Nonviral Vectors for Gene Therapy: Methods and Protocols (Methods in Molecular Medicine)

M E T H O D S I N M O L E C U L A R M E D I C I N E TM Nonviral Vectors for Gene Therapy Methods and Protocols Edited b...

Author: Mark A. Findeis

49 downloads 1216 Views 6MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

M E T H O D S I N M O L E C U L A R M E D I C I N E TM

Nonviral Vectors for Gene Therapy Methods and Protocols Edited by

Mark A. Findeis

Humana Press

Synthesis of Polyampholyte Comb-Type Copolymer

1

1 Synthesis of Polyampholyte Comb-Type Copolymers Consisting of Poly(L-lysine) Backbone and Hyaluronic Acid Side Chains for DNA Carrier Atsushi Maruyama and Yoshiyuki Takei 1. Introduction Polycations have been used as nonviral gene carriers because the polycations and DNAs form stable complexes in a noncovalent manner (1–3). The polycations, e.g., poly(L-lysine) (PLL), are reported to be conjugated with several ligands for targeted gene delivery (4–8). The physicochemical properties of the DNA complexes have been described as factors that influence transfection activity (9–11). The authors have reported (12,13) several comb-type copolymers consisting of a PLL backbone and hydrophilic dextran side chains for controling the assembling structure of DNA–copolymer complexes. The dextran chains grafted onto PLL are found to reduce aggregation of the resulting complexes and to increase the solubility of the complexes. Furthermore, the grafting degree of the copolymer affects the DNA conformation in the complex, allowing regulation of DNA compaction. The comb-type copolymers with a higher degree of grafting induce little compaction of DNA and stabilize DNA duplexes and triplexes by shielding the repulsion between phosphate anions of DNA. Moreover, the grafted chains reduce the nonspecific interaction of the PLL backbone with proteins (14). The comb-type copolymers therefore fulfill several requirements for the cellspecific carrier of DNA, if the copolymers are provided with cell-specific ligands. Hyaluronic acid (HA) is an unbranched high-molecular-weight polysaccharide consisting of alternating N-acetyl-`-D-glucosamine and `-D-glucuronic acid residues linked at the 1.3 and 1.4 positions, respectively (15). Liver sinusoidal endothelial cells possess the receptors that recognize and internalize HA (16,17). More than 90% of HAs in the blood stream are known to be taken up and metaboFrom: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

1

2

Maruyama and Takei

lized by SECs. The authors are therefore interested in HA as the ligand to deliver the DNA to the SEC. The authors’ recent study (18) shows that the complexes between PLL–HA conjugates and reporter genes were distributed exclusively in SECs, leading to gene expression in vivo. This chapter describes preparation of PLL-graft-HA (PLL-g-HA) comb-type copolymers. For the synthesis of the comb-type copolymers, high-molecularweight HA was hydrolyzed, then the HAs were covalently coupled with ¡-amino groups of PLL at their reducing end by reductive amination reaction. 2. Materials 2.1. Enzymatic Hydrolysis of HA 1. High-molecular-weight HA (5.9 × 102 kDa), obtained as its sodium salt (sodium hyaluronate), was a gift from Denki Kagaku Kogyo (Tokyo, Japan). 2. Bovine testicular hyaluronidase (EC 3.2.1.35; Type I-S, Sigma, St. Louis, MO). 3. Syringe filters, 0.45 µm (New Steradisc 25, Kurabo, Osaka, Japan).

2.2. Synthesis of PLL-g-HA Comb-Type Copolymers 1. PLL, obtained as its chloride or bromide salt, was purchased from Peptide Institute (Osaka, Japan). 2. Sodium borate buffer-NaCl: 0.1 M, pH 8.5, 0.4–1 M NaCl. 3. Sodium cyanoborohydride (NaBH3CN). 4. NaCl solution (0.5 M). 5. Dialysis membrane (Spectra/Por 7, mol wt cut-off 25,000). 6. Distilled water.

2.3. Size Exclusion Chromatography (SEC)–Multiangle Laser Light Scattering (MALLS) Apparatus 1. 2. 3. 4.

Chromatography pumping system operating at 1.0 mL/min. Size exclusion chromatography column(s). NaNO3 (0.1 M) containing 5 mM sodium phosphate buffer, pH 8.0. Na2SO4 (0.2 M) containing 5 mM sodium phosphate buffer, pH 8.0.

3. Methods 3.1. Enzymatic Hydrolysis of HA 1. Dissolve hyaluronic acid (1 g) in 120 mL water. 2. Add 20 mg bovine testicular hyaluronidase and stir at 50°C. 3. Use a small portion of the reaction to trace HA molecular weight change by SEC–MALLS. 4. When the desired molecular weight of the HA is reached, boil the mixture for 5 min to terminate the reaction. 5. After cooling down to room temperature, filter the mixture through a 0.45-µm filter to remove the denatured enzyme. The resulting HA fragments were obtained by freeze-drying.

Synthesis of Polyampholyte Comb-Type Copolymer

3

(mL) Fig. 1. Time-course of the hydrolysis of HA by hyaluronidase detected by SEC– MALLS. HA (5.9 × 102 kDa; 1 g) was hydrolyzed by hyaluronidase (20 mg) at 50°C.

Figure 1 shows the time course of the HA hydrolysis determined by a SEC– MALLS apparatus (see Subheadings 2.3. and 3.3.). The rate of hydrolysis depends on enzyme activity, so that preliminary experiment on a small scale is recommended to estimate hydrolysis rate before a large-scale reaction. The rate of hydrolysis can be regulated by changing enzyme concentration. For graft copolymer synthesis, a molecular weight ranging from 3000 to 10,000 is favorable. Because the hydrolyzed product has a large distribution in molecular weight, it is also recommended to fractionate fragments by dialysis or ultrafiltration.

3.2. Synthesis of PLL-g-HA Comb-Type Copolymers The obtained HA fragments were conjugated to PLL by reductive amination using NaBH3CN as a reducing agent (Scheme 1). The reaction proceeded through two steps. First is the Schiff’s base (-CH=N-) formation between a reductive (aldehyde) end of HA and primary amino groups of PLL. Second is reduction of the Schiff’s base to form secondary amino groups (-CH2-NH). Although the Schiff’s base is unstable and reversible, its reduced product is an irreversible covalent product. 1. Dissolve the PLL (60–120 mg) in 15 mL sodium borate buffer (0.1 M, pH 8.5) containing 0.4–1 M NaCl. 2. Add the HA fragment (100–300 mg) to the solution. If turbidity or precipitation was observed, increase NaCl concentration. PLL and HA may form an interpolyelectrolyte complex, which is unfavorable for graft copolymer synthesis. The complex formation can be avoided by increasing NaCl concentration. 3. Stir the mixture at 40°C for a few hours for Schiff’s base formation.

4

Maruyama and Takei

Scheme 1. Synthesis of PLL-g-HA comb-type copolymers. Reprinted with permission from ref. 19. Copyright 1998, American Chemical Society.

4. Add NaBH3CN to the mixture and allow to stand at 40°C for 2 d. Approximately 10 molar excess of NaBH3CN to HA is recommended. 5. Sample the solution and trace the reaction with SEC–MALLS (see Subheading 3.3. for SEC–MALLS procedure). 6. Purify the mixture by dialysis against 0.5 M NaCl aqueous solution using a Spectra/Por 7 membrane (mol wt cut-off = 25,000). 7. Desalt the sample by dialysis against distilled water, the resulting copolymer is obtained by freeze-drying. The resulting copolymer would be precipitated during the dialysis.

Figure 2 shows the time-course of the coupling reaction between PLL and HA traced by SEC–MALLS. The reaction can be detected as a decrease in peak area of free HA, increase in peak of the copolymer and in molecular weight of the copolymer. The coupling was almost completed within a few days of incubation. Note that the free HA is almost eliminated after the reaction.

Synthesis of Polyampholyte Comb-Type Copolymer

5

Fig. 2. Time-course of coupling reaction between PLL and HA detected by SEC– MALLS.

3.3. SEC–MALLS SEC was carried out using a JASCO 880-PU pumping system (Tokyo, Japan) at the flow rate of 1.0 mL/min at 25°C, with Ultrahydrogel series (Japan Waters, Tokyo, Japan) or Shodex OH pack SB-series (Showa Denko, Tokyo, Japan). A suitable combination of mobile phase and columns must be chosen, because polyelectrolytes including HA and PLL are liable to interact with the column packings, leading to delay in elution volume. In such case, molecular weight determination using the calibration curve based on molecular weight standard samples such as polyethyleneglycol and pullulan is not reliable. The choice of the mobile-phase rely on gel permeation chromatography (GPC) columns. It is highly recommended to provide a light-scattering (LS) detector system such as MALLS (Multiangle laser light scattering detector, Dawn-DSP, Wyatt Technology, Santa Barbara, CA). By using a LS detector, a direct estimation of the molecular weight is possible. The mobile phases we used are 0.1 M NaNO3 for HA fragment analysis and 0.2 M Na2SO4 containing 5 mM sodium phosphate buffer (pH 8.0) for copolymer analysis. For typical analysis, 200 µL of each sample was picked up from the reaction mixture and injected into the columns. Eluate was detected by a refractive index (RI) detector (830-RI, JASCO) and a MALLS detector. RI and LS signals were transferred to a computer to calculate number-average and weight-average molecular weight according to the instruction manual (Wyatt Technology) for Dawn-DSP.

3.4. 1H Nuclear Magnetic Research (NMR) Spectroscopic Analyses Each copolymer was dissolved in D2O (Deuterium content: 99.95% Merck, Darmstadt, Germany) containing 0.35 M NaCl. 1H-NMR spectra (400 MHz) were

6

Maruyama and Takei

Fig. 3. 1H NMR spectra of PLL (A), HA (B), and PLL-g-HA (C) in D2O. For the PLL-g-HA, D2O containing 0.35 M NaCl was used. Reprinted with permission from ref. 19. Copyright 1998, American Chemical Society.

obtained by a Varian Unity 400plus spectrometer (Palo Alto, CA), at a probe temperature of 298 K. The chemical shifts are expressed as parts/million using internal HDO molecules (b = 4.7 ppm in D2O) as a reference. As shown in Fig. 3, the 1H-NMR spectra of the comb-type copolymer showed the characteristic signals of both PLL and HA moieties: PLL, b 1.4–1.8 (`, a, b-CH2), 3.0 (¡-CH2), 4.3 (_-CH); HA, b 2.0 (NAc-CH3), 3.3–3.9 (H-2,3,4,5,6), 4.4–4.6 (H-1). From the signal ratio of methyl protons (2.0 ppm) of the N-acetyl groups of the HA-grafts to ¡-methylene protons (3.0 ppm) of the PLL backbone, the content (wt % and grafting-%) of HA in the copolymer was determined. The results of the synthesis of PLL-g-HA comb-type copolymers are summarized in Table 1. Coupling efficiency was more than about 70%. Consequently, the authors have easily prepared the various PLL–HA conjugates with a well-defined combtype structure by combining enzymatic hydrolysis and the reductive amination.

In feed

Copolymer

PLL Sample

7

1 2 3 4 5 6 7 8

Mol wtb

HA

Mnb/104

mg

Mnb/103

mg

wt%

Mn/104

Mw/Mn

4.2 4.2 7.2 7.2 7.2 4.2 7.2 7.2

61.2 61.2 122 61.2 61.2 61.2 122 61.2

2.3 2.3 3.8 3.8 3.8 1.6 1.6 1.6

94.3 189 189 189 283 94.3 94.3 189

61 76 61 76 82 61 44 76

8.5 11 15 23 31 9.3 12 24

1.5 1.4 1.4 1.6 1.5 1.4 1.5 1.4

Coupling

HA Contentc

efficiencyd Yield

Charge wt% Grafting-% ratio 50 63 53 69 77 55 38 71

5.7 9.4 3.8 7.6 11 9.8 4.9 19

0.34 0.56 0.40 0.83 1.3 0.49 0.22 1.0

%

%

66 55 73 73 71 80 80 77

67 57 55 83 57 38 51 85

aReducing reagent, 0.3 M NaBH CN; reaction temperature, 40°C; reaction time, 56 h (samples 1 and 2), 80 h (samples 3–5), 75 h (samples 6–8); 3 solvent, 0.1 M sodium borate buffer (pH 8.5) containing 0.4 M NaCl (samples 1 and 2) or 1 M NaCl (samples 3–8). b Molecular weight and its distribution (M /M ) were determined by SEC–MALLS. w n cDetermined by 1H-NMR; grafting-% = (mol fraction of the lysine residues grafted with HA) × 100%; charge ratio = [carboxyl group] / HA [amino group]PLL in copolymer. d [HA] copolymer/[HA]in feed × 100%. Reprinted from ref. 19. Copyright 1998, American Chemical Society.

Synthesis of Polyampholyte Comb-Type Copolymer

Table 1 Synthesis of PLL-g-HA Comb-Type Copolymersa

7

8

Maruyama and Takei

References 1. Kabanov, A. V., Astafyeva, I. V., Chikindas, M. L., Rosenblat, G. F., Kiselev, V. I., Severin, E. S., and Kabanov, V. A. (1991) DNA interpolyelectrolyte complexes as a tool for efficient cell transformation. Biopolymers 31, 1437–1443. 2. Boussif, O., Lezoualc’h, F., Zanta, M. A., Mergny, M. D., Scherman, D., Demeneix, B., and Behr, J. P. (1995) A versatile vector for gene and oligonucleotide transfer into cells in culture and in vivo: Polyethylenimine. Proc. Natl. Acad. Sci. USA 92, 7297–7301. 3. Page, R. L., Butler, S. P., Subramanian, A., Gwazdauskas, F. C., Johnson, J. L., and Velander, W. H. (1995) Transgenesis in mice by cytoplasmic injection of polylysine/DNA mixtures. Transgenic Res. 4, 353–360. 4. Wu, G. Y. and Wu, C. H. (1987) Receptor-mediated in vitro gene transformation by a soluble DNA carrier system. J. Biol. Chem. 262, 4429–4432. 5. Wagner, E., Cotten, M., Foisner, R., and Birnstiel, M. L. (1991) Transferrinpolycation-DNA complexes: the effect of polycations on the structure of the complex and DNA delivery to cells. Proc. Natl. Acad. Sci. USA 88, 4255–4259. 6. Huckett, B., Ariatti, M., and Hawtrey, A. O. (1990) Evidence for targeted gene transfer by receptor-mediated endocytosis: stable expression following insulindirected entry of neo into HepG2 cells. Biochem. Pharmacol. 40, 253–263. 7. Trubetskoy, V. S., Torchilin, V. P., Kennel, S. J., and Huang, L. (1992) Use of N-terminal modified poly(L-lysine)-antibody conjugate as a carrier for targeted gene delivery in mouse lung endothelial cells. Bioconjugate Chem. 3, 323–327. 8. Martinez-Fong, D, Mullersman, J. E., Purchio, A. F., Armendariz-Borunda, J., and Martinez-Hernandez, A. (1994) Nonenzymatic glycosylation of poly-L-lysine: a new tool for targeted gene delivery. Hepatology 20, 1602–1608. 9. Perales, J. C., Grossmann, G. A., Molas, M., Liu, G., Ferkol, T., Harpst, J., Oda, H., and Hanson, R. W. (1997) Biochemical and functional characterization of DNA complexes capable of targeting genes to hepatocytes via the asialoglycoprotein receptor. J. Biol. Chem. 272, 7398–7407. 10. Wolfert, M. A., Schacht, E. H., Toncheva, V., Ulbrich, K., Nazarova, O., and Seymour, L. W. (1996) Characterization of vectors for gene therapy formed by self-assembly of DNA with synthetic block co-polymers. Hum. Gene Ther. 7, 2123–2133. 11. Kabanov, A. V. and Kabanov, V. A. (1995) DNA complexes with polycations for the delivery of genetic material into cells. Bioconjugate Chem. 6, 7–20. 12. Maruyama, A., Katoh, M., Ishihara, T., and Akaike, T. (1997) Comb-type polycations effectively stabilize DNA triplex. Bioconjugate Chem. 8, 3–6. 13. Maruyama, A., Watanabe, H., Ferdous, A., Katoh, M., Ishihara, T., and Akaike, T. (1998) Characterization of interpolyelectrolyte complexes between doublestranded DNA and polylysine comb-type copolymers having hydrophilic side chains. Bioconjugate Chem. 9, 292–299. 14. Maruyama, A., Ishihara, T., Kim, J. S., Kim, S. W., and Akaike, T. (1997) Nanoparticle DNA carrier with poly(L-lysine) grafted polysaccharide copolymer and poly(D,L-lactic acid). Bioconjugate Chem. 8, 735–742.

Synthesis of Polyampholyte Comb-Type Copolymer

9

15. Balazs, E. A., Laurent, T. C., and Jeanloz, R. W. (1986) Nomenclature of hyaluronic acid. Biochem. J. 235, 903. 16. Forsberg, N. and Gustafson, S. (1991) Characterization and purification of the hyaluronan-receptor on liver endothelial cells. Biochim. Biophys. Acta 1078, 12–18. 17. Yannariello-Brown, J., Frost, S. J., and Weigel, P. H. (1992) Identification of the Ca2+-independent endocytic hyaluronan receptor in rat liver sinusoidal endothelial cells using a photoaffinity cross-linking reagent. J. Biol. Chem. 267, 20,451–20,456. 18. Takei, Y., Maruyama, A., Nogawa, M., Asayama, S., Ikejima, K., Hirose, M., et al. (1999) A novel gene delivery system for genetic manipulation of sinusoidal )endothelial cells by triplex DNA technology: evaluation of targetability and ability to stabilize triplex formation. Hepatology 30, 298A. 19. Asayama, S., Nogawa, M., Takei, Y., Akaika, T., Maruyama, A. (1998) Synthesis of novel polyampholyle comb-type copolymers consisting of poly (L-lysine) backbone and hyaluronic acid side chains for a DNA carrier. Bioconjugate Chem. 9, 476–481.

Cationic _-Helical Peptides

11

2 Cationic _-Helical Peptides for Gene Delivery into Cells Takuro Niidome and Haruhiko Aoyagi 1. Introduction Development of nonviral gene transfer techniques has progressed, particularly the use of several kinds of cationic lipids and cationic polymers such as polylysine derivatives, polyethyleneimines, polyamidoamine dendrimers, and so on, which electrostatically form a complex with the negatively charged DNA, which can be taken up by the cells. Furthermore, targeted gene transfer has also been realized by modification of the gene carriers using cell-targeting ligands such as asialoorosomucoid, transferrin, insulin, or galactose. Recently, novel gene transfer techniques have been reported, in which an amphiphilic _-helical peptide, containing cationic amino acids is used as a gene carrier into cells. Wyman et al. (1) employed a peptide, KALA (WEAKLAKA-LAKA-LAKH-LAKA-LAKA-LKAC-EA), which is derived from the sequence of the N-terminal segment of the HA-2 subunit of the influenza virus hemagglutinin involved in the fusion of the viral envelope with the endosomal membrane. This peptide showed several functions in the transfection process, e.g., condensing DNA and causing an endosome-membrane perturbation, which enables it to deliver the incorporated DNA to the cytosol, which is essential for efficient transfection. Similarly, the authors also found the transfection technique, which is mediated by some amphiphilic _-helical peptides (e.g. Ac-LARL-LARL-LARL-LRAL-LRAL-LRAL-NHCH3 [46] and KLLK-LLLK-LWKK-LLKL-LK [Hel]) as shown in Table 1 (2–5). After that, for the purpose of refining of the peptide structure, we investigated the influence of the peptide chain length on gene transfer ability. As a result, 16 and 17 amino acid residues were sufficient to form aggregates with the DNA, From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

11

12

Niidome and Aoyagi

Table 1 Structures of Amphiphilic _-Helical Peptides Peptide

Sequence

46 Ac-LARL-LARL-LARL-LRAL-LRAL-LRAL-NH2 4668 Ac-LARL-LRAL-LRAL-LRAL-NH2 Hel KLLK-LLLK-LWKK-LLKL-LK Hel61 LLK-LLLK-LWKK-LLKL-LK

Chain length

Cationic charge

24 16 18 17

6 4 7 6

and transfer the DNA into the cells in the deletion series of 46 and Hel, respectively (Table 1; 6). In addition, the authors succeeded in constructing a multiantennary galactose-modified peptide containing four galactose residues that serve for efficient binding to the asialoglycoprotein receptor on hepatoma cells (7). This chapter focuses on synthesis of the peptides and a method of gene transfer using them. As is well known, a peptide is readily synthesized because of the development of an automatic peptide synthesis apparatus and reagents for synthesis. From this point of view, it is expected that the gene transfer method mediated by the peptide is easily accepted by many researchers taking part in the gene therapy study. 2. Materials 2.1. Peptide Synthesis 1. Peptide synthesis apparatus (ABI 431A, PE Biosystems). 2. Fmoc-Lys(Boc) preloaded Wang resin, Rink amide resin (100–200 mesh) (Calbiochem-Novabiochem, CA). 3. Fmoc protected amino acids, 2-(1H-benzotriazole-1-yl)-1,1,3,3,-tetramethyluronium hexafluorophosphate (HBTU), N,N,-diisopropylethylamine, NMP, dichloromethane (DCM), piperidine (Watanabe Chemical, Hiroshima, Japan). 4. Thioanisole, m-Cresol, ethandithiol, trifluoracetic acid (TFA), acetonitrile (Wako Chemicals, Osaka, Japan). 5. High-performance liquid chromatography (HPLC) apparatus (Hitachi L7100 System, Tokyo, Japan). 6. Reverse-phase (RP)-HPLC column (YMC-Pack C4, q10 × 150 mm, Kyoto, Japan). 7. Matrix-assisted Laser Desorption Ionization-Time of Flight-Mass Spectra (MALDI TOF-MS) apparatus (Voyager DE STR, PE Biosystems).

2.2. Preparation of Plasmid DNA 1. Plasmid DNA, which contains a luciferase gene and SV40 promoter (PicaGene control vector, PGV-C), was purchased from Toyo Ink (Tokyo, Japan). 2. Plasmid DNA (pCMVluc), containing a luciferase gene under control of cytomegalovirus enhancer/promoter, was prepared by removing the BglII and

Cationic _-Helical Peptides

13

HindIII insert of the plasmid PGV-C (Toyo Ink), and ligating with the BglII and HindIII fragment from the pRc/CMV (Invitrogen), which contains cytomegalovirus promoter. 3. Closed circular plasmid DNA was purified by ultracentrifugation in cesium chloride gradients. The plasmid preparations showed a major band of closed circular DNA and minor amount (<20%) of nicked plasmid.

2.3. Cell Culture and Gene Transfer into Cells 1. COS-7 cells (a monkey kidney cell line, RCB accession no. RCB0539), HeLa cells (a human cervix, RCB accession no. RCB0271) and CHO cells (a Chinese hamster ovary, RCB accession no. RCB0285) (RIKEN Cell Bank, Tsukuba, Japan). 2. HuH-7 cells (a human hepatoma cell line, JCRB accession no. JCRB0403) (Health Science Research Resources Bank, Osaka, Japan). 3. Dulbecco’s modified Eagle’s medium, RPMI 1640, media supplements, and heatinactivated fetal calf serum (IWAKI Glass, Chiba, Japan). 4. PicaGene luminescence kit (Toyo Ink). 5. Luminometer (Maltibiolumat LB9505, Berthold, Germany).

3. Methods 3.1. Peptide Synthesis As shown in Table 1, the authors recommend two kinds of peptides, 4668 and Hel61, which were refined from two original _-helical peptides, 46 and Hel, respectively (6).

3.1.1. Elongation of Peptide Chain on Resin Peptides can be manually synthesized by the stepwise elongation of Fmoc protected amino acid on Rink amide resin (for 46; 4-(2',4'-dimethoxyphenylFmoc-aminomethyl)-phenoxy resin, 0.43 mmol/g resin) or Fmoc-Lys(Boc) preloaded Wang resin (for Hel; Fmoc-Lys(Boc)-p-benzoyloxy alcohol resin, 0.58 mmol/g resin) in 0.1 mmol scale as described by Fields et al. (8), and in Catalog & Peptide Synthesis Handbook of Calbiochem-Novabiochem. The Fmoc amino acid derivatives used are as follows: Ala, Arg(Pbf), Leu, Lys(Boc), and Trp. The coupling protocol is shown as follows: 1. 2. 3. 4. 5. 6.

Wash: 2 mL DCM (3×). Wash: 2 mL NMP (3×). Wash: 2 mL of 20% piperidine/N,N,-dimethylformamide (1×). Deprotection of Fmoc group: 2 mL 20% piperidine/DMF (15 min). Wash: 2 mL NMP (5×). Coupling of amino acid; Fmoc-amino acid (0.3 mmol), HBTU (0.3 mmol), HOBt (0.3 mmol), and N,N,-diisopropylethylamine (DIEA) (0.6 mmol) in 3 mL NMP:DMF (1:1) (15 min). 7. Wash: 2 mL DMF (3×). 8. Kaiser test (9): When the coupling is incomplete, the protocol is repeated from step 6.

14

Niidome and Aoyagi

As a matter of course, it is possible to elongate the peptide chain using automatic peptide synthesizer (PE Biosystems, ABI 431A). In this case, the peptides are also synthesized by FastMoc method in 0.1 mmol scale. Even if single coupling of amino are applied for all elongation steps, satisfactory peptides in purity are obtained. In the case of modifying by acetyl group to N-terminal of peptide, 46, the Fmoc-deprotected peptide on resin is treated with 1 mmol acetic anhydride in 2 mL NMP for 20 min. After the resin is washed by DCM and methanol, they are dried in vacuo.

3.1.2. Deprotection and Cleavage of Peptide Resin The protecting groups and the resin are removed with TFA (2.0 mL) in the presence of m-cresol (60 µL), ethandithiol (180 µL), and thioanisole (360 µL), in a round flask. After swirling at room temperature for 60 min, the resin is removed by filtration under reduced pressure and washed by TFA. Twenty milliliters of cold diethylether is added to the filtrate, then, the flask can be cooled with ice to further assist precipitation. Crude peptide is isolated by filtration under reduced pressure and washed by cold diethylether. The crude peptides can be purified by RP-HPLC on a YMC-Pack C4 column (q10 × 250 mm) with a linear gradient of water–acetonitrile containing 0.1% TFA. Peptide 4668 and Hel61 are eluted at about 80 and 55% of acetonitrile, respectively. The authors recommend that if possible, the crude peptide be passed through a column of Sephadex G-10 or G-15 (q10 × 250 mm) with 10% acetic acid before purification by RP-HPLC to avoid damaging the HPLC column. The peptide fractions are collected, then lyophilized. The yields obtained after purification are about 80 (34 µmol) and 100 mg (36 µmol) in the cases of 4668 (as 4TFA salt) and Hel61 (as 6TFA salt), respectively. The purified peptides can easily be identified by MALDI-TOF-MS ([M + H]+ = 1874.4 [4668], 2105.7 [Hel61]).

3.2. Preparation of Peptide–DNA Complex Complex of peptide and DNA is prepared by mixing 2.5 µg plasmid DNA (PGV-C) with the peptides at peptide:DNA charge ratio of 2.0 in serum-free medium. The authors protocol is shown as follows: 1. Prepare the DNA solution: 2.5 µg plasmid DNA in 250 µL serum-free culture medium. 2. Prepare the peptide solution: 1.6 mM (as cationic groups) peptide aqueous solution. Practically, 930 µg 4668 • 4TFA or 743 µg Hel61 • 6TFA are dissolved in 1.0 mL sterilized water. 3. 10 µL Peptide solution is added to 250 µL plasmid DNA solution. 4. Stand for 15 min at room temperature.

3.3. Gene Transfer Protocol into Cultured Cells 1. Plate cells in 24-well tissue culture dishes (q16 mm) at 1 × 105 cells/well and grow overnight in an atmosphere of 5% CO2 at 37°C.

Cationic _-Helical Peptides

15

2. 3. 4. 5.

Wash twice with 1 mL serum free medium. The peptide–DNA complex as described above is poured gently to the cells. After incubation for 3 h at 37°C, 1 mL medium containing serum is added. After incubation for 12 h at 37°C, the medium is replaced with 1 mL fresh medium containing serum and the cells are further incubated for 24 h. 6. Harvest of cells and luciferase assays are performed as described in the protocol of PicaGene luminescence kit using a luminometer. 7. The protein concentrations of the cell lysates are measured by Bradford assay using bovine serum albumin as a standard. The light unit values shown in the figures represent the specific luciferase activity (relative light units/mg protein), which is standardized for total protein content of the cell lysate. The measurement of gene transfer efficiency is performed in triplicate.

To date, the authors have tested gene transfer efficiencies of the peptides into several cell lines, such as COS-7, HeLa, CHO, and HuH-7 cells. Figure 1 shows the results for each cell line. To compare the efficiencies of the peptides to a commercially available gene transfer agent, we employed Lipofectin (Gibco-BRL). As a result, the efficiencies of the peptides were similar to or about 10-fold lower than those of Lipofectin. Except for the case of needing a large amount of expression product, it can be said that the peptides are enough to use as a gene carrier. On the other hand, the peptides showed different efficiencies depending on the cell lines. These peptides can be chosen according to the cell lines. The authors also evaluated cytotoxic activities of the peptide–DNA complexes, using Alamer BlueTM, under the same conditions as those in the transfection procedure. As a result, little cytotoxic activity of the complexes could be observed. However, when the complex is prepared at peptide:DNA charge ratio of 4 and more, considerable cytotoxic activities are observed, which the authors consider to result from membrane perturbation activities originated from the amphiphilic structures of the peptides (3,5).

3.4. Active Targeting of Gene into Hepatoma Cells Using Galactose Modification of Peptide As described in the Introduction, peptide is easily available because of the development of its synthetic technology. Therefore, this allows design and synthesis of functional gene carrier molecules such as carbohydrate-modified peptide for targeted gene delivery. Furthermore, peptide-based gene carrier enables construction of well-defined molecules, which cannot be achieved by polymer-based molecule. Here is introduced a synthesis of a multiantennary galactose-modified peptide and its application to a human hepatoma cell line.

3.4.1. Preparation of Multiantennary Galactose-Modified Peptide The synthesis method is described as follows and is summarized in Fig. 2.

16

Niidome and Aoyagi

Fig. 1. Gene transfer efficiencies of the peptides into several cell lines.

1. _-Helical peptide portion can be synthesized by ordinary Fmoc solid-phase method on Rink amide resin (see Subheading 3.1.1.). 2. After coupling Fmoc–2-(2-aminoethoxy) ethanol as a linker, Fmoc-`AlaLys(Fmoc) is coupled twice. Fmoc groups of secondary coupled Fmoc-`AlaLys(Fmoc) are removed by piperidine. At this step, the peptide has four amino groups in the molecule. 3. After cleavage from resin and deprotection, the peptide is purified by RP-HPLC as similar condition to the case of peptide 46 (see Subheading 3.1.2.) 4. The amino groups of the peptide can be modified with lactose by aldimine formation, followed by reduction with NaBH3CN of the secondary amines as follows. The purified peptide (11 µmol, 50 mg as 10TFA salt) is mixed with a solution of lactose (530 µmol, 190 mg as monohydrate) in 500 µL 10 mM aqueous

Cationic _-Helical Peptides

17

Fig. 2. Outline of synthesis of galactose-modified peptide, Gal4–46. sodium acetate, pH 5.0, at 37°C, then NaBH3CN (44 µmol, 2.8 mg each) is added to the solution at 12-h intervals. After total incubation for 60 h at 37°C, the resultant galactose-modified-peptide is purified by RP-HPLC. Yield is 20 mg (38%). 5. Identification is performed by MALDI-TOF-MS ([M + H]+ = 4816.7). The peptide concentration in solution is determined from UV-absorbance of Trp at 280 nm in a buffer containing 6 M Gu • HCl (¡ = 5500).

3.4.2. Gene Transfer Using Galactose-Modified Peptide into Hepatoma Cell Line, HuH-7 Protocol for gene transfer into HuH-7 cells, a human hepatoma cell line, is similar to that described in Subheadings 3.2. and 3.3. Complex of peptide and DNA is prepared by mixing 2.5 µg plasmid DNA (pCMVluc) with Gal4–46 at peptide:DNA charge ratio of 2.0 in serum free medium. 1. Prepare the DNA solution: 2.5 µg plasmid DNA in 250 µL serum free RPMI 1640.

18

Niidome and Aoyagi

2. Prepare the peptide solution: 1.6 mM (as cationic groups) peptide aqueous solution. Practically, 1.5 mg Gal4–46 • 10TFA is dissolved in 1.0 mL sterilized water. 3. 10 µL peptide solution is added to 250 µL plasmid DNA solution. 4. Stand for 15 min at room temperature. 5. The peptide–DNA complex as described above is poured gently on to the cells, which are washed twice with 1 mL serum-free RPMI 1640, beforehand. Following procedures are same to steps 4–7 in Subheading 3.3.

As a result of measurement of luciferase activity in HuH-7 cells, the activity of Gal4–46 was 300-fold higher than that of 46. In addition, the authors could confirm that the DNA complex of Gal4–46 was internalized into the cells via the asialoglycoprotein receptor (see Subheading 3.5.). This chapter introduced a galactose-modified peptide containing 2-(2aminoethoxy)ethanol and `Ala as a linker, which allows the galactose residues to be easily recognized by the receptor. However, the authors found that a Gal4–46 derivative without any linker showed a high efficiency into HuH-7 cells, as similar to that of Gal4–46. Insertion of linker into the peptide system is not always necessary for recognition of the galactose residues by the receptor.

3.5. Commentary In the authors’ series of studies, it has become clear that the hydrophobic region on the amphiphilic structure of the peptides plays an important role in binding to the plasmid DNA and formation of aggregates with the DNA. It is likely that the hydrophobic region of the peptides induces stable oligomers with a well-defined number of monomers by self-association with their hydrophobic faces in aqueous solution. As a result, the oligomer would behave like a polycation, which can form aggregates with DNA (Fig. 3). Furthermore, the authors indicate that the hydrophobic region is also important for the disruption of the endosomal membrane in the cell, which can transfer the incorporated DNA to cytosol and prevent the degradation of the DNA in the lysosomal vesicles (3,5). In order to clear detail translocation pathway of the DNA in the cell, the authors evaluated effects of several endocytosis inhibitors on the gene transfer efficiencies of peptide 46. Treatment with cytochalasin B, which depolymerizes the microfilaments of actin and blocks the uncoated pit-mediated endocytosis (macropinocytotic process) (10), reduced to 25% of the original efficiency of 46; no effect was observed in the case of treatment with chlorpromazine, an inhibitor of clathrin-dependent, receptor-mediated endocytosis (11). Furthermore, N-ethylmaleimide (NEM), which inhibits fusion between endosomes at an early stage of the endocytic pathway (12), reduced to 50% of the original efficiency, and the microtubule-depolymerizing agent, nocodazole, which interferes with transport from the early to late endosome mediated by endocytic carrier vesicles (13), increased the efficiency of the peptide by sixfold. From

Cationic _-Helical Peptides

19

Fig. 3. Formation of peptide–DNA complex and its transfer pathway into cell.

these results, it is reasonable to suppose that the complexes of peptide 46 and plasmid DNA were internalized by macropinocytotic process, and a part of the complexes could be translocated from the endosomal compartment to the cytosol during an early endosome step (Fig. 3). However, the other part of the complexes would be translocated into degradation compartments such as late endosome and lysosome, where the complex could no longer be translocated to the cytosol. In order to enhance gene transfer efficiency, it will be necessary to consider active translocation to the cytosol of the DNA complex. Because there is now no information for translocation into nucleus of the complex, elucidation of this transfer mechanism in the cell will give a clue to construct a novel gene carrier with higher efficiency. The authors have also studied the transfer pathway of the DNA complex with Gal4–46. At first, the competitive effects of asialofetuin, which is internalized into hepatoma cells via the asialoglycoprotein receptor-mediated endocytosis, and fetuin, which is not recognized by the receptor, on the transfer efficiencies of the peptides were examined. As a result, the transfer efficiency of Gal4–46 is reduced to 1% of the original efficiency in the presence of the asialofetuin, but no effect was observed in 46. However, fetuin showed weak effect compared with asialofetuin. In addition, the authors evaluated effects of several endocytosis inhibitors on the gene transfer efficiencies of Gal4-46. Treatment with chlorpromazine significantly reduced the efficiency; no significant reduction was observed in the case of treatment with cytochalasin B. This result suggested that the

20

Niidome and Aoyagi

internalization of Gal4–46 was mediated by the clathrin-dependent, receptormediated endocytosis. Furthermore, NEM reduced to 50% of the original efficiency, and nocodazole increased the efficiency by twofold. From these results, DNA complex with the galactose modified peptide, Gal4–46, would be translocated from the endosomal compartment to the cytosol during an early endosome step as in the case of 46. Finally, to apply these galactose-modified peptides for targeted gene delivery in vivo, it is necessary to further examine the cell selectivity using several cell lines, stability in blood, capture by the reticuloendothelial system, and so on. When these points are solved, this delivery system will be one of the powerful tools for gene therapy. References 1. Wyman, T. B., Nicol, F., Zelphati, O., Scaria, P. V., Plank, C., and Szoka, F. C. (1997) Design, synthesis, and characterization of a cationic peptide that binds to nucleic acids and permeabilizes bilayers. Biochemistry 36, 3008–3017. 2. Ohmori, N., Niidome, T., Hatakeyama, T., Mihara, H., and Aoyagi, H. (1998) Interaction of _-helical peptides with phospholipid membrane: effects of chain length and hydrophobicity of peptides. J. Peptide Res. 51, 103–109. 3. Niidome, T., Ohmori, N., Ichinose, A., Wada, A., Mihara, H., Hirayama, T., and Aoyagi, H. (1997) Binding of cationic _-helical peptides to plasmid DNA and their gene transfer abilities into cells. J. Biol. Chem. 272, 15,307–15,312. 4. Kiyota, T., Lee, S., and Sugihara, G. (1996) Design and synthesis of amphiphilic _-helical model peptides with systematically varied hydrophobic-hydrophilic balance and their interaction with lipid- and bio-membranes. Biochemistry 35, 13,196–13,204. 5. Ohmori, N., Niidome, T., Kiyota, T., Lee, S., Sugihara, G., Wada, A., Hirayama, T., and Aoyagi, H. (1998) Importance of hydrophobic region in amphiphilic structures of _-helical peptides for their gene transfer-ability into cells. Biochem. Biophys. Res. Commun. 245, 259–265. 6. Niidome, T., Takaji, K., Urakawa, M., Ohmori, N., Wada, A., Hirayama, T., and Aoyagi, H. (1999) Chain length of cationic _-helical peptide sufficient for gene delivery into cells. Bioconjugate Chem. 10, 773–780. 7. Niidome, T., Urakawa, M., Sato, H., Takahara, Y., Anai, T., Hatakayama, et al. (2000) Gene transfer into hepatoma cells mediated by galactose-modified alphahelical peptides. Biomaterials 21, 1811–1819. 8. Fields G. B. and Noble R. L. (1990) Solid phase peptide synthesis utilizing 9-fluorenylmethoxycarbonyl amino acids. Int. J. Peptide Protein Res. 35, 161–214. 9. Kaiser, E., Colescott, R. L. and Bossinger, C. D. (1970) Color test for detection of free terminal amino groups in the solid-phase synthesis of peptides. Anal. Biochem. 34, 595–598. 10. Paccaud J.-P., Siddle K., and Carpentier J.-L. (1992) Internalization of the human insulin receptor. The insulin-independent pathway. J. Biol. Chem. 267, 13,101–13,106.

Cationic _-Helical Peptides

21

11. Orlandi P. A. and Fishman P. H. (1998) Filipin-dependent inhibition of cholera toxin: evidence for toxin internalization and activation through caveolae-like domains. J. Cell Biol. 141, 905–915. 12. Rothman J. E. (1994) Mechanisms of intracellular protein transport. Nature 372, 55–63. 13. Lemichez E., Bomsel M., Devilliers G., vanderSpek J., Murphy J. R., Lukianov E. V., Olsnes S., and Boquet P. (1997) Membrane translocation of diphtheria toxin fragment A exploits early to late endosome trafficking machinery. Mol. Microbiol. 23, 445–457.

