPY (Seeger and Payne, 1992a). The cytoplasmic tail of Vps 1 Op contains several tyrosine residues that may interact with clathrin adaptor complexes to facilitate the assembly of a clathrin coat in a manner similar to that of the tyr-based signals present in the mammalian mannose-6-phosphate receptor. In addition to clarifying the issue of vesicle coats, analyses of tail mutants of Vps 1 Op may reveal factors involved in the segregation and packaging of receptor-ligand complexes.
VI. A PROTEIN KINASE AND A PHOSPHATIDYLINOSITOL 3-KINASE ARE REQUIRED FOR VACUOLAR PROTEIN SORTING Strains deleted for either the VPS 15 or the VPS34 gene exhibit a set of common phenotypes, suggesting that their products may act at a similar step in the vacuolar protein sorting pathway. These phenotypes include severe defects in the delivery of multiple soluble vacuolar hydrolases, a temperature-sensitive growth defect, and defects in osmoregulation and in vacuole segregation at mitosis (Herman et al., 1992). Genetic and biochemical analyses have demonstrated that the VPS 15 and VPS34 genes encode a serine/threonine protein kinase and a phosphatidylinositol3-kinase, respectively, that form a membrane-associated complex that is required for the sorting of soluble vacuolar proteins. These findings suggest that protein and phospholipid phosphorylation events are involved in the vesicular delivery of proteins to the vacuole. A. The Vpsi 5 Protein Kinase
The VPSI5 gene encodes a 1455-amino acid protein whose NH2-terminal 300 residues exhibit significant sequence similarity to the serine/threonine family of protein kinases. Vpsl5p appears to be an active protein kinase, inasmuch as in vivo and in vitro labeling experiments have demonstrated that it is able to catalyze an autophosphorylation reaction. This reaction requires an intact Vpsl5p protein kinase domain because mutations altering highly conserved residues in the kinase domain of Vps 15p result in severe defects in the autophosphorylation reaction both in vivo and in vitro (Herman et al., 1991a,b; Stack and Emr, 1994). In addition, it was found that mutations that eliminated Vpsl 5p autophosphorylation also resulted in the missorting and secretion of CPY as the Golgi-modified p2 precursor (Herman
140
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
et al., 1991a,b). Mutations that alter residues in VpslSp that are less conserved among protein kinases have correspondingly less severe phenotypes (Herman et al., 1991b). The correlation between VpslSp catalytic activity and CPY sorting indicates that Vpsl5p protein kinase activity plays a role in the sorting of soluble vacuolar proteins. Analysis of a temperature-sensitive (ts) vpsl5 allele has also provided evidence for a direct role of Vpsl5p in the sorting of soluble vacuolar hydrolases (Herman et al., 1991b). This vpsl5 allele (vpsI5AC30) encodes a truncated Vpsl5 protein that lacks the COOH-terminal 30 amino acids of the wild-type protein. Yeast vpsI5AC30 mutants exhibit a severe temperature-conditional defect in vacuolar delivery of the soluble hydrolases CPY and PrA. The defect is apparently specific for soluble vacuolar hydrolases inasmuch as maturation (and presumably vacuolar delivery) of the vacuolar membrane protein ALP proceeds with essentially wildtype kinetics at the restrictive temperature. The very rapid onset and relative specificity of the ts sorting defects observed in the vpsI5AC30 mutant strongly suggest that the primary role of Vpsl5p in yeast cells is to mediate the delivery of soluble vacuolar proteins. Interestingly, CPY is not secreted from vpslSACSO cells immediately after shifting to the restrictive temperature. CPY instead accumulates as the p2 precursor within a compartment distinct from the vacuole (Herman et al., 1991b). Upon reversal of the temperature block, the intracellular p2CPY is efficiently processed to its mature form, suggesting that it has been delivered to the vacuole. The blocked p2CPY may therefore reside within a normal intermediate of the vacuolar protein sorting pathway. Analysis of this accumulated intracellular p2CPY using sucrose density gradients has indicated that it cofractionates with Kex2p, a resident late Golgi membrane protein (Vida et al., 1993). These data suggest that the loss of Vps 15pfiinctionleads to an immediate block in vacuolar protein sorting at the level of the Golgi. Because the late Golgi has been shown to be the site where the sorting of vacuolar proteins occurs (Graham and Emr, 1991), this suggests the direct involvement of VpslSp in regulating early events in the vacuolar protein sorting pathway. B. The Vps34 Phosphatidylinositol 3-Kinase
Characterization of the VPS34 gene revealed that it encodes an 875-amino acid protein required for the sorting of vacuolar proteins and the partitioning of the vacuolar compartment during cell division (Herman and Emr, 1990). A possible biochemical activity for Vps34p was suggested by a sequence comparison to the pi 10 catalytic subunit of mammalian 3-kinase (PI 3-kinase), which showed it to be highly homologous to Vps34p (Hiles et al., 1992). In mammalian cells, PI 3-kinase phosphorylates phosphatidylinositol (Ptdlns) and other more highly phosphorylated phosphoinositides [PtdIns(4)P and Ptdlns(4,5)2] at the D-3 position of the inositol ring. Mammalian PI 3-kinase associates with many signal transducing
Protein Sorting to the Yeast Vacuole
.
141
receptor tyrosine kinases and is postulated to be involved in the generation of key second messenger molecules important for regulating cell growth and proliferation (Cantley et al., 1991). Mammalian PI 3-kinase is composed of two subunits: an 85-kDa targeting subunit and the 110-kDa catalytic subunit (Escobedo et al., 1991; Otsu et al., 1991; Skolnik et al., 1991) The sequence of the pi 10 subunit of bovine PI 3-kinase is 33% identical and 55% similar to Vps34p over a stretch of approximately 450 amino acids that constitutes the COOH-terminal half of each protein (Hiles et al., 1992). This region of homology includes conserved residues that are also shared with the catalytic domain of protein kinases and may therefore indicate a common ancestry between protein and lipid kinases. The possibility that Vps34p may be a lipid kinase suggested that covalent modification of specific membrane phospholipids may be involved in regulating the delivery of proteins to the yeast vacuole. S. cerevisiae has been shown to contain Ptdlns 3-kinase activity (Auger et al., 1989a), and strains deleted for the VPS34 gene are extremely defective for this activity (Schu et al., 1993). Alteration of conserved residues in the lipid kinase domain of Vps34p results in severe defects in both Ptdlns 3-kinase activity and vacuolar protein sorting (Schu et al., 1993). These results indicate that Vps34p is a functional Ptdlns 3-kinase and that this activity is important for vacuolar protein sorting. In addition, analysis of a temperature-conditional allele of VPS34 has demonstrated that Vps34p is directly involved in the sorting of soluble vacuolar proteins. A shift of these vps34 ts cells to the nonpermissive temperature leads to an immediate block in CPY sorting (Stack et al., 1995). The vacuolar membrane protein alkaline phosphatase is matured normally in the vps34 ts strain incubated at the nonpermissive temperature, indicating that, like Vpsl5p, inactivation of Vps34p leads to a selective block in the sorting of soluble vacuolar proteins. This rapid block in protein sorting appears to be the result of loss of Ptdlns 3-kinase activity, because the vps34 ts strain shifted to the nonpermissive temperature is defective for this activity. These results indicate that the vps34 ts strain is temperature-sensitive for both CPY sorting and Ptdlns 3-kinase activity and directly implicates Vps34p and Ptdlns 3-kinase activity in the delivery of proteins to the vacuole. Biochemical characterization has shown that, unlike mammalian pi 10, Vps34p is only able to utilize Ptdlns, as a substrate and is inactive toward PtdIns(4)P and PtdIns(4,5)P2 (Stack and Emr, 1994). These results may have important implications for the role(s) of PI 3-kinase in mammalian cells. The products of mammalian PI 3-kinase [PtdIns(3)P, PtdIns(3,4)P2, PtdIns(3,4,5)P3] can be detected in cells labeled with [^H]inositol (Auger et al., 1989b). The notion that these phosphoinositides have different effects in vivo is suggested by the fact that formation of PtdIns(3,4)P2 and PtdIns(3,4,5)P3 is stimulated by growth factor addition, whereas PtdIns(3)P levels remain relatively constant (Auger et al., 1989b). It has been suggested that PtdIns(3,4)P2 and PtdIns(3,4,5)P3 formed by PI 3-kinase may act as intracellular second messengers to signal cell proliferation in response to growth
142
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
factor stimulation (Auger et aL, 1989b; Cantley et al., 1991; Fantl et al., 1992). The possibility that the different 3'-phosphorylated phosphoinositides may have distinct functional properties in mammalian cells is supported by recent reports of multiple PI 3-kinase activities with very different substrate specificities (see below). These distinct PI 3-kinase activities imply that multiple PI 3-kinase enzymes with different functional properties exist in mammalian cells. The role for Vps34p in vacuolar protein sorting and its characterization as a Ptdlns-specific 3-kinase led to the proposal that the production of PtdIns(3)P is specifically involved in regulating intracellular protein sorting pathways (Stack and Emr, 1994). Vps34p is highly resistant to inhibitors of mammalian PI 3-kinase such as wortmannin and LY294002 (Stack and Emr, 1994). The substrate specificity, wortmannin resistance, and other biochemical characteristics of its Ptdlns 3-kinase activity suggest that Vps34p may be similar to a Ptdlns-specific 3-kinase activity recently characterized from mammalian cells (Stack and Emr, 1994; Stephens et al., 1994). This may indicate that this mammalian Ptdlns 3-kinase is the functional homologue of Vps34p or that it is a member of a family of Vps34p-like mammalian PI 3-kinases. A PI 3-kinase may function in mammalian cells to regulate the sorting of lysosomal proteins in a manner directly analogous to the function of Vps34p. One can speculate that Pl-specific 3-kinases may function in regulating intracellular protein trafficking rather than in signaling cell proliferation because of their substrate specificity. It will be interesting to learn the intracellular distribution of mammalian PI 3-kinases, and the cloning of such molecules may reveal a protein more closely related to Vps34p than to pi 10. The identification of Vps34p as a yeast Ptdlns 3-kinase and mutational analysis of Vps34p (Schu et al., 1993) indicate that PI 3-kinase activity regulates membrane vesicle trafficking in yeast and suggest the possibility that PI 3-kinase activity may function similarly in mammalian cells. The mammalian pllO/p85 PI 3-kinase is known to associate with activated receptor tyrosine kinases (Cantley et al., 1991). A recent report of a PI 3-kinase activity distinct from pi 10/p85 that is activated by G protein Py subunits (Stephens et al., 1994; Thomason et al., 1994) indicates that PI 3-kinases may associate with numerous types of cell surface receptor proteins. PI 3-kinase activity in mammalian cells may be involved in regulating the intracellular fate of internalized cell surface receptors. Receptors for platelet-derived growth factor (PDGF) have been shown to be internalized as a complex with pllO/p85 PI 3-kinase (Kapeller et al., 1993). Mutant PDGF receptors specifically lacking the binding site for PI 3-kinase fail to accumulate intracellularly, suggesting a role for PI 3-kinase activity in the normal endocytic trafficking of cell surface receptors (Joly et al., 1994). In addition, it has been shown that mutant CSF receptors that do not associate with PI 3-kinase are internalized but fail to be delivered to the lysosome for degradation (Downing et al., 1989; Carlberg et al., 1991). PI 3-kinase activity may therefore be involved in the sorting step for mammalian cell surface receptors, at which the decision is made to recycle to the cell surface or be diverted to the lysosome for turnover. A role for PI 3-kinase in
Protein Sorting to the Yeast Vacuole
143
regulating intracellular membrane trafficking may also be indicated by the observation that PI 3-kinase activity is required for the insulin-stimulated translocation to the cell surface of vesicles containing glucose transporters (Okada et al., 1994; Thelenetal., 1994). The question arises of how a specific phosphorylation event on a membrane phospholipid is involved in regulating the vesicular delivery of proteins to the yeast vacuole. One possibility is that Vps34p-mediated phosphorylation of membrane Ptdlns and subsequent incorporation of this modified phospholipid into transport vesicles designate these vesicles for delivery to the vacuole. Vesicle traffic has been proposed to occur between membrane-bound organelles throughout the secretory pathway, and the fidelity of the pathway is dependent on the docking and fusion of such vesicles with the correct target organelle (reviewed in Fryer et al., 1992; Rothman and Orci, 1992). Mechanisms involving specific membrane receptors that recognize PtdIns(3)P-tagged vesicles and mediate the docking/fusion of the vesicles with the appropriate membrane may facilitate vacuolar protein sorting in yeast. Another model proposes that phospholipids could play a dynamic role in regulating the activity or membrane association of proteins involved in the vacuolar protein sorting pathway. It has been shown that the association of SecA with the E. coli inner membrane and activation of its ATPase activity require acidic phospholipids (Lill et al., 1990; Hendrick and Wickner, 1991). Acidic phospholipids also have been proposed to be involved in other processes such as the import of precursor proteins into mitochondria (Eilers et al., 1989), DNA replication in E. coli (Yung and Romberg, 1988), regulation of phospholipase C activity (Bell and Bums, 1991), and association of annexins with the plasma membrane (Lin et al., 1992). Phosphoinositides have also been postulated to play a role in regulating the activity or membrane association of numerous proteins. PtdIns(4,5)P2 has been shown to bind to and/or activate molecules such as profilin, protein 4.1, and protein kinase CC, (Anderson and Marchesi, 1985; Lassing and Lindberg, 1985; GoldschmidtClermont et al., 1990; Nakanishi et al., 1993). In addition, a recent report has suggested that the pleckstrin motif found in many different proteins may serve in association with phosphoinositides such as PtdIns(3)P, PtdIns(4)P, and PtdIns(4,5)P2 (Harlan et al., 1994). Interestingly, many pleckstrin motif-containing proteins are known to be membrane-associated, suggesting that binding to specific modified phospholipids may contribute to the intracellular localization of these proteins. Vps34p-mediated phosphorylation of membrane Ptdlns may catalyze the recmitment of accessory proteins involved in the budding or transport of vesicles from the sorting compartment. Candidates for such vesicle accessory proteins include clathrin and clathrin-associated adaptors. In addition to its role in endocytosis, clathrin has been shown to be required for the sorting of both mammalian lysosomal and yeast vacuolar proteins (Komfeld and Mellman, 1989; Seeger and Payne, 1992b). Interestingly, it has been reported that adaptins will bind to some 3'-phosphorylated forms of inositol polyphosphates (Voglmaier et al., 1992), suggesting the possibility that adaptins may be able to bind to 3'-phosphorylated
144
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
phosphoinositides. Other candidates include cytoskeletal elements and molecular motors, which may be involved in vesicular trafficking between the Golgi and the vacuole. It is also possible that PtdIns(3)P may be involved in the segregation or clustering of receptor—ligand complexes such that they may be efficiently packaged into vesicular carriers. A final model suggests that phosphorylation of membrane Ptdlns by Vps34p could result in changes in the curvature of the membrane bilayer due to increased charge repulsion between the phospholipid polar heads, as predicted by the membrane couple hypothesis (Sheetz and Singer, 1974), possibly stimulating the budding of transport vesicles. Genetic epistasis experiments between vps mutants that accumulate vesicles (e.g., vps21 and vps45\ Cowles et al., 1994; Horazdovsky et al., 1994) and vps34 mutants may help distinguish between roles for PtdIns(3)P in vesicle formation and targeting. C. Interactions Between VpsISp and Vps34p
Vpsl5p and Vps34p have been shown by genetic and biochemical criteria to interact as a complex that is associated with the cytoplasmic face of an intracellular membrane fraction most likely corresponding to a late Golgi compartment (Herman et al., 1991a; Stack et al., 1993). These experiments showed that vpsl5 kinase domain mutations are suppressed by the overproduction of Vps34p. Although vpsl5 kinase domain mutants are extremely defective for vacuolar protein sorting (>95% of CPY is missorted and secreted as the p2 precursor), overproduction of Vps34p allows approximately 50% of CPY to be delivered to the vacuole (Stack et al., 1993). In contrast, the protein sorting defects of strains from which the VPSJ 5 gene has been deleted are not suppressed by the overproduction of Vps34p. The lack of suppression of strains carrying a VPS J 5 gene deletion indicates that the overproduction of Vps34p cannot bypass the requirement for Vps 15p in vacuolar protein sorting. These genetic data indicate that the Vps 15 and Vps34 proteins interact functionally in the cell. Native immunoprecipitation and chemical crosslinking experiments have demonstrated that Vps 15p and Vps34p physically interact in vivo as a complex (Stack et al., 1993). Sucrose gradient analysis has shown that the Vps 15p-Vps34p complex is associated with an intracellular membrane fraction that also contains the Kex2 protease, which is a resident late Golgi membrane protein. Subcellular fractionation experiments have shown that Vpsl5p is responsible for the membrane association of Vps34p (Stack et al., 1993). The combined biochemical and genetic data therefore argue that the Vps 15 and Vps34 proteins act together within a hetero-oligomeric complex to facilitate yeast vacuolar protein delivery. In addition to recruiting Vps34p to the membrane, Vps 15p also serves to activate Vps34p, because Ptdlns 3-kinase activity is very defective in Avps]5 and vps J 5 kinase domain mutant strains (Stack et al., 1993, 1995). The fact that the great majority of Vps34-mediated Ptdlns 3-kinase activity in a wild-type cell is found in a pelletable fraction, presumably because of interaction of Vps34p with Vps 15p,
Protein Sorting to the Yeast Vacuole
145
also supports a direct role for Vpsl5p in regulating Vps34p activity (Schu et al., 1993; Stack et al., 1993). Analysis of catalytically inactive forms of VpslSp and Vps34p has emphasized the regulatory relationship between the two proteins. Kinase-defective forms of Vps34p result in a dominant-negative phenotype when overproduced in a wild-type strain. These strains missort -50% of p2CPY and exhibit defects in Ptdlns 3-kinase activity (Stack et al., 1995). This mutant phenotype appears to be the result of the sequestration of Vpsl5p into nonfunctional complexes by the Vps34 mutant proteins. In contrast to catalytically inactive Vps34p, overproduction of kinase-defective Vpsl5 mutants does not result in a dominant-negative phenotype. Vpsl5p kinase domain mutants were found to be defective in association with Vps34p, suggesting that overproduction of kinase-defective forms of Vps 15p is unable to titrate Vps34p and consequently does not result in dominant interference (Stack et al., 1995). In addition, these observations indicate that an intact Vpsl5p protein kinase domain is required for association with Vps34p and subsequent stimulation of its Ptdlns 3-kinase activity. The fact that Vps 15 kinase domain mutants are unable to associate with Vps34p also presents a molecular explanation for the observation that overproduction of Vps34p will suppress the vacuolar protein sorting defects of vps 15 protein kinase domain mutant strains (Stack et al., 1993). In this scenario, the decreased affinity of Vps 15p kinase domain mutants for Vps34p can be partially overcome by the 20-30-fold overproduction of Vps34p. Therefore, the increased concentration of Vps34p will allow formation of sufficient Vps 15p-Vps34p complexes such that the severe vacuolar protein sorting defects of vpsl5 kinase domain mutants are partially suppressed. Collectively, these data demonstrate that the functionally active form of Vps34p is in a complex with Vpsl5p and indicate that formation of a stable complex between Vpsl5p and Vps34p is absolutely required for the efficient localization of soluble vacuolar proteins. A number of models can be suggested that incorporate a membrane-associated complex of Vps 15p and Vps34p in the vesicular delivery of proteins to the vacuole. Subcellular fractionation of Vpsl5p and Vps34p (Herman and Emr, 1990; Herman et al., 1991a) and the localization of precursor vacuolar proteins in yeast cells lacking Vpsl5p function (Vida et al., 1993) suggest that Vpsl5p and Vps34p act at the level of the late Golgi, most likely at the sorting compartment for vacuolar proteins. One model suggests that the Vps 15p-Vps34p complex may be able to associate (directly or indirectly) with the cytoplasmic tails of receptors for vacuolar protein precursors (Stack et al., 1993). In a manner analogous to that of cell surface receptor proteins, ligand binding to vacuolar protein receptors in the sorting compartment may transduce a signal that promotes receptor association with and/or activation of the Vps 15 protein kinase. Activation of Vpsl5p results in the stimulation of Vps34p Ptdlns 3-kinase activity. Phosphorylation of membrane Ptdlns by Vps34p would then trigger a cascade of events that ultimately results in the vesicular delivery of receptor—ligand complexes to the vacuole directly or through an endosomal intermediate. In this model, Vpsl5p and Vps34p effectively act as
146
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
components of a signal transduction complex that couples the signal received by specific transmembrane receptor proteins with a second messenger molecule [PtdIns(3)P] that catalyzes the action of as yet unidentified effector molecules. Vps 15p protein kinase activity also may be required for other aspects of vacuolar protein delivery such as regulating the packaging of vacuolar proteins into their specific vesicular carriers or mediating the delivery and/or fusion of transport vesicles with the endosome or the vacuole (Herman et al., 1991a,b, 1992). In mammalian cells, protein phosphorylation has been implicated in the regulation of the sorting of specific transmembrane receptors, including the polymeric immunoglobulin receptor, within an early endosome (Casanova et al., 1990). In addition, phosphorylation of the cytoplasmic tail of mammalian mannose-6-phosphate receptors is correlated with the exit of the receptor from the trans-Golgi network and has been suggested to stabilize their association with clathrin-associated adaptor proteins (Meresse et al., 1990; Meresse and Hoflack, 1993). The functional analyses of the Vps 15 protein kinase and the Vps34 Ptdlns 3-kinase has led to unexpected and novel mechanisms for regulating protein trafficking pathways. Many mechanistic aspects of the specific roles for protein and phospholipid phosphorylation in membrane trafficking remain to be revealed, and it is likely that work on vacuolar protein sorting in yeast will continue to yield exciting and important contributions to the regulation of intracellular protein localization in eukaryotic cells.
VII. GTP-BINDING PROTEINS IN THE SORTING OF VACUOLAR PROTEINS The movement of proteins through many protein localization pathways involves vesicle-mediated transport systems. Because of the directionality and complexity of these systems, the targeting and fusion of vesicular transport intermediates must be specifically and tightly regulated. Genetic approaches as well as the use of in vitro reconstitution assays have uncovered a large number of soluble and membrane-associated components required for the formation, targeting, and fusion of transport vesicles as they move through the secretory pathway (Fryer et al., 1992; Rothman and Orci, 1992). One group of proteins required in this process are members of the ras-like small GTP-binding protein family, the rab proteins (Goud et al., 1988; Pfeffer, 1992; Schwaninger et al., 1992). Distinct members of this family regulate specific steps in the secretion pathway by promoting the targeting and/or fusion of vesicular intermediates with the appropriate membrane targets in a GTP-dependent manner (Pryer et al., 1992; Walworth et al., 1992). This level of specificity is essential to maintaining the unique characteristics of the organelles in the secretory pathway. In the yeast secretory pathway, the small GTP-binding proteins Yptlp and Sec4p are required for vesicle transport steps from the ER to the Golgi and the Golgi to the plasma membrane, respectively (Goud et al., 1988; Becker et al., 1991). Mutations in the genes encoding these proteins result in a defect
Protein Sorting to the Yeast Vacuole
147
in protein secretion and the accumulation of vesicle intermediates (Goud et al., 1988; Segev et al., 1988). Similarly, mutations in the genes encoding small GTP-binding proteins involved in the delivery of proteins from the Golgi to the vacuole would be expected to lead to a vacuolar protein sorting defect and possibly the accumulation of vesicle carriers. A number of small GTP-binding proteins have been shown to play a role in vacuolar protein targeting. A. Vps21p Cloning and sequencing of the VPS21 gene revealed that it encodes a rab5-like small GTP-binding protein (Horazdovsky et al., 1994; Singer-Kruger et al., 1994). vps2J mutants exhibit severe vacuolar protein sorting defects; both CPY and PrA are missorted and secreted in Avps2] strains. Subcellular fractionation experiments showed that Vps21 p is associated with multiple membrane compartments, possibly corresponding to Golgi, vesicular, and endosomal membranes (Horazdovsky et al., 1994). A role for Vps21 p in the targeting of transport vesicles was suggested by the observation that Avps21 cells accumulate 40-50-nm vesicles (Horazdovsky et al., 1994; Singer-Kruger et al., 1994). This result is similar to that obtained with mutants of the rab homologue Sec4p, which acts at the Golgi-to-plasma membrane step in the yeast secretory pathway (Salminen and Novick, 1987). Although the transport vesicles that accumulate in sec4 mutant strains are distinct from those in Avps2] strains (80—100 nm vs. 40-50 nm) and Avps2I cells are clearly competent for secretion, inasmuch as they efficiently secrete missorted p2CPY, the accumulation of vesicles in the two mutants suggests that Sec4p and Vps2 Ip may execute similar functions in their respective pathways. Many small GTP-binding proteins are associated with intracellular membranes through isoprenyl modification of a COOH-terminal cysteine; in many cases, isoprenylation is essential for membrane association (Glomset et al., 1990; Powers, 1991). Vps21p appears to acquire geranylgeranyl isoprenyl modification at its COOH terminus. In addition to containing a consensus sequence for COOH-terminal geranylgeranylation, the membrane association of Vps21p is abolished if its acceptor cysteines are mutated or if the yeast geranylgeranyl transferase P-subunit, Bet2p, is mutated (Horazdovsky et al, 1994). These data indicate that geranylgeranylation of Vps21p is required for its membrane association. Site-directed mutagenesis of the VPS21 gene has further established that domains corresponding to GTP-binding domain I and the putative effector domain are required for vacuolar protein sorting (Horazdovsky et al., 1994). Mutation of residues within GTP-binding domain I result in the inactivation of Vps21p because these mutant strains missort and secrete p2CPY. In addition, these mutant proteins are unable to bind GTP in a blot binding assay. Mutation of residues corresponding to the putative effector domain also result in severe defects in CPY sorting. As mentioned earlier, strains containing a mutation of the COOH-terminal geranylgeranyl addition site in Vps21p are very defective for CPY sorting; however, these
148
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
mutant Vps21 proteins are able to bind GTP at wild-type levels. Collectively, the mutagenesis data indicate that multiple functional domains in Vps2lp are required for the efficient sorting of vacuolar proteins. In addition, these defined point mutants will serve as valuable reagents in genetic and biochemical screens for upstream regulators and downstream effectors of Vps21p. In mammalian cells, the Vps21p homologue rab5 has been implicated in vesicle targeting and fusion events in the early stages of the endocytic pathway. However, a role for Vps21p in early stages of endocytosis is unlikely because a-factor is internalized in t^vpsll strains at a rate nearly identical to that of wild-type strain (Horazdovsky et al., 1994). These analyses indicate two possible sites of action for Vps21p: it may be involved in Golgi-to-endosome traffic or in endosome-to-vacuole transport. The accumulation of vesicles in t^vps2l cells further suggests that Vps21p is specifically required for the targeting and/or fusion of transport vesicles with their target membrane (endosomal or vacuolar). Further biochemical characterization is necessary to precisely define the site and mechanism of action of Vps21p in vacuolar protein sorting. B. Ypt7p
Ypt7p is a yeast homologue of mammalian rab7. Deletion of the FPrZgene leads to defects in vacuolar morphology and vacuolar protein maturation, while protein secretion is unaffected (Wichmann et al., 1992; B. Horazdovsky, unpublished observations). Endocytic uptake of the mating pheromone a-factor is also severely inhibited xnyptl mutant cells. On this basis, it was concluded that Ypt7p acts in the endocytic pathway leading to the vacuole (Wichmann et al., 1992). Analysis of internalized a-factor using Nycodenz gradients suggested that a-factor accumulates in late endosomes in b^yptl cells (Schimmoller and Riezman, 1994). Collectively, these data indicate that YptTp most likely acts at the endosome-to-vacuole step and therefore would affect both biosynthetic traffic from the Golgi and endocytic traffic from the cell surface. C.
Vpsip
The cloning and sequencing of the VPSl gene revealed that it encodes a protein containing a tripartite GTP-binding domain conserved in a number of known GTP-binding proteins. In addition, Vpslp was shown to be homologous to the Mx family of antiviral proteins (Rothman et al., 1990), and subsequent isolation and characterization of the mechanochemical enzyme dynamin found that it is homologous to Vpslp (Obar et al., 1990). The homology between Vpslp, the Mx proteins, and dynamin is most pronounced in their NH2-terminal regions where the GTPbinding domain is present. The Mx proteins were identified in a number of vertebrate species as interferon-inducible proteins that appear to interfere with viral infection. The mechanisms responsible for the antiviral properties of the Mx proteins are not known and the significance of its homology to Vpslp is unclear.
Protein Sorting to the Yeast Vacuole
149
Dynamin was isolated as a microtubule-associated mechanochemical enzyme and postulated to be involved in microtubule-based motility (Obar et al., 1990; Shpetner and Vallee, 1992). Although subsequent studies have cast doubt on this activity (Maeda et al., 1992), a role for dynamin in endocytosis is suggested by its homology to the Drosophila shibire gene product (Chen et al., 1991; van der Bliek and Meyerowitz, 1991). Flies with a shibire mutation appear to exhibit a generalized defect in endocytosis that is manifested as a paralytic phenotype in mutant flies due to defects in vesicle-mediated events at nerve terminals. A direct role for dynamin in endocytosis was demonstrated by the expression in mammalian cells of dominant mutant forms of dynamin-containing alterations in the GTP-binding domain that led to a block in receptor-mediated endocytosis (Herskovits et al., 1993; van der Bliek et al., 1993). In addition to a role for dynamin GTPase activity in endocytosis, these studies also suggested that the basic COOH terminus of dynamin may be responsible for interaction with other components of the endocytic pathway. The COOH terminus of dynamin has been shown to be involved in several interactions that may regulate its GTPase activity and thereby affect endocytosis. Dynamin was shown to bind to and be activated by a subset of SH3 domains, including those found in receptor tyrosine kinases and proteins associated with these receptors (Gout et al., 1993; Scaife et al., 1994). These findings suggest that the binding of dynamin to SH3 domain-containing proteins may be important for the formation of a protein complex required for the endocytic internalization of activated cell surface receptor tyrosine kinases. In addition, the COOH terminus of dynamin has been found to be phosphorylated by protein kinase C, and this phosphorylation results in a 10-fold activation of the GTPase activity of dynamin (Robinson et al., 1993). Finally, a recent report showed that dynamin GTPase activity is also stimulated by acidic phospholipids and membrane vesicles, possibly because of interaction with the basic COOH terminus of dynamin (Tuma et al., 1993). Collectively, these observations indicate that dynamin is involved in the endocytic trafficking of cell surface molecules and is subject to multiple levels of post-translational regulation and further suggest that it is a key regulator of endocytosis. Mutations in the VPSl gene result in a defect in the sorting of soluble vacuolar proteins (Rothman et al., 1990). In addition, VPSl was found to be identical to the S. cerevisiae SP015 gene, mutations in which lead to a defect in meiotic cell division required for sporulation (Yeh et al., 1991). The sporulation defect observed in vpsl cells is reminiscent of that observed in pep4 mutants (PEP4 encodes PrA, which is required for the proteolytic activation of several important vacuolar proteolytic enzymes). The vacuolar protein sorting defect in vpsl cells is likely to lead to a gross deficiency in vacuolar protease activity and result in a sporulation defect similar to that ofpep4 cells. Because dynamin originally was reported to be a microtubule-dependent mechanoenzyme and Spol 5p could associate irreversibly with microtubules in vitro, Vater et al. attempted to determine whether microtubules are required for vacuolar protein sorting in yeast. Although cells treated with the microtubule-depolymerizing drug nocodazole or containing mutations in the yeast
150
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
P-tubulin gene lacked any visible microtubules by immunofluorescence, they did not exhibit any vacuolar protein sorting defect (Vater et al., 1992). In addition, immunofluorescence experiments detected no colocalization of Vpslp and microtubules, and treatment with nocodazole did not lead to a redistribution of Vpslp within the cell. These data indicate that Vpslp most likely does not associate with microtubules in vivo and microtubules do not appear to be essential for vacuolar protein sorting. Vpslp expressed in E. coli has been demonstrated to bind GTP in vitro, and Vpslp immunoprecipitated under nondenaturing conditions from yeast catalyzed the hydrolysis of GTP (Vater et al., 1992). Therefore, Vpslp appears to be able to bind GTP and possesses an intrinsic GTPase activity. Mutational analysis of Vps 1 p led to the proposal that the protein is composed of two functionally distinct domains. Mutations in VPS I generated by hydroxylamine mutagenesis fell into two categories: I) dominant-negative mutations that cause a protein sorting defect in wild-type cells and 2) recessive loss of function mutations (Vater et al., 1992). Mapping of the mutation responsible for these phenotypes yielded a surprising result. Mutations resulting in a dominant-negative phenotype all mapped to the NH2-terminal region of Vpslp, whereas recessive loss of function mutations gave rise to unstable or truncated protein products. To test whether the dominant-negative phenotype required an intact Vpslp COOH terminus, a point mutation which on its own results in a dominant-negative phenotype was combined with a frame shift mutation that causes a COOH-terminal truncation. Transformation of the double mutant into wild-type cells did not result in a vacuolar protein sorting defect, indicating that deletion of the COOH terminus of dominant-negative mutants eliminates dominant interference (Vater et al., 1992). This mutational analysis of Vpslp led to the proposal that any mutation in the NH2 terminus that results in a nonftmctional protein presumably by interfering with GTP binding or hydrolysis, will give rise to a dominant-negative phenotype. The requirement of the COOHterminal region for a dominant-negative phenotype suggests that Vpslp may interact with other components of the sorting machinery through its COOH-terminal domain. A role for Vpslp in the retention of Golgi membrane proteins has been recently described. Wilsbach and Payne performed a genetic screen to isolate yeast mutants defective in the retention of the Golgi membrane protein Kex2p. Kex2p residues in a late Golgi compartment and is involved in the proteolytic processing of the mating pheromone a-factor. In mutants exhibiting defective a-factor processing, Kex2p was found to be delivered to the vacuole and degraded in a PEP^-dependent manner (Wilsbach and Payne, 1993b). The laml mutant (for low a-factor maturation) was also defective for the sorting of CPY to the vacuole and was found to be allelic to VPS I. A model for Vps 1 p action was presented whereby Vps 1 p is involved in the recycling of membrane proteins from a post-Golgi prevacuolar endosomelike intermediate compartment. Loss of Vps Ip function would then lead to delivery of normally recycled membrane proteins from the intermediate compartment to the
Protein Sorting to the Yeast Vacuole
151
vacuole (see Vacuolar Membrane Proteins). Included in the mislocalized membrane proteins would be the putative transmembrane receptor protein(s) for soluble vacuolar proteins (see Receptor-Mediated Vacuolar Protein Sorting). Blockage of sorting receptor recycling back to the Golgi would then result in the vacuolar protein sorting defect observed in vpsJ cells (Wilsbach and Payne, 1993a,b). An alterative explanation is that Vpslp may act at the late Golgi to facilitate the sorting of vacuolar proteins, and loss of Vpslp function may compromise the integrity of the Golgi compartment such that resident membrane proteins are not retained and are delivered to the vacuole by default. Such a model is consistent with immunofluorescent localization experiments of Vps Ip that suggest an association with the Golgi (Rothman et al., 1990). Analysis of a temperature-sensitive allele of VPSJ, in which the protein is immediately inactivated upon shift to the non-permissive temperature, should allow resolution of direct vs. indirect models for the role of Vpslp in membrane protein trafficking.
Vm. VESICLE TARGETING EVENTS IN VACUOLAR PROTEIN LOCALIZATION In addition to GTP-binding proteins, many other molecules have been implicated in the formation and docking/fusion of vesicles with their appropriate target membrane. Analyses of these molecules have revealed remarkable similarities that point to conserved mechanisms regulating vesicular movement throughout the secretory pathway in all eukaryotic cells. Rothman and co-workers have extensively characterized molecules involved in intra Golgi vesicular trafficking: these molecules include N-ethylmaleimide sensitive factor (NSF), soluble NSF attachment proteins (SNAPs), ADP ribosylation factor (ARF), and COP non-clathrin coat proteins (Rothman and Orci, 1992). Characterization of proteins associated with NSF-SNAP complexes revealed that they included components previously identified as being involved in synaptic vesicle docking/fiision with the plasma membrane (Bennett and Scheller, 1993, Sollner et al., 1993a,b). This merging of mechanisms and molecules initially characterized from different processes led Rothman and colleagues to propose a generalized hypothesis for the regulation of transport vesicle docking and fusion with the target membrane (Rothman and Warren, 1994). In this model, soluble SNAP proteins bind to their receptors (SNAREs) on both the transport vesicle (v-SNARE) and the target membrane (t-SNARE). Specificity in the targeting step would result from unique SNAREs present on distinct membranes. The SNARE protein family is typified by the well-characterized synaptic vesicle VAMP/synaptobrevin v-SNAREs (Trimble et al., 1988; Baumert et al., 1989) and the plasma membrane syntaxin (Bennett et al., 1992) and SNAP-25 (Oyler et al., 1989) t-SNAREs. VAMP has been shown to interact specifically with syntaxn (Calakos et al., 1994), in a complex that also includes SNAP-25 (Sollner et al., 1993b). It is thought that this recognition event docks synaptic vesicles at the
152
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
plasma membrane. Following docking, the vesicles fuse with the plasma membrane, releasing their contents (Bennett and Scheller, 1993). Recent studies have shown that another protein, n-Sec 1, specifically interacts with syntaxin (Hata et al., 1993; Garcia et al., 1994; Pevsner et al., 1994), and it has been proposed that n-Sec 1 might function to regulate formation of the vesicle docking complex (Pevsner et al., 1994). In yeast, seel ts mutant cells accumulate 80-100-nm Golgi-derived secretory vesicles, suggesting a role for Sec 1 p in targeting and/or fusion of secretory vesicles with the plasma membrane (Novick et al., 1980). Interestingly, two multicopy suppressors of di seel ts mutant (SSOl and SS02) have been identified, and their predicted gene products share significant sequence homology with mammalian syntaxin (Aalto et al., 1993). In addition, functional homologues of NSF, SNAPs, ARF, and COPs have been identified that act in the secretory pathway in yeast. These homologues have been implicated at multiple steps in the secretory pathway, including ER-to-Golgi and Golgi-to plasma membrane transport. The conservation of these molecules between different organisms and in multiple vesicle targeting reactions suggests that a combination of unique and common molecules acts at individual vesicle transport steps throughout the cell. Bennett and Scheller have suggested that several unique classes of molecules are utilized at specific steps. These components include a small GTP-binding protein of the rab/ypt family, v-SNARE, t-SNARE, and a Sec 1 protein family member (Bennett and Scheller, 1993). Several of these classes of molecules have been identified as being required for Golgi-to-vacuole transport. Vps21p and Ypt7p are small GTP-binding proteins most homologous to mammalian rabs (see above) (Wichmann et al., 1992; Horazdovsky et al., 1994). Pepl2p (Vps6p) (E. Jones, unpublished observations) is structurally homologous to t-SNAREs, and Vps33p (Banta et al., 1990) and Vps45p(Cowlesetal., 1994) are homologous to Seel p. Identification of two Seclp homologues in the vacuolar protein sorting pathway suggests that multiple membrane trafficking events are involved in protein delivery to this organelle. Although vps33 and vps45 mutants are both defective for delivery of vacuolar hydrolases, their vacuole morphologies are strikingly different. In contrast to the enlarged vacuoles of vps45 mutants, vps33 mutants lack vacuoles (and are therefore designated class C). The subcellular fractionation profile of Vps33p also markedly differs from the one observed for Vps45p (Banta et al., 1990; Cowles et al., 1994). These data suggest separate sites of action for Vps33p and Vps45p. It is possible that one of these proteins may function in Golgi-to-endosome protein transport while the other plays a role in an endosome-to-vacuole protein delivery event (Fig. 2). The present data suggest that Vps45p functions in a Golgi-to-endosome delivery event, and the accumulation of vesicular intermediates seen in vps45 suggests that Vps45p functions at a vesicle targeting/fusion event. Although it is not clear whether distinct components will be required for Golgi to endosome and endosome to vacuole steps, it is apparent that Golgi-to-vacuole transport involves conserved mechanisms regulating vesicular traffic throughout the cell. Therefore, insights gained from
Protein Sorting to the Yeast Vacuole
153
genetic and biochemical analyses of these components should prove generally applicable in a variety of systems and may help illuminate mechanistic aspects of the SNARE hypothesis.
IX. CLATHRIN AND VACUOLAR PROTEIN SORTING Most, if not all, vesicle-mediated intracellular transport events involve vesicle-associated coat proteins (Pryer et al., 1992). In general, components of the coat associate with the cytoplasmic tails of transmembrane proteins, and this serves to concentrate membrane proteins prior to packaging into transport vesicles. In addition, self-assembly of coat proteins onto the donor membrane has been proposed to provide the energy to deform the membrane bilayer to allow vesiculation. Classic clathrin-coated vesicles play a role in receptor-mediated endocytosis of proteins from the plasma membrane to the early endosome and in delivery of proteins from the trans-Golgi network (TGN) to the late endosome in mammalian cells. Non-clathrin-coated vesicles appear to mediate intra-Golgi transport in mammalian cells and ER-to-Golgi transport in yeast (Pryer et al., 1992; Rothman and Orci, 1992; Barlowe et al., 1994). Transport events involving clathrin coats provide the best understood vesicular delivery system. Clathrin-coated vesicles are formed by the association of clathrin and associated adaptin proteins at coated pits (Keen, 1990; Pearse and Robinson, 1990). The specificity of the coat assembly reaction is mediated by adaptor complexes (APs) composed of distinct sets of adaptin proteins. The adaptins associate with the cytoplasmic tails of transmembrane proteins and promote the assembly of clathrin heavy and light chains into the coat. There are two classes of adaptor complexes in mammalian cells that correspond to the requirement for clathrin-coated transport at the plasma membrane and the TGN. Adaptor complex I (AP I) mediates the delivery of proteins from the TGN to the endosome and adaptor complex II (AP II) from the plasma membrane to the endosome. It appears that there may be distinct signals present in the cytoplasmic tails of transmembrane cargo proteins that specify which set of adaptor proteins is utilized. Once coated vesicles are formed, the clathrin coat is disassembled to allow fusion of the vesicle with the target membrane. Initial characterization of the yeast clathrin heavy chain gene CHCl indicated that clathrin was not required for cell growth, protein secretion, or delivery of proteins to the vacuole (Payne et al., 1987, 1988). The latter observation was quite surprising given the well-characterized role of clathrin in the delivery of lysosomal proteins from the TGN to the lysosome. Biochemical characterization of yeast clathrin showed that it is capable of assembling into triskelions, suggesting that it is capable of forming a coat complex. Resolution of this apparent paradox came with the generation of a temperature-conditional allele of the yeast clathrin heavy chain gene {chcl-ts). In contrast to strains deleted for the CHCl gene that show no defect in maturation of CPY, one of the immediate consequences of shifting chcl-ts
154
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
cells to the nonpermissive temperature is the missorting of the soluble vacuolar hydrolases CPY and PrA (Seeger and Payne, 1992a). These same experiments also demonstrated that clathrin is not required for the delivery of resident vacuolar membrane proteins because ALP is matured normally in chcl-ts cells incubated at the nonpermissive temperature. These data indicate that clathrin is required for the sorting of soluble vacuolar proteins. Remarkably, it was found that extended incubation at the nonpermissive temperature prior to labeling allowed chcl-ts cells to recover the ability to sort CPY. After several hours of preincubation at nonpermissive temperature, chcl-ts cells showed essentially no defect in CPY sorting and exhibited the same vacuolar protein sorting phenotype as Ac/zc/ cells. This unusual adaptation phenomenon suggests that a non-clathrin mechanism may functionally substitute for clathrin in the sorting of soluble vacuolar proteins. Delivery of CPY to the vacuole in Ac/zc7 cells does not appear to involve transport of CPY to the cell surface and subsequent endocytosis into the vacuole, seel mutants block Golgi-toplasma membrane transport, and a seel ^chcl double mutant does not block CPY maturation, indicating that CPY is not delivered to the vacuole via the cell surface (Payne etal., 1988). The extremely rapid onset of the sorting defect in chcl-ts cells shifted to the nonpermissive temperature argues for a direct role for clathrin in vacuolar protein sorting. By analogy to lysosomal protein sorting in mammalian cells, it has been suggested that clathrin functions in yeast to concentrate receptors for soluble vacuolar proteins (e.g., Vps 1 Op) into coated pits. A model can be proposed whereby clathrin-coated vesicles are budded off the late Golgi sorting compartment and receptor-ligand complexes are delivered to an endosomal compartment where ligand dissociates from receptor. Vacuolar proteins then continue on to the vacuole and receptors recycle back to the late Golgi for additional rounds of transport. Inactivation of clathrin function due to shifting chcl-ts cells to the nonpermissive temperature would result in a blockage in the formation of coated pits and vesicles. Newly synthesized vacuolar protein precursors would then rapidly saturate the blocked receptors and be efficiently secreted from the cell.
X. CONCLUSIONS Identification of the vps mutants has offered unique opportunity to probe the molecular mechanisms of protein sorting. At the outset, the large number of mutants (>40 complementation groups) that affected this pathway was unanticipated. However, as the analysis of vacuolar protein sorting progresses, it is becoming evident that this localization pathway is very complex, involving the movement of protein and membrane to and from a number of subcellular organelles in a highly regulated manner. In recent years, the biochemical characterization of the yeast vacuolar protein sorting system has uncovered at least one prevacuolar compartment through which proteins travel en route to the vacuole. This discovery indicates that multiple transport events must be carried out to move proteins from the Golgi to the vacuole
Protein Sorting to the Yeast Vacuole
155
(Golgi -> endosome -> vacuole). Identification of the CPY sorting receptor (VpslOp) and the analysis of its movement through the vacuolar protein sorting pathway have indicated a requirement of not only anterograde traffic (Golgi to endosome) to move the CPY-receptor complex forward, but also retrograde traffic (endosome to Golgi) to retrieve VpslOp back to the Golgi for another round of sorting. The movement of these receptor—ligand complexes involves vesicular intermediates whose targeting and fusion are tightly controlled. This is evidenced by the participation of small GTP-binding proteins of the rab family (Vps21p and Ypt7p) and homologues of other proteins involved in the regulation of vesicle targeting and fusion (the Pep 12 t-SNARE protein, the Seclp homologue, Vps45p, as well as the Vpslp dynamin homologue). Further examination of the regulatory role these proteins play in vesicular trafficking in the vacuolar protein localization pathway should help uncover some of the general mechanisms involved in intracellular membrane trafficking. One of the most intriguing results generated from the current studies of vacuolar protein transport is the discovery that a phospholipid kinase (the Vps34 PI 3-kinase) is essential for vacuolar protein transport. Defining the role and regulatory mechanisms of lipid phosphorylation in vacuolar protein sorting will offer new insights into the general contribution of lipid modifications in intracellular membrane and protein trafficking. Further study of the Vps proteins discussed in this chapter, as well as determining the functional role of the remaining VPS gene products, should yield significant contributions to understanding the underlying mechanisms of protein sorting and organelle biogenesis in eukaryotic cells.
REFERENCES Aalto, M. K., Ronne, H., & Keranen, S. (1993). Yeast syntaxins Ssolp and Sso2p belong to a family of related membrane proteins that function in vesicular transport. EMBO J. 12, 4095-^104. Ammerer, G., Hunter, C, Rothman, J. H., Saari, G. C, Vails, L. A., & Stevens, T. H. (1986). PEP4 gene of Saccharomyces cerevisiae encodes proteinase A, a vacuolar enzyme required for processing of vacuolar precursors. Mol. Cell. Biol. 6, 2490-2499. Anderson, R. A. & Marchesi, V. T. (1985). Regulation of the association of membrane skeletal protein 4.1 with glycophorin by a polyphosphoinositide. Nature 318, 295-298. Anraku, Y, Hirata, R., Wada, Y, & Ohya, Y (1992a). Molecular genetics of the yeast vacuolar H''-ATPase. J. Exp. Biol. 172, 67-81. Anraku, Y, Umemoto, N., Hirata, R., & Ohya, Y (1992b). Genetic and cell biological aspects of the yeast vacuolar H -ATPase. J. Bioenerg. Biomembr. 24, 395—405. Auger, K. R., Carpenter, C. L., Cantley, L. C, & Varticovski, L. (1989a). Phosphatidyl inositol 3-kinase and its novel product, phosphatidylinositol 3-phosphate, are present in Saccharomyces cerevisiae. J. Biol. Chem. 264, 20181-20184. Auger, K. R., Serunian, L. A., SoltofiF, S. R, Libby, R, & Cantley, L. C. (1989b). PDGF-dependent tyrosine phosphorylation stimulates production of novel polyphosphoinositides in intact cells. Cell 57,167-175. Bachhawat, A., Suhan, J., & Jones, E. (1994). The yeast homolog of H
58, a mouse gene esstenital for embryogenesis, performs a role in the delivery of proteins to the vacuole. Genes Dev. 8, 1379-1387.
156
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
Bankaitis, V. A., Johnson, L. M., & Emr, S. D. (1986). Isolation of yeast mutants defective in protein targeting to the vacuole. Proc. Natl. Acad. Sci. USA 83, 9075-9079. Banta, L. M., Robinson, J. S., Klionsky, D. J., & Emr, S. D. (1988). Organelle assembly in yeast: Characterization of yeast mutants defective in vacuolar biogenesis and protein sorting. J. Cell Biol. 107, 1369-1383. Banta, L. M., Vida, T. A., Herman, P. K., & Emr, S. D. (1990). Characterization of yeast Vps33p, a protein required for vacuolar protein sorting and vacuole biogenesis. Mol. Cell. Biol. 10, 4638— 4649. Barlowe, C, Orci, L., Yeung, T., Hosobuchi, M., Hamamoto, S., Salama, N., Rexach, M. P., Ravazzola, M., Amherdt, M., & Schekman, R. (1994). COPII: A membrane coat formed by Sec proteins that drive vesicle buddingfromthe endoplasmic reticulum. Cell 77, 895—907. Bauerle, C, Ho, M., Lindorfer, M., & Stevens, T. (1993). The Sacchawmyces cerevisiae VMA6 gene encodes the 36-kDa subunit of the vacuolar H -ATPase membrane sector. J. Biol. Chem. 268, 12749-12757. Baumert, M., Maycox, P. R., Navone, P., De Camilli, P., & Jahn, R. (1989). Synaptobrevin: An integral membrane protein of 18,000 daltons present in small synaptic vesicles of rat brain. EMBO J. 8, 379-384. Becker, J., Tan, T. J., Trepte, H. H., & Gallwitz, D. (1991). Mutational analysis of the putative effector domain of the GTP-binding Yptl protein in yeast suggests specific regulation by a novel GAP activity. EMBO J. 10, 785-792. Bell, R. M. & Bums, D. J. (1991). Lipid activation of protein kinase C. J. Biol. Chem. 266,4661-4664. Bennett, M. K. & Scheller, R. H. (1993). The molecular machinery for secretion is conserved from yeast to neurons. Proc. Natl. Acad. Sci. USA 90, 2559-2563. Bennett, M. K., Calakos, N., & Scheller, R. H. (1992). Syntaxin: A synaptic protein implicated in docking of synaptic vesicles at presynaptic active zones. Science 257, 255-259. Blachly-Dyson, E. & Stevens, T. H. (1987). Yeast carboxypeptidase Y can be translocated and glycosylated without its amino-terminal signal sequence. J. Cell Biol. 104, 1183—1191. Calakos, N., Bennet, M., Peterson, K., & Scheller, R. (1994). Protein-protein interactions contributing to the specificity of intracellular vesicular trafficking. Science 263, 1146-1149. Cantley, L. C, Auger, K. R., Carpenter, C, Duckworth, B., Graziani, A., Kapeller, R., & Soltoff, S. (1991). Oncogenes and signal transduction. Cell 64, 281-302. Carlberg, K., Tapley, P., Haystead, C, & Rohrschneider, L. (1991). The role of kinase activity and the kinase insert region in ligand-induced internalization and degradation of the c-fms protein. EMBO J. 10,877-883. Casanova, J. E., Breitland, P. P., Ross, S. A., & Mostov, K. E. (1990). Phosphorylation of the polymeric immunoglobulin receptor required for its efficient transcytosis. Science 248, 742—745. Chen, M. S., Obar, R. A., Schroeder, C C, Austin, T. W., Poodry, C. A., Wadsworth, S. C, & Vallee, R. B. (1991). Multiple forms of dynamin are encoded by shibire, a Drosophila gene involved in endocytosis. Nature 351, 583-586. Chvatchko, Y., Howard, I., & Riezman, H. (1986). Two yeast mutants defective in endocytosis are defective in pheromone response. Cell 46, 355—364. Conradt, B., Shaw, J., Vida, T, Emr, S., «fe Wickner, W. (1992). In vitro reactions of vacuole inheritance in Sacchawmyces cerevisiae. J. Cell Biol. 119, 1469-1479. Conradt, B., Hass, A., & Wickner, W. (1994). Determination of four biochemically distinct, sequential stages during vacuole inheritance in vitro. J. Cell Biol. 126, 99-110. Cowles, C. R., Emr, S. D., & Horazdovsky, B. P. (1994). Mutations in the VPS45 gene, a SECl homologue, are defective for vacuolar protein sorting and accumulate clusters of membrane vesicles. J. Cell Sci. 107, 3449-3459. Davis, N. G., Horecka, J. L., & Sprague, G. P. (1993). Cis-acting and trans-acting functions required for endocytosis of the yeast pheromone receptors. J. Cell Biol. 122, 53-65.
Protein Sorting to the Yeast Vacuole
157
Downing, J. R., Roussel, M. F., & Sherr, C. J. (1989). Ligand and protein kinase C downmodulate the colony-stimulating factor 1 receptor by independent mechanisms. Mol. Cell. Biol. 9, 2890-2896. Dulic, V. & Riezman, H. (1989). Characterization of the END I gene required for vacuole biogenesis and gluconeogenic growth of budding yeast. EMBO J. 8, 1349-1359. Filers, M., Endo, T., & Schatz, G. (1989). Adriamycin, a drug interacting with acidic phospholipids, blocks import of precursor proteins by isolated yeast mitochondria. J. Biol. Chem. 264, 29452950. Escobedo, J. A., Navankasattusas, S., Kavanaugh, W. M., Milfay, D., Fried, V. A., & Williams, L. T. (1991). cDNA cloning of a novel 85 kd protein that has SH2 domains and regulates binding of PI3-kinase to the PDGF P-receptor. Cell 65, 75-82. Fantl, W. J., Escobedo, J. A., Martin, G. A., Turck, C. W., del Rosario, M., McCormick, F., & Williams, L. T. (1992). Distinct phosphotyrosines on a growth factor receptor bind to specific molecules that mediate different signaling pathways. Cell 69, 413—423. Garcia, E. R, Gatti, E., Butler, M., Burton, J., & De Camilli, R (1994). A rat brain Seel homologue related to Rop and UNCI8 interacts with syntaxin. Proc. Natl. Acad. Sci. USA 91, 2003-2007. Glomset, J. A., Gelb, M. H., & Famsworth, C. C. (1990). Prenyl proteins in eukaryotic cells: A new type of membrane anchor. Trends Biochem. Sci. 15, 139-142. Goldschmidt-Clermont, R J., Machesky, L. M., Baldassare, J. J., & Pollard, T. D. (1990). The actin-binding protein profilin binds to PIP2 and inhibits its hydrolysis by phospholipase C. Science 247, 1575-1578. Gomes de Mesquita, D. S., ten Hoopen, R., & Woldringh, C. L. (1991). Vacuolar segregation to the bud of Saccharomyces cerevisiae: An analysis of morphology and timing in the cell cycle. J. Gen. Microbiol. 137,2447-2454. Goud, B., Salminen, A., Walworth, N., & Novick, R (1988). A GTP-binding protein required for secretion rapidly associates with secretory vesicles and the plasma membrane in yeast. Cell 53, 753-768. Gout, I., Dhand, R., Hiles, I. D., Fry, M. J., Panayotou, G., Das, R, Truong, O., Totty, N. R, Hsuan, J., Booker, G. W., Campbell, I. D., & Waterfield, M. D. (1993). The GTPase dynamin binds to and is activated by a subset of SH3 domains. Cell 75, 25-36. Graham, T. R. & Emr, S. D. (1991). The Sec 18/NSF protein is required at multiple steps in the secretory pathway: Mutants define compartmental organization of protein modification and sorting events in the yeast Golgi. J. Cell Biol. 114, 207-218. Haas, A., Conradt, B., & Wickner, W. (1994). G-protein ligands inhibit in vitro reactions of vacuole inheritance. J. Cell Biol. 126, 87-97. Harlan, J. E., Hajduk, R J., Yoon, H. S., & Fesik, S. W. (1994). Pleckstrin homology domains bind to phosphatidylinositol-4,5-bisphosphate. Nature 371, 168-170. Hasilik, A. & Tanner, W. (1987). Biosynthesis of the vacuolar yeast glycoprotein carboxypeptidase Y. Conversion of precursor into the enzyme. Eur. J. Biochem. 85, 599-608. Hata, v., Slaughter, C.A., & Siidhof, T. C. (1993). Synaptic vesicle fusion complex contains unc-18 homologue bound to syntaxin. Nature 366, 347-351. Hemmings, B. A., Zubenko, G. S., Hasilik, A., & Jones, E. W. (1981). Mutant defective in processing of an enzyme located in the lysosome-like vacuole of Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. USA 78, 435-439. Hendrick, J. P. & Wickner, W. (1991). SecA protein needs both acidic phospholipids and SecY/E protein for functional high-affinity binding to the Escherichia coli plasma membrane. J. Biol. Chem. 266, 24596-24600. Herman, P. K. & Emr, S. D. (1990). Characterization of VPS34, a gene required for vacuolar protein sorting and vacuole segregation in Saccharomyces cerevisiae. Mol. Cell. Biol. 10, 6742-6754. Herman, P. K., Stack, J. H., DeModena, J. A., & Emr, S. D. (1991a). A novel protein kinase homolog essential for protein sorting to the yeast lysosome-like vacuole. Cell 64, 425—437.
158
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
Herman, P. K., Stack, J. H., & Emr, S. D. (1991b). A genetic and structural analysis of the yeast Vpsl5 protein kinase: Evidence for a direct role of Vpsl5p in vacuolar protein delivery. EMBO J. 13, 4049-4060. Herman, P. K., Stack, J. H., & Emr, S. D. (1992). An essential role for a protein and lipid kinase complex in secretory protein sorting. Trends Cell Biol. 2, 363—368. Herskovits, J. S., Burgess, C. C, Obar, R. A., & Vallee, R. B. (1993). Effects of mutant rat dynamin on endocytosis. J. Cell Biol. 122, 565-578. Hiles, I. D., Otsu, M., Volinia, S., Fry, M. J., Guot, I., Dhand, R., Panayotou, G., Ruiz-Larrea, F., Thompson, A., Totty, N. F., Hsuan, J. J., Courtneidge, S. A., Parker, R J., «& Waterfield, M. D. (1992). Phosphatidylinositol 3-kinase: Structure and expression of the 110 kd catalytic subunit. Cell 70, 419-429. Ho, M., Hill, K., Lindorfer, M., & Stevens, T. (1993a). Isolation of vacuolar membrane H -ATPase-deficient yeast mutants—the VMA5 and VMA4 gene are essential for assembly and activity of the vacuolar H^-ATPase. J. Biol. Chem. 268,221-227. Ho, M., Hirata, R., Umemoto, N., Ohya, Y., Takatsuhi, A., & Stevens, T. (1993b). VMAI3 encodes a 54-kDa vacuolar H -ATPase subunit required for activity but not assembly of the enzyme complex in Saccharomyces cerevisiae. J. Biol. Chem. 268, 18286-18292. Horazdovsky, B. F. & Emr, S. D. (1993). The VPS16gQnQ product associates with a sedimentable protein complex and is essential for vacuolar protein sorting in yeast. J. Biol. Chem. 268, 4953-4962. Horazdovsky, B. F., Busch, G. R., & Emr, S. D. (1994). VPS21 encodes a rab5-like GTP binding protein that is required for the sorting of yeast vacuolar proteins. EMBO J. 13, 1297-1309. Johnson, L. M., Bankaitis, V. A., & Emr, S. D. (1987). Distinct sequence determinants direct intracellular sorting and modification of a yeast vacuolar protein. Cell 48, 875-885. Joly, M., Kazlauskas, A., Fay, F. S., & Corvera, S. (1994). Disruption of PDGF receptor trafficking by mutation of its PI-3 kinase binding sites. Science 263, 684-687. Jones, E. (1984). The synthesis and function of proteases in Saccharomyces: Genetic approaches. Annu. Rev. Genet. 18,233-270. Jones, E., Woolford, C , Moehle, C, Noble, J., & Innis, M. (1989). Cascades of Yeast Vacuolar Proteases (Hugh, T, Ed.), pp. 141-147. Alan R. Liss, New York. Jones, E. W. (1977). Proteinase mutants oiSaccharomyces cerevisiae. Genetics 85, 23-33. Jones, E. W., Zubenko, G. S., & Parker, R. R. (1982). PEP4 gene function is required for expression of several vacuolar hydrolases in Saccharomyces cerevisiae. Genetics 102, 665-667. Jones, H. D., Schliwa, M., & Drubin, D. G. (1993). Video microscopy of organelle inheritance and motility in budding yeast. Cell Motil. Cytoskel. 25, 129-141. Kane, P. & Stevens, T. (1992). Subunit composition, biosynthesis, and assembly of the yeast vacuolar proton-translocating ATPase. J. Bioenerg. Biomembr. 59,415-438. Kane, P., Kuehn, M., Howald-Stevenson, I., & Stevens, T. (1992). Assembly and targeting of peripheral and integral membrane subunits of the yeast vacuolar H -ATPase. J. Biol. Chem. 267, 447-^54. Kapeller, R., Chakrabarti, R., Cantley, L., Fay, F., & Corvera, S. (1993). Internalization of activated platelet-derived growth factor receptor-phosphatidylinositol-3' kinase complexes: Potential interactions with the microtubule cytoskeleton. Mol. Cell. Biol. 13, 6052-6063. Keen, J. H. (1990). Clathrin and associated assembly and disassembly proteins. Annu. Rev. Biochem. 59,415-438. Kitamoto, K., Yoshizawa, K., Ohsumi, Y., & Anraku, Y. (1988). Dynamic aspects of vacuolar and cytosolic amino acid pools of Saccharomyces cerevisiae. J. Bacteriol. 170, 2683—2686. Klionsky, D. J. & Emr, S. D. (1989). Membrane protein sorting: Biosynthesis, transport and processing of yeast vacuolar alkaline phosphatase. EMBO J. 8, 2241-2250. Klionsky, D. J., Banta, L. M., & Emr, S. D. (1988). Intracellular sorting and processing of a yeast vacuolar hydrolase: Proteinase a propeptide contains vacuolar targeting information. Mol. Cell. Biol. 8,2105-2116.
Protein Sorting to the Yeast Vacuole
159
Klionsky, D. J., Herman, P. K., & Emr, S. D. (1990). The fungal vacuole: Composition, function and biogenesis. Microbiol. Rev. 54, 266-292. Kohrer, K. & Emr, S. D. (1993). The yeast VPS 17 gene encodes membrane-associated protein required for the sorting of soluble vacuolar hydrolases. J. Biol. Chem. 268, 559-569. Komfeld, S. & Mellman, I. (1989). The biogenesis of lysosomes. Annu. Rev. Cell Biol. 5,483-525. Lassing, I. & Lindberg, U. (1985). Specific interaction between phosphatidylinositol 4,5-bisphosphate and profilactin. Nature 314, 472-474. Lill, R., Dowhan, W., & Wickner, W. (1990). The ATPase activity of SecA is regulated by acidic phospholipids, SecY, and the leader and mature domains of precursor proteins. Cell 60, 271-280. Lin, H. C, Sudhof, T. C, & Anderson, R. G. W. (1992). Annexin VI is required for budding of clathrin-coated pits. Cell 70, 283-291. Maeda, K., Nakata, T., Noda, Y, Sato-Yoshitake, R., & Hirokawa, N. (1992). Interaction of dynamin with microtubules: Its structure and GTPase activity investigated by using highly purified dynamin. Mol. Biol. Cell 3, 1181-1194. Marcusson, E. G., Horazdovsky, B. F., Lin Cereghino, J., Gharakhanian, E., & Emr, S. D. (1994). The sorting receptor for yeast vacuolar carboxypeptidase Y is encoded by the VPSIO gene. Cell 77, 579-586. Mechler, B., Muller, M., Muller, H., Meussdoerffer, R, & Wolf, D. (1982). In vivo biosynthesis of the vacuolar proteinases A and B in the yeast Saccharomyces cerevisiae. J. Biol. Chem. 257, 11203-11206. Meresse, S. & Hoflack, B. (1993). Phosphorylation of the cation-independent mannose 6-phosphate receptor is closely associated with its exit from the trans-Golgi network. J. Cell Biol. 120, 67—75. Meresse, S., Ludwig, T., Frank, R., & Hoflack, B. (1990). Phosphorylation of the cytoplasmic domain of the bovine cation-independent mannose 6-phosphate receptor. Serines 2421 and 2492 are the targets of a casein kinase II associated to the Golgi-derived HAI adaptor complex. J. Biol. Chem. 265, 18833-18842. Moehle, C. M., Tizard, R., Lemmon, S. K., Smart, J., & Jones, E. W. (1987). Protease B of the lysosomelike vacuole of the yeast Saccharomyces cerevisiae is homologous to the subtil isin family of serine proteases. Mol. Cell. Biol. 7,4390-4399. Moehle, C. M., Dixon, C. K., & Jones, E. W. (1989). Processing pathway for protease B of Saccharomyces cerevisiae. J. Cell Biol. 108, 309-324. Nakanishi, H., Brewer, K. A., & Exton, J. H. (1993). Activation of the zeta isozyme of protein kinase C by phosphatidylinositol 3,4,5-trisphosphate. J. Biol. Chem. 268, 13—16. Nothwehr, S. F. & Stevens, T. H. (1994). Sorting of membrane proteins in the yeast secretory pathway. J. Biol. ChQm. 269, 10185-10188. Novick, P., Field, C, & Schekman, R. (1980). Identification of 23 complementation groups required for post-translational events in the yeast secretory pathway. Cell 21, 205-215. Obar, R. A., Collins, C. A., Hammarback, J. A., Shpetner, H. S., & Vallee, R. B. (1990). Molecular cloning of the microtubule-associated mechanochemical enzyme dynamin reveals homology with a new family of GTP-binding proteins. Nature 347, 256-261. Ohya, Y, Ohsumi, Y, & Anraku, Y (1986). Isolation and characterization of Ca ^-sensitive mutants of Saccharomyces cerevisiae. J. Gen. Microbiol. 132, 979-988. Okada, T., Kawano, Y, Sakakibara, T., Hazeki, O., & Ui, M. (1994). Essential role of phosphatidylinositol 3-kinase in insulin-induced glucose transport and antilipolysis in rat adipocytes—studies with a selective inhibitor wortmannin. J. Biol. Chem. 269, 3568-3573. Otsu, M., Hiles, I., Gout, I., Fry, M. J., Ruiz-Larrea, F., Panayotou, G., Thompson, A., Dhand, R., Hsuan, J., Totty, N., Smith, A. D., Morgan, S. J., Courtneidge, S. A., Parker, R J., & Waterfield, M. D. (1991). Characterization of two 85 kd proteins that associate with receptor tyrosine kinases, middle-T/pp60c-5rc complexes, and PI3-kinase. Cell 65, 91-104.
160
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
Oyler, G. A., Higgins, G. A., Hart, R. A., Battenberg, E., Billingsley, M., Bloom, F. E., & Wilson, M. C. (1989). The identification of a novel synaptosomal-associated protein, SNAP-25, differentially expressed by neuronal subpopulations. J. Cell Biol. 109, 3039-3052. Paravicini, G., Horazdovsky, B. F., & Emr, B. F. (1992). Alternative pathways for the sorting of soluble vacuolar proteins in yeast: A vps35 null mutant missorts and secretes only a subset of vacuolar hydrolases. Mol. Biol. Cell 3,415-427. Payne, G. S., Hasson, T. B., Hasson, M. S., & Schekman, R. (1987). Genetic and biochemical characterization of cidXhrm-dtTxcxQwi Saccharomyces cerevisiae. Mol. Cell. Biol. 7, 3888-3898. Payne, G. S., Baker, D., van Tuinen, E., & Schekman, R. (1988). Protein transport to the vacuole and receptor-mediated endocytosis by clathrin heavy chain-deficient yeast. J. Cell. Biol. 106, 1453— 1461. Pearse, B. M. & Robinson, M. S. (1990). Clathrin, adaptors, and sorting. Annu. Rev. Cell Biol. 6, 151-171. Pevsner, J., Shu-Chan, H., & Scheller, R. H. (1994). n-Secl: A neural-specific syntaxin-binding protein. Proc. Natl. Acad. Sci. USA 91, 1445-1449. Pfeffer, S. (1992). GTP-binding proteins in intracellular transport. Trends Cell Biol. 2, 41^6. Powers, S. (1991). Protein prenylation: A modification that sticks. Curr. Biol. 1, 114—116. Pryer, N. K., Wuestehube, L. J., & Schekman, R. (1992). Vesicle-mediated protein sorting. Annu. Rev. Biochem. 61,471-516. Raymond, C. K., O'Hara, P J., Eichinger, G., Rothman, J. H., & Stevens, T. H. (1990). Molecular analysis of the yeast VPS3 gene and the role of its product in vacuolar protein sorting and vacuolar segregation during the cell cycle. J. Cell Biol. Ill, 877-892. Raymond, C, Howald-Stevenson, I., Vater, C, & Stevens, T. (1992). Morphological classification of the yeast vacuolar proteins sorting mutants: Evidence for a prevacuolar compartment in class E vps mutants. Mol. Biol. Cell 3, 1389-1402. Roberts, C, Pohlig, G., Rothman, J., & Stevens, T. (1989). Structure, biosynthesis, and localization of dipeptidyl aminopeptidase B, an integral membrane glycoprotein of the yeast vacuole. J. Cell Biol. 108, 1363-1373. Roberts, C. J., Raymond, C. K., Yamashiro, C. T, & Stevens, T. H. (1991). Methods for studying the yeast vacuole. Methods Enzymol. 194, 644—661. Roberts, C. J., Nothwehr, S. F., & Stevens, T. H. (1992). Membrane protein sorting in the yeast secretory pathway: Evidence that the vacuole may be the default compartment. J. Cell Biol. 119, 69-83. Robinson, J. S., Klionsky, D. J., Banta, L. M., & Emr, S. D. (1988). Protein sorting in Saccharomyces cerevisiae: Isolation of mutants defective in the delivery and processing of multiple vacuolar hydrolases. Mol. Cell. Biol. 8, 4936-4948. Robinson, J. S., Graham, T. R., & Emr, S. D. (1991). A putative zinc finger protein, Saccharomyces cerevisiae Vpsl 8p, affects late Golgi functions required for vacuolar protein sorting and efficient alpha-factor prohormone maturation. Mol. Cell. Biol. 11, 5813-5824. Robinson, R J., Sontag, J. M., Liu, J. R, Fykse, E. M., Slaughter, C, McMahon, H., & Sudhof, T. C. (1993). Dynamin GTPase regulated by protein kinase C phosphorylation in nerve terminals. Nature 365, 163-166. Rothman, J. E. & Orci, L. (1992). Molecular dissection of the secretory pathway. Nature 355,409-415. Rothman, J. E. & Warren, G. (1994). Implications of the SNARE hypothesis for intracellular membrane topology and dynamics. Curr. Biol. 4, 220-233. Rothman, J. H. & Stevens, T. H. (1986). Protein sorting in yeast: Mutants defective in vacuole biogenesis mislocalize vacuolar proteins into the late secretory pathway. Cell 47, 1041-1051. Rothman, J. H., Hunter, C. R, Vails, L. A., & Stevens, T. H. (1986). Overproduction-induced mislocalization of a yeast vacuolar protein allows isolation of its structural gene. Proc. Natl. Acad. Sci. USA 83, 3248-3252. Rothman, J. H., Howald, I., & Stevens, T. H. (1989). Characterization of genes required for protein sorting and vacuolar function in the yeast Saccharomyces cerevisiae. EMBO J. 8, 2057—2065.
Protein Sorting to the Yeast Vacuole
161
Rothman, J. H., Raymond, C. K., Gilbert, T., O'Hara, R J., & Stevens, T. H. (1990). A putative GTP binding protein homologous to interferon-inducible Mx proteins performs an essential function in yeast protein sorting. Cell 61,1063-1074. Salminen, A. & Novick, R J. (1987). A ras-like protein is required for a post-Golgi event in yeast secretion. Cell 49, 527-538. Scaife, R., Gout, I., Waterfield, M. D., & Margolis, R. L. (1994). Growth factor-induced binding of dynamin to signal transduction proteins involves sorting to distinct and separate proline-rich dynamin sequences. EMBO J. 13, 2574-2582. Schimmoller, R & Riezman, H. (1994). Involvement of Ypt7p, a small GTPase, in traffic from late endosome to the vacuole in yeast. J. Cell Sci. 106, 823-830. Schu, R v., Takegawa, K., Fry, M. J., Stack, J. H., Waterfield, M. D., & Emr, S. D. (1993). Phosphatidyl inositol 3-kinase encoded by yeast VPS34 gene essential for protein sorting. Science 260, 88-91. Schwaiger, H., Hasilik, A., von Figura, K., Wiemken, A., & Tanner, W. (1982). Carbohydrate-free carboxypeptidase Y is transferred into the lysosome-like yeast vacuole. Biochem. Biophys. Res. Commun. 104,950-956. Schwaninger, R., Plutner, H., Bokoch, G., & Balch, W. (1992). Multiple GTP-binding proteins regulate vesicular transport from the ER to Golgi membranes. J. Cell Biol. 119, 1077-1096. Seeger, M. & Payne, G. (1992a). A role for clathrin in the sorting of vacuolar proteins in the Golgi complex of yeast. EMBO J. 11, 2811-2818. Seeger, M. & Payne, G. S. (1992b). Selective and immediate effects of clathrin heavy chain mutations on Golgi membrane protein retention in Saccharomyces cerevisiae. J. Cell Biol. 118, 531—540. Segev, N., Mulholland, J., & Botstein, D. (1988). The yeast GTP-binding Yptl protein and a mammalian counterpart are associated with the secretion machinery. Cell 52, 915-924. Shaw, J. M. & Wickner, W. T. (1991). vac2: A yeast mutant which distinguishes vacuole segregation from Golgi-to-vacuole protein targeting. EMBO J. 10, 1741-1748. Sheetz, M. P. & Singer, S. J. (1974). Biological membranes as bilayer couples. A molecular mechanism of drug-erythrocyte interactions. Proc. Natl. Acad. Sci. USA 71, 4457—4461. Shpetner, H. S. & Vallee, R. B. (1992). Dynamin is a GTPase stimulated to high levels of activity by microtubules. Nature 355, 733-735. Singer-Kruger, B., Stenmark, H., Dusterhoft, A., Philippsen, R, Yoo, J. S., Gallwitz, D., & Zerial, M. (1994). Role of three rab5-like GTPases, Ypt51p, Ypt52p, and Ypt53p, in the endocytic and vacuolar protein sorting pathways of yeast. J. Cell Biol. 125, 283-298. Skolnik, E. Y, Margolis, B., Mohammadi, M., Lowenstein, E., Fischer, R., Drepps, A., Ullrich, A., & Schlessinger, J. (1991). Cloning of PI3 kinase-associated p85 utilizing a novel method for expression/cloning of target proteins for receptor tyrosine kinases. Cell 65, 83-90. SoUner, T., Bennett, M. K., Whiteheart, S. W., Scheller, R. H., & Rothman, J. E. (1993a). A protein assembly-disassembly pathway in vitro that may correspond to sequential steps of synaptic vesicle docking, activation, and fusion. Cell 75, 409-418. Sollner, T, Whitehart, S. W., Brunner, M., Erdjumentbromage, H., Geromanos, S., Tempst, P., & Rothman, J. E. (1993b). SNAP receptors implicated in vesicle targeting and fusion. Nature 362, 318-324. Spormann D. O., Heim, J., & Wolf, D. H. (1991). Carboxypeptidase yscS: Gene structure and function of the vacuolar enzyme. Eur. J. Biochem. 197, 399-405. Spormann, D. O., Heim, J., & Wolf, D. H. (1992). Biogenesis of the yeast vacuole (lysosome). The precursor forms of the soluble hydrolase carboxypeptidase yscS are associated with the vacuolar membrane. J. Biol. Chem. 267, 8021-8029. Stack, J. H. & Emr, S. D. (1994). Vps34p required for yeast vacuolar protein sorting is a multiple specificity kinase that exhibits both protein kinase and phosphatidylinositol-specific PI 3-kinase activities. J. Biol. Chem. 269, 31552-31562.
162
BRUCE F. HORAZDOVSKY, JEFFREY H. STACK, and SCOTT D. EMR
Stack, J. H., Herman, P. K., Schu, P. V., & Emr, S. D. (1993). A membrane-associated complex containing the Vpsl5 protein kinase and the Vps34 PI 3-kinase is essential for protein sorting to the yeast lysosome-like vacuole. EMBO J. 12, 2195-2204. Stack, J. H., DeWald, D. B., Takegawa, K., & Emr, S. D. (1995). Vesicle-mediated protein transport: regulatory interactions between the Vpsl5 protein kinase and the Vps34 Ptdlns 3-kinase essential for protein sorting to the vacuole in yeast. J. Cell Biol. 129, 321—334. Stephens, L., Cooke, F. T., Walters, R., Jackson, T., Volina, S., Gout, I., Waterfield, M. D., & Hawkins, P. T. (1994a). Characterization of a phosphatidylinositol-specific phosphoinositide 3-kinase from mammalian cells. Curr. Biol. 4, 203-214. Stephens, L., Smrcka, A., Cooke, P., Jackson, T. R., Stemweis, P. C, & Hawkins, P. T. (1994b). A novel phosphoinositide-3-kinase activity in myeloid-derived cells is activated by G protein beta gamma subunits. Cell 77, 83-93. Stevens, T., Esmon, B., & Schekman, R. (1982). Early stages in the yeast secretory pathway are required for transport of carboxypeptidase Y to the vacuole. Cell 30, 439-448. Stevens, T. H., Rothman, J. H., Payne, G. S., & Schekman, R. (1986). Gene dosage-dependent secretion of yeast vacuolar carboxypeptidase Y. J. Cell Biol. 102, 1551-1557. Tan, P K., Davis, N. G., Sprague, G. P., «fe Payne, G. S. (1993). Clathrin facilitates the internalization of 7 transmembrane segment receptors for mating pheromones in yeast. J. Cell Biol. 123, 1707-1716. Thelen, M., Wymann, M. P., & Langen, H. (1994). Wortmannin binds specifically to 1-phosphatidylinositol 3-kinase while inhibiting guanine nucleotide-binding protein-coupled receptor signaling in neutrophil leukocytes. Proc. Natl. Acad. Sci. US A 91,4960-4964. Thomason, P. A., James, S. R., Casey, V. J., & Downes, C. P. (1994). A G-protein py-subunit-responsive phosphoinositide 3-kinase activity in human platelet cytosol. J, Biol. Chem. 269, 16525—16528. Trimble, W. S., Cowan, D. M., & Scheller, R. H. (1988). VAMP-1: Asynaptic vesicle-associated integral membrane protein. Proc. Natl. Acad. Sci. USA 85, 4538-4542. Tuma, P. L., Stachniak, M. C, & Collins, C. A. (1993). Activation of dynamin GTPase by acidic phospholipids and endogenous rat brain vesicles. J. Biol. Chem. 268, 17240-17246. Vails, L. A., Hunter, C. R, Rothman, J. H., & Stevens, T. H. (1987). Protein sorting in yeast: The localization determinant of yeast vacuolar carboxypeptidase Y resides in the propeptide. Cell 48, 887-897. Vails, L. A., Winther, J. R., & Stevens, T. H. (1990). Yeast carboxypeptidase Y vacuolar targeting signal is defined by four propeptide amino acids. J. Cell Biol. Ill, 361-368. van den Hazel, H. B., Kielland-Brandt, M. C, & Winther, J. R. (1993). The propeptide is required for in vivo formation of stable active yeast proteinase A and can function even when not covalently linked to the mature region. J. Biol. Chem. 268, 18002-18007. van der Bliek, A. M. & Meyerowitz, E. M. (1991). Dynamin-like protein encoded by the Drosophila shibire gene associated with vesicular traffic. Nature 351, 411-414. van der Bliek, A. M., Redelmeier, T. E., Damke, H., Tisdale, E. J., Meyerowitz, E. M., & Schmid, S. L. (1993). Mutations in human dynamin block an intermediate stage in coated vesicle formation. J. Cell Biol. 122,553-563. Vater, C. A., Raymond, C. K., Ekena, K., Howaldstevenson, I., & Stevens, T H. (1992). The Vpsl protein, a homolog of dynamin required for vacuolar protein sorting in Saccharomyces cerevisiae, is a GTPase with 2 functionally separable domains. J. Cell Biol. 119, 773—786. Vida, T. A., Huyer, G., & Emr, S. D. (1993). Yeast vacuolar proenzymes are sorted in the late Golgi complex and transported to the vacuole via a prevacuolar endosome-like compartment. J. Cell Biol. 121, 1245-1256. Voglmaier, S. M., Keen, J. H., Murphy, J. E., Ferris, C. D., Prestwich, G. D., Snyder, S. H., & Theibert, A. B. (1992). Inositol hexakisphosphate receptor identified as the clathrin assembly protein AP-2. Biochem. Biophys. Res. Commun. 187, 158-163.
Protein Sorting to the Yeast Vacuole
163
Walworth, N. C, Brennwald, P., Kabcenell, A. K., Garrett, M., & Novick, R (1992). Hydrolysis of GTP by Sec4 protein plays an important role in vesicular transport and is stimulated by a GTPase-activating protein in Saccharomyces cerevisiae. Mol. Cell. Biol. 12, 2017-2028. Weisman, L. S. & Wickner, W. (1988). Intervacuole exchange in the yeast zygote: A new pathway in organelle communication. Science 241, 589-591. Weisman, L. S. & Wickner, W (1992). Molecular characterization oi VAC 1,3. gene required for vacuole inheritance and vacuole protein sorting. J. Biol. Chem. 267, 618-623. Weisman, L. S., Bacallao, R., & Wickner, W. (1987). Multiple methods of visualizing the yeast vacuole permit evaluation of its morphology and inheritance during the cell cycle. J. Cell Biol. 105, 1539-1547. Weisman, L. S., Emr, S. D., & Wickner, W. T. (1990). Mutants of Saccharomyces cerevisiae that block intervacuole vesicular traffic and vacuole division and segregation. Proc. Natl. Acad. Sci. USA 87, 1076-1080. Wichmann, H., Hengst, L., & Gallwitz, D. (1992). Endocytosis in yeast: Evidence for the involvement of a small GTP-binding protein (Ypt7p). Cell 71, 1131-1142. Wilsbach, K. & Payne, G. S. (1993a). Dynamic retention of TGN membrane proteins in Saccharomyces cerevisiae. Trends Cell Biol. 3,426-432. Wilsbach, K. & Payne, G. S. (1993b). Vpslp, a member of the dynamin GTPase family, is necessary for Golgi membrane protein retention in Saccharomyces cerevisiae. EMBO J. 12, 3049-3059. Winther, J., Steven, T., & Kielland-Brandt, M. (1991). Yeast carboxypeptidase Y requires glycosylation for efficient intracellular transport, but not for vacuolar sorting, in vivo stability, or activity. Eur. J. Biochem. 197,681-689. Woolford, C. A., Daniels, L. G., Park, R J., Jones, E. W, van Arsdell, J. N., & Innis, M. A. (1986). The PEP4 gene encodes an aspartyl protease implicated in the posttranslational regulation of Saccharomyces cerevisiae vacuolar hydrolases. Mol. Cell. Biol. 6, 2500-2510. Woolford, C. A., Dixon, C. K., Manolson, M. R, Wright, R., & Jones, E. W. (1990). Isolation and characterization oi PEPS, a gene essential for vacuolar biogenesis in Saccharomyces cerevisiae. Genetics 125, 739-752. Yeh, E., Driscoll, R., Coltrera, M., Olins, A., & Bloom, K. (1991). A dynamin-like protein encoded by the yeast sporulation gene SP015. Nature 349, 713-715. Yoshihisa, T. & Anraku, Y. (1990). A novel pathway of import of alpha-mannosidase, a marker enzyme of vacuolar membrane, in Saccharomyces cerevisiae J. Biol. Chem. 265, 22418—22425. Yung, B. Y & Komberg, A. (1988). Membrane attachment activates dnaA protein, the initiation protein of chromosome replicatin in Escherichia coli. Proc. Natl. Acad. Sci. USA 85, 7202-7205.
This Page Intentionally Left Blank
BACTERIAL EXTRACELLULAR SECRETION: TRANSPORT OF a-LYTIC PROTEASE ACROSS THE OUTER MEMBRANE OF ESCHERICHIA COU
Amy Fujishige Boggs and David A. Agard
I. II. III. IV.
Introduction Leaving the Cytoplasm One-step Versus Two-Step Mechanism Pro Region-Dependent Secretion A. Background B. a-Lytic Protease as a Model System . C. Folding and Extracellular Secretion D. Signals Targeting Extracellular Secretion E. Kineticsof Two-Step Heterologous Secretion of a-Lytic Protease . . . . IV. Concluding Remarks Acknowledgments References
Membrane Protein Transport Volume 3, pages 165-179. Copyright © 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 1-55938-989-3 165
166 167 167 168 168 169 169 171 172 175 177 177
166
AMY FUJISHIGE BOGGS and DAVID A. AGARD
I. INTRODUCTION Understanding extracellular secretion in Gram-negative bacteria presents a multilayered challenge, for the extracellularly secreted protein must traverse the inner membrane, the periplasmic space, and the outer membrane. Although paradigms for each step are still being established and challenged, advances in this field have enabled us to define some major questions. First, to what extent do exoproteins and periplasmic proteins use the same mechanism in crossing the inner membrane? Among exoproteins, examples of signal sequence-dependent and signal sequenceindependent transport exist. Another issue is whether the exoprotein requires participation of the Sec proteins in the first stages of export. A second question is whether a particular protein traverses the inner and outer membranes in a single step (going directly from the cytoplasm through both membranes), or in two steps (going through the cytoplasmic membrane, into the periplasmic space, and then subsequently through the outer membrane). Have these two mechanisms evolved independently or do they share common protein machinery? The brief stopover in the periplasm that occurs in the two-step mechanism may be helpful to certain proteins for efficient acquisition of native structures. For example, catalyzed post-translational processing such as disulfide bond formation takes place in the periplasm (reviewed by Wtilfing and Pliickthun, 1994). The third question is implicit for all proteins using the two-step mechanism: Do these exoproteins fold in the periplasm or in the extracellular milieu? There is a steady trickle of evidence that some folding of exoproteins actually does occur in the periplasm. If so, then the mechanism of export through the outer membrane is intrinsically different from that for transport across the cytoplasmic membrane, which requires that proteins be maintained in an unfolded configuration. Finally, the cues that target proteins beyond the periplasm into or across the outer membrane have yet to be identified. What are the signals that direct an exoprotein to be exported by the one-step or two-step pathway? If proteins using the two-step mechanism do fold in the periplasm, this presents the possibility that the targeting cues may be three-dimensional rather than linear. Accessory factors required for the extracellular secretion of certain exoproteins have been identified, and these factors may recognize such three-dimensional cues in the exoprotein to assist in the process of crossing the outer membrane. Although the Sec proteins have been shown to help a wide variety of proteins cross the inner membrane, and the ATP-binding cassette (ABC) transporter system may be somewhat interchangeable for exoproteins using the one-step system, interchangeability of accessory factors has not been demonstrated for proteins using the two-step mechanism. In this chapter we explore these four questions, present the answers we have obtained using a-lytic protease as a model system, and suggest some areas warranting further investigation.
Secretion ofa-Lytic Protease by E. coli
167
IL LEAVING THE CYTOPLASM In both eukaryotic and prokaryotic systems, most secreted proteins appear to be translocated from the cytoplasm in a signal-sequence-dependent fashion (Blobel, 1980; Walter et al., 1984; Walter and Lingappa, 1986). Although there seem to be many mechanistic similarities between eukaryotes and prokaryotes, there are clear differences in both efficiency and stringency. In Escherichia coli, secretion usually depends on the members of the Sec protein family. These well-studied processes have been covered in numerous reviews and in other chapters, and will not be included here. It should be noted that there are a few phage proteins (Wickner, 1979) and the heterologously secreted prepromelittin (Cobet et al., 1989), and perhaps others, whose secretion is fully or partially Sec-independent. Interestingly, in bacteria some exotoxins (e.g., hemolysin, leukotoxin, cyclolysin) appear to cross the inner membrane independently of an NH2-terminal signal sequence (Felmlee et al., 1985a,b; Glaser et al., 1988; Strathdee and Lo, 1989). A COOH-terminal signal has been identified in hemolysin (Nicaud et al., 1986; Mackman et al., 1986; Koronakis et al., 1989) as well as in the extracellular proteases B and C of Erwinia chrysanthemi (Delepelaire and Wandersman, 1990). These toxins and proteases use the ABC transporter, and in each case an accessory lipoprotein (e.g., HlyD, LktD, CyaD, PrtE) belonging to a subfamily of the putative membrane fusion protein (MFP) family (Dinh et al., 1994) is also required. Exoproteins of this class can apparently use the HlyBD-TolC system for extracellular secretion when heterologously expressed in E. coli (Mackman et al., 1987; Strathdee and Lo, 1989; Masure et al., 1990). The degree of efficiency of export in these mixed systems appears to be related to the degree of homology of the lipoprotein component to HlyD, rather than to the similarity of the COOH-terminal signals.
111. ONE-STEP VERSUS TWO-STEP MECHANISM There appear to be at least two different mechanisms by which proteins can exit the cytoplasm of the Gram-negative cell. Not coincidentally, there appear to be at least two pathways by which the exoprotein may then proceed. Several of the toxins and proteases mentioned above appear to cross the inner and outer membranes in a single step (Filloux et al., 1990; Mackman et al., 1985; Wandersman and Delepelaire, 1990) at Bayer junctions, which are areas of close apposition between the two membranes. In this case, the ATPase activity of the ABC transporter seems to provide the necessary energy. Other proteins have been shown to cross the two membranes separately in what is called the two-step mechanism (Sen and Nikaido, 1990; Hirst and Holmgren, 1987; Pugsley, 1992). In this mechanism, proteins are first translocated across the cytoplasmic membrane in a signal sequence- and Sec-dependent manner, and only subsequently are inserted into or translocated across the outer membrane. This
168
AMY FUJISHIGE BOGGS and DAVID A. AGARD
two-step pathway is also referred to as the extended general secretory pathway (Pugsley, 1993). Little is known about the energy requirements for two-step extracellular secretion. In the case of secretion of aerolysin from Aeromonas salmonicida, translocation from the periplasm to the extracellular medium was found to be inhibited by low pH and by the proton ionophore, carbonyl cyanide w-chlorophenyl hydrazone (CCCP), indicating the possible involvement of a proton motive force (Wong and Buckley, 1989). Although it is difficult to imagine that a proton gradient could be maintained across the outer membrane, it is possible that the energy generated by the proton motive force across the inner membrane is harvested and coupled through accessory factors to facilitate two-step secretion. The demonstration that 14 genes of thepw/ operon are required for the extracellular secretion of the Klebsiella pneumoniae enzyme pullulanase (Pugsley et al., 1990) highlights the potential complexity of extracellular transport systems. However, the identification of significant homology to the pul genes among genes from more than a dozen Gram-negative species (Pugsley, 1993), as well as the observation of a separate set of Gram-negative interspecies homologies among genes of the putative MFP family (Dinh et al., 1994), suggests that at least some aspects of extracellular transport from Gram-negative bacteria may share common mechanisms. Although in each case homologous genes may be performing analogous roles in targeting or transporting a specific exoprotein, we only have evidence that one set can be exchanged for another in the case of the ABC transporter/MFP family.
IV. PRO REGION-DEPENDENT SECRETION A. Background
It appears that the majority, if not all, of extracellular bacterial proteases are synthesized as preproproteins. Where examined, secretion of such proteins to the extracellular medium appears to be dependent on the presence of the signal (pre) sequence as well as the pro region. Interestingly, in several cases, the presence of the signal sequence and pro region appeared to be sufficient to target the protein to the extracellular milieu when heterologously expressed in E. coli. The Serratia marcescens serine protease is synthesized as a 112-kDa preproenzyme, whose amino-terminal signal sequence and 52-kDa carboxy-terminal pro region are cleaved during export through the inner and outer membranes, respectively (Miyazaki et al., 1989). Similarly, aqualysin I is produced by Thermophilus aquaticus with an amino-terminal signal sequence and pro region in addition to a large carboxy-terminal pro region (Terada et al., 1990). Both of these proteases can be expressed in E. coli and, under proper conditions, become processed and secreted. For the IgA protease of Neisseria gonorrhoeae^ it has been proposed that the pro region forms a pore in the outer membrane through which the protease is translo-
Secretion ofa-Lytic Protease by E. coli
169
cated. Following translocation, the mature enzyme is released from the pro region by proteolysis (Pohlner et al., 1987). B. a-Lytic Protease as a Model System
We have studied the folding and extracellular secretion of a-lytic protease, a serine protease from Lysobacter enzymogenes (Whitaker, 1970) that has been used as a model system for extensive studies of serine protease mechanism (Hunkapiller et al., 1973; Bachovchin et al., 1988; Bone et al., 1987) and for understanding the structural basis of enzyme-substrate specificity (Bone et al., 1989a,b; 1991). Like the extracellularly secreted proteases mentioned above, a-lytic protease bears a large pro region and a short signal peptide. In initial work, heterologous expression of prepro-a-lytic protease in E. coli led to the production of active, mature a-lytic protease in the extracellular medium (Silenetal., 1988,1989). Further experiments in £. co//revealed that the 166-amino acid pro region plays an obligatory role in the folding of the 198-amino acid protease domain. Constructs lacking the pro region (A-pro-a-lytic protease) produce inactive a-lytic protease (Silen et al., 1989). Remarkably, it is possible to complement the folding defect of this A-pro molecule by in vivo coexpression of the pro region in trans (Silen and Agard, 1989). From this work, it was clear that the pro region plays a fundamental role in both the folding and secretion of active a-lytic protease. Similarly, in vitro experiments demonstrate that chemically denatured a-lytic protease can only be refolded in the presence of the pro region (Baker et al., 1992a). As in the in vivo system, the pro region can affect folding in either the presence or the absence of a covalent attachment to the protease region. Based on these in vitro studies, it is clear that the pro region is both necessary and sufficient to promote the folding of the protease domain; no other factors or sources of energy are required. Furthermore, detailed kinetic analysis reveals that the pro region functions by directly catalyzing the folding reaction (Baker et al., 1992b). This contrasts with the molecular chaperones, which facilitate folding by blocking off-pathway reactions such as aggregation (see discussion in Agard, 1993). Based on data from other systems, it now seems that the mechanism of pro-assisted folding is remarkably conserved even for proteins that are not evolutionarily related. (For more detailed reviews of pro-region-dependent folding of a-lytic protease see Baker et al., 1993; Sohl and Agard, 1995.) C. Folding and Extracellular Secretion
Recent studies suggest that certain proteins may fold in the periplasm before inserting into or crossing the outer membrane (Sen and Nikaido, 1990; Pugsley, 1992). In the heterologous expression of the extracellularly secreted a-lytic protease, we found that only properly folded forms of this protease arrived in the external medium. What is the relationship between folding and secretion?
170
AMY FUJISHIGE BOGGS and DAVID A. AGARD
The temperature sensitivity of the in vivo production of a-lytic protease in E. coli and the availability of mutants have aided us in dissecting the process of extracellular secretion. The native host, L. enzymogenes, is not viable above 32°C. When wild-type pre-pro-a-lytic protease is expressed in E. coli at temperatures below 30°C, the mature (proteolytically processed and active) protease first appears in the periplasm and then accumulates in the medium. When the same construct is expressed at temperatures above 30°C, an inactive precursor accumulates in the cells (Silen et al., 1989). In vitro experiments reveal that this temperature sensitivity results from a great retardation in the folding process at temperatures above 30°C due to thermal instability of the pro region (Boggs and Agard, manuscript in preparation). Expression of pre-pro-a-lytic protease that has been inactivated by mutation of the catalytic serine residue to an alanine (proalp(SA195)) results in a similar accumulation of cell-associated precursor, even at permissive temperatures, indicating that a-lytic protease is self-processing in E. coli. Similarly, the misfolded A-pro-a-lytic protease is also found to be cell-associated, independent of expression temperature (Silen et al., 1989). Careful cell fractionation analysis indicates that all misfolded or improperly processed forms remained engaged on the periplasmic side of the outer membrane (Fujishige et al., 1992). That is, the A-pro-a-lytic protease, the mutant inactive precursor, and the wild-type precursor synthesized at the restrictive temperature are all tightly associated with the E. coli outer membrane (Fig. 1). These outer membrane-bound forms of a-lytic protease and its precursors can only be released by conditions that disrupt the outer membrane (Fujishige et al., 1992). In addition, extensive deletion analysis of the pro region revealed that the folding and the secretion functions are not easily separable and that only proteolytically active molecules are efficiently secreted across the outer membrane. A folded but inactive protease produced by mutation of the protease active site in complementation with a wild-type pro region (pro:alp(SA195), Fig. 1) is efficiently transported to the medium, indicating that enzyme activity p^r se is not required for secretion. Given this seemingly inseparable linkage between proper folding and secretion, we proposed that, unlike translocation through the cytoplasmic or the mitochondrial membrane, which requires that the nascent protein be in an unfolded state (Randall and Hardy, 1986; Filers and Schatz, 1986), efficient translocation through the outer membrane appears to require that the protein be correctly folded (Fujishige et al., 1992). Thus, the pro region requirement for extracellular secretion is an indirect consequence of it being necessary for proper folding. The fact that a folded protein can cross the outer membrane presents new possibilities for mechanisms of protein translocation.
Transport ofa-Lytic Protease by E. coli
^ 71 Expres^n Temperature
pre
pro
proalp
Extracellular Secretion
a-lytic protease
proalp
alp
Activity
\y/ya
^^^2HHHHHI
~]
22X
+
+
22X
+
+
22'c
I
37'C
serl95 ->ala proalp (SAT 95)
22X ser195 ->ala
(SA195)
l 2 2 2 2 H i M M M Mi
\2ZA
1
22 C
+
Figure 1. Characteristics of expression constructs. Cells and media were subjected to a-lytic protease assays as well as immunoblot analyses. The wild-type construct (proalp) expressed at permissive temperatures allows proper folding of the protease and cleavage of the pro region. a-Lytic protease activity is required for the cleavage. Both the pro region and the mature protease are secreted into the medium. Physical linkage of the pro region is not required for proper folding and secretion, as shown by the complementation construct proialp. Deletion of the pro region (alp) results in an inactive, eel I-associated molecule. Uncleaved precursors accumulate in the cell upon expression of wild type at nonpermissive temperatures (proalp at 37°C) or by mutation of the active site serine (proalp(SA195)) with growth at permissive temperatures. Mutation of the active site serine in the complementation construct (pro:alp(SA195)) allows secretion of the mature, folded, but inactive protease region.
D. Signals Targeting Extracellular Secretion Unfortunately, very little is currently knov^n about the signals that target a protein for extracellular secretion. Unlike targeting signals for secretion across the cytoplasm or for targeting into various cellular compartments in eukaryotes, linear targeting signals have not been identified for extracellular secretion in Gram-negative organisms. One possibility is that the targeting signals may be contained in a tertiary structural epitope, rather than in a linear sequence of amino acids. Supporting this hypothesis is our discovery that a-lytic protease must be properly folded before it can exit the periplasm. On the other hand, the fact that misfolded forms of a-lytic protease are tightly associated w^ith the outer membrane suggests that, even though they are misfolded, these species contain valid targeting signals. Their failure to be secreted may not reflect a lack of targeting cues, but may imply that translocation can only be completed for correctly folded proteins.
172
AMY FUJISHIGE BOGGS and DAVID A. AGARD
E. Kinetics of Two-Step Heterologous Secretion of a-Lytic Protease In the folded state, a-lytic protease has three disulfide bonds, which are presumably formed after the protein leaves the cytoplasm. The discovery of periplasmic enzymes that catalyze disulfide bond formation (Bardwell et al., 1991; Shevchik et al., 1994; Missiakis et al., 1994) suggested that the two-step pathway of secretion may hold advantages for a-lytic protease. Using pulse-chase techniques, we have confirmed that a-lytic protease uses the two-step mechanism of export when heterologously expressed in E. coli (Boggs and Agard, manuscript submitted; Fig. 2). These experiments also allowed us to examine the kinetics of export across the outer membrane. Although there are few data available for comparison, this process appears to be quite slow for a-lytic protease in E. coli (r,/2 = 4 h). By contrast, similar experiments in the native host, Lysobacter enzymogenes, reveal that the entire secretion process is very rapid (ty2 ^ 3 min; Fig. 3). The half-time for extracellular secretion of a-lytic protease in E. coli is considerably longer than has been observed for homologous extracellular transport in the few other systems that have been characterized: enterotoxin is secreted across the outer membrane of Vibrio cholerae with a half-time o f - 1 3 minutes (Hirst and Holmgren, 1987); export of aerolysin from Aeromonas salmonicida takes between 2 and 15 minutes (Wong et al., 1989; Wong and Buckley, 1989); pectate lyase is exported from Erwinia chrysanthemi in less than 1 minute (He et al., 1991). Our Lysobacter data {1^/2 = 3 min) are quite consistent with the data from other homologous systems. In one clear case of reconstituted heterologous expression, the enzyme pullulanase, which is native to Klebsiella pneumoniae, was secreted from E. coli in less than 5 minutes (Pugsley, 1992). However, 14 gene products are known to be required for this process (Possot et al., 1992). The data described above make clear that, although prepro-a-lytic protease carries the necessary information for extracellular secretion, this process is not optimized in E. coli. There are several explanations for the apparent slow export observed for a-lytic protease in E. coli. One possibility is that the observed secretion is the result of either cell lysis or leakage through the outer membrane. Cell viability data provide no indication of the massive cell lysis that would be required to explain the high levels of a-lytic protease secreted into the media. Although studies on (3-lactamase and alkaline phosphatase localization suggest that there is detectable leakage, a-lytic protease secretion into the medium is approximately 10 times more efficient (Boggs et al., unpublished results). A second possibility is that the expression system used for the pulse-chase experiments, which requires phosphate depletion of the cells, may affect the process. Arguing against this explanation is that the gross kinetics of export of a-lytic protease observed for plasmids with other promoters (cells grown in rich
5' Ih 3h ■■:■
M^,
5h 8h l l h
5c*
periplasm
medium
B
800001
o o
60000
TJ (D
^CO
Q. 40000 'o Q.
O
i 20000 E
time (hours) Figure 2. Pulse-chase of a-lytic protease in £ coli. (A) Cultures were labeled for 5 minutes with [^^S]methionine and chased with cold methionine in fresh media. At the indicated times, aliquots were removed and fractionated to obtain periplasmic and extracellular medium samples. Each fraction was immunoprecipitated with anti-alytic protease antibody on immobilized protein A agarose beads. After SDS-PAGE, the bands were visualized by phosphorimaglng. (B) Quantitation of bands shown in A. Periplasmic concentrations are shown as circles, and triangles denote medium concentration. 173
r
5'
10'
15' 20' 30'
cells
medium B
250000
0)
c D o o "D 0
200000 -
.
150000-
CO Q.
o
P Q.
O C D
b
E
100000-
■ ~
—9^
50000:
time (min) Figure 3. Pulse-chase of a-lytic protease in L. enzymogenes. (A) Cultures were labeled for 1 minute and chased with cold methionine. At the indicated times, aliquots were removed and the cells and media were separated by centrifugation. Soluble cell contents were collected by a combination of export into fresh medium and lysis. Proteins were immunoprecipitated with anti-a-lytic protease antibody on immobilized protein A agarose beads. After SDS-PAGE, the bands were visualized by phosphorimager, shown above. (B) Quantitation of the bands shown in (A). Periplasmic concentrations are shown as circles, and triangles denote medium concentration. The dashed lines show the half-time of export. 174
Secretion ofa-Lytic Protease by E. coli
1 75
media) are similar to those observed with this system (Agard, unpublished observations). Third, given the temperature sensitivity of folding of a-lytic protease (Silen et al., 1989; Boggs et al., manuscript in preparation), these experiments were performed at 22°C. Although the physical impact of lower temperature should only account for a factor of two or three in rates, it is completely possible that the biological impact is greater than this. The doubling time is indeed longer at 22°C, but no obvious indications of detrimental changes in cell physiology were observed (data not shown). Notably, autoprocessing of the precursor to the mature form occurs within 15 seconds (Boggs, unpublished observations), indicating that this is not the rate-limiting step of heterologous secretion of a-lytic protease. A fourth, and perhaps most likely explanation for the slow export observed in E. coli is the possibility that the native host, L. enzymogenes, provides specific accessory proteins to aid in transport across the outer membrane. If these factors are missing in E. coli, export would be quite inefficient. In many ways it is remarkable that a-lytic protease is exported at all from E. coli. The organization of the genes for the accessory factors for pullulanase into a large operon suggests that it might be possible to conveniently find the relevant factors for a-lytic protease near the structural gene. Although these factors are not strictly required, they would be expected to enhance a-lytic protease secretion from E. coli.
IV. CONCLUDING REMARKS Advances in our understanding of extracellular secretion point to two major pathways by which proteins exit the Gram-negative cell. In the first of these, the exoprotein crosses the inner and outer membrane in a single step, with no pool accumulating in the periplasm. This process may not require an NH2~terminal signal sequence, but the exoprotein may have a signal at its COOH terminus. The second pathway shares processes with export to the periplasm and is currently referred to as the extended general secretion pathway. In Gram-negative bacteria, exoproteins using the extended general secretion pathway first cross the inner membrane in a signal-sequence-dependent and probably Sec-dependent manner. The signal sequence is most probably cleaved by signal peptidase as the protein enters the periplasm. Once in the periplasm, it appears that the protein folds and post-translational processing occurs (such as cleavage of the juncture separating the pro region from the mature protease in a-lytic protease). Subsequently, another targeting cue, possibly three-dimensional, is recognized by factors that are not yet characterized but which may assist the exoprotein in crossing the outer membrane (Fig. 4). Certainly in the case of a-lytic protease, but likely in most other cases, the pro region of a pre-pro-protein is required for production of active protein in the media. The a-lytic protease data developed from both in vivo and in vitro studies indicate that the primary pro region requirement is to allow folding to occur in the
176
AMY FUJISHIGE BOGGS and DAVID A. AGARD
proteolytic digestion of pro
Figure 4. Maturation of a-lytic protease. In step 1 the preproenzyme is transported across the inner membrane to the periplasmic space, where the signal (pre) sequence is removed. Upon proper folding, the precursor is cleaved between the pro region and the protease region. a-Lytic protease activity is required for the cleavage. The two regions have a high affinity for one another and therefore probably remain as a complex while they are conveyed across the outer membrane to the medium during step 2. Over a period of time, the pro region is further degraded, leaving the mature protease. If the protein does not fold correctly or cleavage between the pro region and the protease region does not occur, the misfolded/precursor form becomes tightly associated with the outer membrane.
periplasmic space. The apparent necessity of the pro region for secretion is then a consequence of the need to first fold the protein. Based on our data with a-lytic protease, it appears that proteins must be fully folded and properly processed before they can exit the periplasm. If these steps have not been completed, as we found using mutations or by expression at nonpermissive temperatures, the protein is found tightly associated with the outer membrane. This might be expected if the misfolded protein was able to initiate but not complete translocation across the outer membrane. The question of whether this process relies on the specific transport of properly folded proteins, or the specific retention of misfolded proteins, has yet to be worked out. Certain key questions remain. Do all NH2-terminal signal sequence-bearing exoproteins use the two-step pathway, whereas those that lack this signal use the one-step pathway? Is there a coupled energy source for the second step of the two-step secretion pathway? What are the targeting signals for specifying extracellular secretion using either the one-step or two-step pathway? Despite many years of searching, researchers have not been able to identify a linear sequence that will target proteins to the outer membrane or external medium. This in turn has resulted in the notion that such targeting signals may be tertiary epitopes. And perhaps the
Secretion ofa-Lytic Protease by E. coli
177
most intriguing question is, what is the mechanism whereby a fully folded protein can cross the outer membrane? Clearly much work remains to sort out the myriad complexities of extracellular secretion in Gram-negative bacteria.
ACKNOWLEDGMENTS Original work from the Agard laboratory was supported by the Howard Hughes Medical Institute and the National Science Foundation.
REFERENCES Agard, D. A. (1993). To fold or not to fold.... Science 260, 1903-1904. Bachovchin, W. W., Wong, W. Y. L., Farr-Jones, S., Shenvi, A. B., & Kettner, C. A. (1988). Nitrogen-15 NMR spectroscopy of the catalytic-triad histidine of a serine protease in peptide boronic acid inhibitor complexes. Biochemistry 27, 7689-7697. Baker, D., Silen, J. L., & Agard, D. A, (1992a). A protease pro region required for folding is a potent inhibitor of the mature enzyme. Proteins 12, 339-344. Baker, D., Sohl, J. L., & Agard, D. A. (1992b). A protein-folding reaction under kinetic control. Nature 365, 263-265. Baker, D., Shiau, A. K., & Agard, D. A. (1993). The role of pro regions in protein folding. Curr. Opin. Cell Biol. 5, 966-970. Bardwell, J. C, McGovem, K., & Beckwith, J. (1991). Identification of a protein required for disulfide bond formation in vivo. Cell 67, 581-589. Blobel, G. (1980). Intracellular protein topogenesis. Proc. Nat. Acad. Sci. 77, 1496-1500. Bone, R. B., Shenvi, A. B., Kettner, C. A., & Agard, D. A. (1987). Serine protease mechanism: Structure of an inhibitory complex of a-lytic protease and a tightly bound peptide boronic acid. Biochemistry 26, 7609-7614. Bone, R. B., Frank, D., Kettner, C. A., & Agard, D. A. (1989a). Structural analysis of specificity: a-lytic protease complexes with analogues of reaction intermediates. Biochemistry 28, 7600-7609. Bone, R. B., Silen, J. L., & Agard, D. A. (1989b). Structural plasticity broadens the specificity of an engineered protease. Nature 339, 191—195. Bone, R .B., Fujishige, A., Kettner, C. A., & Agard, D. A. (1991). Structural basis for broad specificity in a-lytic protease mutants. Biochemistry 30, 10388-10398. Cobet, W. W. E., Mollay, C, Muller, G., & Zimmerman, R. (1989). Export of honeybee prepromellittin in Escherichia coli depends on the membrane potential but does not depend on proteins secA or secY. J. Biol. Chem. 264, 10169-10176. Delepelaire, P., & Wandersman, C. (1990). Protein secretion in gram-negative bacteria. The extracellular metalloprotease B from Erwinia chrysanthemi contains a C-terminal secretion signal analogous to that oiEscherichia coli alpha-hemolysin. J. Biol. Chem. 265, 17118-17125. Dinh, T., Paulsen, I. T, & Saier, M. H., Jr. (1994). A family of extracytoplasmic proteins that allow transport of large molecules across the outer membranes of Gram-negative bacteria. J. Bacteriol. 176,3825-3831. Eilers, M. & Schatz, G. (1986). Binding of a specific ligand inhibits import of a purified precursor protein into mitochondria. Nature 322, 228-232, Felmlee, T, Pellett, S., Lee, E. Y., & Welch, R. A. (1985a). Escherichia coli hemolysin is released extracellularly without cleavage of a signal peptide. J. Bacteriol. 163, 88-93. Felmlee, T, Pellett, S., & Welch, R. A. (1985b). Nucleotide sequence of an Escherichia coli chromosomal hemolysin. J. Bacteriol. 163, 94—105.