PEG-b-PLL Dendrimer with pDNA

23

3 Supramolecular Self-Assembly of Poly(ethylene glycol)-Block-Poly(L-lysine) Dendrimer with Plasmid DNA Joon Sig Choi and Jong Sang Park 1. Introduction Research and development related to nonviral gene carriers comprising chemically synthesized molecules has increased enormously during the past decade. Polycationic polymers and cationic lipids have constituted the main themes of the studies. Various polymers from synthetic to naturally occurring ones have been introduced and tested for their suitability in the field of gene therapy. Several cationic polymers were found to be promising but their intrinsic drawbacks, such as solubility, cytotoxicity, and low transfection efficiency, limited their use as in vivo gene carriers (1). Among them, however, dendrimers are still very attractive to many scientists for the design of gene carriers because of their well-defined structure and ease of control of surface functionality. Already, both polyamidoamine dendrimer and polyethylenimine dendrimer have been tested for their potential utility and have exhibited high transfection efficiency in vitro and in vivo (2,3). However, these dendrimers have not yet overcome the problems of solubility of the complex with DNA and cytotoxicity. Block copolymers containing poly(ethylene glycol) (PEG) have been used for many drug carriers because of their high solubility in water, nonimmunogenicity, and improved biocompatiblity (4). PEG has been coupled to polycationic polymers (e.g., poly-L-lysine [PLL], polyspermine, and polyethylenimine) (5–8) or liposomes (9) to improve the solubility of complexes with DNA and transfection efficiency. This chapter provides the design of a conceptually new hybrid block copolymer which is capable of polyionic complex formation with plasmid DNA via supramolecular self-assembly. Linear PEG was coupled to the globular From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

23

24

Choi and Park

macromolecule, poly(L-lysine) dendrimer by the repetitive liquid-phase peptide synthesis method (10). Poly( L -lysine) dendrimer (11–13) is another polycationic dendrimer containing a large number of surface amines and is considered to be capable of electrostatic interaction with polyanions, such as nucleic acids (14). The enhanced aqueous solubility of the complexes is an advantage compared to that of homopolymer polycations and cationic lipids. This type of self-assembly is interesting, from both theoretical and practical points of view in designing polymers for gene delivery vectors, because it can serve as a suitable model for polyionic complex formation of other hybrid block copolymers with DNA. 2. Materials 2.1. Synthesis 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

Methoxypoly(ethylene glycol) amine, mol wt 3400 (Shearwater Polymer). N-hydroxybenzotriazole (HOBt). 2-(1H-Benzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HBTU). N-_-N-¡-di-Fmoc-L-lysine (AnaSpec, San Jose, CA). N,N-diisopropylethylamine (DIPEA). N,N-dimethylformamide (DMF). Diethyl ether. 30% Piperidine (in DMF). Round-bottomed 50-mL tubes. Spectra/Por dialysis membrane (mol wt cut-off = 6000–8000, Spectrum, Los Angeles, CA).

2.2. Matrix-Assisted Laser Desorption Ionization–Time of Flight Mass Spectra (MALDI-TOF-MS) 1. 2,5-Dihydroxybenzoic acid (DHB). 2. Ultra-pure water (Milli-Q Biocel system).

2.3. Agarose Gel Electrophoresis 1. 5X HEPES-buffered saline (HBS): 0.1 M HEPES, 0.75 M NaCl, pH 7.4. 2. 6X Loading buffer: 0.25% bromophenol blue, 0.25% xylene cyanol FF, 15% Ficoll (Type 400; Pharmacia) in water. 3. 5X TBE buffer: Tris 27 g, boric acid 14 g, 0.5 M EDTA, pH 8.0, 10 mL in 500 mL of water. 4. 0.7% Agarose in TBE buffer containing 0.5 µg/mL ethidium bromide.

2.4. Atomic Force Microscopy 1. Freshly split mica. 2. Ultra-pure water. 3. Whatman 3MM paper.

PEG-b-PLL Dendrimer with pDNA

25

2.5. DNase I Protection Assay 1. DNase I: 8.9 U/µL, 50% glycerol, 20 mM sodium acetate buffer, pH 6.5, 5 mM CaCl2, 0.1 mM phenylmethylsulfonyl fluoride. 2. Stop solution: 4 M ammonium acetate, 20 mM EDTA, 2 mg/mL glycogen. 3. 1% Sodium dodecyl sulfate. 4. Tris-EDTA saturated phenol. 5. Chloroform. 6. Pure ethanol. 7. Tris-EDTA buffer. 8. 6X Loading buffer. 9. 5X TBE buffer. 10. 0.7% Agarose in TBE buffer containing 0.5 µg/mL ethidium bromide.

2.6. Water Solubility Test 1. 5X HEPES-buffered saline. 2. Quartz cuvet. 3. UV spectrophotometer.

2.7. Cell Culture and In Vitro Cytotoxicity Test 1. 2. 3. 4. 5. 6.

293 Cells. Sterilized Dulbecco’s minimum essential medium. 10% Fetal bovine serum. Falcon tubes. 96-Well plate. (3-[4,5-Dimethylthiazol-2-yl]-2,5 diphenyl tetrazolium bromide) (MTT) stock solution, 2 mg/mL in DPBS. 7. DPBS. 8. Dimethylsulfoxide. 9. Eight Channel pipet.

3. Methods 3.1. Synthesis of Hybrid Block Copolymer 1. The overall synthesis scheme is outlined in Fig. 1. Prepare the reaction mixture in anhydrous DMF: mPEG-amine 0.15 g (30 µmol) (see Note 1). Synthesis scale for each step is described in Table 1. 2. After the coupling reaction reaches completion, precipitate the mixture with a 10-fold excess of cold ether and further wash 2× with ether (see Note 2). 3. Deprotect the Fmoc group of lysine using 30% piperidine. 4. Precipitate in cold ether and wash 2× with excess ether (see Note 3). 5. Dry the precipitates in vacuo and prepare for further coupling reaction. 6. Repeat the coupling and deprotection reactions 4× (see Note 4). 7. Solubilize the fourth generation copolymer in water.

26

Choi and Park

Fig. 1. Outline of stepwise liquid phase synthesis of mPEG-b-PLLD. 8. Dialyze for 1 d against water using Spectra/Por dialysis membrane (mol wt cut-off 6–8 kDa) and filter the solution with 0.45-µM syringe filter. 9. Collect the final product by freeze-drying.

PEG-b-PLL Dendrimer with pDNA

27

Table 1 Synthesis Scale for Each Coupling and Deprotection Step rxn no. 1 2 3 4

FmocLys(Fmoc)-OH (g) 0.074 0.147 0.294 0.294 × 2

HOBt (g)

HBTU DIPEA (g) (µL)

0.017 0.047 40 0.033 0.095 80 0.066 0.19 160 0.066 × 2 0.19 × 2 160 × 2

30% Coupling Piperidine (d) 2 mL 2 mL 2 mL (2 mL) × 2

1 1 1–2 d 3d

Deprotection (h) 2 2 3 3

3.2. MALDI-TOF-MS 1. Prepare the stock solution in water: polymer solution, 2–4 mg/mL; DHB solution 10 mg/mL. 2. Mix the two solutions: polymer solution: DHB matrix solution = 1: 9 µL or 2: 8 µL. 3. Pipet about 1 µL mixture and load onto the plate (see Note 5). 4. Dry in vacuo (see Note 6). 5. Characterize the linear polymer/dendrimer block copolymer by MALDI-TOF-MS (see Fig. 2).

3.3. Agarose Gel Electrophoresis 1. Mix the polymer and pSV-`-gal plasmid DNA in 15 mM HEPES buffer, 0.15 M NaCl, pH 7.4. Incubate at room temperature for 30 min. 2. Prepare 0.7% agarose gel containing ethidium bromide (0.5 µg/mL gel). 3. After 30 min, mix with 6X sample loading buffer and load onto the gel. 4. Perform electrophoresis for 30 min at 100 V. 5. Illuminate the elecrophoresced gel on an UV illuminator to show the location of the DNA and complexes (see Fig. 3A,B and Note 7).

3.4. Atomic Force Microscopy 1. Prepare the freshly split mica. 2. Prepare the polymer solution (G = 4) and DNA solution in water. 3. Mix the two solutions and incubate for 30 min at room temperature (see Note 8). Charge ratio (z) 1 2 4 Polymer solution (µg/µL) 0.03 0.06 0.12 5 µL 5 µL 5 µL DNA solution (0.02 µg/µL) 5 µL 5 µL 5 µL 4. Pipet 1–2 µL from the mixture (see Note 9) and load on the surface of the mica (see Note 10). 5. Leave the solution to absorb for 1–2 min, then remove the residual solution by 3MM paper. 6. Air-dry overnight in preparation for imaging (see Fig. 4A–C).

28

Choi and Park

Fig. 2. MALDI-TOF-MS spectrum of the copolymer. The spectrum was obtained on a PerSeptive Biosystems instrument in the linear mode at 20 kV. The Mw and M n values of this copolymer were 7594 and 7553, respectively (M w/Mn = 1.01).

Fig. 3. Analysis of complex formation at various charge ratios by agarose gel electrophoresis. (A) mPEG-PLLD (generation 3) 1.0 µg pSV-`-gal plasmid DNA only (lane 1), charge ratio of copolymer: DNA = 0.5, 1, 2, and 4 (lanes 2, 3, 4, and 5, respectively). (B) mPEG-PLLD (generation 4) 1.0 µg pSV-`-gal plasmid DNA only (lane 1), charge ratio of copolymer: DNA = 1, 2, 3, 5, and 6 (lanes 2, 3, 4, 5, and 6, respectively).

3.5. DNase I Protection Assay 1. Add the copolymer to 4.0 µg of plasmid DNA at various charge ratios from 0 to 4, in 50 µL of 20 mM HEPES buffer, 0.15 M NaCl, pH 7.4. 2. Incubate for 30 min at room temperature.

PEG-b-PLL Dendrimer with pDNA

29

Fig. 4. The atomic force microscopy images of the mPEG-block-PLLD/pSV-`-gal complex. Charge ratios of copolymer: DNA = 1, 2, and 4 (A, B, and C). The image mode was set to tapping mode. The white color indicates a height more than the designated nm above the mica surface. The x and y dimensions are scaled as shown.

30

Choi and Park

3. Further incubate the mixtures in the presence of 8.9 U of DNase I for 20 min at room temperature. 4. To each mixture, add 75 µL stop solution and keep each tube on ice. 5. After addition of 37 µL 1% sodium dodecyl sulfate, extract DNA with Tris-EDTA buffer-saturated phenol/chloroform. 6. Precipitate the DNA pellet by adding pure ethanol. 7. Dissolve the precipitated DNA in Tris-EDTA buffer and subject to 0.7% agarose gel electrophoresis (see Fig. 5).

3.6. Solubility Test of the Polyplex 1. Add the polymer to 26 µg plasmid DNA at a charge ratio of 4 in 0.5 mL 20 mM HEPES buffer, 0.15 M NaCl, pH 7.4 (see Note 11). 2. Incubate for 30 min at room temperature. 3. Centrifuge each tube for 5 min at 13,000 rpm (16,816g), 10°C. 4. Measure the absorbance of the supernatant at 260 nm. 5. Calculate the absorbance percentage compared to that of DNA only solution (see Fig. 6).

3.7. In Vitro Cytotoxicity Assay 1. Seed the 293 cells (human embryonic kidney cell lines) in 96-well microplates at a density of 10,000 cells/well in 0.1 mL growth medium DMEM containing 10% fetal bovine serum. 2. Introduce the polymer to each well and incubate for 48 h. 3. Remove the old medium and replace with new growth medium containing MTT (26 µL 2 mg/mL stock solution/each well) (see Note 12). 4. Incubate the plate for an additional 4 h at 37°C. 5. Remove each medium prior to the addition of 150 µL of dimethylsulfoxide. 6. Mix up and down using eight-channel pipet. 7. Measure the absorbance at 570 nm using a microplate reader (Molecular Devices, Menlo Park, CA). 8. Calculate the absorbance percentage relative to that of untreated control cells (see Fig. 7).

4. Notes 1. It is important to get a MALDI signal for the starting material, mPEG-amine (weight average molecular weight [Mw]= 5757, number average molecular weight [M n ]= 5697, M w/M n = 1.01, determined by MALDI-TOF mass spectrum). Usually, the specified molecular weight of commercial PEG differs from the experimental value. 2. The progress of each reaction was monitored by ninhydrin test and 1H-nuclear magnetic resonance until completed. 1H-nuclear magnetic resonance (D2O) b 1.64 (broad multiplet [brm], [CH2]3), 3.08 (brm, CH2-N), 3.39 (siglet [s], CH3-O), 3.68 (s, CH2CH2-O), 4.25 (brm, COCH-N).

PEG-b-PLL Dendrimer with pDNA

31

Fig. 5. Stability of mPEG-block-PLL–plasmid DNA complexes to DNase I at various charge ratios. Undigested plasmid DNA (lane 1), Charge ratio of copolymer: DNA = 0, 0.5, 1, 2, and 4 (lanes 2, 3, 4, 5, and 6, respectively). The positions of the open circular (oc), linear (li), and supercoiled (su) forms are indicated on the right.

Fig. 6. Water-solubility test of some polyplexes. The complex of mPEG-PLLD with plasmid DNA showed much more increased solubility than that of PLL or polyethylenimine. 3. In order to remove small traces of excess reagents, the precipitate was recrystalized in pure ethanol. Subsequently, the product is obtained as a purewhite crystalline powder. 4. Sometimes, insoluble precipitates were formed because of low solubility. It is recommended to conduct the fourth coupling reaction with half of the total amount of synthesized third generation copolymer. 5. For polymers, about 100 pmol is required to get an effective MALDI signal.

32

Choi and Park

Fig. 7. Effect of PLL and mPEG-PLLD (G = 4) on 293 cells viability. Relative viability is expressed considering the absorbance at 570 nm intact cells as 100%. Each data point is the average ± SD of six different experiments. 6. Usually, it is better to dry the mixture in vacuo, rather than just air-drying for the preparation of an adequate specimen for MALDI-TOF-MS. 7. In the case of mPEG-PLLD (G = 3), any complete retardation that results from full complexation between polymer and DNA was not observed even though the charge ratio of polymer:DNA was increased to 4. 8. 1.89 × 1015 negative charges are present per 1.0 µg plasmid DNA; the copolymer has 1.26 × 1015 charges/1.0 µg. 9. Usually, about 10 ng plasmid DNA is required for the AFM imaging. The amount of the DNA seems to be critical for obtaining the best image. 10. Try to load the complex solutions in the center of mica surface. After 1–2 min, do not disturb the center and remove the residual solution carefully. 11. 19.2 kDa PLL and 25 kDa polyethylenimine were used as control reagents. All the polymers were mixed with DNA at a charge ratio of 4. 12. MTT solution is sensitive to light. After filtering the stock solution through a 0.2 µm syringe filter, it must be preserved in a light-protected tube. It is advisable to wrap the tube with aluminum foil.

References 1. Ledley, F. D. (1995) Nonviral gene therapy: the promise of genes as pharmaceutical products. Hum. Gene Ther. 6, 1129–1144. 2. Kukowska-Latallo, J. F., Bielinska, A. U., Johnson, J., Spindler, R., Tomalia, D. A., and Baker, J. R., Jr. (1996) Efficient transfer of genetic material into mammalian cells using Starbust polyamidoamine dendrimers. Proc. Natl. Acad. Sci. USA 93, 4897–4902.

PEG-b-PLL Dendrimer with pDNA

33

3. Boussif, O., Lezoualc’h, F., Zanta, M. A., Mergny, M. D., Scherman, D., Demeneix, B., and Behr, J.-P. (1995) A versatile vector for gene and oligonucleotide transfer into cells in culture and in vivo: polyethylenimine. Proc. Natl. Acad. Sci. USA 92, 7297–7301. 4. Kataoka, K., Kwon, G. S., Yokoyama, M., Okano, T., and Sakurai, Y. (1993) Block copolymer micelles as vehicles for drug delivery. J. Controlled Release 24, 119–132. 5. Wolfert, M. A., Schacht, E. H., Toncheva, V., Ulbrich, K., Nazarova, O., and Seymour, L. W. (1996) Characterization of vectors for gene therapy formed by self-assembly of DNA with synthetic block co-polymers. Hum. Gene Ther. 7, 2123–2133. 6. Kataoka, K., Togawa, H., Harada, A., Yasugi, K., Matsumoto, T., and Katayose, S. (1996) Spontaneous formation of polyion complex micelles with narrow distribution from antisense oligonucleotide and cationic block copolymer in physiological saline. Macromolecules 29, 8556–8557. 7. Kabanov, A. V., Vinogradov, S. V., Suzdaltseva, Y. G., and Alakhov, V. Y. (1995) Water-Soluble block polycations as carriers for oligonucleotide delivery. Bioconjugate Chem. 6, 639–643. 8. Nguyen, H. -K., Lemieux, P., Vinogradov, S. V., Gebhart, C. L., Guérin, N., Paradis, G., et al. (2000) Evaluation of polyether-polyethyleneimine graft copolymers as gene transfer agents. Gene Ther. 7, 126–138. 9. Lee, R. J. and Huang, L. (1996) Folate-targeted, anionic liposome-entrapped polylysine-condensed DNA for tumor cell-specific gene transfer. J. Biol. Chem. 271, 8481–8487. 10. Bayer, E. and Mutter, M. (1979) The Liquid-phase method for peptide synthesis, in The Peptides (Gross, E. and Meienhofer, J., eds.), vol. 2, Academic, New York, pp. 285–332. 11. Denkewalter, R. G., Kolc, J., and Lukasavage, W. J. (1981) US Patent 4,289,872, Sept 15. 12. Roy, R., Zanini, D., Meunier, S. J., and Romanowska, A. (1993) Solid-phase synthesis of dendritic sialoside inhibitors of influenza A virus haemagglutinin. J. Chem. Soc. Chem. Commun. 1869–1872. 13. Chapman, T. M., Hillyer, G. L., Mahan, E. J., and Shaffer, K. A. (1994) Hydraamphiphiles: novel linear dendritic block copolymer surfactants. J. Am. Chem. Soc. 116, 11,195–11,196. 14. Choi, J. S., Lee, E. J., Choi, Y. H., Jeong, Y. J., and Park, J. S. (1999) Poly(ethylene glycol)-block-poly(L-lysine) dendrimer: novel linear polymer/ dendrimer block copolymer forming a spherical water-soluble polyionic complex with DNA. Bioconjugate Chem. 10, 62–65.

Cationic Methacrylate Polymers/DNA Complexes

35

4 Water-Soluble Cationic Methacrylate Polymers for Nonviral Gene Delivery Gert W. Bos, Daan J. A. Crommelin, and Wim E. Hennink 1. Introduction The aim of gene therapy is to treat inherited or acquired genetic deficiencies (e.g., cystic fibrosis) or viral diseases (e.g., hepatitis B, HIV) by introduction of DNA encoding a therapeutic protein or a specific virus antigen, respectively, into the nucleus of the target cell. Because naked DNA will barely pass cellular membranes, a carrier system is required for transfection (1–4). Cationic polymers, which condense DNA by ionic interaction, form a promising class of nonviral transfection agents. Well-known examples of these polymers are DEAE dextran, poly(L-lysine), poly(ethylenimine), and poly(2-[dimethylamino]ethyl methacrylate) (pDMAEMA) (5–10). In order to achieve transfection, a plasmid must be delivered into the nucleus, which requires cellular uptake of polymer– DNA complexes, generally referred to as “polyplexes” (11), which most likely occurs via endocytosis, followed by endosomal escape and transport to the nucleus. The polyplex must dissociate, either in the cytosol or in the nucleus, which may be a critical step in the transfection process. pDMAEMA (Fig. 1), a water-soluble cationic polymer, may form selfassembled stable nanoparticles with plasmids at a polymer:plasmid ratio above 2 (w/w) (8–10). The positively charged polyplexes have a size of ~150–200 nm (8–10). The transfection efficiency vs the pDMAEMA:plasmid ratio forms a bell-shaped curve, with a maximum at 3–6 (w/w), depending on the transfected cell type. At low pDMAEMA:plasmid ratios, addition of more polymer to the complexes results in smaller complexes with higher transfection efficiency; at higher pDMAEMA:plasmid ratios, the slight cytotoxicity of the polymer probably results in cell death (8–10). A pronounced effect of the molecular weight of the polymer on the transfection efficiency is observed. An increasing molecular weight results in an increasing number of transfected cells (9). Dynamic From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

35

36

Bos, Crommelin, and Hennink

Fig. 1. Poly(2-[dimethylamino]ethyl methacrylate) (pDMAEMA).

light scattering experiments show that high-molecular weight polymers (mol wt > 300 kDa) are able to condense DNA effectively (particle size, 150–200 nm). In contrast, when plasmid is incubated with low-molecular-weight pDMAEMA, large complexes are formed (size 0.5–1.0 µm). Using optimal conditions, the degree of transfection in vitro (approx 30% of the treated cells are transfected) is higher than that achieved with commercially available cationic lipids such as 1,2-dioleoyl3-trimethylammonium-propane (DOTAP) and Lipofectamine (12). Cells grown in vivo can be transfected ex vivo with pDMAEMA-based polyplexes with an overall transfection efficiency of ~1–2% (13). To further improve transfection performance, random copolymers were synthesized. Random copolymers of DMAEMA with ethoxytriethylene glycol methacrylate (triEGMA), N-vinylpyrrolidone (NVP), methyl methacrylate (MMA), and methacrylic acid of different molecular weights and compositions (comonomer fraction up to 66 mol%) are able to bind DNA, yielding polyplexes (14,15; and Bos and Hennink, unpublished data). However, for random copolymers of DMAEMA with triEGMA, NVP, and MMA, the polymer:plasmid ratio at which small complexes (size 200 nm) are formed increases with the mol fraction of the comonomer (14,15). A copolymer with 20 mol% MMA shows a reduced transfection efficiency and a substantial increased cytotoxicity compared with a homopolymer of the same molecular weight (14). On the other hand, for triEGMA, NVP, and methacrylic acid, the cytotoxicity of the copolymers, either complexed with DNA or in the free form, is inversely proportional to the mol fraction of these comonomers. This reduction is even more than what can be expected based on the DMAEMA mol fraction in the copolymer (14 and Bos and Hennink, unpublished data). NVPDMAEMA copolymers synthesized by polymerization to high conversion show excellent DNA binding, condensing characteristics and transfection capabilities. This is ascribed to a synergistic effect of DMAEMA-rich copolymers and NVP-rich copolymers present in this system on the complex formation with plasmid DNA (14).

Cationic Methacrylate Polymers/DNA Complexes

37

Temperature-sensitive copolymers of DMAEMA and N-isopropylacryl amide (NIPAAm) of various monomer ratios and molecular weights have been evaluated as carrier systems for DNA delivery (16). All copolymers, even with a low DMAEMA content of 15 mol%, were able to bind to DNA at 25°C. Light-scattering measurements indicate that complexation is accompanied by precipitation of the copolymer in the complex caused by a drop of the lower critical solution temperature of the copolymer. The copolymer:plasmid ratio, at which complexes with a size of approx 200 nm are formed, shows a positive correlation with the NIPAAm content of the copolymer and is independent of molecular weight of the copolymer. Complexes containing copolymers of low molecular weight or high NIPAAm content prepared at 25°C aggregate rapidly when the temperature is raised to 37°C. On the other hand, complexes containing copolymers of high molecular weight or lower NIPAAm content are stable at 37°C. The cytotoxicity of the complexes decreases with increasing NIPAAm content and is independent of molecular weight of the copolymer. The transfection efficiency as a function of the copolymer:plasmid ratio shows a bell-shaped curve. The copolymer:plasmid ratio at which the transfection efficiency is maximal rises with increasing NIPAAm content; the maximum transfection efficiency drops with increasing NIPAAm content of the copolymer (16). Besides pDMAEMA, a number of structural analogs, differing in the side chain groups, have been evaluated as transfectant (17,18). Almost all studied cationic methacrylate/methacrylamide polymers are able to condense the structure of pDNA, yielding small polyplexes (100–300 nm) and a slightly positive c potential. However, the transfection efficiency and the cytotoxicity of the polymers differ widely: the highest transfection efficiency and cytotoxicity are observed for pDMAEMA itself. Assuming that polyplexes enter cells via endocytosis, pDMAEMA apparently has advantageous properties to escape the endosome. A possible explanation is that, because of its average pKa value of 7.5, pDMAEMA is partially protonated at physiological pH and behaves as a proton sponge. This might cause a disruption of the endosome, which results in the release of both the polyplexes and cytotoxic endosomal/lysosomal enzymes into the cytosol. In contrast, the analogs of pDMAEMA studied have a higher average pKa value and have, consequently, a higher degree of protonation and a lower buffering capacity, which might be associated with a lower tendency to destabilize the endosome, resulting in both lower transfection efficiency and a lower cytotoxicity (17,18). Furthermore, structural analysis by molecular modeling techniques suggests that, of all studied polymers, pDMAEMA has the lowest number of interactions with DNA. The authors therefore hypothesize that the superior transfection efficiency of pDMAEMA-containing polyplexes can be ascribed to an intrinsic property of pDMAEMA to destabilize endosomes combined with an easy dissociation of the polyplex once present in the cytosol and/or the nucleus (17,18).

38

Bos, Crommelin, and Hennink

Functionalization of pDMAEMA, e.g., by coupling of an antibody (-fragment) or another ligand such as poly(ethylene glycol) is feasible by using a random copolymers of DMAEMA with aminoethyl methacrylate (AEMA) (19). The percentage of incorporated primary amino groups can be controlled by the feed ratio of AEMA:DMAEMA, and is usually below 10 mol%. The ligands can be coupled to the amine groups directly or, for example, via a thiol group. In this case, following the synthesis of the copolymer, protected thiol groups are introduced in a derivatization step with N-succinimidyl 3-(2-pyridyldithio) propionate and subsequent treatment with dithiothreitol. The obtained thiolated p(DMAEMA-co-AEMA) can be conjugated to antibodies and other ligands, e.g., the nuclear localization signal decapeptide Gly-Pro-Lys-Lys-Lys-Arg-Lys-ValGlu-Asp-NH2, via a disulfide linkage. In general, the coupling efficiencies are high (>90%) (19). The thiolated polymers can also be used to determine the apparent kinetic rate constants between plasmid DNA and the nonviral carrier polymers using surface plasmon resonance spectrometry (20). In this case, the polymers are attached to the gold layer through the thiol groups. Freeze-drying of these gene delivery systems can be performed using a controlled two-step drying process and sucrose as lyoprotectant (21). Freezedrying is shown to be an excellent method to preserve the size and transfection potential of pDMAEMA–plasmid complexes (8,22), even after aging at 40°C (23). The concentration of the sugars is an important factor affecting both the size and transfection capability of the complexes after freeze-drying and freezethawing. However, the type of lyoprotectant (sugar) used is of minor importance (24). The DNA topology has been shown to affect pDMAEMAmediated transfection: Circular forms of DNA (supercoiled and open-circular) show higher transfection activity than linear forms (25). Recently, the possibilities and limitations of autoclaving, filtration, and a combination of both methods for sterilization of pDMAEMA-based gene transfer complexes have been assessed (26). Agarose gel electrophoresis and circular dichroism spectroscopy shows that filtration of polyplexes does not change the topology and integrity of the DNA. Moreover, a full preservation of the transfection potential of the filtered polyplexes was observed. Precoating of the filter with polyplexes reduces the material loss and the loss of transfectivity. In contrast, autoclaving dramatically affects physical characteristics of polyplexes, resulting in a complete loss of transfection potential. Addition of sucrose to the preparation protects DNA present in pDMAEMA– DNA complexes, to some extent, from degradation during autoclaving, but the transfection potential is not retained. Filtration or autoclaving of polymer alone does not result in substantial loss of polymer integrity and material, or in decreased transfection potential. Naked DNA can easily be sterilized by filtration as well, although some DNA may be lost. Therefore, separate

Cationic Methacrylate Polymers/DNA Complexes

39

sterilization of polymer and DNA stock solutions, followed by aseptic formation and handling of polyplexes, may be an acceptable and preferred alternative to filtration of polyplexes (26). In this chapter, protocols are provided for the synthesis of pDMAEMA homo- and copolymers in water or toluene, gel permeation chromatography (GPC) analysis of the synthesized polymers, the preparation of pDMAEMA–DNA polyplexes, and subsequent determination of the plasmid integrity after formulation, sterilization of pDMAEMA-based polyplexes, a standard transfection protocol, and determination of cell viability and of transfection efficiency. 2. Materials 2.1. Common 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

12. 13. 14.

Water purified by reversed osmosis (RO water). pH-meter or pH-indicator strips. 70% Ethanol. Sodium chloride, pro analysi (pa). Sodium hydroxide (4.0, 1.0, and 0.1 N). Phosphate buffered saline (PBS): 3.6 mM KH2PO4, 6.4 mM Na2HPO4, and 145 mM NaCl, pH 7.2. 4-(2-Hydroxy-ethyl)-1-piperazine ethane sulfonic acid (HEPES). 0.22-µm Filters. Vortex mixer. Eppendorf tubes sterilized by autoclaving. Cell culture equipment: Biohazard safety cabinet or laminar airflow cabinet, equipped with burner and suction device, CO2 incubator 37°C, Bürker-Türk or Bürker counting chamber (bright lined), microscope (e.g., ×25 objective, ×10 ocular), water bath at 37°C, centrifuge. P-20, P-100, P-200, and P-1000 Gilson pipets with sterile tips. 5- and 10-mL pipets. 15- and 50-mL centrifuge tubes.

2.2. Synthesis of pDMAEMA Homo- and Copolymers in Water 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

DMAEMA stabilized with 0.15% tert-butyl-hydroxytoluene (Fluka). Ammonium peroxodisulphate (APS) (Fluka). 37% Hydrochloric acid, pa. 100 mL Infusion bottles with plastic lids and silicon rubber septa (e.g., Emergo). Needles. Nitrogen–vacuum exchange system. Water bath at 60°C. Vacuum distillation equipment. 1–10 mL Glass syringes, glass pipets. Dialysis tubing, mol wt cut-off 12–14 kDa (e.g., Medicell). Freeze-dryer.

40

Bos, Crommelin, and Hennink

2.3. Synthesis of pDMAEMA in Toluene 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

DMAEMA stabilized with 0.15% tert-butyl-hydroxytoluene (Fluka). _,_'-Azoisobutyronitril (AIBN) (Fluka). Toluene, pa. Petroleum ether, boiling range 40–60°C, technical quality. Dichloromethane. 100-mL Infusion bottles with plastic lids and silicon rubber septa (e.g., Emergo). Needles. Nitrogen–vacuum exchange system. Water bath at 60°C. Vacuum stove with temperature control at 40–60°C. Vacuum distillation equipment. Glass syringes, glass pipets.

2.4. GPC of Water-Soluble pDMAEMA Homo- and Copolymers 1. 2. 3. 4. 5. 6.

7. 8. 9. 10. 11.

Tris-(hydroxymethyl)-aminomethane (Tris). NaNO3, pa. 60% HNO3, pa. 0.22-µm Filter (e.g., Millipore GVWP04700). Degassing setup consisting of vacuum pump and ultrasound bath. High-performance liquid chromatography (HPLC) system, consisting of a pump, an autoinjector, a refractive index detector, and software with GPC option and GPC column set option. GPC column set (e.g., Shodex K80P, KB-80M, and KB802); for molecular weight higher than 10,000 g/mol, the KB802 column can be skipped. 1-mL Poly(propylene) (PP) shell vial with snap cap (Alltech). 2-mL Syringes. Nylon 13-mm syringe filters, 0.45 µm (Alltech). 15-mL PP tubes with cap.

2.5. Amplification and Purification of Plasmid 1. Bacterial strain producing an appropriate marker gene. 2. Plasmid purification kit (e.g., Giga-kit from Qiagen).

2.6. Preparation of pDMAEMA–DNA Polyplexes with Low DNA Concentration 1. Plasmid in TE-buffer (see Subheading 3.4.). 2. pDMAEMA (see Subheadings 3.1. and 3.2.).

2.7. Preparation of pDMAEMA–DNA Polyplexes with High DNA Concentration 1. 2. 3. 4.

Glacial acetic acid, pa. Sucrose. Plasmid in TE-buffer (see Subheading 3.4.). pDMAEMA (see Subheadings 3.1. and 3.2.).

Cationic Methacrylate Polymers/DNA Complexes

41

2.8. Integrity of Plasmid in Polyplexes 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

Pronase. Tris-(hydroxymethyl)-aminomethane (Tris). Glacial acetic acid, pa. Ethylenediamine tetraacetic acid (EDTA). Ethidium bromide (EtBr) or SYBR Green I nucleic acid stain (Molecular Probes). Bromophenol blue. Glycerol. Marker DNA (e.g., h DNA [Gibco] digested with restriction enzyme PstI). Polyaspartic acid (pAsp) (Sigma, mol wt 50,000). Electrophoresis apparatus. UV detection system. Conical flask. Microwave oven.

2.9. Sterilization of pDMAEMA-Based Gene Transfer Complexes 1. 0.22-µm Filters (e.g., cellulose acetate) (Schleicher & Schull GmbH, Dassel, Germany).

2.10. Standard Transfection Protocol 1. 2. 3. 4. 5. 6. 7.

OVCAR 3, COS 7 or other cells. Completed culture medium, depending on the particular cell line. 10X Trypsin–EDTA (Gibco). Trypan blue for vital staining (Sigma). 96-Well plates. Centrifuge. Eight-Channel pipet.

2.11. Determination of Cell Viability 1. Sodium 3'-[1-(phenylaminocarbonyl)-3,4-tetrazolium]-bis(4-methoxy-6-nitro) benzene sulfonic acid hydrate (XTT) (Gibco). 2. N-methyl dibenzopyrazine methylsulfate (PMS) (Gibco). 3. Microplate reader.

2.12. Determination of Transfection Efficiency 1. 2. 3. 4. 5. 6. 7. 8. 9.

Tris(hydroxymethyl)-aminomethane (Tris). 37% Hydrochloric acid, pa. Triton X-100. MgCl2·6H2O. NaH2PO4·H2O. Na2HPO4·2H2O. o-Nitrophenyl-`-D-galactopyranoside (ONPG) (Sigma). `-Galactosidase (Sigma). 0.25% Glutaraldehyde solution.

42 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.

Bos, Crommelin, and Hennink NaH2PO 4·HO. Na2HPO4·2H2O. K4Fe(CN)6·3H2O. K3Fe(CN)6. MgCl2·6H2O. 5-Bromo-4-chloro-3-indolyl-`-D-galactopyranoside (X-Gal, Gibco). Dimethylsulfoxide (DMSO): 99.9% spectrophotometric grade. Microplate reader. Gelatin powder. 85% Glycerol. Phenol. Microwave oven.

3. Methods

3.1 Synthesis of pDMAEMA Homo- and Copolymers in Water Synthesis of pDMAEMA (Fig. 1) is achieved by radical polymerization (9,10), and it can be performed in either an acidified aqueous solution (this procedure) or in toluene (see Subheading 3.2.). Because the ester bond of DMAEMA can be chemically hydrolyzed in water (27), synthesis of pDMAEMA in toluene may be preferred. However, pDMAEMA is routinely synthesized in water, which is especially useful for the synthesis of copolymers of DMAEMA and tolueneinsoluble monomers. After synthesis, the polymers are characterized by nuclear magnetic resonance (NMR) to determine if all monomer has reacted, and in the case of copolymers, also the copolymer composition. Furthermore, the polymers are characterized by GPC to determine the average molecular weight and the molecular weight distribution (see Subheading 3.3.). 1. The monomer DMAEMA should be vacuum-distilled shortly before synthesis to remove the radical scavenger tert-butylhydroxytoluene. After distillation, allow nitrogen into the distillation equipment. In some cases, the comonomer should be vacuum-distilled as well (see Note 1). 2. Dissolve 1.0 g initiator APS in 15 mL water in a 50-mL flask. 3. An initial DMAEMA concentration of 20% (v/v) for the synthesis is recommended. The molecular weight of the polymer will depend on the ratio of monomer and initiator (M:I ratio) used. Several typical examples of mixtures are given in Table 1. Add in 100-mL infusion bottles with septum appropriate amounts of water, 37% HCl and DMAEMA (in that order). Use a glass syringe to transfer DMAEMA into the flask. When copolymers are synthesized, the molar quantities of DMAEMA and the comonomer should match the quantity of DMAEMA given in the Table 1. Adjust the pH of the solution to 5.0 to prevent hydrolysis of the esterbond of DMAEMA. In this stage, chain transfer agents such as mercaptoethanol can be added to obtain low-molecular weight polymers with a functional endgroup. Add the appropriate amount of APS solution (see Table 1).

Cationic Methacrylate Polymers/DNA Complexes

43

Table 1 Reaction Components and Expected Molecular Weight 37% HCl (mL)

Water (mL)

DMAEMA (mL)

APS (mL)

M:I ratio

Mpa (expected) (106 g/mol)

4 4 4 4

34 35 35.5 35.8

10 10 10 10

2 1 0.5 0.2

100 200 400 1000

0.5 ± 0.1 1.5 ± 0.2 2.6 ± 0.4 5.0 ± 1.0

aM p

is the molecular weight at the peak of the GPC curve and is therefore independent of the peak integration. Therefore, it is useful to compare different batches of polymer.

4. 5.

6. 7.

8.

Degas the solution in the bottle with a needle through the septum with the use of a nitrogen–vacuum exchange system until first signs of evaporation of water can be seen (typically, approx 15 s), then flush nitrogen into the flask. Repeat this twice to make sure that the air atmosphere is completely replaced by nitrogen atmosphere. Incubate the flasks under shaking conditions in a water bath at 60°C for 20–22 h. Cool down the viscous solution and transfer it into dialysis tubes. If necessary, dilute the solution with water. Beware osmotic pressure: The volume of the solution may increase threefold during dialysis. Remove the buffer by dialysis for several days at 4°C using at least 5 L of water in the external compartment. Freeze-dry the polymer solution. Determine the weight of empty 500-mL or 1-L round-bottomed flasks and transfer the dialyzed polymer solution into these flasks. Do not fill a flask with more than one-third of its volume. Freeze the polymers in a film on the inner wall of the flask, while rotating the flask in liquid nitrogen. Freeze-dry the polymers overnight to remove the water. Determine the weight of the flasks with the polymer and subtract the weight of the empty flask to estimate the yield of the polymerization (usually 80–90%). Characterize the polymer by GPC (see Subheading 3.3.), and by NMR (see also Notes 2 and 3). 1H-NMR: Dissolve about 20 mg (p)DMAEMA in 1 mL CDCl . Chemical shifts: 3 a. Monomer (DMAEMA): 6.09 (s, 1H, =CH), 5.54 (s, 1H, =CH), 4.23 (t, 2H, OCH2), 2.60 (t, 2H, NCH2), 2.28 (s, 6H, N(CH3)2), 1.93 (s, 3H, C=C-CH3). b. Polymer (pDMAEMA): 4.05 ppm (b, 2H, OCH2), 2.55 ppm (b, 2H, NCH2), 2.28 ppm (b, 6H, N(CH3)2), 2.18–2.30 ppm (bm, 2H, C-CH2), 0.75–2.30 ppm (bm, 3H, C-C-CH3).

3.2. Synthesis of pDMAEMA in Toluene 1. The monomer DMAEMA should be vacuum-distilled 1 d before synthesis, as described in Subheading 3.1.1. 2. Dissolve 0.2428 g initiator AIBN in 5 mL toluene (296 mM). 3. An initial DMAEMA concentration of 20% (v/v) is recommended for the synthesis. The molecular weight of the polymer will depend on the M:I mol ratio used. Several typical examples of mixtures are given in Table 2.

44

Bos, Crommelin, and Hennink Table 2 Reaction Components and Expected Molecular Weight DMAEMA (mL)

Toluene (mL)

AIBN (mL)

M:I ratio

10 10 10 10

38 39 39.5 39.8

2 1 0.5 0.2

100 200 400 1000

Mpa (expected) (104 g/mol) 4±1 6±1 12 ± 2 22 ± 4

aM is the molecular weight at the peak of the GPC curve, and is therefore p independent of the peak integration. Therefore, it is useful to compare different batches of polymer.

4. In one flask, add appropriate amounts of DMAEMA and toluene. Add the appropriate amount of AIBN solution using a glass syringe. Degas the solution in the bottle as described in Subheading 3.1.3. 5. Incubate the flask under shaking conditions in a water bath at 60°C for 20–22 h. 6. After incubation, cool down the viscous solution and slowly add the solution drop-wise (e.g., by gravity force through a Pasteur pipet) under stirring conditions to a 10-fold bigger volume of petroleum ether to precipitate the polymer. 7. Discard the petroleum ether/toluene, wash the precipitate twice with 50–100 mL petroleum ether, and transfer the polymer into a suitable storage can or flask. Dissolve the last part of the polymer in the flask with dichloromethane and transfer the solution into the can or flask. Allow remnants of the organic solvents to evaporate in the hood overnight. 8. Dry the polymers inside a vacuum stove at 40–60°C until no toluene can be smelled and the weight of the flask with the polymer does not further decrease. If complete removal of toluene is essential, the polymer can be dissolved in water and subsequently dialyzed and freeze-dried as described in Subheadings 3.1.5–3.1.7. 9. Characterize the polymers by GPC (see Subheading 3.3.), and by NMR. For 1H-NMR assignments, see Subheading 3.1.8. and Notes 2 and 3.

3.3. GPC of pDMAEMA Homo- and Copolymers This instruction is intended for the molecular weight determination of pDMAEMA and its copolymers. 1. Prepare eluent: 0.7 M NaNO3, 0.1 M Tris-HCl, pH 7.2. Dissolve 119 g NaNO3 and 24.2 g Tris in 1.8 L RO water. Adjust pH to 7.2 with 60% HNO3, and fill up to 2 L with RO water. Filter through 0.22-µm filter, Millipore, GVWP04700. Alternative eluents (check specs of columns for compatibility) are 10 mM NaCl; 5 mM NH4+CH3COO–, pH 5.5; 0.8 M NaNO3. 2. Set HPLC system to appropriate specs, e.g., for the series of Shodex columns: flow 0.5 mL/min; detector sensitivity 256 (for concentration range of 0.5–2.0 mg/mL); detector temperature, 35°C; column oven temperature 30°C; run time 40 min; injection vol 100 µL.

Cationic Methacrylate Polymers/DNA Complexes

45

Table 3 Mp-Values of Dextran Standards for GPC Measurement of pDMAEMA Stnd. Dex1

Dex2

Dex3

Dex4

Dex5

Dex6

Mr~ a 180 991 50,000 750,000 342 1000 80,000 1,800,000 504 5000 150,000 667 12,000 270,000 82 25,000 670,000 M w=4,900,000 M p=4,500,000 M n=1,500,000

Mpb

Fluka no.

180 991 43,500 560,000 342 1080 66,700 1,450,000 504 4440 123,600 667 9890 196,000 829 21,400 401,300

63416 31420 31426 63418 31416 31421 31427 63430 31417 31422 63422 31418 31423 63417 31419 31425 31428

Name Sucrose Maltohexaose Dextran 50,000 Dextran 750,000 D(+)-Maltose Dextran 1000 Dextran 80,000 Dextran 800,000 Maltotriose Dextran 5000 Dextran 150,000 Maltotetraose Dextran 12,000 Dextran 270,000 Maltopentose Dextran 25,000 Dextran 670,000 Dextran 4,900,000

aM

r~ is number on bottle. bUse M values in the weight-loading p

table of your GPC software.

3. Calibrate the GPC columns: a. For every mixture (see Table 3) dissolve 5 mg of each dextran in 5 mL eluent. Allow dextran to dissolve overnight at room temperature. Store the solutions at 4°C. b. Measure and calculate a calibration curve. 4. For new types of copolymers, prepare a stock solution by adding 50 mg polymer to 10 mL eluent in a 15-mL PP tube. Put the tube on a rotating wheel or roller bank at 4°C for 4 d to dissolve completely. Filter the sample through a 0.45-µm syringe filter. 5. Prepare solutions with concentrations of 5, 4, 3, 2, 1, and 0.5 mg/mL by pipeting 1.0, 0.8, 0.6, 0.4, 0.2, and 0.1 mL sample respectively, in a PP vial and add eluent to obtain a final volume of 1.0 mL. 6. Inject 100 µL sample onto the GPC system and determine the molecular weight and area of the different concentrations using the software. Plot area vs concentration. This should give a linear relationship. Furthermore, Mw and M n should be independent of the concentration. 7. For validated polymers, routine analysis can be performed at 1.0 mg/mL.

46

Bos, Crommelin, and Hennink

3.4. Amplification and Purification of Plasmid Plasmids are prepared from bacterial cultures grown in the presence of a selective agent such as an antibiotic, using for instance the standard Luria Bertani medium (27). The plasmid is isolated from the culture using, e.g., an Endofree Qiagen Giga Plasmid Kit (Qiagen) according to the instructions of the manufacturer. After amplification and purification, the plasmid is collected in a Tris-EDTA buffer (TE buffer, 10 mM Tris-HCl, 1 mM EDTA, pH 8.0) to obtain a plasmid stock solution, which can be stored at 4°C for at least 1 yr; the concentration of the DNA (OD280 nm) should be checked every 3 mo. However, high EDTA concentrations may induce aggregation of the pDMAEMA–plasmid complexes at high DNA concentration. Therefore, for preparation of polyplexes with a high DNA concentration (see Subheading 3.6.), after purification, the plasmid should be taken up by a low amount of TE buffer to avoid high EDTA concentrations in the final sample. The plasmid concentration of the plasmid stock solution should therefore be higher than 3 µg/µL (see also Note 4).

3.5. Preparation of pDMAEMA–DNA Polyplexes with Low DNA Concentration Small and stable pDMAEMA–plasmid complexes (polyplexes) of low concentrations can be prepared either in HEPES buffer or in HEPES-buffered saline (HBS) (see Fig. 2 and Note 5). This procedure describes the preparation of 3:1 pDMAEMA–plasmid polyplex formulation (250 µL, containing 2.5 µg plasmid and 7.5 µg pDMAEMA). For polyplexes using a different polymer:DNA ratio, the concentration of the polymer solution should be varied. Preparation of polyplexes with higher DNA concentrations is described in Subheading 3.6. (see also Note 6). 1. Switch on the biohazard or laminar airflow safety cabinet 30 min before use and clean the surface with 70% ethanol. After opening and before closing tubes, bottles, and so on, quickly put cap and neck through a burner’s blue flame. 2. Prepare the following solutions: a. HEPES (HBS) containing 20 mM HEPES (pH 7.4) and 150 mM NaCl: Add 2.383 g HEPES and 4.383 g NaCl to 480 mL water, and adjust the pH to 7.4 with approx 1 mL 4 M NaOH. Adjust the total volume to 500 mL with water. Sterilize the buffer through a filter with 0.22-µm pore size (autoclaving causes some precipitation) (see Note 7). b. HEPES buffer (pH 7.4). Add 2.383 g HEPES to 480 mL water and adjust the pH to 7.4 with NaOH. Adjust the total volume to 500 mL with water. Sterilize the buffer through a filter with 0.22-µm pore size (see Note 7). c. Plasmid stock solution. After amplification and purification, the plasmid is collected in a Tris-EDTA buffer (see Subheading 3.4.). The plasmid stock solution can be stored at 4°C for at least 1 yr; the concentration of the DNA (OD280 nm) should be checked every 3 mo.

Cationic Methacrylate Polymers/DNA Complexes

47

Fig. 2. Preparation of polyplexes (see Subheading 3.5.).

3. 4. 5. 6. 7. 8. 9.

d. Stock solution of pDMAEMA, 5 mg/mL in HEPES buffer. Dissolve 50 mg pDMAEMA in 10 mL HEPES buffer (see step 2b). Allow the pDMAEMA to dissolve for 3–4 d at 4°C, preferably using a rotating device. Sterilize the solution through filters with 0.22-µm pore sizes. Dilute the plasmid stock solution with HBS or HEPES to obtain 200 µL containing 10 µg plasmid (50 µg/mL). Dilute the pDMAEMA stock solution with HBS or HEPES to obtain 800 µL containing 30 µg polymer (37.5 µg/mL). Add 200 µL diluted pDMAEMA solution to 50 µL diluted plasmid solution. A 250-µL sample containing 2.5 µg plasmid and 7.5 µg pDMAEMA is obtained. Immediately vortex the sample for 5 s. Repeat steps 5 and 6 to obtain three batches of polyplexes. After 30 min, the complexes are ready for use. Characterize the complexes by dynamic light scattering (DLS, particle size determination). Polyplexes prepared from newly synthesized polymers should be measured at least threefold. To characterize the polyplexes by electrophoretic mobility (c-potential), they should be prepared using HEPES buffer to dilute the plasmid and pDMAEMA stock solutions instead of HBS, and at least 1 mL should be prepared (see Note 8).

3.6. Preparation of pDMAEMA–DNA Polyplexes with High DNA Concentration pDMAEMA–plasmid complex dispersions (polyplexes) of high concentrations can be prepared without aggregation (29). The complexes are prepared at low pH (20 mM acetate buffer, pH 5.7), low ionic strength, and high viscosity (20 w/v% sucrose solution). This procedure describes the preparation of 3:1 pDMAEMA:plasmid complex dispersion (1 mL, containing 150 µg plasmid and 450 µg pDMAEMA). For polyplexes using a different polymer/DNA ratio, the concentration of the polymer solution should be varied.

48

Bos, Crommelin, and Hennink

1. Switch on the biohazard or laminar airflow safety cabinet 30 min before use and clean the surface with 70% ethanol. After opening and before closing tubes, bottles, and so on, put cap and neck through a burner’s blue flame. 2. Prepare the following solutions: a. HBS as in Subheading 3.5., step 2a. b. Acetate buffer (20 mM acetate). Add 0.57 mL glacial acetic acid to 480 mL water, and adjust the pH to 5.7 with 1.0 N and 0.1 N NaOH. Adjust the total volume to 500 mL with water. Sterilize the buffers through filters with 0.22-µm pore sizes. Opened bottles should be stored in the refrigerator and used within 1 mo. c. Sucrose–acetate buffer (40 w/v% sucrose in 20 mM acetate). Dissolve 4.00 g sucrose (2.6g) in 5 mL acetate buffer (3.2g). Add acetate buffer until the volume is 10 mL. Sterilize the buffers through filters with 0.22-µm pore sizes. d. Plasmid stock solution, as in Subheading 3.5., step 2c. However, high EDTA concentrations induce aggregation of the pDMAEMA–plasmid complexes at high DNA concentration. Therefore, after purification, the plasmid should be taken up by a low amount of TE buffer to avoid high EDTA concentrations in the final sample. The plasmid concentration of the plasmid stock solution should therefore be higher than 3 µg/µL (see Note 4). e. Stock solution of pDMAEMA as in Subheading 3.5., step 2d, using HBS instead of HEPES. 3. Dilute the plasmid stock solution with acetate buffer to obtain 100 µL containing 150 µg plasmid. 4. Add 500 µL sucrose–acetate buffer to the 100 µL diluted plasmid solution. 5. Add 310 µL acetate buffer to 90 µL pDMAEMA stock solution. A solution of 450 µg pDMAEMA in 400 µL buffer is obtained. 6. Add the 400 µL diluted pDMAEMA solution to the 600 µL diluted plasmid solution. A 1-mL sample containing 150 µg plasmid and 450 µg pDMAEMA is obtained. 7. Immediately vortex the sample gently for 5 s. 8. After 30 min, the complexes are ready for use. Alternatively, the polyplexes can be lyophilized. 9. Characterize the complexes by DLS (particle size determination) and by electrophoretic mobility (c-potential) (see Notes 9 and 10). Polyplexes prepared from newly synthesized polymers should be measured at least threefold.

3.7. Integrity of Plasmid in Polyplexes Electrophoresis through agarose is the standard method used to separate and identify DNA fragments. In this procedure, a 0.7% (w/v) agarose gel concentration is used, but different gel concentrations can be used for a broad size range of DNA molecules (see ref. 27). In order to run the plasmid on an agarose gel, the DNA needs to be dissociated from the polymer. This can be done using polyaspartic acid (see Fig. 3) (18,30). 1. Prepare the following solutions: a. 50X TAE buffer: Dissolve 242 g Tris-base, 57.1 mL glacial acetic acid, 100 mL 0.5 M EDTA (pH 8.0) in RO water to make 1 L.

Cationic Methacrylate Polymers/DNA Complexes

49

Fig. 3. Agarose gel of pCMVlacZ (lane 1), pDMAEMA–pCMVlacZ polyplexes before (lane 2) and after (lane 3) dissociation with a 100-fold excess of polyaspartic acid (see Subheading 3.7.).

2.

3.

4.

5. 6.

b. Working buffer: 1X TAE buffer: Dilute 50X TAE buffer 50-fold in RO water. c. Ethidium bromide in water (10 mg/mL). The stock solution should be stored in a bottle wrapped in aluminum foil at room temperature. d. 10X Sample buffer: 0.4% bromophenol blue, 10 mM EDTA, 50% glycerol in water. e. Dissociation buffer: Dissolve 100 mg pAsp in 5 mL PBS. Add dissociation buffer to the polyplex samples, using a 100-fold (w/w) excess of pAsp to DNA. Vortex and incubate 30 min at ambient temperature for freshly prepared polyplexes (within 2–3 h after preparation) or between 1 and 5 d at 4°C for older polyplexes. Add pronase to 1X TAE in a conical flask; e.g., for a small size gel 0.35-g pronase in 50 mL 1X TAE. Heat in a microwave until the agarose dissolves. Cool the solution to about 60°C and, when desired, add ethidium bromide to a final concentration of 0.5 µg/mL. Position combs and pour agarose solution into the mold, check for air bubbles. Leave gel to set (30–45 min at ambient temperature), carefully remove the combs and mount the gel in the electrophoresis tank. Add working buffer 1X TAE to cover the gel. Mix the samples of DNA with the sample buffer. The maximum amount of DNA that can be applied to a slot depends on the number of fragments in the sample and their sizes (15). Marker DNA (1–2 µg) can be loaded as well. Run the gel (e.g., 100 V, current unlimited) until the bromophenol blue has migrated ±6 cm through the gel. If ethidium bromide was present in the gel, examine the gel directly by UV light, or after destaining for 30 min in water. Otherwise, stain the gel by soaking it for 30–45 min in a solution of ethidium bromide (0.5 µg/mL) in water or SYBR Green I in water and destain the gel for 30–45 min in water before examining the gel.

3.8. Sterilization of pDMAEMA-Based Gene Transfer Complexes To sterilize polyplexes, filter at least 6 mL polyplex preparation through 0.22-µm filters and discard the first 3 mL of sample (26). Alternatively, filter

50

Bos, Crommelin, and Hennink

Fig. 4. Transfection protocol (see Subheading 3.9.).

at least 5 mL polymer solution (5 mg/mL) and DNA solution (10 µg/mL) separately. Before preparation of polyplexes from these sterilized solutions, the polymer and DNA concentration should be determined using optical densities 225 and 260, respectively, as measure for the concentration.

3.9. Standard Transfection Protocol This procedure describes the in vitro transfection of OVCAR-3 cells (see Note 11) and COS-7 cells (see Note 12), using cationic carriers, such as the polymer, pDMAEMA, as a vector (Fig. 4; see refs. 8–10 for typical results). The protocol can be used for all adherent cell types. For suspension type cells, cells need to be spun down before changing the medium at a speed suitable for the cell type to be transfected. 1. Switch on the biohazard or laminar airflow safety cabinet 30 min before use and clean the surface with 70% ethanol. After opening and before closing tubes, bottles, and so on, put cap and neck through a burner’s blue flame. 2. Prepare the following solutions: a. HBS as in Subheading 3.5., step 2a. b. Fetal bovine serum (FBS) (Integro), sterile and heat-inactivated (30 min at 56°C). Store at –20°C. c. Plasmid stock solution as obtained in Subheading 3.4. d. Stock solutions of carriers, e.g., 1–5 mg/mL in HBS, filtered through 0.22-µm filters, and stored at –20°C (see Subheading 3.5., step 2d). e. Completed Dulbecco's modified Eagle's medium (DMEM) culture medium: Prepare completed medium normally used for culturing the cells to be transfected. Serum can be present in the medium. f. OVCAR-3 or COS-7 cell suspension, or, alternatively, other cell suspensions. g. 0.05% Trypsin–0.02% EDTA solution (1X). Defrost the 10X trypsin–EDTA in a 37°C water bath. Add 10 mL 10X trypsin–EDTA to 90 mL PBS and store at 4°C. Tenable for 3 mo when stored at 4°C.

Cationic Methacrylate Polymers/DNA Complexes

3. 4.

5.

6.

7. 8. 9.

10. 11. 12.

51

h. 0.5% Trypan blue solution. Weigh 0.5 g trypan blue in a biohazard safety cabinet. Dissolve trypan blue in 100 mL PBS and filtrate through filtration paper to remove possible crystals. Divide in portions and store at –20°C. At room temperature, the solution is tenable for about 1 mo. Passage OVCAR-3 or COS-7 cells or other cell types 3–4 d before the transfection experiment. Detach the cells with trypsin–EDTA solution, and determine the cell number and cell viability using trypan blue. Pipet 50 µL cell suspension into an Eppendorf tube and add 50 µL 0.5% trypan blue solution. Bring some cell–trypan blue mixture into a counting chamber and count (microscope) in a number of squares the number of uncolored (vital) and blue (dead) cells in duplicate. Count at least 100 cells. When more than 10 cells are counted per square, dilute the cell suspension and count again. Determine the number of vital cells. Add 100 µL 1.1 × 105 cells/mL/well of a 96-well plate and incubate for 24 h at 37°C (the cells should be approx 60% confluent). Perform each transfection in at least six wells on two different 96-well plates and use one 96-well plate for estimation of the transfection efficiency and the other for estimation of the cell number and viability. Next day, prepare dilute solutions of the carriers and plasmids to be tested from their stock solutions in HBS. For instance: a. pDMAEMA: Prepare 0.6 mL 0.6 mg/mL from the stock solution, and make of this a series of twofold dilutions in HBS (see first column in Table 4). b. DNA: Dilute the plasmid stock solution to a final concentration of 50 µg/mL in HBS. Prepare polyplexes in threefold by pipeting 50 µL 50 µg/mL plasmid solution in a well of a 96-well plate, followed by 200 µL carrier solution. As reference, add 3× 200 µL 37.5 µg/mL pDMAEMA in HBS to 50 µL 50 µg/mL DNA in wells of a 96-well plate using the same batch of polymer for all experiments. Incubate the polyplexes for 30–60 min at room temperature. Meanwhile, aspirate culture medium from the cells (by pipeting or decanting) and add 100 µL of warm (37°C) completed DMEM. Use medium with at least 10% serum in this step. Overlay the cells (in a 96-well plate) with 100-µL polyplexes (1 µg plasmid/ well), vortex 15 s carefully, and incubate for 1 h at 37°C in the incubator. Aspirate transfection media, and add 100 µL warmed, completed DMEM into each well, and culture the cells in the incubator for 44 ± 2 h at 37°C. Estimate the transfection results. The number and viability of the cells can be estimated by the XTT colorimetric assay (see Subheading 3.10.). Using a lacZ reporter gene, the transfection efficiency can be estimated by the `-galactosidase content with the use of the substrate ONPG, or alternatively, an X-Galbased histochemical color assay can be used to estimate the number of transfected cells (see Subheading 3.11.). Express the data relative to the transfection results mediated by the reference carrier solution of pDMAEMA. The incubation times with polyplexes and the subsequent culture time can be adjusted for different cell types.

52

Bos, Crommelin, and Hennink Table 4 Ratios of Carrier:DNA in Polyplex Preparation Carrier solution (µg/mL)

DNA solution (µg/mL)

600 300 150 75 37.5 18.75 9.38 4.69

50 50 50 50 50 50 50 50

Carrier/DNA (w/w) 48 24 12 6 3 1.5 0.75 0.38

3.10. Determination of Cell Viability To determine the influence of carrier–plasmid complexes on cell viability and proliferation, the number of viable cells can be measured using an XTT colorimetric assay (33; see also Note 13). The assay is based on the cleavage of the yellow tetrazolium salt, XTT, to form an orange formazan dye by dehydrogenase activity in active mitochondria. Therefore, this conversion only occurs in living cells. The formazan dye formed is soluble in aqueous solutions and is directly quantified using a microplate reader. This procedure can also be applied for suspension type cells. 1. Switch on the biohazard or laminar flow safety cabinet 30 min before use and clean the surface with 70% ethanol. After opening and before closing tubes, bottles, and so on, put cap and neck through a burner’s blue flame. 2. Prepare the following solutions: a. XTT stock solution, 1 mg/mL: Dissolve XTT in plain RPMI 1640 at 1 mg/mL. Fill out in 10-mL portions and store in the dark at –20°C. b. Electron-coupling reagent: Dissolve PMS in PBS at 0.383 mg/mL (1.25 mM). Fill out in 250-µL portions and store in the dark at –20°C. c. XTT-solution: Freshly prepared each experiment. Thaw XTT stock solution and electron-coupling reagent in a water bath at 37°C. Mix each vial to obtain a clear solution. For one 96-well plate, add 100 µL electron-coupling reagent to 5 mL XTT stock solution. Mix by vortexing. 3. Make a calibration curve with living cells: Detach adherent cells with trypsin– EDTA solution, determine the cell number and cell viability using trypan blue, according to the procedure described in Subheading 3.9.4. Dilute the cells to 1 × 106 cells/mL (cell stock) with completed culture medium. Make a calibration curve in a 96-well plate, e.g., as in Table 5. 4. Incubate the cells for 2 h at 37°C in the CO2-incubator. 5. Add 50 µL XTT-solution (step 2c) per well. 6. Incubate for 1 h at 37°C in the CO2-incubator.

Cationic Methacrylate Polymers/DNA Complexes

53

Table 5 Cell Viability Calibration Curve Cells/well A B C D E F G H

100 µL cell stock 100 µL culture medium + 100 µL cell stock 100 µL culture medium + 100 µL B 100 µL culture medium + 100 µL C 100 µL culture medium + 100 µL D 100 µL culture medium + 100 µL E 100 µL culture medium + 100 µL F, mix and pipet 100 µL out µL culture medium

100 × 103 50 × 103 25 × 103 12.5 × 103 6.25 × 103 3.13 × 103 1.56 × 103 0

7. Measure the absorbance at 490 nm with a reference wavelength of 655 nm. 8. With the calibration curve of living cells, the number of viable cells can be calculated.

3.11. Determination of Transfection Efficiency The enzyme activity of `-galactosidase, expressed upon transfection of OVCAR-3 cells, COS-7 cells, or other cell types with lacZ reporter gene, can be determined using the substrate o-nitrophenyl-`-D -galactopyranoside (ONPG; 28). For suspension type cells, cells need to be spun down before changing the medium, at a speed suitable for the cell type to be transfected. 1. Switch on the biohazard or laminar airflow cabinet 30 min before use and clean the surface with 70% ethanol. 2. Prepare the following solutions: a. Lysis buffer, containing 50 mM Tris–HCl buffer (pH 8.0), 150 mM NaCl, and 1% Triton X-100. For 100 mL, dissolve 0.6057 g Tris, 0.8766 g NaCl in 80 mL water, add 1.0 mL Triton X-100, and adjust the pH to 8.0 with (diluted in water) hydrochloric acid. Fill up to 100 mL with water. b. 100X MgCl2-solution (0.1 M): Dissolve 0.2033 g MgCl2·6H2O in 10 mL water. c. 0.2 M NaH2PO4: Dissolve 5.52 g NaH2PO4·H2O in 200 mL water. The solution is tenable for 3 mo at room temperature. d. 0.2 M Na2HPO4: Dissolve 17.80 g Na2HPO4·2H2O in 500 mL water. The solution is tenable for 3 mo at room temperature. e. 0.1 M sodium phosphate buffer (pH 7.4): Mix 9.5 mL 0.2 M NaH2PO4 (step 2c) with 40.5 mL 0.2 M Na2HPO4 (step 2d) and 50 mL water. f. 10 mg/mL ONPG solution: Dissolve 1000 mg ONPG in 100.0 mL 0.1 M sodium phosphate buffer, pH 7.5 (step 2e). Fill out in 5-mL aliquots in 5- or 10-mL plastic tubes and store in the dark at –20°C. g. 1000 U/mL `-galactosidase stock solution: Add 1.0 mL lysis buffer to 1000 U `-galactosidase. Fill out in 20-µL aliquots in Eppendorf cups and store in the dark at –20°C. Use immediately after thawing and do not freeze and store more than once.

54

Bos, Crommelin, and Hennink

Table 6 Galactosidase Calibration Curve Sample A B C D E F G H

3. 4. 5.

6. 7.

8.

Lysis buffer (µL) 100 100 100 100 100 100 100

Enzyme solution 200 µL 5 U/mL 100 µL A 100 µL B 100 µL C 100 µL D 100 µL E 100 µL F

`-galactosidase (mU/well) 100 50 25 12.5 6.25 3.13 1.56 0

h. ONPG-staining solution: For estimating enzyme activity in all wells of a 96well plate, mix 18.5 mL 0.1 M sodium phosphate buffer, pH 7.5, 200 µL 100X Mg solution, and 1.35 mL 10 mg/mL ONPG solution. Aspirate the media from the cells in the well plate and wash once with 100 µL cold PBS (4°C). Add 20 µL cold lysis buffer (4°C) to the cells in a well, and incubate for 20 min at 4°C. Meanwhile, make the different dilutions for the calibration curve of `-galactosidase from the 1000 U/mL stock solution (2 g). Add 10 µL 1000 U/mL to 90 µL lysis buffer and an aliquot of 50 µL of this mixture to 950 µL lysis buffer, to obtain an enzyme solution of 5 U/mL. Make 7× twofold dilutions of this final solution in a 96-well plate to obtain standard enzyme solutions, according to the scheme in Table 6. Add 20 µL of the standard enzyme solutions A–H to empty wells in the 96-well plate, which is used for, e.g., transfection. Add 180 µL ONPG-staining solution (2 h) to each well (both standards and samples). Incubate the well plate(s) at 37°C and measure the absorbance at 415 nm in a microplate reader, using the absorbance at 655 nm as a reference, until the highest standard shows an absorbance of 2 or more (approx 30–45 min). Calculate the `-galactosidase amount of a sample in mU/well by comparison with the linear standard curve.

Alternatively, cells containing `-galactosidase upon transfection of cells with a lacZ reporter gene can be histochemically stained in vitro (Fig. 5). For suspension type cells, cells need to be spun down before changing the medium at a speed suitable for the cell type to be transfected. 1. Switch on the biohazard or laminar airflow cabinet 30 min before use and clean the surface with 70% ethanol. 2. Prepare the following solutions:

Cationic Methacrylate Polymers/DNA Complexes

55

Fig. 5. X-Gal stained OVCAR3 cells (see Subheading 3.10.). The blue dots are nuclei of cells transfected with pCMVlacZ. a. Working fixative (0.25% glutaraldehyde solution): Made fresh each time by diluting a 25% glutaraldehyde solution 100-fold with PBS. b. 0.2 M NaH2PO4: Dissolve 5.52 g NaH2PO4·H2O in 200 mL RO water. Stored at room temperature, the stock solution is tenable for 3 mo. c. 0.2 M Na2HPO4: Dissolve 17.80 g (0.1 mol) Na2HPO4·2 H2O in 500 mL RO water. Stored at room temperature, the stock solution is tenable for 3 mo. d. 0.1 M Sodium phosphate buffer (pH 7.4): For 100 mL, mix 9.5 mL 0.2 M NaH2PO4 (step 2b) with 40.5 mL 0.2 M Na2HPO4 (step 2c) and 50 mL RO water. Stored at 4°C, it is tenable for 3 mo. e. 0.5 M Potassium hexacyanoferrat(II): Dissolve 0.564 g K4Fe(CN)6·3H2O in 2.67 mL water in a 15-mL centrifuge tube. Protected from light (e.g., by aluminum foil) and stored at 4°C, the stock solution is stable for 3 mo. f. 0.5 M Potassium hexacyanoferrat(III): Dissolve 0.415 g K3Fe(CN)6 in 2.52 mL water in a 15-mL centrifuge tube. Protected from light and stored at 4°C, the stock solution is stable for at least 3 mo. g. 1.0 M MgCl2: Dissolve 2.03 g MgCl2·6H2O in 10 mL water in a 15-mL centrifuge tube. Stored at room temperature, it is tenable for 3 mo. h. 40 mg/mL X-Gal in DMSO: Dissolve 400 mg X-Gal in 10 mL DMSO (99.9% spectrophotometric grade). Fill out in 500-µL portions (in Eppendorf tubes or glass vials, not in polycarbonate or polystyrene ones) and store in the dark at –20°C. After using part of it, freeze the remainder for use again (may become slightly yellow). Do not repeat thawing and freezing more than 5×. i. X-Gal stain solution, freshly made each time. Required volume is approx 0.1 mL/cm2: 1 mL/well of 6-well plate (9.4 cm2), 0.1 mL/well of 24 well

56

3. 4. 5. 6. 7. 8. 9.

10.

Bos, Crommelin, and Hennink plate (2 cm2), 50 µL/well of a 96-well plate (0.38 cm 2). For 10 mL staining solution, mix: 100 µL 0.5 M (K4Fe(CN) 6·3H2O) (step 2e), 100 µL 0.5 M (K3Fe[CN]6) (step 2f), 20 µL 1 M MgCl2 (step 2g), and 250 µL X-Gal solution (step 2h). Bring the volume to 10 mL with 9.53 mL 0.1 M sodium phosphate buffer (step 2d). Aspirate media from monolayers of cells to be assayed. Rinse gently but thoroughly with PBS (for a 96 well plate, 100 µL per well). After aspirating PBS, overlay the cells with fixative (step 2a) and incubate at 4°C for 5 min (for a 96-well plate, 75 µL/well). Aspirate the fixative and rinse gently twice with PBS at room temperature (for a 96-well plate, 100 µL/well). Aspirate the PBS and overlay the cells with X-Gal stain solution (step 2i) (for a 96-well plate, 50 µL/well). Replace the microplate lids and incubate overnight at 37°C. View cells under microscope. Cells expressing `-galactosidase are detectable by their overall blue color or their blue colored nuclei (when the lacZ-plasmid used has an NLS-signal). Count the number of blue cells/well. After washing with PBS, the well plates can be stored at 4°C for 2 d. The plates can be stored for a longer period of time (several years at room temperature) after preservation (see Subheading 3.12.).

3.12. Conservation of X-Gal Colored Cells After performing a color reaction with cells, the cells can be conserved. At elevated temperatures, the conservation solution is a transparent fluid. It congeals when cooled to room temperature. Overlaid with this substance, cells are covered with a jelly transparent film at room temperature. Therefore, they are not susceptible to bacteria, and can be stored at room temperature for infinite time. It is still possible to study them under a microscope. 1. 2. 3. 4. 5. 6. 7. 8.

Suspend 17.4 g gelatin powder in 71 mL water. Add 128 mL 85% glycerol and mix. Add 2.17 g phenol and dissolve it by stirring. Heat at 60°C (or in a microwave oven) until the mixture is a clear, transparent, and air-bubble-free fluid. Fill out in 50-mL portions and store the tubes at room temperature. Prior to use heat a portion in the microwave oven until the substance is fluid. Wash the cells once with 100 µL PBS/well. Add mixture to the wells with the help of a pipette (just enough to cover the cells). Shortly after adding the fluid, it turns into a jelly transparent film protecting the cells.

4. Notes 1. If necessary, the distilled solution can be stored at –20°C for at least 1 wk. 2. High molecular weight pDMAEMA synthesized in water must be dissolved in

Cationic Methacrylate Polymers/DNA Complexes

3. 4. 5.

6.

7.

8.

9. 10. 11. 12. 13.

57

D2O (20 mg/mL) because it does not dissolve in CDCl3. Chemical shifts are much broader than in CDCl 3: a. Monomer (DMAEMA): 5.95 (s, 1H, =CH), 5.52 (s, 1H, =CH), 4.08 (t, 2H, OCH2), 2.58 (t, 2H, NCH2), 2.09 (s, 6H, N[CH3]2), 1.72 (s, 3H, C=C-CH3). b. Polymer (pDMAEMA): 3.95–4.35 ppm (b, 2H, OCH2), 3.00–3.60 ppm (b, 2H, NCH2), 2.50–2.90 ppm (b, 6H, N[CH3]2), 1.30–2.10 ppm (bm, 2H, CCH2), 0.50–1.10 ppm (bm, 3H, C-C-CH3) (see Note 3). The monomer peaks at 6.09 (5.95 in D 2O) ppm and 5.54 (5.52 in D2O) ppm must be absent in polymer samples. If the concentration of the plasmid intended for use is too low, the plasmid can be precipitated in 70% ethanol (28) and redissolved at higher concentration. HBS (step 2a) should be used for in vitro transfection, whereas HEPES (step 2b) is to be preferred for determining the c-potential. Therefore, stock solutions of pDMAEMA can best be prepared in HEPES buffer. The size of polyplexes is influenced by the quantities of DNA and polymer solutions that are mixed, as well as by the speed at which the polymer is added to the DNA solution. Also, the buffer (HEPES or HBS) influences the size of the polyplexes. Closed bottles with sterilized solutions can be stored at room temperature and may be used up to 1 yr after preparation. Opened bottles should be stored in the refrigerator and used within 2 mo. The obtained particle size and c-potential of the pDMAEMA–plasmid complexes prepared in HEPES buffer are approx 100 nm and 30 mV, respectively; the obtained particle size of the pDMAEMA–plasmid complexes prepared in HBS is approx 160 nm. For the refractive index and viscosity, values of 1.363 and 1.546, respectively, must be used (when measured at 25°C). The obtained particle size and c-potential of the pDMAEMA–plasmid complexes are approx 175 nm and 28 mV, respectively. OVCAR-3 cells are human ovarian carcinomas (adenocarcinoma) and originate from American Type Culture Collection (ATCC) (31,32). COS-7 cells are CV-1 cell lines derived from African green monkey kidney cells. The cell line originates from ATCC. Alternatively, cell numbers can be counted using a Bürker counting chamber.

Acknowledgments The authors wish to thank Dr. Jong-Yuh Cherng, Dr. Petra Van de Wetering, Dr. Herre Talsma, Nancy M. E. Schuurmans-Nieuwenbroek, Dr. Wouter L. Hinrichs, and OctoPlus B.V. for the use of their data and protocols. The European Union (grant PL 970002) funds G. W. Bos. References 1. Behr, J. P. (1993) Synthetic gene transfer vectors. Acc. Chem. Res. 26, 274–278. 2. Ledley, F. D. (1997) Pharmaceutical approach to somatic gene therapy. Pharm. Res. 13, 1595–1614.

58

Bos, Crommelin, and Hennink

3. Godbey, W. T., Wu, K. K., and Mikos, A. G. (1999) Poly(ethylenimine) and its role in gene delivery. J. Control. Release 60, 149–160. 4. De Smedt, S. C., Demeester, J., and Hennink, W. E. (2000) Cationic polymer based gene delivery systems. Pharm. Res. 17(2), 113–126. 5. Kawai, S. and Nishizawa, M. (1984) New procedure for DNA transfection with polycation and dimethyl sulfoxide. Mol. Cell. Biol. 4, 1172–1174. 6. Curiel, D. T., Wagner, E., Cotten M., Birnstiel, M. L., Agarwal, S., Li, C. M., Loechel, S., and Hu, P. C. (1992) High-efficiency gene transfer mediated by adenovirus coupled to DNA- complexes. Hum. Gene Ther. 3, 147–154. 7. Boussif, O., Lezoualc’h, F., Zanta, M. A., Mergny, M. D., Scherman, D., Demeneix, B., and Behr, J. P. (1995) A versatile vector for gene and oligonucleotide transfer into cells in culture and in vivo: polyethylenimine. Proc. Nat. Acad. Sci. USA 92, 7297–7301. 8. Cherng, J. Y., Van de Wetering, P., Talsma, H., Crommelin, D. J. A., and Hennink, W. E. (1996) Effect of size and serum proteins on transfection efficiency of poly([2-dimethylamino] ethyl methacrylate)-plasmid nanoparticles. Pharm. Res. 13, 1038–1042. 9. Van de Wetering, P., Cherng, J. Y., Talsma, H., and Hennink, W. E. (1997) Relation between transfection efficiency and cytotoxicity of poly(2-(dimethylamino)ethyl methacrylate)/plasmid complexes. J. Control. Release 49, 59–69. 10. Hennink, W. E. and Van de Wetering, P. (1997) Cationic polyacrylates and poly(alkyl) acrylates and acrylamides for use as carriers of nucleic acid in the transformation of animal cells. PCT Int. Appl. Patent no. WO9715680A1, p. 37. 11. Felgner, P. L., Barenholz, Y., Behr, J. P., Cheng, S. H., Cullis, P., Huang L., et al. (1997) Nomenclature of synthetic gene delivery systems. Human Gene Ther. 8, 511–512. 12. Fonseca, M. J., Storm, G., Hennink, W. E., Gerritsen, W. R., and Haisma, H. J. (1999) Cationic polymeric gene delivery of `-glucuronidase for doxorubicin prodrug therapy. J. Gene Med. 1, 404–417. 13. Van de Wetering, P., Schuurmans-Nieuwenbroek, N. M. E., Hennink, W. E., and Storm, G. (1999) Comparative transfection studies of human ovarian carcinoma cells in vitro, ex vivo and in vivo with poly(2-(dimethylamino)ethyl methacrylate)-based polyplexes. J. Gene Med. 1, 156–165. 14. Van de Wetering, P., Schuurmans-Nieuwenbroek, N. M., Van Steenbergen, M. J., Crommelin, D. J., and Hennink, W. E. (2000) Co-polymers of 2(dimethylamino)ethyl methacrylate with ethoxytriethylene glycol methacrylate or N-vinyl-pyrrolidone as gene transfer agents. J. Control. Release 64, 193–203. 15. Van de Wetering, P., Cherng, J. Y., Talsma, H., Crommelin, D. J., and Hennink, W. E. (1998) 2-(Dimethylamino) ethyl methacrylate based (co) polymers as gene transfer agents. J. Control. Release 53, 145–153. 16. Hinrichs, W. L., Schuurmans-Nieuwenbroek, N. M., Van de Wetering, P., and Hennink, W. E. (1999) Thermosensitive polymers as carriers for DNA delivery. J. Control. Release 60, 249–259.

Cationic Methacrylate Polymers/DNA Complexes

59

17. Van de Wetering, P., Moret, E. E., Schuurmans-Nieuwenbroek, N. M., Van Steenbergen, M. J., and Hennink, W. E. (1999) Structure-activity relationships of water-soluble cationic methacrylate/methacrylamide polymers for nonviral gene delivery. Bioconjugate Chem. 10, 589–597. 18. Arigita, C., Zuidam, N. J., Crommelin, D. J., and Hennink, W. E. (1999) Association and dissociation characteristics of polymer/DNA complexes used for gene delivery. Pharm. Res. 16, 1534–1541. 19. Van Dijk-Wolthuis, W. N., van de Wetering, P., Hinrichs, W. L., Hofmeyer, L. J., Liskamp, R. M., Crommelin, D. J., and Hennink, W. E. (1999) A versatile method for the conjugation of proteins and peptides to poly(2-[dimethylamino]ethyl methacrylate). Bioconjugate Chem. 10, 687–692. 20. Wink, T., De Beer, J., Hennink, W. E., Bult, A., and Van Bennekom, W. P. (1999) Interaction between plasmid DNA and cationic polymers studied by surface plasmon resonance spectrometry. Anal. Chem. 71, 801–805. 21. Talsma, H., Cherng, J., Lehrmann, H., Kursa, M., Ogris, M., Hennink, W. E., Cotten, M., and Wagner, E. (1997) Stabilization of gene delivery systems by freeze-drying. Int. J. Pharm. 157, 233–238. 22. Cherng, J. Y., Van de Wetering, P., Talsma, H., Crommelin, D. J., and Hennink, W. E. (1997) Freeze-drying of poly([2-dimethylamino]ethyl methacrylate)-based gene delivery systems. Pharm. Res. 14, 1838–1841. 23. Cherng, J. Y., Talsma, H., Crommelin, D. J. A., and Hennink, W. E. (1999) Long term stability of poly([2-dimethylamino]ethyl methacrylate)-based gene delivery systems. Pharm. Res. 16, 1417–1423. 24. Cherng, J. Y., Van de Wetering, P., Talsma, H., Crommelin, D. J., and Hennink, W. E. (1999) Stabilization of polymer-based gene delivery systems. Int. J. Pharm. 183, 25–28. 25. Cherng, J. Y., Schuurmans-Nieuwenbroek, N. M., Jiskoot, W., Talsma, H., Zuidam, N. J., Hennink, W. E., and Crommelin, D. J. (1999) Effect of DNA topology on the transfection efficiency of poly([2-dimethylamino]ethyl methacrylate)plasmid complexes. J. Control. Release 60, 343–353. 26. Bos, G. W., Trullas-Jimeno, A., Jiskoot, W., Crommelin, D. J. A., and Hennink, W. E. (2000) Sterilization of poly(dimethylamino) ethyl methacrylate-based gene transfer complexes. Int. J. Pharm. 15, 211(1–2), 79–88. 27. Van de Wetering, P., Zuidam, N. J., Van steenbergen, M. J., Van der Houwen, O. A. G. J., Underberg, W. J. M., and Hennink, W. E. (1998) A mechanistic study of the hydrolytic stability of poly(2-[dimethyl]aminoethyl methacrylate). Macromolecules 31, 8063–8068. 28. Sambrook, J., Fritsch, E. F., Maniatis, T., eds. (1989) Molecular Cloning, A Laboratory Manual, 2nd ed., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York. 29. Cherng, J. Y., Talsma, H., Verrijk, R., Crommelin, D. J., and Hennink, W. E. (1999) Effect of formulation parameters on the size of poly-([2dimethylamino]ethyl methacrylate)-plasmid complexes. Eur. J. Pharm. Biopharm. 47, 215–224.

60

Bos, Crommelin, and Hennink

30. Katayose S. and Kataoka K. (1997) Water-soluble polyion complex associates of DNA and poly(ethyleneglycol)-poly(L-lysine) block co-polymer. Bioconjugate Chem. 8, 702–707. 31. Hamilton, T. C., Young, R. C., McKoy, W. M., Grotzinger, K. R., Green, J. A., Chu, E. W., et al. (1983) Characterization of a human ovarian carcinomal cell line (NIH:OVCAR-3) with androgen and estrogen receptors. Cancer Res. 43, 5379–5389. 32. Hamilton, T. C., Foster, B. J., Grotzinger, K. R., McKoy, W. M., Young, R. C., and Ozols, R. F. (1983) Development of drug sensitive and resistant human ovarian cancer cell lines. A model system for investigating new drugs and mechanisms of resistance. Proc. Am. Assoc. Cancer Res. 24, 313. 33. Scudiero, D. A., Shoemaker, R. H., Paull, K. D., Monks, A., Tierney, S., Nofziger, T. H., et al. (1988) Evaluation of a soluble tetrazolium/formazan assay for cell growth and drug sensitivity in culture using human and other tumor cell lines. Cancer Res. 48, 4827–4833.