178
AMY FUJISHIGE BOGGS and DAVID A. AGARD
Filloux, A., Bally, M., Ball, G., Akrim, M., Tommassen, J., & Lazdunski, A. (1990). Protein secretion in Gram-negative bacteria: Transport across the outer membrane involves common mechanisms in different bacteria. EMBO J. 9, 432S-4329. Fujishige, A., Smith, K. R., Silen, J. L., & Agard, D. A. (1992). Correct folding of a-lytic protease is required for its extracellular secretion from Escherichia coli. J. Cell Biol. 118, 33-^2. Glaser, P., Sakamoto, H., Bellalou, J., Ullmann, A., & Danchin, A. (1988). Secretion of cyclolysin, the calmodulin-sensitive adenylate cyclase-haemolysin bifunctional protein of Bordetella pertussis. EMBO J. 7, 3997-4004. He, S. Y., Schoedel, C, Chatterjee, A. K., & Collmer, A. (1991). Extracellular secretion of pectate lyase by the Erwinia chrysanthemi Out pathway is dependent upon Sec-mediated export across the inner membrane. J. Bacteriol. 173,4310-^317. Hirst, T. R. & Holmgren, J. (1987). Transient entry of enterotoxin subunits into the periplasm occurs during their secretion from Vibrio cholerae. J. Bacteriol. 169, 1037-1045. Hunkapiller, M. W., Smallcombe, S. H., Whitaker, D. R., & Richards, J. H. (1973). Carbon nuclear magnetic resonance studies of the histidine residue in a-lytic protease. Biochemistry 12, 47324743. Koronakis, V., Koronakis, E., & Hughes, C. (1989). Isolation and analysis of the C-terminal signal directing export of Escherichia coli hemolysin protein across both bacterial membranes. EMBO J. 8, 595-605. Krogfelt, K. A. (1991). Bacterial adhesin: Genetics, biogenesis, and role in pathogenesis of fimbrial adhesins of Escherichia coli. Rev. Infec. Dis. 13, 721-735. Mackman, N., Nicaud, J. M., Gray, L., & Holland, I. B. (1985). Identification of polypeptides required for the export of haemolysin 2001 from E. coli. Mol. Gen. Genet. 201, 529-536. Mackman, N., Nicaud, J. M., Gray, L., & Holland, I. B. (1986). Secretion of haemolysin by Escherichia coliCun. Topics Microbiol. Immunol. 125, 159-181. Mackman, N., Baker, K., Gray, L., Haigh, R., Nicaud, J. M., & Holland, I. B. (1987). Release of a chimaeric protein into the mediumfromEscherichia coli using the C-terminal secretion signal of haemolysin. EMBO J. 6, 2835-2841. Masure, H. R., Au., D. C, Gross, M. K., Donovan, M. G., & Storm, D. R. (1990). Secretion of the Bordetella pertussis adenylate cyclase from Escherichia coli containing the hemolysin operon. Biochemistry 29, 140-145. Missiakis, D., Georgopoulos, C, & Raina, S. (1994). The Escherichia coli dsbC (xprA) gene encodes a periplasmic protein involved in disulfide bond formation. EMBO J. 13, 2013-2020. Miyazaki, H., Yanagida, N., Horinouchi, S., & Beppu, T. (1989). Characterization of the precursor of Serratia marcescens serine protease and COOH-terminal processing of the precursor during its excretion through the outer membrane of Escherichia coli. J. Bacteriol. 171, 6566-6572. Nicaud, J.-M., Jackman, N., Gray, L., & Holland, I. B. (1986). The C-terminal, 23 kDa peptide of E. coli haemolysin 2001 contains all the information necessary for its secretion by the haemolysin (Hly) export machinery. FEBS Lett. 204, 331-335. Pohlner, J., Halter, R., Beyreuther, K., & Meyer, T. F. (1987). Gene structure and extracellular secretion of Neisseria gonorrhoeae IgA protease. Nature 325, 458-462. Possot, O., d'Enfert, C, Reyss, I., & Pugsley, A. P. (1992). PuUulanase secretion in Escherichia coli K-12 requires a cytoplasmic protein and a putative polytopic cytoplasmic membrane protein. Mol. Microbiol. 6,95-105. Pugsley, A. P. (1992). Translocation of a folded protein across the outer membrane in Escherichia coli. Proc. Natl. Acad. Sci. USA 89, 12058-12062. Pugsley, A. P. (1993). The complete general secretory pathway in Gram-negative bacteria. Microbiol. Rev. 57, 50-108. Pugsley, A. P., d'Enfert, C, Reyss, I., & Komacker, M. G. (1990). Genetics of extracellular protein secretion by Gram-negative bacteria. Annu. Rev. Genet. 24, 67—90.
Secretion ofa-Lytic Protease by E. coli
179
Randall, L. L. & Hardy, S. J. S. (1986). Correlation of competence for export with lack of tertiary structure of the mature species: A study in vivo of maltose-binding protein in E. coli. Cell 46, 921-928. Sen, K. & Nikaido, H. (1990). In vitro trimerization of OmpF porin secreted by spheroplasts of Escherichia coli. Proc. Natl. Acad. Sci. USA 87, 743-747. Shevhik, V. E., Condemine, G., & Robert-Badouy, J. (1994). Characterization of DsbC, a periplasmic protein of Erwinia chrysanthemi and Escherichia coli with disulfide isomerase activity. EMBO J. 13,2007-2012. Silen, J. L. & Agard, D. A. (1989). The a-lytic protease pro-region does not require a physical linkage to activate the protease domain in vivo. Nature 341, 462—464. Silen, J. L., McGrath, C. N., Smith, K. R., & Agard, D. A. (1988). Molecular analysis of the gene encoding a-lytic protease: Evidence for a preproenzyme. Gene 69, 237—244. Silen, J. L., Frank, D., Fujishige, A., Bone, R., & Agard, D. A. (1989). Analysis of prepro a-lytic protease expression in Escherichia coli reveals that the pro region is required for activity. J. Bacteriol. 171, 1320-1325. Sohl, J. L. & Agard, D. A. (1995). a-lytic protease: dynamic stability via kinetic control. In: Intramolecular Chaperones and Protein Folding (Shinde, U., & Inouye, M., eds.), pp. 61-83. R.J. Landis Co., Austin, TX. Strathdee, C. A. & Lo, R. Y. C. (1989). Cloning, nucleotide sequence, and characterization of genes encoding the secretion function of the Pasteurella haemolytica leukotoxin determinant. J. Bacteriol. 171,916-928. Terada, I., Kwon, S., Miyata, Y, Matsuzawa, H., & Ohta, T. (1990). Unique precursor structure of an extracellular protease, aqualysin I, with NH2- and COOH-terminal pro-sequences and its processing in Escherichia coli. J. Biol. Chem. 265, 6576-^581. Walter, P. & Lingappa, V. R. (1986). Mechanism of protein translocation across the endoplasmic reticulum membrane. Annu. Rev. Cell Biol. 2,499-516. Walter, P., Gilmore, R., & Blobel, G. (1984). Protein translocation across the endoplasmic reticulum. Cell 38, 5-8. Whitaker, D. R. (1970). The a-lytic protease of a myxobacterium. Methods Enzymol. 19, 599-613. Wickner, W. (1979). The assembly of proteins into biological membranes: The membrane trigger hypothesis. Annu. Rev. Biochem. 48, 23-45. Wong, K. R. & Buckley, J. T. (1989). Proton motive force involved in protein transport across the outer membrane oi Aeromonas salmonicida. Science 246, 654—656. Wong, K. R., Green, M. J., & Buckley, J. T. (1989). Extracellular secretion of cloned aerolysin and phospholipase by Aeromonas salmonicida. J. Bacteriol. 171, 2523—2527. Wulfmg, C. & Pluckthun, A. (1994). Protein folding in the periplasm of Escherichia coli. Mol. Microbiol. 12,685-692.
This Page Intentionally Left Blank
MECHANISMS OF PEROXISOME BIOGENESIS: REGULATION OF PEROXISOMAL ENZYMES, AND THEIR SUBSEQUENT SORTING TO PEROXISOMES
Gillian M. Small
Abstract 182 I. Introduction 182 II. Peroxisome Biology 183 A. Peroxisome Proliferation in Mammals 183 B. Peroxisome Proliferation in Yeast 184 III. Peroxisome Membrane Biogenesis 185 A. Peroxisome Membranes in Normal and Peroxisome-Proliferating Cells . 185 B. Peroxisome Membranes during Peroxisome Proliferation 188 C. Peroxisome Formation in Yeast 191 IV. Transcriptional Regulation of Genes Encoding Peroxisomal Proteins 191 A. Transcriptional Regulation in Mammals 191 B. Regulationof Mammalian Genes Encoding Peroxisomal Proteins . . . . 192 C. Transcriptional Regulation in Yeast 194 D. Regulationof Yeast Genes Encoding Peroxisomal Proteins 194
Membrane Protein Transport Volume 3, pages 181-211. Copyright © 1996 by JAI Press Inc. AH rights of reproduction in any form reserved. ISBN: 1-55938-989-3 181
182
GILLIAN M. SMALL
V. Transport and Targeting of Peroxisomal Proteins A. Peroxisomal Targeting Signals B. Other Factors Involved in Peroxisome Assembly VI. Summary Acknowledgments References
197 198 200 201 202 202
ABSTRACT Compartmentalization is a feature of eukaryote cells and requires the sorting of specific proteins to their correct destination within the cell. This is generally accomplished by the inclusion of topogenic sorting signals in the amino acid sequence of the protein to be sorted. Nuclear-encoded proteins are either deposited in the cytosol or are targeted to the endoplasmic reticulum, the nucleus, mitochondria, chloroplasts, or peroxisomes. The focus of this chapter is on the processes involved in regulating the biogenesis of peroxisomes. Peroxisomes are specialized organelles found in most eukaryote cells, where their major functions are related to cellular respiration and fatty acid oxidation. They mature by the posttranslational incorporation of newly synthesized proteins; they then grow and divide to form new peroxisomes. This is a dynamic process, and the kinetics by which this takes place depend on the organism, the cell type, and the metabolic state of that cell. Under certain conditions peroxisomes are caused to rapidly proliferate; this entails increasing the membrane content of the organelle, activating the transcription of genes encoding peroxisomal proteins, and importing these proteins into peroxisomes. Our understanding of the mechanisms involved in this process is not complete; however, much information regarding this process has been gathered in the last few years.
1. INTRODUCTION The past decade has seen a widely increased interest in peroxisome biology, in the modulation of peroxisome biogenesis, and in the regulation of enzymes encoding peroxisomal proteins. The reasons for this spiraling interest are many-fold but surely include the following: the ambition to understand and treat inherited peroxisomal disorders such as Zellweger syndrome, and the desire to determine whether peroxisomes are formed in a manner similar to that of other organelles such as mitochondria, in which the mechanisms of protein trafficking and targeting to the organelle are much better understood, or whether peroxisomes have their own unique mechanisms for protein import. A more practical reason is the emergence of techniques that enable us to manipulate mammalian cells in culture and make use of the elegant techniques of yeast genetics to ask some of these questions at the molecular level. The purpose of this chapter is to present a brief overview of some of the recent developments in this field, especially with respect to mechanisms
Peroxisome Biogenesis
183
involved in peroxisome proliferation. Because of the rapid development of data on this topic I will limit this discussion to the regulation and biogenesis of mammalian and yeast peroxisomes.
II. PEROXISOME BIOLOGY Peroxisomes are single membrane-bound organelles that are found in almost all eukaryote cells. They contain a variety of enzymes, many of which are oxidases that generate hydrogen peroxide. H2O2 is decomposed within the peroxisome to water and oxygen by peroxisomal catalase. Many of the peroxisomal enzymes take part in metabolic pathways that are involved in fatty acid metabolism. In mammals peroxisomes contain a P-oxidation pathway that is analogous to that found in mitochondria, but the enzymes differ significantly from their mitochondrial counterparts. The first enzyme of the peroxisomal fatty acid oxidation cycle is a flavoprotein, acyl-CoA oxidase. This enzyme catalyzes the dehydrogenation of fatty acyl-CoA; this is the rate-limiting step of the pathway. Abifiinctional protein, enoyl-CoA hydratase/3-hydroxyacyl-CoA dehydrogenase, and 3-ketoacyl-CoA thiolase form the rest of the P-oxidation cycle. Peroxisomes also contain enzymes involved in bile acid synthesis (Krisans et al., 1985) and cholesterol metabolism (Thompson et al., 1987), as well as the first two enzymes involved in the synthesis ofplasmalogens(Hajra and Bishop, 1982) (for reviews see Lazarow, 1987, 1988). The number of peroxisomes found in a cell, as well as the size, volume, and specific enzyme composition, varies widely between organisms and between tissues. Peroxisomes with a diameter of 0.5—1.0 jiim are readily identifiable in liver and kidney tissue, whereas fewer and smaller peroxisomes (sometimes referred to as microperoxisomes) are found in other tissues such as heart and small intestine. The functional importance of peroxisomes in mammalian metabolism is indicated by the existence of several human peroxisomal disorders (reviewed in Lazarow and Moser, 1989). The most severe of these is the genetic disorder Zellweger syndrome, which is characterized by the absence of morphologically distinguishable peroxisomes (Goldfischer et al.,1973). A. Peroxisome Proliferation in Mammals A further fascinating aspect of peroxisome biology is the fact that the number of organelles per cell can vary in response to physiological and pharmacological stimuli. Peroxisomes are induced in the liver of rodents by feeding high-fat diets (Neat et al., 1980, 1981), by starvation (Ishii et al., 1980; Thomassen et al., 1982), by vitamin E deficiency (Reddy et al, 1981), and by administration of a wide range of synthetic compounds that are collectively termed peroxisome proliferators (Rao and Reddy, 1987). These include many hypolipidemic drugs, herbicides, and industrial plasticizers. The mechanisms by which these stimuli cause peroxisome
184
GILLIAN M. SMALL
proliferation are not fully understood; some of the current work investigating this process at the molecular level will be discussed later in this chapter. It is of great interest that all of the peroxisome proliferators tested thus far in long-term studies have been found to induce liver tumors in rodents (for a review see Rao and Reddy, 1987). Peroxisome proliferators do not cause DNA damage directly, according to genotoxic assays, and they appear to have little or no effect on the initiation of carcinogenesis (Popp and Cattley, 1993). They do, however, cause a substantial increase in liver size and appear to have an effect on the growth of preneoplastic lesions and on the conversion of such lesions to tumors (Popp and Cattley, 1993). The mechanisms by which this occurs remain to be elucidated. It is clear that there is a qualitative association between the ability of peroxisome proliferators to cause an increase in the peroxisome population and their ability to induce tumor formation, and there has been much discussion in the literature questioning whether there is a causal relationship between induction of the organelle and carcinogenesis (see Reddy and Lalwani, 1983; Lock et al., 1989). Reddy and co-workers postulated an "oxidative stress" hypothesis to explain peroxisome proliferator-induced tumor formation. This is based on the premise that the increased production of hydrogen peroxide from peroxisomal oxidase reactions would exceed the ability of peroxisomal catalase to decompose this toxic compound, resulting in a gradual accumulation of oxidative damage (Reddy et al., 1982; Reddy and Rao, 1989; Rao and Reddy, 1991). In support of this theory, administration of peroxisome proliferators to rodents eventually leads to enhanced lipid peroxidation, lipofuscin deposition, and increased damage to DNA. However, there is no consistent correlation between the magnitude of hepatic peroxisome proliferation (and presumably oxidative stress) and subsequent liver tumor formation, suggesting that if this mechanism is involved, it is not the sole reason for the carcinogenicity (Lake, 1993). Alternatively, it has been suggested that peroxisome proliferators may act as promoters of tumors, either by acting at a later stage of hepatocarcinogenesis (promoting growth of spontaneously initiated cells) or by having promoter activity involved in the generation of tumors but not foci (Cattley and Popp, 1989). B. Peroxisome Proliferation in Yeast
In yeast the number and size of peroxisomes are also extremely variable; here the governing factor is the nutrient supplied for growth. Peroxisomes can be induced in yeasts by fatty acids, methanol, «-alkanes, and some amines (see Veenhuis and Harder, 1987). The amount of proliferation depends on the yeast species and on the carbon source. For example, the methylotrophic yeasts Hansenula polymorpha, Candida boidini, and Pichia pastoris have one or a few small peroxisomes when grown on glucose (Van Dijken et al., 1975; Veenhuis and Harder, 1987). However, following growth on methanol there is a huge proliferation of peroxisomes such that they become the most abundant organelle in the cell (Douma
Peroxisome Biogenesis
185
et al., 1985; Goodman, 1985; Veenhuis and Harder, 1987). In the yeasts Saccharomyces cerevisiae, Candida tropicalis, and P. pastoris peroxisomes are induced when a fatty acid such as oleate is supplied for growth, whereas in the presence of glucose peroxisomes are scarce (Dommes et al., 1983; Veenhuis et al., 1987; Veenhuis and Harder, 1987). In both mammalian and yeast cells peroxisome proliferation is accompanied by an induction of peroxisomal enzymes, especially those involved in lipid metabolism, or, in the case of methylotrophic yeasts, those involved in the metabolism of methanol. Peroxisome proliferation and induction of peroxisomal enzymes have also been demonstrated in cultured hepatocytes of several species (Foxworthy et al., 1990; Tomaszewski et al., 1990) and in a number of nonhepatic tissues (Small et al., 1982). Induction of peroxisomal enzymes has been shown to be transcriptionally regulated (Reddy et al., 1986; Einerhand et al., 1991; Osumi et al., 1991b; Wang et al., 1992; Einerhand et al., 1993; Godecke et al., 1994).
IIL PEROXISOME MEMBRANE BIOGENESIS Peroxisome biogenesis is considered to occur by the import of both matrix and membrane proteins into pre-existing peroxisomes; new peroxisomes would then form by fission of pre-existing peroxisomes (Lazarow and Fujiki, 1985; Subramani, 1993). If this is the case then there are several implications; first, every cell must contain at least one peroxisome, a premise that is generally accepted (but see Waterham et al., 1993). Then several dynamic mechanisms must operate. There must be an increase in peroxisomal membrane mass to accommodate this process, and the transcriptional regulation of genes encoding peroxisomal proteins must be finely regulated. Last, the newly synthesized peroxisomal proteins must be transported and targeted to the peroxisomes. Each of these processes is discussed below. A. Peroxisome Membranes in Normal and Peroxisome-Proliferating Cells Mammalian Peroxisomal Membrane Proteins
The peroxisome is bounded by a single unit membrane that maintains the integrity of the organelle. The membrane has a unique polypeptide composition (Fujiki et al., 1982a). Some of the peroxisome-associated enzyme activities, including those involved in the biosynthesis pathway of ether lipids, have been assigned to the peroxisomal membrane. However, these are all peripheral membrane proteins (for a review see Causeret et al., 1993). Peroxisomal integral membrane proteins, as defined by their property of not being extractable by alkaline sodium carbonate according to the procedure of Fujiki et al. (1982b), range in size from 15 kDa to 520 kDa in mammalian liver peroxisomes (Causeret et al.,1993). Thus far the specific functions of most of these proteins are unknown. One major protein in peroxisomal membranes from rat, mouse, and human liver is a 68/70 kDa protein that has high homology with the
186
GILLIAN M. SMALL
family of membrane transport ATPases (Fujiki et al., 1982b; Just and Hartl, 1987; Santos et al., 1988b, 1994; Small et al., 1988b; Kamijo et al., 1990; Chen and Crane, 1992). Surprisingly, the 70-kDa peroxisomal membrane protein is present in fibroblasts from patients with Zellweger syndrome (Santos et al., 1988b; Wiemer et al., 1989) and is contained within membrane structures devoid of much matrix content. These structures have been termed "peroxisome ghosts" (Santos et al., 1988a, 1992; Wiemer et al., 1989). Recently a gene that appears to be responsible for X-linked adrenoleukodystrophy (ALD) was cloned and sequenced (Mosser et al., 1993). The predicted amino acid sequence has a high degree of homology with the human and rat 70-kDa peroxisomal membrane protein; thus the ALD protein seems to be a peroxisomal membrane protein. Peroxisome-defective mutants have been isolated in Chinese hamster ovary (CHO) cells (Zoeller et al., 1989; Morand et al., 1990; Tsukamoto et al., 1990; Shimozawa et al., 1992b). These mutants have reduced levels of dihydroxyacetone phosphate (DHAP) acyl transferase, alkyl DHAP synthase, acyl-CoA oxidase, particulate catalase, and plasmalogens. One of the mutants isolated was shown to be deficient in the PAF-1 gene encoding a 35-kDa peroxisomal integral membrane protein (Tsukamoto et al., 1991). A mutation in the human homolog of PAF-1 was shown to be the cause of the peroxisomal defect in fibroblasts from a patient with Zellweger syndrome (Shimozawa et al., 1992a). Furthermore, it has been shown that CHO cells that are defective in plasmalogen biosynthesis and peroxisome assembly have these functions restored when the cDNA encoding rat liver PAF-1 is expressed in these cells (Zoeller and Raetz, 1986; Thieringer and Raetz, 1993). Other reported integral membrane proteins from rodent peroxisomes have apparent molecular masses of 57,41/42,36,28,26,22, and 15 kDa (Koster et al., 1986; Imanaka et al., 1991). The 22-kDa protein has been cloned and sequenced (Kaldi et al., 1993). Based on its hydropathy analysis and accessibility to added proteases, Kaldi et al. (1993) proposed that this protein has four transmembrane segments and that both the amino and carboxyl termini are directed toward the cytosol. Recently two other peroxisomal membrane proteins from rat liver of 450 and 520 kDa have been shown to display A^-ethylmaleimide-sensitive and -resistant ATPase activities (Shimizu et al., 1992). The profile of human liver peroxisome integral membrane proteins is similar to that of rodents; proteins with apparent molecular masses of 147, 112, 79, 69/70, 52/53, 47, 43, 31, 28, 22, and 17 kDa have been reported (Santos et al., 1988a, 1994; Causeret et al., 1993). The nature of the exact function of any of these proteins remains to be elucidated. Peroxisome proliferators have been shown to cause markedly increased levels of mRNA encoding the 70-kDa protein (Hashimoto et al., 1986; Suzuki et al., 1987; Kamijo et al., 1990), with more moderate increases detected in some of the other peroxisomal membrane proteins (Hashimoto et al., 1986; Suzuki et al., 1987).
Peroxisome Biogenesis
187
Yeast Peroxisome Membrane Proteins
Integral membrane proteins have been examined in peroxisomes from the yeasts S. cerevisiae (McCammon et al., 1990b), H. polymorpha (Suiter et al., 1990), C. boidini (Goodman et al., 1986; Suiter et al., 1990), and C tropicalis (Nuttley et al., 1990). Generally, membrane proteins do not appear to be very abundant in yeast peroxisomes. The most prominent protein bands from carbonate-extracted peroxisomal membranes from S. cerevisiae have apparent molecular masses of 32, 31, and 24 kDa (McCammon et al., 1990b). This is similar to the pattern of dominant proteins isolated from C. tropicalis (34, 29, 24 kDa) (Nuttley et al., 1990), and C boidini (47, 32, 31 kDa) (Goodman et al., 1986). Peroxisomal membranes of methanol-grown H. polymorpha have been examined, and those that were not removed by carbonate (and therefore are integral membrane proteins) had apparent molecular masses of 68, 57, 51,42, 35, and 31 kDa (Suiter et al., 1990). A few yeast peroxisomal membrane proteins have been cloned and sequenced by conventional methods such as screening cDNA libraries with antibodies or oligonucleotides. By using an antibody to a membrane protein, PMP20, from C. boidini the cDNA encoding this protein was obtained. However, this protein is also found in the peroxisome matrix and is not as resistant to carbonate extraction as those membrane proteins from this yeast listed above and therefore is probably a peripheral membrane protein (Garrard and Goodman, 1989). The gene encoding the 47-kDa membrane protein from C. boidini was isolated from a genomic library by using oligonucleotide probes designed from peptide sequences of the isolated protein; this appears to be a bona fide integral membrane protein (McCammon et al., 1990a). Recently, two groups have reported cloning genes encoding peroxisomal membrane proteins from S. cerevisiae that have apparent molecular masses of 27 and 24 kDa, respectively (Erdmann and Blobel, 1994; Marshall et al., 1994). Sequence data suggest that this is the same gene (R. Erdmann, personal communication). Gene disruption experiments indicate that the encoded membrane protein is essential for maintaining normal peroxisomal morphology. Other, less abundant yeast peroxisomal membrane proteins have been cloned by fimctional complementation of peroxisome biogenesis mutants. Peroxisome-defective or -deficient mutants have been isolated from S. cerevisiae (Erdmann et al., 1989; Van Der Leij et al., 1992; Elgersma et al., 1993; Zhang et al., 1993b), H. polymorpha (Cregg et al., 1990; Veenhuis, 1992; Waterham et al., 1992), P. pastoris (Gould et al., 1992; Liu et al., 1992), and Y. lipolytica (Nuttley et al., 1993). The PASS gene from S. cerevisiae pas3 mutants has been cloned (Hohfelf et al., 1991). This gene encodes a 48-kDa protein (pas3p), which behaves like an integral membrane peroxisomal protein. By cloning the PASS gene that complements the mutation that is causing the peroxisome abnormality in P pastoris pas8 cells, a peroxisome membrane protein (Pas8p) was cloned (McCollum et al., 1993). This gene encodes a 68-kDa protein that localizes to peroxisome membranes and is induced by methanol and oleate.
188
GILLIAN M. SMALL
Furthermore, the protein binds to the peroxisomal SKL targeting signal (PTS1) (see section on peroxisome protein transport and targeting). A homolog of PASS has been cloned from a peroxisome-deficient S. cerevisiae mutant (pas 10); this is complemented by the PASIO gene from this strain (Van Der Leu et al., 1993). Functional roles in peroxisomal targeting and import have been postulated for the latter two proteins (see below). B. Peroxisome Membranes during Peroxisome Proliferation
Membranous structures have frequently been seen in close contact with peroxisome, and connections between the endoplasmic reticulum and peroxisomes were described based on morphological data (NovikofF and Shin, 1964). However, subsequent studies on peroxisomes in regenerating mouse and rat liver (Gorgas, 1985; Yamamoto and Fahimi, 1987) have demonstrated that these membranes are actually interconnections between peroxisomes. These observations supported the biochemical data that peroxisomal membrane proteins are synthesized on free polyribosomes and are imported into pre-existing peroxisomes (Fujiki et al., 1984; Suzuki et al., 1987). Thus it was postulated that peroxisomes themselves form a reticulum quite distinct from the endoplasmic reticulum (Lazarow et al., 1980). This was confirmed by Yamamoto and Fahimi in their studies on regenerating rat liver following partial hepatectomy, which demonstrated interconnections between several peroxisomes in the early stages of peroxisome proliferation (Yamamoto and Fahimi, 1987). Further evidence of a peroxisome reticulum came from studies on rat liver following treatment with the peroxisome-proliferating drug BM 15766 (Baumgart et al., 1987). These studies identified double-membrane loops in close or direct continuity with the peroxisome membrane. Using antibodies to peroxisomal integral membrane proteins Baumgart et al. (1989) demonstrated that these membrane loops label with an antibody to the 70-kDa peroxisomal membrane protein (see Fig. 1). The loops could not be stained for glucose-6-phosphatase, an endoplasmic reticulum marker, and were also negative for catalase; thus it was suggested that they comprise a membrane reservoir for the proliferation of peroxisomes and for the expansion of this compartment (Luers et al., 1993). In support of this theory are earlier reports of a catalase-negative subpopulation of peroxisomes that are induced in the livers of mice administered with the peroxisome proliferator clofibrate (Klucis et al., 1991). Based on these and other observations noting the heterogeneous nature of the distribution of different peroxisomal enzymes (Le Hir and Dubach, 1980; Fahimi et al., 1982), the incorporation of newly synthesized proteins into peroxisomes has recently been reinvestigated. Using short pulse label times to label proteins in rats that had received a partial hepatectomy, in conjunction with a revised subcellular fractionation protocol, Luers et al. (1993) claim to have isolated two distinct populations of rat liver peroxisomes. Their studies suggest that the "heavy" population, which bands at a density of 1.24 g/cm^ (the expected density for rat liver peroxisomes), incorporates proteins at a slower
Figure 1. Sections of livers from rats treated with BM 15766, embedded in Lowicryl K4M (a) and in LR white (b), and incubated with the 70-kDa PMP antibody. The membranes appear as negative images. (a) A peroxisome (PO) with an invagination of its limiting membrane into the matrix showing heavy labeling for the 70-kDa PMP. (b) Adouble-membraned loop labeled heavily with gold particles representingthe antigenic sites for the 70-kDa PMP. Note the absence of label in mitochondria (MITO) and ER membranes. Reproduced from The)ournal of Cell Biology 1989, 108, 2228, by copyright permission of The Rockefeller University Press.
190
GILLIAN M. SMALL
rate than the "Hght" peroxisome population. Thus, new proteins are preferentially incorporated into the "light" peroxisome fractions, which also differ from the heavy population in size and enzyme composition (see Fig. 2). Similar findings have been reported in rats administered clofibrate prior to sacrifice (Heinemann and Just, 1992). In the latter study two subpopulations of peroxisomes were isolated from a Nycodenz density gradient: the "normal" peroxisome fraction, which equilibrates at a density of 1.22—1.23 g/cm^, and an intermediate fraction at 1.16-1.17 g/cm^. Both fractions were defined as being peroxisomal by the presence of the 22-kDa peroxisomal integral membrane protein. Newly synthesized acyl-CoA oxidase was found predominantly in the intermediate compartment and was gradually chased to the high-density peroxisomes. The differences in enzyme content in the two populations of peroxisomes, combined with the different rates of incorporation of newly isynthesized proteins, argue against the possibility that the light peroxisomal fraction in each of these studies arises from fragmentation of normal peroxisomes. In light of these findings Fahimi and co-workers have postulated a model of peroxisome biogenesis in which the initial stage is formation of membranous
Figure 2. (a-d) High-power views of isolated "light" peroxisomes from regenerating rat liver showing tubular membrane extensions and buds (arrowheads), (e) A peroxisome (P) with a double-membrane loop (LOOP) is depicted. MC, Microsome. Bars, 0.2 ^ M . Reproduced from The Journal of Cell Biology 1993, 121,1274, by copyright permission of The Rockefeller University Press.
Peroxisome Biogenesis
191
attachments on the surface of pre-existing peroxisomes (analogous to the loops that contain the 70-kDa peroxisomal membrane protein). In this model these attachments would become import competent following the incorporation of membrane proteins and would expand to form small peroxisomes (Fahimi et al., 1993). C. Peroxisome Formation in Yeast
Peroxisome assembly has also been studied morphologically and biochemically in the methylotrophic yeast C boidini (Veenhuis and Goodman, 1990). The formation of new peroxisomes was studied in cells that were precultured in glucose and were then diluted into methanol media. Peroxisome formation appeared to occur in three distinct stages; in the first hour the small peroxisomes that were present when the cells were grown in glucose became elongated, and there was a concomitant increase in catalase activity and in the amount of the integral membrane protein PMP47. During the next 2-3 hours the peroxisomes divided to form peroxisome clusters; this was accompanied by induced levels of peroxisomal alcohol oxidase, dihydroxyacetone synthase, and the peripheral membrane protein PMP20. Finally, as the peroxisomes enlarged, there was a decrease in their number per cell and an increase in volume per peroxisome. Thus, the events of peroxisome proliferation in yeast appear to resemble those that occur in mammalian cells undergoing peroxisome proliferation.
IV. TRANSCRIPTIONAL REGULATION OF GENES ENCODING PEROXISOMAL PROTEINS Accompanying peroxisome proliferation is an increase in the levels of several peroxisomal proteins, the most marked increases being in the peroxisomal P-oxidation enzymes. It would appear likely that there is a coordinated regulation of the corresponding genes. To understand this highly controlled event the molecular mechanisms of peroxisome enzyme induction has been investigated in both mammals and yeast. A. Transcriptional Regulation in Mammals
In higher eukaryotes the regulatory sequences, or cis elements, of a gene may be distanced several thousand nucleotides upstream from the transcriptional start site. Activating sequences (enhancers) in mammalian genes can be fairly large elements, often several hundred base pairs. They generally function in either orientation and are position independent, in that they may be upstream of the transcriptional start site, located within a gene, or may be downstream of the stop codon. The regulation of mammalian genes is often very complex and may involve 10—20 regulatory proteins (Miner and Yamamoto, 1991). Some of these proteins are transcriptional activators, whereas others are repressors.
192
GILLIAN M. SMALL B. Regulation of Mammalian Genes Encoding Peroxisomal Proteins
Acyl'CoA Oxidase
The induction of the peroxisomal P-oxidation enzymes in rat liver has been demonstrated to occur at the transcriptional level (Reddy et al., 1986). In searching for CIS elements involved in the regulation of acyl-CoA oxidase, Osumi et al. (1991b) demonstrated that two elements in the 5' region of the gene encoding this protein were involved in its regulation. Site A was found to be an activating element, whereas site B had a negative regulatory role. During the time that this work was being performed the cloning of a transcription factor termed the peroxisome proliferator-activated receptor (PPAR) was reported (Issemann and Green, 1990) (see below). If turned out that site A defined by Osumi's group contained the response element that binds this receptor; thus acyl-CoA oxidase is activated via this transcription factor. Peroxisome Proliferator-Activated Receptor
Mouse PPAR (mPPAR), cloned by Isseman and Green (1990), has been placed in the family of nuclear hormone receptors that includes receptors that bind steroid hormones, thyroid hormones, vitamin D, and retinoic acid (for a review see Carson-Jurica et al., 1990; Green and Wahli, 1994). mPPAR contains regions homologous to the functional portions of these receptors. PPAR homologs have subsequently been cloned from rat liver (Gottlicher et al., 1992), from Xenopus laevis (Dreyer et al., 1992), and from human liver (Schmidt et al., 1992; Andre and Small, 1993; Sher et al., 1993; Mukherjee et al., 1994). Three putative receptors were isolated from X. laevis, each having highly conserved DNA-binding and ligand-binding domains, and one (XPPARa) is very closely related (77% homology at the amino acid level) to the mouse receptor (Dreyer et al., 1992). These findings suggest that there are families of PPAR within species. Tissues demonstrating high levels of expression of mouse PPAR (liver and kidney) are those that also have the greatest abundance of peroxisomes (Issemann and Green, 1990). Weak expression was detected in heart, skeletal muscle, small intestine, testis, and thymus. This pattern is consistent with the tissue-specific induction of acyl-CoA oxidase by peroxisome proliferators (Nemali et al., 1988). In adult Xenopus, transcripts of xPPAR were found in all organs tested (liver, kidney, fat body, muscle, brain, and spleen). However, in contrast to rodent tissues, the highest levels occurred in ovaries and kidney, with lower levels in liver (Dreyer et al., 1992). Studies in my own laboratory have also demonstrated that expression of human PPAR is extremely low in liver and is much higher in kidney and skeletal muscle (Andre and Small, 1993, and unpublished observations). However, Mukherjee et al. (1994) recently reported high levels of hPPAR in heart, liver, and skeletal muscle. Receptors belonging to the superfamily of nuclear hormone receptors are intracellular proteins that contain a ligand-binding domain and a DNA-binding domain.
Peroxisome Biogenesis
193
The latter recognizes short DNA motifs termed hormone response elements (HREs), which are usually located upstream of target genes and behave as transcriptional enhancers. Some members of the steroid/thyroid hormone receptor superfamily form complexes with other proteins, either to activate or to enhance function. To determine whether PPAR is involved in the action of peroxisome proliferators, chimeric receptors have been constructed that contain the ligandbinding domain of mPPAR and the DNA-binding domain of either the estrogen or glucocorticoid receptors. In the presence of the peroxisome proliferators nafenopin, Wy-14,643, methylclofenapate, or clofibric acid, these chimeric receptors were able to activate estrogen or glucocorticoid-responsive genes, respectively (Issemann and Green, 1990). Thus, mPPAR mediates the biological effects of peroxisome proliferators. It has also been demonstrated that mPPAR can mediate the action of peroxisome proliferators in mouse hepatoma cells, causing a 20-40-fold induction of a reporter gene consisting of the promoter region of rat acyl-CoA oxidase upstream of the gene encoding chloramphenicol acetyl transferase (CAT) (Tugwood et al., 1992). When these experiments were performed in primary rat hepatocytes, in human HepG2 cells, or in monkey kidney cell lines the induction was 4—10-fold, indicating that the mouse receptor functions but is less efficient in these heterologous systems. mPPAR was shown to recognize a response element in the 5' flanking sequence of the rat acyl-CoA oxidase gene (Tugwood et al., 1992). This element contains a direct repeat of the motifs TGACCT and TGTCCT, the former of which is also recognized by a number of other nuclear hormone receptors, including that for retinoic acid (Mangelsdorf et al., 1991). Both thyroid and retinoic acid receptors have been shown to form complexes with the retinoic acid X receptor (RXR) (Zhang et al., 1992). Formation of such heterodimers enhances the transcriptional activity of each. It was recently shown that RXR also forms a heterodimer with rat PPAR and enhances the induction of acyl-CoA oxidase by the hypolipidemic drug clofibrate (Kliewer et al., 1992). Peroxisome proliferator-response elements have now been localized in the 5' flanking regions of a number of genes encoding peroxisome proteins (Zhang et al., 1992). The natural ligand for PPAR is not yet known; peroxisome proliferators include many structurally unrelated compounds. However, many of these compounds contain a carboxylate functional group, and it has been shown that these compounds are activated to coenzyme A thioesters (Bronfman et al., 1986,1992), which further activate protein kinase (Bronfman et al., 1989). It is possible that these are the initial steps of a cascade of reactions that eventually form the ligand. In the case of those compounds that are activated to the thioester, this activation appears to be necessary for peroxisome proliferation (Tomaszewski and Melnick, 1994). However, the large number of different compounds that are able to activate PPAR make it seem unlikely that they are metabolized to a single common ligand. Alternative suggestions are that the fatty acids themselves bind to PPAR, or that they induce the formation or
194
GILLIAN M. SMALL
release of an unknown ligand involved in this mechanism (Green and Wahli, 1994; Hertz etal., 1994). C. Transcriptional Regulation in Yeast
The mechanism and control of transcription of yeast genes bears many similarities to that of mammalian genes. Most yeast promoters are composed of a TATA box as well as one or more upstream sequence elements. Yeast upstream elements are known as upstream activating or repressing sequences (UAS or URS, respectively). They function in either orientation at various distances upstream from the transcription start site but are usually within a few hundred base pairs (Guarente, 1992). Unlike mammalian enhancers, yeast regulatory sites do not function downstream from the TATA box (Guarente and Hoar, 1984). Yeasts utilize a variety of sugars for carbon and energy. They are also able to use several nonfermentable carbon sources, such as ethanol, acetate, and lactate, which are converted to acetyl-CoA or pyruvate and thus bypass glycolysis. The expression of genes encoding enzymes for the utilization of carbon sources that are alternatives to glucose is tightly regulated. D. Regulation of Yeast Genes Encoding Peroxisomal Proteins Genes from Saccharomyces cerevisiae
Many peroxisomal enzymes of the yeast S. cerevisiae are repressed when the yeast is grown on glucose, derepressed during growth on nonfermentable carbon sources, and induced when a fatty acid such as oleate is supplied as the carbon source. To understand peroxisome biogenesis as well as the regulated repression/induction of peroxisomal proteins, both genetic and molecular biology techniques have been used. The genes encoding acyl-CoA oxidase (POXI), the trifiinctional enzyme (F0X2), 3-oxoacyl-CoA thiolase (POTI/FOX3), and peroxisomal catalase (CTAl) have been cloned and sequenced (Cohen et al., 1988; Dmochowska et al., 1990; Igual et al., 1991; Hiltunen et al., 1992). Regulation of these genes has been shown to occur at the transcriptional level (Skoneczny et al,, 1988; Einerhand et al., 1991; Simon et al., 1991, Hiltunen et al., 1992; Igual et al., 1992; Simon et al., 1992; Wang et al, 1992; Einerhand et al., 1993). CTAl
Expression of the CTAl gene is sensitive to glucose repression (Cohen et al., 1988). It has recently been shown that the ADRI gene acts as a positive regulator controlling the expression of CTAl (Simon et al., 1991). ADRI is also known to function in the activation of the ADH2 gene, which encodes alcohol dehydrogenase 2 (Shore andNasmyth, 1987). Gel retardation experiments demonstrated thai ADRI binds to an upstream region of CTAl, and it was found that adrl null mutants had reduced CTAl expression (Simon et al., 1991). Two genes encoding enzymes of
Peroxisome Biogenesis
195
peroxisomal P-oxidation (POTJ/FOX3 and FOX2) and a gene involved in peroxisome assembly (PASI) were also shown to be negatively affected by the adrl null mutation. An ^Z)/? 7-binding cis element was identified for ^Z)/? 7, deletion of which abolished ADRI activation but did not affect oleate induction (Simon et al., 1991). It was subsequently shown that two other genes, 5A^7^7 and SNF4, which are known to regulate derepression of glucose-repressible genes, are also involved in the derepression of C7:47, FOX2, FOX3/POT1, and PASI (Simon et al., 1992). This effect was most pronounced on the 7^610 transcript and was substantially less on the transcripts of the FOX2 and CTAl genes. In the same study peroxisome morphology was investigated in snfl and snf4 mutants by electron microscopy and immunofluorescence. Typical peroxisome structures were rare in snf mutants, suggesting that these genes are required for peroxisome proliferation. POTUFOX3
Further information regarding glucose repression and oleate induction of peroxisomal enzymes was gained by dissecting the promoter of the POT1/FOX3 gene encoding peroxisomal thiolase. A 5' flanking region of POT] was found to be important for its regulation; deletion of this region abolished both glucose repression and oleate induction (Einerhand et al., 1991). A sequence motif within this region, designated the "p-oxidation box," consists of a palindrome of two repeats (see below). This was subsequently demonstrated to be the binding site for a protein(s) responsible for oleate induction (Einerhand et al., 1993). Thus far, no proteins involved in this regulation have been isolated; however, purification of these trans-2iciing factors is under way in my own and other laboratories. One of the peroxisome-deficient mutants of 5. cerevisiae mentioned earlier was shown to negatively affect the binding of a protein to the oleate-responsive element of this promoter in a band-shift assay (Van Der Leij et al., 1992). The gene complementing this pas 14 mutant was found to be SNF], thus confirming that the SNF genes are involved in regulating peroxisomal proteins in S. cerevisiae. POX1
POX] encodes fatty acyl-CoA oxidase, which is the rate-limiting enzyme of the peroxisomal p-oxidation cycle. We have mapped the promoter region of POX] and have used deletion analysis, DNA band-shift assays, and DNA footprinting to identify regulatory elements in this promoter (Wang et al., 1992). We have shown that there are at least three regulatory elements, two repressing sequences and one activating sequence, suggesting that the regulation of this gene is complex (Wang et al., 1994). We were not able to identify specific ADR] binding sites. However, UASs that bind the same proteins may have strikingly dissimilar sequences (Guarente, 1992). Therefore it is possible that the glucose repression mechanism of POX] is similar to that of other genes that are repressed by glucose. The UAS in the POX] promoter is similar to the sequence described in POT], except that instead of a single palindrome the POX] sequence contains two
GILLIAN M. SMALL
196
Fatty Acid Glucose
Fatty acid activation Glucose repression
Interacting PiomDlly^" complex factors
URS1
URS2 UAS
POX1
TATAA mRNA
Figure 3. Cartoon depicting some of the features regulating peroxisome biogenesis, using POX1 from S. cerevisiae as a representative gene. POXl undergoes glucose repression and oleate induction. The encoded protein, acyl-CoA oxidase, is targeted to peroxisomes by an as yet unidentified peroxisomal targeting signal (PTS). Peroxisomes grow and divide to form new peroxisomes.
Peroxisome Biogenesis
197
palindromes consisting of four repeats. This may reflect the important regulatory role of acyl-CoA oxidase. The consensus sequences for the four repeats in the POXI promoter and the two repeats in the POT1/FOX3 promoter are shown below: POXI
T/A A T / A N N C C G T/AT/A
P0T1/F0X3
T A T/A C C/A C C G T/C C/A
It is possible that these two genes are regulated in a similar manner, and it will be interesting to compare the proteins that are binding to the activating elements as they are identified and characterized. In my own laboratory we are employing standard protein purification techniques to purify the rra«5-acting factor that is binding to the oleate-specific UAS, as well as yeast mutagenesis to identify and clone other factors involved in this pathway. A model for the regulation of POXI, based on the elements identified thus far, is depicted in Figure 3. Genes from Hansenula polymorpha H. polymorpha is a methylotrophic yeast in which peroxisomes are induced in methanol growth conditions. The regulation of the MOX gene, which encodes peroxisomal methanol oxidase, has been studied by dissecting the promoter region of this gene (Godecke et al., 1994). By using deletion analysis and DNA band-shift and footprinting methods, a methanol-specific UAS was located between positions —507 and-481. The sequence identified bears marginal resemblance to the UASs found in POXI and POT1/FOX3, but in the case ofMOXl there is no evidence of a palindromic sequence. However, further analysis of this promoter region revealed a U AS2 that does contain internal symmetry, suggesting that the trans-dLoXmg factor may bind as a homodimer (Godecke et al., 1994). Upstream repression sequences were also identified in the MOX] promoter.
V. TRANSPORT AND TARGETING OF PEROXISOMAL PROTEINS Whereas the targeting and transport of proteins to the endoplasmic reticulum, mitochondria, chloroplasts, and the nucleus have been topics of extensive research over many years, the transport of proteins to peroxisomes is a relatively new field of investigation. Indeed, the first peroxisomal targeting sequence was reported just eight years ago, and many of the features of this process remain a unknown. However, with the increased use of modem molecular biology techniques and with the studies on human peroxisomal disorders ^id on both mammalian and yeast peroxisome biogenesis mutants, much information has been gained in the last few years. This research has been the subject of many recent reviews (De Hoop and Geert, 1992; Roggenkamp, 1992; Subramani, 1993; Purdue and Lazarow, 1994). I will summarize some of the key features that have been uncovered thus far and highlight areas where much information remains to be elucidated.
198
GILLIAN M. SMALL A. Peroxisomal Targeting Signals
Having accepted the premise that peroxisomal proteins are synthesized on free polyribosomes and are transported to the peroxisome post-translationally, it became apparent that there must be some topogenic signal involved to target the protein to the correct organelle. The definition used for such a signal is that it must be completely necessary and sufficient to target the protein to peroxisomes. The first such peroxisomal targeting signal to be identified was the tripeptide serine-lysineleucine (SKL) at the carboxy terminus of firefly luciferase, a protein that is targeted to peroxisomes (Gould et al., 1987; Keller et al., 1987). This signal, designated PTSl for peroxisomal targeting signal 1, is found at the carboxy termini of many peroxisomal proteins from mammals, plants, and yeast (Gould et al., 1990; Keller et al., 1991). Certain conservative changes in the SKL motif are permissible for the tripeptide to remain functional as a targeting signal. For example, alanine can be substituted for serine in the first position, and arginine (or less efficiently histidine) is able to replace lysine in position two. Methionine will function, although less efficiently than leucine, in position three (Gould et al., 1989; Swinkels et al., 1992). In yeast species SKL, or conservative variations of this motif, also function as peroxisomal targeting signals. Peroxisomal citrate synthase of S. cerevisiae has SKL at its carboxy terminus, whereas the multifunctional p-oxidation enzyme of C. tropicalis ends in AKI (Rosenkrantz et al., 1986; Nuttley et al., 1988; Aitchison et al., 1991b); both are essential for peroxisomal targeting (Aitchison et al., 1991a; Singh et al., 1992). The two abundant proteins found in methanol-grown H. polymorpha end in ARF and NKL, respectively; again both will act as peroxisomal targeting signals (Hansen et al., 1992; Roggenkamp, 1992). Although it is clear that PTSI acts as a targeting signal for many peroxisomal proteins it clearly is not the only signal for directing proteins to this organelle. Many peroxisomal proteins, including all peroxisomal integral membrane proteins sequenced thus far, do not contain this signal. A second type of signal has been defined as an 11-amino acid sequence within the first 36 amino acids at the amino termini of rat and human thiolases (Osumi et al., 1991a; Swinkels et al., 1991). This signal has been designated PTS2, and unlike PTS 1, it is cleaved from the mature protein. The sequences at the amino termini of yeast thiolases resemble PTS2, but they appear not to be cleaved upon import (Erdmann, 1994; Glover et al., 1994b). Other peroxisomal targeting signals have yet to be defined. However, we do have evidence that they may not be confined to either the carboxy or amino termini of peroxisomal proteins. Acyl-Co A oxidase from C tropicalis contains two redundant targeting regions, both of which are internally located in the amino acid sequence (Small et al., 1988a; Kamiryo et al., 1989). Peroxisomal catalase from S. cerevisiae also appears to have redundant targeting information; one signal resides at the carboxy terminus and resembles PTS 1 (SKF), and the other is located within the amino-terminal third of the protein (Kragler et al., 1993). There is also evidence for an internal peroxisomal targeting sequence in human L-alanine-glyoxylate
Peroxisome Biogenesis
199
aminotransferase 1. The localization of this protein to either peroxisomes or mitochondria is species dependent (Noguchi and Takada, 1979; Okuno et al., 1979; Yokota et al., 1987; Cooper et al., 1988). In humans the enzyme is normally localized in peroxisomes; however, a substitution at amino acid 170 interferes with this peroxisomal targeting (Purdue et al., 1990). Targeting signals for peroxisome membrane proteins have yet to be defined. None of the proteins that have been cloned thus far (discussed earlier) have sequences that resemble PTSl at their carboxy termini or PTS2 at their amino termini; thus it appears that a different type of signal directs these proteins to the peroxisomal membrane. PMP47, an integral membrane protein from C. boidini, sorts to the peroxisomal membrane of 5. cerevisiae when the protein is expressed in this yeast (McCammon et al., 1990a). To determine the peroxisomal sorting signal in this protein fusions were prepared between portions of PMP47 and dihydrofolate reductase (DHFR). When expressed in S. cerevisiae, hybrids containing the first 267 amino acids of PMP47 fused to DHFR sorted to peroxisomes, whereas hybrids containing 199 or 88 amino acids of PMP47 and DHFR were cytosolic. This suggests that there is peroxisomal targeting information between amino acids 200 and 267 of PMP47; however, these amino acids alone did not appear to be sufficient for peroxisomal localization (McCammon et al., 1994). It may seem surprising that there appear to be at least three different types of peroxisomal targeting signals, and therefore, one supposes, three sets of import machinery, or at least parts thereof. Yet evidence accumulating from studies of peroxisome-deficient mutants in yeast and mammals (discussed earlier) seems to confirm this idea. Studies on Zellweger cells demonstrate that generally they cannot package any of the peroxisomal matrix proteins, yet can clearly assemble some peroxisomal membrane proteins, thus forming "membrane ghosts" (Santos et al., 1988a; Wiemer et al., 1989). There are S. cerevisiae mutants that fail to package proteins with PTSl-type signals, and some that fail to package proteins with PTS2-type signals. There are also mutants in these studies that fail to package any of the matrix proteins tested (Erdmann et al., 1989; Van Der Leij et al., 1992; Zhang et al., 1993a,b). As more of the genes that complement these mutants are cloned we should gain a better understanding of the translocation processes that are involved. Indeed, the recent cloning and characterization of the PASS gene from P. pastoris (McCollum et al., 1993) and the homologous PASIO from S. cerevisiae (Van Der Leu et al., 1993) indicate that the encoded proteins may be the PTSl receptor (McCollum et al., 1993). The PASS gene from S. cerevisiae has been cloned. Based on its predicted amino acid sequence and on protease digestion experiments Pas3p is predicted to be an integral membrane protein that is anchored in the membrane at its amino terminus, with the majority of the protein located on the cytosolic side of the membrane (Hohfelf et al., 1991). With such a topology it is possible that this protein is a component of the translocation apparatus; however, there is as yet no evidence for this function.
200
GILLIAN M. SMALL
Figure 3 summarizes schematically some of the features discussed in this chapter regarding the regulation of peroxisome biogenesis. The transcriptional regulation of POX 1 from S. cerevisiae is used as an example of the types of control mechanisms that regulate the expression of peroxisomal proteins in yeast. Transport and targeting of peroxisomal proteins to peroxisomes appear to be well conserved between yeast and mammals. B. Other Factors Involved in Peroxisome Assembly
One of the reasons for increased research in peroxisome biogenesis, alluded to at the beginning of this chapter, was the need to understand molecular defects in peroxisomal disorders such as Zellweger syndrome. The recent knowledge we have gained has taken us some way toward that goal; however, much remains to be understood. The yeast mutants described above resemble, in part, the deficiencies seen in some peroxisomal disorders. Yeast mutants that fail to package any peroxisomal matrix proteins tested resemble cells from patients with Zellweger syndrome in which there is a failure to import peroxisomal matrix proteins. However, ghosts have not been directly demonstrated morphologically in yeast peroxisome mutants; therefore we cannot be certain whether peroxisomal membrane proteins are present in "ghost-like" structures as seen in Zellweger cells (Santos et al., 1988a, 1992; Wiemer et al., 1989). There is some evidence for a similar type of structure in P. pastorispas5 cells (Spong and Subramani, 1993), and candidate "ghost" structures have been suggested in S. cerevisiae peb4 mutants (Zhang et al., 1993b). In a peroxisome-deficient mutant of//. polymorpha peroxisomal membrane proteins are found in protein-phospholipid aggregates, suggested to represent remnants of the peroxisomal membrane (Suiter et al., 1993). The severest of the human peroxisomal disorders, characterized by a general loss of peroxisomal function, include Zellweger syndrome (Goldfischer et al., 1973), neonatal adrenoleukodystrophy (Kelley et al., 1986), hyperpipecolic acidemia (Gatfield et al., 1968), and infantile Refsum's disease (Scotto et al., 1982). Complementation studies involving restoration of several peroxisomal functions using fibroblasts from these patients have revealed at least ten complementation groups (see Subramani, 1993). All of these complementation groups are deficient in plasmalogen biosynthesis (Roscher et al., 1989) and in p-oxidation of fatty acids (van den Bosch et al., 1992). Thus it appears that peroxisome assembly is a complex process that requires the presence of many gene products to occur correctly. The fact that peroxisome targeting signals may be located at an internal position in the protein suggests that the protein should either be in an unfolded state to expose the signal or that the signal must be exposed on the outer surface of a folded protein. It is generally thought that fiiUy folded proteins are not competent for import into organelles (Pfanner et al., 1988). Cytoplasmic chaperones are involved in protein folding and transport to organelles such as mitochondria and the endoplasmic reticulum (Deshaies et al., 1988; Murakami et al, 1988; Kang et al, 1990). They
Peroxisome Biogenesis
201
are thought to maintain the protein in an unfolded state for the initial stage of translocation, after which folding takes place. In Escherichia coli the leader sequence has been shown to have a negative influence on protein folding, thus facilitating the chaperone in its ability to maintain the transport-competent conformation (Liu et al., 1989). It is possible that this could also be a function of mitochondrial presequences and of PTS2 of mammalian peroxisomal thiolase. In some cases refolding of a protein following translocation has also been shown to require the assistance of chaperones, as is the case with mitochondrial proteins that require hsp60 and ATP to assume their refolded state (Ostermann et al., 1989). The degree to which peroxisomal proteins must be unfolded during translocation and whether a chaperone is involved in this process are not yet clear. It has recently been reported that hsp70 is associated with rat liver peroxisomes (Walton et al., 1994). However, its role, if any, in import is unclear; it was postulated that hsp70 could interact with the peroxisome targeting signal to allow subsequent access to the receptor. Although the premise that proteins must be in an unfolded state to be importcompetent is generally accepted in the case of proteins entering mitochondria or the endoplasmic reticulum, this has not been proved for peroxisomal proteins. Many peroxisomal proteins are oligomeric, and there is recent evidence suggesting that some of these proteins may be imported as oligomers. When octameric alcohol oxidase from P. pastoris peroxisomes was microinjected into mammalian cells it appeared to be incorporated into punctate structures that were identified as peroxisomes by immunofluorescent analysis (Walton et al., 1992). Peroxisomal thiolase oiS. cerevisiae forms a homodimer. By utilizing different forms of monomers (wild type, truncated, and epitope tagged) Glover et al. (1994a) provided evidence suggesting that dimerization takes place in the cytosol and that these dimers are then translocated into peroxisomes. Furthermore, McNew and Goodman have recently reported that chloramphenicol acetyltransferase (CAT) can be targeted to peroxisomes of 5. cerevisiae if the SKL tripeptide is attached to one subunit of this teteromeric protein (McNew and Goodman, 1994). They demonstrated that trimerization of CAT occurred soon after its synthesis, whereas import occurred over a much longer time course, suggesting that CAT is imported into peroxisomes as an oligomer.
VI. SUMMARY It appears that our knowledge in the field of peroxisome biogenesis and regulation of peroxisome proUferation is growing in an exponential fashion. The next few years should prove to be exciting ones, as genes complementing peroxisome-deficient mutants continue to be cloned and analyzed. The regulation ofperoxisome proliferation will be better understood at the molecular level as transcription factors involved in this process are identified in both mammals and yeast. Carefiil dissection ofthe components involved in the signal transduction pathway leading to activation of genes encoding
202
GILLIAN M. SMALL
peroxisomal proteins should shed light on the mechanisms involved in peroxisome assembly, disassembly, and proliferation. With this knowledge in hand we will be in a stronger position to understand, diagnose, and treat human peroxisomal disorders.
ACKNOWLEDGMENTS I would like to thank Drs. Avrom Caplan and Jing Wei Zhang for critical reading of the manuscript. The work carried out in my own laboratory was supported by grants from the American Heart Association (AHA 92-850) and from the ILSI Risk Science Institute. The author is a recipient of an American Heart Association Established Investigatorship.