Steric Stabilization of DNA Complexes

61

5 Stabilization of Polycation–DNA Complexes by Surface Modification with Hydrophilic Polymers David Oupicky, Martin L. Read, and Thierry Bettinger 1. Introduction Polycation–DNA complexes represent promising synthetic vectors for gene delivery, showing good transfection activities in vitro and safety in vivo. However, simple polycation–DNA complexes suffer from several disadvantages that limit their potential usefulness in vivo. Advances in this field thus rely on better control of the structure, colloidal, and surface properties of condensed DNA particles. Physicochemical stability of simple polycation–DNA complexes is limited. The complexes are usually stable in water or in low-concentration buffers, because of their positive charge; in physiological saline, however, this charge is suppressed by the screening effect of salt and, as a result, the complexes aggregate. Their positive charge (beneficial for transfection efficiency in vitro [1]) also means that they are prone to interaction with various proteins, which represents another major drawback for their use in vivo. When distribution kinetics on nonviral vectors based on polycations (or lipids) have previously been examined in vivo, they were invariably cleared quickly from the bloodstream (2). This is thought to result from binding of serum proteins, which promotes the uptake of complexes by the liver. Systemic application of these complexes would require (1) formation of small particles (~100 nm) facilitating diffusion, extravasation through vascular fenestration, and cellular uptake (endocytosis), (2) formation of neutral particles to minimize nonspecific interactions with proteins and negatively charged surfaces of cell membranes. Apart from reducing nonspecific interactions in vivo, possible specific uptake of the complexes by components of reticuloendothelial system should also be avoided. (3) Cell-specific binding and internalization are also required, and are achievable by their enhanced circulation times in the blood. From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

61

62

Oupicky, Read, and Bettinger

In order to overcome the previously mentioned problems of simple polycation– DNA complexes, a range of block and graft copolymers of polycations and hydrophilic nonionic polymers, (such as polyethylene glycol (PEG), dextran, and poly(N-[2-hydroxypropyl]methacrylamide) (PHPMA) (3–7) have been synthesized and tested. Complexes formed with these polymers, indeed, showed improved stability against aggregation and interaction with proteins, compared to simple polycation–DNA complexes, but no improvement of in vivo properties (blood circulation times) was observed (3,7,8). Using block copolymers in constructing nonviral gene delivery vectors is discussed in Chapter 3 by Choi and Park. An alternative approach, inspired by experiences with liposomes and nanoparticles in the field of drug delivery systems, has been adopted recently into the field of nonviral gene delivery vectors. The approach relies on surface modification of preformed complexes with hydrophilic polymers. The rationale behind this approach, also called “steric stabilization,” is to provide the complexes with a protecting surface polymer layer, which significantly affects interparticle forces. First, it influences van der Waals attractive forces; and second, it can give rise to repulsion between the particles. The magnitude of repulsion arising from the presence of the layer depends on the density with which it covers the surface; the more thinly it is spread, the smaller its effectiveness in preventing the particles from approaching one another (9). The concept of steric stabilization has been exploited in the field of drug delivery as a way of preventing the particles from aggregation, but also as a way of reducing their interactions with opsonins and components of the reticuloendothelial system, which effectively increased circulation times of these particles (10). Steric stabilization of particles or liposomes generally increases biocompatibility, reduces immune response, increases in vivo stability, and delays clearance by the reticuloendothelial system. Modification of DNA complexes (as with liposomes, proteins) can thus provide many benefits for both in vivo and in vitro applications. Covalent coupling of hydrophilic polymers to complexes can alter their surface and solubility properties, effectively masking the intrinsic character of the surface. Although a range of suitable polymers is available, DNA complexes have so far been modified (coated) with the most frequently used polymer, PEG, and also by PHPMA (11–16). The purpose of this chapter is to describe basic methods used for surface coating of DNA complexes with hydrophilic polymers and their subsequent characterization. 2. Materials 1. 2. 3. 4.

Plasmid DNA. 10 mM HEPES buffer, pH 7.4. Poly-L-lysine (PLL). Polyethylenimine (PEI).

Steric Stabilization of DNA Complexes 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.

63

mPEG(5000)-SPA. N-(2-hydroxypropyl) methacrylamide (HPMA). _,_'-Azoisobutyronitrile (AIBN). 3-Mercaptopropionic acid (3-MPA). Acetone. Ether. Methanol. Dimethylformamide (DMF). N,N'-Dicyclohexylcarbodiimide (DCCD). 4-Dimethylaminopyridine (DMAP).

3. Methods 3.1. Coating of DNA Complexes An important prerequisite for coating DNA complexes is the availability of free primary (or secondary) amino groups. Polycations based on tertiary and quaternary amino groups have been successfully used for gene delivery in vitro, but they cannot be as easily chemically modified (17,18). Among the described polycations containing primary or secondary amino groups, the most widely used are PLL and branched PEI.

3.2. DNA Complexes Suitable for Coating and Their Formation In order to prepare small complexes (~100 nm) it is vital to keep the concentration of DNA and salts low during formation. If possible, the complexes should be made in water or in low-concentration buffers (10–50 mM). PLL, if obtained from Sigma, is supplied as a hydrobromide salt and complexes can be prepared directly in water. PEI, however, is usually not supplied as a salt, and needs to be protonated first by dissolving in appropriate buffer (HEPES, pH 7.4). PLL complexes with amine:phosphate (N:P) ratios higher than 1.2 and PEI complexes with N:P higher than 3 are generally suitable for coating reactions as DNA is fully condensed, and the complexes already show positive c potential, indicating the presence of free amino groups on the surface of the complexes (12,19,20). In order to increase the size stability of the complexes, however, it is advisable to work at higher N:P ratios (~2 for PLL and 4–5 for PEI), because complexes around neutrality (N:P ~1 for PLL:DNA and N:P ~2–3 for PEI) tend to aggregate more easily. The following protocol describes a standard method of preparing PLL and PEI complexes in the authors’ laboratories. 1. Prepare solution of DNA in 10 mM HEPES, pH 7.4, at a of concentration 20 µg/mL. (All methods described in this chapter have been performed with plasmid DNA– circular 5 kb expression vector containing a cytomegalovirus promoter-driven luciferase reporter and ampicillin resistance gene.)

64

Oupicky, Read, and Bettinger

2. Add PLL solution (2.5 mg/mL in water) in a single addition to achieve the desired N:P ratio and mix solution quickly by inverting several times. 3. Alternatively, prepare solution of DNA at 40 µg/mL in 10 mM HEPES, pH 7.4, and mix with equal volume of PEI in 10 mM HEPES, pH 7.4, containing the calculated amount of polycation. 4. Leave the complexes for at least 30 min prior to use.

3.3. Coating with Monofunctional Polymers 3.3.1. PEG-Coated Complexes The presence of primary amino groups in the PLL–DNA and PEI–DNA complexes predetermines the choice of the chemistry useful for coating reactions. As mentioned in the introduction the most widely used polymer in modification of particulate drug delivery systems is PEG. Amino-reactive analogs of PEG suitable for coating reactions are available in a wide range of molecular weights and types of reactive groups. They can be obtained from Shearwater Polymers (www.swpolymers.com), who specialize in synthesis of various PEG derivatives. The most commonly used reagents are those based on the active esters of carboxylic acid groups and carbonates. Carboxyl groups activated as N-hydroxysuccinimidyl (NHS) esters are highly reactive toward amine nucleophiles, forming an amide bond and releasing free NHS. The reaction of NHS esters with thiol or hydroxyl groups is possible, but does not yield stable conjugates. Several types of mPEG-NHS are available, but we prefer to use succinimidyl esters of methoxy-PEG propionic acid (mPEG-SPA) because of their increased hydrolytic stability compared to other derivatives. mPEG-SPA, with mol wt 2000, 5000, and 20,000, can be obtained from Shearwater Polymers; PEG(5000)-SPA is also available from Fluka (no. 85969). If a limited amount of accessible amino groups is available, fork-like (PEG)2-NHS containing two mPEG molecules attached to amino groups of lysine, is the reagent of choice to increase PEG density on the coated complexes (also available from Shearwater Polymers). The reaction should be performed in buffers of pH 7.0–9.0 (hydrolysis halflife for mPEG-SPA at pH 8.0 at 25°C is 16.5 min). We have found pH 7.4–8.0 suitable. Increasing pH above 8.0 (together with higher salt concentration) can lead to aggregation of the complexes because of charge screening. The following protocol describes coating of DNA complexes (Fig. 1) using monofunctional mPEG(5000)-SPA (for structure, see Fig. 2). 1. Prepare PLL–DNA complexes at N:P ratio 2, or PEI–DNA complexes at N:P ratio 3–5, in 10–50 mM HEPES, pH 7.4–7.8 (final DNA concentration, 20 µg/mL). 2. Dissolve mPEG(5000)-SPA in water at concentration of 20 mg/mL prior to use; alternatively, make stock solution in dry dimethylsulfoxide (DMSO) and store at –80°C.

Steric Stabilization of DNA Complexes

65

Fig. 1. Schematic presentation of coating (and attachment of targeting ligands to) DNA complexes with multifunctional (PHPMA-ONp) and monofunctional polymers (PHPMA-NHS and mPEG[5000]-SPA). 3. Add the PEG solution to complex solution at desired concentration (0.05–1 mg/mL final concentration is usually optimal). 4. Allow the reaction to proceed at room temperature for at least 3 h: Complexes are then ready to use.

3.3.2. Complexes Coated with Semitelechelic PHPMA We have used PHPMA as an alternative monofunctional (semitelechelic) hydrophilic polymer to PEG. PHPMA is known to be a nontoxic and nonimmunogenic polymer that has previously been used in the field of drug delivery as a carrier of low-molecular-weight drugs (21), and also in

66

Oupicky, Read, and Bettinger

Fig. 2. Chemical structures of polymers used for coating of DNA complexes.

combination with cationic polymers for the formation of DNA complexes (7). The properties of this polymer are in many ways similar to those of PEG. However, it has significant advantages in terms of its structural versatility, enabling easy incorporation of various functionalities. Monofunctional PHPMA can be prepared by radical polymerization in the presence of chain transfer agents, for introducing carboxylic acid endgroup 3-MPA (7) is suitable. The following protocol describes, in detail, synthesis of PHPMA containing one carboxylic endgroup and its activation to NHS reactive ester (Fig. 2). 3.3.2.1. SYNTHESIS OF PHPMA CONTAINING REACTIVE SUCCINIMIDYL ESTER GROUP (PHPMA-NHS) 1. Dissolve 1.2 g HPMA (Polysciences, no. 08242), 5 mg AIBN (initiator of radical polymerization) (Fluka, no. 11630), and 10 mg 3-MPA (Fluka, no. 63770) in 10 mL methanol.

Steric Stabilization of DNA Complexes

67

2. Bubble nitrogen through the solution for 10 min to remove dissolved oxygen. 3. Polymerize in sealed ampule at 50°C for 24 h. 4. Precipitate the solution into 200 mL acetone/ether (3:1 v/v) and isolate the polymer by filtration through sintered glass filter. 5. Purify the polymer by redissolving in methanol and precipitating into acetone/ ether (3:1). 6. Dry polymer in vacuum and determine number of end carboxylic groups by titration with 0.05 M NaOH (molecular weight should be 8000–10,000 g/mol). 7. Convert the carboxylic acid end-group into succinimidyl active ester by dissolving 0.2 g of the obtained polymer in 1 mL dry DMF; add 25 mg NHS. 8. Cool the solution to –20°C, and add a solution of 52 mg DCCD (Fluka, no. 36650) and 2 mg DMAP (Fluka, no. 39405) in 0.3 mL DMF. 9. Leave the solution overnight at 0°C. 10. Remove the precipitated N,N'-dicyclohexylurea by filtration, and precipitate the polymer into 20 mL mixture of dry acetone–diethyl ether (3:1). 11. Isolate the precipitated polymer by filtration and dry it under vacuum at room temperature. 12. The polymer is now ready for use in the coating reaction; however, it can be stored for a limited period of time at –20°C under argon.

3.3.2.2. COATING PROCEDURE 1. Prepare PLL–DNA complexes, at N:P ratio 2 or PEI:DNA complexes at N:P ratio 3–5, in 10 mM HEPES, pH 7.4 2. Dissolve PHPMA–NHS in water at a concentration 15 mg/mL. 3. Add the PHPMA–NHS solution to complex solution at desired concentration (0.2–1 mg/mL final concentration is usually optimal). 4. Let the reaction proceed at room temperature for at least 3 h: Complexes are then ready to use.

3.4. Coating with Multifunctional Polymers Coating with multifunctional polymers was recently developed in this laboratory as an alternative approach to the more common coating with monofunctional PEG. An advantage of coating with multifunctional polymer is that it provides complexes not only with better stability, because of increased hydrophilic character of their surface, but also with lateral stabilization, which was found to be an important factor influencing their biological properties (15). We have used a statistical copolymer of HPMA with methacryloylated tetrapeptide, GlyPheLeuGly, having terminal carboxyl group activated as reactive 4-nitrophenyl ester (PHPMA-ONp) (Fig. 2). Multifunctional PHPMA is not commercially available, and needs to be synthesized. The synthesis is well described by Ulbrich (22), and its description here would be beyond the scope of this chapter. The coating procedure using multifunctional PHPMA itself is similar to that with monofunctional polymers. Schematic representation of the differences are

68

Oupicky, Read, and Bettinger

illustrated in Fig. 1. The polymer of typical mol wt 20,000 contains on average 12 reactive 4-nitrophenyl groups/molecule, and this enables multivalent attachment to the surface of the complex, which is assumed to be cooperative in nature. The coating then stabilizes the complex by crosslinking its surface, introducing another degree of stabilization compared to monofunctional polymers. The following protocol describes a procedure used to coat DNA complexes with multifunctional PHPMA-ONp. 1. Dissolve PHPMA-ONp in pure water (5 mg/mL) and add to complexes (prepared according to the standard protocol in 10 mM HEPES, pH 7.4), at a final concentration of 200 µg/mL (i.e., 40 µL/mL). 2. Accelerate the reaction by addition of HEPES buffer, pH 7.8, 0.5 M to 50 mM concentration. 3. Incubate for at least 3 h at ambient temperature or overnight at 4°C. 4. Reaction can be monitored by measuring the disappearance of 4-nitrophenyl esters at 274 nm (¡274 = 10,500 L/[mol/cm]). The PHPMA-ONp hydrolysis halflife at pH 8.0 at 25°C is about 30 min. 5. Before use, the remaining 4-nitrophenyl groups can be aminolyzed by addition of aminoethanol solution 2 µL/mL of complexes to a final aminoethanol concentration of 0.0004% v/v.

3.5. Attaching of Targeting Agents Simple polycation–DNA complexes show good transfection activity in vitro, mostly because of their nonspecific uptake by adsorptive endocytosis. Coating the complexes with hydrophilic polymers changes their surface properties, namely, the positive charge responsible for cellular uptake. As a result, the complexes show decreased uptake into cells and a reduction in transfection activity (15,23). This is in fact desirable because it enables retargeting of complexes to selected cells or tissues. In order to achieve targeted delivery, it is necessary to attach targeting ligands to the coated complexes. Both types of described polymers can in principle be used for attaching targeting ligands to the surface of the complexes. However, only the use of multifunctional PHPMA-ONp has been published so far. The principle of attachment of targeting ligands onto DNA complexes by means of multifunctional PHPMA-ONp is schematically shown in Fig. 1.

3.5.1. Attachment of Targeting Agents via Multifunctional PHPMA Polymers with multivalent reactive sites can be used to couple numerous smaller molecules to complexes, which can serve as targeting ligands for binding sites on cells. PHPMA contains several reactive groups, part of which can be used for the coating reaction and the rest for attaching targeting ligands (Fig. 1). Using multifunctional PHPMA for attaching targeting ligands to the

Steric Stabilization of DNA Complexes

69

complexes is easy, and the attachment is realized through amino groups. Targeting ligands, if of a protein nature, usually contain amino groups in sufficient amount. The disadvantage is the limited control on the specificity of binding, and also a possibility of crosslinking the complexes via the multifunctional polymer. For efficient attachment of targeting ligands, the coating reaction has to be stopped by addition of the targeting ligand at a suitable time in order to preserve sufficient amounts of reactive 4-nitrophenyl groups, and at the same time allow proper coating of the complexes (15). The following protocol describes the coating of PLL-DNA complexes and attachment of the targeting ligand transferrin. 1. Dissolve PHPMA-ONp in pure water at the concentration of 5 mg/mL and add to complexes at a final concentration of 200 µg/mL (i.e., 40 µL/mL). 2. Add HEPES buffer (pH 7.8, 0.5 M) to 50 mM concentration to accelerate the reaction. 3. Leave the reaction mixture for 2 h at 4°C. 4. Add 10 µL of transferrin solution (10 mg/mL) to the complex solution and incubate overnight at 4°C. 5. Other targeting ligands may require different timing of addition to the coated complexes and different concentration of ligand.

3.5.2. Attachment of Targeting Agents via Heterobifunctional PEG Another possibility for attaching targeting agents to coated complexes would be the use of heterobifunctional PEG. An advantage of using heterobifunctional PEG could be better control over the product structure, as reactive groups toward two different functionalities are used. This approach has been successfully used for attaching targeting ligands (antibodies) to liposomes (24). However, to the authors’ knowledge, no paper describing the use of this approach for targeting of polycation–DNA complexes has yet been published.

3.6. Properties and Analysis of Coated Complexes 3.6.1. Colorimetric Determination of Degree of Modification The efficiency of the coating reaction can be estimated by measuring the decrease in the number of amino groups in complexes. A wide range of reagents is available for this purpose including 2,4,6-trinitrobenzenesulfonic acid (TNBS) (25), ninhydrin (13), and fluorescamine (12). The advantage of using TNBS is that the assay is performed in fully aqueous solution (0.05 M borate buffer); both fluorescamine and ninhydrin assays are performed in aqueous–organic solutions, which raises the question of possible influence of the presence of organic solvents on the complex properties. Fluorescence-based fluorescamine assay, on the other hand, is at least 10-fold more sensitive and can be performed more quickly than absorption–based TNBS and ninhydrin assays.

70

Oupicky, Read, and Bettinger

The following two protocols are for determination of amino groups in DNA complexes. The fluorescamine assay is described in Chapter 11 by Read et al. 3.6.1.1. TNBS ASSAY 1. Mix 0.4 mL of complexes (made at 20 µg/mL DNA concentration) with 0.4 mL 0.1 M Na2B4O7·10H2O, and add 0.02 mL TNBS solution diluted from 1 M stock solution (Fluka, no. 92822) to 15 mg/mL with water. 2. Incubate for 45 min at room temperature and measure absorbance at 420 nm (absorbance of the solution made from PLL (20,000)–DNA N:P = 2 is usually about 0.35). 3. As a blank, add 0.02 mL TNBS to 0.8 mL borate buffer. 4. Compare absorbances of the coated complexes with those of the original complexes and calculate % of modified amino groups.

3.6.1.2. NINHYDRIN ASSAY 1. Prepare solution A (2.5 g ninhydrin in 50 mL ethanol) and solution B (1.3 mg potassium cyanide in 2 mL water and 98 mL pyridine). 2. Dilute sample containing ~10 µg polycation (PLL, PEI) with 0.1 mL solution B and 0.075 mL solution A. 3. Stop the reaction after heating for 10 min at 95°C by addition of 0.75 mL 60% ethanol and measure the absorption of the resulting blue solution at 570 nm (the assay gives identical results with free or DNA-bound PEI).

3.6.2. Size of Coated Complexes The introduction of a surface layer of hydrophilic polymer causes an increase in the size of the complexes that can be easily measured by dynamic light scattering (photon correlation spectroscopy [PCS]). Under the previously described conditions used for coating reactions, the authors have usually observed an increase in diameter following modification with mPEG(5000)-SPA of about 15–20 nm and slightly higher (about 20–25 nm) in the case of multifunctional PHPMA-ONp (mol wt 20,000). The thickness of the coating polymer layer is related to a reduction in phagocytic uptake, as demonstrated on polystyrene nanoparticles. For particles of the size of typical DNA complexes (~100 nm), a layer thickness of approx 10 nm should be adequate to confer particle nonrecognition in vivo (10). The typical changes in size of PLL–DNA and PEI–DNA complexes, following their modification with mPEG(5000)-SPA and multifunctional PHPMA-ONp, are shown in Fig. 3. The increase of size mostly results from the presence of the surface polymer layer, but could also partly reflect crosslinking of complexes when using multifunctional PHPMA-ONp. According to Davis (10), the thickness (L) of the PEG layer on the surface of modified nanoparticles is linearly related to the number of ethylene oxide groups (EO): L (nm) = 0.125 EO

Steric Stabilization of DNA Complexes

71

Fig. 3. Effect of coating PLL (20,000)–DNA and PEI (25,000)–DNA complexes with mPEG(5000)-SPA and multifunctional PHPMA-ONp on the size of the complexes (measured by PCS).

This equation gives 14-nm thickness for PEG(5000) (i.e., increase in diameter of 28 nm), which is slightly more than the observed increase of the size of the complexes as documented in Fig. 3. Details about measuring sizes by PCS are described in Chapter 11 by Read et al. and Chapter 22 by Wiethoff et al.

3.6.3. Altered c Potential Using excess polycation for complex formation results in generating positively charged particles. Attachment of hydrophilic polymer (PEG) to the surface of the complexes screens (masks) the surface charge and as a result, a decreased c potential is usually observed. In the case of multifunctional PHPMA-ONp, recharging is observed because some reactive 4-nitrophenyl ester groups on PHPMA-ONp do not react with amino groups on the surface of the complex, but hydrolyze leaving a free carboxylic acid group, giving the complexes a negative c potential. The influence of coating on c potential of DNA complexes is shown in Fig. 4. The protocol for measuring c potential of DNA complexes is in Chapter 11 by Read et al.

3.6.4. Increased Stability Against Salt-Induced Aggregation The size of polycation–DNA complexes depends on their N:P ratio, and on the ionic strength of the solution. At high N:P ratios, simple polycation:DNA complexes are usually well-stabilized in low-ionic-strength solutions by

72

Oupicky, Read, and Bettinger

Fig. 4. Effect of coating PLL (20,000)–DNA and PEI (25,000)–DNA with mPEG(5000)-SPA and multifunctional PHPMA-ONp on the c potential of the complexes.

electrostatic repulsion. With increasing ionic strength of the solution, the attractive van der Waals forces become stronger than the electrostatic repulsion, because of the screening effect of salt suppressing the charged layer surrounding the particle. As a result, complexes aggregate with increasing salt concentrations. Several techniques can be used to study the stability of the coated complexes against salt-induced aggregation. The simple one suitable for routine testing is to follow changes in turbidity of the complex solution after addition of salt. For this purpose, a fluorimeter with excitation and emission wavelengths set to the same value (600 nm) is the easiest option. The uncoated polycation–DNA complexes usually precipitate very fast, forming large aggregates shortly after the addition of NaCl, so that the quality of coating is easily checked. Properly coated complexes should be stable in 0.15 M NaCl for an extended period of time (at least 24 h with only minimal change in size). However, distinct differences between properly coated and uncoated complexes are observed after several minutes in 0.15 M NaCl solution. Better information about the changes in the properties of the complexes in salt solution can be obtained by dynamic light scattering or a combination of static and dynamic light scattering techniques (7).

3.6.5. Morphology of Coated Complexes To obtain the full picture about the influence of the coating on the properties of the complexes, it is advisable to use several analytical techniques. Light scattering techniques (PCS) are powerful tools, but it is not always easy to

Steric Stabilization of DNA Complexes

73

obtain information about morphology of the complexes. Electron microscopy, on the other hand, gives direct image of the complexes (limitations of this technique are described in Chapters 11 and 12). Special attention needs to be paid to the possibility that coating of the complexes will disrupt their condensed structure. As a result, extended open structures can be obtained. Various extended and open structures were observed for DNA complexes of PEG-PLL and PEG-PEI graft and block copolymers (3,23). The presence of such structures has not been reported for coated complexes (PEG, PHPMA-ONp, PHPMA-NHS), but the possibility of disrupting and consequently decreasing the stability of the complexes as a direct result of coating cannot be completely ruled out.

3.6.6. Albumin-Induced Turbidity Assay Simple polycation–DNA complexes are cleared quickly from the bloodstream after intravenous administration (2). This is thought to result from binding of serum proteins. Reducing the amount of proteins binding to the complexes is thus believed to be essential for successful systemic gene delivery. The addition of albumin to simple polycation–DNA complexes in water results in significant turbidity, which may be utilized as a convenient and simple measure of the stability of the complexes toward albumin (or other proteins) interaction. Albumin is able to bind to simple polycation–DNA complexes forming ternary complexes (2). At certain ratios, the ternary complexes are hydrophobic and excessive aggregation can be observed. However, at high concentrations, albumin is able to reduce the aggregation of PLL–DNA complexes, acting as a surface stabilizer (1,11), which suggests that, although the simple PLL–DNA complexes show only low size stability in saline solution or in low concentrations of albumin, their size could be well stabilized in the presence of higher concentrations of proteins in the blood. However, the adsorbed protein still represents a problem for successful systemic delivery. Although the addition of low concentrations of albumin (<100 µg/mL) causes significant aggregation in uncoated complexes, in the case of coated complexes, no turbidity is observed up to high albumin concentration (11,15). The albumin-induced turbidity assay thus shows that coating the PLL–DNA complexes with PEG, monofunctional PHPMA-NHS, or multifunctional PHPMA-ONp prevents albumin from causing aggregation of the complexes, which demonstrates that coating the complexes with hydrophilic polymers can significantly reduce nonspecific binding of albumin and proteins generally. 3.6.6.1. ALBUMIN-INDUCED TURBIDITY ASSAY 1. Prepare solutions of complexes to be tested and stock solution of albumin (30 mg/mL). 2. Set the fluorimeter excitation and emission wavelengths to 600 nm.

74

Oupicky, Read, and Bettinger

Fig. 5. Effect of coating PLL (20,000)–DNA complexes with PHPMA-ONp on the stability of the complexes against interaction with albumin. 3. Add albumin solution stepwise (up to 10 mg/mL) to the complex solution, and monitor changes in signal (typical results are shown in Fig. 5 for PHPMA-ONpcoated PLL (20,000)–DNA complexes).

3.6.7. Decreased Macrophage Uptake To address the possibility that the coated complexes may be able to evade phagocytic capture, we have employed an in vitro model, representing phagocytic components of the RES (11). Cell association of YOYO-1-labeled complexes was monitored using primary cultures of mouse peritoneal macrophages. Substantial levels of cell association are observed for simple polycation– DNA complexes, which may reflect their net positive surface charge, promoting nonspecific association with cells and membranes and easy binding of various proteins that can also promote uptake. The particulate nature of PLL– DNA complexes can also promote physical attachment and subsequent phagocytosis. All coated complexes described previously (PEG, PHPMA-NHS, multifunctional PHPMA-ONp) show significantly decreased levels of capture (Fig. 6), mostly because of their higher hydrophilicity (manifested by increased stability against salt-induced precipitation) and reduced c potential, which may inhibit complex-cell attachment, leading to a decrease in cell capture. Coating of the complexes may also create a physical barrier between the complex and cell surface, resulting in steric hindrance analogous to stealth technology (26), leading to reduced uptake. The following protocol can be used to investigate the phagocytic uptake of DNA complexes in vitro.

Steric Stabilization of DNA Complexes

75

Fig. 6. Effect of coating PLL (20,000)–DNA complexes with PHPMA-ONp and PHPMA-NHS on the phagocytic uptake of the complexes by murine peritoneal macrophages in vitro.

3.6.7.1. ISOLATION OF MURINE PERITONEAL MACROPHAGES 1. Adult female Balb/c mice are killed by cervical dislocation and injected intraperitoneally, under sterile conditions, with 5 mL medium 199 containing 20% fetal calf serum (Gibco, Paisley, UK). 2. Agitate the abdomen gently, expose and breach the peritoneum, and remove the medium, using a syringe. 3. Centrifuge medium (500 g, 10 min), and resuspend the pellet in medium 199 containing 50% fetal calf serum. 4. Plate the cell suspension out into multiwell 6-well plates (480,000 cells/well). 5. Allow the macrophages to adhere for 2 h before medium (containing nonadherent cells) is removed. 6. Add fresh medium containing amphotericin B (25 µg/mL) (Sigma, Poole, UK), penicillin (100 U/mL), and streptamycin (0.1 mg/mL) (Sigma, Poole, UK) to the cells. 7. After 40 h, the cells are ready for use in cellular association studies.

3.6.7.2. STUDY OF COMPLEX ASSOCIATION WITH MACROPHAGES 1. Label DNA with intercalating dye, YOYO-1, by adding its DMSO stock solution (Molecular Probes) to DNA solution (one YOYO-1 molecule/300 bp): Leave to incubate for 1 h before making complexes. 2. Form the complexes with YOYO-1-labeled plasmid DNA using the standard protocol described previously (at a DNA concentration of 20 µg/mL). 3. Add 75 µL complexes/well to plates containing adherent mouse macrophages in 50% 199 media, 50% fetal calf serum.

76

Oupicky, Read, and Bettinger

4. Remove the cells after desired time of incubation by standard trypsin treatment, wash and resuspend in phosphate-buffered saline. 5. Association of the complexes with cells is determined by the amount of YOYO1–DNA detected with a Coulter EPICS XL flow cytometer using an argon ion laser set for excitation (488 nm) and emission (520 nm) (cell groups were defined to exclude debris and duplets, and each cell type is gated individually against a negative control of cells without addition of complexes).

3.6.8. Concentration of Coated Complexes Simple polycation–DNA complexes need to be made at low DNA concentrations to prevent formation of large particles and their aggregation. Polymer coating increases solubility and size stability of the complexes to such an extent that they can be concentrated to high concentrations. We have used concentrators available from VivaScience (VivaSpin20 mol wt cutoff 100,000) that allow up to 20 mL solution to be concentrated down to 0.2–0.4 mL. If the complexes are properly coated (show good stability against salt-induced aggregation), they can be concentrated without significant change in their size (as confirmed by PCS and transmission electron microscopy), up to at least a DNA concentration of 1 mg/mL. During the concentration of the complexes, the range of reaction side products (NHS, 4-nitrophenol, unreacted polymers) can be at least partially removed from the complex solution, thus allowing purification of the coated complexes. 1. Prepare and coat the complexes with PEG or PHPMA according to previously described protocols. 2. Rinse the concentrator first with 10 mL water by centrifuging at 800g for 5 min. 3. Concentrate up to 20 mL complexes in centrifuge at 800g. 4. Concentration takes 10–45 min, depending on the sample volume and type of sample (generally PHPMA-coated complexes tend to take longer than PEGcoated ones). 5. VivaSpin20 concentrators give excellent recovery of coated complexes (>90%).

3.6.9. Freeze-Drying of Coated Complexes It is practical to store all pharmaceutical formulations in solid state. Successful freeze-drying of the simple polycation–DNA complexes requires the presence of cryoprotectants, such as sucrose, in order to preserve original size and biological properties (27). Multifunctional PHPMA-ONp-coated complexes can be easily freeze-dried and stored in solid state even without the presence of cryoprotectants during freeze-drying. We have successfully freeze-dried both concentrated PHPMA-coated and diluted complexes. The presence of 0.5% lactose in the solution has only a minimal effect on the sizes of reconstituted complexes.

Steric Stabilization of DNA Complexes

77

References 1. Ogris, M., Steinlein, P., Kursa, M., Mechtler, K., Kircheis, R., and Wagner, E. (1998) Size of DNA/transferrin-PEI complexes is an important factor for gene expression in cultured cells. Gene Ther. 5, 1425–1433. 2. Dash, P. R., Read, M. L., Barrett, L. B., Wolfert, M. A., and Seymour, L. W. (1999) Factors affecting blood clearance and in vivo distribution of polyeletrolyte complexes for gene delivery. Gene Ther. 6, 643–650. 3. Wolfert, M. A., Schacht, E. H., Toncheva, V., Ulbrich, K., Nazarova, O., and Seymour, L. W. (1996) Characterization of vectors for gene therapy formed by self-assembly of DNA with synthetic block copolymers. Hum. Gene Ther. 7, 2123–2133. 4. Kabanov, A. V. and Kabanov, V. A. (1995) DNA complexes with polycations for the delivery of genetic material into cells. Bioconjugate Chem. 6, 7–20. 5. Katayose, S. and Kataoka, K. (1997) Water-soluble polyion complex associates of DNA and poly(ethylene glycol)-poly(L-lysine) block copolymer. Bioconjugate Chem. 8, 702–707. 6. Seymour, L.W., Kataoka, K., and Kabanov, A. V. (1998) Cationic block copolymers as self-assembling vectors for gene delivery, in Self-assembling Complexes for Gene Delivery: From Laboratory to Clinical Trials (Kabanov, A. V., Felgner, P. L., and Seymour, L. W., eds.), Wiley, New York, pp. 219–239. 7. Oupicky, D., Konak, C., Dash, P. R., Seymour, L. W., and Ulbrich, K. (1999) Effect of albumin and polyanion on the structure of DNA complexes with polycation containing hydrophilic nonionic block. Bioconjugate Chem. 10, 764–772. 8. Toncheva, V., Wolfert, M. A., Dash, P. R., Oupicky, D., Ulbrich, K., Seymour, L. W., and Schacht, E. H. (1998) Novel vectors for gene delivery formed by self-assembly of DNA with poly(L-lysine) grafted with hydrophilic polymers. Biochim. Biophys. Acta 1380, 354–368. 9. Everett, D. H. (1988) Basic Principles of Colloid Science. The Royal Society of Chemistry, London. 10. Davis, S. S. and Illum, L. (1988) Polymeric microspheres as drug carriers. Biomaterials 9, 111–115. 11. Oupicky, D., Howard, K. A., Konak, C., Dash, P. R., Ulbrich, K., and Seymour, L. W. (2000) Steric stabilization of poly-L-lysine/DNA complexes by covalent attachment of semitelechelic poly[N-(2-hydroxypropl)methacrylamide]. Bioconjugate Chem. 11(4), 492–501. 12. Read, M. L., Etrych, T., Ulbrich, K., and Seymour, L. W. (1999) Characterisation of the binding interaction between poly( L-lysine) and DNA using the fluorescamine assay in the formation of non-viral gene delivery vectors. FEBS Lett. 461, 96–100. 13. Ogris, M., Brunner, S., Schuller, S., Kircheis, R., and Wagner, E. (1999) PEGylated DNA/transferrin-PEI complexes: reduced interaction with blood components, extended circulation in blood and potential for systemic gene delivery. Gene Ther. 6, 595–605.

78

Oupicky, Read, and Bettinger

14. Kircheis, R., Schuller, S., Brunner, S., Ogris, M., Heider, K.-H., Zauner, W., and Wagner, E. (1999) Polycation-based DNA complexes for tumor-targeted gene delivery. J. Gene Med. 1, 111–120. 15. Dash, P .R., Read, M. L., Fisher, K., Howard, K. A., Wolfert, M., Oupicky, D., et al. (2000) Decreased binding to proteins and cells of polymeric gene delivery vectors surface modified with a multivalent hydrophilic polymer and retargeting through attachment of transferrin. J. Biol. Chem. 275, 3793–3802. 16. Plank, C., Mechtler, K., Szoka, F. C., and Wagner, E. (1996) Activation of the complement system by synthetic DNA complexes: a potential barrier for intravenous gene delivery. Hum. Gene Ther. 7, 1437–1446. 17. Wolfert, M. A., Dash, P. R., Nazarova, O., Oupicky, D., Seymour, L. W., Smart, S., et al. (1999) Polyelectrolyte vectors for gene delivery: influence of cationic polymer on biophysical properties of complexes formed by self-assembly with DNA. Bioconjugate Chem. 10, 993–1004. 18. van de Wetering, P., Moret, E. E., Schuurmans-Nieuwenbroek, M. E., van Steenbergen, M. J., and Hennink, W. E. (1999) Structure-activity relationship of water-soluble cationic methacrylate/methacrylamide polymers for nonviral gene delivery. Bioconjugate Chem. 10, 589–597. 19. Bettinger, T., Remy, J.-S., and Erbacher, P. (1999) Size reduction of galactosylated PEI/DNA complexes improves lectin-mediated gene transfer into hepatocytes. Bioconjugate Chem. 10, 558–561. 20. Zou, S.-M., Erbacher, P., Remy, J.-S., and Behr, J.-P. (2000) Systemic linear polyethylenimine (L-PEI)-mediated gene delivery in the mouse. J. Gene Med. 2(2), 128–134. 21. Duncan, R. and Ulbrich, K. (1993) Development of N-(2-hydroxypropl)methacrylamide copolymer conjugates for delivery of cancer chemotherapy. Macromol. Chem. Macromol. Symp. 70/71, 157–162. 22. Ulbrich, K., Subr, V., Strohalm, J., Plocova, D., Jelinkova, M., and Rihova, B. (2000) Polymeric drugs based on conjugates of synthetic and natural macromolecules. I. Synthesis and physico-chemical characterisation. J. Control. Release 64, 63–79. 23. Erbacher, P., Bettinger, T., Belguise-Valladier, P., Zou, S., Coll, J.-L., Behr, J.-P., and Remy, J.-S. (1999) Transfection and physical properties of various saccharide, poly(ethylene glycol), and antibody-derivatized polyethylenimines (PEI). J. Gene Med. 1, 210–222. 24. Mastrobattista, E., Koning, G. A., and Storm, G. (1999) Immunoliposomes for the targeted delivery of antitumor drugs. Adv. Drug Delivery Rev. 40, 103–127. 25. Snyder, S. L. and Sobocinski, P. Z. (1975) An improved 2,4,6-trinitrobenzenesulfonic acid method for the determination of amines. Anal. Biochem. 64, 284–288. 26. Lasic, D. D., Martin, F. J., Gabizon, A., Huang, S. K., and Papahadjopoulos, D. (1991) Sterically stabilized liposomes: a hypothesis on the molecular origin of the extended circulation times. Biochim. Biophys. Acta 1070, 187–192. 27. Cherng, J .Y., Van De Wetering, P., Talsma, H., Crommelin, D. J. A., and Hennink, W. E. (1997) Freeze-drying of poly([2-dimethylamino]ethyl methacrylate)-based gene delivery systems. Pharm. Res. 14, 1838–1841.

Disulfide Cationic Lipids

79

6 Use of Disulfide Cationic Lipids in Plasmid DNA Delivery Fuxing Tang and Jeffrey A. Hughes 1. Introduction Gene therapy provides a paradigm of the treatment of human diseases. The ultimate goal of gene therapy is to cure both inherited and acquired disorders by removing the original causes, i.e., adding, blocking, correcting, or replacing genes. Although gene therapy trials have been initiated worldwide for more than two decades, little has been achieved in clinically curing diseases. One of the major hurdles for gene therapy is the lack of an efficient gene delivery system. An ideal gene delivery system should be specifically targeting, biodegradable, nontoxic, nonimmunogenic, and stable for storage. Cationic liposomes are the most extensively investigated nonviral vectors. It is generally believed that DNA–liposome complexes enter cells via endocytosis, although other pathways such as membrane fusion may exist (1,2). The barriers involved in the transfection process in vitro generally include the following aspects (3): 1. 2. 3. 4. 5. 6.

Formation of the liposome–DNA complexes. Entry of complexes into cell. Escape of DNA from the endosomes. Dissociation of DNA from liposomes. Entry of DNA into nucleus. DNA transcription.

The strategy for overcoming any of the above barriers should increase transgene expression, and the formulations, which overcome the major barriers, will result in greater transgene expression. Toxicity is one of the major barriers that limits the application of cationic lipids in clinical trials. The use of ester, amide, and carbamate linkages to tether From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

79

80

Tang and Hughes

polar and hydrophobic domains of cationic lipid is a common strategy to lower toxicity. Ester bonds are biodegradable, but the introduction of the ester bond may also decrease the stability of liposomes in systemic circulation, when the liposomes are used in clinical trials. For example, Aberle et al. (4) constructed a tetraester cationic lipid and the liposome demonstrated lower toxicity in NIH 3T3 cells than did 3`(N-[N',N' dimethyl-aminoethane] carbamoyl) cholesterol (DC-Chol). However, this liposome must be used within 2 h of preparation. Carbamate bond is biodegradable and more stable than an ester bond in aqueous solution. Huang’s group (5) first introduced carbamate in the cationic lipid DC-Chol. DC-Chol was the first lipid used in clinical trials because of its combined properties of transfection efficiency, stability, and low toxicity (2). Another barrier for cationic lipid-mediated plasmid DNA (pDNA) delivery is the low transfection efficiency compared to the viral system. To solve these problems, the authors designed a class of disulfide cationic lipids that can enhance transfection activity and decrease the toxicity, but not sacrifice the stability of liposomes in aqueous solution (6,7). The strategy was to take advantage of the high intracellular reductive environment to use a disulfide linkercontaining cationic lipid that is selectively stable outside cells; it can be reduced in cells by intracellular reductive substances such as glutathione. The reduction of the disulfide linker resulted in the collapse of complexes, thus decreasing toxicity (Fig. 1) and enhancing the release of DNA from DNA–liposome complexes (Fig. 2). Dissociation of DNA from DNA–liposome complexes is one of the major barriers for cationic liposome-mediated gene transfection (1,3,8). The enhancement of the release of DNA was expected to increase transgene expression. Introduction of disulfide linker into cationic lipids enhanced transgene expression and decreased the toxicity of cationic lipid-mediated plasmid delivery (Figs. 3 and 4; 6,7,9,10). Disulfide conjugate techniques have been widely used in drug delivery to achieve high delivery efficiency (11–13). The normal method used in bioconjugation involves crosslinking or modification reactions using disulfide exchange processes to form disulfide linkage with sulfhydryl-containing molecules such as 3-(2-pyridyldithio)propionic acid n-hydroxy-succinimide ester (SPDP) (11–13). This method results in more stable disulfide compounds than the original compound, and thus is not suitable for the purpose of decreasing toxicity. The authors proposed a direct conjugation method to introduce disulfide lipid in cationic lipids-1', 2'-dioleoyl-sn-glycero-3'-succinyl-2-hydroxyetheyl disulfide ornithine conjugate (DOGSDSO) and cholesteryl hemidithiodiglycolyl Tris(aminoethyl)amine conjugate (CHDTAEA) (6,7). This method applied routine chemical synthesis, which is more economical, compared with the disulfide exchange method. For example, the price of SPDP, which is the key compound used in the classic disulfide exchange method, is thousands of times more expensive than dithioglycolic acid, which is used to introduce a disulfide

Disulfide Cationic Lipids

81

Fig. 1. Cytotoxicity of the liposome–DNA complexes was studied in CHO and SKnSH cells. A fixed dose of 1 µg/well pDNA was mixed with increasing amounts of cationic liposomes and used in the toxicity assay. Cell viability was calculated as percentage of survival cells as stated in Subheading 3. (A) Toxicity in CHO cells. (B) Toxicity in SKnSH cells. 䉬, CHDTAEA–DOPE, 䊏, CHSTAEA–DOPE; 䉱, DCChol–DOPE. Data is shown as mean ± SD (n = 3).

linker in the authors’ methods. This economical method will demonstrate an advantage in scale-up clinical trials and in pharmaceutical industry applications. Another advantage of this method is that disulfide bonds with different stability can be chosen for various applications. 2. Materials 1. 2-Hydroxyethyl disulfide (Aldrich, Milwaukee, WI), dithiodiglycolic acid (Pfaltz & Bauer, Waterbury, CT). 2. 1,2-Dioleoyl-sn-glycero-3-succinate, cholesterol (Avanti Polar Lipids, Ala– baster, AL).

82

Tang and Hughes

Fig. 2. Gel electrophoretic analysis of the release of pDNA from cationic liposomes–DNA complexes in 10 mM glutathione in phosphate-buffered saline. (A) DNA; (B) CHDTAEA–DOPE + DNA (2:1,w/w); (C) CHSTAEA–DOPE + DNA (2:1,w/w).

Fig. 3. Transgene expression in primary rat neuronal cultures transfected with a fixed amount of pDNA (3 µg) and various liposomal formulations. Number of replicates, n = 4. Each experiment was repeated at least 3×. The error bars represent mean ± SD.

3.

L-Ornithine,

2-(tert-butoxycarbonyloxyimino)-2-phenylacetonitrile (BOC-ON), triethylamine (TEA), 1,3-dicyclohexylcarbodiimide (DCCD), 1,4-dioxane, trifluoroacetic acid, 4-pyrrolidinopyridine, ethyl acetate, dichloromethane and methanol (Aldrich). 4. Chromatographic silica gel (200–425 mesh) (Fisher, Fair Lawn, NJ). 5. 1,2-Dioleoyl phosphatidylethanolamine (DOPE) (Avanti Polar Lipids).

Disulfide Cationic Lipids

83

Fig. 4. Comparison of the transfection of pGL3-luciferase plasmid by CHDTAEA–DOPE, CHSTAEA–DOPE, and DC-Chol in CHO and SKnSH cells. A fixed dose of 2 µg/well pGL3 DNA was mixed with increasing weight ratios of cationic liposomes (calculation based on cationic lipid) and used for transfection. (A) Transfection in CHO cells; (B) Transfection in SKnSH cells. 䉬, CHDTAEA– DOPE, 䊏, CHSTAEA–DOPE; 䉱, DC-Chol–DOPE. Data is shown as mean ± SD (n = 3). 6. 7. 8. 9.

Rotary evaporator Buchi 011(Buchi, Switzerland). Kontes ChromaflexTM chromatography column (Kontes,Vineland, NJ). Sonic Dismembrator 60 (Fisher Scientific). LiposoFastTM (Avestin, Ottawa, Canada).

84

Tang and Hughes

3. Methods 3.1. Synthesis of DOGSDSO (see Fig. 5) 1. Dissolve a total of 0.5 g A (ornithine) (0.0030 mol) and TEA in 6 mL water and 4 mL dioxane. 2. Add a total of 1.6 g BOC-ON (0.0066 mol) to the solution and stir the mixture at room temperature overnight. 3. Add 6 mL saturated sodium bicarbonate solution to the reaction mixture and extract the solution 3× with 20 mL ethyl acetate. Product B should be in aqueous layer at this step. 4. To the aqueous layer, add 20 mL 5% citric acid solution. Extract product B with 3× 20 mL ethyl acetate. 5. Collect the organic layers and dry over anhydrous sodium sulfate. Evaporate the organic solvent on a rotary evaporator. Pure product B is a colorless oil. 6. 0.388 g B (1.167 mmol), 0.360 g 2-hydroxyethyl disulfide (2.334 mmol), and 0.0173 g 4-pyrrolinopridine are dissolved in 20 mL methylene chloride under nitrogen (see Note 1). 7. To the solution, 1.284 mL 1 M DCC in dichloromethane is added via syringe under nitrogen. The mixture is stirred at room temperature under nitrogen overnight. 8. The white precipitate N,N-dicyclohexyl urea is filtered and the solvent is evaporated. 9. Product C is purified by chromatography on silica gel with hexane:ethyl acetate (1:1, v/v). Rf = 0.25. 10. 0.075 g 1,2-dioleoyl-sn-glycero-3-succinate (D) (0.104 mmol), 0.0788 g C (0.156 mmol), and 0.002 g 4-pyrrolidinopridine (0.0156 mmol) are mixed in dichloromethane under nitrogen. 11. 0.156 mL 1 M DCCD in dichloromethane is added to the mixture via syringe under nitrogen. The reaction mixture is stirred at room temperature under nitrogen overnight. 12. White precipitate N,N-dicyclohexyl urea is filtered and product E is purified by chromatography on silica gel with an eluant of hexane:ethyl acetate (3:1). Rf = 0.25. 13. To remove the BOC group, 2 mg E is dissolved in dichloromethane in a 50-mL round bottomed flask. The solvent is evaporated on a rotary evaporator and the flask is cooled on ice for 10 min to 0°C. 14. 5 mL trifluoroacetic acid is added to the flask at 0°C and the flask is incubated at room temperature for 5 min. Excess trifluoroacetic acid is evaporated on a rotary evaporator and evaporated further using a stream of nitrogen for 30 min. The product DOSGDSO is ready for preparation of liposomes (see Note 2).

3.2. Synthesis of CHDTAEA (see Fig. 6) The strategy of synthesis of CHDTAEA is similar to that of DOGSDSO with some modification. The electron-withdrawing properties of two _-carboxyl groups at the `-position of disulfide bond weaken the disulfide linker in CHDTAEA and make DNA easily releasable by reductive substances.

Disulfide Cationic Lipids

85

Fig. 5. Scheme of synthesis of 1', 2'-dioleoyl-sn-glycero-3'-succinyl-2-hydroxyethyl disulfide ornithine conjugate (DOGSDSO). 1. A total of 2.0 g (6 mmol) cholesterol and 1.9 g dithiodiglycolic acid (12 mmol) are dissolved in 30 mL ethyl acetate under nitrogen (see Note 3). 2. A total of 12 mmol DCCD, 0.19 g 4-pyrrolidinopyridine (1.2 mmol), and 1.43 mL TEA are added to the solution at 0°C. 3. The mixture is warmed to room temperature by standing and stirred under nitrogen overnight.

86

Tang and Hughes

Fig. 6. Scheme of synthesis of cholesteryl hemidithiodiglycolyl Tris(aminoethyl)amine (CHDTAEA). 4. The reaction mixture is filtered and washed by a 5% solution of citric acid, followed by brine, 3×. 5. The organic layer is collected and dried over sodium sulfate. 6. The organic solvent was evaporated on rotary evaporator. 7. The product cholesteryl hemidithiodiglycolate is purified on silica gel with an eluant of hexane:ethyl acetate:acetic acid (30:10:1). Rf = 0.3. 8. N,N-BOC2-Tris(2-aminoethyl) amine is prepared by treating Tris(2-aminoethyl) amine with two equivalents of BOC-ON in wet tetrahydrofuran. 9. A total of 0.3 g cholesteryl hemidithiodiglycolate (0.5 mmol), 0.24 g N,NBOC2-Tris(2-aminoethyl) amine (0.65 mmol), and 0.007 g 4-pyrrolidinopridine (0.005 mmol) is dissolved in 20 mL of dichloromethane. 10. A total of 0.6 mmol DCCD in dichloromethane and 0.06 mL TEA is added dropwise at 0°C. 11. The mixture is stirred at room temperature under nitrogen overnight. 12. The product N,N-BOC2–CHDTAEA is separated and purified on silica gel with a developer of dicholomethane:methanol (10:1). Rf = 0.35. 13. To remove BOC group, 2 mg N,N-BOC2 –CHDTAEA is dissolved in dichloromethane in a 50-mL round -bottomed flask. The solvent is evaporated on a rotary evaporator and the flask is cooled on ice for 10 min to 0°C. 5 mL TFA is added to the flask at 0°C and the flask is incubated at room temperature for 5 min. Excess TFA is evaporated on a rotary evaporator and evaporated further using a stream of nitrogen for 30 min. The product CHDTAEA is ready for preparation of liposomes.

3.3. Liposome Preparation 1. 5 mg Disulfide lipid DOGSDSO or CHDTAEA is dissolved in chloroform in a 250-mL round-bottomed flask.

Disulfide Cationic Lipids

87

2. One molar equivalent of DOPE is mixed with DOGSDSO or CHDTAEA in a flask. 3. Chloroform is evaporated on a rotary evaporator at room temperature and residual solvent is dried using a stream of nitrogen for 10 min (see Note 4). 4. Sterile water is added to the flask to hydrate the lipid film (see Note 5). 5. The flask is shaken at 30°C for 30 min. 6. The liposomes are hydrated at 4°C overnight. 7. The liposome suspension is sonicated using a Sonic Dismembrator for 2 min at 4°C to form homogenized liposomes (see Note 6). 8. When certain sizes of liposomes are required, the liposome suspension is passed through a membrane of specific sizes using LiposoFastTM (Avestin). 9. The liposomes are stored at 4°C until used for plasmid delivery. The liposomes do not lose activity for at least 1 yr.

4. Notes 1. The esterifcation reaction is hard to complete at room temperature. 4-pyrrolidinopyridine is a catalyst that can assist the reaction to be completed. 2. Disulfide lipid is not stable on chromatography column via separation. Because this step was reported to be quantitative (14), no further purification was attempted. 3. Ethyl acetate is a solvent with moderate polarity. Ethyl acetate was a better solvent than N,N-dimethylformide in this special case. 4. Temperature should be controlled to not exceed 30°C in considering the stability of disulfide bond. 5. The amount of water added was decided by the concentration required in the experiments. The concentration used in this laboratory is 1 mg/mL for in vitro or 5 mg/mL for in vivo experiments. In in vivo experiments, 5% dextrose solution is recommended to keep isotonic. Liposomes can still be made in sterile water (more stable) at a high concentration and diluted to concentration required in experiments, by dextrose solution of high concentration to reach a final concentration of 5% dextrose. 6. When the disulfide liposomes were made using sonicator, the size of liposomes was in a range of 130 ± 20 nm.

References 1. Rolland, A. P. (1998) From genes to gene medicines: recent advances in nonviral gene delivery. Crit. Rev. Ther. Drug Carrier Syst. 15, 143–198. 2. Gao, X. and Huang, L. (1995) Cationic liposome-mediated gene transfer. Gene Ther. 2, 710–722. 3. Zabner, J., Fasbender, A. J., Moninger, T., Poellinger, K. A., and Welsh, M. J. (1995) Cellular and molecular barriers to gene transfer by a cationic lipid. J. Biol. Chem. 270, 18,997–19,007. 4. Aberle, A. M., Tablin, F., Zhu, J., Walker, N. J., Gruenert, D. C., and Nantz, M. H. (1998) A novel tetraester construct that reduces cationic lipid-associated cytotoxicity. Implications for the onset of cytotoxicity. Biochemistry 37, 6533–6540. 5. Gao, X. and Huang, L. (1991) A novel cationic liposome reagent for efficient transfection of mammalian cells. Biochem. Biophys. Res. Commun. 179, 280–285.

88

Tang and Hughes

6. Tang, F. and Hughes, J. A. (1999) Use of dithiodiglycolic acid as a tether for cationic lipids decreases the cytotoxicity and increases transgene expression of plasmid DNA in vitro. Bioconjug. Chem. 10, 791–796. 7. Tang, F. and Hughes, J. A. (1998) Introduction of a disulfide bond into a cationic lipid enhances transgene expression of plasmid DNA. Biochem. Biophys. Res. Commun. 242, 141–145. 8. Escriou, V., Ciolina, C., Helbling-Leclerc, A., Wils, P., and Scherman, D. (1998) Cationic lipid-mediated gene transfer: analysis of cellular uptake and nuclear import of plasmid DNA. Cell. Biol. Toxicol. 14, 95–104. 9. Tang, F., Wang, W., and Hughes, J. A. (1999) Cationic liposomes containing disulfide bonds in delivery of plasmid DNA. J. Liposome Res. 9, 331–347. 10. Ajmani, P. S., Tang, F., Krishnaswami, S., Meyer, E. M., Sumners, C., and Hughes, J. A. (1999) Enhanced transgene expression in rat brain cell cultures with a disulfide-containing cationic lipid. Neurosci. Lett. 277, 141–144. 11. Boutorine, A. S. and Kostina, E. V. (1993) Reversible covalent attachment of cholesterol to oligodeoxyribonucleotides for studies of the mechanisms of their penetration into eucaryotic cells. Biochimie 75, 35–41. 12. Legendre, J. Y., Trzeciak, A., Bohrmann, B., Deuschle, U., Kitas, E., and Supersaxo, A. (1997) Dioleoylmelittin as a novel serum-insensitive reagent for efficient transfection of mammalian cells. Bioconjug. Chem. 8, 57–63. 13. Trail, P. A., Willner, D., Knipe, J., Henderson, A. J., Lasch, S. J., Zoeckler, M. E., et al. (1997) Effect of linker variation on the stability, potency, and efficacy of carcinoma-reactive BR64-doxorubicin immunoconjugates. Cancer Res. 57, 100–105. 14. Behr, J. P., Demeneix, B., Loeffler, J. P., and Perez-Mutul, J. (1989) Efficient gene transfer into mammalian primary endocrine cells with lipopolyamine-coated DNA. Proc. Natl. Acad. Sci. USA 86, 6982–6986.

Lipid–ODN Interaction with LDL

89

7 Interactions of Lipid–Oligonucleotide Conjugates with Low-Density Lipoprotein Erik T. Rump, Erik A. L. Biessen, Theo J. C. van Berkel, and Martin K. Bijsterbosch 1. Introduction The ability of antisense oligonucleotides to interdict, sequence-specifically, the expression of pathogenic genes affords an exciting new strategy for therapeutic intervention (1–3). Oligonucleotides with physiological phosphodiester internucleotide bonds are rapidly degraded, predominantly by exonucleases. Numerous oligonucleotide analogs have therefore been synthesized to confer resistance toward nuclease activity (3). The phosphorothioate analog is the most extensively studied, and phosphorothioate oligodeoxynucleotides have been shown to be potent inhibitors of the expression of their target genes in vitro and in vivo (1,3). However, phosphorothioate oligodeoxynucleotides also bind avidly and nonspecifically to proteins, thus provoking a variety of non-antisense effects (4). Oligonucleotide analogs that do not bind to proteins are therefore expected to display less nonantisense side effects. However, protein binding also affects the in vivo disposition of oligonucleotides. Nonphosphorothioate oligonucelotide analogs generally do not bind to serum proteins, and are therefore rapidly cleared from the circulation, protein-bound phosphorothioate oligodeoxynucelotides circulate much longer (5,6). The authors’ aim is to prolong the circulation time of nonphosphorothioate oligonucleotides, in order to increase the exposure of target cells to the oligonucleotides. The approach is to conjugate oligonucleotides with lipids, with the objective of inducing association of the oligonucleotide with longcirculating lipid particles (lipoproteins and lipoprotein-like particles). The biological fate of the particle-associated oligonucleotide will then be determined by the lipid carrier. The authors’ studies focused on low-density From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

89

90

Rump et al.

lipoprotein (LDL), the main cholesterol-transporting vehicle in human circulation. LDL is a spherical particle (diameter 23 nm), consisting of an apolar core of cholesteryl esters and triglycerides, which is surrounded by a shell of cholesterol and phospholipids (7). A large part of the surface is covered by apoprotein B100, which is recognized by specific LDL receptors. LDL is slowly cleared from the circulation (7,8), which makes it suitable for prolonging the circulation of associated oligonucleotides. A second reason to choose LDL as carrier is that it has been shown that a variety of tumor cell types (e.g., leukemic cells) internalize large amounts of LDL via the LDL receptor (9). Association of an oncogene-specific antisense oligonucleotide with LDL may lead to a higher uptake of the oligonucleotide by tumors, and consequently higher therapeutic efficacy. This chapter describes the conjugation of various lipids to a c-myb-directed oligonucleotide. The association of the lipid–oligonucleotide conjugates (lipid– ODNs) with LDL is characterized, as well as the stability of the lipid-ODN–LDL complexes in vitro in rat plasma and in vivo in rats. 2. Materials 2.1. Synthesis of Activated Lipid Structure 1. 2. 3. 4. 5. 6. 7. 8.

Lithocholic acid. Oleic acid. 3_-(oleoyloxy)-5`-cholanic acid. 3_,7_-bis(oleoyloxy)-5`-cholanic acid. 3`-(oleoyloxy)-5-cholenic acid. Pentafluorophenol. Dicyclohexylcarbodiimide. Cholesterol chloroformate.

2.2. Synthesis and Purification of 3'-Lipid-ODN 1. An 18-mer antisense oligonucleotide complementary to c-myb (10) (5'-GTG CCG GGG TCT TCG GGC-3') was from Eurogentec (Seraing, Belgium) and had a phosphodiester backbone. The antisense oligonucleotide had three phosphorothioate linkages at the 5' end and a C7-amino linker at the 3'-end. 2. [3H]-Labeled 3'-amine 18-mer antisense oligonucleotide, radiolabeled with 3H by heat-catalyzed exchange at the C8 positions of the purine nucleotides (see Note 1 and refs. 11,12). 3. [3H]ISIS-9388, a 3'-cholesteryl-conjugated phosphorothioate oligodeoxynucleotide specific for murine intercellular adhesion molecule-1 (13) was kindly provided by Dr. M. Manoharan (ISIS Pharmaceuticals, Carlsbad, CA). 4. Na125I (carrier-free) and 3H2O were from Amersham (Amersham, UK). 5. Low-melting multipurpose agarose and Agarase from Pseudomonas atlantica were obtained from Boehringer (Mannheim, Germany). 6. Tween-20 was from Merck (Darmstadt, Germany).

Lipid–ODN Interaction with LDL 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17.

91

Lithium perchlorate (LiClO4). Acetone. Dimethylformamide. Dioxane. N,N-Diisopropylethylamine (DIPEA). Dichloromethane. 50 mM Triethylamine ammonium acetate, pH 7.0. Agarase. 30 mM Bis-Tris, containing 10 mM EDTA, pH 6.5. Acetonitrile. 45 mM Tris-borate buffer, containing 0.1 mM EDTA, pH 8.4.

2.3. Determination of Melting Temperatures 1. A 28-mer sense oligonucleotide (5'-CCA TGG CCC GAA GAC CCC GGC ACA GCA T-3') was from Eurogentec and had a phosphodiester backbone. 2. Phosphate-buffered saline (PBS): 10 mM sodium phosphate buffer, pH 7.4, containing 0.15 M NaCl.

2.4. LDL and Lipid-ODNs 1. Human LDL: density 1.024–1.063 g/mL. 2. Hionic Fluor TM scintillation cocktail. 3. Superose 6 Precision chromatography column, 3.2 × 300 mm (Pharmacia).

2.5. Plasma Clearance and Liver Association 1. Male Wistar rats (180–230 g). 2. Ketamine hydrochloride was from Eurovet (Bladel, the Netherlands). 3. HypnormTM (0.315 mg/mL fentanyl citrate and 10 mg/mL fluanisone) was from Janssen-Cilag (Sauderton, UK). 4. Thalomonal TM (0.05 mg/mL fentanyl and 2.5 mg/mL droperidol) was from Janssen-Cilag. 5. Hionic Fluor and Emulsifier SafeTM scintillation cocktails and Soluene-350TM were from Packard (Downers Grove, IL). 6. Heparinized collection tubes.

3. Methods 3.1. Conjugation of Oligonucleotides with Lipid Structures The authors synthesized a series of conjugates of lipids with a 18-mer antisense oligonucleotide complementary to the c-myb proto-oncogene. The oligonucleotide had a phosphodiester backbone with three phosphorothioate linkages at the 5' end as protection against 5'-exonucleases. At the 3' end, the oligonucleotide was provided with an amino group to enable conjugation. Figure 1 shows the structures of the lipids that were used for conjugation. To enable coupling to the 3'-amino-tailed oligonucleotide, the lipids were

92

Rump et al.

Fig. 1. Lipid structures of lipid-ODNs.

activated. Except cholesterol (ODN-4), all lipids were activated at their carboxylic acid functionalities with pentafluorophenol (see ref. 11 for full details of the synthesis of the activated lipid structures). Cholesterol chloroformate was used for the conjugation of cholesterol. Conjugation of the

Lipid–ODN Interaction with LDL

93

amino-terminated oligonucleotide with the activated lipids was performed in a mixture of water and the aprotic solvents, dimethylformamide and 1,4-dioxane (1:4:4; v/v/v), in which the organic base DIPEA was present to create mild basic conditions. After conjugation, the lipid-ODNs were purified. ODN-2, ODN-3, and ODN-4 (conjugates of oligonucleotide with litocholic acid, oleic acid, and cholesterol, respectively) were purified by reversed-phase high-performance liquid chromatography (RP-HPLC). Purification of conjugates of the oligonucleotide with the oleoyl steroid ester structures (ODN-5, ODN-6, and ODN-7) was not possible by RP-HPLC; no conjugated products were recovered. These lipid-ODNs were separated from unconjugated oligonucleotide by electophoresis in a 1% agarose gel, containing 0.1% Tween-20. Separation is accomplished by the formation of micelles of lipid-ODNs and Tween-20. The lipid-ODNs were retrieved from the gel by melting the gel, followed by digestion of the gel material by Agarase.

3.1.1. Synthesis of Activated Lipid Structure 5`-Cholanic acid 3_-ol pentafluorophenyl ester, oleic acid pentafluorphenyl ester, 3_-(oleoyloxy)-5`-cholanic acid pentaflorophenyl ester, 3_,7`-bis(oleoyloxy)-5`-cholanic acid pentaflorophenyl ester, and 3`(oleoyloxy)-5-cholenic acid pentaflorophenyl ester were synthesized as described in full detail earlier by activating lithocholic acid, oleic acid, 3_(oleoyloxy)-5`-cholanic acid, 3_,7`-bis(oleoyloxy)-5`-cholanic acid, and 3`-(oleoyloxy)-5-cholenic acid, respectively, with pentafluorophenol (11). The identity of all compounds was verified by mass spectrometry and nuclear magnetic resonance.

3.1.2. Synthesis and Purification of 3'-Lipid-ODN 1. The amino-terminated antisense oligonucleotide is precipitated as a Li-salt with 10 vol of 3% LiClO4 in acetone. 2. The oligonucleotide is subsequently dissolved in H2O and precipitated again with 10 vol acetone to remove final traces of LiClO4. 3. In a typical derivatization experiment, 15 nmol of oligonucleotide is dissolved in 350 µL H2O/dimethylformamide/dioxane (1:4:2, v/v/v). Subsequently, 1 µmol activated lipid, dissolved in 100 µL dioxane, and 35 µmol DIPEA are added. For the preparation of ODN-2, ODN-3, and ODN-4, the mixtures were incubated for 48 h at 37°C, and, for the preparation of ODN-5, ODN-6, and ODN-7, the mixtures were incubated for 48 h at 56°C. 4. Solvents are removed in a Speed-Vac concentrator, and the residue is taken up in 200 µL dichloromethane and 200 µL H2O. The layers are separated by centrifugation, and the organic layer is washed twice with 200 µL H2O. 5. The aqueous phases are combined and freeze-dried. 6. Derivatization of (3H)oligonucleotide is performed as described in steps 1–5 with 2 nmol labeled oligonucleotide and 4 nmol unlabeled oligonucleotide.

94

Rump et al.

7. Oligonucleotides conjugated with lithocholic acid, oleic acid or cholesterol are purified by RP-HPLC on a Waters C4 column (5 µm, 300 A, 300 × 3.9 mm) by applying a gradient of 1% CH3CN/min in 50 mM triethyl ammonium acetate (pH 7.0) at a flow rate of 1 mL/min. The gradient (5–50%) is started after elution for 5 min at 5% CH3CN. Oligonucleotides are detected at 260 nm. 8. All other lipid-conjugated oligonucleotides are purified by gel electrophoresis in a 1% (w/v) low-melting multi purpose agarose gel, containing 0.1% Tween-20, at pH 8.4 (45 mM Tris-borate buffer, containing 0.1 mM EDTA). Gel slices containing lipid-ODNs are melted for 5 min at 65°C. The agarose is digested for 2 h with Agarase (40 U/mL gel) at 45°C in 30 mM Bis-Tris, containing 10 mM EDTA. The lipidODNs are precipitated with 10 vol of acetone. To remove traces of undigested agarose, the precipitate is taken up in 200 µL H2O and passed over a filter paper (no. 589, Schleicher and Schüll). Lipid-ODNs are isolated in a yield of 35–75%.

3.2. Interaction of Lipid–ODNs with Their Target Sequences: Determination of Melting Temperatures To ascertain that the attached lipid structures do not interfere with the interaction of the antisense oligonucleotide with the target sequence, the authors determined the melting temperatures of antisense–sense hybrids. A 28-mer sense oligonucleotide was utilized with five overhang nucleotides at both the 3'- and 5'-end, to assess a possible interaction of the lipid with the singlestranded part of the hybrid. The melting temperatures of the hybrids with the conjugated ODNs differed not appreciably (maximally 2°C) from that of the hybrid with the unconjugated ODN-1 (Table 1). Thus, the bulky steroid lipids (particularly the oleoyl steroid esters) do not significantly interfere with the overhang nucleotides of the target sequence. These findings are consistent with reports of other lipid-ODNs (14). Melting temperatures of hybrids of the antisense lipid-ODNs with an 28-mer sense oligonucleotide were determined using a Perkin-Elmer spectrophotometer equipped with a PTP-6 thermal programmer. 1. Equimolar amounts of both oligonucleotides are dissolved to a concentration of 3.6 µM in PBS. 2. The mixtures are placed for 2 min at 96°C and subsequently slowly cooled to room temperature to allow annealing. 3. Then, the temperature is adjusted to 35°C and hybrids are melted by increasing the temperature to 95°C at a rate of 0.5°C/min (see Note 2).

3.3. Interaction of Lipid-ODNs with LDL To examine which of the conjugated lipid structures is able to associate the oligonucleotide with LDL, radiolabeled lipid-ODNs were incubated for 2 h at 37°C in PBS with equimolar amounts of radioiodinated LDL. Aliquots of the incubation mixtures were analyzed by agarose gel electrophoresis at pH 8.8 in

Lipid–ODN Interaction with LDL

95

Table 1 Melting Temperatures of Hybrids of Antisense Lipid-ONDs and Sense Oligonucleotide Oligonucleotide ODN-1 ODN-2 ODN-3 ODN-4 ODN-5 ODN-6 ODN-7

Melting temperature (°C) 71 72 72 73 70 69 70

Equimolar amounts of antisense lipid-ODNs and 28-mer sense oligonucleotide, dissolved in PBS, were heated for 2 min at 96°C, and subsequently slowly cooled to room temperature to allow annealing. Then, the temperature was adjusted to 35°C and hybrids were melted by increasing the temperature to 95°C at a rate of 0.5°C/min. Differences in melting temperatures measured in duplicate runs were <1°C.

75 mM Tris-hippuric acid buffer (Fig. 2). The electrophoretic mobility (Rf) of LDL in this gel matrix is 0.2; oligonucleotides (nonconjugated as well as lipidconjugated) have Rf values close to 1.0. Figure 2 shows that >95% of the [3H]ODN-4, [3H]ODN-5, and [3H]ODN-7 comigrated in the gel with radioiodinated LDL (Rf, 0.2). Thus, conjugation of the oligonucleotide with cholesterol or the oleoyl esters of lithocholic acid or cholenic acid induces spontaneous and almost complete association of the oligonucleotide with LDL. These steroids meet the structural requirements for LDL anchors as defined by Firestone et al. (15; see Note 3). When lipid-ODNs were associated at higher molar ratios, the increase in the electrophoretic mobility was more clear (data not shown). Figure 2 also shows that conjugation of lithocholic acid and oleic acid (ODN-2 and ODN-3) did not induce association of the oligonucleotide with LDL at all. Only a proportion of the [3H]oligonucleotide conjugated to the bis-oleoyl steroid ester (ODN-6) comigrated with LDL. A substantial proportion of 3H-radioactivity migrated at Rf 0.2–1.0, which suggests that the lipid-ODN does associate with LDL, but that the complex slowly dissociates during electrophoresis. The lipid moiety of ODN-6 probably does not partition with its complete steroid structure into the lipids of LDL, but only with the two oleoyl chains. The complete bisoleoyl steroid ester structure may be too bulky to associate spontaneously with LDL. However, ODN-6 may well be incorporated in the recently developed artificial LDL-like carrier systems (16,17; see Note 4).

Fig. 2. Association of lipid-ODNs with LDL; analysis by agarose gel electrophoresis. Equimolar amounts of 125I-LDL and lipid(1.7 µM), dissolved in PBS + 1 mM EDTA, pH 7.4, were incubated for 2 h at 37°C. Aliquots of the incubation mixtures were subjected to gel electrophoresis in a 0.75% (w/v) agarose gel in 75 mM Tris-hippuric acid buffer (pH 8.8). After electrophoresis, the gel was cut into slices that were assayed for 3H-radioactivity (䊊), and 125I-radioactivity (䊉). The results are expressed as percentages of the recovered radioactivities. Recoveries were >95%.

(3H)ODNs

96 Rump et al.

96

Lipid–ODN Interaction with LDL

97

3.3.1. Isolation and Radioiodination of LDL Human LDL (density 1.024–1.063 g/mL) was isolated from the serum of fasted volunteers by density gradient ultracentrifugation (18) and were dialyzed against PBS containing 1 mM EDTA. Radioiodination was performed at pH 10.0 with carrier-free 125I as described by McFarlane (19). Protein concentrations of the lipoproteins were determined by the method of Lowry et al. (20) with bovine serum albumin as standard.

3.3.2. Determination of Association of Lipid–ODNs with LDL 1. Equimolar amounts of 125I-labeled LDL and 3H-labeled lipid-ODNs (1.7 µM), dissolved in PBS containing 1 mM EDTA, are incubated for 2 h at 37°C (see Note 5). 2. Aliquots of the incubation mixtures are subjected to gel electrophoresis in a 0.75% (w/v) agarose gel in 75 mM Tris-hippuric acid buffer, pH 8.8. 3. After electrophoresis, the gel is cut into slices, and the 125I radioactivity is counted after addition of 0.5 mL Soluene-350. 4. The gel slices are allowed to dissolve for 24 h at room temperature. 5. Then, 3 mL Hionic Fluor is added and samples are counted for 3H-radioactivity. The measured values of 3H-radioactivity are corrected for the contribution of 125I-radioactivity.

3.4. Exchange of Lipid–ODNs from LDL to Plasma Proteins The preceding subheading demonstrated that only the conjugates of the oligonucleotide with cholesterol or oleoyl steroid esters (ODN-4, ODN-5, and ODN-7) associate quantitatively with LDL. To examine the stability of the complexes of the lipid-ODNs with LDL in a biological matrix, the authors studied the exchange of the lipid-ODNs from preformed lipid-ODN–LDL complexes to components in rat plasma. Lipid-[3H]ODN–LDL complexes were incubated for 5 or 25 min with rat plasma and subsequently subjected to size-exclusion chromatography. The fractions were monitored for 3H-radioactivity, and the results are depicted in Fig. 3. Although HDL is the main lipoprotein present in rat plasma, no significant redistribution of ODN-4, ODN-5 and ODN-7 from LDL to HDL (elution vol 1.57 mL) was seen. The lipid-ODNs redistributed to some extent to triglyceride-rich lipoproteins (elution vol 0.9 mL). The oleoyl steroid ester-conjugated oligonucleotides (ODN-5 and ODN-7) (Fig. 3B,D) appeared to be most stably complexed with LDL, because after 25 min of incubation, >70% of the radioactivity still was associated with the LDL fractions. At that time, only approx 50% of the cholesteryl-conjugated ODN (ODN-4) (Fig. 3A) was found to be associated with LDL. The lipid-ODNs, ODN-4, ODN-5, and ODN-7, have a partially phosphorothioate (PS)-modified backbone (17% PS linkages). Many studies have found that PS linkages in oligonucleotides induce binding to serum proteins (22,23). Therefore, the authors also examined the stability of complexes of LDL and [3H]ISIS-9388, a full

98

Rump et al.

Fig. 3. Chromatographic profiles after incubation of lipid-ODN–LDL complexes with rat plasma. Preformed lipid-ODN/LDL complexes were incubated with rat plasma for 5 min (䊊) or for 25 min (䊉) at 37°C. The samples were subsequently injected onto a Superose 6 column (Pharmacia) and the fractions were assayed for 3H-radioactivity. The 3H-radioactivity in the fractions are expressed as percentage of recovered radioactivity (recoveries were >95%). The gray zones indicate the fractions containing 90% of the lipid-ODN–LDL complex at t = 0 min. (A) ODN-4; (B) ODN-5; (C) ODN-7; (D) ISIS-9388.

PS oligonucleotide that is conjugated at the 3' end with cholesterol. Upon incubation with rat plasma, ISIS-9388 dissociated more rapidly from a preformed complex with LDL than the partially PS-modified lipid-ODNs. After 25 min incubation, less then 20% of ISIS-9388 was found to be associated with LDL (Fig. 3D; see Note 6).

3.5. Determination of Exchange of Lipid–ODNs from LDL to Proteins in Plasma 1. Equimolar amounts of 125I-labeled LDL and 3H-labeled lipid-ODNs (200 pmol), dissolved in 100 µL of PBS containing 1 mM EDTA, are incubated for 2 h at 37°C. 2. Subsequently, 50 µL of the mixtures are injected onto a Superose 6 Precision Column (3.2 × 300 mm) (Smart System, Pharmacia) and eluted at a flow rate of 50 µL/min with PBS.

Lipid–ODN Interaction with LDL

99

3. 50 µL Fractions are collected and the three main fractions containing the lipidODN–LDL complexes are pooled. 4. 50 µL Aliquots of the pooled fractions are subsequently incubated at 37°C with 40 µL citrated rat plasma. 5. After incubation, the mixtures are injected onto the Superose 6 column and the column is eluted as described previously. The fractions are assayed for 3H-radioactivity.

3.6. Behavior of Lipid–ODN–LDL Complexes In Vivo To establish which of the lipid steroid structures is the most effective LDL anchor in vivo, the authors examined in the rat the plasma clearance of complexes of LDL and lipid-(3H)ODNs, ODN-4, ODN-5, and ODN-7. When injected without LDL, these lipid-ODNs were rapidly cleared from the circulation (Fig. 4A,C,E). At 5 min after injection, more than 95% of the dose was cleared, and a significant proportion of the radioactivity (15–40% of the dose) was recovered in the liver (Fig. 4B,D,F). The plasma clearance of the lipid-ODNs complexed with LDL was studied utilizing double-labeled complexes (lipid-(3H)ODN/(125I)LDL), which allows monitoring of both the lipid-ODN and the LDL carrier. Fig. 4A,C,E show that (125I)LDL was slowly cleared from the circulation, with a concomitant low liver uptake. These findings are consistent with the previously reported half-life of 5–6 h for LDL in the rat (8). The lipid-ODNs in the lipid-[3H]ODN/[125I]LDL complexes were cleared much slower than the noncomplexed lipid-ODNs (Fig. 4A,C,E). At 5 min after injection of the LDLcompexed lipid-ODNs, only 51, 39, and 24% of ODN-4, ODN-7, and ODN-5, respectively, had been cleared from the circulation (vs >95% for the noncomplexed lipid-ODNs). Thus, when complexed with LDL, both oleoyl steroid ester-conjugated oligonucleotides (ODN-5 and ODN-7) were more slowly cleared than the cholesteryl-conjugated ODN-4. Accordingly, the plasma area under the curve (AUC) of the LDL-complexed ODN-4 (1.49 ± 0.37 µg/min/mL) was significantly lower that the LDL-associated ODN-5 and ODN-7 (6.82 ± 1.34 µg/min/mL and 4.61 ± 0.38 µg/min/mL, respectively). The noncomplexed lipidODNs had plasma AUCs < 0.4 µg/min/mL. Compared to the free lipid-ODNs, the liver uptake of LDL-complexed lipid-ODNs was reduced (Fig. 4B,D,F). However, the lipid-ODN–LDL complexes were not completely stable, since the clearance of the lipid-ODNs did not completely resemble the clearance of LDL (Fig. 4A,C,E). The reduction of the clearance rate of the lipid-ODNs, achieved by complexation with LDL, was most evident for the oleoyl steroid ester-conjugated ODNs (ODN5 and ODN-7). The cholesteryl-conjugated ODN (ODN-4) displayed the highest leakage from LDL. 3.6.1. Determination of Plasma Clearance and Liver Association 1. Male Wistar rats are anesthetized by subcutaneous injection of a cocktail of ketamine-HCl, fentanyl, droperidol, and fluanisone (75, 0.04, 1.1, and 0.75 mg/kg, respectively).

100

Rump et al.

Fig. 4. Plasma clearance and liver association of lipid-ODNs and lipid-ODN–LDL complexes. Rats were injected with the free lipid-[3H]ODNs (䉱) or lipid-(3H)ODN/ (125I)LDL complexes (3H,䊊; 125I, 䊉). At the indicated times, the amounts of radioactivity in plasma (A,C,E) and liver (B,D,F) were determined. Values are means ± SEM of two (free lipid-ODNs) or three (ODN–LDL complexes) separate experiments. A and B: ODN-4; C and D: ODN-5; E and F: ODN-7. 2. The abdomen is opened. Free lipid-[3H]ODNs (4 µM in PBS), or complexes of lipid-[3H]ODNs with (125I)LDL (4 µM, complexes prepared as described above), are injected via the vena penis at a dose of 5 µg lipid-ODN/kg body wt.

Lipid–ODN Interaction with LDL

101

3. At the indicated times, 250 µL blood samples are taken from the inferior vena cava and collected in heparinized tubes. The blood samples are centrifuged for 10 min at 500g, and the plasma is assayed for radioactivity. 4. The total amounts of radioactivity in plasma are calculated using the equation: Plasma vol (mL) = [0.0291 × body wt (g)] + 2.54 (24) 5. At the indicated times, liver lobules are tied off and excised, and at the end of the experiment, the remainder of the liver is removed. 6. The amount of radioactivity in the liver at each time-point is calculated from the radioactivities and weights of the liver samples and is corrected for radioactivity in plasma present in the tissue at the time of sampling (85 µL/g fresh weight). 7. The plasma concentration-time AUC is calculated by computerized nonlinear fitting (Graphpad Prism, Graphpad Software, San Diego, CA).

3.6.2. Determination of Radioactivity Samples containing 3H are counted in a Packard Tri-Carb 1500 liquid scintillation counter. Liquid samples are counted without further processing using Emulsifier Safe or Hionic Fluor scintillation cocktails. Agarose gel slices are first dissolved in Soluene-350. Tissue samples are processed using a Packard 306 Sample Oxidizer. Samples containing both 125I and 3H, are first assayed for 125I-radioactivity using a Packard Auto-Gamma 5000 counter. The 3H-radioactivity is subsequently measured as described above, and corrected for the contribution of 125I-radioactivity.