REFERENCES Aitchison, J. D., Murray, W. W., & Rachubinski, R. A. (1991a). The carboxyl-terminal tripeptide Ala-Lys-Ile is essential for targeting Candida tropicalis trifunctional enzyme to yeast peroxisomes. J. Biol. Chem. 266, 23197-23203. Aitchison, J. D., Sloots, J. A., Nuttley, W. M., & Rachubinski, R. A. (1991b). Sequence of the gene encoding Candida tropicalis peroxisomal trifunctional enzyme. Gene 105, 135-136. Andre, P. J. & Small, G. M. (1993). Characterization of human peroxisome proliferation-activated receptor. Mol. Biol. Cell 4, 106a. Baumgart, E., Stegmeier, K., Schmidt, F. H., & Fahimi, H. D. (1987). Proliferation of peroxisomes in pericentral hepatocytes of rat liver after administration of a new hypocholesterolemic agent (BM 15766). Sex-dependent ultrastructural differences. Lab. Invest. 56, 554-564. Baumgart, E., Volkl, A., Hashimoto, T, & Fahimi, H. D. (1989). Biogenesis of peroxisomes: Immunocytochemical investigation of peroxisomal membrane proteins in proliferating rat liver peroxisomes and in catalase-negative membrane loops. J. Cell Biol. 108, 2221-2231. Bronfman, M., Amigo, L., & Morales, M. N. (1986). Activation of hypolipidemic drugs to acyl-coenzyme Athioesters. Biochem. J. 239, 781-784. Bronfman, M., Orellana, A., Morales, M. N., Bieri, F., Waechter, F., Staubli, W., & Bentley, R (1989). Potentiation of diacylglycerol-activated protein kinase C by acyl-coenzyme A thioesters of hypolipidemic drugs. Biochem. Biophys. Res. Commun. 159, 1026-1031. Bronfman, M., Morales, M. N., Amigo, L., Orellana, A., Nunez, L., Cardenas, L., & Hidalgo, P. C. (1992). Hypolipidaemic drugs are activated to acyl-CoA esters in isolated rat hepatocytes. Biochem. J. 284, 289-295. Carson-Jurica, M. A., Schrader, W. T, & O'Malley, B. W. (1990). Steroid receptor family: structure and functions. Endocr. Rev. 11, 201-218. Cattley, R. C. & Popp, J. A. (1989). Differences between the promoting activities of the peroxisome proliferator WY-14,643 and phenobarbital in rat liver. Cancer Res. 49, 3246-3251. Causeret, C, Bentejac, M., & Bugaut, M. (1993). Proteins and enzymes of the peroxisomal membrane in mammals. Biol. Cell 77, 89-104. Chen, N. & Crane, D. I. (1992). Induction of the major integral membrane protein of mouse liver peroxisomes by peroxisome proliferators. Biochem. J. 283, 605-610. Cohen, G., Rapatz, W., & Ruis, H. (1988). Sequence of the Saccharomyces cerevisiae CTAl gene and amino acid sequence of catalase A derived from it. Eur. J. Biochem. 176, 159-163. Cooper, P. J., Danpure, C. J., Wise, P. J., & Guttridge, K. M. (1988). Immuno-electron microscope localization of alanine-glyoxylate aminotransferase in normal human liver and type 1 hyperoxaluric liver. Biochem. Soc. Trans. 16, 627-628.
Peroxisome Biogenesis
203
Cregg, J. M., van der Klei, I. J., Suiter, G. J., Veenhuis, M., & Harder, W. (1990). Peroxisome deficient mutants of Hansenula polymorpha. Yeast 6, 87—97. De Hoop, M. J. & Geert, A. B. (1992). Import of proteins into peroxisomes and other microbodies. Biochem. J. 286, 657-669. Deshaies, R. J., Koch, B. D., Werner-Washbume, M., Craig, E. A., & Schekman, R. (1988). A subfamily of stress proteins facilitates translocation of secretory and mitochondrial precursor polypeptides. Nature 332, 800-805. Dmochowska, A., Dignard, D., Maleszka, R., & Thomas, D. Y. (1990). Structure and transcriptional control of the Saccharomyces cerevisiae POXl gene encoding acyl-coenzyme A oxidase. Gene 88,247-252. Dommes, R, Dommes, V., & Kunau, W-H. (1983). p-Oxidation in Candida tropicalis. J. Biol. Chem. 258, 10846-10852. Douma, A. C, Veenhuis, M., de Koning, W., Evers, M., & Harder, W. (1985). Dihydroxyacetone synthase is localized in the peroxisomal matrix of methanol-grown Hansenula polymorpha. Arch. Microbiol. 143,237-243. Dreyer, C, Krey, G., Keller, H., Givel, F., Helftenbein, G., & Wahli, W. (1992). Control of the peroxisomal p-oxidation pathway by a novel family of nuclear hormone receptors. Cell 68, 879-887. Einerhand, A. W. C, Voom-Brouwer, T. M., Erdmann, R., Kunau, W-H., & Tabak, H. F. (1991). Regulation of transcription of the gene coding for peroxisomal 3-oxoacyl-CoA thiolase of Saccharomyces cerevisiae. Eur. J. Biochem. 200, 113—122. Einerhand, A. W. C , Kos, W T., Distel, B., & Tabak, H. F. (1993). Characterization of a transcriptional control element involved in proliferation of peroxisomes in yeast in response to oleate. Eur. J. Biochem. 214, 323-331. Elgersma, Y, Van den Berg, M., Tabak, H. F., & Distel, B. (1993). An efficient positive selection procedure for the isolation of peroxisomal import and peroxisome assembly mutants of Saccharomyces cerevisiae. Genetics 135, 731—740. Erdmann, R. (1994). The peroxisomal targeting signal of 3-oxoacyl-CoA thiolase from Saccharomyces cerevisiae. Yeast 10, 935-944. Erdmann, R. & Blobel, G. (1994). PMP27 encodes a peroxisomal membrane protein from Saccharomyces cerevisiae involved in peroxisome proliferation. Mol. Biol. Cell 5, 477a. Erdmann, R., Veenhuis, M., Mertens, D., & Kunau, W-H. (1989). Isolation of peroxisome-deficient mutants oiSaccharomyces cerevisiae. Proc. Natl. Acad. Sci. USA 86, 5419-5423. Fahimi, H. D., Reinicke, A., Sujatta, M., Yokota, S., & Ozel, M. (1982). The short- and long-term effects of bezafibrate in the rat. Ann. N.Y Acad. Sci. 386, 111-135. Fahimi, H. D., Baumgart, E., & Volkl, A. (1993). Ultrastructural aspects of the biogenesis of peroxisomes in rat liver. Biochimie 75, 201-208. Foxworthy, R S., White, S. L., Hoover, D. M., & Eacho, R I. (1990). Effect of ciprofibrate, bezafibrate, and LY171883 on peroxisomal p-oxidation in cultured rat, dog, and rhesus monkey hepatocytes. Toxicol. Appl. Pharmacol. 104, 386-394. Fujiki, Y, Fowler, S., Shio, H., Hubbard, A. L., & Lazarow, R B. (1982a). Polypeptide and phospholipid composition of the membrane of rat liver peroxisomes: comparison with endoplasmic reticulum and mitochondrial membranes. J. Cell Biol. 93, 103—110. Fujiki, Y, Hubbard, A. L., Fowler, S., & Lazarow, P. B. (1982b). Isolation of intracellular membranes by means of sodium carbonate treatment: application to endoplasmic reticulum. J. Cell Biol. 93, 97-102. Fujiki, Y, Rachubinski, R. A., & Lazarow, P. B. (1984). Synthesis of a major integral membrane polypeptide of rat liver peroxisomes on free polysomes. Proc. Natl. Acad. Sci. USA 81,7127—7131. Garrard, L. J. & Goodman, J. M. (1989). Two genes encode the major membrane-associated protein of methanol-induced peroxisomes from Candida boidini. J. Biol. Chem. 264, 13929-13937.
204
GILLIAN M. SMALL
Gatfield, P. D., Taller, E., & Hinton, G. G. (1968). Hyperpipecolacetemia: a new metabolic disorder associated with neuropathy and hepatomegaly. Can. Med. Assoc. J. 99, 1215—1233. Glover, J. R., Andrews, D. W., & Rachubinski, R. A. (1994a). Saccharomyces cerevisiae peroxisomal thiolase is imported as a dimer. Proc. Natl. Acad. Sci. USA 91, 10541-10545. Glover, J. R., Andrews, D. W., Subramani, S., & Rachubinski, R. A. (1994b). Mutagenesis of the amino targeting signal of Saccharomyces cerevisiae 3-ketoacyl-CoA thiolase reveals conserved amino acids required for import into peroxisomes in vivo. J. Biol. Chem. 269, 7558—7563. Godecke, S., Eckart, M., Janowicz, Z. A., & Hollenberg, C. P. (1994). Identification of sequences responsible for transcriptional regulation of the strongly expressed methanol oxidase-encoding gene in Hensenula polymorpha. Gene 139, 35-42. Goldfischer, S., Moore, C. L., Johnson, A. B., Spiro, A. J., Valsamis, M. P., Wisniewski, H. K., Ritch, R. H., Norton, W. T., Rapin, L, & Gartner, L. M. (1973). Peroxisomal and mitochondrial defects in the cerebro-hepato-renal syndrome. Science 182, 62-64. Goodman, J. M. (1985). Dihydroxyacetone synthase is an abundant constituent of the methanol-induced peroxisome of Candida boidini. J. Biol. Chem. 260, 7108-7113. Goodman, J. M., Maher, J., Silver, P. A., Pacifico, A., & Sanders, D. (1986). The membrane proteins of the methanol-induced peroxisome oi Candida boidini. J. Biol. Chem. 261, 3464-3468. Gorgas, K. (1985). Serial section analysis of mouse hepatic peroxisomes. Anat. Embryol. 172, 21—32. Gottlicher, M., Widmark, E., Li, Q., & Gustafsson, J-A. (1992). Fatty acids activate a chimera of the clofibrate acid-activated receptor and the glucocorticoid receptor. Proc. Natl. Acad. Sci. USA 89, 4653-4657. Gould, S. J., Keller, G-A., & Subramani, S. (1987). Identification of a peroxisomal targeting signal at the carboxy terminus of firefly luciferase. J. Cell Biol. 105, 2923-2931. Gould, S. J., Keller, G-A., Hosken, N., Wilkinson, J., & Subramani, S. (1989). A conserved tripeptide sorts proteins to peroxisomes. J. Cell Biol. 108, 1657—1664. Gould, S. J., Keller, G-A., Schneider, M., Howell, S. H., Garrard, L. J., Goodman, J. M., Distel, B., Tabak, H., & Subramani, S. (1990). Peroxisomal protein import is conserved between yeast, plants, insects and mammals. EMBO J. 9, 85—90. Gould, S. J., McCoUum, D., Spong, A. R, Heyman, J. A., & Subramani, S. (1992). Development of the yeast Pichia pastoris as a model organism for a genetic and molecular analysis of peroxisome assembly. Yeast 8, 613-628. Green, S. & Wahli, W. (1994). Peroxisome proliferator-activated receptors: finding the orphan a home. Mol. Cell. Endocrinol. 100, 149-153. Guarente, L. (1992). Mechanism and regulation of transcriptional activation in eukaryotes: conserved features from yeasts to humans. In: Transcriptional Regulation (McKnight, S. L., & Yamamoto, K. R., eds.), pp. 1007-1036. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. Guarente, L. & Hoar, E. (1984). Upstream activation sites of the CYCl gene oi Saccharomyces cerevisiae are active when inverted but not when placed downstream of the "TATA box." Proc. Natl. Acad. Sci. USA 81, 7860-7864. Hajra, A. K. & Bishop, J. E. (1982). Glycerolipid biosynthesis in peroxisomes via the acyl dihydroxyacetone pathways. Ann. N.Y. Acad. Sci. 386, 170-182. Hansen, H., Didion, T., Thiemann, A., Veenhuis, M., & Roggenkamp, R. (1992). Targeting sequences of the two major peroxisomal proteins in the methylotrophic yeast Hansenula polymorpha. Mol. Gen. Genet. 135,269-278. Hashimoto, T., Kuwabara, T., Usuda, N., & Nagata, T. (1986). Purification of membrane polypeptides of rat liver peroxisomes. J. Biochem. 100, 301—310. Heinemann, R & Just, W. W. (1992). Peroxisomal protein import. FEBS Lett. 300,179-182. Hertz, R., Berman, I., & Bar-Tana, J. (1994). Transcriptional activation by amphipathic carboxylic peroxisomal proliferators is induced by the free acid rather than the acyl-CoA derivative. Eur. J. Biochem. 221, 611-615.
Peroxisome Biogenesis
205
Hiltunen, J. K., Wenzel, B., Beyer, A., Erdmann, R., Fossa, A., & Kunau, W. H. (1992). Peroxisomal multifunctional p-oxidation protein of Sacchammyces cerevisiae. Molecular analysis of the FOX2 gene and gene product. J. Biol. Chem. 267, 6646-6653. Hohfelf, J., Veenhuis, M., & Kunau, W-H. (1991). PAS3, a Sacchammyces cerevisiae gene encoding a peroxisomal integral membrane protein essential for peroxisome biogenesis. J. Cell Biol. 114, 1167-1178. Igual, J. C, Matallana, E., Gonzalez-Bosch, C, Franco, L., & Perez-Ortin, J. E. (1991). A new glucose-repressible gene identified from the analysis of chromatin structure in deletion mutants of yeast SUC2 locus. Yeast 7, 379-389. Igual, J. C, Gonzales-Bosch, C, Franco, L., & Perez-Ortin, J. E. (1992). The POTJ gene for yeast peroxisomal thiolase is subject to three different mechanisms of regulation. Mol. Microbiol. 6, 1867-1875. Imanaka, T., Lazarow, P. B., & Takano, T. (1991). A novel 57 kDa peroxisomal membrane polypeptide detected by monoclonal antibody (PXMla/207B). Biochim. Biophys. Acta 1062, 264-270. Ishii, H., Horie, S., & Suga, T. (1980). Physiological role of peroxisomal p-oxidation in liver of fasted rats. J. Biochem. 87, 1855-1858. Issemann, I, & Green, S. (1990). Activation of a member of the steroid hormone receptor superfamily by peroxisome proliferators. Nature 347, 645-650. Just, W. W. & Hartl, F. U. (1987). Biogenesis of rat liver peroxisomal membrane polypeptides. In: Peroxisomes in Biology and Medicine (Fahimi, H. D. & Sies, H., eds.), pp. 402-416. SpringerVerlag, Berlin. Kaldi, K., Diestelkotter, R, Stenbeck, G., Auerbach, S., Jakle, U., Magert, H-J., Wieland, F. T., & Just, W. W. (1993). Membrane topology of the 22 kDa integral peroxisomal membrane protein. FEBS Lett. 315, 217-222. Kamijo, K., Taketani, S., Yokota, S., Osumi, T., & Hashimoto, T. (1990). The 70-kDa peroxisomal membrane protein is a member of the Mdr (P-glycoprotein)-related ATP-binding protein superfamily. J. Biol. Chem. 265, 4534-4540. Kamiryo, T., Sakasegawa, Y, & Tan, H. (1989). Expression and transport of Candida tropicalis peroxisomal acyl-CoA oxidase in the yeast Candida maltosa. Agric. Biol. Chem. 53, 179-186. Kang, P-J., Ostermann, J., Shilling, J., Neupert, W., Craig, E. A., & Pfanner, N. (1990). Requirement for hsp70 in the mitochondrial matrix for translocation and folding of precursor proteins. Nature 348, 137-143. Keller, G-A., Gould, S., Deluca, M., & Subramani, S. (1987). Firefly luciferase is targeted to peroxisomes in mammalian cells. Proc. Natl. Acad. Sci. USA 84, 3264—3268. Keller, G-A., Krisans, S., Gould, S. J., Sommer, J. M., Wang, C. C, Schliebs, W., Kunau, W., Brody, S., & Subramani, S. (1991). Evolutionary conservation of a microbody targeting signal that targets proteins to peroxisomes, glyoxysomes and glycosomes. J. Cell Biol. 114, 893-904. Kelley, R. I., Datta, N. S., Dobyns, W. B., Hajra, A. K., & Moser, A. B. (1986). Neonatal adrenoleukodystrophy: new cases, biochemical studies, and differentiation from Zellweger and related peroxisomal polydistrophy syndromes. Am. J. Med. Genet. 23, 869-901. Kliewer, S. A., Umesono, K., Noonan, D. J., Heyman, R. A., & Evans, R. M. (1992). Convergence of 9-cis retinoic acid and peroxisome proliferator signalling pathways through heterodimer formation of their receptors. Nature 358, 771—774. Klucis, E., Crane, D. I., Hughes, J. L., Poulos, A., & Masters, C. J. (1991). Identification of a catalase-negative sub-population of peroxisomes induced in mouse liver by clofibrate. Biochim. Biophys. Acta 1074,294-301. Koster, A., Heisig, M., Heinrich, P. C, & Wilhelm, W. J. (1986). In vitro synthesis of peroxisomal membrane polypeptides. Biochem. Biophys. Res. Commun. 137, 626-632. Kragler, F., Langeder, A., Raupachova, J., Binder, M., & Hartig, A. (1993). Two independent peroxisomal targeting signals in catalase A of Saccharomyces cerevisiae. J. Cell Biol. 120, 665-673.
206
GILLIAN M. SMALL
Krisans, S. K., Thompson, S. L., Pena, L. A., Kok, E., & Javitt, N. B. (1985). Bile acid synthesis by rat liver peroxisomes: metabolism of 26-hydroxy-cholesterol to 3 p-hydroxy-5-cholenoic acid. J. Lipid Res. 26, 1324-1332. Lake, B. G. (1993). Role of oxidative stress and enhanced cell replication in the hepatocarcinogenicity of peroxisome proliferators. In: Peroxisomes: Biology and Importance in Toxicology and Medicine (Gibson, G. & Lake, B., eds.), pp. 595-618. Taylor and Francis, London. Lazarow, P. B. (1987). The role of peroxisomes in mammalian cellular metabolism. J. Inherit. Metab. Dis. 10, 11-22. Lazarow, P. B. (1988). Peroxisomes. In: The Liver: Biology and Pathobiology (Arias, I. M., Jakoby, W. B., Popper, H., Schachter, D., & Schafritz, D. A., eds.), pp. 241-254. Raven Press, New York. Lazarow, P. B. & Fujiki, Y. (1985). Biogenesis of peroxisomes. Annu. Rev. Cell Biol. 1,489-530. Lazarow, P. B. & Moser, H. W. (1989). Disorders of peroxisomal biogenesis. In: The Metabolic Basis of Inherited Diseases (Scriver, C. R., Beaudet, A. L., Sly, W. S., & Valle, D., eds.), pp. 1479-1509. McGraw-Hill, New York. Lazarow, P. B., Shio, H., & Robbi, M. (1980). Biogenesis of peroxisomes and the peroxisome reticulum hypothesis. In: Biological Chemistry of Organelle Formation (Bucher, T., Sebald, W., & Weiss, H., eds.), pp. 187-206. Springer-Verlag, Berlin. Le Hir, M. & Dubach, U. C. (1980). The activities of peroxisomal oxidases in periportal and prevenous zones of the rat liver acinus. Histochemistry 69, 95-99. Liu, G., Topping, T. B., & Randall, L. L. (1989). Physiological role during export for the retardation of folding by the leader peptide of maltose-binding protein. Proc. Natl. Acad. Sci. USA 86, 9213-9217. Liu, H., Tan, X., Veenhuis, M., McCollum, D., & Cregg, J. M. (1992). An efficient screen for peroxisome-deficient mutants of Pichia pastoris. J. Bacteriol. 174, 4943-4951. Lock, E. A., Mitchell, A. M., & Elcombe, C. R. (1989). Biochemical mechanisms of induction of hepatic peroxisome proliferation. Annu. Rev. Pharmacol. Toxicol. 29, 145-163. Luers, G., Hashimoto, T, Fahimi, H. D., & Volkl, A. (1993). Biogenesis of peroxisomes: isolation and characterization of two distinct peroxisomal populations from normal and regenerating rat liver. J. Cell Biol. 121, 1271-1280. Mangelsdorf, D. J., Umesono, K., Kliewer, S. A., Borgmeyer, U., Ong, E. S., & Evans, R. M. (1991). A direct repeat in the cellular retinol-binding protein type 11 gene confers differential regulation by RXR and RAR. Cell 66, 555-561. Marshall, R A., Krimkevich, Y I., Lark, R., & Goodman, J. M. (1994). PMP24 from Saccharomyces cerevisiae is homologous to PMPs 31-32 of Candida boidini. Mol. Biol. Cell 5, 477a. McCammon, M. T., Dowds, C. A., Orth, K., Moomaw, C. R., Slaughter, C. A., 8c Goodman, J. M. (1990a). Sorting of peroxisomal membrane protein PMP47 from Candida boidinii into peroxisomal membranes of Saccharomyces cerevisiae J. Biol. Chem. 265, 20098-20105. McCammon, M. T., Veenhuis, M., Trapp, S. B., & Goodman, J. M. (1990b). Association of glyoxylate and Beta-oxidation enzymes with peroxisomes of Saccharomyces cerevisiae. J. Bacteriol. 172, 5816-5827. McCammon, M. T., McNew, J. A., Willy, P. J., & Goodman, J. M. (1994). An internal region of the peroxisomal protein PMP47 is essential for sorting to peroxisomes. J. Cell Biol. 124, 915—925. McCollum, D., Monosov, E., & Subramani, S. (1993). IhtpasS mutant of Pichia pastoris exhibits the peroxisomal protein import deficiencies of Zellweger syndrome cells: The PAS8 protein binds to the COOH-terminal tripeptide peroxisomal targeting signal, and is a member of the TPR protein family. J. Cell Biol. 121, 761-774. McNew, J. A. & Goodman, J. M. (1994). An oligomeric protein is imported into peroxisomes in vivo. J. Cell Biol. 127, 1245-1257. Miner, J. N. & Yamamoto, K. R. (1991). Regulatory crosstalk at composite response elements. Trends Biochem. Sci. 16, 423-426.
Peroxisome Biogenesis
207
Morand, O. H., Allen, L-A. H., Zoeller, R. A., & Raetz, C. R. H. (1990). A rapid selection for animal cell mutants with defective peroxisomes. Biochim. Biophys. Acta 1034, 132-141. Mosser, J., Douar, A-M., Sarde, C-0., FCioschis, R, Feil, R., Moser, H., Poustka, A-M., Mandel, J-L., & Aubourg, R (1993). Putative X-linked adrenoleukodystrophy gene shares unexpected homology with ABC transporters. Nature 361, 726-730. Mukherjee, R., Jow, L., Noonan, D., & McDonnell, D. P. (1994). Human and rat peroxisome proliferator activated receptors (PPARs) demonstrate similar tissue distribution but different responsiveness to PPAR activators. J. Steroid Biochem. Mol. Biol. 51, 157-166. Murakami, H., Pain, D., & Blobel, G. (1988). 70-kD heat shock-related protein is one of at least two distinct cytosolic factors stimulating protein import into mitochondria. J. Cell Biol. 107, 2051— 2057. Neat, C. E., Thomassen, M. S., & Osmundsen, H. (1980). Induction of peroxisomal P-oxidation in rat liver by high fat diets. Biochem. J. 186, 369-371. Neat, C. E., Thomassen, M. S., & Osmundsen, H. (1981). Effects of high-fat diets on hepatic fatty acid oxidation in the rat. Isolation of rat liver peroxisomes by vertical-rotor, iso-osmotic, Percoll gradient. Biochem. J. 196, 149-159. Nemali, M. R., Usuda, N., Reddy, M. K., Oyasu, K., Hashimoto, T., Osumi, T, Rao, M. S., & Reddy, J. K. (1988). Comparison of constitutive and inducible levels of expression of peroxisomal p-oxidation and catalase genes in liver and extrahepatic tissues of rat. Cancer Res. 48,5316-5324. Noguchi, T. & Takada, Y. (1979). Peroxisomal localization of alanine:glyoxylate aminotransferase in human liver. Arch. Biochem. Biophys. 196, 645-647. Novikoff, A. B. & Shin, W.-Y. (1964). The endoplasmic reticulum in the Golgi zone and its relations to microbodies, Golgi apparatus and autophagic vacuoles in rat liver cells. J. Microsc. 3, 187—206. Nuttley, W. M., Aitchison, J. D., & Rachubinski, R. A. (1988). cDNA cloning and primary structure determination of the peroxisomal trifunctional enzyme hydratase-dehydrogenase-epimerase from the yeast Candida tropicalis pK233. Gene 69, 171-180. Nuttley, W. M., Bodnar, A. G., Mangroo, D., & Rachubinski, R. A. (1990). Isolation and characterization of membranes from oleic acid-induced peroxisomes of Candida tropicalis. J. Cell Sci. 95, 463-470. Nuttley, W. M., Brade, A. M., Gaillardin, C, Eitzen, G. A., Glover, J. R., Aitchison, J. D., & Rachubinski, R. A. (1993). Rapid identification and characterization of peroxisomal assembly mutants in Yarrowia lipolytica. Yeast 9, 507-517. Okuno, E., Minatogawa, Y, Nakanishi, J., Nakamura, M., Kamoda, N., Makino, M., & Kido, R. (1979). The subcellular distribution of alanine-glyoxylate aminotransferase and serine-pyruvate aminotransferase in dog liver. Biochem. J. 182, 877-879. Ostermann, J., Horwich, A. L., Neupert, W., & Hartl, F-U. (1989). Protein folding in mitochondria requires complex formation with hsp60 and ATP hydrolysis. Nature 341, 125—130. Osumi, T., Tsukamoto, T., Hata, S., Tokota, S., Miura, S., Fujiki, Y, Hijikata, M., Miyazawa, S., & Hashimoto, T. (1991a). Amino-terminal presequence of the precursor of peroxisomal 3-ketoacylCoA thiolase is a cleavable signal peptide for peroxisomal targeting. Biochem. Biophys. Res. Commun. 181,947-954. Osumi, T., Wen, J.-K., & Hashimoto, T. (1991b). Two c/j-acting regulatory sequences in the peroxisome proliferator-responsive enhancer region of rat acyl-CoA oxidase gene. Biochem. Biophys. Res. Commun. 175,866-871. Pfanner, N., Hartl, F.-U., & Neupert, W. (1988). Import of proteins into mitochondria: a multi-step process. Eur. J. Biochem. 175, 205-212. Popp, J. A. & Cattley, R. C. (1993). Peroxisome proliferators as initiators and promoters of rodent hepatocarcinogenesis. In: Peroxisomes: Biology and Importance in Toxicology and Medicine (Gibson, G. & Lake, B., eds.), pp. 653-665. Taylor and Francis, London. Purdue, P. E. & Lazarow, P. B. (1994). Peroxisomal biogenesis: multiple pathways of protein import. J. Biol. Chem. 269, 30065-30068.
208
GILLIAN M. SMALL
Purdue, P. E., Takada, Y., & Danpure, C. J. (1990). Identification of mutations associated with peroxisome-to-mitochondrion mistargeting of alanine/glyoxylate aminotransferase in primary hyperoxaluria type 1. J. Cell Biol. 111, 2341-2351. Rao, M. S. & Reddy, J. K. (1987). Peroxisome proliferation and hepatocarcinogenesis Carcinogenesis 8,631-636. Rao, M. S. & Reddy, J. K. (1991). An overview of peroxisome proliferator-induced hepatocarcinogenesis. Environ. Health Perspect. 93, 205-209. Reddy, J. K. & Lalwani, N. D. (1983). Carcinogenesis by hepatic peroxisome proliferators: Evaluation of the risk of hypolipidemic drugs and industrial plasticizers to humans. Crit. Rev. Toxicol. 12, 1-58. Reddy, J. K. & Rao, M. S. (1989). Oxidative DNA damage caused by persistent peroxisome proliferation: its role in hepatocarcinogenesis. Mutat. Res. 214, 63-68. Reddy, J. K., Lalwani, N. D., Dabholkar, A. S., Reddy, M. K., & Qureshi, S. A. (1981). Increased peroxisomal activity in the liver of vitamin E deficient rats. Biochem. Int. 3, 41—49. Reddy, J. K., Lalwani, N. D., Reddy, M. K., & Qureshi, S. A. (1982). Excessive accumulation of autofluorescent lipofuscin in the liver during hepatocarcinogenesis by methyl clofenepate and other hypolipidemic peroxisome proliferators. Cancer Res. 42, 259-266. Reddy, J. K., Goel, S. K., Nemali, M. R., Carrino, J. J., Laffler, T. G., Reddy, M. K., Sperbeck, S. J., Osumi, T., Hashimoto, T., Lalwani, N. D., & Rao, M. S. (1986). Transcriptional regulation of peroxisomal fatty acyl-CoA oxidase and enoyl-CoA hydratase/3-hydroxyacyl-CoA dehydrogenase in rat liver by peroxisome proliferators. Proc. Natl. Acad. Sci. USA 83, 1747—1751. Roggenkamp, R. (1992). Targeting signals for protein import into peroxisomes. Ceil Biochem. Funct. 10, 193-199. Roscher, A. A., Hoefler, S., Hoefler, G., Paschke, E., Paltauf, R, Moser, A. B., & Moser, H. W. (1989). Genetic and phenotypic heterogeneity in disorders of peroxisome biogenesis—a complementation study involving cell lines from 19 patients. Pediatr. Res. 26, 67—72. Rosenkrantz, M., Alam, T., Kim, K-S., Clark, B. J., Srere, P A., & Guarente, L. P (1986). Mitochondrial and nonmitochondrial citrate syntheses in Saccharomyces cerevisiae are encoded by distinct homologous genes. Mol. Cell. Biol. 6, 4509-4515. Santos, M. J., Imanaka, T., Shio, H., Small, G. M., & Lazarow, P. B. (1988a). Peroxisomal membrane ghosts in Zellweger syndrome—aberrant organelle assembly. Science 239, 1536-1538. Santos, M. J., Imanaka, T, Shio, H., & Lazarow, P. B. (1988b). Peroxisomal integral membrane proteins in control and Zellweger fibroblasts. J. Biol. Chem. 263, 10502-10509. Santos, M. J., Hoefler, S., Moser, A. B., Moser, H. W., & Lazarow, P. B. (1992). Peroxisome assembly mutations in humans: Structural heterogeneity in Zellweger syndrome. J. Cell. Physiol. 151, 103-112. Santos, M. J., Kawada, M. E., Espeel, M., Figueroa, C, Alvarez, A., Hidalgo, U., & Metz, C. (1994). Characterization of human peroxisomal membrane proteins. J. Biol. Chem. 269, 24890-24896. Schmidt, A., Endo, N., Rutledge, S. J., Vogel, R., Shinar, D., & Rodan, G. A. (1992). Identification of a new member of the steroid hormone receptor superfamily that is activated by a peroxisome proliferator and fatty acids. Mol. Endocrinol. 6, 1634—1641. Scotto, J. M., Hadchouel, M., Odievre, M., Laudat, M. H., Saudabray, J. M., Dulac, O., Beucler, I., & Beaune, P. (1982). Infantile phytanic acid storage disease, a possible variant of Refsum's disease: three cases including ultrastructure studies of the liver. J. Inherit. Metab. Dis. 5, 83—90. Sher, T., Yi, H-F., McBride, W., & Gonzalez, F. J. (1993). cDNA cloning, chromosomal mapping and functional characterization of the human peroxisome proliferator activated receptor. Biochemistry 32,5598-5604. Shimizu, S., Imanaka, T., Takano, T., & Ohkuma, S. (1992). Induction and characterization of two types of ATPase on rat liver peroxisomes. J. Biochem. 112, 376-384.
Peroxisome Biogenesis
209
Shimozawa, N., Tsukamoto, T., Suzuki, Y, Orii, T., Shirayoshi, Y., Mori, T., & Fujiki, Y (1992a). A human gene responsible for Zellweger syndrome that affects peroxisome assembly. Science 255, 1132-1134. Shimozawa, N., Tsukamoto, T., Suzuki, Y, Orii, T., & Fujiki, Y (1992b). Animal cell mutants represent two complementation groups of peroxisome-defective Zellweger syndrome. J. Clin. Invest. 90, 1864-1870. Shore, D. & Nasmyth, K. (1987). Purification and cloning of a DNA binding protein from yeast that binds to both silencer and activator elements. Cell 51, 721-732. Simon, M., Adam, G., Rapatz, W., Spevak, W., & Ruis, H. (1991). The Saccharomyces cerevisiaeADRl gene is a positive regulator of transcription of genes encoding peroxisomal proteins. Mol. Cell. Biol. 11,699-704. Simon, M., Binder, M., Adam, G., Hartig, A., & Ruis, H. (1992). Control of peroxisome proliferation in Saccharomyces cerevisiae by ADRl, SNFl (CATl, CCRl) and SNF4 {CA T3). Yeast 8,303-309. Singh, K. K., Small, G. M., & Lewin, A. S. (1992). Alternative topogenic signals in peroxisomal citrate synthase of Saccharomyces cerevisiae. Mol. Cell. Biol. 12, 5593-5599. Skoneczny, M., Chelstowska, A., & Rytka, J. (1988). Study of the coinduction by fatty acids of catalase A and acyl-CoA oxidase in standard and mutant Saccharomyces cerevisiae strains. Eur. J. Biochem. 174,297-302. Small, G. M., Burdett, K., & Connock, M. J. (1982). Clofibrate-induced changes in enzyme activities in liver, kidney and small intestine of male mice. Ann. N.Y Acad. Sci. 386,460-464. Small, G. M., Szabo, L. J., & Lazarow, R B. (1988a). Acyl-CoA oxidase contains two targeting sequences each of which can mediate protein import into peroxisomes. EMBO J. 7, 1167—1173. Small, G. M., Santos, M. J., Imanaka, T., Poulos, A., Danks, D. M., Moser, H. W., & Lazarow, P. B. (1988b). Peroxisomal integral membrane proteins in livers of patients with Zellweger syndrome, infantile Refsum's disease and X-linked adrenoleukodystrophy. J. Inherit. Metab. Dis. 11, 358371. Spong, A. P. & Subramani, S. (1993). Cloning and characterization of PAS5: A gene required for peroxisome biogenesis in the methylotrophic yeast Pichiapastoris. J. Cell Biol. 123, 535-548. Subramani, S. (1993). Protein import into peroxisomes and biogenesis of the organelle. Annu. Rev. Cell Biol. 9, 445-478. Suiter, G. J., Looyenga, L., Veenhuis, M., & Harder, W. (1990). Occurrence of peroxisomal membrane proteins in methylotrophic yeasts grown under different conditions. Yeast 6, 35-43. Suiter, G. J., Vrieling, E. G., Harder, W., & Veenhuis, M. (1993). Synthesis and subcellular location of peroxisomal membrane proteins in a peroxisome-deficient mutant of the yeast Hansenula polymorpha EMBO J. 12, 2205-2210. Suzuki, Y, Orii, T., Takiguchi, M., Mori, M., Hijikata, M., & Hashimoto, T. (1987). Biosynthesis of membrane polypeptides of rat liver peroxisomes. J. Biochem. 101, 491—496. Swinkels, B. W., Gould, S. J., Bodnar, A., Rachubinski, R. A., & Subramani, S. (1991). A novel, cleavable peroxisomal targeting signal at the amino-terminus of the rat 3-ketoacyl-CoA thiolase. EMBO J. 10,3255-3262. Swinkels, B. W., Gould, S. J., & Subramani, S. (1992). Targeting efficiencies of various permutations of the concensus C-terminal tripeptide peroxisomal targeting signal. FEBS Lett. 305, 133—136. Thieringer, R. & Raetz, R. H. (1993). Peroxisome-deficient Chinese hamster ovary cells with point mutations in peroxisome assembly factor-1. J. Biol. Chem. 268, 12631—12636. Thomassen, M. S., Christiansen, E. N., & Norum, K. R. (1982). Characterization of the stimulatory effect of high fat diets on peroxisomal p-oxidation in rat liver. Biochem. J. 206, 195—202. Thompson, S. L., Burrows, R., Laub, R. J., & Krisans, S. K. (1987). Cholesterol synthesis in rat liver peroxisomes: conversion of mevalonic acid to cholesterol. J. Biol. Chem. 262, 17420-17425. Tomaszewski, K. E. & Melnick, R. L. (1994). In vitro evidence for involvement of CoA thioesters in peroxisome proliferation and hypolipidaemia. Biochim. Biophys. Acta 1120,118-124.
210
GILLIAN M. SMALL
Tomaszewski, K. E., Heindel, S. W., Jenkins, W. L., & Melnick, R. L. (1990). Induction of peroxisomal acyl-CoA oxidase activity and lipid peroxidation in primary rat hepatocyte cultures. Toxicology 65, 49-60. Tsukamoto, T., Yokota, S., & Fujiki, Y. (1990). Isolation and characterization of Chinese hamster ovary cell mutants defective in assembly of peroxisomes. J. Cell Biol. 110, 651-660. Tsukamoto, T, Miura, S., & Fujiki, Y. (1991). Restoration by a 35K membrane protein of peroxisome assembly in a peroxisome-deficient mammalian cell mutant. Nature 350, 77-81. Tugwood, J. D., Issemann, I., Anderson, R. G., Bundell, K. R., McPheat, W. L., & Green, S. (1992). The mouse peroxisome proliferator activated receptor recognizes a response element in the 5' flanking sequence of the rat acyl CoA oxidase gene. EMBO J. 11, 433-439. van den Bosch, H., Schutgens, R. B. H., Wanders, R. J. A., & Tager, J. M. (1992). Biochemistry of peroxisomes. Annu. Rev. Biochem. 61,157-197. Van Der Leij, I., Van den Berg, M., Boot, R., Franse, M., Distel, B., & Tabak, H. F. (1992). Isolation of peroxisome assembly mutants from Saccharomyces cerevisiae with different morphologies using a novel positive selection procedure. J. Cell Biol. 119, 153—162. Van Der Leij, I., Franse, M. M., Elgersma, Y, Distel, B., & Tabak, H. F. (1993). PAS 10 is a tetratricopeptide-repeat protein that is essential for the import of most matrix proteins into peroxisomes of Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. USA 90, 11782-11786. Van Dijken, J. R, Veenhuis, M., Kreger-Van Rij, N. J. W., & Harder, W. (1975). Microbodies methanol-assimilating yeasts. Arch. Microbiol. 102, 41—44. Veenhuis, M. (1992). Peroxisome biogenesis and function in Hansenula polymorpha. Cell Biochem. Funct. 10, 175-184. Veenhuis, M. & Goodman, J. M. (1990). Peroxisomal assembly: membrane proliferation precedes the induction of the abundant matrix proteins in the methylotrophic yeast Candida boidini. J. Cell Sci. 96, 583-590. Veenhuis, M. & Harder, W. (1987). Metabolic significance and biogenesis of microbodies in yeast. In: Peroxisomes in Biology and Medicine (Fahimi, H. D. and Sies, H., eds.), pp. 435-457. SpringerVerlag, Berlin. Veenhuis, M., Mateblowski, M., Kunau, W. H., & Harder, W. (1987). Proliferation of microbodies in Saccharomyces cerevisiae. Yeast 3, 77—84. Walton, P A., Gould, S. J., Rachubinski, R. A., Subramani, S., «& Feramisco, J. R. (1992). Transport of microinjected alcohol oxidase from Pichiapastoris into vesicles in mammalian cells: Involvement of the peroxisomal targeting signal. J. Cell Biol. 118, 499-508. Walton, P A., Wendland, M., Subramani, S., Rachubinski, R. A., & Welch, W. J. (1994). Involvement of 70-kD heat-shock proteins in peroxisomal import. J. Cell Biol. 125, 1037-1046. Wang, T. W., Lewin, A. S., & Small, G. M. (1992). A negative regulating element controlling transcription of the gene encoding acyl-CoA oxidase in Saccharomyces cerevisiae. Nucleic Acids Res. 20, 3495-3500. Wang, T, Luo, Y, & Small, G. M. (1994). The POXl gene encoding peroxisomal acyl-CoA oxidase in Saccharomyces cerevisiae is under the control of multiple regulatory elements. J. Biol. Chem. 269, 24480-24485. Waterham, H. R., Titorenko, V I., Van der Klei, I., Harder, W., & Veenhuis, M. (1992). Isolation and characterization of peroxisomal protein import (Pim-) mutants of Hansenula polymorpha. Yeast 8, 961-972. Waterham, H. R., Titorenko, V. I., Swaving, G. J., Harder, W, & Veenhuis, M. (1993). Peroxisomes in the methylotrophic yeast Hansenula polymorpha do not necessarily derive from pre-existing organelles. EMBO J. 12, 4785-4794. Wiemer, E. A., Brul, S., Just, W. W., Van Driel, R., Brouwer-Kelder, E., Van den Berg, M., Weijers, P J., Schutgens, R. B. H., van den Bosch, H., Scgram, A., Wanders, R. J. A., & Tager, J. M. (1989). Presence of peroxisomal membrane proteins in liver and fibroblasts from patients with the
Peroxisome Biogenesis
211
Zellweger syndrome and related disorders: evidence for the existence of peroxisomal ghosts. Eur. J. Cell Biol. 50, 407-417. Yamamoto, K. & Fahimi, H. D. (1987). Three-dimensional reconstruction of a peroxisome reticulum in regenerating rat liver: evidence of interconnections between heterogeneous segments. J. Cell Biol. 105,71^-722. Yokota, S., Oda, T., & Ichiyama, A. (1987). Immunocytochemical localization of serine-pyruvate aminotransferase in peroxisomes of the human liver parenchymal cells. Histochemistry 87, 601-606. Zhang, B., Marcus, S. L., Sajjadi, F. G., Alvares, K., Reddy, J. K., Subramani, S., Rachubinski, R. A., & Capone, J. R (1992). Identification of a peroxisome proliferator-responsive element upstream of the gene encoding rat peroxisomal enoyl-CoA hydratase/3-hydroxyacyl-CoA dehydrogenase. Proc. Natl. Acad. Sci. USA 89, 7541-7545. Zhang, J. W., Luckey, C, & Lazarow, R B. (1993a). Three peroxisome protein packaging pathways suggested by selective permeabilization of yeast mutants defective in peroxisome biogenesis. Mol. Biol. Cell 4, 1351-1359. Zhang, J. W., Han, Y, & Lazarow, R B. (1993b). Novel peroxisome clustering mutants and peroxisome biogenesis mutants of Saccharomyces cerevisiae, J. Cell Biol. 123, 1133—1147. Zhang, X-K., Hoffmann, B., Tran, R B-V, Graupner, G., & Pfahl, M. (1992). Retinoid X receptor is an auxiliary protein for thyroid hormone and retinoic acid receptors. Nature 355, 441-446. Zoeller, R. & Raetz, R. H. (1986). Isolation of animal cell mutants deficient in plasmalogen biosynthesis and peroxisome assembly. Proc. Natl. Acad. Sci. USA 83, 5170-5174. Zoeller, R. A., Allen, L. A., Santos, M. J., Lazarow, R B., Hashimoto, T., Tartakoff, A. M., & Raetz, C. R. H. (1989). Chinese hamster ovary cell mutants defective in peroxisome biogenesis. J. Biol. Chem. 264, 21872-21878.
This Page Intentionally Left Blank
PEROXISOMAL TOPOGENIC SIGNALS AND THE ETIOLOGY OF PEROXISOME-DEFICIENT DISEASE
Yukio Fujiki
I. Introduction 213 II. Topogenic Signals of Peroxisomal Proteins 215 A. In Vitro Approach 215 B. In Vivo Approach 217 III. Peroxisome-Deficient Mammalian Cell Mutants and Complementing Genes . 220 A. Isolation of Peroxisome-Deficient Mammalian Cell Mutants 220 B. Cloning of Genes Required for Peroxisome Biogenesis and the Primary Defects of Human Peroxisome-Deficient Disorders . . . 222 Acknowledgments 226 References 226
1. INTRODUCTION The peroxisome is a ubiquitous intracellular organelle present in eukaryotes. It is classically defined as a subcellular compartment containing catalase and at least
Membrane Protein TVansport Volume 3, pages 213-229. Copyright © 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 1-55938-989-3 213
214
YUKIOFUJIKI
one H202-producing oxidase (de Duve and Baudhuin, 1966). Peroxisomes function in the catabolism of a wide variety of substrates such as fatty acids, D-amino acids, L-a-hydroxy acids, uric acid, and polyamines. Several human genetic disorders with evidence of the absence of peroxisomes are linked to various biochemical dysfunctions. Hence, the peroxisome plays a crucial metabolic role, including the catabolism of very long chain fatty acids by the p-oxidation system, the biosynthesis of ether-linked glycerolipids such as plasmalogens, the metabolism of cholesterol and phytanic acid, and the synthesis of bile acids (Lazarow and Moser, 1989; van den Bosch et al., 1992). Among these disorders, cerebro-hepato-renal syndrome (Zellweger syndrome) is a typical, severe disease. Many lines of biochemical and morphological evidence are consistent with the idea that peroxisomes are formed by division of pre-existing peroxisomes after post-translational import of newly synthesized proteins (Lazarow and Fujiki, 1985; Borst, 1986). Peroxisomal proteins, including membrane polypeptides, are synthesized on free polyribosomes in the cytosol, mostly at their final sizes. Post-translational import of several proteins into peroxisomes has been reproduced in vitro (Fujiki and Lazarow, 1985; Imanaka et al., 1987; Small et al., 1988; Miyazawa et al., 1989; Miura et al., 1992, 1994). Targeting signals have been noted in vivo and in vitro for peroxisomal proteins such as luciferase (Gould et al., 1989) and acyl-CoA oxidase (AOX) of rat liver (Miyazawa et al., 1989; Miura et al., 1992) as well as from Candida tropicalis (Small et al., 1988). One type of topogenic signal identified for a number of enzymes resides at the extreme COOH terminus and comprises the -Ser-Lys-Leu-COOH (SKL)-related tripeptide sequence (Gould et al., 1989; Miyazawa et al., 1989; Miura et al., 1992; Subramani, 1993). Investigations of peroxisome biogenesis and peroxisomal disorders have progressed tremendously in the last decade; these have included the identification of peroxisome targeting signals, isolation of many genes essential for peroxisome assembly, and the discovery of the primary defects in several peroxisomal diseases. At present, peroxisomal diseases include 12 different disorders, which are classified into three groups (Lazarow and Moser, 1989; van den Bosch et al., 1992): (1) those in which peroxisomes are virtually absent and generalized impairment of peroxisomal functions are noted (cerebro-hepato-renal Zellweger syndrome, neonatal adrenoleukodystrophy, and infantile Refsum disease); (2) those in which peroxisomes are present but several peroxisomal functions are impaired (rhizomelic chondrodysplasia punctata and combined peroxisomal P-oxidation enzyme deficiency); (3) those in which peroxisomes are present and only a single peroxisomal function is impaired (X-linked adrenoleukodystrophy, peroxisomal 3-ketoacyl-CoA thiolase deficiency, and acyl-CoA oxidase deficiency). In this chapter I discuss two topics in the study of peroxisome biogenesis; peroxisomal topogenic signals and pathogenic genes of human peroxisome-deficient disorders.
Peroxisomal Topogenic Signals and Disease
215
ir. TOPOGENIC SIGNALS OF PEROXISOMAL PROTEINS To address the question of the nature and location of the topogenic signals of peroxisomal proteins, use has been made of in vitro and in vivo protein transport systems in combination with recombinant DNA technology (Osumi and Fujiki, 1990;Subramani, 1993). A. !n Vitro Approach
Biosynthesis of peroxisomal proteins at their final sizes implies that the topogenic signal(s) for transport of newly synthesized polypeptides to peroxisomes resides in the internal sequence of the proteins. In vitro assay systems of peroxisomal import have been achieved for several proteins, including AOX of rat liver (Fujiki and Lazarow, 1985; Imanaka et al., 1987) and Candida tropicalis (Small et al., 1988), rat liver catalase (Fujiki and Lazarow, 1985), and rat liver 3-ketoacyl-CoAthiolase (Miura et al., 1994). In vitro transcribed and translated, radiolabeled products of a cloned cDNA encoding a protein such as AOX or any constructed cDNA to be examined for the transport of its product to peroxisomes, are incubated with purified peroxisomes. Peroxisomal import is assessed by co-sedimentation of these polypeptides with peroxisomes, by resistance of the polypeptides to proteolysis with externally added protease such as proteinase K, and by abolishing this resistance by addition of a detergent prior to the protease treatment. Peroxisomal import is time dependent, occurs at 26°C or 37°C but not at 0°C, and requires energy (ATP). Minimal Targeting Signal of AOX We investigated the peroxisome-targeting signal in vitro, with various deletion and fusion mutants of AOX. In our earlier studies, we found that the peroxisometargeting signal of AOX resides at the COOH terminus within five (LQSKL) or fewer amino acid residues of the extreme terminus (Miyazawa et al., 1989: Osumi and Fujiki, 1990). To elucidate the minimum sequence and positional requirement of the SKL translocation signal and to determine whether the analogous COOH-terminal sequences such as SHL and SRL found in peroxisomal enzymes function as targeting signals, we carried out in vitro import experiments of AOX variants, including those with modified SKL sequence, as well as chimera proteins (Miura et al., 1992). Wild-type, fiill-length, and various types of truncated AOX were synthesized in a cell-free system, and were then post-translationally incubated for 60 min at 26°C withfi-eshlyand highly purified rat liver peroxisomes. Full-length AOX was imported into peroxisomes, partly with conversion of the 75-kDa Achain to 53-kDa B and 22-kDa C chains by proteolytic cleavage at residues 428-429 as noted in vivo (Furuta et al., 1982; Miura et al., 1984). An AOX variant in which the COOH-terminal tripeptide SKL was removed fi-om the wild-type AOX was not
216
YUKIOFUJIKI
taken up by peroxisomes. Furthermore, when AOX was deleted about 70 to 230 amino acid residues from the COOH terminus, the resulting polypeptides were not imported into peroxisomes. When the tripeptide SKL was linked to the carboxyl terminus of the truncated and import-incompetent AOX, the resulting proteins were translocated to peroxisomes, as verified by the resistance to the digestion with protease. Linkage of the COOH-terminal pentapeptide LQSKL likewise restored the import of AOX deleted in the COOH-terminal region. Next, rat liver catalase with deletion of about 150 residues from the COOH terminus was used as a heterologous protein for fusion with AOX COOH-terminal residues of 3 and 5 amino acids. Two types of fusion polypeptides with SKL and LQSKL were recovered with peroxisomes and were found to be partly resistant to externally added protease, indicating that SKL and LQSKL possessed the topogenic activity to translocate the import-negative truncated catalase into peroxisomes. Taken together, we concluded that the minimum sequence of a peroxisome-targeting signal is within the tripeptide SKL. In attempts to identify the minimum peroxisome-targeting signal, various AOX mutants in which the SKL sequence was either mutated or deleted by site-directed mutagenesis were examined for their import activity (Miura et al., 1992). None of the AOX mutants with deletion of any one or two residues in the SKL was translocated to peroxisomes. .The AOX variants, with the mutation of lysine in the SKL to arginine or histidine, were imported into peroxisomes with lesser import activity; 34% with SHL, 86% with SRL of the wild-type SKL (100%). With substitution of glutamic acid for lysine, however, the AOX mutant became importnegative. A C-terminal sequence, -Ala-Lys-Leu-COOH similar to SKL, is found for nonspecific lipid transfer protein (sterol carrier protein 2) localized in rat liver peroxisomes (Tsuneokaetal., 1988; Keller etal., 1989; Mori etal., 1991). We tested if the AKL sequence functions as a peroxisome-targeting signal, by the mutation of serine to alanine in the SKL of AOX. The AOX-AKL was imported to peroxisomes as much (78%) as the wild type. These resuhs indicate that the topogenic signal requires all residues of SKL, and that lysine is replaceable by other basic amino acids, arginine and histidine, but cannot be substituted by an acidic residue such as glutamic acid. The sequence AKL is also functional as a targeting signal. Inhibition of Import by Synthetic Peptides
To confirm the finding that SKL is a minimum peroxisome-targeting sequence, we studied the effect on AOX import of various synthetic peptides from which SKL was modified in vitro (Miura et al., 1992). The wild-type peptide comprising ten COOH-terminal amino acid residues of AOX, KHLKPLQSKL-COOH, at 20 ILIM, reduced import of full-length AOX by 40%, and by 70% as the concentration of peptide was increased to 200 |LIM. The inhibition by peptide of AOX import may occur at the step of binding of AOX to peroxisomes. In contrast, the peptide of the AOX sequence 649-658, SYHKHLKPLQ-COOH, lacking COOH-terminal SKL
Peroxisomal Topogenic Signals and Disease
217
but containing 10 amino acid residues, did not have this effect, even at 200 \xM. Another 10-amino acid peptide, KHLKSKLPLQ-COOH, in which SKL had been inserted in the middle of the sequence, apparently had no inhibitory effect on AOX import. Similar findings were reported, in that mutation of SKL to SKLS or SKLIK abolishes the translocation of luciferase into peroxisomes in vivo (Gould et al., 1989). The authentic peptide, but with the a-carboxyl group of COOH-terminal leucine amidated, did not decrease the peroxisomal uptake of AOX, indicating the importance of the free a-carboxyl group of the COOH-terminal leucine. The peptide with glutamic acid substituted for lysine in the SKL sequence did not affect the import of wild-type AOX. Accordingly, we proved by the in vitro import studies that the COOH-terminal SKL-related tripeptide, the SKL motif, functions as the peroxisomal topogenic signal when in full sequence and when located at the extreme COOH terminus. The tripeptide SKL motif is found in a number of peroxisomal proteins thus far sequenced. All of the in vitro findings discussed here are essentially in good agreement with those obtained in in vivo experiments (see below), where the mutated and fused firefly luciferase genes were transfected to mammalian cells (Gould et al., 1989). B. In Vivo Approach The SKL Motif
Keller et al. (1987) introduced cloned firefly luciferase into CV-1 monkey kidney cells in culture and found by double immunoelectron microscopy that the expressed luciferase was localized in peroxisomes. A number of mutant luciferase proteins were likewise tested for the import to peroxisomes. When the carboxyl-terminal 12 amino acid residues and the terminal SKL were deleted, the mutant proteins were no longer imported to peroxisomes. These carboxyl-terminal sequences directed fusion proteins to peroxisomes, when linked to the carboxyl terminus of cytosolic proteins such as dihydrofolate reductase and chloramphenicol acetyltransferase. These results suggested that the tripeptide SKL functions in vivo as a peroxisometargeting signal (Gould et al., 1989). This SKL sequence functions not only as a peroxisomal topogenic signal in animal, plant, and yeast (Gould et al., 1990), but also as a targeting signal to the glycosome (a member of the peroxisome family) of a unicellular parasite, Trypanosome{B\sinnQrQt2i\., 1992; Sommeretal., 1992). As a glycosomal targeting signal, however, a much broader spectrum of the tripeptide sequence is accepted (Sommer et al., 1992). In the SKL peroxisomal targeting signal, certain conservative substitutions are permitted without loss of the targeting activity, in the first and second amino acids (serine and lysine, respectively) but not at the COOH-terminal leucine. In mammalian cells, serine functions as efficiently as alanine, but better than cysteine in the first position; lysine and arginine work equally well and more efficiently than histidine in the second position; leucine is more efficient than methionine in the last position (Swinkels et al., 1992). Accordingly, the consensus sequence could be defined as -S/A/C-K/R/H-L/M.
218
YUKIO FUJIKI
Studies in in vitro or in vivo systems have their respective advantages and disadvantages and are complementary. The immunofluorescence assay, employed in the in vivo experiments such as those described above, is qualitative. It is, therefore, rather difficult to evaluate the import efficiency among the various proteins. Failure in vivo to detect low-efficiency import may result in accumulation in the cytosol of the polypeptides examined. On the other hand, the in vitro import assay allows us to discriminate the activities of polypeptides tested, although the fundamental issue may remain obscure: to what extent do the findings in in vitro studies reflect the physiological occurrence? Nevertheless, the investigations both in vivo and in vitro have reached the same conclusion on the COOH-terminal peroxisome-targeting signal (Table 1). The notion of the SKL motif sequence present in mammals, plants, insects, and yeast implies that the mechanism of protein translocation to peroxisomes using the SKL motif signal has been conserved throughout eukaryotic evolution (Osumi and Fujiki, 1990; Subramani, 1993). A targeting signal with a short and well-conserved motif is rather unique, in contrast to other topogenic signals responsible for the transport of newly synthesized polypeptides to organelles such as mitochondria and endoplasmic reticulum mediated by NH2-terminal cleavable extra peptides with longer, but no consensus sequence. AminO'Terminal Peroxisome Targeting Signal
More than 60% of peroxisomal proteins contain carboxyl-terminal sequences unrelated to the SKL motif, implying that non-SKL peroxisomal topogenic signals function in their transport (Subramani, 1993). Peroxisomal 3-ketoacyl-CoA thiolase, a fatty acid P-oxidation enzyme, is exceptional in that it is synthesized as a larger precursor (Miura et al., 1984; Fujiki et al., 1985), with an amino-terminal presequence of 26 amino acid residues or 36 residues derivedfromthe thiolase gene B or A, respectively (Table 1) (Hijikata et al., 1990). Osumi and his associates
Table 1. Peroxisome Targeting Signals Type C-terminal SKL motif -Ser/Ala-Lys/Arg/His-Leu-COOH and related tripeptide sequence [. Presequence of 3-ketoacyl-CoA thiolase -36 -26 -20 -10 -1 A MSESVGRTSAMHRLQVVLGHLAGRPESSSALQAAPC B MHRLQVVLGHLAGRSESSSALQAAPC
Cleavage
Topogenic activity proven
No
In vitro/in vivo
Yes
In vivo
Note: Amino acid residues indispensable for the topogenic activity in the thiolase presequences are highlighted by holding.