3.7. Discussion This chapter describes the conjugation of an amino-terminated oligonucleotide with several lipid structures in solution phase. The conjugated lipid structures did not affect the association of the oligonucleotide with its target sequence, as judged by the lack of effects on the melting temperatures of the antisense–sense hybrids. Several of the lipid-ODNs, namely those conjugated with cholesterol or oleoyl steroid ester moieties, associate readily with LDL. The stability of the complexes of these lipid-oligonucleotides with LDL was investigated in vitro by incubation with rat plasma and in vivo in rats. Detailed reports of the experiments presented here were published earlier (11,25). In vitro, the authors examined the exchange of the lipid-oligonucleotides from preformed lipid-ODN–LDL complexes to components in rat plasma. The more lipidic oleoyl steroid ester structures appeared to be better LDL-anchors than the cholesteryl moiety. ISIS-9388 (a 3'-cholesteryl-conjugated phosphorothioate oligonucleotide) redistributed to a much higher extent from LDL to plasma proteins than the partially phosphorothioate-modified lipid-ODNs (ODN-4, ODN-5, and ODN-7). This is likely to be the result of the high affinity of the full phosphorothioate oligonucleotide for plasma proteins (22,23).

102

Rump et al.

When the lipid-ODNs were injected in rats without LDL, they were rapidly cleared from the circulation, meaning that lipid-conjugation alone is not sufficient to achieve a prolonged half-life in the circulation. A substantial amount of the lipidODNs was recovered in the liver, which may be ascribed to recognition by scavenger receptors on liver cells (13,26). When the lipid-ODNs were administered as complexes with LDL, the plasma clearance of the lipid-ODNs was considerably delayed and their liver uptake reduced. The oligonucleotides containing oleoyl steroid ester structures (ODN-5 and ODN-7) were more slowly cleared than the cholesteryl-conjugated ODN-4. The clearance and liver uptake of the LDL particles in the complexes was not altered, which indicates that complexation with the lipidODNs does not affect the integrity of the particle. Taken together, of all the steroid structures tested, the lithocholic acid-3_oleate structure (present in ODN-5) most effectively reduced clearance of the oligonucleotide in vivo and the exchange from LDL in vitro. The strong association of this steroid ester lipid anchor with the lipids of LDL is probably primarily responsible for the prolonged circulation.The backbone chemistry of the oligonucleotide is also important. A modification with low affinity for plasma proteins (e.g., oligoncucleotide with little phosphorothioate linkages or nonphosphorothioate oligonucleotides, such as morpholino or peptide nucleic acid oligomers [27,28]), will reduce dissociation of lipid-conjugated oligonucleotides from lipoproteins. Oligonucleotides conjugated with the oleoyl steroid esters remain associated to LDL and the retarded plasma clearance results in a higher exposure to target cells. Oligonucleotides associated with LDL (or LDL-like particles [16,17]) may be taken up along with the particles via LDL receptors that are overexpressed on various types of tumor cells, e.g., leukemic cells (9). The firm association of lipid-ODNs with LDL or LDL-like particles thus has the additional advantage that enhanced exposure is accompanied by enhanced rate of uptake by LDL receptor-overexpressing tumor cells. 4. Notes 1. Radiolabeling of oligoncucleotides. The 3'-amine antisense oligonucleotide was radiolabeled with 3H by heat-catalyzed exchange at the C8 positions of the purine nucleotides as described previously (11,12). The radiolabeled oligonucleotide was stored at –20°C. No loss of radioactivity from the oligonucleotide was detected during 6 mo of storage at –20°C. The specific activity of the radiolabeled oligonucleotide was 80 × 106 dpm/mg. 2. Differences in melting temperatures measured in duplicate runs were <1°C. 3. Association of the lipid-ODNs with LDL also slightly broadened the band of LDL in the gel, probably because of an increase of the overall negative charge of the complex. 4. In these systems, lipid-ODNs can be incorporated during preparation of the particles, in contrast to the spontaneous association with LDL in this study.

Lipid–ODN Interaction with LDL

103

5. For these experiments, the molecular weight of an LDL particle was related to the molecular weight of the apoprotein, B-100 (21) because each LDL particle contains only one copy of the apoprotein. 6. No radioactivity eluted at the column volume after incubation of ODN-4, ODN5, or ODN-7 with plasma. This implies that no radiolabeled oligonucleotides were generated during the incubation, which indicates that the 3 PS-linkages at the 5' end and the amino linker–steroid lipid structures at the 3' end effectively protect the lipid-ODNs against nuclease activity in the plasma.

References 1. Akhtar, S. and Agrawal, S. (1997) In vivo studies with antisense oligonucleotides. Trends Pharm. Sci. 18, 12–18. 2. Branch, A. D. (1996) Hitchhiker’s guide to antisense and nonantisense biochemical pathways. Hepatology 24, 1517–1529. 3. Szymkowski, D. E. (1996) Developing antisense oligonucleotides from the laboratory to the clinic. Drug Discov. Today 1, 415–428. 4. Stein, C. A. (1995). Does antisense exist? Nature Med. 1, 1119–1121. 5. Sands, H., Gorey-Feret, L. J., Cocuzza, A, J., Hobbs, F. W., Chidester, D., and Trainor, G. L. (1994) Biodistribution and metabolism of internally 3H-labeled oligonucleotides. I. Comparison of a phosphodiester and a phosphorothioate. Mol. Pharmacol. 45, 932–943. 6. Nolting, A., DeLong, R. K., Fisher, M. H., Wickstrom, E., Pollack, G. M., Juliano, R. L., and Brouwer, K. L. (1997) Hepatic distribution and clearance of antisense oligonucleotides in the isloated perfused rat liver. Pharm. Res. 14, 516–521. 7. Brown, M. S. and Goldstein, J. L. (1986) A receptor-mediated pathway for cholesterol homeostasis. Science 232, 34–47. 8. Harkes, L. and van Berkel, T. J. C. (1983) Cellular localization of the receptordependent and receptor-independent uptake of human LDL in the liver of normal and 17 alpha-ethinyl estradiol-treated rats. FEBS Lett. 154, 75–80. 9. Firestone, R. A. (1994) Low-density lipoprotein as a vehicle for targeting antitumor compounds to cancer cells. Bioconjugate Chem. 5, 105–113. 10. Majello, B., Kenyon, L. C., and Dalla-Favera, R. (1986) Human c-myb protooncogene: nucleotide sequence of cDNA and organization of the genomic locus. Proc. Nat. Acad. Sci. USA 83, 9636–9640. 11. Rump, E. T., de Vrueh, R. L. A., Sliedregt, L. A. J. M., Biessen, E. A. L., van Berkel, T. J. C., and Bijsterbosch, M. K. (1998) Preparation of conjugates of oligonucleotides and lipid structures, and their interaction with low-density lipoprotein. Bioconjugate Chem. 9, 341–349. 12. Graham, M. J., Freier, S. M., Crooke, R. M., Ecker, D. J., Maslova, R. N., and Lesnik, E. A. (1993) Tritium labeling of antisense oligonucleotides by exchange with tritiated water. Nucleic Acids Res. 21, 3737–3743. 13. Bijsterbosch, M. K., Rump, E. T., De Vrueh, R. L. A., Dorland, R., Van Veghel, R., Tivel, K. L., et al. (2000) Modulation of plasma protein binding and in vivo liver cell uptake of phosphorothioate oligodeoxynucleotides by cholesterol modification. Nucleic Acids Res. 28, 2717–2725.

104

Rump et al.

14. Manoharan, M., Tivel, K. L., and Cook, P. D. (1995) Lipidic nucleic acids. Tetrahedron Lett. 36, 3651–3654. 15. Firestone, R. A., Pisano, J. M., Falck, J. R., McPaul, M. M., and Krieger, M. (1984) Selective delivery of cytotoxic compounds to cells by the LDL pathway. J. Med. Chem. 27, 1037–1043. 16. Rensen, P. C. N., Herijgers, N., Netscher, M. H., Meskers, S. C. J., van Eck, M., and van Berkel, T. J. C. (1997) Particle size determines the specificity of apolipoprotein E-containing triglyceride-rich emulsions for the LDL receptor versus hepatic remnant receptor in vivo. J. Lipid Res. 38, 1070–1084. 17. Rensen, P. C. N., Schiffelers, R. M., Versluis, A. J., Bijsterbosch, M. K., van KuijkMeuwissen, M. E. M. J., and van Berkel, T. J. C. (1997) Human recombinant apolipoprotein E-enriched liposomes can mimic low density lipoproteins as carriers for the site-specific delivery of anti-tumour agents. Mol. Pharmacol. 52, 445–454. 18. Redgrave, T. G., Roberts, D. L. K., and West, C. E. (1975) Separation of plasma lipoproteins by density gradient ultracentrifugation. Anal. Biochem. 65, 42–49. 19. McFarlane, A. S. (1958) Efficient trace-labelling of proteins with iodine. Nature 182, 53–54. 20. Lowry, O. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J. (1951) Protein measurement with the Folin phenol reagent. J. Biol. Chem. 193, 265–275. 21. Chen, S. H., Yang, C. Y., Chen, P. F., Setzer, D., Tanimuara, M., Li, W. H., Gotto, A. M., Jr., and Chan, L. (1986) The complete cDNA and amino acid sequence of human apolipoprotein B-100. J. Biol. Chem. 261, 12,918–12,921. 22. Cossum, P. A., Sasmor, H., Dellinger, D., Truong, L., Cummins, L., Owens, S. R., et al. (1993) Disposition of the C-14-labeled phosphorothioate oligonucleotide ISIS 2105 after intravenous administration to rats. J. Pharmacol. Exp. Ther. 267, 1181–1190. 23. Srinivasan, S. K., Tewary, H. K., and Iversen, P. L. (1995) Characterization of binding sites, extent of binding, and drug interactions of oligonucleotides with albumin. Antisense Res. Dev. 5, 131–139. 24. Bijsterbosch, M. K., Duursma, A. M., Bouma, J. M. W., and Gruber, M. (1981) The plasma volume of the Wistar rat in relation to the body weight. Experientia 37, 381–382. 25. Rump, E. T., de Vrueh, R. L. A., Manoharan, M., Waarlo, I. H. E., van Veghel, R., Biessen, E. A. L., van Berkel, T. J. C., and Bijsterbosch, M. K. (2000) modification of the plasma clearance and liver uptake of steroid ester-conjugated oligodeoxynucleotides by association with (lactosylated) low-density lipoprotein. Biochem. Pharmacol. 59, 1407–1416. 26. Biessen, E. A. L., Vietsch, H., Kuiper, J., Bijsterbosch, M. K., and van Berkel, T. J. C. (1998) Liver uptake of phosphodiester oligodeoxynucleotides is mediated by scavenger receptors. Mol. Pharmacol. 53, 262–269. 27. Summerton, J. (1999) Morpholino antisense oligomers: the case for an RNAse H-independent structural type. Biochim. Biophys. Acta 1489, 141–158. 28. Larsen, H. J., Bentin, T., and Nielsen, P. E. (1999) Antisense properties of peptide nucleic acid. Biochim. Biophys. Acta 1489, 159–166.

Synthesis of Plasmid-NLS Conjugates

105

8 Coupling of Nuclear Localization Signals to Plasmid DNA Carole Neves, Daniel Scherman, and Pierre Wils 1. Introduction This chapter focuses on a methodology for covalently associating nuclear localization signal (NLS) peptides to DNA, in which cationic NLS peptides are covalently bound to plasmid DNA by photoactivation. Described here are the synthesis and characterization of these conjugates. The NLS of the SV40 large T antigen possesses five basic amino acids and efficiently induces nuclear targeting of proteins (1). Moreover, nuclear import is increased when several independent SV40 large T antigen NLS peptides are conjugated to a protein (2). Karyopherin _ (also called importin _ or m-importin) binds to NLS sequence in the cytoplasm, then interacts with karyopherin ` (also called importin `). The resulting complex binds to the nuclear pore and is translocated inside the nucleus in a mechanism involving the small guanosine triphosphatase Ran and other proteins (3). Association of NLS peptides to plasmid DNA is thus an attractive strategy for improving nuclear import of plasmid DNA, which is an important limiting step during nonviral gene transfer (4). Various noncovalent or covalent techniques have been described so far for modification of plasmid DNA with fluorescent molecules or targeting peptides. DNA-intercalating dyes such as ethidium bromide (5) or cyanine dyes (6) have been traditionally used to label DNA. Recently, fluorescent peptide nucleic acids (PNA) were used to label functionally intact plasmid DNA (7). These PNA can be used to associate targeting sequences to DNA, as described with PNA–NLS peptide conjugates (8). All methods using molecules noncovalently associated with plasmid DNA raise the possibility that the labeling or targeting molecule dissociates from plasmid DNA during intracellular trafficking. Some covalent labeling methods use nucleic acid modifying enzymes in order to From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

105

106

Neves, Scherman, and Wils

incorporate labeled nucleotides into DNA. Thus, polymerase chain reaction and nick translation were used to incorporate fluorescent nucleotides into plasmid DNA (9,10). However, these techniques induce relaxation or linearization of plasmid DNA. Direct covalent coupling of signal peptides by chemical means, which maintain structural integrity of plasmid DNA, represents an alternative approach (11). The use of photoactive molecules is an efficient method for covalent association of molecules, such as biotin or ethidium bromide to plasmid DNA (12,13). It was shown (14) that tetrafluorophenylazido reagents can modify more chemical groups than their nonfluorinated analogs. The authors developed a new chemical strategy for covalent coupling of NLS peptides to plasmid DNA: For this purpose, a p-azido-tetrafluoro-benzyl–NLS peptide conjugate was synthesized and used to covalently associate NLS peptides to plasmid DNA by photoactivation (Fig. 1). 2. Materials 2.1. Peptide Synthesis 1. Applied Biosystems 431A automatic synthesizer. 2. Dichloromethane. 3. Lissamine rhodamine B sulfonyl chloride (purchased from Aldrich, mixture of two isomers). 4. Diisopropylethylamine (DIEA, pH 10.0). 5. Dimethylaminopyridine (DMAP).

2.2. Preparation of p-Azido-Tetrafluoro-Benzyl-Peptide Conjugates 1. N-(4-azido-2,3,5,6-tetrafluorobenzyl)-6-maleimidylhexanamide (TFPAM-6) (Molecular Probes, Eugene, OR). 2. Dimethylsulfoxide. 3. Triethylammonium acetate, pH 7.5.

2.3. High-Performance Liquid Chromatography 1. Merck-Hitachi gradient pump equipped with an AS-2000A autosampler, a L-6200A intelligent pump, and a UV-visible detector L-4000. 2. Method A: For analytical separations, 218 TP54 Vydac C18 column (Interchim, Montluçon, France) with a gradient of acetonitrile/water 0.1% trifluoroacetic acid (from 0 to 41% acetonitrile in 25 min) and a flow rate of 1 mL/min. For semipreparative separation, a 218 TP 1022 (10 mm) Vydac C18 column with a gradient of acetonitrile/water 0.1% trifluoroacetic acid (from 5 to 50% acetonitrile in 40 min) and a flow rate of 7 mL/min. Detection by UV absorbance at 254 nm. 3. Method B: C8 column Vydac 250 × 4.6 mm with a gradient of 5–50% acetonitrile over 40 min in 0.1 M triethylammonium acetate, pH 7.5 (flow rate = 1 mL/min). Detection by UV absorbance at 250 nm (absorbance of TFPAM-6).

Synthesis of Plasmid-NLS Conjugates

107

Fig. 1. Synthesis of p-azido-tetrafluoro-benzyl-peptide conjugates. A lissamine rhodamine group can be coupled at the N-terminal position of the peptide. X = K or N, for NLS and mNLS peptide, respectively.

2.4. Covalent Coupling of Peptide Conjugates to dsDNA 1. Plasmid (about 7000 bp) purified using Wizard Megaprep kit (Promega, Madison, WI). 2. Ice-water bath. 3. Sun lamp (Rad Free UV lamp-365 nm, Schleicher and Schuell, Ecquevilly, France). 4. Ion-exchange chromatography columns (PCR Purification kit, Qiagen, Hilden, Germany).

3. Methods 3.1. Peptide Synthesis 1. The SV40 large T NLS sequence (NLS), `ACGAGPKKKRKV, or its mutant form, `ACGAGPKNKRKV (mNLS), were synthesized using Rink amide resin and Fmoc strategy. Introduction of a point mutation in the NLS sequence (asparagine instead of lysine) induces loss of NLS targeting activity (1). The sequence CGAG was used as a spacer, and cysteine provided a free thiol group for coupling to the photoactivable linker. 2. Alternatively, a fluorescent probe, Lissamine rhodamine B, was coupled to the N-terminal function of the NLS or mNLS peptide. This probe makes it possible to follow the intracellular traffic of plasmid–peptide conjugates by fluorescence microscopy, and allows a easy quantification of the number of peptides linked to plasmid DNA by spectrofluorimetry. 3. A solution containing 10 mol/equivalent excess of Lissamine rhodamine B sulfonyl chloride in dichloromethane, DIEA (pH 10.0), and a catalytic amount of DMAP was reacted with the peptide on the resin. 4. The peptides were cleaved from the Rink resin, purified, and analyzed by highperformance liquid chromatography using method A. 5. The final products were analyzed by mass spectroscopy.

3.2. Synthesis of p-Azido-Tetrafluoro-Benzyl-Peptide Conjugates 1. The crosslinker TFPAM-6 bears a maleimide group that reacts with the nucleophilic thiol residues on the peptides and a tetrafluoro-azido moiety for subsequent photoactivation at 360 nm. 2. Lyophilized TFPAM-6 was dissolved in dimethylsulfoxide at a concentration of 1 mg/mL.

108

Neves, Scherman, and Wils

3. Lyophilized peptides were dissolved in 100 mM triethylammonium acetate, pH 7.5, at a concentration of 1 mg/mL. 4. Peptide and TFPAM-6 (1:1 molar ratio) were reacted for 2 h at room temperature. 5. The reaction was followed by HPLC (method B), which showed complete disappearance of TFPAM-6 after 2 h. The product was purified using the same HPLC method and analyzed by mass spectroscopy. The yield of the reaction was 90%.

3.3. Covalent Coupling of Peptide Conjugates to dsDNA 1. p-azido-tetrafluoro-benzyl-peptide conjugate was mixed with dsDNA (final concentration 0.3 µg/µL) at peptide:DNA mol ratios varying from 5 to 1000, depending on the level of DNA modification needed. 2. The mixture was cooled in an ice-water bath and illuminated for 15 min at 365 nm in a dark room. 3. Excess p-azido-tetrafluoro-benzyl-peptide conjugate was removed by ion-exchange chromatography.

3.4. Plasmid–Peptide Conjugates Characterization 1. If fluorescent peptide was used, the average number of peptides attached to plasmid DNA was measured by fluorimetry (excitation, 544 nm; emission, 590 nm). Free illuminated p-azido-tetrafluoro-benzyl-peptide conjugate was used for the standard curve. 2. In the final plasmid–peptide conjugates, about 20% of the added p-azidotetrafluoro-benzyl-peptide conjugates were found to have reacted with plasmid DNA. Depending on the molar ratio used for plasmid modification, between 1 and 300 peptide molecules were bound per plasmid molecule (7257 bp). For example, when a 500-fold mol excess of NLS peptide over DNA was used, an average number of 100 NLS peptides were linked per plasmid DNA molecule. 3. If nonfluorescent peptides were used, the number of peptides bound to DNA could be estimated using a protein assay kit such as BCA (Pierce). 4. Modified DNA was analyzed by agarose gel electrophoresis. When less than 100 peptides were linked per plasmid DNA molecule, peptide–plasmid conjugates migrated on gels as supercoiled DNA. With higher numbers of peptides per plasmid DNA molecule, plasmid DNA migration was retarded (15). 5. Conjugates were shown to specifically interact with NLS receptor, importin _, using an in vitro binding assay. Reporter gene was expressed after transfection of the NLS–plasmid conjugates into cultured fibroblasts. Fluorescent NLS–plasmid conjugates were used for intracellular trafficking studies (15).

References 1. Lanford, R. E., Kanda, P., and Kennedy, R. C. (1986) Induction of nuclear transport with a synthetic peptide homologous to th SV40 T Antigen transport signal. Cell 46, 575–582. 2. Dworetzky, S. I., Lanford, R. E., and Feldherr, C. M. (1988) The effects of variations in the number and sequence of targeting signals on nuclear uptake. J. Cell Biol. 107, 1279–1287.

Synthesis of Plasmid-NLS Conjugates

109

3. Nigg, E. A. (1997) Nucleocytoplasmic transport: signals, mechanisms and regulation. Nature 386, 779–787. 4. Scherman, D., Bessodes, M., Cameron, B., Herscovici, J., Hofland, H., Pitard, B., et al. (1998) Application of lipids and plasmid design for gene delivery to mammalian cells. Curr. Opin. Biotech. 9, 480–485. 5. Le Pecq, J. B. (1971) Use of ethidium bromide for separation and determination of nucleic acids of various conformational forms and measurement of their associated enzymes. Methods Biochem. Anal. 20, 41–86. 6. Haugland, R. P. (1996) Nucleic acid stains, in Handbook of Fluorescent Probes and Research Chemicals (Spence, M. T. Z., ed.), Molecular Probes, Eugene, OR, pp. 144–156. 7. Zelphati, O., Liang, X., Hobart, P., and Felgner, P. L. (1999) Gene chemistry: functionally and conformationally intact fluorescent plasmid DNA. Hum. Gene Ther. 10, 15–24. 8. Branden, L. J., Mohamed, A. J., and Smith, C. I. E. (1999) A peptide nucleic acidnuclear localization signal fusion that mediates nuclear transport of DNA. Nature Biotech. 17, 784–787. 9. Hagstrom, J. E., Luddtke, J. J., Bassik, M. C., Sebestyén, M. G., Adam, S. A., and Wolff, J. A. (1997) Nuclear import of DNA in digitonin-permeabilized cells. J. Cell Sci. 110, 2323–2331. 10. Escriou, V., Ciolina, C., Lacroix, F., Byk, G., Scherman, D., and Wils, P. (1998) Cationic lipid-gene transfer: effect of serum on cellular uptake and intracellular fate of lipopolyamine/ DNA complexes. Biochim. Biophys. Acta 1368, 276–288. 11. Sebestyen, M. G., Ludtke, J. J., Bassik, M. C., Zhang, G., Budker, V., Lukhtanov, E. A., Hagstrom, J. E., and Wolff J. A. (1998) DNA vector chemistry: the covalent attachment of signal peptides to plasmid DNA. Nature Biotech. 16, 80–85. 12. Forster, A. C., McInnes, J. L., Skingle, D. C., and Symons, R. H. (1985) Nonradioactive hybridization probes prepared by the chemical labeling of DNA and RNA with a novel reagent, photobiotin. Nucl. Acids Res. 13, 745–761. 13. Tseng, W., Purvis, N. B., Haselton, F. R., and Giorgio, T. D. (1996) Cationic liposomal delivery of plasmid to endothelial cells measured by quantitative flow cytometry. Biotech. Bioeng. 50, 548–554. 14. Levina, A. S., Tabatadze, D. R., Dobrikov, M. I., and Shishkin, G. V. (1996) Sitespecific photomodification of single-stranded DNA targets by arylazide and perfluoroarylazide derivatives of oligonucleotides. Antisense Nucleic Acid Drug Dev. 6, 119–126. 15. Ciolina, C., Byk, G., Blanche, F., Thuillier, V., Scherman, D., and Wils, P. (1999) Coupling of nuclear localization signals to plasmid DNA and specific interaction of the conjugates with importin _. Bioconjugate Chem. 10, 49–55.

Synthetic Virus for Liver-Directed Delivery

111

9 Progress Toward a Synthetic Virus A Multicomponent System for Liver-Directed DNA Delivery Bo-Hua Zhong, George Y. Wu, and Catherine H. Wu 1. Introduction Vectors for gene transfer can be categorized as viral and nonviral. The advantages of nonviral carriers are their ease of preparation and scale-up, flexibility regarding the size of DNA to be transferred, and safety in vivo. Despite these advantages, nonviral vectors need to be further optimized for their efficiency is generally low. Thus, the future of non-viral vectors will be dependent on the possibility of creating synthetic efficient systems. A possible and reasonable approach is to develop artificial nucleic acid carriers that incorporate functional elements mimicking viruses. Viruses contain core proteins in their interiors that package their nucleic acid genome into small compact structures. The surface proteins of many viruses contain two further functions essential for the delivery of the viral genome into the cytoplasm of the target cells: the binding of the virus particle to a cellular surface receptor that mediates endocytosis of the virion into endosomes; and, after endocytosis, the release of the viral genome from endosomes into the cytoplasm by disruption of the endosome membranes. Based on an understanding of these natural components, ligand–polycation DNA delivery systems have been developed (1–3). In these systems, a DNA-binding polycation, such as polylysine was employed to compact plasmid DNA to a size that could be taken up by cells. To allow internalization by receptor-mediated endocytosis, cell binding ligands such as asialoglycoproteins for hepatocytes, antiCD3 and anti-CD5 antibodies for T cells, and transferrin for some cancer cells have been covalently attached to polylysine. The liver is an important target organ for gene therapy because it plays a central role in the metabolism and production of serum proteins. There are From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

111

112

Zhong, Wu, and Wu

many metabolic diseases that result from a defect or deficiency of hepatocytederived gene products. Acquired diseases such as hepatocellular carcinomas and viral hepatitis may also serve as targets for hepatic gene therapy. Advancements in hepatic gene therapy depend to a large degree on the development of delivery systems capable of efficiently introducing genes into the hepatocytes. Parenchymal liver cells (hepatocytes) are useful target cells for gene delivery because they can perform a host of posttranslational modifications that may be required for the activity of certain gene products (4). The cells are highly active metabolically. The liver has a rich blood supply that can be useful for the delivery of genes to the liver, and also for the distribution of gene products from the liver to the systemic circulation. Finally, mammalian hepatocytes are the only cells that possess large numbers of high-affinity cell-surface receptors that can bind asialoglycoproteins (5). Work in this laboratory (1,2) revealed for the first time that DNA could be delivered specifically to and expressed in the liver in vivo with an asialoglycoproteinmediated system. However, the efficiency in vivo has been poor. The authors have previously shown that incorporation of an endosome disruptive peptide into the delivery system could greatly increase the specific gene expression to liver in vivo. However, because of the activation of the complement system by the positively charged surface of the DNA complexes, and weak compaction of DNA by the conjugate, the DNA complex formed was not stable enough for efficient use in vivo. Furthermore, the tendency to form macromolecular aggregates at higher concentrations made it difficult to deliver large quantities of DNA in small volumes. The objective of the present study was to construct a DNA delivery system that has high water solubility, serum stability, and high gene expression efficiency. First, poly-L-lysine (PLL) was used as DNA-compacting component. To PLL was attached polyethylene glycol (PEG) to provide a biocompatible protective coating for the DNA complex. An endosomolytic peptide derived from vesicular stomatitis viral G protein (VSV) was further introduced to produce the conjugate PEG-PLL-VSV. VSV can induce membrane changes at low pH, allowing the internalized DNA to escape from lysosomal digestion. A combination of complexes containing two conjugates and four components provided incorporation of PLL and VSV for DNA compaction and endosome disruption, respectively, and a coating of PEG was added to the DNA complex to decrease interaction with biological systems, and increase serum stability and water solubility. Finally, an asialoglycoprotein, asialoorosomucoid (AsOR), was added as a hepatocyte-targeting ligand. 2. Materials 1. Plasmid DNA (e.g., a plasmid pCMVluc, containing a firefly luciferase gene driven by a cytomegalovirus immediate early promoter was amplified in Escherichia coli, isolated by alkaline lysis and purified by cesium chloride gradient centrifugation [6]).

Synthetic Virus for Liver-Directed Delivery 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17.

113

PLL hydrobromide salt (Sigma, mol wt 3970). Phosphate buffered saline (PBS), at varied concentrations of NaCl, pH 7.2. Aqueous NaCl, varied concentrations for chromatography. 2 N NaOH. PEG-succinyl ester, mol wt 5000 (Sigma, M3152). Ion exchange column, TSK-GEL CM-650, Supelco, 40–90 µ. Bio-gel P-6 gel filtration column (Bio-Rad). 0.2 M NH4HCO3. 0.5 M EDTA. VSV G peptide TIVFPHNQKGNWKNVPSNYHYCP (Immune Response, Carlsbad, CA). Orosomucoid. Succinimidyl 3-(2-pyridyldithio) propionate (SPDP) (Pierce). 0.2 µm Syringe filters. Huh7 cells. Dulbecco's modified Eagle’s medium (DMEM) cell culture media. Tissue lysis buffer (Promega).

3. Methods 3.1. Synthesis of PEG-PLL Conjugates 1. Dissolve PLL-HBr (30 mg) in 1.5 ml PBS (0.1 M, pH 7.2), adjusted to ~pH 7.0–8.0 with addition of 2 N NaOH. 2. To the solution add 10 mg PEG-succinyl ester and incubate at room temperature (23°C) for 5 h. 3. Dilute the reaction solution with 10 mL water, and chromatograph on a 2 × 10 cm TSK-GEL CM-650 ion exchange column (Supelco, 40–90 µm). The sample is eluted with 50 mL water, then with 200 mL 0–5.0 M NaCl gradient. Absorbance is monitored by UV at 230 nm. The second peak is collected and freeze-dried. 4. The powder is dissolved in 2 mL water. The sample is gel-filtered through a 2 × 50-cm Bio-gel P-6 (Bio-Rad) column, and eluted with 0.2 M NH4HCO3. The first peak containing PEG-PLL is collected in 3 mL and freeze-dried to give 10 mg white powder.

3.2. Synthesis of PEG-PLL-VSV Conjugate 1. PEG-PLL, 8.5 mg, is dissolved in 1.0 mL 0.2 M PBS (pH 7.2). 2. To the solution, 1.2 mg SPDP in 0.2 mL tetrahydrofuran is added. 3. After stirring at 23°C for 3 h, the product is gel-filtered through a 2 × 50-cm Biogel P-6 column and eluted with 0.2 M NH4HCO3 buffer. 4. The first peak is collected and freeze-dried to give PEG-PLL-dithiopyridyl (DTP). 5. PEG-PLL-DTP, 7.0 mg, is dissolved in 1.0 mL PBS (pH 7.2). To the solution is added 0.1 mL 0.5 M EDTA and 9.2 mg VSV G protein peptide (TIVFPHNQKGNWKNVPSNYHYCP) in 0.2 mL of water and incubated at 23°C for 24 h. 6. The reaction solution is then gel-filtered through a Bio-gel P-6 column (2 × 50 cm), eluted with 0.2 M NH4HCO3.

114

Zhong, Wu, and Wu

7. The first peak containing PEG-PLL-VSV is collected and freeze-dried to give 6.0 mg powder.

3.3. Synthesis of Asialoorosomucoid-Polylysine (AsOR–PLL) Conjugates AsOR was prepared by desialylation of orosomucoid from pooled human serum, and coupled to PLL, as described previously (1,7). Briefly, AsOR was modified with SPDP to introduce a DTP group, which then reacted with the thiol group in PLL introduced with SPDP to form a conjugate with a disulfide linkage. The conjugate was purified on an ion exchange column, 2 × 10 cm (TSK-GEL CM-650, Supelco, 40–90 µ).

3.4. Formation of Multicomponent DNA Complexes with PEG-PLL-VSV–AsOR-PL Complexes of DNA with conjugates were formed in saline: 30 µg DNA in 1.0 mL 0.15 M saline was first complexed with PEG-PLL-VSV in 1.0 mL saline and further complexed with AsOR-PLL in 1.0 mL saline. The complex formed was filtered through a 0.2-µ filter and the DNA concentration was determined by measuring UV absorption at 260 nm. The filtered solution was stored at 4°C for all further experiments.

3.5. Measurement of DNA Binding and Compaction To measure the compaction of DNA after complexation with various conjugates, fluorescence of ethidium bromide (EtBr) excluded from DNA complexes was used. Fluorescence studies were performed using a Perkin-Elmer luminescence spectrometer at an excitation wavelength of 516 nm (slit width 6 nm) and an emission wavelength of 598 nm (slit width 10 nm). To 2.5 mL saline solution, containing 30 µg DNA, was added 3.0 µg ethidium bromide, and a baseline fluorescence was determined. Ethidium bromide fluorescence of DNA complexed to various amounts of different conjugates was measured. The fluorescences of the complexes were corrected for dilution as a result of the addition of the conjugate solutions, and normalized to the fluorescence of free DNA before complexation, which was assigned a value of 100.

3.6. Particle Size and c-Potential To determine the effective hydrodynamic diameter and the net charge of DNA complexes, a 90 Plus Particle Size Analyzer (Brookhaven) was used. All samples were dissolved in 0.15 M NaCl, and measurements performed in triplicate at 23°C.

Synthetic Virus for Liver-Directed Delivery

115

3.7. Cell Transfections and Luciferase Activity Assays AsOR-receptor-positive Huh7 cells were seeded into 24-well plates. After 20 h, old media were removed and 0.5 mL DMEM media added, and 100 µL (1 µg DNA) DNA as complexes was added. To assess specificity, a 100-fold molar excess of free AsOR, more than that calculated to be present in the complexes, was mixed with complexes prior to administration to cells. After incubation for 6 h, 50 µL fetal bovine serum was added to each well, the cells were further incubated for 20 h, then the media were removed, the cell layers washed with PBS, homogenized with 200 µL tissue lysis buffer (Promega), and centrifuged at 8000g for 5 min. 20 µL Supernatant solution was mixed with 50 µL luciferase substrate, and relative light units (RLU) measured with a luminometer (Monolight 2001, Analytical Luminescence) for 30 s. All assays were performed in triplicate and results expressed as means ± SD in units of RLU.

3.8. Stability of Complexes Stability of complexed DNA was determined by incubation of DNA complexes with fresh rat serum or saline for 30 min. To determine the status of the DNA in the complexes, DNA was released from the complexes by heparin (2000 U/mL) after 30 min of incubation. All samples were analyzed on 1% agarose gel.

3.9. Characterization of Conjugates PLL was first conjugated with PEG to form conjugate PEG–PLL. The number of amino groups in the conjugate was determined by ninhydrin assay (8), and, from that, the content of PLL in the conjugate was calculated. The average ratio of PEG:PLL was determined to be 1.2:1. The conjugate was then modified with SPDP to introduce dithiopyridyl (DTP) groups for conjugating with VSV, which has a free thiol group in the C-terminal cysteine residue. The ratio of DTP:PEG-PLL was determined to be 2.7. The conjugation of PEG-PLLDTP with VSV was monitored by measuring the absorption at 343 nm resulting from the release of 2-mercaptopyridine (7). The content of VSV in the conjugate was determined from UV absorption at 280 nm, and the ratio of components in the conjugate PEG-PLL-VSV to be 1.2:1:1.6, respectively.

3.10. Ethidium Bromide Exclusion and Particle Size of Complexes The compaction of DNA by PEG-PLL-VSV was assessed by ethidium bromide exclusion and particle size analysis. In general, the higher exclusion of free ethidium bromide and the smaller the particle size, the better the DNA was compacted. As shown in Table 1, with the increase in ratio of conjugate to DNA, both ethidium bromide fluorescence of bound DNA in the complex and

116

Zhong, Wu, and Wu

Table 1 Formulation and Properties of DNA Complexes with PEG-PLL-VSV Complex no. 1 2 3 4 5 6 7

DNA (µg)

PEG-PLL-VSV (µg)

30 30 30 30 30 30 30

50 100 150 200 250 300 350

Particle sizea (nm) 145.5 137.7 136.8 123.6 123.0 115.1 109.3

EtBr fluorescence (% free DNA) 80.9 69.9 53.0 38.7 29.0 25.0 21.0

c Potential (mV) –10.16 –2.89 13.44 ND ND ND ND

aEffective hydrodynamic diameter and the c potential were measured at 23°C by a 90 Plus Particle Size Analyzer (Brookhaven Instrument) with all samples in 0.15 M NaCl. ND, not determined.

particle size of the bound DNA decreased, demonstrating increasingly effective DNA compaction. Compared with the complex with AsOR-PLL or PLL, the complex formed between DNA and PEG-PLL-VSV possessed superior solubility in saline. At DNA concentration as high as 50 µg/mL, the recovery rate was as high as 86%; that of AsOR-PLL alone was 65% or less (data not shown). No turbidity of the complex solution appeared after 2 wk of storage at 4°C. The effect on size and ethidium bromide exclusion of DNA complexes with the addition of the cell targeting component, AsOR-PLL, to DNA-PEG-PLL-VSV complex can be seen in Table 2. The addition of AsOR-PLL decreased the size of DNA complexes from a mean diameter of 145.5 to 112.8 nm (Table 1, complex 1 vs Table 2, complex 1) and ethidium bromide fluorescence of complexed DNA from 81 to 24%. The charge of the complex was also significantly less negative (c potential, –10.16 mV vs 0.14 mV) with the addition of AsOR-PLL to the complex. The solubility of the DNA–PEG-PLL-VSV–AsOR-PLL complex was similar to that of DNA–PEG-PLL-VSV complexes (data not shown). 3.11. Effect of Increasing VSV Content in DNA Complexes on Transfection In Vitro As seen from Fig. 1, the transfection activity as reflected by RLU increased as the ratio of PEG-PLL-VSV:DNA increased. This could have resulted from more VSV incorporated in the complex to increase endosomolysis, or from excess positive charge of the complex with binding to the negatively charged cell membranes nonspecifically, rather than through the receptor-mediated endocytosis. The transfection was not be inhibited by addition of excess AsOR (data not shown). 3.12. Transfection of Multicomponent DNA Complexes After incorporation of the targeting component AsOR-PL, the multicomponent complex DNA–PEG-PLL-VSV–AsOR-PL was formed. As shown in Fig. 2, the

Synthetic Virus for Liver-Directed Delivery

117

Table 2 Formulation and Properties of DNA Complexes with PEG-PLL-VSV and AsOR-PLL Complex no.

1 2 3 4

DNA PEG-PLL-VSV AsOR-PLL (µg) (µg) (µg)

30 30 30 30

44 88 88 0

14 14 7 45

Particle sizea (nm)

112.8 107.1 133.9 204.6

EtBr fluorescence c Potential (% free DNA) (mV)

24.0 18.9 49.6 20.8

0.14 12.08 3.12 0.43

aThe

effective hydrodynamic diameter and the c potential were measured at 23°C by a 90 Plus Particle Size Analyzer (Brookhaven Instrument) with all samples in 0.15 M NaCl.

Fig 1. Effects of increasing VSV ratios in DNA–PEG-PLL-VSV complexes on gene expression. AsOR-receptor-positive Huh7 cells were seeded into 24-well plates. After incubation for 20 h, old media were removed and 0.5 mL DMEM media added, and 100 µL (1 µg DNA) of samples to be tested were added in triplicate. After incubation for 6 h, 50 µL fetal bovine serum was added to each well and the cells were further incubated for 20 h. Luciferase activity was determined as described in Subheadings 2. and 3. Weight ratios of DNA to carrier are presented along the horizontal axis, and targeted gene expression as RLU/mg cell protein along the vertical axis.

118

Zhong, Wu, and Wu

Fig. 2. Targeted transfection by multicomponent carriers in vitro. AsOR-receptor-positive Huh7 cells were seeded into 24-well plates. After incubation as described for Fig. 1, luciferase activity was determined as described in Subheadings 2. and 3. Lane 1, DNA– PEG-PLL-VSV–AsOR-PLL (30:44:14 µg); lane 2, DNA–PEG-PLL-VSV–AsOR-PLL (30:44:14 µg) + 100-fold molar excess AsOR; lane 3, DNA–PEG-PLL-VSV–AsOR-PLL (30:88:14 µg); lane 4, DNA–PEG-PLL-VSV–AsOR-PL (30:88:14 µg + 100-fold molar excess AsOR; lane 5, DNA–PEG-PLL-VSV–AsOR-PLL (30:88:7 µg); lane 6, DNA– PEG-PLL-VSV–AsOR-PLL (30:88:7 µg)+ 100-fold molar excess AsOR; lane 7: DNA– AsOR-PLL (30:4 µg); lane 8, DNA–AsOR-PL (30:4 µg) + 100-fold molar AsOR.

transfection activity and specificity were affected by the ratios of three components. Complex 3, DNA–PEG-PLL-VSV–AsOR-PLL (30:88:7 µg) (lane 5), in which DNA was not well compacted (as seen from the EtBr exclusion in Table 2), possessed substantial activity, but low specificity (revealed by the AsOR competition of about 40%, lane 6). Complex 2, DNA–PEG-PLLVSV–AsOR-PLL (30:88:14 µg) (lane 3), with high ratios of all three components relative to DNA, also had strong activity, but low specificity, as reflected by only a 40% AsOR competition (lane 4). In contrast, complex 1, DNA–PEGPLL-VSV–AsOR-PLL (30:44:14µg) (lane 1), had the highest transfection activity, which was 100× as that of the DNA complex formed with the conventional AsOR-PLL system (lane 7). Furthermore, the expression was greatly (more than 90%) inhibited by excess of AsOR, lane 2, indicating that the gene expression was mediated by the asialoglycoprotein receptor.

Synthetic Virus for Liver-Directed Delivery

119

Fig. 3. Stability of DNA complexes in serum. Stability of complexed DNA was determined by incubation of DNA complexes with fresh rat serum or saline as control. Samples taken after 30 min incubation with serum were dissociated by heparin, and the released DNA analyzed by agarose gel electrophoresis in the presence of ethidium as described in Subheadings 2. and 3. Lanes 1–4, DNA–PEG-PLL-VSV:AsOR-PL; 5–8, free DNA; 9–12, DNA–AsOR-PL. Lanes 3, 4, 7, 8, 11, and 12 contain samples incubated in saline only. Lanes 1, 2, 5, 6, 9, and 10 contain samples incubated in serum. Lanes 2, 4, 6, 8, 10, and 12 contain samples treated with heparin.

3.13. Stability of Multicomponent DNA Complexes For a DNA delivery system to work well in vivo, the DNA complexes formed must be stable in serum. The complex DNA–PEG-PLL-VSV–AsORPLL (30:44:14 µg), which was found to have the highest transfection activity in cell culture, was further assessed for its stability in serum. As seen in Fig. 3, after 30 min incubation in saline, both the free DNA (lanes 7 and 8) and DNA in complexes (lanes 3 and 4, 11 and 12) remained intact. In contrast, after 30 min incubation in rat serum, all free DNA was degraded to small oligonucleotides (lanes 5 and 6); no significant degradation of DNA in the complex 1, DNA– PEG-PLL-VSV–AsOR-PLL (30:44:14 µg) was observed (lanes 1 and 2), which was as stable as the prototype complex of AsOR-PLL (30:45 µg), lanes 9 and 10. This life-span in serum is theoretically long enough for the DNA complex to be delivered and taken up in vivo as a targeted distribution, as shown previously (5). DNA in complexes not treated with heparin was retarded at the well (lanes 1, 3, 9, and 11) and revealed that no free DNA had dissociated. 3.14. Discussion PEG is a noncharged polymer, the attachment of which, to other molecules and surfaces, provides a biocompatible, protective coating. This has been shown previously to decrease rejection in biological systems. As a result, the DNA complexes had both increased serum stability and water solubility (9). For polylysine-DNA complexes, complement activation was also shown to be considerably decreased by modifying the surface of preformed DNA complexes with PEG (10).

120

Zhong, Wu, and Wu

Block copolymers of PLL with PEG can compact DNA into a water-soluble complex (11,12). These systems have been shown previously (13,14) to form micelle-like aggregates that contained a hydrophobic core of neutralized polyions, and a hydrophilic shell of hydrated nonionic chains in aqueous solutions. In order to make the complexes of DNA–PEG-PLL targetable to hepatocytes, the complex was linked to a targeting conjugate, AsOR-PLL. However, because of lysosomal degradation, the complex of DNA formed with PEG-PLL and AsORPLL had low transfection activity in vitro. For this reason, an endosomal disruptive agent such as VSV peptide was further incorporated by conjugating VSV to PEGPLL through a disulfide linkage. Compared with AsOR-PLL, this multicomponent system possessed several advantages, which may benefit its in vivo application. The transfection activity was increased approx 100-fold. The water solubility was increased, making it possible to deliver a large amount of DNA, and the serum stability of DNA complex formed was increased. Finally, as seen from Table 2, the particle size of the complex of DNA–PEG-PLL-VSV–AsOR was much smaller compared to AsOR-PLL or PLL complexes. This is another important property for possible in vivo applications. The sinusoids of the liver possess a fenestrated endothelium with a mean diameter of fenestrations of 100 nm (15). Therefore, the size of the particle is an important factor for efficient delivery. DNA complexes tend to form aggregates. Thus, many special approaches, such as gradient dialysis of complexes (1), formation of DNA complexes at high concentrations of salt (16), extrusion of the complexes through membranes (17), and so on, have been used in attempts to maintain small size. c Potential is another feature worth mentioning because it measures the overall charge of DNA complexes in solution. It is known that the efficiency of transfections in vitro can be enhanced by increasing the positive charge of the complexes. However, this is a potential obstacle for targeted in vivo applications, because positively charged complexes can bind nonspecifically to cell surfaces and activate the alternate complement system. Interaction with the complement system followed by uptake by cells of the reticuloendothelial system, would be expected to result in a rapid clearance of the transfection complexes (10). An ideal vector should be nontoxic, noninflammatory, nonimmunogenic, and have a large capacity for genetic material, efficient transfection, target cell specificity, and the ability to be produced in high concentration and at low cost. Through the use of multicomponent DNA carrier systems, many of these objectives can be achieved, making them attractive candidates for in vivo applications. Acknowledgments The secretarial assistance of M. Schwartz are gratefully acknowledged. This work was supported in part by a grant from the National Institutes of Health, DK-42182 (G. Y. W.), the Immune Response Corporation (C. H. W.), and the Herman Lopata Chair for Hepatitis Research.

Synthetic Virus for Liver-Directed Delivery

121

References 1. Wu, G. Y. and Wu, C. H. (1987) Receptor-mediated in vitro gene transformation by a DNA carrier system. J. Biol. Chem. 262, 4429–4432. 2. Wu, G. Y. and Wu, C. H. (1988) Receptor-mediated gene delivery and expression in vivo. J. Biol. Chem. 263, 14,621–14,624. 3. Wagner, E., Zenke, M., Cotton, M., Beug, H., and Birnstiel, M. L. (1990) Transferrin-polycation conjugates as carriers for DNA uptake into cells. Proc. Natl. Acad. Sci. USA 87, 3410–3414. 4. Ledley, F. D. (1993) Hepatic gene therapy: present and future. Hepatology 18, 1263–1273. 5. Wu, G. Y. and Wu, C. H. (1998) Receptor-mediated delivery of foreign genes to hepatocytes. Adv. Drug Delivery Rev. 29, 243–248. 6. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. 7. Carsson, J., Drevin, D., and Axen, R. (1978) Protein thiolation and reversible protein-protein conjugation. N-succinimidyl 3-(2-pyridyldithio)-propionate, a new heterobifuctional reagent. Biochem. J. 173, 723–737. 8. Moore, S. (1968) Amino acid analysis: aqueous dimethyl sulfoxide as solvent for the ninhydrin reaction. J. Biol. Chem. 243, 6281. 9. Harris, J. M. (1992) Poly (Ethylene Glycol) Chemistry: Biotechnical and Biomedical Applications. Plenum, New York. 10. Plank, C., Mechtler, K., Szoka, F., and Wagner, E. (1996) Activation of the complement system by synthetic DNA complexes: a potential barrier for intravenous gene delivery. Hum. Gene Ther. 7, 1437–1446. 11. Katayose, S. and Kataoka, K. (1997) Water-soluble polyion complexes associates of DNA and polyethylene-glycol poly(L-lysine) block copolymer. Bioconjug. Chem. 8, 702–707. 12. Choi, Y. H., Liu, F., Kim, J.-S., Choi, Y. K., Park, J. S., and Kim, S. W. (1998) Polyethylene glycol-grafted poly-L-lysine as polymeric gene carrier. J. Control. Release 54, 39–48. 13. Harada, A. and Kataoka, K. (1995) Formation of polyion complex micelles in aqueous milieu from a pair of oppositely charged block copolymers with poly(ethylene glycol) segments. Macromolecules 28, 5294–5299. 14. Kabanov, A. V., Bronich, T. K., Kabanov, V. A., Yu, K., and Eisenberg, A. (1996) Soluble stoichiometric complexes from poly(N-ethyl-4-vinylpyridinium) cations and poly(ethylene oxide)-block-poly(methacrylate) anions. Macromolecules 29, 6797–6802. 15. Nopanitaya, W., Lamb, J. C., Grisham, J. W., and Carson, J. L. (1976) Effect of hepatic venous outflow obstruction on pores and fenestration in sinusoidal endothelium. Br. J. Exp. Pathol. 57, 604–609. 16. Perales, J. C., Ferkel, T., Beegen, H., Ratnoff, O. D., and Hanson, R. W. (1994) Gene transfer in vivo-sustained expression and regulation of genes introduced into the liver by receptor targeted uptake. Proc. Natl. Acad. Sci. USA 91, 4086–4090. 17. Hara, T., Tan, Y., and Huang, L. (1997) In vivo delivery to the liver using reconstituted chylomicron remnants as a novel nonviral vector. Proc. Natl. Acad. Sci. USA 94, 14,547–14,552.

Polyampholyte Comb-Type Copolymers

123

10 Characterization of Polyampholyte Comb-Type Copolymer DNA Carriers Yoshiyuki Takei, Atsushi Maruyama, Toshihiro Akaike, and Nobuhiro Sato 1. Introduction The authors have reported several comb-type copolymers, consisting of a polycation backbone such as poly-L-lysine (PLL) and hydrophilic side chains of polysaccharides, for controling the assembling structure of DNA–copolymer complexes (1–3). The DNA–copolymer system consists of two phases: compacted DNA–polycation complex and highly hydrated glycocalyx. The latter plays multiple roles in reducing the aggregation of the resulting complexes and increasing the solubility of the complexes (2–3). Furthermore, the grafting degree of the copolymer affects the DNA conformation in the complex, permitting regulation of DNA compaction. The comb-type copolymers with a higher degree of grafting induce little compaction of DNA and stabilize DNA duplexes and triplexes by shielding the repulsion between phosphate anions of DNA. The grafting chains also reduce the nonspecific interaction of the PLL backbone with proteins. This chapter focuses on the characterization of polyampholyte comb-type copolymer having hyaluronic acid (HA) side chains as DNA carriers in the context of their physicochemical properties and complex formation. 2. Materials 1. 2. 3. 4.

Plasmid DNA. Oligonucleotides. Comb-type copolymers. Phosphate-buffered saline (PBS), 8.1 mM Na2HPO4, 2.68 mM KCl, 1.47 mM KH 2PO4, and 1 M NaCl. 5. PBS. 6. 0.5 M EDTA. From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

123

124

Takei et al.

7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19.

1 M NaCl. Spectra/Por 7 membrane (mol wt cut-off 1000). Deuterium oxide (D2O). Agarose. Tris-acetate–EDTA buffer for gel electrophoresis. Ethidium bromide. Fluorescein isothiocyanate (FITC)-labeled DNA. Ether. Phenol. Chloroform. Ethanol. Polymerase chain reaction (PCR) primers. Fluorogenic `-galactosidase substrates (e.g., 5-dodecanoylamino-fluorescein di-`D-galacto-pyranoside [C12FDG][Molecular Probes, Eugene, OR]). 20. Glutaraldehyde 1% solution.

3. Methods 3.1. Assessment of Interaction of Comb-Like Copolymers with DNA

3.1.1. Preparation of Comb-Like Copolymer–DNA Complex 1. Mix DNAs (plasmids or oligonucleotides) (1–100 µg/mL) with various amounts (see Note 1) of comb-type copolymers (1–5 mg/mL in PBS) in PBS containing 8.1 mM Na2HPO4, 2.68 mM KCl, 1.47 mM KH2PO 4, and 1 M NaCl and allowed to stand at 4°C for at least 1 h. 2. Dilute the mixtures to a final NaCl concentration of 154 mM with NaCl-free PBS and allow to stand at 4°C for more than 24 h before use (see Note 1).

3.1.2. Turbidity Measurement of Comb-Like Copolymer–DNA Mixtures Turbidity is a simple and useful marker of solubility of the copolymer–DNA complex, which is related to interpolyelectrolyte complex formation (IPEC) between the graft copolymer and DNA. 1. Following the procedure for forming a comb-type copolymer–DNA complex as described above, prepare PBS solutions containing 50 µg/mL DNA and various proportions of a comb-like copolymer. 2. Measure the turbidity of the solutions by absorbance at 340 nm with a spectrophotometer.

3.1.3. Estimation of Copolymer–DNA Interaction by 1H Nuclear Magnetic Resonance Spectroscopy 1. Mix the stock solution of DNA (1 µg/mL) in 10 mM sodium phosphate buffer, pH 7.2, containing 0.5 mM EDTA and 1 M NaCl comb-like copolymer solution (1 mg/mL) in 1 M NaCl at various P:D ratios (copolymer to DNA charge ratio: [amino group]copolymer:[phosphate group]DNA [usually 0–5.0]).

Polyampholyte Comb-Type Copolymers

125

Fig. 1. 1H-NMR spectra of the PLL-g-HA comb-type copolymer in the presence or absence of DNA. (A) PLL-g-HA alone in D2O containing 350 mM NaCl, (B) and PLL-g-HA in the absence (B) or presence (C) of DNA in D2O without NaCl. The signal of ¡-methylene protons of PLL (3.0 ppm) and the H-2 signal of glucuronic acid of HA (3.3 ppm) are represented. Reprinted with permission from ref. 3. Copyright 1998, American Chemical Society. 2. Desalt the mixture by step-down dialysis using a Spectra/Por 7 membrane (mol wt cut-off, 1000), followed by freeze-drying. 3. Dissolve the resulting sample in D2 O (Deuterium content: 99.95) (Merck, Darmstadt, Germany) containing 0.35 M NaCl. 4. Obtain 1H-NMR spectra (400 MHz) with a spectrometer (Varian Unity 400plus, Palo Alto, CA) at a probe temperature of 298 K. The chemical shifts are expressed as parts per million using internal HDO molecules (=4.7 ppm in D2O) as a reference (2).

For example, the 1H-NMR spectra of the copolymer PLL-graft-HA (PLL-gHA), which contains PLL and HA oligomers as a hydrophilic polysaccharide side chain, with or without DNA, are shown in Fig. 1 (3). The signals of ¡-methylene protons of the PLL moiety in the PLL-g-HA broadened in the presence of DNA. This is caused by IPEC formation between the PLL and DNA. Although the PLL signals broadened, the HA signals remained unchanged, or rather, became sharper.

126

Takei et al.

The sharp signals of HA in the presence of DNA under low-ionic-strength conditions (c) seem to be similar to those of PLL-g-HA alone at high ionic strength, when the PLL-g-HA does not form the IPEC; i.e., HA chains are free. These results collectively suggest that the PLL backbone selectively forms the IPEC with DNA even in the presence of the HA side chains that have negative charges. The preferential complex formation can be further demonstrated by the acrylamide or agarose gel retardation assay (see Subheading 3.1.4.).

3.1.4. Agarose Gel Retardation Assay To determine the proportions of nonviral synthetic vectors, such as comblike copolymers, that should be mixed with plasmid DNAs to optimize the DNA (plasmid or oligonucleotide) content of the complex, agarose gel retardation assay is used. 1. Add increasing amounts of copolymers to a fixed amount (e.g., 1 µg) of DNA at various P:D ratios (usually 0–5.0), all in 1 M NaCl. 2. Allow samples to stand for 12 h at 4°C and diluted to 154 mM NaCl with ddH2O. 3. Electrophoresis samples containing an equal amount of DNA (~100 ng) on a 0.8% agarose gel in Tris-acetate–EDTA buffer, followed by staining with ethidium bromide to visualize DNA.

DNA migration into the agarose gel is completely retarded when the whole amount of DNA interacts with the copolymer. For example, using a PLL-g-HA, the titration point representing the minimum proportion of PLL-g-HA required to retard completely the DNA at a physiological ionic condition (154 mM NaCl) occurs at a P:D charge ratio of approx 1.0, showing that despite the presence of HA graft chains, which bear negative charges, the PLL backbone can efficiently interact with DNA molecules. This selectivity of inter-polyelectrolyte formation indicates that when the PLL-g-HA forms complex with DNA, the negatively charged HA side chains of the PLL-g-HA copolymer remain free. This property is important because the free HA chains are crucial for solubility of the complex and HA receptor-mediated delivery to the liver endothelial cells.

3.1.5. Circular Dichroism Measurement Circular dichroism is used to examine the structural changes of DNAs when associated with comb-type copolymers. This information is useful for predicting whether the copolymer-associated DNAs preserve the ability to form duplex or triplexes. For detailed procedures, refer to Chapter 17 by Torigoe and Maruyama.

3.1.6. UV-Thermal Denaturation Profile The effect of comb-like copolymers on stabilizing duplex or triplex DNAs can be analyzed by recording thermal denaturation curves with a UV spectrophotometer equipped with a micromelting temperature apparatus (Beckman), as described in Chapter 17.

Polyampholyte Comb-Type Copolymers

127

3.1.7. Electrophoretic Mobility Shift Assay Triplex formation involving DNA sequences of interest can be analyzed by electrophoretic mobility shift assay, as described in Chapter 17.

3.2. Targetability to Organ Some glycosaminoglycans such as HA are known to be ligands for receptors that are expressed uniquely in some cell types. Indeed, through this mechanism, HA is taken up exclusively by the liver sinusoidal endothelial cells (4). This characteristic of sinusoidal endothelial cells has been utilized to develop a copolymer (PLL-g-HA) targeting foreign DNA selectively to the sinusoidal endothelial cells (3). To ensure that the PLL-g-HA–DNA complexes were delivered to the sinusoidal endothelial cells in vivo, the following experiments should be performed.

3.2.1. Organ Distribution of PLL-g-HA–DNA Complex After In Vivo Transfection 1. Label DNA with 32P. For labeling of plasmid DNA, multiprimer method is recommended. Follow the instruction of kits commercially available (e.g., Random Primer DNA Labeling Kit version 2, TaKaRa, Kyoto, Japan). End-labeling is suitable for labeling DNA oligunucleotides. 2. Prepare a PLL-g-HA–DNA complex as described in Subheading 3.1.1. DNA plasmids or DNA oligomers can be used. 3. Inject 32P-labeled DNA (10–500 µg) complexed to PLL-g-HA in saline (~500 µL) into rats via the tail vein. 4. After various time periods, sacrifice animals and harvest samples of blood, liver, spleen, gut, heart, lung, kidney, and thymus. Weigh tissue samples, followed by homogenization in saline. The homogenates can be solubilized with a tissue solubilizer (Soluene-350, Packard Instrument, Meriden, CT). Mix the solubilized homogenates with aqueous scintillant (High-Ionic Fluor, Packard) and count them on a liquid scintillation analyzer (TRI-CARB 1500, Packard).

3.2.2. Assessment of PLL-g-HA–Complexes in Organ The following is a procedure to evaluate intralobular distribution of the PLLg-HA–DNA complexes in the liver. 1. Prepare a complex composed of an FITC-labeled DNA oligonucleotide and PLLg-HA as described in Subheading 3.1.1. 2. After various time periods, harvest the liver and freeze it in liquid nitrogen. Prepare 8-µm-thick sections from the frozen sample and view the sections under an epifluorescent microscope with a filter set for FITC.

3.3. Detection of Gene Expression An expression plasmid encoding reporter genes such as lacZ can be used (e.g., pSV `-galactosidase expression vector, Promega, Madison, WI).

128

Takei et al.

3.3.1. Detection of Transgene in Cytosolic Fraction of Liver Tissue 1. To detect the transgene in the target organs such as liver after in vivo transfection, administer intravenously in rats via the tail or penile vein, 50–500 µg reporter gene plasmid, either complexed to comb-like copolymers or alone. 2. After various time periods, anesthetize the animals with ether, open the abdominal cavity, perfuse livers briefly with saline to remove blood, then homogenize the organ in a solution containing 154 mM NaCl and 10 mM EDTA. 3. Prepare the cytosolic fraction by ultracentrifugation (2 × 104g, 15 h) with an ultracentrifuge according to conventional methods (5). 4. Extract the transgene from the cytosolic fractions by phenol, phenol:chloroform (1:1, v/v), and chloroform, then ethanol-precipitate, and resuspend in a Tris-EDTA buffer. 5. Perform PCR to amplify the plasmid DNA using a standard kit (e.g., GeneAmp, Perkin Elmer, Norwalk, CT) and an appropriate primer set.

3.3.2. Detection of lacZ mRNA with Reverse TranscriptionPolymerase Chain Reaction (RT-PCR) 1. Isolate total hepatic RNA by a standard procedure (6), following in vivo transfection. Reverse transcribe it by a standard method or with a kit (e.g., You-Prime First-Strand Beads, Pharmacia Biotech, Uppsala, Sweden). 2. Amplify 1 µL of the resultant first-strand complementary DNA (cDNA) pool using an appropriate primer. For comparison with the amount of the transgene remaining in the liver tissue, the same RNA samples not converted to cDNA by reverse transcriptase should be also amplified by PCR. 3. Separate the PCR products by agarose gel electrophoresis.

3.4. Detection of `-Galactosidase mRNA with RT-PCR After Transfection In Vivo 3.4.1. Reporter Gene Expression After In Vivo Transfection lacZ is a popular reporter gene for detection of gene expression after in vivo transfection. For evaluation of localization of gene expression in organ sections, use fluorogenic `-galactosidase substrates such as 5-dodecanoylamino-fluorescein di-`-D-galactopyranoside (7). 1. After in vivo transfection with a lacZ reporter gene such as pSV `-galactosidase reporter plasmid in a PLL-g-HA-complexed form or DNA alone, sacrifice rats and harvest organs, then freeze the organs in liquid nitrogen and prepare 8-µm sections. 2. Fix the sections with 1% glutaraldehyde, wash with PBS, and incubate at 37°C for 1 h in PBS containing 33 µM C12FDG (7). View the section under an epifluorescent microscope with an appropriate filter set.

4. Notes 1. The proportions of comb-like copolymers that should be mixed with plasmid DNAs to optimize the DNA (plasmid or oligonucleotide) content of the complex should be determined before application of the copolymer-DNA system for in vivo or in vitro gene transfection (see Subheading 3.1.4.).

Polyampholyte Comb-Type Copolymers

129

References 1. Maruyama, A., Ishihara, T., Kim, J. -S., Kim, S. W., and Akaike, T. (1997) Nanoparticle DNA carrier with poly(L-lysine) grafted polysaccharide copolymer and poly(D, L-lactic acid). Bioconjugate Chem. 8, 735–742. 2. Maruyama, A., Watanabe, H., Ferdous, A., Katoh, M., Ishihara, T., and Akaike, T. (1998) Characterization of interpolyelectrolyte complexes between doublestranded DNA and polylysine comb-type copolymers having hydrophilic side chains. Bioconjugate Chem. 9, 292–299. 3. Asayama, S., Nogawa, M., Takei, Y., Akaike, T., and Maruyama, A. (1998) Synthesis of novel polyampholyte comb-type copolymers consisting of a poly(Llysine) backbone and hyaluronic acid side chains for a DNA carrier. Bioconjugate Chem. 9, 476–481. 4. Smedsrød, B., Pertoft, H., Eriksson, S., Fraser, J. R. E., and Laurent, T. C. (1984) Studies in vitro on the uptake and degradation of sodium hyaluronate in rat liver endothelial cells. Biochem. J. 223, 617–626. 5. Chowdhury, N. R., Wu, C. H., Wu, G. Y., Yarneni, P. C., Bommineni, V. R., and Chowdhury, J. R. (1993) Fate of DNA targeted to the liver by asialoglycoprotein receptor-mediated endocytosis in vivo. Prolonged persistence in cytoplasmic vesicles after partial hepatectomy. J. Biol. Chem. 268, 11,265–11,271. 6. Chomczynski, P. and Sacchi, N. (1987) Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal. Biochem. 162, 156–159. 7. Zhang, Y. Z., Naleway, J. J., Larison, K. D., Huang, Z., and Haugland, R. P. (1991) Detection of lacZ gene expression in living cells with new lipophilic, fluorogenic `-galactosidase substrates. FASEB J. 5, 3108–3113.

Physiochemical Properties of Polyplex

131

11 Methods for Studying Formation of Polycation–DNA Complexes and Properties Useful for Gene Delivery Martin L. Read, Thierry Bettinger, and David Oupicky 1. Introduction There is an urgent requirement in the field of gene therapy for gene transfer vectors that are both safe to use and able to efficiently deliver therapeutic genes to target cells in vivo. Viral vectors, such as retrovirus, adenovirus, and herpes simplex virus, are efficient in transducing a broad range of cells, but they often lead to an inflammatory response against successfully transduced tissues, along with a strong immunogenicity of the virus itself (1). A further problem is the often expensive and laborious procedure required to produce the virus in sufficient quantities. In the past decade, several nonviral gene transfer vectors based on polycations (2) and liposomes (3,4), have been developed in order to overcome such problems. These vectors are becoming increasingly popular for use in delivering DNA to target cells both in vitro and in vivo because they are generally nonimmunogenic and easier to manufacture in bulk quantities. This chapter focuses on the use of polycations in gene delivery vectors.

1.1. Polycation–DNA Complexes as Nonviral Gene Transfer Vectors A large number of polycations have so far been used to form the core of nonviral gene transfer vectors. They include polyvinylpyridine (5); various polypeptides, such as poly(L-lysine) (PLL) (6) and poly(L-ornithine) (7,8); polyamidoamine dendrimers (9,10); and the polyethylenimines (PEIs) (11). The most commonly used polycations for gene delivery are PLL (12) and pEI (13). PLL contains only primary amino groups, and is commercially available in a wide range of molecular weights from 1 to 450 kDa. In contrast, branched PEI contains primary, secondary, and tertiary amino groups at a molar ratio of From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

131

132

Read, Bettinger, and Oupicky

1:2:1, and is commercially available in a range of molecular weights from 0.6 kDa up to 800 kDa. PEI is also available in a linear form that contains only secondary amino groups and two terminal primary amino groups. Polycation–DNA complexes (or polyplexes) are formed by the self-assembly reaction between positively charged polycations and DNA. Binding of the polycation to the DNA is dependent on the electrostatic interaction between the protonated amino groups of the polycation and negatively charged phosphate groups of the DNA. Neutralization of the electrostatic charges induces collapse of the DNA molecule into a discrete particle (also known as “DNA condensation”) at a critical molar ratio of polycation to DNA. The population of discrete particles that are observed are usually between 50 and 100 nm in diameter. The amount of polycation:DNA used to prepare complexes is normally represented as the molar ratio of polycation nitrogen to DNA phosphate, i.e., the N:P ratio (see Note 1). In the case of polycations such as 20-kDa PLL, for example, condensation with plasmid or calf thymus DNA is typically observed at N:P ratios of 0.8–1.2 (14,15).

1.2. Development of Nonviral Vectors to Improve Transfection Efficiency A limitation of nonviral gene transfer systems is the inefficient transgene expression in cells, compared to viral vectors. Several barriers have been identified as restricting transgene expression, including targeting of vectors to specific cell types, release of DNA from the endosomal system, and delivery of DNA into the nucleus. In attempts to improve the level of transgene expression, specific targeting ligands, e.g., transferrin (16), or endosomal release peptides, e.g., influenza virus hemagglutinin (17), have been incorporated into the design of the vector. Although such modifications have enhanced the transfection efficiency of synthetic vectors by several orders of magnitude, further development of the vector is still required. Indeed, nonviral systems will ideally need to be produced with transfection efficiencies rivaling viral-mediated gene transfer before they are capable of mediating sufficient gene expression to enable their application for treatment of disease. The purpose of this chapter is to describe methods that are commonly used to characterize polycation-based nonviral gene transfer vectors. Proper characterization of vectors is essential in order to develop nonviral systems with improved transfection abilities. First, it is important to monitor the self-assembly reaction between the polycation and DNA, to demonstrate that complexation has occurred. Methods are described in Subheading 3.1. that can be used to study the formation of polycation–DNA complexes, and the influence of reaction conditions, such as the ionic strength, pH, and time of incubation on the selfassembly procedure. Second, the transfection properties of gene delivery vectors

Physiochemical Properties of Polyplex

133

Fig. 1. Techniques used to study the formation and stability of polycation–DNA complexes.

are dependent on their stability, size, and surface charge. Later subheadings describe methods for investigating the physiochemical properties and stability of vectors, in order to interpret their behavior in biological systems. An overview of the methods described, and the different stages of vector development at which they can be applied, are shown in Fig. 1. All of these methods can be used with simple polycation–DNA vectors, such as those based on PLL or PEI, or with more complex vectors in which targeting ligands and/or peptides have been incorporated into the design.

134

Read, Bettinger, and Oupicky

2. Materials 2.1. Fluorescence Measurements 1. 2. 3. 4. 5. 6. 7.

Calf thymus DNA (1 mg/mL in water, Sigma). PLL (2.5 mg/mL in water, Sigma). PEI (100 mM) (see Note 1) (Polysciences or Aldrich). Ethidium bromide (EtBr) (1 mg/mL, store at 4°C in dark). Fluorescamine (0.01% in acetone, Molecular Probes). Polymethacrylate cuvets (Sigma). Fluorimeter (Perkin-Elmer LS 50B).

2.2. Agarose Gel Electrophoresis 1. 2. 3. 4.

Agarose. 5X TBE buffer. Film, Kodak. Loading buffer: 0.25% (w/v): bromophenol blue, 0.25% (w/v) xylene cyanol, 5% glycerol. 5. Gel equipment.

2.3. Physicochemical Analysis 1. 2. 3. 4.

Semimicrocuvets, 1.6 mL capacity (Fisherbrand). Nanosphere size standards: 204 nm ± 6 nm, 2% w/v (Duke Scientific). c Potential transfer standard: –50 ± 5 mV (Malvern Instruments). Zetamaster system (Malvern Instruments).

2.4. Stability of Polycation–DNA Complexes 1. 2. 3. 4.

Mung bean nuclease (Promega). Poly(L-aspartic) acid (1 mg/mL, Sigma). Heparin sulfate (Sigma). Albumin (10 mg/mL, Sigma).

3. Methods 3.1. Formation of Polycation–DNA Complexes There are several factors that can influence the self-assembly process, including the size and type of the polycation, the presence of ligands or peptides conjugated to the polycation, and the reaction conditions. Full complexation is always necessary to form small, discrete complexes, and to protect the DNA against nuclease degradation in biological systems. One must therefore determine the amount of polycation required to fully condense DNA in order to produce polycation–DNA complexes useful for gene delivery. Shorter polycations (e.g., 4 kDa PLL) are usually poorer at forming discrete complexes, and higher N:P ratios are required to achieve full condensation. Ligands or peptides conjugated to the polycation generally interfere with the condensation

Physiochemical Properties of Polyplex

135

of DNA (18). If problems are encountered, then larger polycations (e.g., 800 kDa PEI) can be used in combination with the ligand or peptide to achieve full complex formation. High concentrations of salt or elevated pH can also inhibit the formation of complexes when small molecular weight polycations (e.g., 0.7 kDa PEI) are used.

3.1.1. Ethidium Bromide Exclusion Assay The most commonly used technique to analyze the formation of polycation– DNA complexes is the ethidium bromide exclusion assay (19,20). The principle behind this assay is that ethidium bromide intercalates between stacked base pairs of double-stranded DNA to give a 10–15-fold enhancement of fluorescence. Upon addition of the polycation, ethidium bromide molecules are displaced from the DNA, and there is a corresponding reduction in the fluorescence signal. A typical DNA condensation curve using branched 25-kDa PEI or linear 22-kDa PEI is shown in Fig. 2A. An important characteristic to note is the large fall in fluorescence between N:P ratios of 1 and 4, which indicates the collapse of the DNA molecule to form the discrete particle. The inhibitory concentration of 50% (IC50) value corresponds to the N:P ratio required to produce 50% inhibition of ethidium bromide fluorescence, and is used to compare the abilities of different polycations to condense DNA. In a typical assay, ethidium bromide is used at subsaturation levels, i.e., 400 ng/mL for 10 µg/mL DNA, which corresponds to 1 ethidium molecule per 30 DNA phosphate groups. Problems may arise if higher concentrations of ethidium bromide are used, since the DNA becomes saturated with ethidium molecules. This has the effect of reducing the ability of polycations to condense DNA, because the backbone of the DNA is now rigid, as the result of the presence of the dye molecules. Concentrations of ethidium bromide must be within the linear range, the limit of which is approx 2 µg/mL of ethidium bromide per 10 µg/mL DNA. Other fluorescent dyes such as propidium iodide can also be used successfully in exclusion assays to study the formation of complexes (see Note 2). Increased turbidity caused by the formation of large particulates can sometimes interfere with the fluorescence signal. In such cases, using an appropriate cut-off filter helps to reduce the turbidity signal. For example, if excitation and emission wavelengths of 510 and 590 nm are used to measure ethidium bromide fluorescence, then a 515-nm cutoff filter can be used. If problems with turbidity cannot be overcome, then an alternative method, such as the gel shift assay, should be used to study the condensation process. 1. Set up the following in a polymethacrylate fluorimeter cuvet: 10 µg/mL DNA (calf thymus or plasmid); 400 ng/mL ethidium bromide; dH2O distilled water up to a final volume of 2 mL.

136

Read, Bettinger, and Oupicky

Fig. 2. (A) Condensation of plasmid DNA by branched PEI (25 kDa) and linear PEI (22 kDa) as monitored by the ethidium bromide exclusion assay. Increasing the amount of PEI added to DNA results in a reduction of the ethidium bromide fluorescence. The IC50 values observed for linear and branched PEI are 1.8 and 2.2, respectively. At N:P ratios * 4, 90% of ethidium bromide is excluded from the DNA. (B) Agarose gel electrophoresis of PEI (25 kDa)–DNA complexes at different N:P ratios. At N:P ratios ) 1, the DNA is not fully condensed, but at N:P ratios * 2, the DNA is retained in the wells. A progressive decrease of the fluorescence in the wells is observed at higher N:P ratios, because of the inability of ethidium bromide to intercalate between base pairs of DNA. 2. Measure the fluorescence of the DNA alone (Fdna) at an excitation wavelength of 510 nm and an excitation wavelength of 590 nm. Use slit widths set at 10 nm and an integration time of 3 s. Perform all experiments in triplicate. 3. Add aliquots of the polycation to the DNA solution (in at least 0.2:1 N:P ratio increments), mix gently by inversion, and measure the fluorescence (Fx). 4. Measure the background fluorescence (Fbg) with ethidium bromide alone. 5. Calculate the % inhibition of fluorescence (%Finh), using the formula: %Finh = (Fx – Fbg)/(Fdna–Fbg) × 100 6. Plot the graph %Finh vs N:P ratio, and calculate the IC50 value for each polycation used.

3.1.2. Agarose Gel Retardation Assay The gel retardation (or gel shift) assay is another technique that is used to study the formation of polycation–DNA complexes. In this assay, the polycation and DNA are mixed at different N:P ratios, and the mobility of the DNA determined by agarose gel electrophoresis. Complexation is indicated by complete retention of the DNA in the wells of the gel. In contrast, at lower N:P ratios prior to condensation, DNA still freely migrates into the gel. The location of free DNA is detected using fluorescent dyes, such as ethidium bromide (Fig. 2B). In

Physiochemical Properties of Polyplex

137

addition to retarding the movement of the DNA, the polycation also reduces binding of the ethidium bromide dye; therefore, the presence of the DNA must be demonstrated by using polyanions, such as heparin sulfate or poly(L-aspartic) acid to disrupt complexes trapped in the wells and release the DNA (21). 1. Aliquot calf thymus or plasmid DNA (10 µg/mL) in a microcentrifuge tube containing 1 mL water. 2. Add polycation to the microcentrifuge tubes to form complexes at different N:P ratios. 3. Incubate at room temperature for 30 min to allow the complexes to form properly. 4. Aliquot 20 µL complexes to fresh microcentrifuge tubes containing 2 µL gel loading buffer. 5. Load the complexes onto a 0.8% agarose gel containing ethidium bromide at 0.5 µg/mL in 0.5X TBE buffer. 6. Electrophorese the gel in 0.5X TBE buffer at 5 V/cm for 60 min. 7. Visualize and photograph the gel on a UV transilluminator. 8. Add heparin sulfate (500 U/well), incubate for 30 min, and rephotograph the gel to show presence of DNA.

3.1.3. Amino Group Analysis: Fluorescamine Assay The self-assembly reaction between DNA and polycations can also be monitored using reagents that specifically react with amino groups of the polycation such as fluorescamine, ninhydrin, and 2,4,6-trinitrobenzenesulphonic acid (TNBS). The principle behind this approach is that when polycations bind to DNA, the amino groups become inaccessible because of the formation of salt linkages with DNA phosphate groups. This results in a reduction in the number of amino groups that are available for detection. However, higher levels of amino groups are detected when the polycation is added after complex formation because of the inability of the amino groups to bind to DNA within condensed structures (15). The fluorescamine assay is preferentially used to determine the availability of amino groups because it is at least 10-fold more sensitive than the TNBS assay, and unlike TNBS, there is no reaction between fluorescamine with DNA. The range of the assay is linear, up to 10 µg polycation, and the fluorescence signal produced by the reaction of fluorescamine with amino groups is steady for at least 20 min. An advantage of using this approach to study complex formation is that the assay is less susceptible to changes in pH and salt conditions as compared to the ethidium bromide exclusion assay. In the presence of 1 M salt, for example, there is still a fluorescence signal approx 40% of that observed in no salt conditions. The reaction between fluorescamine and polycations proceeds quickly at pH 9.0–9.5, at which it is complete within 1 min. In contrast, the reaction takes longer to reach completion at pH 7.0–8.0, and a steady signal is generated after 5–10 min. Furthermore, the actual level of fluorescence generated with polycations is approx 2–3-fold higher at pH 9.0–9.5 than at pH 7.0.

138

Read, Bettinger, and Oupicky

Fig. 3. Amino group availability of polycation–DNA complexes at different N:P ratios. 459 kDa PLL was incubated with calf thymus DNA for 10 min prior to analysis of free amino groups using the fluorescamine assay. At low N:P ratios < 0.8, prior to condensation, there are few amino groups detected. In contrast, at N:P ratios * 0.8, after DNA condensation, the fluorescence signal is much higher because of unbound polycation.

A typical DNA condensation curve using fluorescamine to detect the availability of amino group in 459 kDa PLL, is shown in Fig. 3. The most important characteristic to note is that at low N:P ratios, there are few amino groups detected prior to DNA condensation because the polycation is bound tightly to the DNA. However, after complex formation, in this case at N:P ratios > 0.8, the fluorescence signal is much higher because of the presence of unbound polycation. Determination of amino group availability also provides an indication of the binding efficiency between the DNA and polycation (see Note 3). 1. Prepare polycation–DNA complexes at a range of N:P ratios using a DNA concentration of 10 µg/mL. 2. Incubate at room temperature for 30 min to allow complexes to form properly. 3. Dilute 100-µL aliquots of each sample in 1.4 mL of assay buffer (100 mM boric acid-NaOH, pH 7–9.5). 4. Add 500 µL 0.01% fluorescamine (prepared in acetone). 5. Mix samples rapidly by inversion 4–5× and incubate at room temperature for 2–10 min. 6. Measure Fx using a Perkin-Elmer LS 50B fluorimeter using hex 392 nm, hem 480 nm with 5–10 nm slit widths. 7. Use polycation alone to determine 100% value (Fpc) and background fluorescence using assay buffer (Fbg). 8. Calculate the % free amino groups (%Fag) using the formula: %Fag = (Fx–Fbg)/(Fpc–Fbg) × 100

Physiochemical Properties of Polyplex

139

3.1.4. Physicochemical Analysis of Polycation–DNA Complexes There are several techniques that are commonly used to study the physicochemical properties of polycation–DNA complexes, including photon correlation spectroscopy (PCS), electrophoretic mobility, transmission electron microscopy (TEM), and atomic force microscopy (AFM). These techniques can be used to monitor the formation of complexes, and to provide direct proof of DNA compaction. PCS permits determination of the mean diameter of complexes and the particle size distribution; electrophoretic mobility gives information on the surface charge (or c potential). Microscopy-based techniques such as TEM or AFM are used to assess the morphology and size of complexes. The application of these techniques to study polycation–DNA complexes is described in the following subheadings. 3.1.4.1. ANALYSIS OF PARTICLE SIZE: PCS PCS (or dynamic light scattering) is a straightforward way of analyzing the size of polycation–DNA complexes. In simple terms, this method measures the Brownian motion of particles, and relates this to their size. This is achieved by measuring the intensity fluctuations of scattered light arising from the illumination of particles with a vertically polarized laser light. Small particles cause the intensity to fluctuate more rapidly than larger ones. The fluctuations in the intensity of scattered light are then converted into electrical pulses, and a correlation function is generated that can be used to calculate the mean diameter and size distribution of particles. Various algorithms, of increasing resolution, such as monomodel (cumulants), multimodal, Contin, and nonnegative least-squares analysis, are used to obtain the mean particle diameter and size distribution from the correlation function. The higher the resolution of the algorithm that is used, the better the quality of data needs to be as well. At least three measurements of a sample should be performed to determine the reproducibility of the data. If the results are not repeatable, then the data must be reanalyzed using a lower-resolution algorithm. The processing time and concentration of sample required to give good quality data are dependent on the power of the laser and the sensitivity of detector used. If a suitably powerful laser is used, such as a 50–100 mW external laser, then a sample can be processed within a few minutes. The limit of detection using this type of laser is usually at a concentration between 5 and 10 µg/mL polycation–DNA complexes. One limitation of PCS is that the method is only suitable for measuring the size of spherical particles. The presence of any extended structures on particles will be largely ignored with this technique, leading to inaccurate results (18); therefore, the morphology of complexes must be determined using TEM or AFM, as well as characterizing their size and distribution using PCS. The following protocol is for determining the size of complexes using a Zetasizer system with a 70-mW external laser.