Peroxisomal Topogenic Signals and Disease
219
reported, in collaboration with our group, that the NH2-terminal presequences of 26 and 36 amino acids target a reporter protein to peroxisomes in vivo as assessed in both transient and stable fusion protein-expression experiments (Osumi et al., 1991; Tsukamoto et al., 1994). The fusion proteins of the NH2-terminal thiolase presequences linked to the NH2 terminus of Escherichia coli dihydrofolate reductase were detected inside the peroxisomes by immunofluorescence and immunoelectron microscopy when the fusion protein cDNAs were expressed in cultured Chinese hamster ovary cells. Moreover, the fusion proteins were correctly processed to their expected mature forms. The charged residues, arginine at position —24 (starting at the NH2-terminal residue of mature thiolase as 1) and histidine at —17 in the B-type, 26-residue presequence, were both indispensable and were not replaceable even with other basic amino acids. Both of these amino acids are also required in a longer, 36-residue presequence of the thiolase A. The importance of these basic residues is also noted in the 37-residue NH2-terminal extrapeptide of glyoxysomal malate dehydrogenase of watermelon (Gietl et al., 1994). Taken together, the thiolase presequence signal peptide is a newly defined type of peroxisome-targeting signal recognized by a mechanism presumably distinct from that for the SKL motif signal. This second type of peroxisome translocation signal was also identified by Subramani's group (Swinkels et al., 1991). Proteins possessing the SKL motif and the thiolase presequence make up nearly 40% of total peroxisomal proteins (Subramani, 1993). Thereby, it is quite plausible to assume that other types of peroxisomal translocation signals exist and remain to be identified. Small et al. (1988) have found, by in vitro import assay, that AOX (PXP4) of the yeast Candida tropicalis contains a topogenic signal(s) in two regions. A similar conclusion is obtained in in vivo experiments (Kamiryo et al., 1989). An SKL sequence is noted internally in this protein but not in the segments that appear to contain the targeting information. Other type(s) of peroxisome-targeting signal may exist in C tropicalis. Concentration-dependent inhibition of in vitro AOX import by synthetic peptides with the COOH-terminal SKL suggests the presence of some membrane component(s), presumably such as those interacting with the SKL signal (Miura et al., 1992). Although biochemical approaches such as those using protein cross-linking, affinity chromatography, and anti-idiotype antibody, in combination with the sensitive in vitro import assay systems, have been proven successful for the identification of import machinery of mitochondria (Vestwever et al., 1989; SoUner et al., 1989) and chloroplasts (Pain et al., 1988), such methods have not worked out for peroxisomes. Alternatively, an approach by genetic complementation of yeast mutants that are defective in the import of SKL-signal proteins to peroxisomes has been successful in identifying genes apparently encoding a putative SKL receptor including PASS of Pichia pastoris (McCollum et al., 1993) and PASIO of Saccharomyces cerevisiae (Van der Leij et al., 1993).
220
YUKIO FUJIKI
III. PEROXISOME-DEFICIENT MAMMALIAN CELL MUTANTS AND COMPLEMENTING GENES Human peroxisome-deficient disorders are autosomal recessive and include at least nine different genotypes reported to date (Brul et al., 1988; Roscher et al., 1989; Yajima et al., 1992; Shimozawa et al., 1992b, 1993a), with three distinct phenotypes, i.e., cerebro-hepato-renal Zellweger syndrome, neonatal adrenoleukodystrophy, and infantile Refsum disease (Lazarow and Moser, 1989). Goldfischer et al. (1973) reported that no peroxisomes were observed in electron microscopy of liver and kidney from patients with Zellweger syndrome. Recent findings are consistent with the notion that these disorders are closely or tightly linked to the dysfunction of peroxisomes, which causes anomalies including accumulation of very long chain fatty acids in body fluid and very low levels of plasmalogen in various tissues such as brain and erythrocytes (Lazarow and Moser, 1989; van den Bosch et al., 1992). Santos et al. (1988) and Wiemer et al. (1989) noted apparently empty peroxisomal membrane vesicles, "membrane ghosts," in fibroblasts from Zellweger patients by immunoelectron microscopic analysis using antibody to peroxisomal membrane protein, suggesting that import of peroxisomal proteins is impaired in these patients. In general, however, it is not an easy task to work with specimens from patients and manipulate the cells, even fibroblasts, in the investigation of peroxisome biogenesis and the pathogenesis and molecular bases of the primary defects in these disorders. For such a purpose, mammalian somatic cell mutants that phenotypically and genotypically resemble the cells from the patients have been shown to be more useful (Tsukamoto et al., 1991; Shimozawa et al., 1992a). A. Isolation of Peroxisome-Deficient Mammalian Cell Mutants
Two methods were developed for isolation of Chinese hamster ovary (CHO) cell mutants defective in peroxisome biogenesis by Raetz and his colleagues (Zoeller and Raetz, 1986; Morand et al., 1990). The isolated CHO cell mutants are shown to mimic the cellular phenotypes noted in fibroblasts from patients with peroxisome-defective disorders (Zoeller and Raetz, 1986; Zoeller et al., 1989; Tsukamoto et al., 1990; Shimozawa et al., 1992b). Isolation by Autoradiographic Screening
CHO cell mutants defective in dihydroxyacetonephosphate acyltransferase (DHAP-ATase), a key peroxisomal enzyme in the synthesis of ether-linked glycerolipids such as plasmalogens, are isolated by a colony autoradiographic screening procedure (Zoeller and Raetz, 1986). Mutants deficient in DHAP-ATase cannot convert ^^P-labeled DHAP to trichloroacetic acid-insoluble palmitoylDHAP in the presence of palmitoyl-CoA, whereas normal cells do. Thus mutants can be isolated by comparing autoradiograms of cells grown on filter papers and cell-colony staining. In these mutants, catalase, a matrix enzyme of the peroxisome.
Peroxisomal Topogenic Signals and Disease
221
is fully active but not particle-bound. By further morphological and biochemical characterization, the isolated mutants are found to be defective in peroxisomes, as noted in biopsied tissues and in fibroblasts from patients with Zellweger syndrome (Zoeller et al., 1989; Tsukamoto et al., 1990; Shimozawa et al., 1992b). Selection by P90H/UV
Method
A new method for the direct selection of animal cell mutants defective in peroxisome biogenesis was devised by Morand et al. (1990). The procedure is based on the idea that wild-type animal cells such as CHO cells incorporate fluorescent, pyrene-labeled fatty alcohol analogs such as 9-(r-pyrene)nonanol (P90H), a photosensitizer, into lipids and thereby become susceptible to photosensitized killing with oxygen radicals generated by excitation of the pyrene with UV light. In contrast, CHO mutants defective in plasmalogen and/or peroxisomes would not incorporate P90H or would do so very poorly, if at all, thereby rendering them resistant to killing by UV light. By the two methods described above, we have isolated three phenotypically similar but genotypically distinct peroxisome-deficient CHO cell mutants, Z24 and Z65 (by screening), and ZP92 (by the selection method) that resemble fibroblasts from patients with peroxisome-deficient disease, in the defect of peroxisome assembly (Tsukamoto et al., 1990; Shimozawa et al., 1992b). Several CHO mutant cell clones of the same complementation group as Z65 have also been isolated by Raetz's group (Zoeller et al., 1989; Thieringer and Raetz, 1993). Peroxisomal membrane ghosts are seen by immunochemical staining with anti-70 kDa peroxisomal integral membrane protein in all of the CHO mutants, as reported for fibroblasts from patients with peroxisome-deficient disease. Biogenesis of peroxisomal enzymes is also impaired in the mutants, e.g., normal synthesis but a rapid degradation of acyl-CoA oxidase and the unprocessed but rather stable form of thiolase precursor (Tsukamoto et al., 1990; Shimozawa et al., 1992b). Thus, assembly of peroxisomes is defective in the mutants, whereas the synthesis of peroxisomal proteins is normal. However, in the hybrid cells between the mutants, normal assembly of peroxisomes and proper biogenesis of enzymes are evident, as in the wild-type cells, suggesting that these mutants contain different lesions in gene(s) encoding factor(s) required for peroxisome assembly. Peroxisomes are also detectable in fused cells of the wild type with each of the mutants, indicating that the mutations are recessive. A complete set of complementation analyses by cell fusion between three CHO mutants and nine groups of cultured fibroblasts from patients with generalized peroxisomal disorders such as Zellweger syndrome revealed that three CHO mutants, Z24, Z65, and ZP92, represent the human complementation groups E in Japan (the same as group I in the United States and group 2 in Europe), F (the same as group X in the United States and group 5 in Europe), and C (the same as group IV in the United States and group 3 in Europe), respectively (Table 2) (Shimozawa et al., 1992a,b, 1993a). Accordingly, these CHO cell mutants are a relevant animal
222
YUKIO FUJIKI
Table 2. Complementation Analysis between CHO Cell Mutants and Human Fibroblasts from Patients with Peroxisome-deficient Disorders Humanfibroblast(group) Japan"
USA^
A B C E F
VIII IV I X II
in VI
CHO cell mutants
Europe^
Phenotype
Z24
Z65
ZP92
+ + +
+ + + +
+ +
3 2 5 4
ZS,NALD,IRD ZS ZS ZS,NALD,IRD ZS ZS,NALD ZS NALD
+ + + +
+ + +
+ + + + +
Notes: From Shimozawa et al. (1992b, 1993). "From Yajima et al. (1992), Shimozawa et al. (1992a,b). *Roscheretal.(1989). ^Bruletal.(1988). ZS, Zellweger syndrome; NALD, neonatal adrenoleukodystrophy; IRD, infantile Refsum disease. Complementation group V is no longer available. +, complemented; - , not complemented.
cell model for studies on the molecular bases and primary defects of human peroxisome-deficient diseases. B. Cloning of Genes Required for Peroxisome Biogenesis and the Primary Defects of Human Peroxisome-Deficient Disorders
Factors essential for organelle biogenesis have been elucidated, including cisacting topogenic signals such as the peroxisomal COOH-terminal SKL motif. However, only a little is known of membranous components involved in peroxisome assembly. Pathogenic genes responsible for peroxisome-deficient disorders had not been elucidated until our initial report (see below). Successful isolation of animal cell mutants prompted us to search for the genes essential for peroxisome assembly and human pathogenic genes by gene transfection and genetic complementation analysis using the CHO cell mutants. cDNA Cloning of Peroxisome Assembly Factor
We searched for the gene encoding a factor that complements the defects of one of the CHO cell mutants, Z65, by transfection of a rat liver cDNA library con-
Peroxisomal Topogenic Signals and Disease
223
structed in a mammalian expression vector, pcD2 (Tsukamoto et al., 1991). Revertant cells were selected by the P12/UV (12 (r-pyrene)dodecanoic acid/ultraviolet) method (Zoeller et al., 1988), where --90% of the wild-type CHO cells survived but only 0.02% or less of the peroxisome-defective mutant cells such as Z65 were viable, presumably because of a hypersensitivity to the reactive oxygen species formed by irradiation of P12. P12/UV-resistant cells from the transfectants mostly contained peroxisomes, as assessed by immunocytochemical microscopy. The plasmids recovered from these cells exhibited peroxisome-restoring activity when introduced into Z65. An open reading frame in the plasmid encoded a novel 35-kDa protein of 305 amino acids, termed peroxisome assembly factor 1 or PAF-1 (Tsukamoto et al., 1991). Biochemical abnormalities were also complemented: catalase was latent and biogenesis of acyl-CoA oxidase and peroxisomal thiolase was normal. Peroxisomal DHAP-ATase was also restored in its activity. Taken together, the 35-kDa PAF-1 was proved to be sufficient to restore peroxisome formation in the mutant Z65. However, the PAF-1 could not rescue the other CHO mutants, Z24 and ZP92. Subcellular fractionation and immunoblots suggest that the PAF-1 protein is localized in peroxisomes as an integral membrane protein, consistent with the presence of two possible membrane-spanning segments (Fig. 1, underlined). PAF-1 does not contain the SKL motif at its carboxy terminus, implying that PAF-1 is targeted by another type of topogenic signal. Several possible fimctions of PAF-1 in peroxisome biogenesis can be speculated upon, including being a component of a putative signal receptor or of transmembrane import machinery for newly synthesized proteins. Pathogenic Gene for Zellweger Syndrome Disturbances in the assembly of peroxisomes are likely to be the primary defect of peroxisome-deficient disorders comprising nine complementation groups thus far reported, such as Zellweger syndrome and neonatal adrenoleukodystrophy, implying that nine or more genes are required for peroxisome assembly (Shimozawa et al., 1992b, 1993a). To search for the etiology of generalized peroxisomal disorders, we transfected rat PAF-1 cDNA to the fibroblasts from a female patient with Zellweger syndrome of complementation group F that had been found to be in the same group as the CHO mutant Z65 (Shimozawa et al., 1992a). Numerous peroxisomes were evident in the transfected cells. Restoration of peroxisome assembly and peroxisomal protein biogenesis was likewise achieved with a cloned cDNA coding for human PAF-1 (Fig. 1). The human and rat nucleotide sequences of PAF-1 cDNAs are 86% homologous; the deduced amino acid sequences are 88% homologous. Both sequences encode two highly conserved, putative membranespanning segments and seven cysteine residues, and a RING finger motif (Freemont et al., 1991) in the COOH-terminal region. When fibroblasts from patients in the other complementation groups were transfected with human PAF-1 cDNA, no peroxisomes were detected, implying that the other complementation groups represent genes other than PAF-1 responsible for the biogenesis of peroxisomes.
224
YUKIO FUJIKI
60 MASRKENAKSANRVLRISQLDALELNKALEQLVWSQFTQCFHGFKPGLLARFEPEVKACL ^120 WVFLWRFTIYSKNATVGQSVLNIKYKNDFSPNLRYQPPSKNQKIWYAVCTIGGRWLEERC 180 YDLFRNHHLASFGKVKQCVNFVIGLLKLGGLINFLIFLQRGKFATLTERLLGIHSVFCKP 240 QNIREVGFEYMNRELLWHGFAEFLIFLLPLINVQKLKAKLSSWCIPLTGAPNSDNTLATS 300 GKEOALOGEWPTMPHTIGGEHIFOYFQAKSSFLFDVYFTSPK^GTEVHSLQPLKSGIEMS EVNAL Figure 1. Amino acid sequence of human PAF-1. Two membrane-spanning segments are underlined; cysteine and histidine residues in the RING finger motif are highlighted by shading. The amino acid changed by a nonsense mutation in patients with Zellweger syndrome is indicated by an arrowhead.
Dysfunction of PAF-1 in the patient of group F was investigated (Shimozawa et al., 1992a). Northern (RNA) blot analysis showed a 1.9-kb RNA band in the patient's fibroblasts, similar in size and quantity to those detected in fibroblasts from an unaffected person and patients with Zellweger syndrome of other complementation groups. Thus, transcription of PAF-1 in this patient is normal. The nucleotide sequence of PAF-1 cDNA from the patient's fibroblasts was determined. Nucleotide C of a codon CGAfor ^ '^Arg was mutated to T in all of the cDN A clones, resulting in the creation of a termination codon, TGA. No peroxisomes were detected in the patient's fibroblasts in which the PAF-1 cDNA from this patient had been transfected back. Accordingly, we concluded that the point mutation of PAF-1 was homozygous. Next, the nucleotide sequence in the region of PAF-1 containing the mutation site of her parents' genomes was determined by polymerase chain reaction to determine whether the homozygous mutation was derived from the parents. T was detected in place of C in two and three of six genomic DNA clones, from the mother and the father, respectively, indicating that both father and mother were heterozygous for the same mutation in PAF-1 and that the patient inherited the mutation from both parents. Taken together, for the first time we were able to identify the primary defect causing a human peroxisome-deficient disorder, one complementation group of Zellweger syndrome, and detect its carrier. Another Zellweger patient of the same complementation group was recently analyzed and was found to carry the same homozygous point mutation at the same site (Shimozawa et al., 1993a). We also conducted prenatal diagnosis of Zellweger syndrome using amniocytes by polymerase chain reaction for the second child of parents whose first offspring had experienced the first case of PAF-1 dysfunction (Shimozawa et al., 1993b). The mutation in the PAF-1 gene was heterozygous, the same type of mutation as noted for the parents. This child was bom healthy and
Peroxisomal Topogenic Signals and Disease
225
grew up. This is the first prenatal carrier detection of Zellweger syndrome using DNA analysis. Structural Features ofPAFrl PAF-ls from three species, human, rat, and Chinese hamster, are currently available. Chinese hamster PAF-1 comprises 304 amino acids with one amino acid deletion at alignment position 5, whereas human and rat PAF-ls consist of 305 residues (Tsukamoto et al., 1991, 1994; Shimozawa et al., 1992a; Thieringer and Raetz, 1993). Chinese hamster PAF-1 shows homology to those of rat and human— 90% and 86% in the nucleotide sequence, 92% and 90% at the deduced amino acid sequence level, respectively, indicating a high degree of conservation of PAF-1 among at least these three mammalian species. All PAF-ls also contain two membrane-spanning segments, suggesting the importance of these regions for the topology and function of PAF-1, consistent with the results of PAF-1 truncation mutant studies (Tsukamoto et al., 1994). Acysteine-rich region in the COOH-terminal part of PAF-1, at the alignment positions 244 through 305, contains a RING finger motif, C3HC4, with possible involvement in specific protein-protein interaction through binding with zinc or divalent metal ions (Fig. 1) (Lovering et al., 1993). The RING finger motif is also noted in the complementing genes PAS-4 and PAS-5 for peroxisome assembly-defective p^^ mutants of the yeast S. cerevisiae (Hohfeld et al., 1992). To our surprise, however, this riNG finger does not appear to be required for the peroxisome-restoring activity of PAF-1, as deduced from truncation studies. The functional significance of this zinc finger remains to be investigated. Thieringer and Raetz (1993) reported that two types of point mutations of PAF-1, ^^"^Arg to a stop codon and ^"^^Cys in the RING finger to Tyr, were noted in peroxisome-deficient CHO mutants, ZR-82 and ZR-78, respectively, both belonging to the same complementation group as Z65. The former mutation is similar to the incidence in Z65 and is consistent with the results of the PAF-1 deletion mutant study. The latter disagrees with our results where the RING motif appears to be dispensable, although the possibility of instability of PAF-1 caused by this mutation cannot be excluded. The primary sequence of PAF-1, with the deletion of amino acid residues 2 through 19 at the NH2 terminus and residues 214 to 305 in the COOH-terminal part, appears to be essential for peroxisome-restoring activity, as assessed by transfection of the cDNAs for various truncation mutants of PAF-1 to both the CHO mutant Z65 and the fibroblasts from a Zellweger patient (Tsukamoto et al., 1994). Nonsense point mutations in PAF-1 gene have been elucidated, one at * ^^Arg in human PAF-1 (Shimozawa et al., 1992a) and others at ^^"^Arg and ^^"^Trp in CHO mutants (Thieringer and Raetz, 1993; Tsukamoto et al., 1994). The mutation in this region of the PAF-1 gene in human and CHO cells may not be coincidental, but may be more frequent than those in other parts of the sequence. Dysfiinction of PAF-1 caused by these premature terminations is inferred from the truncation studies described above.
226
YUKIO FUJIKI
Isolation of more CHO cell mutants, i.e., those of other complementation groups, and gene transfection followed by genetic complementation analysis using the mutants, including the previously isolated Z24 and ZP92, may lead to identification of pathogenic genes in peroxisome-deficient disorders, as proved to be useful in the case of PAF-1 cDNA cloning. Other Pathogenic Genes of Peroxisome-Deficlent Disorders Gartner et al. (1992) postulated that 70-kDa peroxisomal integral membrane protein is involved in a pathogenic gene of peroxisome-deficient disease. Of 21 patients of one complementation group (group I in the United States), two point mutations (a donor splice site mutation and a missense mutation) in the gene for 70-kDa peroxisomal membrane protein, one each in one allele of two patients with Zellweger syndrome, were noticed. The incidence, only two cases in 21 patients, is apparently a paradox. The functional significance of the 70-kDa membrane protein in peroxisome biogenesis remains to be elucidated.
ACKNOWLEDGMENTS I thank S. Miura, T. Mori, N. Shimozawa, T. Tsukamoto, and all of the other members in the laboratory for their devoted work at the Meiji Institute of Health Science. I am also grateful to T. Hashimoto, T. Orii, T. Osumi, S. Yokota, and their associates for their long-term collaboration. This work was supported in part by grants-in-aid (07408016) for Scientific Research and by grants from Uehara Memorial Foundation, Mitsubishi Foundation, and Nagase Science and Technology Foundation.
REFERENCES Blattner, J., Swinkels, B., Dorsam, H., Prospero, T., Subramani, S., & Clayton, C. (1992). Glycosome assembly in trypanosomes: Variations in the acceptable degeneracy of a COOH-terminal microbody targeting signal. J. Cell Biol. 119, 1129-1136. Borst, P. (1986). Review: How proteins get into microbodies (peroxisomes, glyoxysomes, glycosomes). Biochim. Biophys. Acta 866, 17^203. Brul, S., Westerveld, A., Strijland, A., Wanders, R. J. A., Schram, A. W., Heymans, H. S. A., Schutgens, R. B. H., van den Bosch, H., & Tager, J. M. (1988). Genetic heterogeneity in the cerebrohepatorenal (Zellweger) syndrome and other inherited disorders with a generalized impairment of peroxisomal functions. J. Clin. Invest. 81,1710-1715. de Duve, C. & Baudhuin, P. (1966). Peroxisomes (microbodies and related particles). Physiol. Rev. 46, 323-357. Freemont, P. S., Hanson, I. M., & Trowsdale, J. (1991). Letter: A novel cysteine-rich sequence motif. Cell 64, 483-^84. Fujiki, Y. & Lazarow, P. B. (1985). Post-translational import of fatty acyl-CoA oxidase and catalase into peroxisomes of rat liver in vitro. J. Biol. Chem. 260, 5603—5609. Fujiki, Y., Rachubinski, R. A., Mortensen, R. M., & Lazarow, P. B. (1985). Synthesis of 3-ketoacyl-CoA thiolase of rat liver peroxisomes on free polyribosomes as a larger precursor. Induction of thiolase mRNA activity by clofibrate. Biochem. J. 226, 697-704.
Peroxisomal Topogenic Signals and Disease
227
Furuta, S., Miyazawa, S., & Hashimoto, T. (1982). Biosynthesis of enzymes of peroxisomal p-oxidation. J. Biochem. 92, 319-326. Gartner, J., Moser, H., & Valle, D. (1992). Mutations in the 70K peroxisomal membrane protein gene in Zellweger syndrome. Nature Genet. 1, 16-23. Gietl, C , Faber, K. N., van der Klei, I. J., & Veenhuis, M. (1994). Mutational analysis of the N-terminal topogenic signal of watermelon glyoxysomal malate dehydrogenase using the heterologous host Hansenulapolymorpha. Proc. Natl. Acad. Sci. USA 91, 3151-3155. Goldfischer, S., Moore, C. L., Johnson, A. B., Spiro, A. J., Valsamis, M. P., Wisniewski, H. K., Ritch, R. H., Norton, W. T., Rapin, I., & Gartner, L. M. (1973). Peroxisomal and mitochondrial defects in the cerebro-hepato-renal syndrome. Science 182, 62-64. Gould, S. J., Keller, G.-A., Hosken, N., Wilkinson, J., & Subramani, S. (1989). A conserved tripeptide sorts proteins to peroxisomes. J. Cell Biol. 108, 1657-1664. Gould, S. J., Keller, G.-A., Schneider, M., Howell, S. H., Garrard, L. J., Goodman, J. M., Distel, B., Tabak, H., & Subramani, S. (1990). Peroxisomal protein import is conserved between yeast, plants, insects and mammals. EMBO J. 9, 85-90. Hijikata, M., Wen, J.-K., Osumi. T., & Hashimoto, T. (1990). Rat peroxisomal 3-ketoacyl-CoA thiolase gene. Occurrence of two closely related but differentially regulated genes. J. Biol. Chem. 265, 4600-4606. Hohfeld, J., Mertens, D., Wiebel, F. F, & Kunau, W.-H. (1992). Defining components required for peroxisome assembly in Saccharomyces cerevisiae. In: Membrane Biogenesis and Protein Targeting (Neupert, W. & Lill, R., eds.), pp. 185-207. Elsevier Science Publishers B.V., Amsterdam. Imanaka, T., Small, G. M., & Lazarow, P. B. (1987). Translocation of acyl-CoA oxidase into peroxisomes requires ATP hydrolysis but not a membrane potential. J. Cell Biol. 105, 2915-2922. Kamiryo, T., Sakasegawa, Y, & Tan, H. (1989). Expression and transport of Candida tropicalis peroxisomal acyl-coenzyme A oxidase in the yeast Candida maltosa. Agric. Biol. Chem. 53, 17^^186. Keller, G.-A., Gould, S., DeLuca, M., & Subramani, S. (1987). Firefly luciferase is targeted to peroxisomes in mammalian cells. Proc. Natl. Acad. Sci. USA 84, 3264—3268. Keller, G. A., Scallen, T. J., Clarke, D., Maher, P A., Krisans, S. K., & Singer, S. J. (1989). Subcellular localization of sterol carrier protein-2 in rat hepatocytes: Its primary localization to peroxisomes. J. Cell Biol. 108, 1353-1361. Lazarow, R B. & Fujiki, Y. (1985). Biogenesis of peroxisomes. Annul Rev. Cell Biol. 1,489-530. Lazarow, P. B. & Moser, H. W. (1989). Disorders of peroxisome biogenesis. In: The Metabolic Basis of Inherited Disease (Scriver, C. R., Beaudet, A. I., Sly, W. S., & Valle, D., eds.), 6th edn., pp. 1479-1509. McGraw-Hill, New York. Lovering, R., Hanson, I. M., Borden, K. L. B., Martin, S., O'Reilly, N. J., Evan, G. I., Rahman, D., Pappin, D. J. C, Trowsdale, J., & Freemont, P. S. (1993). Identification and preliminary characterization of a protein motif related to the zincfinger.Proc. Natl. Acad. Sci. USA 90,2112-2116. McCollum, D., Monosov, E., & Subramani, S. (1993). Th^pas^ mutant of Pichia pastoris exhibits the peroxisomal protein import deficiencies of Zellweger syndrome cells—^the PAS8 protein binds to the COOH-terminal tripeptide peroxisomal targeting signal, and is a member of the TPR protein family. J. Cell Biol. 121, 761-774. Miura, S., Mori, M., Takiguchi, M., Tatibana, M., Furuta, S., Miyazawa, S., & Hashimoto, T. (1984). Biosynthesis and intracellular transport of enzymes of peroxisomal P-oxidation. J. Biol. Chem. 259, 6397-6402. Miura, S., Kasuya-Arai, I., Mori, H., Miyazawa, S., Osumi, T., Hashimoto, T., & Fujiki, Y (1992). Carboxyl-terminal consensus Ser-Lys-Leu-related tripeptide of peroxisomal proteins functions in vitro as a minimal peroxisome-targeting signal. J. Biol. Chem. 267, 14405-14411. Miura, S., Miyazawa, S., Osumi, T., Hashimoto, T, & Fujiki, Y (1994). Post-translational import of 3-ketoacyl-CoA thiolase into peroxisomes in vitro. J. Biochem. 115, 1064—1068.
228
YUKIO FUJIKI
Miyazawa, S., Osumi, T., Hashimoto, T., Ohno, K., Miura, S., & Fujiki, Y. (1989). Peroxisome targeting signal of rat liver acyl-coenzyme A oxidase resides at the carboxy terminus. Mol. Cell. Biol. 9, 83-91. Morand, O. H., Allen, L.-A. H., Zoeller, R. A., & Raetz, C. R. H. (1990). A rapid selection for animal cell mutants with defective peroxisomes. Biochim. Biophys. Acta. 1034, 132—141. Mori, T., Tsukamoto, T., Mori, H., Tashiro, Y., & Fujiki, Y. (1991). Molecular cloning and deduced amino acid sequence of nonspecific lipid transfer protein (sterol carrier protein 2) of rat liver: A higher molecular mass (60 kDa) protein contains the primary sequence of nonspecific lipid transfer protein as its C-terminal part. Proc. Natl. Acad. Sci. USA 88, 4338-^342. Osumi, T. & Fujiki, Y. (1990). Topogenesis of peroxisomal proteins. BioEssays 12, 217—222. Osumi, T., Tsukamoto, T., Hata, S., Yokota, S., Miura, S., Fujiki. Y, Hijikata, M., Miyazawa, S., & Hashimoto, T. (1991). Amino-terminal presequence of the precursor of peroxisomal 3-ketoacylCoA thiolase is a cleavable signal peptide for peroxisomal targeting. Biochem. Biophys. Res. Commun. 181,947-954. Pain, D., Kanwar, Y. S., & Blobel, G. (1988). Identification of a receptor for protein import into chloroplasts and its localization to envelope contact zones. Nature 331, 232-237. Roscher, A. A., Hoefler, S., Hoefler, G., Paschke, E., Paltauf, F., Moser, A., & Moser, H. (1989). Genetic and phenotypic heterogeneity in disorders of peroxisomal biogenesis—^A complementation study involving cell lines from 19 patients. Pediatr. Res. 26, 67—72. Santos, M. J., Imanaka, T., Shio, H., Small, G. M., & Lazarow, P. B. (1988). Peroxisomal membrane ghosts in Zellweger syndrome—^abberant organelle assembly. Science 239, 1536-1538. Shimozawa, N., Tsukamoto, T., Suzuki, Y, Orii, T, Shirayoshi, Y, Mori, T., & Fujiki, Y. (1992a). A human gene responsible for Zellweger syndrome that affects peroxisome assembly. Science 255, 1132-1134. Shimozawa, N., Tsukamoto, T, Suzuki, Y, Orii, T., & Fujiki, Y (1992b). Animal cell mutants represent two complementation groups of peroxisome-defective Zellweger syndrome. J. Clin. Invest. 90, 1864-1870. Shimozawa, N., Suzuki, Y, Orii, T, Moser, A., Moser, H. W., & Wanders, R. J. A. (1993a). Standardization of complementation grouping of peroxisome-deficient disorders and the second Zellweger patient with peroxisomal assembly factor-1 (PAF-1) defect. Am. J. Hum. Genet. 52, 843-844. Shimozawa, N., Suzuki, Y, Orii, T., Tsukamoto, T., & Fujiki, Y (1993b). Prenatal diagnosis of Zellweger syndrome using DNA analysis. Prenat. Diagnosis 13, 149. Small, G. M., Szabo, L. J., & Lazarow, P. B. (1988). Acyl-CoA oxidase contains two targeting sequences each of which mediate protein import into peroxisomes. EMBO J. 7, 1167-1173. Sollner, T., Griffiths, G., Pfaller, R., Pfanner, N., & Neupert, W. (1989). MOM 19, an import receptor for mitochondrial precursor proteins. Cell 59, 1061—1070. Sommer, J. M., Cheng, Q.-L., Keller, G.-A., & Wang, C. C. (1992). In vivo import of firefly luciferase into the glycosomes of Trypanosoma brucei and mutational analysis of the C-terminal targeting signal. Mol. Biol. Cell 3, 749-759. Subramani, S. (1993). Protein import into peroxisomes and biogenesis of the organelle. Annul Rev. Cell Biol. 9, 445-^78. Swinkels, B. W., Gould, S. J., Bodnar, A. G., Rachubinski, R. A., & Subramani, S. (1991). A novel, cleavable peroxisomal targeting signal at the amino-terminus of the rat 3-ketoacyl-CoA thiolase. EMBO J. 10,3255-3262. Swinkels, B. W., Gould, S. J., & Subramani, S. (1992). Targeting efficiencies of various permutations of the consensus C-terminal tripeptide peroxisomal targeting signal. FEBS Lett. 305, 133—136. Thieringer, R. & Raetz, C. R. H. (1993). Peroxisome-deficient Chinese hamster ovary cells with point mutations in peroxisome assembly factor-1. J. Biol. Chem. 268, 12631—12636. Tsukamoto, T., Yokota, S., & Fujiki, Y. (1990). Isolation and characterization of Chinese hamster ovary cell mutants defective in assembly of peroxisomes. J. Cell Biol. 110, 651-660.
Peroxisomal Topogenic Signals and Disease
229
Tsukamoto, T., Miura, S., & Fujiki, Y. (1991). Restoration by a 35K membrane protein of peroxisome assembly in a peroxisome-deficient mammalian cell mutant. Nature 350, 77-81. Tsukamoto, T., Hata, S., Yokota, S., Miura, S., Fujiki, Y, Hijikata, M., Miyazawa, S., Hashimoto, T., & Osumi, T. (1994). Characterization of the signal peptide at the amino terminus of the rat peroxisomal 3-ketoacyl-CoA thiolase precursor. J. Biol. Chem. 269, 6001-6010. Tsuneoka, M., Yamamoto, A., Fujiki, Y, & Tashiro, Y (1988). Nonspecific lipid transfer protein (sterol carrier protein-2) is localized in rat liver peroxisomes. J. Biochem. 104, 560-564. van den Bosch, H., Schutgens, R. B. H., Wanders, R. J. A., & Tager, J. M. (1992). Biochemistry of peroxisomes. Annul Rev. Biochem. 61, 157—197. Van der Leij, I., Franse, M. M., Elgersma, Y, Distel, B., & Tabak, H. F. (1993). PAS 10 is a tetratricopeptide-repeat protein that is essential for the import of most matrix proteins into peroxisomes of Saccharomyces cerevisisae. Proc. Natl. Acad. Sci. USA 90, 11782—11786. Vestwever, D., Brunner, J., Baker, A., & Schatz, G. (1989). A 42K outer-membrane protein is a component of the yeast mitochondrial protein import site. Nature 341, 205-209. Wiemer, E. A. C, Brul, S., Just, W. W., van Diel, R., Brouu'er-Kelder, E., van den Berg, M., Weijers, R J., Schutgens, R. B. H., van den Bosch, H., Schram, A., Wanders, R. J. A., & Tager, J. M. (1989). Presence of peroxisomal membrane proteins in liver and fibroblasts from patients with the Zellweger syndrome and related disorders: Evidence for the existence of peroxisomal ghosts. Eur. J. Cell Biol. 50,407-417. Yajima, S., Suzuki, Y, Shimozawa, N., Yamaguchi, S., Orli, T, Fujiki, Y, Osumi, T., Hashimoto, T., & Moser, H. W. (1992). Complementation study of peroxisome-deficient disorders by immunofluorescence staining and characterization of fused cells. Hum. Genet. 88, 491-499. Zoeller, R. A. & Raetz, C. R. H. (1986). Isolation of animal cell mutants deficient in plasmalogen biosynthesis and peroxisome assembly. Proc. Natl. Acad. Sci. USA 83, 5170-5174. Zoeller, R. A., Morand, O. H., & Raetz, C. R. H. (1988). A possible role for plasmalogens in protecting animal cells against photosensitized killing. J. Biol. Chem. 263, 11590-11596. Zoeller, R. A., Allen, L.-A. H., Santos, M. J., Lazarow, R B., Hashimoto, T., Tartakoff, A. M., & Raetz, C. R. H. (1989). Chinese hamster ovary cell mutants defective in peroxisome biogenesis. Comparison to Zellweger syndrome. J. Biol. Chem. 264, 21872-21878.
This Page Intentionally Left Blank
ATP BINDING CASSETTE PROTEINS IN YEAST
Carol Berkower and Susan Michael is
I. Introduction to ABC Proteins 232 A. The ATP-Binding Cassette Superfamily 232 B. Scope of This Review 235 C. Structure and Function of ABC Proteins 236 D. How Do ABC Proteins Work and How Are They Assembled? 239 E. Designation of a New Protein as a Member of the ABC Superfamily . . . 240 II. STE6'. The ProioiypQ ABC ProtQin in Saccharomyces cerevisiae 241 A. Identification ofthe^red Gene 241 B. The Role of STE6 in a-Factor Export and Mating 242 C. Structure-Function Analysis of STE6 245 D. Life Cycle and Intracellular Trafficking of STE6 248 E. A Study of STE6/CFTR Chimeras 250 III. Other ABC Proteins in Saccharomyces cerevisiae, Schizosaccharomyces pombe, and Candida albicans 251 A. Overview 251 B. S. cerevisiae ABC Proteins 252 C. 5./?owZ>^ ABC Proteins 261 D. C. albicans ABC Proteins 263
Membrane Protein Transport Volume 3, pages 231-277. Copyright © 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 1-55938-989-3 231
232
CAROL BERKOWER and SUSAN MICHAELIS
IV. Expressionof Heterologous ABC Proteins in Yeast: Analysis m v/vo 264 A. Utility of Yeast for Expression of Heterologous Proteins 264 B. Mouse MDR3 Expressed in Yeast Complements a ste6 Mutation and Confers Resistance to the Drug FK520 264 C. Plasmodium Falciparum MDRl: ComplQmQntation of a ste6 DQlQtion . .265 D. Human MDRl Does Not Complement a ste6 Deletion but Confers Valinomycin and FK520 Resistance 265 E. Useof a Drug-Sensitive Yeast Mutant, erg(5 265 V. Use of the Yeast sec6 Mutant to Study Heterologous Transporters: Analysis in vitro 266 A. Isolationof Membrane Proteins in Secretory Vesicles 266 B. Mouse MDR2 Is a Lipid Flippase 267 C. Yeast has an Endogenous Bile Acid Transporter 267 VI. Conclusion: Emerging Perspectives in the Study of the ABC Superfamily . . . 268 Acknowledgments 268 References 268
I. INTRODUCTION TO ABC PROTEINS A. The ATP-Binding Cassette Superfamily Members of the ABC Superfamily The ATP binding cassette (ABC) superfamily of transporters, also called the "traffic ATPases," is an ever-growing collection of membrane proteins that mediate transport and channel functions in prokaryotes and eukaryotes (reviewed in Higgins, 1992). ABC proteins share a similar overall structure and significant sequence homology, and most are thought to function by transporting a substrate across a membrane. Within the ABC superfamily in eukaryotes, notable members include: P-glycoproteins, also known as multidrug resistance (MDR) proteins, which are associated with resistance to a wide range of hydrophobic drugs in the case of mammalian MDRl or MDR3 (Germann et al., 1993; Gottesman and Pastan, 1993; Gros et al., 1991; Schinkel et al., 1994), or with phosphatidylcholine transport, in the case of MDR2* (Ruetz and Gros, 1994b; Smit et al., 1993a); CFTR, the cystic fibrosis transmembrane conductance regulator (Collins, 1992; Welsh and Smith, 1993); TAP proteins, the transporters associated with antigen processing in mammalian cells (Androlewicz et al., 1994; Momburg et al., 1994; Parham, 1992); and STE6, which exports the a-factor mating pheromone of Saccharomyces cerevisiae
'TO maintain consistency with the yeast nomenclature, names of multidrug resistance genes are capitalized and italicized. MDRl = human MDRl, unless otherwise noted. MDR3 = mouse MDR3. MDR2 = human or mouse MDR2, as noted in the text. The product of each gene is referred to in the text by the same name as the gene, but is not italicized.
ABC Proteins in Yeast
233
II II « « MSD
MSD
NBD
Function Histidine uptake (Salmonella typhimurium) Hemolysin export (Escherischia coll) Antigen transport (mammals)
NBD
Gene product(s) HisM
HisQ
HlyB (x2)
TAP1
TAP2
Drug efflux/ Ion channel regulation? (mammals)
MDR1
Chloride ion channel/ Ion channel regulation (mammals)
CFTR
STE6
a-Factor export (Saccharomyces cerevisiae) Pignnent Transport (Drosophila melanogaster)
HisP(x2)
white
Qjxrxfx.
brown
©L/XTL/V
Figure 1. A. Model of an ABC protein. A model for the arrangement of the four modules making up an ABC protein is depicted here. The membrane spanning domain (MSD), usually composed of six predicted transmembrane spans, is shown as a cylinder and the nucleotide binding domain (NBD) is shown as a sphere. The transporter may be encoded as a single polypeptide or as two, three, or four polypeptide modules that assemble into the complete structure. The NBDs are shown partially buried in the membrane, but there is evidence that they may actually be accessible from both faces of the membrane (Baichwal et al., 1993). B. Representative ABC proteins. "Reference proteins" discussed in this review are shown, with wavy lines representing MSDs and spheres representing NBDs. MDR, CFTR, and STE6 function as full-length ABC proteins. TAP1 and TAP2 are examples of "half-molecule" ABC proteins, which function as a heterodimer, while HlyB is thought to function as a homodimer. NBD1 and NBD2 may be identical for certain transporters, as in the case of bacterial HisP (Ames, 1993). Note that the NBD precedes the MSD in the D. melanogaster white protein. The R domain of CFTR is represented by a triangle.
234
CAROL BERKOWER and SUSAN MICHAELIS
(Kuchler and Thomer, 1992b; Michaelis, 1993). Prokaryotic ABC superfamily members include: periplasmic nutrient permeases, such as those responsible for uptake of maltose and histidine in Gram-negative bacteria (MalFGK and HisMPQ, respectively); and toxin exporters, such as those required for export of hemolysin and colicin from Escherichia coli (HlyB and ColV, respectively) (Ames et al., 1992; Path and Kolter, 1993). Because ABC proteins play fundamental roles in many aspects of cellular physiology and because at least several members of this family are of significant medical importance, a great deal of attention has been focused on the structure and function of ABC proteins during the past several years. In general, ABC transporters comprise two homologous halves; each half contains a membrane spanning domain (MSD), predicted in most cases to span the membrane six times, and an ATP nucleotide binding domain (NBD) (Higgins, 1992) (Fig. 1). Within the NBD lie two conserved regions associated with ATP utilization, which define an "ATP binding cassette." The highest degree of sequence similarity among ABC proteins is found within their NBDs, in and around the ATP binding cassette. The conserved overall structure of ABC proteins is likely to reflect a common underlying mechanism, in which nucleotide utilization is coupled to the transport of a substrate across a lipid bilayer. ATP binding and substrate-stimulated ATP hydrolysis have been demonstrated for several family members (Bishop et al., 1989; Davidson and Nikaido, 1990; Sharom et al., 1993). In addition, mutational analysis of diverse family members indicates that conserved residues within the NBDs are universally important for function (e.g., Berkower and Michaelis, 1991; Gregory et al., 1991; Hoof et al., 1994; Panagiotidis et al., 1993; Shyamala et al., 1991). ABC Proteins Can Function as Transporters, Ion Channels, or Flippases
Direct biochemical analysis of purified proteins has been important in demonstrating that certain ABC proteins function as transport pumps that use the energy from ATP hydrolysis to drive substrate across a membrane (reviewed in Gottesman and Pastan, 1993; Higgins, 1992). For example, the bacterial histidine permease (HisMPQ) was the first ABC transporter for which active transport was demonstrated in vitro; for HisMPQ, substrate transport was shown to be tightly coupled to ATP hydrolysis (reviewed in Doige and Ames, 1993). More recently, purified MDR has been shown to exhibit drug-stimulated ATP hydrolysis (Ambudkar et al., 1992; Germann et al., 1993; Sarkadi et al., 1992; Sharom et al., 1993). Although it has not yet been purified, the mammalian TAP transporter (TAPl and TAP2) has also been shown to function as an ATP-dependent pump that transports peptides into ER microsomes in vitro. In vivo, the TAP transporter translocates antigenic peptides across the ER membrane, where the peptides are bound by major histocompatibility complex (MHC) class I heavy chain for transport to the cell surface and presentation to T cells (Androlewicz et al., 1994; Momburg et al., 1994; Neefjes et al., 1993; Shepherd et al., 1993). The recent demonstration that TAPl can be coimmunoprecipitated with class I molecules suggests that translocation of pep-
ABC Proteins in Yeast
235
tides into the ER and their subsequent delivery to the class I molecule may be tightly coupled (Androlewicz et al., 1994; Ortmann et al., 1994; Suh et al., 1994). CFTR functions as a chloride ion channel rather than a pump. Since ion channels do not generally require ATP for function, it has been suggested that in CFTR the NBDs are involved in the regulation of channel activity. Indeed, perturbations of the NBDs cause specific changes in the profile of the current produced by CFTR (Carson et al., 1995; Smit et al., 1993b). In addition to its own channel activity, CFTR appears to regulate a separately encoded channel in airway epithelia, an activity for which CFTR also requires ATP (Egan et al., 1992; Gabriel et al., 1993; Jovov et al, 1995; Schwiebert et al., 1994). Interestingly, a recent study suggests that MDRl may possess a second function distinct from its drug efflux activity, either functioning as an ion channel itself or as a regulator of an endogenous ion channel in epithelial cells (Gill etal., 1992; Hardy etal., 1995; Valverdeetal., 1992). The MDR-dependent drug transport and channel activities are distinct, and can be separated by their requirements for ATP hydrolysis and their pharmacology (Gill et al., 1992; Mintenig et al., 1993). In addition to transport and channel functions, yet a third activity, that of a lipid flippase, has been associated with ABC proteins. Mouse MDR2 in liver cells appears to function as a phosphatidylcholine (PC) flippase responsible for moving PC from one side of the hepatocyte apical membrane to the other during bile formation (Ruetz and Gros, 1994b; Smit et al., 1993a). Thus, although they perform the common task of moving substrates across a membrane, ABC transporters may employ diverse methods for coupling the energy from ATP hydrolysis to their particular biological roles. There is another family of transport proteins in bacteria, called the "major facilitator family" (Marger and Saier, 1993), with which the ABC proteins should not be confused. This family includes glucose carriers of prokaryotes and eukaryotes, the E. colt lactose permease, bacterial tetracycline resistance proteins, and several members that confer multidrug resistance. Members of the major facilitator family possess a common structural motif of 12 transmembrane-spanning a-helices. Nonetheless, they do not possess NBDs, do not use ATP, and are not homologous to ABC proteins. B. Scope of This Review
In this review, we will first provide a brief introduction to the field of ABC proteins, followed by a survey of the ABC superfamily members presently known to exist in 5. cerevisiae, the best characterized of which is STE6. We will discuss STE6 in depth, reviewing current knowledge regarding its structure, function, and intracellular trafficking. We will then discuss the other yeast ABC proteins, including several that have only been identified as open reading frames without any analysis of their product. We will also mention the genes known to encode ABC proteins in the fungi Schizosaccharomyces pombe and Candida albicans. We will
236
CAROL BERKOWER and SUSAN MICHAELIS
conclude with a discussion of experiments in which heterologous ABC proteins have been expressed in yeast. These experiments highlight the potential for taking advantage of yeast genetics and biochemistry to learn a great deal about ABC proteins from other systems. S. cerevisiae provides an excellent system in which to study ABC proteins because of the ease of moving between gene and function in this organism. Furthermore, due to concerted efforts world-wide, approximately 30% of the sequence of the yeast genome is presently available, and the entire sequence could be known by the time this article is published. We anticipate that the yeast genome will contain at least 50 ABC proteins, since 17 ABC proteins have been found in S. cerevisiae to date. The native functions are known for only a few of these. Thus, yeast provides a fertile ground, offering many opportunities to study the functions of diverse members of the ABC superfamily. C. Structure and Function of ABC Proteins A Common Overall Structure and Particular Regions of Sequence Homology Are Diagnostic for ABC Proteins
A notable feature of ABC proteins is their modular design. Figure 1 shows a schematic representation of a generic ABC protein and of particular members of the ABC superfamily, demonstrating various arrangements of the modular subunits. The typical ABC transporter contains two membrane spanning domains (MSDl, MSD2) and two nucleotide binding domains (NBDl, NBD2) (reviewed in Ames et al., 1992; Fath and Kolter, 1993; Higgins, 1992). "Full-length" transporters, such as MDR, CFTR, and STE6, contain all four modules in a single polypeptide in the order MSD1-NBD1-MSD2-NBD2, with a region designated the "linker" between NBDl and MSD2 that connects the two homologous halves of the full-length molecule. In several full-length ABC proteins, there is a region immediately following NBDl that contains consensus sites for phosphorylation by protein kinase. In CFTR, this large (241 amino acid) region is called the R (regulatory) domain; it is phosphorylated in vivo, and is important for regulation of the ion channel (Cheng et al., 1991; Rich et al., 1991; Riordan et al., 1989). Several ABC transporters have the NBDs and MSDs in reverse arrangement (NBDl-MSDlNBD2-MSD2), including two recently identified proteins from S, cerevisiae (SNQ2 and PDR5) and one from S. pombe, BFR1 (Balzi et al., 1994; Bissinger and Kuchler, 1994; Nagao et al., 1995; Servos et al., 1993) (discussed further below). In contrast to full-length transporters, certain ABC superfamily members, including the mammalian TAP proteins and many bacterial exporters, consist of two "half-molecules." Each half-molecule encodes a single MSD-NBD unit (Fig. 1); two such units heterodimerize to form the functional transporter (Androlewicz et al., 1994; Fath and Kolter, 1993;Ortmannetal., 1994). Nearly all known half-molecule ABC proteins are encoded in the order MSD-NBD, with the exception of the Drosophila white, brown, and scarlet gene products, which transport precursors of
ABC Proteins in Yeast
237
the fly eye color pigments and have the order NBD-MSD, as indicated in Figure 1 (Dreesen et al., 1988; O'Hare et al., 1984). Genetic evidence suggests that white/brown and white/scarlet heterodimers may transport distinct substrates (Ewart et al., 1994). For proteins such as the bacterial hemolysin transporter HlyB, where only one half-molecule is known, the functional transport unit is presumed to be a homodimer, ahhough direct evidence for homodimerization is lacking. Among the bacterial permease systems, these half-molecules are often subdivided further into "quarter-molecules," so that HisMPQ, for instance, consists of four separate subunits, each of which contains only an MSD or NBD (Ames, 1993). The NBD Domains of ABC Proteins Contain Conserved Sequences: The WallcerA, Wallcer B^ Signature, and Center Regions of the NBD The ATP binding cassette of the NBD: Walker A and B sites. A wide variety of ATP binding proteins, including members of the ABC superfamily as well as non-members, possess the A and B sites identified by Walker et al. (1982) as forming a consensus for nucleotide binding (Higgins, 1992; Higgins et al., 1986). Among ABC proteins, the Walker A site has the sequence GxxGxGKS/T, where x is any amino acid (Fig. 2). This region is also known as the "P loop" ("P" designating phosphate), based on structural and biochemical data that place it in direct contact with the nucleotide triphosphate (Saraste et al., 1990). In particular, the Lys residue of the Walker A site is thought to stabilize the p and y nucleotide phosphates of ATP. The Walker B site has the less stringent consensus "RX(^_g)(I)OOOD," where x is any amino acid and O is any hydrophobic residue. The conserved Asp in the Walker B site is thought to interact with the Mg-H- cation and is present in all ABC proteins (Mimura et al., 1991). Certain ATP-binding proteins contain an extended Walker B site with the consensus Rx.^_g)OOOODEATSALD. Few ABC proteins possess the EATSALD sequence exactly, but most show significant conservation within these seven residues. Mutagenesis of conserved residues in either the Walker A site or the Walker B site of the ATP binding cassette usually impairs function (e.g., Berkower and Michaelis, 1991; Gregory et al., 1991; Hoof et al., 1994; Panagiotidis et al., 1993; Shyamala et al., 1991). Walker A site
Center region
GxxGxGKSn' I
Signature region LSGGQ
90 - 120 a.a.
Walker B site RX(e^)D i
Figure 2. Model of the ATP binding cassette of an ABC protein. Boxes represent regions that are highly conserved among the NBDs of ABC proteins; the consensus sequence for each region is represented below the box, except for the Center region, for which a consensus sequence is lacking. The significance of conserved residues within the ATP binding cassette is discussed in the text. This figure is reprinted from Michaelis and Berkower (1995).
238
CAROL BERKOWER and SUSAN MICHAELIS
While both ABC and non-ABC proteins contain Walker A and B sites, ABC proteins generally show a conserved spacing between the two sites of 90-120 amino acids (Fig. 2). Although there are exceptions, appropriate spacing between the Walker A and B sites is an important diagnostic feature of the ABC superfamily. The ''Signature region'' of the NBD is a diagnostic feature of ABC proteins. As noted above, the presence of Walker A and B sites is not unique to ABC proteins. However, among ABC proteins, there is a highly conserved region immediately upstream of the B site with the consensus "LSGGQ" (Fig. 2) (Ames et al., 1992; Hyde et al., 1990; Mimura et al., 1991). This "Signature region," conserved among ABC proteins, distinguishes them from most other proteins that bind ATP. The Signature region is also referred to in the literature as the C region (Cutting, 1994) or as the "Linker peptide" (Shyamala et al., 1991) (not to be confused with the poorly conserved "linker" region that connects the two halves of full-length ABC proteins). The Center region of the NBD. A fourth region that can be identified in ABC proteins occurs about midway between the Walker A and B sites and is sometimes called the "Center region" (Fig. 2) (Berkower and Michaelis, 1991; Kerem et al., 1989). The Center region is notable because it contains the site of the most common cystic fibrosis mutation, AF508 of CFTR (Kerem et al., 1989). There is no consensus of highly conserved residues in the Center region, but the NBDs of many ABC proteins can be aligned so as to have a Phe or Tyr in a position equivalent to that of F508 of CFTR. Several residues in the Center region are also conserved among particular subsets of ABC proteins (see, for example, Michaelis and Berkower, 1995). Deletion of a residue analogous to F508 of CFTR disrupts the function of some ABC proteins, including yeast YCFl (Szczypka et al., 1994) and human MDRl (Hoof et al., 1994). Nonetheless, other ABC proteins, including yeast STE6 (Berkower and Michaelis, 1991), can tolerate deletions of analogous residues in the Center region without a significant loss of function. The MSDs of ABC Proteins Contain Multiple Membrane Spans with Connecting Loops of Similar Size
A final characteristic of ABC transporters is the possession of two MSDs. In most (but not all) ABC proteins, hydropathy analysis predicts that each of these MSDs contains six membrane spanning segments. Certain groups of functionally related ABC transporters share regions of limited homology within their MSDs. One of these, a short hydrophilic segment with the consensus EAAX3GX9IXLP, located in a cytoplasmic loop between two predicted membrane spans, is conserved among the MSDs of many bacterial periplasmic permeases (Kerppola and Ames, 1992; Saurin et al., 1994). Additional residues in this region are conserved among certain subgroups of bacterial permeases and can be used to classify the individual transporters according to the type of substrate transported (Saurin et al., 1994).
ABC Proteins in Yeast
239
Comparison of the MSDs from a diverse collection of ABC proteins, including CFTR, MDR, STE6, HlyB, and HAM 1 (a mammalian TAP protein), reveals several conserved features of length and sequence (Manavalan et al., 1993). Most notably, the approximate length of each loop connecting a given pair of predicted membrane spans is the same among all these proteins. In addition, there are three conserved sequence motifs present in or near the loops of all five proteins, with the exception of CFTR, which has prompted the suggestion that CFTR may be in a separate category from the other transporters examined. Within the actual transmembrane spans, no regions of strong homology were observed. Topological studies of several ABC proteins have revealed a complex picture that has suggested interesting alternatives to the "6 spans x 2" model. For instance, gene fusion studies of the integral membrane components of the E. coli maltose transporter, malF and malG, indicate eight spans for malF and six spans for malG, totaling 14 spans for this transporter (Boyd et al., 1987; 1993). Several studies have been carried out on human and mouse MDRl (Bibi and Beja, 1994; Skach et al., 1993; Zhang et al., 1993; Zhang and Ling, 1991; 1993), and in no case was the deduced topology exactly as predicted; for instance, one study suggested that the topology of MDRl may be variable, so that it can actually adopt two distinct conformations in the membrane. The topologies of the MSDs of the bacterial oligopeptide and histidine permeases (OppBC and HISMQ, respectively) have also been examined, and are consistent with the existence of six spans each for OppB and OppC but only five spans each for HisM and HisQ (Kerppola and Ames, 1992; Pearce et al., 1992). Two separate groups employed gene fusions to analyze the topology of HlyB and both arrived at a prediction of six to eight membrane spans, although the locations of the deduced spans differed significantly between the two studies (Gentschev and Goebel, 1992; Wang et al., 1991). To date, most topological analyses have been done in "non-native" conditions, that is, in vitro or in foreign cell types or on partial molecules, making these studies less than conclusive; the reliability of topological studies should improve, however, as they are performed under more physiologically appropriate conditions. D. How Do ABC Proteins Work and How Are They Assembled? Current Questions Concerning the Function of ABC Proteins
Much current work on ABC proteins is directed toward addressing fundamental questions about their function. Subjects of particular concern include the roles of ATP binding and hydrolysis, which may serve different purposes even within a single protein such as MDRl, depending on whether it is functioning as a drug pump or a channel regulator (Gill et al., 1992; Hardy et al., 1995). The mechanism of substrate translocation is also of considerable interest. In addition, nontransporter members of the ABC superfamily such as YEF3 (discussed below) may have a completely different method of coupling ATP hydrolysis to biological function. Results from the study of CFTR mutants indicate that many cystic fibrosis muta-
240
CAROL BERKOWER and SUSAN MICHAELIS
tions result in mislocalization of CFTR and have motivated research into the biogenesis and intracellular trafficking of ABC transporters (Berkower et al., 1994; Cheng et al., 1991; Kolling and Hollenberg, 1994a; Loo and Clarke, 1994a; Find et al., 1994; Ward and Kopito, 1994; Welsh and Smith, 1993). Finally, there is continuing interest in the manner in which ABC transporters recognize substrate. No simple "substrate recognition site" has been defined for any of these proteins. For the TAP proteins and for MDR, crosslinking studies with peptide and drug substrates, respectively, indicate that regions in both halves of these molecules are in contact with substrate (Androlewicz et al., 1994; Gottesman and Pastan, 1993). The results of these studies indicate that several regions throughout the transporter come together to form a binding site for substrate. Additional regions may be involved in its translocation. Subcellular Location of ABC Transporters
ABC proteins are localized to the plasma membrane and to the membranes of various intracellular organelles. An interesting observation can be made regarding the intracellular distribution of full-length and half-molecule ABC transporters. Full-length mammalian ABC proteins (including CFTR, MDRl, and MDR2) have been shown to reside on the plasma membrane. In contrast, half-molecule mammalian ABC proteins reside in intracellular organelles. For example, TAPl and TAP2 are on the ER membrane, whereas PMP70 and ALDP are peroxisomal (Kamijo et al., 1990; Kleijmeer et al., 1992; Mosser et al., 1994). Among the yeast ABC proteins whose localization is known, a similar phenomenon is observed: the full-length S. cerevisiae STE6 and PDR5 proteins travel to the cell surface, whereas in the case of half-molecules, PXAl is in peroxisomes, ATMl is in mitochondria, and HMTl of S. pombe is in vacuoles. As the intracellular locations of more eukaryotic ABC proteins are determined, it will be interesting to see if this pattern is maintained, and to establish its significance. E. Designation of a New Protein as a Member of the ABC Superfamily
When an ABC protein such as STE6 is compared to the collection of known sequences in GenBank and other databases by a homology search, one finds strong homology to the MDR proteins, lesser homology to other ABC transporters, and weak homology to non-ABC ATP-binding proteins, such as myosin. It is reasonable to ask what distinguishes ABC superfamily members, and how a newly identified protein may be classified to reside within or outside of the ABC superfamily. Possession of an NBD containing Walker A and B sites is an obvious requirement, and in particular, the Signature region upstream of the Walker B site is often considered to be a strong diagnostic feature of an ABC protein (Fig. 2). Analysis of representative proteins from several ATP-binding families (including the ABC superfamily) has shown that conserved residues around the Walker B site, in addition to the Signature region, may be sufficient to classify new proteins within
ABC Proteins in Yeast
241
these families (Saitoh et al., 1994). The distance between the Walker A and B sites is also diagnostic and ranges in length between 90 and 120 amino acids among the NBDs of most ABC proteins. Association with an MSD is also important, although the modular nature of many ABC proteins may prevent an accurate assessment of this feature for a separately encoded NBD. And, indeed, there is a subset of ABC proteins that do not contain any MSDs (the YEF3 subgroup, see below). If an MSD is available, then it should be examined for features of loop length and sequence that are conserved among ABC proteins (as described in Manavalan et al., 1993). Finally, a multiple sequence analysis that compares the unknown NBD to the NBDs of known ATP-binding proteins is extremely useful in placing a new sequence within a family, as shown in Michaelis and Berkower, 1995 and Saitoh et al., 1994.
II. STE6: THE PROTOTYPE ABC PROTEIN IN SACCHAROMYCES CEREVISIAE A. Identification of the STE6 Gene The Original ste6 Mutant was Isolated Based on its Defect in Mating
The yeast Saccharomyces cerevisiae has two haploid cell types, MATdi and MATa, which can mate to one another. MATa, andMATa cells each secrete a specific pheromone that binds to a receptor on the opposite cell type. When pheromone binds to its receptor, it activates a program of responses that culminate in mating, which is the fusion of a MATa. cell with a MATa cell to form a MATa/a diploid (Sprague and Thomer, 1992). The STE6 gene was identified in a screen for yeast mutants that were unable to mate and hence were designated sterile (Rine, 1979). The mating defect of a ste6 mutant is cell-type specific; only MATa ste6 mutants are sterile. STE6 has been reisolated in subsequent screens for sterile mutants (Oshima and Takano, 1980; Wilson and Herskowitz, 1987). The STE6 and a-factor Genes are Coregulated
The STE6 gene was cloned by complementation (Wilson and Herskowitz, 1984). Analysis of a STE6-lacZ gene fusion revealed that STE6 is expressed only in MATa haploid cells and not in MATa haploids or in a MATa/a diploid. Repression ofSTE6 in MATa and MATa/a cells was shown to be dependent upon the a2 homeobox protein, a transcriptional repressor encoded by the MATa2 gene (reviewed in Johnson, 1992). These studies led to the identification of a negative regulatory site in the 5' untranslated region (UTR) of STE6 that binds a2 (Johnson and Herskowitz, 1985; Wilson and Herskowitz, 1986). The 5' UTR of STE6 also possesses pheromone response elements, or PREs, sites for positive regulation that enable the activation of STE6 expression upon stimulation of MATa cells by a-factor (reviewed in Sprague and Thomer, 1992). Interestingly, the MEil and MFA2 genes, which encode a-factor, are regulated by al and PRE binding factors
242
CAROL BERKOWER and SUSAN MICHAELIS
in a manner that parallels regulation of STE6. These findings provided the first suggestion that STE6 might carry out a function dedicated to a-factor biosynthesis. The fact that STE6 is transcribed only in MATdi haploid cells and not in MATa haploids or MATdi/a diploids was the first indication that STE6 function is not essential for cell viability. Indeed, a MATa. cell with a deletion of the STE6 gene is viable; its only phenotype is the inability to export a-factor and to mate (reviewed in Michaelis, 1993). STE6 is a Member of the ABC Superfamily
Sequencing of STE6 revealed a 3870 bp open reading frame, which encodes a 145 kD protein with homology to MDR and other ABC superfamily members (Kuchler et al., 1989; McGrath and Varshavskiy, 1989). STE6 is a ftill-length ABC protein, with the arrangement MSD1-NBD1-MSD2-NBD2 (Fig. 1). The similarity of STE6 to MDR proteins suggested that STE6 might also function as a transporter. The STE6 gene is located on Chromosome XI next to the UBAI gene, which encodes an ubiquitin-activating enzyme (McGrath and Varshavsky, 1989). STE6 and UBAI are transcribed toward one another so that their mRNAs overlap by -300 nucleotides, with the STE6 RNA polyadenylation site inside the UBAI coding region (McGrath and Varshavsky, 1989). This configuration does not appear to result in coordinate regulation of the two gene products, since the abundance of an mRNA transcript likely to encode UBAI (Band X in Wilson and Herskowitz, 1984) is not altered in cells of different mating types, unlike STE6. However, it is worth noting that STE6 is ubiquitinated, and this modification appears to affect its stability in the cell (discussed further below). B. The Role of STE6 in a-Factor Export and Mating Biogenesis and Structure of the STE6 Substrate^ a-Factor
The co-regulation of STE6 and a-factor and the absence of any ste6 mutant phenotypes other than those related to mating suggested early on that STE6 was a dedicated pump for a-factor. The structure of the STE6 gene lent further support to this notion. Indeed, by metabolic labeling and immunoprecipitation of a-factor, it has been definitively shown that a-factor export is defective in a ste6 mutant (Berkower and Michaelis, 1991; Kuchler et al., 1989), although direct biochemical evidence proving that STE6 is the a-factor transporter is still lacking. Purification of STE6 and in vitro reconstitution of transport will be necessary to determine whether STE6 functions alone or in concert with another protein. The STE6 substrate, a-factor, is encoded by two redundant genes, MFAl and MFA2; either one is sufficient to achieve wild-type levels of mating, but a MATa strain lacking both genes is sterile (Michaelis and Herskowitz, 1988). a-Factor is initially synthesized as a 36 or 38 amino acid precursor. The precursor undergoes extensive posttranslational modification to generate the mature pheromone, which is a 12 amino acid peptide that contains the lipid famesyl (C15) covalently attached
ABC Proteins in Yeast
243
via thioether linkage to its C-terminal Cys and a methyl group esterified to the carboxyl group of the same Cys (Anderegg et al., 1988; Michaelis, 1993). a-Factor lacks a cleaved signal sequence and does not appear to exit the cell via the classical secretory pathway (Chen, 1993; McGrath and Varshavsky, 1989; P. Chen, C. Berkower, and S. Michaelis, unp. obs.) (see below). Although the famesyl modification is likely to be important for recognition of a-factor by STE6, its significance has been difficult to analyze in vivo, because unfamesylated a-factor does not undergo any of the subsequent processing steps (including C-terminal methylation and N-terminal cleavage) that are essential for its export via STE6 (He et al., 1991). Methylation of a-factor is performed by the STE14 gene product. In a stel4 mutant, in which a-factor lacks its methyl group, the N-terminal cleavage of a-factor is normal but its export is blocked (Sapperstein et al., 1994). Thus, the methyl group may play a direct role in mediating the interaction of a-factor with STE6. Alternatively, but less likely, STE14 could serve as an essential component of the transporter, possibly interacting directly with STE6. Assays for STEG-Mediated a-factor Export There are two commonly used methods to detect a-factor export. The first, a biochemical assay, involves immunoprecipitation of the pheromone from the culture fluid of metabolically labeled cells, followed by SDS-PAGE and quantitation by phosphorimager analysis. By immunoprecipitation, mature a-factor is observed in the culture medium of the wild-type strain, but no a-factor can be detected in the culture fluid of the ste6 mutant (Berkower and Michaelis, 1991; Kuchler et al., 1989). Alternatively, the culture fluid from growing cells can be concentrated and tested for biological activity by a halo assay, which relies on the ability of a-factor to arrest the growth of a supersensitive strain of MATa cells (Berkower and Michaelis, 1993; Michaelis and Herskowitz, 1988). The biological halo assay is more sensitive than immunoprecipitation, and can be performed quantitatively down to very low levels of function. The halo assay can also be performed using yeast cell patches growing on petri plates. By this plate assay, a ste6 mutant exhibits a low level of a-factor halo activity, presumably due to the lysis of a small portion of cells on the plate and release of intracellular a-factor (Berkower and Michaelis, 1991). For a more detailed discussion of assays for STE6 function, see Berkower and Michaelis (1993). In an analysis of STE6 mutants altered in conserved regions of the NBDs, we demonstrated a direct correlation between the amount of exported a-factor and the probability of mating (Berkower and Michaelis, 1991). However, it should be pointed out that in wild-type cells, STE6 appears to be present in excess, based on our finding that STE6 levels do not become rate-limiting for a-factor export even when a-factor is highly over-expressed (C. Berkower, P. Chen, and S. Michaelis, unp. obs.). Thus, mutations that produce only a modest effect on STE6 function are not likely to result in a defect in mating. It should also be noted that, while mating occurs at the level of individual yeast cells, it is only possible to quantify a-factor
244
CAROL BERKOWER and SUSAN MICHAELIS
in terms of the amount released from a large population of cells. Therefore, it is most accurate to state that the probability that any one yeast cell carrying a ste6 mutation will mate is directly correlated with the amount of a-factor exported by a population of cells with the same mutation. a-Factor Export is Not Dependent on the Classical Secretory Pathway
Yeast sec mutants can be used to examine the dependence of a-factor export on components of the yeast secretory pathway. We have quantified a-factor exported into the medium after subjecting sec mutants to non-permissive temperature at which their secretion defect is manifest (Chen, 1993; C. Berkower and S. Michaelis, unp. obs.). For a-factor, secretion is completely blocked under these conditions (Julius et al., 1984). In the case of a-factor, we found that mutations that disrupt either intra-Golgi trafficking (sec14) or ftision of vesicles with the plasma membrane (seel, sec6) do not dramatically affect export of newly synthesized a-factor. STE6 is rapidly turned over in the cell (see below), and the modest reduction in a-factor export that we do observe in these three sec mutants might be accounted for by a depletion of STE6 from the cell surface over the course of the experiment. In contrast to the mild effects of these sec mutants, however, in a sec 18 mutant at restrictive temperature, a-factor export is reduced to 10-15% of the wild-type level. SEC 18 ftinctions at various stages in secretion as well as endocytosis; it is therefore not immediately apparent precisely how the secl8 block affects a-factor export, or even whether the effect is a direct one. Characteristics of Mutant Forms of a-Factor as Substrates for STE6
To determine which regions of a-factor are important for STE6-mediated export, our laboratory performed an extensive mutagenesis of the a-factor gene MFAl and recovered mutants that produce the mature, fully processed form of a-factor but are unable to mate (Kistler et al., manuscript in preparation). Approximately 20 different a-factor mutants altered in the 12-amino-acid mature a-factor peptide were recovered. Of these, most were capable of exiting the cell normally, indicating that the defect was likely to affect a step downstream of export, such as the interaction of a-factor with its receptor (STE3) on the surface ofMATa cells. The unhampered export of nonfunctional mutant forms of a-factor by STE6 suggests that those determinants within a-factor that are needed for its recognition by STE6 are not as specific as those needed for binding and stimulation of STE3. Thus, STE6 resembles the multidrug resistance protein MDRl in its ability to transport variant forms of substrate. An Additional Function for STE6 in Mating?
In addition to the export of a-factor, STE6 appears to have a second, poorly understood function in yeast mating. Such a function was suggested by the observation that a-factorless MATo. cells (i.e., cells deleted for the genes encoding a-factor) are unable to mate with MATa cells, even when physiological levels of
ABC Proteins in Yeast
245
exogenous a-factor are provided to the mating mix (Michaelis and Herskowitz, 1988). Thus, MAT?L cells must actively produce and secrete a-factor in order to mate. However, when extremely high (non-physiological) concentrations of purified a-factor are added to a mating mixture consisting of the a-factorless MATa mutant cells and MATa cells, a low level of mating can occur (Marcus et al., 1991). In contrast, if the a-factorless MATa strain is replaced with a MATa strain lacking STE6, then even when an extremely high level of exogenous a-factor is added to the mating mixture, no mating occurs (Marcus et al., 1991). Evidently, STE6 contributes directly to mating, in addition to its primary role in a-factor export, though what role it might play is not understood. Several models are possible; for instance, STE6 could establish a gradient of a-factor that emanates from each cell (or from a particular region on the cell's surface), thereby making a trail of pheromone that could direct the MATa mating partner to its source. Alternatively, STE6 might serve as a cell surface marker that allows a neighboring MATa cell to specifically recognize the MATa cell as being of the opposite mating type; STE6 may even facilitate fusion of the two cells. A recent observation (L. Marsh, personal communication) supports the latter conjecture; in a search for sterile mutants with a specific defect in cell-cell fusion, several mutations were found that mapped to the STE6 gene. Unlike previously studied ste6 mutants, these fusion-defective (and therefore mating-defective) mutants could export a-factor at nearly the wild-type level. Further study of these mutants may uncover a new and interesting function for STE6. Is STE6 a Drug Transporter?
The initial characterization of a ste6 deletion mutant indicated that strains lacking STE6 were hypersensitive to the cytotoxic drug valinomycin (Kuchler et al., 1989). In our laboratory, we have compared a strain expressing high levels of STE6 with a ste6 deletion strain and see no difference in valinomycin toxicity. We have performed this analysis in three different genetic backgrounds, and are still unable to reproduce the reported result (Berkower et al., unpublished observations). In addition, we have tested STE6 for its ability to confer resistance to the MDR drug substrates doxorubicin, daunomycin and ethidium bromide in a yeast mutant that is hypersensitive to these drugs, and we see no effect (Taglicht and Michaelis, unpublished observations). Nonetheless, it is possible that STE6 may be able to transport other, as yet untested, drugs. C. Structure-Function Analysis of STE6 Analysis of STE6 NBD Mutants
The high degree of sequence conservation within the NBDs of ABC proteins suggests that these regions are critical for function. To evaluate the importance of the NBDs of STE6, we generated mutant versions of STE6 bearing single residue alterations in conserved residues in either NBDl or NBD2 (Berkower and
246
CAROL BERKOWER and SUSAN MICHAELIS
Michaelis, 1991; 1993). Several of the mutations we generated were analogous to cystic fibrosis mutations of CFTR. We found that mutations in the Walker A site or Signature region of either NBDl or NBD2 disrupted STE6 function. In no case did a substitution of a single residue completely abolish the ability of STE6 to promote mating. Rather, the mutants exhibited varying levels of residual mating activity, ranging from 0.3% to 26% of the wild-type level. Levels of residual mating correlated with variations in the amount of a-factor exported, as quantitated by halo assay and immunoprecipitation of extracellular a-factor. Unlike the Walker A and B site mutations, mutations in the Center region, including deletions of residues in STE6 analogous to F508 of CFTR, did not disrupt function. The AF508 mutation causes CFTR to be retained in the ER, perhaps because of a defect in folding. The lack of effect on STE6 function of mutations analogous to AF508 suggests that the Center region is not as sensitive to disruption in folding or function within the context of STE6 as it is within the context of CFTR. It is also possible that this region does play a role in STE6 folding and/or function, but since STE6 is not rate-limiting for a-factor export under normal physiological conditions (Chen et al., unpublished observations), a relatively minor alteration would not produce a visible defect. Analysis ofSTEG Nonsense Mutants has Revealed a Surprising Example of ^^Natural'^ Nonsense Suppression
Studies of nonsense mutations in STE6 in a region adjacent to the Signature region of NBDl have led to an examination of the phenomenon of nonsense suppression in yeast (Fearon et al., 1994). It was found that when residue Q511 of STE6 was replaced with an opal, ochre, or amber stop codon, each of these nonsense mutants expressed full-length, functional STE6, albeit at a lower level than wildtype. Further analysis of a lacZ reporter construct containing the Q511 codon and flanking sequences from STE6 showed that the two codons flanking Q511 contain much of the information required to confer nonsense suppression in a completely novel context. Interestingly, two cystic fibrosis mutations generate in-frame termination codons in CFTR, near the Signature region of NBDl (G542X, R553X). It is possible that in human cells these mutations, which have been associated with a relatively mild pulmonary pathology, could also be suppressed at a low level in a manner comparable to the STE6 nonsense mutants described by Fearon et al. (1994). Analysis ofSTE6 Partial Molecules Reveals the Modular Nature ofSTES
To probe the modular nature of STE6, we severed the STE6 gene to obtain two separately encoded half-molecules, MSDl-NBDl (N-half) and MSD2-NBD2 (Chalf) (Berkower and Michaelis, 1991). Structurally, these half-molecules resemble the two subunits of the TAP transporter. Neither STE6 half-molecule expressed individually could promote a-factor export or mating. However, when both halves were co-expressed in the same strain, STE6 function was restored. Thus, as with
ABC Proteins in Yeast
247
the TAP transporter and certain bacterial exporters, such as HlyB, STE6 does not require covalent linkage of its two halves in order to function. Recently, it was shown that CFTR can likewise be encoded as separate subunits that associate in vivo to form a functional channel (Ostedgaard et al., 1994). Similarly, for MDR half-molecules, although each half alone can carry out ATP hydrolysis, the drugstimulated ATPase activity characteristic of full-length MDR requires simultaneous expression of both halves (Loo and Clarke, 1994b). Thus, for STE6, MDR, and CFTR, when separately encoded half-molecules are co-expressed, these halves can reconstitute a functional ABC protein. We have also generated the partial molecules MSDl (N-1/4) andNBDl-MSD2NBD2 (C-3/4). In this case, as for half-molecules, neither partial molecule worked on its own, but in a strain coexpressing N-1/4 and C-3/4, STE6 function was restored (Berkower et al., manuscript submitted). Therefore, the membrane spanning domains of STE6 seem able to mediate their own interaction, in the manner of bacterial permeases (such as HisMPQ and MalFGK) that encode each MSD on a separate polypeptide. We also performed the converse severing experiment, generating MSD1-NBD1-MSD2 (N-3/4) and NBD2 (C-1/4). In this case, STE6 function was not obtained upon co-expression of N-3/4 and C-1/4. A likely explanation for the lack of complementation between these two partial molecules is their inability to associate with one another. Indeed, we have found that N-3/4 associates with yeast membranes, while C-1/4 is in the soluble fraction. This result supports the notion that interaction between modules of STE6 occurs via its MSDs rather than its NBDs. ^'Half-molecule Rescue^^
In a separate experiment, we found that function could be restored to a defective full-length STE6 protein containing a point mutation in NBD2 by co-expressing a half-molecule containing the wild-type NBD2 (Berkower and Michaelis, 1993). We have subsequently tested a total of six point mutants, three in each half of STE6, and all were rescued by co-expression of the appropriate wild-type half-molecule (Berkower, 1995). This result suggests that the region connecting the two halves of STE6 is sufficiently flexible to allow a productive interaction to form between one half of the full-length protein and a lone half-molecule. It also provides an interesting model for in vivo rescue of a mutant ABC transporter, CFTR for instance, by ectopic expression of a half-molecule. Flexibility ofSTEG to N H j - and COOH-Terminal Tags
In order to enable detection of STE6 with monoclonal antibodies for immunoprecipitation. Western analysis, and immunofluorescence, sequences encoding "epitope tag" peptides have been inserted at various locations within the STE6 gene. Epitopes from c-myc (MYC) and influenza hemagglutinin (HA), as well as the "FLAG" tag, have been inserted into three regions in the N-Half of STE6 (at codons 7,70 and 282) and at the extreme C-terminus just preceeding the termination codon,
248
CAROL BERKOWER and SUSAN MICHAELIS
with little or no effect on function (Berkower et al., 1994; Kuchler et al., 1993). A much larger tag, consisting of a 238-residue portion of the green fluorescent protein (GFP) of the jellyfish Aequorea victoria, has also been added to the C-terminus of STE6 without disrupting its function. The resulting STE6-GFP fusion protein can be visualized by direct fluorescence microscopy, circumventing the need to perform indirect immunofluorescence (Tam and Michaelis, unpublished observations). Similarly, MDRl with a large enzyme (adenosine deaminase) fused to its C-terminus can still confer drug resistance on a drug-sensitive cell line (Germann et al., 1990a). Given the results from epitope-tagged STE6 and MDRl, ABC proteins may, as a rule, tolerate additional functions at their C-termini. D. Life Cycle and Intracellular Trafficking of STE6 Life Cycle of STE6 STE6 is a short-lived protein that is internalized from the plasma membrane and transported to the vacuole for degradation (Berkower et al., 1994; Rolling and HoUenberg, 1994a). STE6 has a half-life of about 35 minutes at 30°C, which is comparable to that of the pheromone receptors STE2 and STE3, but is much shorter than the half-lives of most other plasma membrane proteins, which do not appear to undergo internalization. STE6 travels to the plasma membrane via the secretory pathway (ER-Golgi-secretory vesicles). Once at the plasma membrane, STE6 is internalized in a constitutive manner by an endocytic process. In endS, end4, and sac6 mutants (all of which are defective in endocytosis), STE6 is retained on the cell surface, as determined by indirect immunofluorescence and by subcellular fractionation in the case of end4 (Berkower et al., 1994; Rolling and HoUenberg, 1994a; Paddon et al., 1996). Following its internalization, STE6 is transported to the vacuole, where it is degraded in a manner dependent on the vacuolar protease PEP4. In a Apep4 mutant, the half-life of STE6 increases to over 5 hours and undegraded STE6 accumulates in the vacuole (Berkower et al., 1994; Rolling and HoUenberg, 1994a). There was no reason to anticipate that STE6 might undergo constitutive endocytosis and degradation in the vacuole. One explanation for this dynamic life cycle is that yeast cells in the wild switch mating type as often as once every generation. Efficient mating-type switching from MATa. to MATa must also involve clearing the cell surface of mating type-specific molecules residing at the plasma membrane. Thus, the ability of a mating type-specific molecule such as STE6 to be rapidly turned over would facilitate the mating-type switching process. Subcellular Localization ofSTES In mutants defective in endocytosis, STE6 shows a cell-surface immunofluorescence staining pattern (Berkower et al., 1994; Rolling and HoUenberg, 1994a). However, in a wild-type strain the STE6 staining pattern is punctate, suggesting that it is located in dispersed intracellular compartments. (In addition, STE6 does
ABC Proteins in Yeast
249
not co-fractionate with the major plasma membrane protein, the plasma membrane ATPase (Rolling and Hollenberg, 1994a).) These intracellular compartments appear to represent mainly Golgi, based on co-immunofluorescence of STE6 with the late Golgi marker KEX2 (Loayza and Michaelis, manuscript in preparation) and on fractionation experiments (Kolling and Hollenberg, 1994a). These studies suggest that accumulation of STE6 in the Golgi represents a kinetically slow step in its transit to the plasma membrane. It is possible that STE6 may enter endosomal vesicles during part of its life cycle. It is not known whether STE6 is functional inside the cell, or whether it is only competent to mediate a-factor transport once it reaches the plasma membrane. STE6 is phosphorylated in vivo, but the cellular location of phosphorylation and its significance for STE6 function are obscure (Kuchler et al., 1993). Based on observations of a STE6-invertase fusion protein, it was recently reported that STE6 may contain a cleaved N-terminal signal sequence, with the cleavage site following the first predicted membrane span (Kolling and Hollenberg, 1994b). This assertion relied on the deduced size of the STE6-invertase fusion protein and on a predictive algorithm for locating potential signal sequence cleavage sites. However, work from our laboratory, including pulse-chase analysis of STE6 containing an N-terminal or C-terminal epitope tag, indicates that there is no signal sequence cleavage in STE6 (Geller et al., 1996). Furthermore, examination of a large set of STE6-invertase fusions indicates that none of the fusion proteins undergo signal sequence cleavage (Geller et al., 1996). Rather, the hydrophobic N-terminus of STE6 is retained as a membrane anchor, as is the case with most polytopic membrane proteins. Is STE6 Localized to the Shmoo Tip?
When MAT3. cells are exposed to the a-factor pheromone, they arrest growth and distend one region of the cell surface to form a mating projection, or shmoo. Immunofluorescence of STE6 in pheromone-treated cells suggested that STE6 accumulates at the shmoo tip (Kuchler et al., 1993). This may occur by redistribution of STE6 from its intracellular location or from elsewhere on the plasma membrane to the shmoo tip during pheromone stimulation. However, since STE6 is rapidly turned over (even during pheromone stimulation; Berkower, 1995), and trafficking of newly synthesized cell surface proteins is polarized toward the shmoo tip (Baba et al., 1989), it seems likely that accumulation of STE6 on the shmoo tip would be the natural effect of directed deposition of nascent proteins at the shmoo tip combined with endocytosis of older molecules that had diffused furthest away from the shmoo tip. According to this view, most of the STE6 gathered at the shmoo tip would have been synthesized after the initiation of polarization as opposed to older molecules of STE6 redistributing themselves there. Experimental evidence that directly distinguishes between these two possibilities is currently lacking.
250
CAROL BERKOWER and SUSAN MICHAELIS
Ubiquitination ofSTES: A Novel Signal for Vacuolar Degradation?
It was recently shown that STE6 is conjugated to ubiquitin (Kolling and Hollenberg, 1994a). In a ubc4 ubc5 double mutant, which lacks two ubiquitin conjugating enzymes, the half-life of STE6 is increased threefold, suggesting that ubiquitination of STE6 could be important either for internalization of STE6, for its delivery to the vacuole, or for recognition by vacuolar proteases. Any one of these possibilities would represent an as-yet undescribed role for ubiquitination. Alternatively, it is also possible that STE6 stabilization in the ubc4 ubc5 double mutant may be an indirect effect, related to the severely reduced growth rate of this strain. It will be of interest to ascertain the ubiquitination pattern of STE6 in the ubc4 ubc5 strain. Biogenesis and Trafficking ofSTEG: Exit to the Plasma Membrane
Our laboratory has isolated mutant forms of STE6 that are not properly targeted to the cell surface (Loayza et al., unpublished observations). Rather, these mutants are retained in the ER, presumably due to misfolding, and have a reduced ability to promote mating. Using these ER-retained ste6 alleles as a starting point in genetic screens, we expect to uncover elements of the ER machinery that are involved in the retention of misfolded membrane proteins. Ideally, such studies may reveal a process in yeast akin to that by which mutant CFTR molecules are retained in the ER of mammalian cells. E. A Study of STE6/CFTR Chimeras
To generate a yeast model for studying CFTR, a novel approach was used involving STE6/CFTR chimeric proteins (Teem et al., 1993). The aim was to obtain a phenotype in yeast for the F508 mutation of CFTR. (It had previously been shown that deletion of L455 from STE6, analogous to AF508 of CFTR, did not affect STE6 function (Berkower and Michaelis, 1991).) A series of chimeras was generated in which portions of the STE6 NBDl were replaced by corresponding portions of CFTR. One fusion, designated H5, provided STE6 function, albeit not at the full level of wild-type STE6, and exhibited an 80-fold drop in mating frequency when the F508 mutation was introduced into the chimera. (Shorter substitutions, although functional, were not affected by AF508, and longer substitutions were not functional.) Starting with the H5-AF508 chimera, two intragenic suppressor mutations were isolated in which mating was enhanced. These mutations, which occurred in the CFTR portion of the chimera, altered a single residue (R553) within a six-residue stretch immediately downstream of the Signature region. The mutations were introduced into full-length mammalian CFTRAF508 to create the double mutants CFTRAF508/R553M and CFTRF508/R553Q. When the CFTR double mutants were expressed in HeLa cells, a chloride current was observed on the surface of cells expressing the double CFTR mutants that was absent from cells expressing the CFTRAF508 single mutant. Therefore, revertants of a mutant STE6/CFTR chimera obtained genetically in yeast
ABC Proteins in Yeast
251
were effective at altering the phenotype of full-length mutant CFTR in a mammalian cell line. Although the STE6/CFTR chimeras produced intriguing results, their physiological significance has not yet been established. First, the fact that CFTRAF508 is mislocalized and subsequently degraded in mammalian cells does not necessarily mean that the STE6/CFTR AF508 chimera must be mislocalized in yeast. Indeed, we have examined epitope-tagged versions of the wild-type and mutant chimeras and have observed localization indistinguishable from that of wild-type STE6 for both the wild-type chimera STE6/CFTR, and for the mutant chimera STE6/CFTRAF508, thus indicating that the mutant chimera is not ER-retained in yeast (Paddon et al., 1996). Second, intragenic suppression of the mating defect of STE6/CFTRAF508 by the R553Q and R553M mutations, as observed by Teem et al. (1993), might be interpreted as an indication that, in CFTR, the residues F508 and the R553 physically interact. However, there is no evidence that suppression is allele-specific. Thus, the suppressor mutation could result in acquisition of a novel function in CFTR that compensates for the lowering of function caused by AF508. To distinguish between these two possibilities, it would be very interesting to combine the suppressor mutations R553Q and R553M with loss-of-function mutations in other regions of CFTR, aside from AF508. In such an experiment, one might learn whether the suppressors act by creating a "relaxed" channel that is more tolerant to mutations elsewhere in the molecule, or whether they are indeed specific to AF508. As they now stand, the STE6/CFTR chimeras may provide a model for certain aspects of cystic fibrosis in yeast, but they should be interpreted with caution. As a yeast ABC transporter with a well-characterized substrate, a-factor, STE6 represents a valuable prototype within the ABC field. In addition, STE6 offers a unique opportunity for studying all aspects of an ABC protein, including structure, function, regulation, intracellular targeting, and biochemistry.
III. OTHER ABC PROTEINS IN SACCHAROMYCES CEREVISIAE, SCHIZOSACCHAROMYCES POMBE, AND CANDIDA ALBICANS A. Overview
Several approaches have led to the identification of ABC proteins in yeast. Overexpression cloning and mutant hunts have been valuable in identifying genes that confer drug or heavy metal resistance. Molecular strategies have also been useful; these take advantage of the high degree of conservation within the NBDs and utilize PCR primers corresponding to conserved regions to amplify portions of new genes. Finally, the computer offers a supremely sensitive method for identifying ABC superfamily members, and as the yeast genome is sequenced, new genes encoding ABC proteins are frequently revealed. Having a yeast gene in hand, it is
252
CAROL BERKOWER and SUSAN MICHAELIS
possible to determine the localization of its product, generate knockout mutations, and test for a variety of phenotypes, thus enabling researchers to gain a sense of the function of a new gene product with relative ease. For instance, S. cerevisiaeATMl (described below) was identified by PCR amplification and was subsequently shown to encode a protein essential for yeast viability that localized to the mitochondrion; studies of this gene may provide important information regarding the essential nature of mitochondria (Leighton and Schatz, 1995). Using the STE6 protein sequence, we have carried out a survey of the National Center for Biotechnology Information (NCBI) protein sequence data base to identify all ABC proteins presently known (as of April, 1995) (Michaelis and Berkower, 1995). This survey revealed 17 ABC members in S. cerevisiae (including STE6), three in S. pombe, and two in C. albicans. All twenty-two yeast ABC proteins are listed in Fig. 3, where they are represented graphically, along with information regarding each protein's size and the chromosomal location of its gene. Because new proteins are constantly being discovered as the result of the genome project, many more yeast ABC proteins are likely to be identified by the time this article is published. We performed a multiple sequence analysis of these twenty-two known yeast proteins, together with representative ABC proteins from other species. This analysis defined five subgroups based on extended homology within their NBDs (Michaelis and Berkower, 1995). These subgroups are named according to their most prominent yeast or mammalian member: PDR5, ALDP, CFTR, MDR, and YEF3 (Fig. 3). A notable characteristic of certain subgroups is that they appear to contain functionally related proteins. For instance, multiple members of the PDR5 subgroup have been implicated in drug resistance and both members of the ALDP subgroup are involved in peroxisomal function. A summary of the information presently known about each ABC family member in S. cerevisiae, S. pombe, and C albicans is presented below. B. S. cerevisiae ABC Proteins S. cerevisiae
PDRS/STSI/YDRt/LEMI
The phenomenon of multiple, or pleiotropic, drug resistance in yeast has been studied since the 1960s. Many genes have been identified that are implicated in the pleiotropic drug resistance phenotype (Balzi and Goffeau, 1991; Balzi and Gofifeau, 1994). These genes were identified either as overexpressed clones that render yeast resistant to a wide variety of drugs or as genomic mutations that result in drug hypersensitivity. The PDR5 gene, which encodes an ABC protein, was identified in one screen based on its ability, when overexpressed, to confer pleiotropic drug resistance (PDR5) (Leppert et al., 1990), and in three others based on its capacity as a sporidesmin toxicity suppressor (STSl) (Bissinger and Kuchler, 1994), as a determinant of yeast multiple drug resistance (YDRl) (Hirata et al., 1994), and as
ABC Proteins in Yeast
253
a ligand effect modulator (LEMl) that selectively modulates the ability of cells to accumulate glucocorticoids (Kralli et al., 1995). Here we refer to the gene as PDR5. PDR5 encodes a full-length ABC protein with two NBDs and two MSDs (Balzi et al., 1994; Bissinger and Kuchler, 1994; Hirata et al., 1994; Kralli et al., 1995). Unlike STE6, the four modules of PDR5 are arranged in "reverse" order: NBDlMSD1-NBD2-MSD2 (Fig. 3). NBDl of PDR5 is extremely degenerate, with the conserved Lys within the Walker A site (GxxGxGKS/T) replaced by a Cys. This substitution is surprising because the Walker A site Lys is thought to interact directly with ATP; indeed, this Lys is highly sensitive to mutation, so that in nearly all cases examined, alteration of this residue disrupts function (e.g., Azzaria et al., 1989; Berkower and Michaelis, 1991; Gregory et al., 1991; Panagiotidis et al., 1993; Shyamala et al., 1991). Nonetheless, all members of the PDR5 subgroup are distinguished by possessing a Cys instead of Lys in the Walker A site of NBDl (Michaelis and Berkower, 1995). Another unusual feature present exclusively in NBDl of members of the PDR5 subgroup is a strongly conserved sequence comprising 17 residues immediately upstream of the Signature region, with the consensus GLxHTxNTxVGNDxVRG. In addition, in NBD2 of these proteins the Signature sequence is absent and in its place is a different consensus motif (Michaelis and Berkower, 1995; Prasad et al., 1995). Disruption of PDR5 causes hypersensitivity to cycloheximide, chloramphenicol, cerulenin, staurosporine, sporidesmin, and other drugs, while overexpression of PDR5 increases resistance (Bissinger and Kuchler, 1994; Hirata et al., 1994; Leppert et al., 1990). The spectrum of drugs tested in these yeast drug resistance studies is quite distinct from the spectrum of drugs analyzed in studies of mammalian multidrug resistance. The main reason for this difference is that S. cerevisiae is insensitive to most of the well characterized MDR substrates, so that yeast cannot easily be tested for acquisition of resistance to MDR drugs (see below, however, for a discussion of the drug-sensitive erg6 mutant). Aside from PDR5, most of the pleiotropic drug resistance genes that have been isolated to date have turned out to be transcription factors; of these, the best-studied is PDRl. The PDRl gene product regulates the expression of PDR5: a disruption of the PDRl gene reduces the level of PDR5 mRNA (Balzi et al., 1994; Meyers et al., 1992). Conversely, mutations in PDRl that were selected for their ability to confer pleiotropic drug resistance increase the level of PDR5 mRNA (Balzi et al., 1994). Thus, it appears that PD/?7-mediated drug resistance is due, at least in part, to overexpression of PDR5. So far, PDR5 itself has demonstrated no phenotypes other than those associated with drug resistance. A PDRl mutant that causes overexpression of PDR5 has enabled the production of large amounts of PDR5 for purification. PDR5 has been detergent-solubilized from isolated yeast plasma membranes and partially purified by glycerol gradient centrifugation (Decottignies et al., 1994). Partially purified PDR5 hydrolyzes a range of di- and tri-nucleotides in a manner similar to that of MDRl. Several inhibitors of MDRl, including vanadate and oligomycin, were shown to inhibit the
Gene name
Predicted structure
--
Qrcrcn9rcrcrc BFRl SN02
-
NvQ ' -"J-uw
@
Size (a.a.)
Yeast
1511
Chrom.
Reference
Accession no.
Sc
Balzi et al. 1994 B i i n g e r (L Kuchler 1994
PIR A49730
1501
Ca
Prasad et al. 1995
EMBL X77589
1530
sp
-
Turi (L Rose (unp.) Nagao et al. 1995
EMBL X82891
1501
Sc
IV
Servos et al. 1993
EMBL 66732
1411
Sc
IX
Barrell et al. (unp.) [Chr IX sequence]
EMBL Z47047
1049
Sc
111
Purnelle et al. 1991
EMBL X59720
758
Sc
XVI
Shani et al. 1995 Swaltzman et al. (unp.)
GB L38491
853
Sc
XI
Bossier et al. 1994
PIR S34682
1515
Sc
IV
Szypka e! al. 1994
GB L35237 GB U11583
1592
Sc
Vlll
Johnston et al. (unp.) [Chr Vlll sequence]
1218
Sc
XI
Dujon et al. 1994
EMBL 228328
306
Sc
XI
Dupn et al. 1994
EMBL 228329
NCBl ID
MDR
I
STE6
1290
Sc
XI
McGrath (L Varshavsky 1989 Kuchler et al. 1989
GB M26376
134967
PMDl
1362
SP
-
Nishi et al. 1992
SP P36619
548528
830
SP
-
Ortiz et al. 1992
EMBL 214055
399917
694
Sc
Xlll
Leighton (L Schatz 1995
EMBL X82612
575392
696
Sc
XI1
Dean et al. 1994
GB L16958
462581
812
Sc
XVI
Dean et al. 1994
GB L16959
462582
1044
Sc
XI1
Qin et al. 1990
GB J05197
119180
1049
Ca
Myers et at. 1992
SP P25997
119179
752
Sc
VI
Vazquez de Aldana et al. 1995
-
643479
610
Sc
V
Dietrich et al. (unp.) [Chr V sequence]
GB U18796
603269
HMT1 ATMl
L
MDL2
Jv-@
JJ'n' @
-N'4a
Di Domenico et at. 1992
Figure 3. ABC proteins in yeast. All sequences either known or predicted to encode ABC proteins from S. cerevisiae (Sc), S. pombe (Sp) or C. albicans (Ca) as of April, 1995 are represented in this table. NBDs are drawn as spheres and MSDs by wavy lines; note that these are not meant to imply the existence of precisely six membrane spans per MSD. The highly degenerate NBD1 of YIL013c i s represented by a solid circle. The two EGF3 repeats predicted to occur in the first extracellular loop of ADPl are represented by a rectangle. The R domain of YCFl is shown as a triangle. A dash indicates that the chromosomal location of a particular gene has not been reported. The complete sequence of each gene may be obtained via its accession number in one or more databases. We have provided only one accession number for each sequence, though some sequences are represented in multiple databases and have alternative accession numbers. All of these sequences can be accessed by their National Center for Biotechnology Information (NCBI) ID number, which is also provided. This figure is reprinted from Michaelis and Berkower (1995).
256
CAROL BERKOWER and SUSAN MICHAELIS
NTPase activity of PDR5, while sodium azide, which does not affect MDR1, is also ineffective against PDR5. It has been suggested that properties shared by MDR and PDR5 might reflect an underlying similarity in the way the two proteins utiHze ATP for drug transport. S. cerevisiae YCF1
The YCFl (yeast cadmium factor) gene was identified on the basis of its ability to confer resistance to high levels of cadmium when overexpressed (Szczypka et al., 1994). The YCFl gene encodes a full-length ABC protein that closely resembles human CFTR and human MRPl, a multidrug resistance protein distinct from MDRl (Cole et al., 1992; Grant et al., 1994). YCFl, MRPl, and CFTR are distinguished from other ABC proteins by the possession of an R domain that contains consensus sites for Ser phosphorylation. Mutagenesis of a single Ser in the R domain of YCFl renders it nonfunctional for cadmium resistance, suggesting that phosphorylation of the R domain may play a role in YCFl function, just as it does in the case of CFTR. YCF1 also has a Phe residue (F713) in the Center Region of NBDl analogous to F508 of CFTR, and deletion of F713 in YCFl likewise destroys its function. Thus, YCFl may be a useful model for certain aspects of CFTR function in yeast. The YCFl gene product is not essential for viability, but yeast lacking YCFl are hypersensitive to cadmium. While YCFl localization has not yet been determined, it is possible that YCFl confers resistance to cadmium by a mechanism similar to that proposed for S. pombe HMTl (discussed below), that is, by sequestering metal in the vacuole and thereby protecting cells from its toxicity. S. cerevisiae SNQ2
The SNQ2 gene was identified on the basis of its ability to confer resistance to the mutagen 4-nitroquinoline-N-oxide (4-NQO) when overexpressed (Servos et al., 1993). High-copy SNQ2 also confers resistance to the mutagen Trenimon and to the chemicals sulphomethuron methyl and phenanthroline. SNQ2 encodes a fulllength ABC protein with the "reverse" arrangement NBDI-MSDI-NBD2-MSD2 and is most similar to the D. melanogaster half-molecules, white and brown. Based on homology in its NBDs, SNQ2 falls into the PDR5 subgroup of ABC proteins (Fig. 3). Recently SNQ2 has been purified and studies on its nucleoside triphosphatase activity and regulation are underway; like PDR5, SNQ2 appears to be regulated by PDRl (S. Moye-Rowley, personal communication). SNQ2 was independently identified as a clone designated BAD6, based on its ability to confer resistance to the microtubule-destabilizing drug benomyl when overexpressed (A. Murray, unp. obs.). The BAD6 gene was mapped to the right arm of chromosome IV (Kistler et al., unpublished observations).
ABC Proteins in Yeast
257
S. cerevisiae PXA1/SSH2/PAL1
The human PMP70 and ALD genes, which are associated with Zellweger's syndrome and adrenoleukodystrophy, respectively, encode peroxisomal membrane ABC proteins (Kamijo et al., 1990; Mosser et al., 1994). Homologues of these proteins were sought using degenerate PCR primers corresponding to the conserved Walker A and B sites of ABC proteins to amplify cDNA from yeast cells that had been induced to form peroxisomes (Shani et al., 1995). The most prominent product obtained was named PXAl (peroxisomal ABC transporter), based on its localization to peroxisomes and its homology to ABC proteins. Yeast PXAl is most similar to human ALDP (the product of the ALD gene), with 28% identity overall and 47% identity in a 178 amino acid region surrounding the Walker A and B sites. Like ALD and PMP70, PXA1 encodes a half-molecule transporter with the order MSD-NBD. PXAl is identical to SSH2 and PALI, mentioned below. Like ALDP, PXA 1 is localized to peroxisomes. Metabolism of oleic acid requires p-oxidation, which occurs in the peroxisome; yeast lacking PXAl grow poorly on oleic acid. In addition, peroxisomes derived from pxal null yeast are defective for P-oxidation in vitro (Shani et al., 1995). Similarly, ALD patients show reduced P-oxidation of very long chain fatty acids (Valle and Gartner, 1993), suggesting that human ALDP and yeast PXAl may carry out similar functions. The precise functions of ALDP and PXAl are not known. Both proteins contain an EAA motif, a sequence located between the fourth and fifth putative transmembrane spans of prokaryotic transporters. In addition, the N-termini of both proteins contain a region of homology to fatty acid binding proteins, which is consistent with the possibility that they play a direct role in the transport of very long chain fatty acids into peroxisomes. Interestingly, this same region of homology can also be found in MDRl, which transports lipophilic drugs; this common motif may indicate that all three proteins possess similar sites for recognition of hydrophobic substrates (Shani et al., 1995). 5. cerevisiae YKL741/PXA2
YKL741 was identified as an open reading frame during the sequencing of the left arm of chromosome XI (Bossier et al., 1994; Wiemann et al., 1993). It encodes a half-molecule transporter with strong homology to human ALDP and PMP70. All three gene products share homologous regions extending beyond the conserved nucleotide binding domains of ABC proteins in general. In an 89 amino acid stretch, YKL741 displays 48% identity with PMP70 and ALDP. In combination with PXAl, these proteins form a subgroup of the ABC superfamily (Fig. 3). It has been suggested that YKL 741 be renamed PXAl if it is shown to localize to the peroxisome (N. Shani, personal communication). It has been postulated that PXAl and YKL741 could heterodimerize (Shani et al., 1995). Certain half-molecule ABC transporters, such as TAPl and TAP2, heterodimerize to form a fiinctional transporter; as a result, loss of TAPl, TAP2, or
258
CAROL BERKOWER and SUSAN MICHAELIS
both have identical phenotypic consequences (Kelly et al., 1992; Neefjes et al., 1993; Shepherd et al., 1993; Spies et al., 1992). Likewise, yeast carrying a disruption of the YKL741 gene have the same growth defect on oleic acid as do yeast disrupted for PXAJ, and the double mutant {ykl741 pxal) has the same phenotype as the single mutants (Shani et al., 1995). The identical and non-additive phenotypes of single and double mutants suggests that the products of the PXAl and YKL741 genes may contribute to the same function and is consistent with the notion that they are subunits of a single transporter. A similar relationship has been postulated for PMP70 and ALDP in humans (Valle and Gartner, 1993). 5. cerevisiae ATM1
To identify ABC proteins involved in mitochondrial function, Schatz and coworkers amplified DNA fragments from yeast by PCR, using degenerate oligonucleotide primers corresponding to the Walker A and B sites (Leighton and Schatz, 1995). Often DNA fragments they sequenced, three corresponded to known genes and the other seven were novel, indicating that there are numerous ABC proteins still unidentified in yeast. Five putative ABC genes were disrupted and one of these was found to be essential. This gene product was localized to mitochondria by subcellular fractionation and immunofluorescence, and was called ATM I for ABC transporter of mitochondria. ATM I is a "half-size" ABC protein organized as MSD-NBD. It contains an N-terminal signal sequence sufficient for targeting a cytoplasmic protein to the mitochondrion. ATMl is the only known mitochondrial ABC protein. Based on its structure, localization, and essential nature, ATM 1 could function as a pump that transports substrates into or out of mitochondria. Mitochondria from an atml mutant lack all major cytochromes and thus produce colonies that are white. The atml mutant also experiences rapid loss of mitochondrial DNA. It has been suggested that ATM I may function as one half of an export pump that prevents the buildup of toxic substances in the mitochondrial matrix, or that releases mitochondrially derived peptides into the cytosol for communication to the nucleus (Leighton and Schatz, 1995). Alternatively, ATMl may import an essential compound into mitochondria. S. cerevisiae MDL1, MDL2/SSH1 and SSH2/PAU
To identify new genes encoding ABC proteins in S. cerevisiae, two groups amplified pieces of chromosomal DNA, using degenerate oligonucleotide primers based on conserved regions in the NBD portions of ABC proteins (Dean et al., 1994; Kuchler et al., 1992). Each group found two genes, which they called MDLl and MDL2 for multidrug resistance-like, and SSHl and SSH2 for sterile six homologue. MDL2 and SSHl are the same gene. All three gene products (MDLI, MDL2/SSH1 and SSH2) are encoded as half-molecules with the order MSD-NBD. MDL2/SSH1 has extensive homology with the MDR class of ABC proteins in the Walker A, Walker B and Signature sites of its NBD. SSH2 is the same gene as PXAl, described
ABC Proteins in Yeast
259
above, and was recently renamed PALI (Swartzman et al., 1996). None of the three genes is essential for viability, and an mdll mdl2 double knockout is also viable (Dean et al., 1994; Kuchler et al., 1992; M. Dean and S. Michaelis, unp. obs.). S. cerevisiae ADP1
ADPl (ATP-dependent permease) was identified during the sequencing of chromosome III (Goffeau et al, 1990; Goffeau et al., 1993; Oliver and others, 1992; Pumelle et al., 1991). The proposed structure of ADPl is unusual for an ABC protein, since there are nine predicted membrane spans. Despite its size, ADPl has only one ATP binding motif, and it occurs in an unusual position in the cytoplasmic domain between the second and third transmembrane spans. ADPl has strong homology to the D. melanogaster white protein in a region extending beyond its Walker A and B sites. Unlike any other known ABC protein, ADPl has two epidermal growth factor (EGF) motifs composed of cysteine-rich repeats, both occurring in a large (300 a.a.) extracellular domain between the first and second transmembrane spans. This is the first time such a motif has been reported in yeast, and it suggests a possible function for ADPl in cell adhesion or differentiation. Sequence analysis of the NBD of ADPl places it in the PDR5 family of ABC proteins (Fig. 3). Analysis of the 5' untranslated region of ADPl reveals two consensus sites for binding of the transcription factor ABFl/GFl (Buchman et al., 1988; Dorsman et al., 1991), and an RPG box, which binds the regulatory protein TUF (Huet and Sentenac, 1987). Disruption of the.4DP7 gene indicates that it is not essential for viability. Work is continuing to determine the function of its intriguing product. S. cerevisiae YEF3
The YEF3 gene encodes a yeast translation elongation factor that is essential for viability and necessary for protein synthesis by yeast ribosomes in vitro (Skogerson, 1979). YEF3 has two predicted nucleotide binding domains, both of which contain the Walker A and B sites and the Signature region (LSGGQ) (Qin et al., 1990). The spacing between the Walker A and B sites of NBD 1 is 85 amino acids, in agreement with the results from other ABC proteins. In contrast, the Walker A and B sites of NBD2 are separated by 200 residues, which is unusually large for ABC proteins. Despite the fact that the NBDs of YEF3 (particularly NBDl) bear strong resemblance to the ABC superfamily of transporters, YEF3 is a soluble protein and it does not appear to associate with an MSD. Therefore, YEF3 is unlikely to function as a transmembrane transporter. Nonetheless, because the NBDs of YEF3 are closely related to those of other ABC proteins, we have chosen to include YEF3 and its relatives (GCN20, YER036p, and CEF3) in this survey. The function of YEF3 may be to transduce the energy from ATP hydrolysis into mechanical energy for translocating cellular components engaged in the elongation step of protein synthesis (Qin et al., 1990). It has also been suggested that YEF3 stimulates translation elongation by facilitating the release of uncharged tRN A from
260
CAROL BERKOWER and SUSAN MICHAELIS
the ribosomal E site, dependent on NTP hydrolysis (Triana et al., 1994). In this respect, YEF3 may provide a novel type of ABC activity, in which it shares with ABC transporters the capacity to translocate a substrate from one position to another, thus not entirely diverging from the transporter model, in addition, by using ATP as its energy source (rather than GTP, which is the favored energy source for protein synthesis in higher eukaryotes), YEF3 could serve as a sensor of the cell's ATP level and thus regulate the rate of protein elongation. The 5' untranslated region of YEF3 contains three sites characteristic of genes that encode ribosomal proteins, suggesting that YEF3 is co-regulated with other genes encoding components of the translational apparatus in yeast (Qin et al., 1990). Two of these regulatory regions are recognized by the TUF DNA-binding protein, which is also predicted to regulate transcription ofADPI (above). The YEF3 protein has a consensus site for phosphorylation by cAMP-dependent protein kinase near its N-terminus. The C-terminus of YEF3 is very hydrophilic and has three clusters of Lys residues, suggesting interaction with RNA. Though elongation factor 3 is apparently ubiquitous in fungi, it has never been detected in mammals. The essential nature of YEF3 in S. cerevisiae makes this protein a promising target for antifungal drug development. 5. cerevisiae GCN20
The GCN20 gene product, identified in a genetic screen, plays a role in a yeast regulatory pathway known as general amino acid control, in which the transcription of a variety of amino acid biosynthetic enzymes is increased upon amino acid starvation (reviewed in Hinnebusch, 1992). In this pathway, a protein kinase (GCN2) phosphorylates the yeast translation initiation factor 2 (eIF-2), which in turn stimulates translation of the transcription factor (GCN4) that governs general amino acid control. GCN20 is an ABC protein that interacts with a partner, GCN1. GCN20 and GCNl are required for GCN2-mediated phosphorylation of eIF-2. They are thereby thought to couple the activity of GCN2 to the availability of amino acids (Vazquez de Aldana et al., 1995). GCN20 has two NBDs but no predicted membrane spans (Vazquez de Aldana et al., 1995). GCN20 is most closely related to the predicted product of YER036p, an open reading frame recently discovered during the sequencing of the yeast genome. GCN20 and YER036p fall into the yeast elongation factor (YEF3) subgroup of ABC proteins (Fig. 3). As is the case for YEF3 (discussed above), none of the members of this subgroup contain membrane spans. GCN20 is shorter than YEF3 by about 300 residues at its N-terminus (Vazquez de Aldana et al., 1995). Interestingly, a segment of approximately 200 residues in YEF3 that is not generally conserved among ABC proteins (and not present in GCN20) is highly homologous to a segment of GCNl. It has been suggested that GCNl and GCN20 work together, in a manner analogous to that of the single protein YEF3, to stimulate translation of GCN4 (Vazquez de Aldana et al., 1995). An alternative mechanism that has been proposed is that GCN20 could interact with
ABC Proteins in Yeast
261
a transmembrane protein, for instance a vacuolar membrane protein, to carry out an amino acid transport function. Whatever its function, GCN20, along with YEF3, seems likely to offer a new twist on the use of the ATP binding cassette to couple ATP hydrolysis to a highly regulated biological event. Putative ABC Proteins: New Genes from ttie Saccharomyces cerevisiae Genome Sequencing Project
As the yeast genome sequencing project progresses, new open reading frames (ORFs) encoding putative ABC proteins periodically appear in the on-line DNA sequence databases. We encountered five such ORFs in a search of the NCBI database for sequences bearing homology to STE6. The predicted products of these ORFs are included in the list of yeast ABC proteins in Fig. 3. They are YER036p, YHL035C, YILOJSc, YKRlOSw, and YKR104w. Some of these ORFs have been assigned several names in the past, but all can be accessed from NCBI by the accession numbers provided in Fig. 3. The functions of these ORFs are obscure. YER036p is a member of the YEF3 subgroup and is the closest relative of GCN20. By our multiple sequence analysis, YHL035C, YKRI03W, and YKR104w fall in the CFTR subgroup of ABC proteins (Fig. 3). The N-terminal NBDs of CFTR, YCFl, and YHL035c, which are fulllength transporters, all cluster together in this analysis. Their C-terminal NBDs form a separate cluster (Michaelis and Berkower, 1995). During the examination of these ORFs, we noted an interesting feature for two of them, YKR103w and YKR104w. These ORFs encode unique structures for eukaryotic ABC proteins, with YKR103w organized as a "3/4-molecule" with the predicted structure MSD1-NBD1-MSD2 and YKR104w as a "1/4-molecule" consisting of just a single NBD with no MSD. It is possible that these two gene products are unrelated in function. Alternatively, they may co-assemble to form a functional unit. It is also conceivable that, because these two ORFs are located directly next to one another on chromosome XI, they are pieces of a single gene, artifactually separated by a sequencing error. The possibility that these gene products form a single functional unit is supported by our multiple sequence analysis, which groups the NBD of YKR103w with the N-terminal NBDs of CFTR-related proteins, and that of YKR104W with the C-terminal NBDs of the same proteins (Michaelis and Berkower, 1995). If YKR104w is indeed a separate protein from YKR103w, it would be the first example of a eukaryotic ABC transporter in which a single subunit encodes a solo NBD, without any MSD. C. 5. pombe ABC Proteins 5. pombe HMT1
In order to identify the genes involved in heavy metal tolerance in S. pombe, a screen was carried out for mutants that were hypersensitive to cadmium (Ortiz et al., 1992). One gene, cloned by its ability to rescue the mutant phenotype, was named HMTl for
262
CAROL BERKOWER and SUSAN MICHAELIS
heavy metal tolerance. The HMTl gene encodes a "half-molecule" ABC protein with the structure MSD-NBD (Fig. 3). The MSD is predicted by hydropathy analysis to span the membrane from six to ten times. S. pombe cells that display wild-type levels of cadmium resistance form a complex of cadmium bound to phytochelatins, which are peptides synthesized from glutathione. Overexpression of the HMTl gene results in higher levels of intracellular cadmium while enhancing the cadmium tolerance of cells, leading to the suggestion that HMTl might be involved in pumping cadmium or cadmium-phytochelatin complexes into an intracellular organelle, rather than secreting it (Ortiz et al., 1992). Indeed, it has recently been shown that HMTl is located in the vacuolar membrane and that it exhibits ATP-dependent phytochelatin transport in vitro (Ortiz et al., 1995). Therefore, HMTl appears to function by transporting complexed metal into the S. pombe vacuole, where it is prevented from damaging the cell, either by sequestration or by vacuole-mediated inactivation. S. pombe PMD1
The PMDI gene ("5. pombe mdrl-like") was obtained by screening a high-copy genomic library for clones that increased the resistance of 5. pombe to leptomycin B, an antifungal drug (Nishi et al., 1992). PMDI encodes aftill-lengthABC protein with the structure MSD1-NBD1-MSD2-NBD2. PMDI displays a high degree of similarity to mouse MDRl, with extensive regions of identity in the NBDs (Fig. 3). Overexpression of PMDI increases resistance to a variety of MDR substrates, including cycloheximide and valinomycin. However, PMDI does not affect the sensitivity of cells to brefeldin A, distinguishing it from S. pombe BFRl (below). PMDI is not essential, but a null mutant is hypersensitive to several cytotoxic drugs, including leptomycin B, cycloheximide, valinomycin, and actinomycin D. PMDI could be an S. pombe counterpart of mammalian MDRl or of 5. cerevisiae PDR5. S. pombe BFRl
BFRl, which was cloned on the basis of its ability to confer brefeldin A resistance when overexpressed (Nagao et al., 1995), encodes a full-length ABC protein whose modules are arranged in the "reverse" order (NBD1 -MSD 1-NBD2-MSD2), as was also observed for S. cerevisiae SNQ2 and PDR5 (Fig. 3). Not surprisingly, BFRl shows greatest homology to these proteins and to the white protein of Drosophila, a half-molecule ABC protein whose NBD is N-terminal to its MSD. BFRl is not essential, but a strain carrying a chromosomal deletion of BFRl is hypersensitive to brefeldin A. Cells carrying BFRl on a high-copy plasmid exhibit increased resistance to other drugs, including actinomycin D, cerulenin and cytochalasin B, while sensitivity to oligomycin and cycloheximide is unaffected. Based on the distinct drug resistance profiles of PMDI and BFRl, both proteins may have MDR-like activities in S. pombe that differ with respect to substrate and function. The
ABC Proteins in Yeast
263
sequence of 5F/?7 is in the GenBank database under the name HBA2 (Turi and Rose, 1994). The S. pombe M-factor Transporter
S. pombe has two haploid mating types, designated Plus (P) and Minus (M) (reviewed in Nielsen, 1993). Cells of the M mating type produce a pheromone, M-factor, that resembles a-factor in its size, lipid modification and C-terminal methylation (Davey, 1992). A gene involved in the processing of M-factor has been cloned; this gene, MAM4, bears strong homology to S. cerevisiae STEM which is required for methylation and export of a-factor (Sapperstein et al., 1994; M. Yamamoto, personal communication). Based on the similar structure and biogenesis of a-factor and M-factor, it has seemed reasonable to expect that M-factor exits the cell via a STE6-like transporter. Indeed, it was recently demonstrated that an S. pombe mam I mutant is defective in the secretion of M-factor and that MAMl encodes an ABC protein (J. Davey, personal communication). Comparison between S. cerevisiae STE6 and S. pombe MAMl could facilitate the analysis of substrate specificity determinants in these transporters. D. C. albicans ABC Proteins C albicans CEF3
The gene encoding C. albicans elongation factor 3 (CEF3) was cloned by probing a C. albicans DNA library with a region of the S. cerevisiae YEF3 gene (Colthurst et al., 1992; Di Domenico et al., 1992; Myers et al., 1992). CEF3 bears a strong resemblance to YEF3 in its structure and sequence, with 78% identity overall. The two proteins also share functional homology, as evidenced by the ability of the C. albicans CEF3 gene to rescue the lethal phenotype of a YEF3 gene knockout in S. cerevisiae (Myers et al., 1992). C albicans CDR1
The CDRI (Candida drug resistance) gene was isolated based on its ability to complement an S. cerevisiae PDR5 disruptant, which is hypersensitive to cycloheximide and chloramphenicol (Prasad et al., 1995). Expression of CDRI increased resistance to cycloheximide, chloramphenicol and the antifungal drug miconazole, while at the same time making yeast hypersensitive to oligomycin, nystatin and the protonophore DNP. The CDRI gene product is a full-length ABC protein with the "reverse" arrangement NBD1-MSD1-NBD2-MSD2 and falls in the PDR5 subgroup of ABC proteins (Fig. 3). CDRI displays strong conservation with S. cerevisiae PDR5 and SNQ2 along its entire length. Interestingly, a tandem repeat of heat-shock consensus elements is found in the promoters of C albicans CDRI, S. cerevisiae PDR5 and mammalian MDR genes, suggesting analogous regulation of these genes (Balzi and Goffeau, 1994; Prasad et al., 1995).
264
CAROL BERKOWER and SUSAN MICHAELIS
In their screen for C. albicans suppressors of the S. cerevisiae pdr5 mutation, Prasad et al. (1995) uncovered several genes in addition to CDRl that increased drug resistance. Analysis of CDRl and these genes may provide further insight into the molecular aspects of drug resistance in C. albicans, a common human pathogen.
IV. EXPRESSION OF HETEROLOGOUS ABC PROTEINS IN YEAST: ANALYSIS IN VIVO A. Utility of Yeast for Expression of Heterologous Proteins
There are many potential advantages to studying non-yeast ABC proteins in yeast, particularly in cases where an easily assayable phenotype, such as drug resistance or a-factor transport, can be conferred by the heterologous protein. Expression of the heterologous ABC protein can be engineered to be either constitutive or inducible. Intragenic mutations are straightforward to generate, allowing dissection of functional elements within an ABC protein, and one can use classical genetic methods to identify interacting proteins. Finally, if the protein of interest is transported to the cell surface, it should be relatively easy to isolate tightly sealed secretory vesicles containing large quantities of that protein, which can be used for functional assays. This last feature takes advantage of a secretion-defective yeast strain, and is outlined in Section V below. B. Mouse MDR3 Expressed in Yeast Complements a ste6 Mutation and Confers Resistance to the Drug FK520
The structural similarity of STE6 and MDR, combined with the similar hydrophobic properties of their substrates, suggested that a-factor might serve as a substrate for MDR. To test this possibility, Raymond et al. (1992) asked whether mouse MDR3 could complement the mating defect of a yeast strain carrying a deletion of the ste6 gene. High-copy expression of MDR3 was found to partially complement the ste6 defect, enabling yeast to mate at ~ 1 % of the wild-type level, indicative of a low but significant level of function compared to the lack of function (<0.001% of the wild-type level) exhibited by the ste6 deletion strain (Raymond et al., 1992). This finding supports the notion that a-factor can be a substrate for MDR. Expression of MDR3 in yeast was also shown to confer increased resistance to the immunosuppressive drug FK520 (Raymond et al., 1994). Amutation in the eleventh transmembrane span that decreases MDR3-mediated drug resistance in mammalian cells abolishes its ability to promote a-factor export, mating, and drug resistance in yeast, suggesting that MDR3 functions in a similar manner in yeast and mammalian cells. Testing drug resistance conferred by MDR in mammalian cells poses numerous experimental difficulties. In contrast, drug sensitivity assays can be performed in yeast with relative ease. In addition, the yeast mating assay, which can distinguish
ABC Proteins in Yeast
265
levels of mating from 100% down to 0.001%, provides a dramatic increase in sensitivity for the analysis of MDR isoforms that partially complement the mating defect of a ste6 mutant. Thus, yeast appears to be an ideal vehicle for carrying out detailed studies on MDR. C. Plasmodium Falciparum MDR1: Complementation of a steS Deletion
ThepJMDRI gene, which has been associated with drug resistance in the malarial parasite Plasmodium falciparum, was recently expressed in yeast and shown to complement a ste6 mutant for a-factor export and mating (at a level somewhat higher than that observed for mouse MDR3) (Volkman et al., 1995). Introduction into p/MDR of mutations associated with chloroquine resistance abolishes a-factor export and mating, suggesting that, as with mouse MDR3, pfMDRl operates in a similar manner in yeast as in its native organism. D. Human MDR1 Does Not Complement a ste6 Deletion but Confers Valinomycin and FK520 Resistance
Expression of either wild-type human MDRl or a mutant MDRl (G185 V), which has an altered pattern of drug resistance in mammalian cells, results in increased resistance of yeast to valinomycin (Kuchler et al., 1992; Kuchler and Thorner, 1992a; Ni et al., personal communication; Taglicht and Michaelis, unpublished observations). It has been reported that wild-type human MDRl is also able to restore mating to a ste6 mutant, albeit only when it is mated to a particular partner that is hypersensitive to pheromone (MATa sst2) (Kuchler et al., 1992). However, we have been unable to repeat this observation. In our tests, human MDRl confers valinomycin resistance but does not complement a ste6 mutant, whereas a control construct, pfMDRl (described above), does confer a substantial level of mating (Taglicht and Michaelis, unpublished observations). It may be that differences in the plasmid vector, strain background, or particular AfD/?/ construct influence the outcome of this test. Interestingly, both wild-type (Glyl85) and mutant (Vail85) MDRl are localized primarily to the ER in yeast (Kuchler and Thorner, 1992a; Saeki et al., 1991; Taglicht and Michaelis, unpublished observations). The localization of mouse MDR3 in yeast has not been reported, but if it is primarily on the cell surface, this may account for those functional differences between MDRl and MDR3 that are observed in yeast. E. Use of a Drug-Sensitive Yeast Mutant, ergS
One difficulty encountered when studying MDR function in yeast is the inherent resistance of yeast to many of the most common MDR substrates, such as colchicine, daunorubicin, and doxorubicin. Resistance may be due in part to differences in the lipid composition of the yeast and mammalian plasma membranes;
266
CAROL BERKOWER and SUSAN MICHAELIS
yeast contains ergosterol instead of cholesterol. An erg6 mutant, which is defective in synthesis of ergosterol and instead inserts an ergosterol precursor into the plasma membrane, exhibits hypersensitivity to a number of MDR drugs, including daunorubicin, doxorubicin, actinomycin D, and valinomycin (Ni et al., personal communication; Taglicht and Michaelis, unpublished observations). The heightened drug sensitivity of an erg6 mutant could be due to a heightened permeability to these drugs, or to the decreased activity of a hypothetical resident efflux pump. The erg6 mutant promises to be useful for assaying the function of MDR expressed in yeast. Ni et al. have expressed both wild-type (G185) and mutant (VI85) forms of human MDRl in wild-type and erg6 mutant yeast (Ni et al., personal communication). Expression of either wild-type or mutant MDRl in the erg6 strain increases resistance to actinomycin D, daunorubicin, doxorubicin, and valinomycin, although resistance to valinomycin is much greater for the strain expressing wild-type MDRl than for that expressing the VI85 mutant form of the protein. Thus, human MDRl has a similar profile of multidrug resistance in yeast as it does in cultured mammalian cells. It was also shown that vesicles derived from membranes of an ERG6 (wild-type) strain expressing wild-type MDRl demonstrate an ATP-dependent increase in vinblastine uptake (Ni and Gottesman, personal communication).
V. USE OF THE YEAST sec6 MUTANT TO STUDY HETEROLOGOUS TRANSPORTERS: ANALYSIS IN VITRO A. Isolation of Membrane Proteins in Secretory Vesicles
Newly synthesized plasma membrane proteins in yeast are transported from the Golgi to the cell surface by secretory vesicles. In the mutants sec6 and seel, vesicles carrying proteins to the cell surface are prevented from fusing with the plasma membrane at the restrictive temperature (Novick et al., 1980; Novick and Schekman, 1983). Slayman and coworkers have shown that these vesicles accumulate in the cell and can be isolated in significant yields at high purity with a minimal amount of manipulation (Nakamoto et al., 1991). The isolated vesicles are tight enough to maintain a proton gradient across their membrane. An ABC transporter that is destined for the plasma membrane would accumulate in these inside-out vesicles, such that the NBD is accessible to the buffer, and appropriate substrates should be transported into the vesicle interior. We shall refer to this system as the sec vesicle system. A study examining the function of mouse MDRl and MDR3 proteins expressed in sec6 mutant yeast has provided compelling evidence of the utility of the sec vesicle system. When mouse MDRl, MDRl, or MDR3 is expressed in a sec6 strain, the secretory vesicles that accumulate contain the corresponding MDR protein in the sec vesicle preparation (Ruetz and Gros, 1994a). Ruetz and Gros have demonstrated ATP-dependent translocation of drugs into vesicles carrying MDRl or
ABC Proteins in Yeast
267
MDR3 (but not MDR2). Drug transport was dependent on ATP, but was independent of a proton gradient across the membrane, and drugs were actively moved against a concentration gradient, indicating that these transporters were acting as pumps (Ruetz and Gros, 1994a). The absence of transport into vesicles containing MDR2 demonstrated that the sec vesicle transport assay is a reliable indicator of function, since the MDR2 gene has never been associated with multidrug resistance in vivo. Ruetz et al. had previously demonstrated transport of drugs into plasma membrane-derived vesicles from yeast expressing MDRl (Ruetz et al., 1993); however, quantitation of intravesicular drug concentration only become possible when they turned to the sec vesicle system. B. Mouse MDR2 Is a Lipid Flippase
MDR2 is not associated with drug resistance. Instead, evidence from mouse knock-out experiments suggested that MDR2 might be involved in phosphatidylcholine (PC) movement across the apical membrane of hepatocytes (Smit et al., 1993a). In order to test mouse MDR2 and MDR3 for the ability to act as a lipid flippase, Ruetz and Gros developed a method for quantitating asymmetric lipid distribution between the inner and outer leaflets ofsec6 vesicles containing MDR2 (Ruetz and Gros, 1994b). A fluorescent phosphatidylcholine analogue (NBD-PC) was incorporated into the outer leaflet of purified sec vesicles containing MDR2. Transfer of NBD-PC to the inner leaflet of secretory vesicles could be monitored as an increase in fluorescence associated with the inner leaflet (i.e., resistance to a membrane impermeant quenching reagent). In this way, it was shown that MDR2 (but not MDR3) is able to mediate flipping of NBD-PC from the outer to the inner leaflet of yeast sec vesicles in an ATP-dependent manner. This result is consistent with the phenotype of the mdr2 mutant mouse, whose bile is devoid of PC (Smit et al., 1993a), and supports the notion that MDR2 translocates PC across the hepatocyte membrane via a flippase mechanism. C. Yeast has an Endogenous Bile Acid Transporter
Additional work in seel and sec6 mutants has revealed an endogenous yeast activity that transports the bile acid taurocholate (TC) and the glutathione conjugate GS-DNP into sec vesicles (St-Pierre et al., 1994). TC and GS-DNP are normally substrates for the canalicular transporters of mammalian liver cells. Their uptake into yeast sec vesicles is temperature-dependent and saturable, and requires ATP and Mg"*""*". Characterization of uptake indicates that yeast possesses ATP-dependent transporter activities that are fiinctionally similar to the exporter(s) of bile acids and glutathione conjugates found in mammalian liver cells. The yeast transporter(s) that carry out TC and GS-DNP transport remain to be identified. It will be interesting to learn whether they are members of the ABC superfamily. Purification of the yeast activities or cloning of their gene(s) may aid in the identification of the mammalian bile canalicular organic anion transporters.
268
CAROL BERKOWER and SUSAN MICHAELIS
VI. CONCLUSION: EMERGING PERSPECTIVES IN THE STUDY OF THE ABC SUPERFAMILY ABC proteins continue to turn up in interesting places, and in roles that are relevant to many aspects of cell physiology and disease. Yeast has proven to be an ideal system for studying ABC transporters. The genetic and biochemical potential of yeast for analysis of heterologous ABC proteins from less tractable organisms has only just begun to be tapped. The use of the sec vesicle system to obtain biochemically active ABC proteins has already made a significant contribution to our understanding of how several MDR proteins function, and will undoubtedly continue to illuminate the functions of other ABC proteins over the next few years. The availability of altered substrates, such as mutant forms of a-factor that are not exported by STE6, will permit detailed studies of substrate specificity. Studies on the biogenesis and intracellular trafficking of ABC proteins in yeast should shed light on the cellular machinery involved in folding these proteins and in directing them to the appropriate intracellular compartments. Finally, the yeast genome project has provided a cornucopia of new ABC proteins that can now be dissected using the biochemical and genetic methods available in yeast, and these may provide new paradigms for the ABC superfamily as a whole.
ACKNOWLEDGMENTS S. M. is supported by grants GMS1508 and DK48977 from the NIH.
REFERENCES Ambudkar, S. V., Lelong, I. H., Zhang, J., Cardarelli, C. O., Gottesman, M. M., «fe Pastan, I. (1992). Partial purification and reconstitution of the human multidrug-resistance pump: Characterization of the drug-stimulatable ATP hydrolysis. Proc. Natl. Acad. Sci. USA 89, 8472-8476. Ames, G. F.-L. (1993). Bacterial periplasmic permeases as model systems for multidrug resistance (MDR) and the cystic fibrosis transmembrane conductance regulator (CFTR). In: Molecular Biology and Function of Carrier Proteins (Reuss, L., Russell, J. M., & Jennings, M. L., eds.), pp. 77—94. The Rockefeller University Press. Ames, G. F.-L., Mimura, C. S., Holbrook, S. R., & Shyamala, V. (1992). Traffic ATPases: A superfamily of transport proteins operating from Escherichia coli to humans, pp. 1-^7. In: Advances in Enzymology Related Areas of Molecular Biology (Meister, A., ed.), John Wiley & Sons, New York. Anderegg, R. J., Betz, R., Carr, S. A., Crabb, J. W., & Duntze, W. (1988). Structure of Saccharomyces cerevisiae mating hormone a-factor: Identification of S-famesyl cysteine as a structural component. J. Biol. Chem. 263, 18236-18240. Androlewicz, M. J., Ortmann, B., van Endert, P. M., Spies, T., & Cresswell, P. (1994). Characteristics of peptide and major histocompatibility complex class I/p2-microglobulin binding to the transporters associated with antigen processing (TAPl and TAP2). Proc. Natl. Acad. Sci. USA 91, 12716-12720.
ABC Proteins in Yeast
269
Azzaria, M., Schurr, E., & Gros, P. (1989). Discrete mutations introduced in the predicted nucleotidebinding sites of the mdrl gene abolish its ability to confer multidrug resistance. Mol. Cell. Biol. 9, 5289-5297. Baba, M., Baba, N., Ohsumi, Y, Kanaya, K., & Osumi, M. (1989). Three-dimensional analysis of morphogenesis induced by mating pheromone a factor in Saccharomyces cerevisiae. J. Cell Science 94, 207-216. Baichwal, V., Liu, D., & Ames, G. F.-L. (1993). The ATP-binding component of a prokaryotic traffic ATPase is exposed to the periplasmic (external) surface. Proc. Natl. Acad. Sci. USA 90,620-624. Balzi, E. & Goffeau, A. (1991). Multiple or pleiotropic drug resistance in yeast. Biochim. Biophys. Acta 1073,241-252. Balzi, E. & Goffeau, A. (1994). Genetics and biochemistry of yeast multidrug resistance. Biochim. Biophys. Acta 1187, 152^162. Balzi, E., Wang, M., Leterme, S., Van Dyck, L., & Goffeau, A. (1994). PDR5, a novel yeast multidrug resistance conferring transporter controlled by the transcription regulator PDRl. J. Biol. Chem. 269,2206-2214. Berkower, C. (1995). Analysis of STE6, a Saccharomyces cerevisiae ATP binding cassette (ABC) protein. Ph. D. Thesis, Johns Hopkins University, Baltimore, MD. Berkower, C, Loayza, D., & Michaelis, S. (1994). Metabolic instability and constitutive endocytosis of STE6, the a-factor transporter of Saccharomyces cerevisiae. Mol. Biol. Cell 5, 1185-1198. Berkower, C. & Michaelis, S. (1991). Mutational analysis of the yeast a-factor transporter STE6, a member of the ATP binding cassette (ABC) protein superfamily. EMBO J. 10, 3777-3785. Berkower, C. & Michaelis, S. (1993). Effects of nucleotide binding fold mutations on STE6, a yeast ABC protein. In: Molecular Biology and Function of Carrier Proteins (Reuss, L., Russell, J. M., & Jennings, M. L., eds.), pp. 130-146. The Rockefeller University Press. Bibi, E. & Beja, O. (1994). Membrane topology of multidrug resistance protein expressed in Escherichia coli. J. Biol. Chem. 269, 19910-19915. Bishop, L., Agbayani, R., Jr., Ambudkar, S. V., Maloney, P C, & Ames, G. F.-L. (1989). Reconstitution of a bacterial periplasmic permease in proteoliposomes and demonstration of ATP hydrolysis concomitant with transport. Proc. Natl. Acad. Sci. USA 86, 6953-6957. Bissinger, P. H. & Kuchler, K. (1994). Molecular cloning and expression of the Saccharomyces cerevisiae STS1 gene product: A yeast ABC transporter conferring mycotoxin resistance. J. Biol. Chem. 269,4180-^186. Bossier, P., Femandes, L., Vilela, C, & Rodrigues-Pousada, C. (1994). The yeast YKL741 gene situated on the left arm of chromosome XI codes for a homologue of the human ALD protein. Yeast 10, 681-686. Boyd, D., Manoil, C, & Beckwith, J. (1987). Determinants of membrane protein topology. Proc. Natl. Acad. Sci. USA 84, 8525-8529. Boyd, D., Traxler, B., & Beckwith, J. (1993). Analysis of the topology of a membrane protein by using a minimum number of alkaline phosphatase fusions. J. Bacteriol. 175, 553—556. Buchman, A. R., Kimmerly, W. J., Rine, J., & Romberg, R. D. (1988). Two DNA-binding factors recognize specific sequences at silencers, upstream activating sequences, autonomously replicating sequences, and telomeres in Saccharomyces cerevisiae. Mol. Cell. Biol. 8, 210-225. Carson, M. R., Travis, S. M., & Welsh, M. J. (1995). The two nucleotide-binding domains of cystic fibrosis transmembrane conductance regulator (CFTR) have distinct functions in controlling channel activity. J. Biol. Chem. 270, 1711-1717. Chen, P. (1993). Ph.D. Thesis. The Johns Hopkins University School of Medicine, Baltimore, MD. Cheng, S. H., Rich, D. P, Marshall, J., Gregory, R. J., Welsh, M. J., & Smith, A. E. (1991). Phosphorylation of the R domain by cAMP-dependent protein kinase regulates the CFTR chloride channel. Cell 66, 1027-1036.
270
CAROL BERKOWER and SUSAN MICHAELIS
Cole, S. P. C, Bhardwaj, G., Gerlach, J. H., Mackie, J. E., Grant, C. E., Almquist, K. C, Stewart, A. J., Kurz, E. U., Duncan, A. M., & Deeley, R. G. (1992). Overexpression of a transporter gene in a multidrug-resistant human lung cancer cell line. Science 258, 1650-1654. Collins, F. (1992). Cystic fibrosis: Molecular biology and therapeutic implications. Science 256, 774-779. Colthurst, D. R., Schauder, B. S., Hayes, M. V., & Tuite, M. F. (1992). Elongation factor 3 (EF-3) from Candida albicans shows both structural and functional similarity to EF-3 from Saccharomyces cerevisiae. Mol. Microbiol. 6, 1025—1033. Cutting, G. R. (1994). Mutations that cause cystic fibrosis. NIDDK workshop on the novel ATPases and disease-ATP-dependent transporters, an emerging superfamily. National Institutes of Health, Bethesda, MD. Davey, J. (1992). Mating pheromones of the fission yeast Schizosacchawmyces pombe: Purification and structural characterization of M-factor and isolation and structural characterisation of two genes encoding the pheromone. EMBO J. 11, 951-960. Davidson, A. L. & Nikaido, H. (1990). Overproduction, solubilization and reconstitution of the maltose transport system from Escherichia coli. J. Biol. Chem. 265, 4254—4260. Dean, M., AUikmets, R., Gerrard, B., Stewart, C, Kistler, A., Shafer, B., Michaelis, S., & Strathem, J. (1994). Mapping and sequencing of two yeast genes belonging to the ATP-binding cassette superfamily. Yeast 10, 377-383. Decottignies, A., Kolaczkowski, M., Balzi, E., & Goffeau, A. (1994). Solubilization and characterization of the overexpressed PDR5 multidrug resistance nucleotide triphosphatase of yeast. J. Biol. Chem. 269, 12797-12803. Di Domenico, B., Lupisella, J. A., Sandbaken, M. G., & Chakraburty, K. (1992). Isolation and sequence analysis of the gene encoding translation elongation factor 3 from Candida albicans. Yeast 8, 337-352. Doige, C. A. 8L Ames, G. F.-L. (1993). ATP-dependent transport systems in bacteria and humans: Relevance to cystic fibrosis and multidrug resistance. Annu. Rev. Microbiol. 47, 291—319. Dorsman, J. C, Gozdzicka-Jozefiak, A., van Heeswijk, W. C , & Grivell, L. A. (1991). Multi-functional DNA proteins in yeast: The factors GFI and GFII are identical to the ARS-binding factor ABFI and the centromere-binding factor CPFl respectively. Yeast 7,401-412. Dreesen, T. D., Johnson, D. H., & HenikofF, S. (1988). The brown protein of Drosophila melanogaster is similar to the white protein and to components of active transport complexes. Mol. Cell. Biol. 8,5206-5215. Egan, M., Flotte, T., Afione, S., Solow, R., Zeitlin, R L., Carter, B. J., & Guggino, W. B. (1992). Defective regulation of outwardly rectifying CI- channels by protein kinase Acorrected by insertion of CFTR. Nature (London) 358, 581-584. Ewart, G. D., Cannell, D., Cox, G. B., & Howells, A. J. (1994). Mutational analysis of the traffic ATPase (ABC) transporters involved in uptake of eye pigment precursors in Drosophila melanogaster. JBC269, 10370-10377. Fath, M. J. & Kolter, R. (1993). ABC transporters: Bacterial exporters. Microbiological Reviews 57, 995-1017. Fearon, K., McClendon, V., Bonetti, B., & Bedwell, D. M. (1994). Premature translation termination mutations are efficiently suppressed in a highly conserved region of yeast Ste6p, a member of the ATP-binding cassette (ABC) transporter family. J. Biol. Chem. 269, 17802-1 ^808. Gabriel, S. E., Clarke, L. L., Boucher, R. C, & Stutts, M. J. (1993). CFTR and outward rectifying chloride channels are distinct proteins with a regulatory relationship. Nature (London) 363, 263-266. Geller, D., Taglicht, D., Edgar, R., Tam, A., Pines, O., Michaelis, S., & Bibi, E. (1996). Comparative topology studies in Saccharomyces cerevisiae and in Escherichia coli of the N-terminal half of the yeast ABC protein, Ste6. JBC (in press).
ABC Proteins in Yeast
271
Gentschev, I. & Goebel, W. (1992). Topological and functional studies on HlyB of Escherichia coli. Mol. Gen. Genet. 232,40-48. Germann, U. A., Chin, K.-V., Pastan, I., & Gottesman, M. M. (1990a). Retroviral transfer of a chimeric multidrug resistance-adenosine deaminase gene. FASEB J. 4, 1501-1507. Germann, U. A., Pastan, I., & Gottesman, M. M. (1993). P-glycoproteins: Mediators of muhidrug resistance. Seminars in Cell Biol. 4, 63—76. Germann, U. A., Willingham, M. C, Pastan, I., & Gottesman, M. M. (1990b). Biochemistry 29, 2295-2303, Gill, D. R., Hyde, S. C, Higgins, C. F., Valverde, M. A., Mintenig, G. M., & Sepulveda, F. V, (1992). Separation of drug transport and chloride channel functions of the human multidrug resistance P-glycoprotein. Cell 71, 23-32. Goflfeau, A., Ghislain, M„ Navarre, C, Pumelle, B., & Supply, R (1990). Novel transport ATPases in yeast. Biochim. Biophys. Acta 1018, 200-202. Goffeau, A., Nakai, K., Slominski, P., & Risler, J.-L. (1993). The membrane proteins encoded by yeast chromosome III genes. FEBS Letters 325, 112-117. Gottesman, M. M. & Pastan, I. (1993). Biochemistry of multidrug resistance mediated by the multidrug transporter. Annu. Rev. Biochem. 62, 385-427. Grant, C. E., Valdimarsson, G., Hipfner, D. R., Almquist, K. C, Cole, S. R, & Deeley, R. (1994). Overexpression of multidrug resistance-associated protein (MRP) increases resistance to natural product drugs. Cancer Research 54, 357—361. Gregory, R. J., Rich, D. P., Cheng, S. H., Sbuza, D. W., Paul, S., Manavalan, P, Anderson, M. R, Welsh, M. J., & Smith, A. E. (1991). Maturation and function of cystic fibrosis transmembrane conductance regulator variants bearing mutations in putative nucleotide-binding domains 1 and 2. Mol. Cell. Biol. 11,3886-3893. Gros, P., Dhir, R., Croop, J., & Talbot, F. (1991). A single amino acid substitution strongly modulates the activity and substrate specificity of the mouse mdrl and mdr3 drug efflux pumps. Proc. Natl. Acad. Sci. USA 88, 7289-7293. Hardy, S. R, Goodfellow, H. R., Valverde, M. A., Gill, D. R., Sepulveda, F. V., & Higgins, C. F. (1995). Protein kinase C-mediated phosphorylation of the human multidrug resistance P-glycoprotein regulates cell volume-activated chloride currents. EMBO J. 14, 68-75. He, B., Chen, R, Chen, S.-Y, Vancura, K. L., Michaelis, S., & Powers, S. (1991). RAM2, an essential gene of yeast, and RAMI encode the two polypeptide components of the famesyltransferase that prenylates a-factor and Ras proteins. Proc. Natl. Acad. Sci. USA 88, 11373-11377. Higgins, C. F. (1992). ABC transporters: From microorganisms to man. Annu. Rev. Cell Biol. 8,67—113. Higgins, C. R, Hiles, I. D., Salmond, G. R C, Gill, D. R., Downie, J. A., Evans, I. J., Holland, I. B., Gray, L., Buckel, S. D., Bell, A. W., & Hermondson, M. A. (1986). A family of related ATP-binding subunits coupled to many distinct biological processes in bacteria. Nature (London) 323,448-450. Hinnebusch, A. G. (1992). General and pathway-specific regulatory mechanisms controlling the synthesis of amino acid biosynthetic enzymes in Saccharomyces cerevisiae. In: The Molecular and Cellular Biology of the Yeast Saccharomyces: Gene Expression (Jones, E. W., Pringle, J. R., & Broach, J. R., eds.), pp. 319-414. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York. Hirata, D., Yano, K., Miyahara, K., & Miyakawa, T. (1994). Saccharomyces cerevisiae YDRl, which encodes a member of the ATP-binding cassette (ABC) superfamily, is required for multidrug resistance. Current Genetics 26, 285-294. Hoof, T., Demmer, A., Hadam, M. R., Riordan, J. R., & Tummler, B. (1994). Cystic fibrosis-type mutational analysis in the ATP-binding cassette transporter signature of human P-glycoprotein MDRl. J. Biol. Chem. 269, 20575-20583. Huet, J. & Sentenac, A. (1987). TUF, the yeast DNA-binding factor specific for UASrpg upstream activating sequences: Identification of the protein and its DNA-binding domain. Proc. Natl. Acad. Sci. USA 84, 364^-3652.
272
CAROL BERKOWER and SUSAN MICHAELIS
Hyde, S. C, Emsley, P., Hartshorn, M. J., Mimmack, M. M., Gileadi, U., Pearce, S. R., Gallagher, M. P., Gill, D. R., Hubbard, R. E., & Higgins, C. F. (1990). Structural model of ATP-binding proteins associated with cystic fibrosis, multidrug resistance and bacterial transport. Nature (London) 346, 362-365. Johnson, A. D. (1992). A combinatorial regulatory circuit in budding yeast. In: Transcriptional Regulation (McKnight, S. L. & Yamamoto, K. R., eds.), pp. 975-1006. Cold Spring Harbor Laboratory Press, Plainview, New York. Johnson, A. D. & Herskowitz, I. (1985). A repressor (MATa2 product) and its operator control expression of a set of cell type specific genes in yeast. Cell 42, 237-247. Jovov, B., Ismailov, L L, & Benos, D. J. (1995). Cystic fibrosis transmembrane conductance regulator is required for protein kinase A activation of an outwardly rectified anion channel purified from bovine tracheal epithelia. J. Biol. Chem. 270, 1521-1528. Julius, D., Schekman, R., & Thomer, J. (1984). Glycosylation and processing of prepro-a-factor through the yeast secretory pathway. Cell 36, 309-318. Kamijo, K., Taketani, S., Yokata, S., Osumi, T., & Hashimoto, T. (1990). The 70-kDa peroxisomal membrane protein is a member of the Mdr (P-glycoprotein)-related ATP-binding protein superfamily. J. Biol. Chem. 265, 4534^540. Kelly, A., Powis, S. H., Kerr, L. A., Mockridge, L, Elliott, T, Bastin, J., Uchanska-Ziegler, B., Ziegler, A., Trowsdale, J., & Townsend, A. (1992). Assembly and function of the two ABC transporter proteins encoded in the human major histocompatibility complex. Nature (London) 355,641-644. Kerem, B. S., Rommens, J. M., Buchanan, J. A., Markiewicz, D., Cox, T. K., Chakravarti, A., Buchwald, M., & Tsui, L. C. (1989). Identification of the cystic fibrosis gene: Genetic analysis. Science 245, 1073-1080. Kerppola, R. E. & Ames, G. F.-L. (1992). Topology of the hydrophobic membrane-bound components of the histidine periplasmic permease. J. Biol. Chem. 267, 2329-2336. Kleijmeer, M. J., Kelly, A., Geuze, H. J., Slot, J. W., Townsend, A., & Trowsdale, J. (1992). Location of MHC-encoded transporters in the endoplasmic reticulum and d5-Golgi. Nature (London) 357, 342-344. Kolling, R. & Hollenberg, C. P. (1994a). The ABC-transporter Ste6 accumulates in the plasma membrane in a ubiquitinated form in endocytosis mutants. EMBO J. 13, 3261-3271. Kolling, R. & Hollenberg, C. R (1994b). The first hydrophobic segment of the ABC-transporter, Ste6, functions as a signal sequence. FEBS Letters 351, 155-158. Kralli, A., Bohen, S. R, & Yamamoto, K. R. (1995). LEMl, an ATP-binding-cassette transporter, selectively modulates the biological potency of steroid hormones. Proc. Natl. Acad. Sci. USA 92, 4701-^705. Kuchler, K., Dohlman, H. G., & Thomer, J. (1993). The a-factor transporter (STE6 gene product) and cell polarity in the yeast Saccharomyces cerevisiae. J. Cell Biol. 120, 1203—1215. Kuchler, K., Goransson, H. M., Viswanathan, M. N., & Thomer, J. (1992). Dedicated transporters for peptide export and intercompartmental traffic in the yeast Saccharomyces cerevisiae. Cold Spring Harbor Symposia on Quantitative Biology 57, 579-592. Kuchler, K., Sterne, R. E., & Thomer, J. (1989). Saccharomyces cerevisiae STE6gene product: A novel pathway for protein export in eukaryotic cells. EMBO J. 8, 3973—3984. Kuchler, K. & Thomer, J. (1992a). Functional expression of human mdrl in the yeast Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. USA 89, 2302-2306. Kuchler, K. & Thomer, J. (1992b). Secretion of peptides and proteins lacking hydrophobic signal sequences: The role of adenosine triphosphate-driven membrane translocators. Endocrine Reviews 13,499-514. Leighton, J. & Schatz, G. (1995). An ABC transporter in the mitochondrial inner membrane is required for normal growth of yeast. EMBO J. 14, 188-195.
ABC Proteins in Yeast
273
Leppert, G., McDevitt, R., Falco, S. C, Van Dyk, T. K., Ficke, M. B., & Golin, J. (1990). Cloning by gene amplification of two loci conferring multiple drug resistance in Saccharomyces. Genetics 125, 13-20. Loo, T. W. & Clarke, D. M. (1994a). Prolonged association of temperature-sensitive mutants of human P-glycoprotein with calnexin during biogenesis. J. Biol. Chem. 269, 28683—28689. Loo, T. W. & Clarke, D. M. (1994b). Reconstitution of drug-stimulated ATPase activity following co-expression of each half of human P-glycoprotein as separate polypeptides. J. Biol. Chem. 269, 7750-7755. Manavalan, P., Smith, A. E., & McPherson, J. M. (1993). Sequence and structural homology among membrane-associated domains of CFTR and certain transporter proteins. J. Prot. Chem. 12, 279-290. Marcus, S., Caldwell, G. A., Miller, D., Xue, C.-B., Naider, F., & Becker, J. M. (1991). Significance of C-terminal cysteine modifications to the biological activity of the Saccharomyces cerevisiae a-factor mating pheromone. Mol. Cell. Biol. 11, 3603-3612. Marger, M. D. & Saier, M. H. J. (1993). A major superfamily of transmembrane facilitators that catalyze uniport, symport and antiport. TIBS 18, 13—20. McGrath, J. P. & Varshavsky, A. (1989). The yeast STE6 gene encodes a homologue of the mammalian multidrug resistance P-glycoprotein. Nature (London) 340, 400-404. Meyers, S., Schauer, W., Balzi, E., Wagner, M., GoflFeau, A., & Golin, J. (1992). Interaction of the yeast pleiotropic drug resistance genes PDRl and PDR5. Current Genetics 21, 431-436. Michaelis, S. (1993). STE6, the yeast a-factor transporter. Seminars in Cell Biology 4, 17-27. Michaelis, S., & Berkower, C. (1995). Sequence comparison of yeast ABC proteins. In: Cold Spring Harbor Symposium on Quantitative Biology 60: Protein kinesis- the dynamics of protein trafficking and stability, pp. 291-307. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York. Michaelis, S. & Herskowitz, 1.(1988). The a-factor pheromone of Saccharomyces cerevisiae is essential formating. Mol. Cell. Biol. 8, 1309-1318. Mimura, C. S., Holbrook, S. R., & Ames, G. F.-L. (1991). Structural model of the nucleotide-binding conserved component of periplasmic permeases. Proc. Natl. Acad. Sci. USA 88, 84—88. Mintenig, G. M., Valverde, M. A., Sepulveda, F. V., Gill, D. R., Hyde, S. C, Kirk, J., & Higgins, C. F. (1993). Specific inhibitors distinguish the chloride channel and drug transporter functions associated with the human multidrug resistance P-glycoprotein. Receptors & Channels 1, 305-313. Momburg, F., Neefjes, J. J., & Hammerling, G. J. (1994). Peptide selection by MHC-encoded TAP transporters. Current Opinion in Immunology 6, 32—37. Mosser, J., Lutz, Y., Stoeckel, M. E., Sarde, C.-O., Kretz, C , Douar, A.-M., Lopez, J., Aubourg, R, & Mandel, J.-L. (1994). The gene responsible for adrenoleukodystrophy encodes a peroxisomal membrane protein. Hum. Molec. Genet. 3, 265-271. Myers, K. K., Fonzi, W. A., & Sypherd, P. S. (1992). Isolation and sequence analysis of the gene for translation elongation factor 3 from Candida albicans. Nucleic Acids Res. 20, 1705—1710. Nagao, K., Taguchi, Y, Arioka, M., Kadokura, H., Takatsuki, A., Yoda, K., & Yamasaki, M. (1995). bfrl-^, a novel gene of Schizosaccharomyces pombe which confers brefeldin A resistance, is structurally related to the ATP-binding cassette superfamily. J. Bacteriol. 177, 1536-1543. Nakamoto, R. K., Rao, R., & Slayman, C. W. (1991). Expression of the yeast plasma membrane [H+]ATPase in secretory vesicles. J. Biol. Chem. 266, 7940-7949. Neefles, J. J., Momburg, F., & Hammerling, G. J. (1993). Selective and ATP-dependent translocation of peptides by the MHC-encoded transporter. Science 261, 769-771. Nielsen, O. (1993). Signal transduction during mating and meiosis in S. pombe. Trends in Cell Biology 3, 60-65. Nishi, K., Yoshida, M., Nishimura, M., Nishikawa, M., Nishiyama, M., Horinouchi, S., & Beppu, T. (1992). A leptomycin B resistance gene oiSchizosaccharomyces pombe encodes a protein similar to the mammalian P-glycoproteins. Mol. Microbiol. 6, 761—769.
274
CAROL BERKOWER and SUSAN MICHAELIS
Novick, P., Field, C, & Schekman, R. (1980). Identification of 23 complementation groups required for post-translational events in the yeast secretory pathway. Cell 21, 205—215. Novick, P. & Schekman, R. (1983). Export of major cell surface proteins is blocked in yeast secretory mutants. J. Cell Biol. 96, 541-547. O'Hare, K., Murphy, C, Levis, R., & Rubin, G. M. (1984). DNA sequence of the white locus of Drosophila melanogaster. J. Mol. Biol. 180, 437-455. Oliver, S. G. & a. 1. others. (1992). The complete DNA sequence of yeast chromosome III. Nature (London) 357, 38-46. Ortiz, D. R, Kreppel, L., Speiser, D. M., Scheel, G., McDonald, G., & Ow, D. W. (1992). Heavy metal tolerance in the fission yeast requires an ATP-binding cassette-type vacuolar membrane transporter. EMBO J. 11, 3491-3499. Ortiz, D. F., Ruscitti, T., McCue, K. F., & Ow, D. W. (1995). Transport of metal-binding peptides by HMTl, a fission yeast ABC-type vacuolar membrane protein. JBC 270, 4721-4728. Ortmann, B., Androlewicz, M. J., & Cresswell, P. (1994). MHC class I/p2-microglobulin complexes associate with TAP transporters before peptide binding. Nature (London) 368, 864—867. Oshima, T. & Takano, I. (1980). Mutants showing heterothallism from a homothallic strain of Saccharomyces cerevisiae. Genetics 94, 841-857. Ostedgaard, L. S., Rich, D. P, DeBerg, L. G., & Welsh, M. J. (1994). Association between domains of CFTR (abstract). Mol. Biol. Cell 5, 191a. Paddon, C, Loayza, D., Vangelista, L., Solari, R., «fe Michaelis, S. (1996). Analysis of the localization of STE6/CFTR chimeras in a Saccharomyces cerevisiae model for the cystic fibrosis defect CFTRAF508, Mol. Micro. 19, 1007-1017. Panagiotidis, C. H., Reyes, M., Sievertsen, A., Boos, W., & Shuman, H. A. (1993). Characterization of the structural requirements for assembly and nucleotide binding of an ATP-binding cassette transporter: The maltose transport system oiEscherichia coli. J. Biol. Chem. 268, 23685—23696. Parham, P (1992). Flying the first class flag. Nature (London) 357, 193-194. Pearce, S. R., Mimmack, M. L., Gallagher, M. P, Gileadi, U., Hyde, S. C, & H. C. F. (1992). Membrane topology of the integral membrane components, OppB and OppC, of the oligopeptide permease of Salmonella typhimurium. Mol. Microbiol. 6,47—57. Pind, S., Riordan, J. R., & Williams, D. B. (1994). Participation of the endoplasmic reticulum chaperone calnexin (p88, IP90) in the biogenesis of the cystic fibrosis transmembrane conductance regulator. J. Biol. Chem. 269, 12784-12788. Prasad, R., De Wergifosse, P., Goflfeau, A., & Balzi, E. (1995). Molecular cloning and characterization of a novel gene of Candida albicans, CDRl, conferring multiple resistance to drugs and antifungals. Current Genetics 27, 320-329. Pumelle, B., Skala, J., & GofFeau, A. (1991). The product of the YCR105 gene located on the chromosome III from Saccharomyces cerevisiae presents homologies to ATP-dependent permeases. Yeast 7, 867-872. Qin, S., Xie, A., Bonato, M. C. M., & McLaughlin, C. S. (1*^90). Sequence analysis of the translation elongation factor 3 from Saccharomyces cerevisiae. J. Biol. Chem. 265, 1903-1912. Raymond, M., Gros, P., Whiteway, M., & Thomas, D. Y. (1992). Functional complementation of yeast ste6 by a mammalian multidrug resistance mdr gene. Science 256, 232—234. Raymond, M., Ruetz, S., Thomas, D. Y, & Gros, P. (1994). Functional expression of P-glycoprotein in Saccharomyces cerevisiae confers cellular resistance to the immunosuppressive and antifungal agent FK520. Mol. Cell. Biol. 14, 277-286. Rich, D. R, Gregory, R. J., Anderson, M. R, Manalavan, R, Smith, A. E., & Welsh, M. J. (1991). Effect of deleting the R domain on CFTR-generated chloride channels. Science 253, 205—207. Rine, J. (1979). Regulation of transposition of cryptic mating type genes in Saccharomyces cerevisiae. Ph.D. Thesis. University of Oregon. Riordan, J. R., Rommens, J. M., Kerem, B S., Alon, N., Rozmahel, R., Grzelczak, Z., Zielenski, J., Lok, S., Plavsic, N., Chou, J. L., Drumm, M. L., lannuzzi, M. C, Collins, F. S., & Tsui, L. C. (1989).
ABC Proteins in Yeast
275
Identification of the cystic fibrosis gene: Cloning and characterization of complementary DNA. Science 245, 1066-1073. Ruetz, S. & Gros, P. (1994a). Functional expression of P-glycoproteins in secretory vesicles. J. Biol. Chem. 269, 12277-12284. Ruetz, S. & Gros, P. (1994b). Phosphatidylcholine translocase: A physiological role for the mdr2 gene. Cell 77, 1071-1081. Ruetz, S., Raymond, M., & Gros, P. (1993). Functional expression of P-glycoprotein encoded by the mouse mdrS gene in yeast cells. Proc. Natl. Acad. Sci. USA 90, 11588-11592. Saeki, T., Shimabuku, A. M., Azuma, Y, Shibano, Y., Komano, T., & Ueda, K. (1991). Expression of human P-glycoprotein in yeast cells—effects of membrane component sterols on the activity of P-glycoprotein. Agric. Biol. Chem. 55, 1859-1865. Saitoh, N., Goldberg, I. G., Wood, E. R., & Eamshaw, W. C. (1994). Sell: An abundant chromosome scaffold protein is a member of a family of putative ATPases with an unusual predicted tertiary structure. J. Cell Biol. 127, 303-318. Sapperstein, S., Berkower, C, & Michaelis, S. (1994). Nucleotide sequence of the yeast STEM gene, which encodes famesylcysteine carboxyl methyltransferase, and demonstration of its essential role in a-factor export. Mol. Cell. Biol. 14, 1438-1449. Saraste, M., Sibbald, R R., & Wittinghofer, A. (1990). The P-loop—A common motif in ATP- and GTP-binding proteins. Trends Biochem. Sci. 15, 430-434. Sarkadi, B., Price, E. M., Boucher, R. C, Germann, U. A., & Scarborough, G. A. (1992). J. Biol. Chem. 267, 4854-4858. Saurin, W., Koster, W., & Dassa, E. (1994). Bacterial binding protein-dependent permeases: Characterization of distinctive signatures for functionally related integral cytoplasmic membrane proteins. Mol. Microbiol. 12, 993-1004. Schinkel, A. H., Smit, J. J. M., van Telligen, O., Beijnen, J. H., Wagenaar, E., van Deemter, L., Mol, C. A. A. M., van der Valk, M. A., Robanus-Maandag, E. C , te Riele, H. P J., Bems, A. J. M., & Borst, P. (1994). Disruption of the mouse mdrla P-glycoprotein gene leads to a deficiency of the blood-brain barrier and to increased sensitivity to drugs. Cell 77, 491—502. Schwiebert, E. M., Flotte, T, Cutting, G. R., & Guggino, W. B. (1994). Both CFTR and outwardly rectifying chloride channels contribute to cAMP-stimulated whole cell chloride currents. Am. J. Physiol. 266, C1464-C1477. Servos, J., Haase, E., & Brendel, M. (1993). Gene SNQ2 of Saccharomyces cerevisiae, which confers resistance to 4-nitroquinoline-N-oxide and other chemicals, encodes a 169 kDa protein homologous to ATP-dependent permeases. Mol. Gen. Genet. 236,214—218. Shani, N., Watkins, P. A., & Valle, D. (1995). PXAl, an S. cerevisiae homologue of the human adrenoleukodystrophy gene. Proc. Natl. Acad. Sci. USA 92, 6012-6016. Sharom, F. J., Yu, X., & Doige, C. A. (1993). Functional reconstitution of drug transport and ATPase activity in proteoliposomes containing partially purified P-glycoprotein. J. Biol. Chem. 268, 24197-24202. Shepherd, J. C, Schumacher, T. N. M., Ashton-Rickardt, P. G., Imaeda, S., Ploegh, H. L., Janeway Jr., C. A., & Tonegawa, S. (1993). TAP 1-dependent peptide translocation in vitro is ATP dependent and peptide selective. Cell 74, 577—584. Shyamala, V., Baichwal, V., Beall, E., & Ames, G. F.-L. (1991). Structure-function analysis of the histidine permease and comparison with cystic fibrosis mutations. J. Biol. Chem. 266, 18714— 18719. Skach, W. R., Calayag, M. C, & Lingappa, V. R. (1993). Evidence for an alternate model of human P-glycoprotein structure and biogenesis. J. Biol. Chem. 268, 6903-6908. Skogerson, L. (1979). Separation and characterization of yeast elongation factors. Methods Enzymol. 60, 676-685. Smit, J. J. M., Schinkel, A. H., Oude Elferink, R. P. J., Groen, A. K., Wagenaar, E., van Deemter, L., Mol, C. A. A. M., Ottenhofer, R., van der Lugt, N. M. T, van Roon, M. A., van der Valk, M. A.,
276
CAROL BERKOWER and SUSAN MICHAELIS
Offerhaus, G. J. A., Bems, A. J. M., & Borst, P. (1993a). Homozygous disruption of the murine mdr2 P-glycoprotein gene leads to a complete absence of phospholipid from bile and to liver disease. Cell 75, 451^62. Smit, L. S., Wilkinson, D. J., Mansoura, M. K., Collins, F. S., & Dawson, D. C. (1993b). Functional roles of the nucleotide-binding folds in the activation of the cystic fibrosis transmembrane conductance regulator. Proc. Natl. Acad. Sci. USA 90, 9963-9967. Spies, T., Cerundolo, V., Colonna, M., Cresswell, P., Townsend, A., & DeMars, R. (1992). Presentation of viral antigen by MHC class I molecules is dependent on a putative peptide transporter heterodimer. Nature (London) 355, 644—646. Sprague, G. F. & Thomer, J. (1992). Pheromone response and signal transduction during the mating process of Saccharomyces cerevisiae. In: The Molecular and Cellular Biology of the Yeast Sacchawmyces (Jones, E. W., Pringle, J. R., & Broach, J. R., eds.), pp. 657-744. Cold Spring Harbor Laboratory Press, Plainview, New York. St-Pierre, M. V., Ruetz, S., Epstein, L. F., Gros, R, & Arias, I. M. (1994). ATP-dependent transport of organic anions in secretory vesicles of Saccharomyces cerevisiae. Proc. Nad. Acad. Sci. USA 91, 9476-9479. Suh, W.-K., Cohen-Doyle, M. F., Fruh, K., Wang, K., Peterson, P A., & Williams, D. B. (1994). Interaction of MHC class I molecules with the transporter associated with antigen processing. Science 264, 1322-1326. Swartzman, E. E., Viswanathan, M. N., & Thomer, J. (1996). The PALI gene product is a peroxisomal ATP-binding cassette transporter in the yearst Saccharomyces cerevisiae. J. Cell Biol. 132, 549-563. Szczypka, M. S., Wemmie, J. A., Moye-Rowley, W. S., & Thiele, D. J. (1994). A yeast metal resistance protein similar to human cystic fibrosis transmembrane conductance regulator (CFTR) and multidrug resistance-associated protein. J. Biol. Chem. 269, 22853-22857. Teem, J. L., Berger, H. A., Ostedgaard, L. S., Rich, D. P, Tsui, L.-C, & Welsh, M. J. (1993). Identification of revertants for the cystic fibrosis F508 mutation using STE6-CFTR chimeras in yeast. Cell 73, 335-346. Triana, F. J., Nierhaus, K. H., Ziehler, J., & Chakraburtty, K. (1994). Defining the function of EF-3 from low fungi. In: The Translational Apparatus (Nierhaus, K. H., ed.), pp. 000-000. Elsevier, New York. Turi, T. G. & Rose, J. K. (1994). Characterization of a novel S. pombe multidrug resistance transporter conferring brefeldin A resistance. Unpublished. Valle, D. «& Gartner, J. (1993). Penetrating the peroxisome. Nature (London) 361, 682-683. Valverde, M. A., Diaz, M., Sepulveda, F. V., Gill, D. R., Hyde, S. C, & Higgins, C. F. (1992). Volume-regulated chloride channels associated with the human multidrug-resistance P-glycoprotein. Nature (London) 355, 830-^33. Vazquez de Aldana, C. R., Marton, M. J., & Hinnebusch, A. G. (1995). GCN20, a novel ABC protein, and GCNl reside in a complex that mediates activation of the eIF-2a kinase GCN2 in amino acid-starved cells. Submitted. Volkman, S. K., Cowman, A. F., & Wirth, D. F. (1995). Functional complementation of the ste6 gene of Saccharomyces cerevisiae with ihtpfindrl gene of Plasmodium falciparum. PNAS in press. Walker, J. E., Saraste, M., Runswick, M. J., & Gay, N. J. (1982). Distantly related sequences in the aand p-subunits of ARP synthase, myosin, kinases and other ATP-requiring enzymes and a common ' nucleotide binding fold. EMBO J. 1, 945-951. Wang, R., Seror, S. J., Blight, M., Pratt, J. M., Broome-Smith, J. K., & Holland, I. B. (1991). Analysis of the membrane organization of an Escherichia coli protein translocator, HlyB, a member of a large family of prokaryote and eukaryote surface transport proteins. J. Mol. Biol. 217, 441—454. Ward, C. L. & Kopito, R. R. (1994). Intracellular turnover of cystic fibrosis transmembrane conductance regulator. J. Biol. Chem. 269, 25710-25718.
ABC Proteins in Yeast
277
Welsh, M. J. & Smith, A. E. (1993). Molecular mechanisms of CFTR chloride channel dysfunction in cystic fibrosis. Cell 73, 1251-1254. Wiemann, S., Voss, H., Schwager, C, Rupp, T., Stegemann, J., Zimmermann, J., Grothues, D., Sensen, C, Erfle, H., Hewitt, N., Banrevi, A., & Ansorge, W. (1993). Sequencing and analysis of 51.6 kilobases on the left arm of chromosome XI from Saccharomyces cerevisiae reveals 23 open reading frames including the FASl gene. Yeast 9, 1343-1348. Wilson, K. L. & Herskowitz, I. (1984). Negative regulation of STE6 gene expression by the a2 product of Saccharomyces cerevisiae. Mol. Cell. Biol. 4, 2420-2427. Wilson, K. L. & Herskowitz, I. (1986). Sequences upstream of the STE6gQne required for its expression and regulation by the mating type locus in Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. USA 83, 2536-2540. Wilson, K. L. & Herskowitz, 1.(1987). STEI6, a new gene required for pheromone production by a cells of Saccharomyces cerevisiae. Genetics 115, 441—449. Zhang, J.-T., Duthie, M., & Ling, V. (1993). Membrane topology of the N-terminal half of the hamster P-glycoprotein molecule. J. Biol. Chem. 268, 15101-15110. Zhang, J.-T. & Ling, V. (1991). Study of membrane orientation and glycosylated extracellular loops of mouse P-glycoprotein by in vitro translation. J. Biol. Chem. 266, 18224—18232. Zhang, J.-T. & Ling, V. (1993). Membrane orientation of transmembrane segments 11 and 12 of MDRand non-MDR-associated P-glycoproteins. Biochim. et Biophys. Acta 1153, 191—202.
This Page Intentionally Left Blank
MEMBRANE PROTEIN TRANSPORT IN EUKARYOTIC SECRETION CELLS
Kaarin K. Goncz and Stephen S. Rothman
I. II. III. IV. V. VI. VII. VIII.
Introduction Secretion in Pancreatic Acinar Cells Permeability ofthe Zymogen Granule Membrane to Protein The Permeability ofthe Zymogen Granule Membrane and X-ray Microscopy Mechanism ofMembrane Protein Transport The Contribution ofMembrane Transport to Protein Secretion By Acinar Cells The Formation and Fining of Zymogen Granules Final Comments References
279 281 282 283 286 288 289 291 291
I. INTRODUCTION As the many articles in this series attest, we are beginning to realize that the transport of protein molecules through biological membranes is a ubiquitous process. It occurs for many different proteins, ranging widely in structure and function, across many membranes and in virtually all cell types. Indeed, membrane
Membrane Protein Transport Volume 3, pages 279-293. Copyright © 1996 by JAI Press Inc. All rights of reproduction in any form reserved. ISBN: 1-55938-989-3 279
280
KAARIN K. GONCZ and STEPHEN S. ROTHMAN
protein transport is critical to the organization and viability of biological cells. Nonetheless, in spite of this understanding and seemingly without sufficient explanation, there has been a great deal of resistance to the idea that such processes exist in at least one particular case, the secretion of proteins by eukaryotic cells. For these events, which include secretion of a wide variety of molecular species including hormones, structural proteins, and enzymes, it is widely believed that membrane transport does not occur and that the membranes involved, such as those enclosing secretion granules or the plasma membrane itself, are impermeable to secreted proteins; that is, they do not contain mechanisms for protein transport. In the past, the simple assumption that, as a matter of principle, the transport of individual proteins across bio-membranes was impossible formed the basis for this conclusion. But today, such an a priori position is untenable given the number and variety of membrane transport processes that have been identified for proteins. Granted, the majority of these processes have been found in prokaryotes or for intracellular transport in eukaryotic cells, such as across nuclear and mitochondrial membranes, and not for secretion in eukaryotes, but then this is where they have been sought. To our knowledge there is nothing in the fundamental nature of secreted proteins in eukaryotes, or in the character of the membranes involved in their secretion, that suggests significant physical or chemical differences from the proteins and membranes involved in these other processes. Indeed, given the presence of membrane protein transport for secretion in prokaryotes and for intracellular transport in eukaryotes, its absence in secretion by eukaryotes would represent a special case, one in which evolutionary forces led to the loss of a prior mechanism. That is, one would have to suppose that membrane protein transport was deleted during the evolution of secretion in eukaryotes by a specific selective pressure against such processes, replacing them with a different, far more complex, and apparently nonselective mechanism. This raises the second reason why membrane protein transport has not been considered for secretion in eukaryotic cells; another mechanism, vesicle transport, is thought to be responsible. Rather than crossing membranes individually, secretory proteins are thought to be carried en masse from compartment to compartment within cells in small, 40-50-nm-diameter, membrane-bound vesicles, from the endoplasmic reticulum, where they are synthesized, to the Golgi network, through its various stacks, where they are further modified, and from the Golgi, to membrane-bound storage vesicles, called secretion granules, where they are held until the cell is stimulated to secrete. Finally, the product is released from the cell by exocytosis. In exocytosis, the membrane of the secretion granule fuses with the cell membrane, followed by the fission of both membranes, which in turn allows for the release of product into the extracellular milieu (Palade, 1975; Beaudoin and Grondin, 1992). Although there is a wide range of qualitative evidence that supports different elements in this vesicle pathway for secretion, for the majority of secretory products and cells, most aspects of this theory remain a matter for speculation. Direct, no
Membrane Protein Transport in Eul<aryotic Secretion Cells
281
less quantitative evidence for their occurrence has yet to be obtained, even after many years of extensive study. As such, despite popular opinion, it cannot be assumed that vesicle protein transport alone accounts for secretion. For this reason, alternative mechanisms, in particular membrane transport processes, must be considered.
II. SECRETION IN PANCREATIC ACINAR CELLS In fact, a body of evidence already exists that indicates that membrane transport plays a role in the secretion of protein by eukaryotic cells. For example, the secretion of the pheromone peptide, a-factor, occurs by a nonstandard (that is, nonvesicular) route in yeast (McGrath and Varshavsky, 1989). Membrane protein transport in this case and for analogous processes in other cells seems to involve a particular type of pore transporter, in particular, homologous membrane proteins of the ATP-binding cassette (ABC) superfamily, like P-glycoprotein (Bradley et al., 1988). However, the most substantial evidence for the membrane transport of protein in secretion by eukaryotic cells is for the secretion of digestive enzymes by the acinar cells of the mammalian pancreas. Ironically, it is this same cell in which much of the evidence for the vesicle theory of secretion was originally adduced (as above). These cells are extremely efficient protein factories that secrete the approximately 20 different enzymes that digest the food we eat in the small intestine. The idea that these proteins are directly and individually transported across membranes was proposed almost 30 years ago by one of us (Rothman, 1967) as an explanation for an observation on protein secretion by this gland. Different molecular species (different digestive enzymes) stored in the same secretion granule (the zymogen granule) were, under particular circumstances, secreted simultaneously at different rates. The vesicle model, as envisioned at that time, could not account for this because each secretion granule was thought to contain a mixture of all of the digestive enzymes whose release, it was thought, occurred en masse in a random fashion. As such, it was not possible for the rates of secretion of different proteins to vary relative to each other in a reproducible manner. Nonetheless, in spite of this observation and many other similar ones, and with only indirect and qualitative evidence for most of the vesicular events, the vesicle transport system outlined above has remained the operant paradigm, viewed widely as the only mechanism for protein secretion from these and all other eukaryotic cells. Although the original vesicle model has undergone a number of revisions to explain discordant observations, such as the idea that enzymes are segregated into specific granules in the pancreas to allow for the secretion of different proteins at different rates (Adelson and Miller, 1989), this, as well as other auxiliary and ad hoc hypotheses that have been proposed to account for numerous anomalous experimental observations, remain, as a rule, either untested or unproven. Over the years, results from a range of experiments have strengthened the view that membrane transport occurs in these cells. Much of the evidence fits into four
282
KAARIN K. GONCZ and STEPHEN S. ROTHMAN
broad categories: 1) the independent secretion of various proteins that are stored in common secretion granules (called nonparallel transport); 2) the existence of a cytoplasmic pool of secreted protein; 3) the permeability of the cell membrane to secreted protein; and finally 4) the permeability of the membrane of the secretion granule to its contained proteins. This evidence is discussed in a variety of places (Rothman, 1975a, 1985; Rothman and Ho, 1985; Rothman and Liebow, 1985; Rothman et al., 1991), including a recent general review on the transport of proteins across membranes (Isenman et al., 1995). Here we will focus on the last of these lines of evidence, the transport of proteins across the membrane of the secretion or zymogen granule.
III. PERMEABILITY OF THE ZYMOGEN GRANULE MEMBRANE TO PROTEIN The first direct evidence to demonstrate that individual protein molecules are transported across membranes appears to be the report by Liebow and one of us (SSR) in Nature (1972). In this paper, evidence for the permeability of the zymogen granule membrane to the proteins it contains was presented. Granules were isolated and suspended in different volumes of isologous media. Over time, digestive enzymes were detected in the suspending solution, indicating that protein had been released from the granules. By increasing the volume of the solution, the amount of enzyme released was also increased. In this simple way, it was possible to vary the amount of protein in the granules and medium by over two orders of magnitude, from virtually all protein being contained in the granules to almost all being released into the suspending medium. The proteins did not appear to be held in place by an absolute membrane barrier. Moreover, release was not haphazard, as might be expected if it was due to damaged membranes. Rather, protein release was controlled by and dependent upon the concentration of these proteins in the suspending medium. Further evidence for membrane protein transport came from other experiments on isolated granules. For example, the reuptake of protein was demonstrated subsequent to its release (Liebow and Rothman, 1972, 1976). In addition, differences in the rate of release of different molecular species were observed when granules were suspended in different solutions, including those containing various digestive end products. For example, when the suspending solution contained amino acids or glucose analogs, protein release was found to be selective (Grendell and Rothman, 1981; Niederau et al., 1986). To explain these observations, granule proteins had to move independently into and out of the granules. Nonetheless, and in spite this evidence, it was commonly considered that membrane transport did not occur and the results were attributed to the rupture of granule membrane, with the resulting solubilization of its protein contents. Such "granule lysis" has been thought to be the cause of protein release by these secretion granules under any and all circumstances. It is presumed to occur even under conditions
Membrane Protein Transport in Eukaryotic Secretion Cells
283
where there is no clear reason to expect lytic rupture, as when granules are suspended in "physiological" solutions such as at neutral pH, in isotonic ionic media, or in solutions that contain low concentrations of various amino acids and glucose analogs. Moreover, lysis has been proffered as the cause of protein release without explicit evidence of its occurrence, this conclusion being based on the supposition that release could only occur in this fashion. It was the need to resolve this question of granule lysis that led to the following experiments. In these studies, x-ray microscopy was used to observe isolated zymogen granules suspended in aqueous solution, at high resolution (-50 nm). The observation of individual granules over time as they released protein and the simultaneous quantitative measure of the amount of protein in each granule were possible using this method. We hoped to determine whether protein release was an all-or-none phenomenon, as the lytic hypothesis proposed, or gradual, as predicted by the membrane transport model. In the former case, protein release would be accompanied by a reduction in the total number of granules in suspension, whereas in the latter, each granule would lose a fraction of its contents while otherwise maintaining its character and structure.
IV. THE PERMEABILITY OF THE ZYMOGEN GRANULE MEMBRANE AND X-RAY MICROSCOPY In these experiments, high-resolution digital images of isolated zymogen granules were obtained with a scanning transmission x-ray microscope (Jacobsen et al., 1991). Using a specially designed sample chamber (Goncz et al., 1991), granules were suspended in one of a variety of aqueous solutions that were in the microscope. The solution could be changed at any point during an experiment. During an exposure, a field of granules was scanned by a small spot (~50 nm) of focused X-rays, and the number of photons transmitted through the sample at each point, or pixel, was detected and recorded digitally. The X-rays in the energy range used (2.3-^.4 nm) are absorbed more by protein (carbon) than water by a factor of 10. As a consequence of this natural difference in absorption, it is possible to observe the structure of hydrated biological material without the use of stains or other artificial enhancers. These x-ray images were therefore able to show, for the first time at such high resolution, the internal structure and morphology of zymogen granules in a natural state. A typical field of granules, imaged over a period of 5 hours, is shown in Figure 1. In these images, each granule can be readily identified in all of the frames. The amount of protein within each granule decreased over time, although at different rates. These observations, though seemingly simple, were among our most exciting, demonstrating, as they did, that the release of protein from the granules was not the result of lysis. From the same images, changes in granule structure were appraised as protein was released. Little in the way of morphologic change was seen, except that granules losing content also tended to decrease in size. Importantly, such
284
KAARIN K. GONCZ and STEPHEN S. ROTHMAN
Figure /. X-ray images of a field of zymogen granules suspended in an aqueous environment observed over a period of 5 hours at approximately 50 nm resolution (1-5). Two types of granules are seen in this figure: nonuniform (NUG) and uniform (UG), or in traditional terminology, condensing vacuoles and zymogen granules. Uniform granules are those whose contents are uniformally dense; that is, they do not contain obvious lucent regions. The remainder, nonuniform granules, in unusually high abundance in this field, contain regions of variable density, from highly dense areas to quite lucent ones. Note the characteristic cap structures of many of the NUG.
changes were observed for all granules, although, as noted above, occurring to varying degrees in different objects. These observations also confirmed previous in situ results in which granule shrinkage had been observed; granules become smaller during active secretion (Ermak and Rothman, 1981). A quantitative analysis of these observations provided additional support for the view that granules were permeable to the proteins they contained. This analysis also opened the door to an understanding of the mechanism of protein transport and the role such processes play in naturally occurring secretion. Information concerning the diameter of each granule and its protein contents was determined from the digital information stored in the x-ray image (Goncz, 1994). Average values for these two parameters, as well as granule protein concentration (granule protein content divided by volume), for a population of granules is given in Table 1. The granules are separated into two types based on their appearance in x-ray images. They are referred to as uniform and nonuniform granules, or UG and NUG, respectively. In traditional terms, they would be referred to as zymogen granules
Membrane Protein Transport in Eukaryotic Secretion Cells
285
Table 1. Average Values (± SEM) for Three Parameters from a Population of 388 Zymogen Granules Granule UG {n = 300) NUG {n = 88) Total {n = 388) Note:
Diameter (\im)
Protein concentration (mg/ml)
Protein content (fg)
1.00 ±0.01 1.36 + 0.04 1.08 ±0.02
294 ±17 115± 6 253 ± 7
153 ± 5 142 ± 10 150 ± 4
The granules have been divided into two subclasses, uniform and non-uniform, as described in the legend of Figure I.
and condensing vacuoles (the precursor form of zymogen granules). The reason for not using this latter terminology will become clear later. Quantitative analysis v^as performed for granules in several different experiments under conditions similar to those of the earlier experiments of Liebow and Rothman (1972), as well as other observations (Hokin, 1955; Rothman, 1971; Burwen and Rothman, 1972; Liebow and Rothman, 1976) in which protein release or uptake had been observed. These conditions included the suspension of granules in water, as well as in non-ionic solutions that were iso- or hyperosmotic (0.3 and 0.6 M sucrose). Isosmotic solutions at different pH (5.0-7.5) were also used, as were solutions that contained different concentrations of the digestive enzyme chymotrypsinogen. In other experiments, granules were exposed to low concentrations of detergent or were suspended in isotonic ionic solutions. Granules were generally imaged at approximately hourly intervals over time periods of up to 5 hours. The results from only a few of these experiments will be mentioned here. The others are discussed elsewhere (Goncz and Rothman, 1992; Goncz et al., 1992; Goncz, 1994). That protein release is the result of simple mass action is demonstrated by the results of the following experiments. When granules were suspended in isosmotic solution, the rate at which protein was released decreased over time and approached zero. This meant that a steady state had been achieved in that there was no net protein movement between the two compartments (granule and solution). When this suspending solution was replaced with fresh medium to remove protein that had accumulated, thereby increasing the protein gradient, protein release was seen again. Finally, the addition of protein (chymotrypsinogen) to the suspending solution led to its uptake by the granules. In addition to determining the rate of protein release for each granule, a membrane permeability coefficient to protein (Pg) was also calculated. This value is widely used to characterize the ease of transport of different molecules across membranes. The coefficient for each granule is calculated from the x-ray data and Pick's first law of diffusion:
286
KAARIN K. GONCZ and STEPHEN S. ROTHMAN
J/A=P^Aiq„-C,J,
(1)
where J is protein efflux, A is the area of the granule membrane, and C-^ and C^^^ are the concentrations of soluble protein inside and outside of the membrane, respectively. Protein efflux (J) is calculated as the difference in protein content of a granule between the first two images, divided by the amount of elapsed time and can be considered the initial rate of release. The area of the granule membrane (A) is calculated from its diameter as measured in the initial image. These calculations were performed on data from granules in which the suspending solution was continuously changed. As such, the concentration of soluble protein outside the granules (Q^^) was zero or close to zero. The concentration of soluble protein inside the granule (Cjn) was set at 75 |Lig/ml according to results from other studies on isolated granules (Liebow and Rothman, 1976). An average value for P^ of 2.8 ± 0.3 x 10"^ cm/sec (« = 51) was calculated. The magnitude of this value gives some indication of the type of mechanism that could be involved in transport. For example, it is unlikely that transport occurs as a result of the simple "solubilization" of folded protein in the lipid phase of the membrane; a much smaller value for P would be expected (<10~^^ cm/sec). It is also unlikely that transport is completely unhindered, as would be the case if free diffusion occurred, due, for example, to the absence of a membrane barrier or the presence of very large aqueous channels (Pg > 10~^ cm/sec). As such, the results suggest a specific transport mechanism in the granule membrane that is responsive to mass action but that restricts diffusion.
V. MECHANISM OF MEMBRANE PROTEIN TRANSPORT With this understanding, we began efforts to identify a mechanism of transport that fits this description. The simplest possibility was a membrane pore. Such pore mechanisms have been identified in other systems for the transport of a variety of proteins (Isenman et al., 1995). There was evidence that this might also be the case for the zymogen granule. Freeze-fracture electron microscopic images of zymogen granule membrane showed large, presumably proteinaceous, pore-like structures, with a 5-nm-diameter lucent center, that are present at a number density of 26/iim^ memt^rane area (Cabana et al., 1988). A structure of this size would be able to accommodate all of the secreted proteins without the need for unfolding, the largest protein being just small enough to pass through the pore. We considered the possibility that protein transport occurred via this structure by calculating the theoretical permeability coefficient (P^) for a membrane that contains these pores in the observed numbers and comparing it to the experimental value (Pg). Briefly, the coefficient was calculated for each of the approximately 20 different digestive enzymes. Each value was then weighted according to the proportion of that protein present in the pancreas (Schick et al., 1984). The final
Membrane Protein Transport in Eul<aryotic Secretion Cells
287
value of Pj was an average of these weighted values. A more detailed discussion of this calculation can be found elsewhere (Goncz and Rothman, 1995). Pj for granules suspended in distilled water is shown in Table 2 alongside the value calculated for P as described above. The values matched extremely well, and as such provide strong support for the hypothesis that transport occurs by diffusion through a pore with the characteristics of the one found by Cabana et al. (1988). That is, using realistic numbers for pore size and number, in addition to protein size, number, and proportions, a theoretical estimate of permeability matched that actually observed. The idea that passage occurs by diffusion through a water-filled channel was tested further. For this to be true, a change in the viscosity of the suspending medium, that affects the rate of diffusion, should also affect the permeability coefficient. Specifically, an increase in viscosity will lower the permeability coefficient. As such, for a solution with a high viscosity (0.6 M sucrose) the value of P should be lower than for water (Table 2). This was indeed the case, and Pg was reduced to 1.5 x 10"^ cm/sec (Table 2). How does this value compare to that predicted by the pore model? Table 2 shows that they are essentially the same; that is, the theoretical value and experimental observations were again well matched. Thus, the data warrant the conclusion that the structure seen in freeze-fracture micrographs is the transporter and that it is a water-filled channel that carries these proteins. These results also provide additional support for the conclusion that any damage that the granules may have incurred as a result of x-ray absorption did not alter in any substantive way, the natural permeability of these objects to protein. Thus, it appears that transport of all of the different proteins occurs through one type of pore. As a result, we are left with the question of how proteins can be released in a selective manner by such a mechanism, as is known to occur. In spite of the fact that all proteins travel through a single structure, selectivity could still be attributed to the pore transporter, because none of the proteins pass freely and diffusion is differentially restricted. Small variations in either the structure of the protein or the bore of the channel would alter their relative rates of transport, even
Table 2, A Comparison of Predicted and Measured Permeability Coefficients for the Transport of Proteins Across the Membrane of Pancreatic Zymogen Granules Suspended in Solutions of Different Viscosity Suspending solution
Permeability coefficient (cm/sec) Measured (P^
Distilled water (pH 6.0) 0.6 M sucrose (pH 6.0) Note:
2.8 ± 0.3 X 10"^ (51) 1.5±0.3x 10"^(11)
Predicted (PJ 2.7 ± 0.5 X 10"^ 1.4±0.3x 10"^
The errors shown for the predicted values were determined from the variability in the pore diameter measured by Cabana et al. Measured values are given as average ± SEM (number of granules).
288
KAARIN K. GONCZ and STEPHEN S. ROTHMAN
if no selective chemical events occurred during passage, which of course they might. In addition, almost all of the granule's protein contents are in an aggregated form. By changing environmental circumstances, the binding constant or partition coefficient for each protein would vary independently. In this way their concentration in solution within the granule, that concentration presented to the transporter, would also vary independently. Control at this level would also lead to the differential release of the different proteins.
VI. THE CONTRIBUTION OF MEMBRANE TRANSPORT TO PROTEIN SECRETION BY ACINAR CELLS Given that digestive enzymes are transported across granule membranes, two questions remain concerning the role of this process in secretion by the pancreas. The first is: What additional evidence is there that secretion occurs by a pathway from the secretion granule, through the cytoplasm, and across the cell's plasma membrane? Here are four examples. First, the independent secretion of different molecular species has been demonstrated in situ. Because all of these proteins reside in common secretion granules (that is, not in vesicles that contain single protein species) their independent secretion requires membrane transport (Rothman et al., 1991). Second, regarding the presence of a cytoplasmic pool of secretory protein, secretion has been found to occur by the transfer of material between granules and extracellular fluid through a third component, the cytoplasm, interposed between them (Rothman and Isenman, 1974). Third, regarding the permeability of the plasma membrane, labeled chymotrypsinogen is taken up by pancreatic tissue, through a soluble compartment, en route to the zymogen granule (Liebow and Rothman, 1974). The specific radioactivity of this protein in the secretion granule equilibrates with that in the medium, indicating the complete exchange of protein between granule and medium, as predicted by a membrane transport mechanism. Finally, it is known that secretion from the pancreas continues at highly augmented rates even after the degranulation of cells, that is, when secretion granules are no longer present (Rothman, 1975b). These and other examples, when viewed alongside evidence for the permeability of the granule membrane, represent substantial experimental support for a nonvesicular route of secretion. Thus, not only does membrane transport exist for proteins in the general sense that this treatise demonstrates, but its existence includes membrane transport in eukaryotic secretion cells that have traditionally been thought to secrete their products by means of vesicle transport and such transport alone. But accepting this, of course, does not mean that membrane transport represents a quantitatively significant element in protein secretion. This brings us to the second question, whether the rate of membrane protein transport is sufficient to account for the amount of secretion. From x-ray microscopy measurements of protein efflux, we can estimate the contribution of membrane transport to protein secretion, assuming that granule stores are the major source of secreted protein and that the rate at which
Membrane Protein Transport in Eukaryotic Secretion Cells
289
protein leaves the granule across its membrane represents the maximum contribution of a membrane transport pathway to protein secretion overall. According to the data, the net efflux of protein from the granule is maximally about 30% of granule protein contents per hour, which corresponds to a loss of 46 fg/hr for an average zymogen granule (Table 1). If we estimate the number of granules per cell and the number of cells in a particular gland, then we can predict the maximum rate of secretion attributable to this source, given that this process occurs at the same rate in situ and that this step in the transport sequence is rate limiting. After an overnight fast, secretion granules occupy between 15% and 25% of the cell volume in adult rats. Given that an average granule is 1 |Lim in diameter and that the cell diameter is 10 |Lim, these cells would contain between 150 and 250 granules each on average. There are, on average, 1.5 x 10^ cells in the rat pancreas. Thus, the product of the secretory capacity of a single cell and the number of cells in the gland provides an estimate of the potential contribution of membrane transport processes to secretion. When this calculation is performed, a value of 10-17 mg of protein secreted per hour is obtained. How does this compare to actual rates of protein secretion from rat pancreas in situ? Protein output from fasted anesthetized rats without exogenous stimulation is approximately 1.0 mg/hr. It is about 10 times this rate, 10 mg/hr, for secretion in a highly stimulated state (Rothman, 1975b). Thus, the potential exists for the membrane transport or nonvesicular pathway to account fully for protein secretion by this gland. Several lines of in situ evidence suggest that nonvesicular mechanisms contribute substantially to protein secretion—^at least 25%, and perhaps much more. This having been said, this result cannot be taken as evidence against vesicle transport any more than qualitative evidence of vesicle transport excludes the possibility of membrane transport. Indeed, it is likely that both mechanisms play a role in protein secretion. Certainly the whole of the published literature on the subject supports such a conclusion.
VII. THE FORMATION AND FILLING OF ZYMOGEN GRANULES The question of whether protein secretion occurs by membrane or vesicle transport or both raises questions about the formation and filling of zymogen granules. The traditional view, which is both a source of the vesicle theory and evidence in its support, is that mature zymogen granules (ZGs) develop from precursor granules, condensing vacuoles (CVs). CVs are thought to bud from Golgi membranes as empty or almost empty structures. Secretory protein is added subsequently as membrane-bound micro vesicles carry product from the trans-Golgi network to the filling CV. When the granule is filled, it undergoes a metamorphosis to its mature form. As a result of this process, the irregularly shaped, although still roughly spherical CV, whose heterogeneous structure usually contains variable quantities and forms of electron dense clumps and lucent spaces, becomes homogeneous,
290
KAARIN K. GONCZ and STEPHEN S. ROTHMAN
dense, tightly spherical, and lacks obvious lucent regions. This is thought to be the result of protein condensation and the extrusion of water from the vesicle. As a result, mature granules contain a dense, homogeneous mass of aggregated-protein. Once this transformation occurs, the granule is considered to be full and is no longer able to take on additional material. This hypothesis provides an explanation for the formation and filling of zymogen granules from the standpoint of the vesicle theory of secretion. When one considers the secretion granule from the point of view of membrane transport, the results are different. Formation and filling are not necessarily congruent events, as in the vesicle model, nor does each granule fill irreversibly to reach a maximum capacity. This is because granules are not formed anew with every secretion cycle. They act as storage capacitors, variably collecting secretory protein or discharging their load, depending upon the extant physiological circumstances (such as the stimulation of secretion). In this model, the protein contents, and as it turns out the size of granules, vary with ftinctional state. Granules lose contents and become smaller during active protein secretion, and take on protein and increase in size during the recovery phase. As a result, for the majority of granules there is no need for an immature or precursor form. Granules simply exist at variable levels of filling, depending upon the physiological circumstances. A similar idea has been proposed in a vesicle transport context for the secretion of acetylcholine at the cholinergic synapse. What is the evidence for this perspective? Or does the evidence indicate granule formation and filling as the vesicle theory predicts? The evidence that CVs are immature ZGs is both morphological and biochemical. Morphologically, as already noted, condensing vacuoles most often appear to be only partially filled or even almost wholly empty in electron micrographs, compared to the homogeneously dense ZG. CVs are also found in greatest abundance near the Golgi region of the cell, where they are thought to be formed. The biochemical evidence comes from a by now standard experiment in which protein is pulse labeled with radioactive amino acids, chased by the addition of nonradioactive amino acids, and then followed during its travels through and out of the cell by cell fractionation or autoradiography. From such measurements, it was found that labeled protein first accumulated over CVs and only subsequently over ZGs, a sequence consistent with the conclusion that CVs are precursor forms of the ZG. We were able to test two simple predictions of the precursor model of granule filling using the data from the x-ray microscopy experiments discussed above. First, according to this model, although CVs should contain a range of protein contents, they should on average contain less protein than ZGs because they are ZGs in the process of being filled. Second, having been filled, ZGs should contain protein at a common concentration, set by the condensation process. Neither prediction was supported by the data (Goncz et al., 1995). With regard to the former prediction, CVs actually contained as much protein as ZGs on average (Table 1). The latter prediction was also found to be incorrect. Protein concentration varied over a
Membrane Protein Transport in Eul<aryotic Secretion Cells
291
sixfold range for ZGs and an order of magnitude for CVs. Such a range of values is predicted by the membrane transport model. In addition, another piece of evidence offers support for the notion that granules are storage capacitors that are capable of variable filling. It was discovered by Ermak and Rothman (1983) and confirmed independently by Sjostrand (1985, 1990) that in the fetus, the formation of zymogen granules in the cells of the pancreas does not appear to take place as predicted by the precursor model. Two things are observed. First, the number and location of condensing vacuoles remain relatively constant during these events, appearing much as in adult animals. Second, as zymogen granules increase in number over time, they become larger and store additional material, reaching sizes far larger than normally seen after birth. That is, granules already formed, according to classical notions, appear to continue to take up protein. Moreover, these filling granules, located at first only at the cell's apex, are increasingly found throughout the cytoplasm, including basolateral regions of the cell, as filling progresses and their number and size increase (to levels that occupy close to 50% of cell volume). These observations support the idea that condensing vacuoles are not precursors to zymogen granules, but are rather another, different and somewhat larger object that also stores protein but at a lower concentration. It is not uncommon when such morphologically distinct vesicles are seen in cells that different ftinctions and contents are ascribed to them, for example, in nerve and pituitary cells, as opposed to a precursor/product relationship (Moore et al., 1988).
VIII. FINAL COMMENTS The evidence we have discussed has led us to the following conclusions: 1) the membrane of the zymogen granule is permeable to its contained proteins, as has been proposed for many years; 2) if granule lysis occurs, it does not account for the examples of protein release from zymogen granules that we have studied; 3) the mechanism of protein transport into and out of these objects appears to involve diffusion through a membrane channel of about 5 nm diameter, 4) this channel carries all of the digestive enzymes that are secreted; 5) the rate of transport across this membrane is consonant with the rate of protein secretion by acinar cells; and 6) the notion that condensing vacuoles are "immature" precursors of "mature" zymogen granules is not borne out by the evidence.
REFERENCES Adelson, J. W. & Miller, P. E. (1989). Heterogeneity of the exocrine pancreas. Am. J. Physiol. 256, G817-G825. Beaudoin, A. R. & Grondin, G. (1992). Zymogen granules of the pancreas and the partoid gland and their role in secretion. Int. Rev. Cytol. 132, 177-222. Bradley, G., Juranka, P. F., & Ling, V. (1988). Mechanism of multidrug resistance. Biochim. Biophys. Acta 948, 87-128.
292
KAARIN K. GONCZ and STEPHEN S. ROTHMAN
Burwen, S. J. & Rothman, S. S. (1972). Zymogen granules: osmotic properties, interactions with ions, and some structural implications. Am. J. Physiol. 222, 1177—1181. Cabana, C, Magny, P., Nadeau, D., Grondin, G., & Beaudoin, A. (1988). Freeze-fracture study of the zymogen granule membrane of pancreas: two novel types of intramembrane particles. Eur. J. Cell Biol. 45, 51-66. Ermak, T. H. & Rothman, S. S. (1981). Zymogen granules of pancreas decrease in size in response to feeding. Cell Tissue Res. 214, 51-66. Ermak, T. H. & Rothman, S. S. (1983). Increase in zymogen granule volume accounts for increase in volume density during prenatal development of pancreas. Anat. Rec. 207,487-507. Goncz, K. K. (1994). A comprehensive study of the physical properties of isolated zymogen granules using scanning transmission X-ray microscopy. Ph.D. thesis. University of California, Berkeley. Goncz, K. K. & Rothman, S. S. (1992). Protein flux across the membrane of single secretion granules. Biochim. Biophys. Acta 1109, 7-16. Goncz, K. K. & Rothman, S. S. (1995). A trans-membrane pore can account for protein movement across zymogen granule membranes. Biochim. Biophys. Acta. 1238, 91-93. Goncz, K. K., Batson, R, Ciarlo, D., Loo, B. W., Jr., & Rothman, S. S. (1991). An environmental sample chamber for the X-ray microscope. J. Microsc. 168, 101-110. Goncz, K. K., Moronne, M., Lin, W., & Rothman, S. S. (1992). Measuring changes in the mass of single subcellular organelles using X-ray microscopy. SPIE 1741, 342—350. Goncz, K. K., Behrsing, R., & Rothman, S. S. (1995). The protein content and morphogenesis of zymogen granules. Cell Tissue Res. 280, 519-530. Grendell, J. H. & Rothman, S. S. (1981). Digestive end products mobilize secretory proteins from subcellular stores in the pancreas. Am. J. Physiol. 241, G67-G73. Hokin, L. E. (1955). Isolation of the zymogen granule of dog pancreas and a study of their properties. Biochim. Biophys. Acta 18, 379-388. Isenman, L., Liebow, C, & Rothman, S. (1995). Transport of proteins across membranes. A paradigm in transition. Biochim. Biophys. Acta. 1241, 341-370. Jacobsen, C, Williams, S., Anderson, E., Browne, M. T, Buckley, C. J., Kern, D., Kirz, J., Rivers, M., & Zhang, X. (1991). Dififracuon-limited imaging in a scanning transmission X-ray microscope. Opt. Commun. 86, 351-364. Jamieson, J. D. & Palade, G. E. (1968). Intracellular transport of secretory proteins in pancreatic exocrine cell. III. Dissociation of intracellular transport from protein synthesis. J. Cell Biol. 39, 580-588. Liebow, C. & Rothman, S. S. (1972). Membrane transport of proteins. Nature 240, 176-178. Liebow, C. & Rothman, S. S. (1974). Transport of bovine chymotrypsinogen into rabbit pancreatic cells. Am. J. Physiol. 226, 1077-1081. Liebow, C. & Rothman, S. S. (1976). Equilibration of pancreatic digestive enzymes across zymogen granule membranes. Biochim. Biophys. Acta 455, 214—253. McGrath, J. P. & Varshavasky, A. (1989). The yeast STE6 gene encodes a homologue of the mammalian multidrug resistance P-glycoprotein. Nature 340, 400-404. Moore, H.-R, Orci, L., & Oster, G. F. (1988). Biogenesis of secretory vesicles. In: Protein Transfer and Organelle Biogenesis (Das, R. C. & Robbins, P. W., eds.), pp. 521-561. Academic Press, New York. Niederau, C, Grendel, J. H., & Rothman, S. S. (1986). Characteristics of rat pancreatic zymogen granules prepared by different methods. Am. J. Physiol. 215, G421-G429. Palade, G. E. (1975). Intracellular aspects of the process of protein synthesis. Science 189, 347-358. Rothman, S. S. (1967). "Non-parallel transport" of enzyme protein by the pancreas. Nature 213, 215-218. Rothman, S. S. (1971). The behavior of isolated granules: pH dependent release and reassociation of protein. Biochim. Biophys. Acta 241, 567-577. Rothman, S. S. (1975a). Protein transport by the pancreas. Science 190, 747-753.
Membrane Protein Transport in Eul<aryotic Secretion Cells
293
Rothman, S. S. (1975b). Enzyme secretion in the absence of zymogen granules. Am. J. Physiol. 228, 1828-1834. Rothman, S. S. (1985). Protein Secretion: A Critical Analysis of the Vesicle Model. John Wiley and Sons, New York. Rothman, S. S. & Ho, J. J. L., eds. (1985). Nonvesicular Transport. John Wiley and Sons, New York. Rothman, S. S. & Isenman, L. D. (1974). Secretion of digestive enzyme derived from two parallel intracellular pools. Am. J. Physiol. 226, 1082-1087. Rothman, S. S. & Liebow, C. (1985). Permeability of zymogen granule membrane to protein. Am. J. Physiol. 248, G385-G392. Rothman, S. S., Liebow, C, & Grendell, J. (1991). Non-parallel transport and mechanisms of secretion. Biochim. Biophys. Acta 1071, 159-173. Schick, J., Kern, H., & Scheele, G. (1984). Hormonal stimulation in the exocrine pancreas results in coordinate and anticoordinate regulation of protein synthesis. J. Cell. Biol. 99, 1569-1574. Sjostrand, F. S. (1985). Intracellular transport in exocrine pancreas. In: Nonvesicular Transport (Rothman, S. S. & Ho, J. J. L., eds.), pp. 275-286. John Wiley and Sons, New York. Sjostrand, F. S. (1990). Deducing Function from Structure. Academic Press, San Diego.
This Page Intentionally Left Blank
INDEX
ABC see Adenosine triphosphatebinding cassette Acinar cells, 285-286, 292-293, 295 Acyl-CoA oxidase (AOX), 183, 186, 190-198, 196, 214-216, 219223 Adenosine triphosphate (ATP), 3,4, 15, 36-38, 52, 83, 89, 93, 106, 109, 201, 234-235, 239240, 247, 257, 260, 263-266, 270-271 Adenosine triphosphate-binding cassette (ABC), 83-89, 8687, 105, 166, 167, 232-236, 233, 238-268, 252, 270-272, 285 bacterial toxin transport and, 84, 85-89, 86-87, 88,91, 94, 102103, 104, 110, 111 function of, 236-240, 237 Alkaline phosphatase (ALP), 91, 94, 126, 129, 135, 138, 140, 154, 172 ALP see Alkaline phosphatase a-helix, 52-55, 53, 58, 99-100 a-lytic protease, 166, 169, 170, 171, 172-175, 173, 174, 176, 176 Amino-terminal extensions, see Mitochondrial presequences Amphiphilicity, 53-55, 58, 73-74, 100-103, 107
ATP see Adenosine triphosphate ATPase, 109, 143, 167, 185-186, 247 Bacterial extracellular secretion, 166175, 171, 173, 174 Bacterial toxin transport, 82-111 )3-galactosidase, 91, 94, 103, 106, 108 /?-oxidation, 183, 191, 192, 194-195, 198, 214, 218, 261 Binding curve, 60, 61-62, 63-64, 64 Ca^*-binding, 104, 107-108, 110 Candida albicans, 235, 253, 267-268 CAN/NUP214, 27, 28-29, 30, 34, 4142 Carboxyl terminus, 198, 199, 216, 223 Carboxypeptidase Y (CPY), 121126, 123, 128-130, 135-139, 144, 147, 153-155 CD see Circular dichroism Center region, 237, 237, 238, 246, 255 Central plug, 7, 8-12, 10, 15-18, 16, 18, 27, 28, 34, 36, 38, 39,41 Central pore, 7, 8, 15, 18, 30 Channels, 36, 38, 94, 104, 110, 234235, 236, 239, 291, 295 Circular dichroism (CD), 55, 101, 106 Clathrin, 139, 143, 146, 153-154
295
296 Colicin V, 89, 98, 108, 234 Colloidal gold, 25-27, 28-29, 30 Complementation, 105, 200, 219, 221, 222, 223-224, 225-226, 269 Condensing vacuoles (CVs), 288, 294-295 Consensus sequence, 52-53, 74, 197, 217, 237, 238 COOH-terminal, 23, 25,26,30, 31,33, 35, 82-83, 85, 86-87, 90, 96105,100, 107,108,110, 126, 140, 147, 149, 150, 167, 175, 214-218, 222, 223, 247-248 CPY see Carboxypeptidase Y C region see Signature region Cyclolysin, 89, 98, 100, 101, 104, 167 Cystic fibrosis transmembrane conductance regulator (CFTR), 104, 232, 235, 236, 238, 239240, 246, 247, 250-251, 260 Cytoplasm, 2-4, 4, 38, 50, 71, 72, 90, 94, 108, 167, 170, 171, 172, 200, 238, 286, 292, 295 Cytoplasmic family, 5-9, 6, 7, 11-15, 13, 14, 17, 18, 27-39, 28-29, 30, 34, 39, 41, 84, 85, 104111, 127, 139, 145,146, 153, 263 Cytosol, 35, 36-38, 39, 40, 51, 71, 186, 199, 201, 214, 217, 262 Drosophila, 236-237, 266 Dynamin, 148-149, 155 E. chrysanthemi. see Erwinia chrysanthemi E. coli. see Escherichia coli Electron microscopy (EM), 5, 8, 12, 13, 14, 18, 134, 195, 220, 290, 294 see also Immunoelectron microscopy
INDEX
EM see Electron microscopy Endoplasmic reticulum (ER), 3, 4, 52, 91, 109, 120-126, 121, 152, 153, 188, 189, 197, 200, 201, 234-235, 240, 248, 269, 284 Endosomes, 121, 122-123, 123, 128, 135-136, 138, 144-145, 148, 150, 152 Enzymes, 182, 183, 190, 194-195, 198, 214, 215, 220-221 digestive, 285, 286, 291, 292, 295 Epitopes, 33-35, 34, 176, 201, 247248 ER see Endoplasmic reticulum Erwinia chrysanthemi, 90, 93, 98, 101, 106, 167, 172 Escherichia coli, 82-84, 84, 89-90, 95, 97, 98-105, 100, 107, 143, 150, 167-175, 173, 201, 218, 234, 239 Eukaryotes, 2, 84-85, 89, 93, 109, 120, 167, 171, 183, 191, 232, 264, 265 Eukaryotic secretion cells, 284-295 Fatty acids, 183, 184, 193-194, 200, 214, 218, 220 Fluorescence, 60-62, 61-64, 66, 68, 70, 74, 150, 195, 218, 221, 247-248 Fluorescence quenching, 56, 60, 6162, 66, 271 Folding, 19, 52, 72, 200-201, 272, 290 bacterial extracellular secretion and, 166, 169-172, 171, 175177, 176 Golgi complex, 138, 139-140, 152, 154, 244, 248, 270 GTP-binding and, 146, 147, 148, 150
Index
vacuolar proteins and, 122,154,155 yeast vacuole and, 120, 121, 123, 123, 125, 127, 129, 130, 135, 144,145 see also Trans-Golgi network Gram-negative bacteria, 82, 83, 166, 177, 234 Gram-negative cells, 167, 168, 171, 175 Granules, secretion, 284, 285-290, 292-294 GTP-binding, 146-151, 150-151, 152, 155 Half-molecules, 233, 236, 240, 246, 247, 261, 266 Hemolysin, 82-107, 83-88, 92, 96, 97, 100,109-111,167,233,234, 237, 239, 246-247 Hemolysin transport, 82-110, 83, 84, 86-87, 88, 92, 96, 97, 100 Hydrolases, vacuolar, 120-121, 123, 124, 129, 135, 137, 140 Immunoelectron microscopy, 27, 2829, 30, 31, 33, 217, 219, 220 see also Electron microscopy Infantile Refsum disease, 214, 220 Inner membrane, 166, 167, 176 see also Mitochondrial inner membrane Integral membrane protein, 19-24, 23, 31-33, 127, 138, 185-188, 198, 199, 226 Invertase, 124-129, 137 Lipids, 52, 62, 67, 68, 70-71, 74, 93, 95, 106, 109, 141, 184, 216, 221, 234, 235, 266, 269-270, 271 LktA see P. haemolytica leukotoxin Localization pathway, 121, 146, 154, 155
297 Lumen, 3, 4, 9, 19, 23, 24, 34, 122 Lumenal domain, 9, 10, 19, 23, 24, 127 Lysobacter enzymogenes, 172, 174, 175 Lysosomes, 120, 122-123, 124, 137, 142, 153, 154 Mass, molecular, 7-12, 8, 14, 15, 17, 18, 23, 24, 34, 35, 41, 51-52, 187 MAT cells, 244-245, 248 Mating, 232, 241-246, 248, 251, 252, 266, 268, 269 MDR see Multidrug resistance Mediated transport, 36, 38, 40-42, 41 Membrane association, 82, 104, 106, 139, 143, 147 Membrane binding, 58-65, 61-62, 6265, 63-64 Membrane fusion protein (MFP) family, 94, 167, 168 Membrane potential, 52, 65, 67, 6870, 69, 71-72, 92 Membrane spanning domain (MSD), 233, 234, 236-241, 246, 247, 252, 257, 265, 266 Membranous structures, 188, 190, 190-191 Metalloprotease, 89, 93, 98, 101, 102, 103, 106, 107 MIM see Mitochondrial inner membrane Mitochondria, 143, 182, 189, 197, 199, 200, 219, 240, 253, 262 Mitochondrial inner membrane (MIM), 51-52,55,65,67, 72, 73-74 Mitochondrial matrix, 50-52, 51, 55, 74 Mitochondrial membrane proteins, 51, 72-74
298 Mitochondrial membranes, 52, 56, 59-62,61,65,67,68, 170, 284 Mitochondrial outer membrane (MOM), 51-52, 60, 65, 66, 67, 72, 73 Mitochondrial presequences, 50-74, 201 Mitosis, 4, 5, 135, 136 Molecules, small, 3, 4, 9, 19, 36, 85 MOM see Mitochondrial outer membrane MSD see Membrane spanning domain Multidrug resistance (MDR), 85, 232, 233, 235, 236, 238-240, 244, 247, 248, 257-258, 260263,266,268-269,268-271 NE see Nuclear envelope Neonatal adrenoleukodystrophy, 214, 220, 223, 261 N-ethylmaleimide-sensitive factor (NSF), 122, 151, 186 NLS see Nuclear localization signals NMR see Nuclear magnetic resonance NPC see Nuclear pore complex NSF see N-ethylmaleimide-sensitive factor Nuclear basket, 8, 11-18, 13, 14, 17, 18,27,28-29,34,35,41,42 Nuclear envelope (NE), 2-5, 4, 6, 10, 11,34 Nuclear face, 5, 6, 12, 13, 27, 28-29 Nuclear import, 36, 38-40, 39, 41, 42 Nuclear localization signals (NLS), 37, 38-40, 39, 41 Nuclear magnetic resonance (NMR), 55,101,102 Nuclear membranes, 3-5, 4, 19, 284 Nuclear periphery, 13, 28-29, 34
INDEX
Nuclear pore complex (NPC), 2-42, 131 Nuclear ring, 7, 7-9, 12, 17, 18, 41 Nucleotide binding domain (NBD), 233,235,237-238,241,246, 247,251,252,257,261-267 O-linked glycoproteins, 24-30, 26, 31,33,37 One-step secretory pathway, 166, 167-168, 175, 176 Outer membrane, 166, 168-169, 170, 171, 172, 175, 176, 177 see also Mitochondrial outer membrane P. haemolytica leukotoxin (LktA), 98, 101, 102, 104-106, 108 PAP see Peroxisome assembly factor Pancreas, 285-286, 290-293, 291, 295 Partition coefficient, 60, 61-62, 6364, 63-64, 292 Partitioning, 60, 61-62, 63-64, 71 PDR see Pleiotropic drug resistance Peripheral membrane proteins, 19, 24-31,26, 127, 185, 191 Periplasm, 83, 84, 91, 94-96, 105108, 130, 166, 168-176, 173, 174, 176, 238 Permeability, 286-295, 288, 289, 291 Permease, 85, 93, 105, 234, 238 Peroxisomal enzymes see Peroxisomes Peroxisomal ghosts, 186, 199, 200, 220 Peroxisomal proteins, 182, 191-202, 196, 214, 215-219, 218 Peroxisome assembly factor (PAF), 200-202,214,221,222-226
Index
Peroxisome-deficiency, 199, 200, 201,214,220-226,222,224 see also Zellweger syndrome; Neonatal adrenoleukodystrophy; Infantile Refsum disease Peroxisome proliferator-activated receptor (PPAR), 192, 193194 Peroxisomes, 182-202, 214-226 Peroxisome targeting signal (PTS), 198-201,214-219,218 P-glycoprotein (Pgp), 84, 85, 89, 91, 93,104,109, 111,232 Phosphatidylinositol 3-kinase (PI 3kinase), 139, 140-146 Phosphoinositides, 140-145 Phospholipids, 53, 56, 59, 68-70, 69, 70,71, 101, 141, 143,149, 155 Phosphorylation, 5, 139-140, 143, 144, 145, 146, 149, 236, 264 Pleiotropic drug resistance (PDR), 253-260, 254, 255, 256, 258259, 266, 268, 269 Plug-spoke complex see Central plug; Spoke complex Polypeptides, 18, 23, 24-25, 26, 3035, 32, 34, 53, 215, 216, 218, 233 Precursor proteins, 54, 57, 58, 65, 72, 73, 143, 145 Precursors, 50, 52, 70-71, 74, 170, 171 Prenatal carrier detection see Zellweger syndrome Presequences see Mitochondrial presequences Prevacuolar compartment, 123, 138, 154 Prokaryotes, 85, 232, 234, 261, 284 Pro region, 168-176, 171, 173, 174, 176
299 Protease, 98, 130, 167, 168-169, 186, 199, 215, 216, 248 Proteinase, 121, 122, 123-126, 128, 129, 135-138, 140, 147, 149 Protein kinase, 139-140, 144-146, 149 Protein sorting, vacuolar see Vacuole, yeast Proton motive force, 92-93, 109, 168 PrtSM see Serratia marcescens Rab family, 146, 148, 155 Receptors, 36, 37, 41, 42, 72-74, 128, 137-139, 142, 144-146, 149, 153-155, 201 Repeat toxin (RTX), 83-85, 88, 89, 93, 94, 96, 98-100, 102-104, 106, 108, 110, 111 Ribonucleoprotein particles (RNP), 3, 36, 38-42 Ribosomes, 3, 9, 214, 263-264 RNP see Ribonucleoprotein particles RTX see Repeat toxin Saccharomyces cerevisiae, 120, 141, 149, 185, 187, 188, 194, 195, 196,198-201,219,225,232, 235, 236, 240-266, 252, 254, 255, 256, 258-259 Schizosaccharomyces pombe, 235, 236, 252, 253, 258-259 Sec family, 96, 107, 109, 122, 125, 130, 143, 146, 147, 152, 154, 155, 166, 167, 175, 244, 270271 Secretory pathway, 120, 121, 121122, 126, 130, 134, 143, 146, 147, 152, 243, 244, 248 Serratia marcescens (PrtSM), 98, 99, 108, 168
300
Signals, 3, 36, 40-42, 85, 96, 96-104, 97, 100, 108, 127-128, 140141, 146, 166, 175, 176, 201, 223 see also Targeting; topogenic signals Signature region, 237, 238, 240-241, 246, 256, 257, 262, 263 SKL, 188, 201, 215-219, 222, 223 SNARE proteins, 151, 152, 153, 155 Spoke complex, 6, 7, 9-10, 18, 19, 36,41 STE6, 233, 235, 236, 238, 239, 240, 241-251,253,257,266,272 Substrates, 85, 89, 104-106, 108-110, 142, 169, 240 Surface activity, mitochondrial, 54, 56-58, 71, 74 Surface potential, 62-65, 63-64 Synthetic presequences, 54-59, 65, 73 TAP proteins, 232, 233, 236, 239, 240, 261-262 TAP transporters, 234-235, 246-247 Targeting, 50, 52-53, 57, 70, 72, 125, 127, 128, 144, 146-148, 166, 168, 171, 172, 175, 176, 182, 185, 188, 193, 197-201, 214219,218 Terminal ring, 8, 12, 14, 27, 28, 2829, 34, 35 3-D structures, 7-9, 11, 15, 16, 17, 18, 25-27, 33-36, 34, 40-41 TolC, 82-83, 84, 89,90,94-95, 98, 107 Topogenic signals, 198, 214-219, 222 Trans-Golgi network (TGN), 146, 153, 284, 294
INDEX
Vacuolar protein mutants, 120, 128129,130,131-132,133, 134, 138-152 Vacuole, yeast, 120-155, 240, 248, 265-266 Valinomycin, 245, 266, 269, 270 Vesicles, 55-59, 62-65, 63-64, 68-70, 69, 70, 74, 122, 123, 128, 138, 139, 142-146, 148, 149, 151-153, 155, 220, 244, 248, 270, 295 Walker sites, 237-238, 240, 241, 246,254,256,257,261, 262, 263 Xenopus nucleoplasmin, 37, 192 Xenopus oocyte nuclear envelope (NE), 3, 5,6,7, 11-13, 13, 14, 16, 27, 28-29, 31, 35, 36 Yeast, 60, 64, 73, 217, 232-272, 233, 237, 252, 258-259 nuclear pore complex (NPC) and, 19,24,25,31-35,32,34,41, 51 peroxisomes and, 182-183, 194197, 196, 198, 199, 200, 201, 225 see also Vacuole, yeast; STE6 Zellweger syndrome, 182, 183, 186, 199, 200, 214, 220, 221, 223224, 225, 226, 261 Zymogen granule (ZG), 285, 286, 2S7-295, 288, 289, 291
Membrane Protein Transport A Multi-Volume Treatise Edited by Stephen S. Rothman, University of California, San Francisco Volume 1,1995,286 pp. ISBN 1-55938-907-9
$109.50
CONTENTS: Preface, Stephen S. Rottiman. The Energetics of Bacterial Protein Translocation, Robert Arkowitz. Transport of Pertussis Toxin Across Bacterial and Eukaryotic Membranes, Drusilla L Burns. Transport of Protein-Nucleic Acid Complexes Within and Between Plant Cells, Vitaly Citovsky and Patricia Zambryski. Selective Degradation of Cytosolic Proteins by Direct Transport into Lysosomes, J. Fred Dice. The Conformation and Path of Nascent Proteins in Ribosomes, Boyd Hardesty, Ada Yonath, Gisela Kramer, O.W. Odom, Miriam Eisenstein, Francois Franceschi, and Wieslaw Kudlicki. Protein Import Into Mitochondria, l\/lartin Horst and Nafsika G. Kronidou. Selective Secretion by Lysosomes, Lois Isenman. Colicin Transport, Claude J. Lazdunski. The Mechanism of Diphtheria Toxin Translocation Across Membranes, Erwin London. Protein Translocation Into Choloroplasts, Marinus Pilon, Twan America, Ron van't Hot, Ben de Kruijff, and Peter Weisbeek. Internationalization of Pseudomonas Exotoxin a Utilizes the a2 Macroglobulin. Receptor/Low Density Lipoprotein Receptor Related Protein, Randal E. Morris and Catharine B. Saelinger. Index. Volume2, 1995, 272 pp. ISBN 1-55938-983-4
$109.50
CONTENTS: Preface. The Role of Molecular Chaperones in Transport of Proteins Across Membranes, Elizabeth A. Craig, B. Diane Gambill, Wolfgang Voos, and Nikolaus Pfanner. The Nuclear Pore Complex in Yeast, Paolo Grand! and Eduard C Hurt. ABC Transporters In Yeast: From Mating to Multidrug Resistance, RalfEgner, Yannick Mahe, Rudy Pandjaitan, Veronika Huter, Andrea Lamprecht, and Karl Kuchler. Caveolar Clustering Underlies the Apical Sorting of Glycosylphosphatidylinositol-Linked Proteins, Michael P. Lisanti, ZhaoLan Tang, Philipp E. Scherer, and Massimo Sarglacomo. Caveolae: Portals for Transmembrane Signaling and Cellular Transport, Michael P. Lisanti, ZhaoLan Tang, and Massimo Sarglacomo. Nuclear Transport of Uracil-Rich Small Nuclear Ribonucleoprotein Particles, Ellsa Izaurralde, Lain W. Mattaj, and David S. Goldfarb. Phosphorylation-Mediated Regulated Regulation of Signal-Dependent Nuclear Protein Transport: The "CcN Motif", David A. Jans. Membrane Protein Topogenesis in Escherichia coll, Gunnarvon Heljne. Model for Integrating P-Type ATPases into Endoplasmic Reticulum, Randolph Addison and dialing Lin. Nuclear Transport as a Function of Cellular Activity, Carl M. Feldherr and Debra Akin.
J A I P R E S S
J A I
Advances In Cell and Molecular Biology of Membranes and Organelles (Prevlosly published as Advances In Cell and Molecular Biology of Membranes) Edited by Alan M. Tartakoff, Institute of Pattiology, Case Western Reserve University Volume 4, Protein Export and Membrane Biogenesis 1995.276 pp. ISBN 1-55938-924-9
$109.50
Edited by Ross E. Dalbey, Department of Chemistry, The Ohio State University
P R E S S
CONTENTS: Introduction to the Series, Alan M. Tartakoff. Preface, Ross E. Dalbey. Membrane Protein Assembly, Paul Whitley and Gunnar von Heijne. Membrane Insertion of Small Proteins: Evolutionary and Functional Aspects, Dorothee Kiefer and Andreas Kuhn. Protein Translocation Genetics, Koreaki Ito. Biochemical Analyses of Components Comprising the Protein Translocation Machinery of Escherichia coli, Shin-ichi Matsuyama and Shoji Mizushima. Pigment Protein Complex Assembly in Rhodobacter sphaeroides and Rhodobacter capsulatus, AmyR. Vargas and Samuel Kaplan. Identification and Reconstitution of Anion Exchange Mechanisms in Bacteria, AtuI Varadhachary and Peter C. Maloney Helix Packing in the C-Terminal Half of Lactose Permease, H. Ronald Kaback, Kirsten Jung, Heinrich Jung, Jianhua Wu, Gilbert C. Prive, and Kevin Zen. Export and Assembly of Outer Membrane Proteins in E. coli, Jan Tommassen and Hans de Cock. StructureFunction Relationships in the Membrane Channel Porin, Georg E. Schulz. Role of Phospholipids in Escherichia coli Cell Function, William Dowhan. Mechanism of Transmembrane Signaling in Osmoregulation, Alfaan A. Rampersaud. Index. Also Available: Volumes 1-3 (1993-1994)
$109.50 each
JAI PRESS INC. 55 Old Post Road No. 2 - P.O. Box 1678 Greenwich, Connecticut 06836-1678 Tel: (203) 661 - 7602 Fax: (203) 661 -0792