140

Read, Bettinger, and Oupicky

1. Turn on Zetasizer system and external laser. Adjust power of external laser to 70 mW. 2. Set parameters for software. Select a wavelength of 488 nm, number of measurements to three, temperature at 25°C, and duration of measurement to rapid. If complexes are in 150 mM NaCl set viscosity centipoise (cP) to 1.145 cP and refractive index (RI) medium to 1.3402; if in water, set viscosity to 1 cP and RI medium to 1.335; and if in 5% glucose, set viscosity to 1.014 cP and RI medium to 1.34. RI particle is set to 1.45 and the temperature to 25°C, whatever the solution. 3. Check system using standard latex beads of diameter 204 ± 6 nm. The standard needs be diluted approx 1000-fold in water to be at a suitable concentration for measurement. 4. Aliquot 500 µL of complexes at 20 µg/mL into 1.6-mL semimicrocuvets and measure the mean particle size, polydispersity, % merit, and % in range. 5. If the signal is <5 kilo counts per second (Kcps), then increase power of laser if possible, and/or set duration of measurement to automatic. Alternatively, the concentration of complexes can be increased to 40 µg/mL to give a stronger signal. Check quality of data, using values obtained for polydispersity, % merit, and % in range. 6. Check repeatability of data. Use lower-resolution algorithm, e.g., monomodal, if necessary.

3.1.4.2. ANALYSIS OF SURFACE CHARGE OF COMPLEXES

A widely used method to determine the surface charge of complexes is to measure the c potential. This provides a useful indicator to study the formation of complexes and their interaction with serum proteins such as albumin (see Subheading 3.2.3.). In modern equipment the c potential is not measured directly; instead, laser Doppler velocimetry is used to determine the electrophoretic mobility of complexes. Laser Doppler velocimetry measures the velocity of particles through a fluid in an electrophoresis experiment. The electrophoretic mobility is then converted to c potential using theoretical considerations. Using this technique, it is possible to monitor the condensation process by determining the c potential at different N:P ratios of polycation and DNA. At a low N:P ratio, the c potential is negative because of the large abundance of phosphate groups on the DNA. However, following neutralization of the electrostatic charges by polycations, there is an increase in the c potential with rising N:P ratio. Complexation of the DNA with simple polycations such as PLL or PEI, results in the formation of positively charged particles. For example, 20-kDa PLL/DNA complexes, formed in water at a DNA concentration of 20 µg/mL and at a N:P ratio of 2, have a c potential of approx 35 mV. The size of the polycation used does not significantly alter the c potential of the polycation–DNA complexes (22). Two factors that can significantly affect the c potential of complexes are the pH and salt conditions used. A c potential measurement should always be quoted with the pH value and the salt conditions used. The main technical problem in determining the c potential of complexes is that excess polycation can adhere to the walls of the capillary cell and interfere with the

Physiochemical Properties of Polyplex

141

consistency and reproducibly of the data obtained. Therefore, it is advisable to regularly wash the cell thoroughly with water and check the correct operation of the instrument with a c potential standard. If problems are still encountered, then the cell can be flushed through with 1 M NaCl to remove any bound polycation. The following protocol is for determining the c potential of particles using a Zetamaster system containing a 5-mW internal laser. 1. Turn on the instrument and select parameters for software. Set number of measurements to five, c range to unknown, and duration of measurement to automatic. If complexes are in 150 mM NaCl, set cP to 1.145, RI medium to 1.3402, and dielectric constant (¡) to 79; if in water set cP to 1, RI medium to 1.335, and ¡ to 78; and, if in 5% glucose, set cP to 1.014, RI medium to 1.34, and ¡ to 80.37. RI particle is set to 1.45 and the temperature to 25°C, whatever the solution. 2. Check system using c potential transfer standards of –50 ± 5 mV. This standard is comprised of carboxylated polystyrene latex dispersed in a pH 9.22 buffer. 3. Inject 2.5 mL complexes at 20 µg/mL and measure c potential. If signal is low or data is not reproducible, then thoroughly wash the cell and use complexes prepared at a higher concentration, such as 40 µg/mL.

3.1.4.3. ANALYSIS OF MORPHOLOGY AND SIZE OF PARTICLES

TEM provides a powerful method for analyzing the morphology and size of polycation–DNA complexes. TEM is capable of magnifying approx 200,000×, and has a resolution limit of about 1 nm. The excellent resolving power of TEM is mostly a function of the short wavelength of electrons accelerated by an applied electric field. An example of 25-kDa PEI–DNA complexes as visualized by TEM, is shown in Fig. 4. Complexes are directly mounted by adsorption onto the surface of carbon support films, stained with uranyl acetate, and visualized with the electron microscope. The staining procedure with uranyl acetate must be kept shorter than 1 min; otherwise, the uranyl salt can help to condense the DNA (23). Complexes must be visualized in several electron microscope grids and not just within one or two. This enables the complete range of complexes of differing shapes and sizes, which are normally present within a given population to be properly assessed. 1. Prepare carbon film on freshly cleaved mica by the evaporation of carbon rods under vacuum. 2. Cover electron microscope copper–rhodium grids with the carbon film using a flotation technique. 3. Dry grids overnight, and keep them on blotting paper placed in a Petri dish. 4. Just before adding the samples, glow discharge the grids (110 mV, 25–30 µA, 25 s). 5. Add a 5-µL drop of complex solution onto the grid. After 1 min, stain the complexes with 30 µL of an aqueous uranyl–acetate solution (1%, w/w) for 20 s. 6. Remove the excess liquid with blotting paper. 7. Visualize the samples in the electron microscope at 80 kV.

142

Read, Bettinger, and Oupicky

Fig. 4. TEMs of PEI 25 kDa–DNA complexes for N:P = 20 (A), and PEI (25 kDa)polyethylene glycol3400 1.7%–DNA complexes for N:P = 10 (B), in 150 mM NaCl. (A) shows the formation of discrete DNA particles of approx 50–100 nm in diameter, which correlates well to measurements obtained by PCS. (B) shows partially compacted micrometric structures made of 5–10-nm-thick polymorphic strings. The size measurement by PCS for these structures gives a value around 100 nm, which highlights the reasons why PCS values must be interpreted with caution. The bar represents 100 nm.

3.2. Determining Properties of Complexes Useful for Gene Delivery After monitoring the self-assembly reaction between the polycation and DNA, as described in Subheading 3.1., the next step is to investigate whether the polycation–DNA complexes have properties that are useful for gene delivery. In particular, it is necessary to determine the stability of complexes in physiological conditions and how they interact with individual components of biological systems such as proteins or charged molecules. The study of such interactions helps to interpret the behavior of the complexes in biological systems. By using such an approach, it should be possible to identify traits in the design of complexes that can be modified in order to change their interaction with biological components, and hence improve their overall transfection ability. The following subheadings describe methods that are used to investigate the stability of polycation–DNA complexes and their interaction with proteins and charged molecules.

3.2.1. Degradation by Nucleases One of the prerequisites for an efficient gene delivery system is that the DNA is still intact upon reaching the target cell type within the body, and has not been degraded by nucleases. For instance, serum nucleases in the bloodstream are known to rapidly degrade nucleic acids, and so free plasmid DNA

Physiochemical Properties of Polyplex

143

cannot be used for systemic applications of gene therapy (24). Complexation with polycations such as PLL has been shown to protect DNA against degradation by serum nucleases and purified nucleases such as mung bean nuclease (14,25). Therefore, a vital step in the development of any new delivery system based on polycations is to demonstrate stability of the DNA against nuclease attack. The most common method used is to incubate complexes in the presence of either serum nucleases or purified nucleases and determine the integrity of the DNA by agarose gel electrophoresis. 1. 2. 3. 4.

5. 6.

7. 8. 9. 10. 11.

Make polycation–DNA complexes according to standard protocols. Add mung bean nuclease (2.5 U/µg DNA). Incubate at 37°C for at least 1 h. Denature nuclease by heating the sample to 70°C for 15 min. Dissociate complexes by adding either an equal volume of trypsin (for PLL–DNA complexes) or 1.5 vol 1% sodium dodecyl sulfate and NaOH. Incubate at 37°C for 30 min. Extract DNA with an equal volume of equilibrated phenol–chloroform. Precipitate the DNA with 2 vol of 100% ethanol and 0.1 vol 3 M sodium acetate, pH 5.2. Incubate at 4°C for at least 30 min and centrifuge at 16,000 g at 4°C for 30 min to pellet the DNA. Wash the DNA pellet with 500 µL 70% ethanol and allow to air-dry for at least 10 min. Resuspend DNA in 20 µL TE buffer or water. Load the complexes onto a 1.5% agarose gel containing ethidium bromide at 0.5 µg/mL in 0.5X TBE buffer. Electrophorese the gel in 0.5X TBE buffer at 5 V/cm for 30 min. Compare nuclease-treated DNA with untreated DNA.

3.2.2. Destabilization of Complexes by Polyanions There is a wide range of molecules that are capable of disrupting polycationbased complexes, leading to the release of free DNA. The glycosaminoglycans, for example, are a group of negatively charged molecules, such as heparin sulfate, which exist in the extracellular matrix and as components of cell surfaces. In a recent study (26), the ability of these molecules to disrupt the transfection activity of complexes was demonstrated. Hence, one of the challenges in developing new polycation-based delivery vectors is to ensure that they are more resistant to disruption by charged molecules. Methods used to compare the stability of complexes are based on the disruption of complexes with polyanions, such as poly(L-aspartic) acid or heparin sulfate (14). The degree to which DNA is released from complexes can be monitored either by the restoration of ethidium bromide fluorescence using a fluorimetry-based approach or by agarose gel electrophoresis. Misleading results can be obtained if there is excess polycation present because the polyanion will bind preferentially to the free polycation. This occurs if

144

Read, Bettinger, and Oupicky

complexes are formed at N:P ratios that are much greater than what is required for condensation. In such cases, complexes appear more stable to polyanionic disruption than they actually are, and it is necessary to use appropriate controls. The following protocol is a fluorimetry-based method for studying the stability of complexes to polyanionic disruption. 1. Set up the following in a total volume of 2 mL in a polymethacrylate fluorimeter cuvet: 10 µg/mL polycation–DNA complexes in water; 400 ng/mL ethidium bromide. 2. Measure the Fx of the complexes alone at an excitation wavelength of 510 nm and an excitation wavelength of 590 nm. Use slit widths set at 10 nm and an integration time of 3 s. Perform all experiments in triplicate. 3. Add aliquots of poly(L-aspartic) acid to the complexes, mix gently by inversion and measure the Fx. 4. Measure the Fbg with ethidium bromide alone and the Fdna at 10 µg/mL with ethidium bromide. 5. Calculate the %F using the formula: %F = (Fx–Fbg)–(Fdna–Fbg) × 100 6. Plot the graph %F vs N:P ratio.

3.2.3. Stability of Complexes in Physiological Conditions For successful application, DNA delivery systems must be less than 100 nm in diameter to permit extravazation through vascular fenestration and endocytosis into cells. Although the mean diameter of polycation–DNA complexes are usually below this size threshold, they tend to aggregate in physiological conditions, which limits their usefulness in vivo. For example, 20-kDa PLL–DNA complexes, at a N:P ratio of 2, aggregate rapidly in 150 mM salt to diameters greater than 500 nm in 15 min. Larger particles are sometimes better at transfecting cells in vitro, where high levels of gene expression can be achieved (27). Several polycation-based delivery systems have been described that demonstrate increased stability in physiological conditions, including systems based on high-molecular-weight PLL (22), branched PEI at N:P ratios greater than 5 (18) and the use of hydrophilic polymers covalently attached to the surface of complexes, resulting in steric stabilization (see Chapter 5 by Oupicky et al.). Complexes formed with linear PEI of molecular weight similar to branched PEI are not stable in physiological salt conditions. A number of techniques including PCS (see Subheading 3.1.4.1.), electrophoretic mobility (see Subheading 3.1.4.2.), and turbidometric analysis are available for determining the suitability of polycation-based delivery systems for in vivo applications. Stability in physiological salt conditions can be determined by incubating complexes in the presence of 150 mM NaCl and monitoring for changes in particle size over time using PCS. Any complexes that do not significantly increase in size within the first 30 min are good candidates for

Physiochemical Properties of Polyplex

145

in vivo applications. In such experiments, the parameters for the viscosity and refractive index have to be adjusted within the software to take into account the different media used. Serum proteins can also cause aggregation of polycation– DNA complexes. However, the high background signal from serum means that the highest concentration of serum that can be used to investigate the stability of particles is approx 5%. A more common approach is to monitor the changes in particle size in the presence of individual serum proteins such as albumin. PCS has been used successfully to assess the stability of complexes to crosslinking by albumin at concentrations up to 1 mg/mL (28). It is possible to monitor the interaction of complexes with serum proteins by determining the changes in c potential. Complexes based on PLL or PEI are highly positively charged, and bind proteins such as albumin by electrostatic interactions, which results in shielding of the positive charges on the surface of complexes and a reduction in the observed c potential (29). The increased turbidity, resulting from the aggregation of complexes by salt or proteins, can also be measured using turbidometric analysis (30). The details and protocol for this method are described in Chapter 5. Polycation–DNA complexes with useful in vivo properties for systemic gene delivery should demonstrate no change in c potential or increase in turbidity in the presence of serum proteins such as albumin. 4. Notes 1. The amount of polycation to DNA used for each complex is represented as the mol ratio of polycation nitrogen to DNA phosphate (i.e., the N:P ratio). 1 µg DNA is equivalent to 3.08 nmol phosphate groups; 1 µg PLL–HBr is equivalent to 4.78 nmol nitrogen. To form pLL–DNA complexes at a N:P ratio of 2, for example, 25.6 µg PLL–HBr should be added to 20 µg DNA in water. In the case of PEI, 4.3 mg PEI should be dissolved in 1 mL water to give a 100 mM solution of PEI with respect to the nitrogen content. Complexes between PEI and DNA should be formed in a suitable buffer, such as 20 mM HEPES-NaOH, pH 7.4. To form PEI–DNA complexes, the plasmid (20 µg) and the desired amount of polymer solution (for N:P = 10, 6 µL 100 mM amino group PEI solution) are each diluted in 500 µL 20 mM HEPES-NaOH, pH 7.4, and vortexed (13). After 10 min, the two solutions are mixed and the resulting solution is vortexed. Incubate at room temperature for 15 min, to allow complexes to form properly. The final concentration of the polyplex is 20 µg/mL DNA. 2. Ethidium bromide has been widely used to study complex formation between double-stranded DNA and polycations (20). Other fluorescent dyes such as propidium iodide can also be used. A 3–4-fold higher level of sensitivity is achieved using propidium iodide, compared to ethidium bromide in studying the formation of oligonucleotide-based complexes (31). The lower level of sensitivity observed with ethidium bromide results from the high background fluorescence of the dye alone and because ethidium molecules only bind weakly to oligonucle-

146

Read, Bettinger, and Oupicky

otides. In contrast, the better fluorescence properties of propidium iodide most likely result from the higher positive charge associated with the propidium molecule, enabling it to bind more avidly to oligonucleotide DNA. 3. The determination of amino group availability can provide an indication of the binding efficiency between the DNA and polycation. For example, in the case of higher-molecular weight polycations (e.g., 200-kDa pLL), there are usually fewer amino groups detected by the fluorescamine reagent at low N:P ratios because the polycation binds efficiently to the DNA. However, there is an increase in the number of amino groups detected at low N:P ratios, if the pH is raised to 9.0–9.5, because polycations become unprotonated and can no longer bind electrostatically to the DNA. In contrast, the formation of condensed structures stabilizes the interaction between the polycation and DNA at higher N:P ratios, and increasing the pH to 9.0–9.5 does not significantly affect the amino group availability.

References 1. Quantin, B., Perricaudet, L. D., Tajbakhsh, S., and Mandel, J. L. (1992) Adenovirus as an expression vector in muscle cells in vivo. Proc. Natl. Acad. Sci. USA 89, 2581–2584. 2. Kabanov, A. V., Szoka, F. C. J., and Seymour, L. W. (1998) Interpolyelectrolyte complexes for gene delivery: polymer aspects of transfection activity, in Selfassembly Complexes for Gene Delivery: From Laboratory to Clinical Trial. Wiley, Chichester, pp. 197–218. 3. Zabner, J. (1997) Cationic lipids used in gene transfer. Adv. Drug Delivery Rev. 27, 17–28. 4. Smith, J., Zhang, Y. L., and Niven, R. (1997) Toward development of a non-viral gene therapeutic. Adv. Drug Delivery Rev. 26, 135–150. 5. Kabanov, A. V., Astafyeva, I. V., Chikindas, M. L., Rosenblat, G. F., Kiselev, V. I., Severin, E. S., and Kabanov, V. A. (1991) DNA interpolyelectrolyte complexes as a tool for efficient cell transformation. Biopolymers 31, 1437–1443. 6. Wagner, E., Zenke, M., Cotten, M., Beug, H., and Birnstiel, M. L. (1990) Transferrin-polycation conjugates as carriers for DNA uptake into cells. Proc. Natl. Acad. Sci. USA 87, 3410–3414. 7. Bond, V. C. and Wold, B. (1987) Poly-ornithine-mediated transformation of mammalian cells. Mol. Cell. Biol. 7, 2286–2293. 8. Pouton, C. W., Lucas, P., Thomas, B. J., Uduehi, A. N., Milroy, D. A., and Moss, S. H. (1998) Polycation-DNA complexes for gene delivery: a comparison of the biopharmaceutical properties of cationic polypeptides and cationic lipids. J. Controlled Release 53, 289–299. 9. Haensler, J. and Szoka, F. C. (1993) Polyamidoamine cascade polymers mediate efficient transfection of cells in culture. Bioconjugate Chem. 4, 372–379. 10. Kukowska-Latallo, J. F., Bielinska, A. U., Johnson, J., Spindler, R., Tomalia, D. A., and Baker, J. R. (1996) Efficient transfer of genetic material into mammalian cells using Starburst polyamidoamine dendrimers. Proc. Natl. Acad. Sci. USA 93, 4897–4902.

Physiochemical Properties of Polyplex

147

11. Boussif, O., Lezoualch, F., Zanta, M. A., Mergny, M. D., Scherman, D., Demeneix, B., and Behr, J. P. (1995) A versatile vector for gene and oligonucleotide transfer into cells in culture and in vivo: polyethylenimine. Proc. Natl. Acad. Sci. USA 92, 7297–7301. 12. Zauner, W., Ogris, M., and Wagner, E. (1998) Polylysine-based transfection systems utilizing receptor-mediated delivery. Adv. Drug Delivery Rev. 30, 97–113. 13. Remy, J. S., Abdallah, B., Zanta, M. A., Boussif, O., Behr, J. P., and Demeneix, B. (1998) Gene transfer with lipospermines and polyethylenimines. Adv. Drug Delivery Rev. 30, 85–95. 14. Dash, P. R., Toncheva, V., Schacht, E., and Seymour, L. W. (1997) Synthetic polymers for vectorial delivery of DNA: characterisation of polymer-DNA complexes by photon correlation spectroscopy and stability to nuclease degradation and disruption by polyanions in vitro. J. Controlled Release 48, 269–276. 15. Read, M. L., Etrych, T., Ulbrich, K., and Seymour, L. W. (1999) Characterisation of the binding interaction between poly( L-lysine) and DNA using the fluorescamine assay in the preparation of non-viral gene delivery vectors. FEBS Lett. 461, 96–100. 16. Wagner, E., Curiel, D., and Cotten, M. (1994) Delivery of drugs, proteins and genes into cells using transferrin as a ligand for receptor-mediated endocytosis. Adv. Drug Delivery Rev. 14, 113–135. 17. Plank, C., Oberhauser, B., Mechtler, K., Koch, C., and Wagner, E. (1994) The influence of endosome-disruptive peptides on gene transfer using synthetic viruslike gene transfer systems. J. Biol. Chem. 269, 12,918–12,924. 18. Erbacher, E., Bettinger, T., Belguise-Valladier, P., Zou, S., Coll, J.-L., Behr, J.-P., and Remy, J.-S. (1999) Transfection and physical properties of various saccharide, poly(ethylene glycol), and antiboby-derivatized polyethylenimine. J. Gene Med. 1, 210–222. 19. Cain, B. F., Baguley, B. C., and Denny, W. A. (1978) Potenial antitumor agents. 28. Deoxyribonucleic acid polyintercalating agents. J. Med. Chem. 21, 658–668. 20. Wolfert, M. A. and Seymour, L. W. (1996) Atomic force microscopic analysis of the influence of the molecular weight of poly(L)lysine on the size of polyelectrolyte complexes formed with DNA. Gene Ther. 3, 269–273. 21. Wolfert, M. A., Schacht, E. H., Toncheva, V., Ulbrich, K., Nazarova, O., and Seymour, L. W. (1996) Characterization of vectors for gene therapy formed by self-assembly of DNA with synthetic block co-polymers. Hum. Gene Ther. 7, 2123–2133. 22. Xu, B., Wiehle, S., Roth, J. A., and Cristiano, R. J. (1998) The contribution of poly- L-lysine, epidermal growth factor and streptavidin to EGF/PLL/DNA polyplex formation. Gene Ther. 5, 1235–1243. 23. Böttcher, C., Endisch, C., Fuhrhop, J.-H., Catterall, C., and Eaton, M. (1998) High-yield preparation of oligomeric C-type DNA toroids and their characterization by cryoelectron microscopy. J. Am. Chem. Soc. 120, 12–17. 24. Kawabata, K., Takakura, Y., and Hashida, M. (1995) The fate of plasmid DNA after intravenous injection in mice: involvement of scavenger receptors in its hepatic uptake. Pharm. Res. 12, 825–830.

148

Read, Bettinger, and Oupicky

25. Chiou, H. C., Tangco, M. V., Levine, S. M., Robertson, D., Kormis, K., Wu, C. H., and Wu, G. Y. (1994) Enhanced resistance to nuclease degradation of nucleic acids complexed to asialoglycoprotein-polylysine carriers. Nucleic Acids Res. 22, 5439–5446. 26. Ruponen, M., YlaHerttuala, S., and Urtti, A. (1999) Interactions of polymeric and liposomal gene delivery systems with extracellular glycosaminoglycans: physicochemical and transfection studies. Biochim. Biophys. Acta 1415, 331–341. 27. Ogris, M., Steinlein, P., Kursa, M., Mechtler, K., Kircheis, R., and Wagner, E. (1998) The size of DNA/transferrin-PEI complexes is an important factor for gene expression in cultured cells. Gene Ther. 5, 1425–1433. 28. Dash, P. R., Read, M. L., Fisher, K. D., Howard, K. A., Wolfert, M., Oupicky, D., et al. (2000) Decreased binding to proteins and cells of polymeric gene delivery vectors surface modified with a multivalent hydrophilic polymer and retargeting through attachment of transferrin. J. Biol. Chem. 275, 3793–3802. 29. Dash, P. R., Read, M. L., Barrett, L. B., Wolfert, M. A., and Seymour, L. W. (1999) Factors affecting blood clearance and in vivo distribution of polyelectrolyte complexes for gene delivery. Gene Ther. 6, 643–650. 30. Ward, C. M., Fisher, K. D., and Seymour, L. W. (1999) Turbidometric analysis of polyelectrolyte complexes formed between poly(L-lysine) and DNA. Colloid Surface B 16, 253–260. 31. Read, M. L., Dash, P. R., Clark, A., Howard, K. A., Oupicky, D., Tonchera, V., et al. (2000) Physiochemical and biological characterisation of an antisense oligonucleotide targeted against the bcl-2 mRNA complexed with cationic-hydrophilic copolymers. Eur. J. Pharm. Sci. 10, 169–177.

AFM of DNA Condensates

149

12 Characterization of DNA Condensates by Atomic Force Microscopy Ye Fang and Jan H. Hoh 1. Introduction In recent years, interest in in vitro DNA condensation has been revived by efforts to develop vectors for nonviral gene therapy (1,2). One of the critical elements for successful and versatile delivery of specific genes into targeted cells is that DNA vectors of several kilobase pairs must be compressed and packaged into small particles. Ideally, the size of these particles should be similar to that of viruses, typically less than 100 nm in diameter (3,4). It is well known that several classes of multivalent cations can be used to condense DNA in vitro to form well-defined structures, particularly toroids and rods (5,6). Because of its simplicity, this type of in vitro condensation of DNA has become a widely studied system for investigating mechanisms of DNA condensation. Techniques for characterizing the structure and properties of DNA condensates include gel electrophoresis, circular dichroism and fluorescence spectroscopy, light scattering, and electron and fluorescence microscopy. Atomic force microscopy (AFM) (7), as an emerging technique has been applied to directly visualize various sequence- and/or environment-dependent structures of DNA molecules in three-dimension with subnanometer height resolution (8–10). The structural characteristics of DNA molecules confined to solid surfaces have been used to shed light on the structure of DNA in solution, such as kinked (11), supercoiled (12), cruciform (13), bent (14), and looped (15) DNA. In contrast to electron microscopy, the AFM imaging can be performed on a wide range of surfaces and in different environments (8,9). The AFM has been used to visualize the structure of DNA molecules with surfaces such as mica (12), silicon (16), lipid bilayers (17,18), self-assembled From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

149

150

Fang and Hoh

monolayers (16,19), amorphous carbon (20), minerals (21), and sapphire (22), in ambient, low-pressure, or near-physiological conditions. Further, AFM imaging of DNA can be performed without staining or other contrast enhancement of the sample. Electron microscopy of unstained DNA has been reported, but this is difficult to achieve (23). There are a variety of agents that can induce DNA condensation by neutralizing the negative charge of DNA chain. These agents include Co(NH 3)6 3+, polyamines, cationic lipids, cationic polymers, highly positive-charged polypeptide, and proteins. Neutral polymers and certain solvents, such as ethanol, can also condense DNA by excluding the solvent from the DNA molecules (5). AFM has been used to examine the structure of DNA condensates induced by a number of these agents, including spermidine (24,25), protamine (26), polylysine (27,28), cationic lipids (29), cationic silanes (30,31), and ethanol (32). AFM has also been used to examine two specific examples, cationic silane-modified surface (16) and supported cationic lipid membranes (17,18), in which the surfaces play a dominant role in the DNA condensation. Here is demonstrated how AFM is used to characterize DNA condensates under ambient (dry) conditions or in aqueous solutions using condensates formed by cationic silanes (Fig. 1) and polyamines (Fig. 2) as examples. 2. Materials

2.1. DNA and Chemicals 1. 2. 3. 4. 5. 6. 7. 8.

pBR322 DNA (Sigma, St. Louis, MO). N-(2-aminohexyl)-3-aminopropyl-trimethoxysilane (AHA) (Gelest, Tullytown, PA). Deionized water (>18 M1; MilliQ-UV, Millipore, Bedford, MA). NaCl. Glacial acetic acid. Spermine (Sigma). Compressed gas (Vari-Air, Peca, Janesville, WI). Ruby muscovite mica (grade V2 or better) (Asheville-Schoomaker Mica, Newport News, VA).

2.2. Atomic Force Microscopy 1. Silicon cantilevers for tapping mode imaging (model TESP, Digital Instruments, Santa Barbara, CA). 2. Silicon nitride cantilevers (Microlevers, ThermoMicroscopes, Sunnyvale, CA). 3. Nanoscope IIIa controller with a Multimode AFM and a J-type scanner (maximum x, y range ~150 × 150 µm) (Digital Instruments). 4. Ambient tapping mode cantilever holder (Digital Instruments). 5. Fluid-tapping cell (Digital Instruments). 6. Ultraviolet (UV) light cleaner (UVO-Cleaner, Jelight, Laguna Hills, CA).

AFM of DNA Condensates

151

Fig. 1. Ambient tapping mode AFM images of condensates formed by 2 µg/mL pBR322 in the presence of (A) 180 µM AHA at neutral pH, and (B) 90 µM AHA at acidic pH (~4.5). At neutral pH, condensation is incomplete and complex condensation intermediates are seen. The lower pH promotes silane crosslinking and results in fully formed toroidal and rod-shaped condensates. Adapted with permission from ref. 30.

Fig. 2. Fluid tapping mode AFM image of the condensates formed by 2 µg/mL pBR322 in the presence of 250 µM spermine at neutral pH. These condensates appear essentially identical to those seen under ambient conditions (24).

152

Fang and Hoh

3. Methods 3.1. General Concerns DNA condensation is a delicate process (5). Factors that effect the DNA condensation include: 1. DNA concentration. For intramolecular condenstion, low DNA concentrations (0.1–10 µg/mL) should be used. Concentrations below 0.1 µg/mL typically result in too few molecules on the surface for imaging; DNA concentrations above 10 µg/mL result in intermolecular interactions that lead to aggregations rather than to condensation (5). 2. Cation concentration. A certain critical concentration of a cation is needed to complete DNA condensation, depending on cation species, DNA-binding constant, and environmental conditions (5,33,34). High cation concentration can induce aggregation or the formation of giant condensates (35). In addition, silanes can form islands and other structures on the surface (Fig. 3A). 3. Time. Intramolecular DNA condensation is a rapid process that generally takes less than a few minutes to complete (36). Long storage time of the DNA–cationic complexes can result in aggregation caused by the interaction between condensates (24). 4. Buffer. The presence of other ions effects DNA condensation, e.g., higher concentration of multivalent cations is needed to complete DNA condensation in the presence of NaCl (34), and the presence of some divalent cations may cause DNA aggregation rather than condensation. 5. Rinsing sample. Salt crystals that form upon drying can obscure DNA on the surface. The amount of salt can be reduced by rinsing with water. However, decondensation during a water rinse can be rapid, and hence water rinses should be avoided if possible (6,24). An example of this is shown in Fig. 3B. The best approach the authors have found to dealing with this is to use rapid blow-drying with compressed air, after adsorption to mica, to remove excess liquid and salt. 6. Other factors may alter the outcome of DNA condensation, including DNA size, structure, sequence and topology, and cation structure and hydrophobicity (5). The balance among these factors determines the structure and size of final DNA condensates formed.

3.2. Preparation of DNA–Silane Complex The preparation of DNA–cation complexes is straightforward. A solution with an appropriate concentration of DNA is mixed with an equal volume of a solution containing the cation. Using equal volumes of the solutions results in the smallest local concentration gradients for any of the components, which produces a more reproducible outcome of the condensation. Note that pH is a particularly critical parameter (see Note 1). 3.2.1. Preparation of DNA–Silane Complex at Neutral pH 1. Measure the DNA concentration by spectrophotometry at 260 nm. 2. Dilute the DNA samples to 4 µg/mL with water.

AFM of DNA Condensates

153

Fig. 3. Ambient tapping mode AFM images of two DNA samples that exhibit problems. (A) High concentrations of cationic silane interfere with characterizing DNA condensates by forming small islands on the mica surface. The islands can grow and eventually obscure the DNA. Conditions: 2 µg/mL pBR322, 750 µM AHA, neutral pH, 2 min adsorption on mica. (B) Rinsing with water can cause rapid decondensation. Condensates form using 2 µg/mL pBR322 in the presence of 180 µM AHA at neutral pH (see Fig. 1A) were rinsed with ~5 mL water over ~10 s and blown dry. The resulting DNA is almost entirely decondensed.

3. Prepare a 400-µM solution of AHA (see Note 2). 4. Add 10 µL silane solution to 10 µL DNA solution in microcentrifuge tube and gently mix. This can be done by slowly pipeting up and down a couple times or tapping the tube with a finger after addition of the silane. 5. Incubate the mixture at room temperature for 10–15 min. 6. Proceed immediately to DNA adsorption onto mica (see Subheading 3.3.).

3.2.2. Preparation of DNA–Silane Complex at Acidic pH Crosslinking of silane monomers can be promoted by reducing the pH to 4.5 using acetic acid. Such crosslinking stabilizes the final condensate (Fig. 1B; Fang and Hoh, unpublished observation). 1. 2. 3. 4.

Follow steps 1–5 of Subheading 3.2.1. Add 2 µL 1 mM acetic acid to the mixture. Incubate additional 1 min at room temperature. Proceed immediately to DNA adsorption onto mica (see Subheading 3.3.).

3.3. Adsorption of DNA Complexes onto Mica The adsorption of DNA complexes onto mica is the most problematic aspect of examining DNA complexes by AFM, because of concerns that the surface modifies the final structures. Confidence in the results are improved by

154

Fang and Hoh

reproducing the same result on different surfaces, and correlating the results with other methods. However, there is no single way of definitely detecting or excluding surface effects. This is simply a limitation of the approach, and there is no single best way to deal with it. However, in the authors’ experience, short incubation times at lower concentrations of cations give results in which one can have most confidence. 1. Cleave a mica substrate (8–15 mm disk) with a piece of tape. 2. Immediately deposit the DNA–silane mixture onto the mica surface and incubate 1–2 min at room temperature. 3. Blow-dry the sample with compressed gas for several seconds. 4. Proceed to AFM imaging. If the sample needs to be stored, it should be kept in a desiccator. High humidity will in some cases allow rearrangement of the DNA on the surface.

3.4. Tapping Mode AFM Imaging in Air 1. Clean the cantilever by exposure to high intensity UV light. There are varying opinions on the length of cleaning: Some investigators prefer short times (3–5 min) and others prefer longer times (15–20 min). 2. Mount the cantilever and set up the AFM for imaging. 3. Mount the DNA sample in the microscope. 4. Tune the resonance frequency of cantilever (typically ~200–300 kHz) and adjust the free amplitude of the cantilever to about 10–50 nm. 5. Engage and select an imaging size of about 3 × 3 µm. 6. Optimize the imaging parameters. The best optimization depends on the instrument used, and details should be obtained from the manufacturer. The authors typically adjust the feedback gains to be as high as possible without ringing in the image, adjust the set point to about 80–90% of the free drive amplitude, and finally increase scan speed as much as possible, while maintaining an optimal image quality. Note that long shadows in the amplitude image indicate that the set point is too high, the scan rate is too high or the feedback gains are too low. 7. Collect and analyze images. Examples are shown in Fig. 1.

3.5. Tapping Mode AFM Imaging in Solution To avoid problems that might arise upon drying of the sample and to allow for visualization of molecular dynamics, AFM imaging can be performed in solution. This is technically more complicated than ambient imaging, from the point of operating the instrument and preparing the sample. It may be particularly difficult to adequately immobilize the sample on a surface, in such a way that it does not move too much during imaging. On the other hand, a certain amount of movement might be desirable. In general, DNA condensed with cations binds well to mica, but also has been shown to have some mobility (31). Varying concentrations of various ions can also be used to modulate the surface binding (37).

AFM of DNA Condensates

155

For tapping mode imaging in solution, silicon nitride cantilevers with spring constants on the order of 0.1 N/m are typically used. The authors use the 85-µmlong F-cantilevers (Microlevers) from ThermoMicroscopes, which have a resonant frequency in solution of about 30 kHz and nominal spring constant of 0.5 N/m. Again, there is a great deal of variability between instruments. To demonstrate imaging in solution, the authors use spermine instead of the cationic silanes. The presence of high concentrations of cationic silanes in solution for extended periods of time results in the formation of a monolayer of silanes on the mica surface, and eventually obscures the DNA. 1. Follow steps 1–5 of Subheading 3.2.1., but, replace the silane solution with a 500 µM spermine. 2. Deposit the 20-µL mixture onto mica, and incubate 2 min at room temperature. 3. Add 250 µL 250 µM spermine. 4. Mount the sample in the microscope. The authors typically use an “open cell,” in which the liquid is trapped between the sample and the fluid cell, for imaging. 5. Tune the resonant frequencies of the F cantilever (~30 kHz) and adjust the free amplitude to about 2–4 nm. The tuning can be difficult because the frequency sweep typically reveals dozens of peaks. The authors’ approach derives from the work of Michael Allen (personal communication). In this approach, one of the many peaks within the natural resonant frequency is used. The natural frequency can typically be identified in the phase signal at small amplitudes. The final tune of the cantilever must be performed with the tip within 10 µm of the surface. 6. Engage and select imaging size of 3 × 3 µm. 7. Optimize the imaging parameters. This is similar to ambient tapping, although more attention needs to be paid to the tapping amplitude in liquids, which should be minimized. 8. Collect and analyze images: An example is shown in Fig. 2.

4. Notes 1. The silane monomers with trimethoxy or triethoxy group are stable on the timescale of hours (or more) in neutral aqueous solutions. However, at low or high pH, they hydrolyze rapidly (minutes) to form reactive species that polymerize into silicon gels (38,39). Thus, the silane solution should be prepared fresh and used immediately. Furthermore, the properties of those silanes that allow for crosslinking between silanes can be used to form stable nanoparticles of the DNA vector. For example, treatment of a dilute solution of pBR322 plasmid DNA with the cationic silane, AHA (at 180 µM), at neutral pH results in the formation of a variety of complex structures that appear to be stable intermediates of the condensation of DNA into toroids and rods (30). However, the formation of complete rods and toroids can be significantly enhanced by lowering the pH during silane treatment to ~4.5, and at that pH complete toroids and rods form readily at 90 µM AHA (Fig. 1). These toroids and rods are stable against the dilution by adding water or high concentration of NaCl.

156

Fang and Hoh

2. As noted above, silanes are sensitive to water. As such, care should be taken to exclude moisture (air) from the silane container during storage. Containers should be flushed with argon or dry nitrogen prior to resealing after use.

References 1. Behr, J. P. (1994) Gene-transfer with synthetic cationic amphiphiles: prospects for gene-therapy. Bioconjug. Chem. 5, 382–389. 2. Kay, M. A., Liu, D., and Hoogerbrugge, P. M. (1997) Gene therapy. Proc. Natl. Acad. Sci. USA 94, 12,744–12,746. 3. Blessing, T., Remy, J.-S., and Behr, J.-P. (1998) Monomolecular collapse of plasmid DNA into stable virus-like particles. Proc. Natl. Acad. Sci. USA 95, 1427–1431. 4. Perales, J. C., Ferkol, T., Beegen, H., Ratnoff, O. D., and Hanson, R. W. (1994) Gene-transfer in vivo: sustained expression and regulation of gene introduced into the liver by receptor-targeted uptake. Proc. Natl. Acad. Sci. USA 91, 4086–4090. 5. Bloomfield, V. A. (1996) DNA condensation. Curr. Opin. Struct. Biol. 6, 334–341. 6. Marquet, R. and Houssier, C. (1991) Thermodynamics of cation-induced DNA condensation. J. Biomol. Struct. Dynamics 9, 159–167. 7. Binnig, G., Quate, C. F., and Gerber, C. H. (1986) Atomic force microscope. Phys. Rev. Lett. 56, 930–933. 8. Hansma, H. G. and Hoh, J. H. (1994) Biomolecular imaging with the AFM. Annu. Rev. Biophys. Biomol. Struct. 23, 115–139. 9. Bustamante, C. and Rivetti, C. (1996) Visualizing protein-nucleic acid interactions on a large scale with the scanning force microscope. Annu. Rev. Biophys. Biomol. Struct. 25, 395–429. 10. Hansma, H. G., Vesenka, J., Siegerist, C., Kelderman, G., Morrett, H., Sinsheimer, R. L., et al. (1992) Reproducible imaging and dissection of plasmid DNA under liquid with the atomic force microscope. Science 256, 1180–1184. 11. Han, W. H., Dlakic, M., Zhu, Y. W. J., Lindsay, S. M., and Harrington, R. E. (1997) Strained DNA is kinked by low concentrations of Zn2+. Proc. Natl. Acad. Sci. USA 94, 10,565–10,570. 12. Rivetti, C., Guthold, M., and Bustamante, C. (1996) Scanning force microscopy of DNA deposited onto mica: equilibration versus kinetic trapping studied by statistical polymer chain analysis. J. Mol. Biol. 264, 919–932. 13. Oussatcheva, E. A., Shlyakhtenko, L. S., Glass, R., Sinden, R. R., Lyubchenko, Y. L., and Potaman, V. N. (1999) Structure of branched DNA molecules: gel retardation and atomic force microscopy studies. J. Mol. Biol. 292, 75–86. 14. Erie, D. A., Yang, G. L., Schultz, H. C., and Bustamante, C. (1994) DNA bending by Cro protein in specific and nonspecific complexes-implications for protein site recognition and specificity. Science 266, 1562–1566. 15. Rippe, K., Guthold, M., von Hippel, P. H., and Bustamante, C. (1997) Transcriptional activation via DNA-looping: visualization of intermediates in the activation pathway of E. coli RNA polymerase sigma(54) holoenzyme by scanning force microscopy. J. Mol. Biol. 270, 125–138. 16. Fang, Y. and Hoh, J. H. (1998) Surface-directed DNA condensation in the absence of soluble multi-valent cations. Nucleic Acids. Res. 26, 588–593.

AFM of DNA Condensates

157

17. Fang, Y. and Yang, J. (1997) Two-dimensional condensation of DNA molecules on cationic lipid membranes. J. Phys. Chem. B 101, 441–449. 18. Mou, J., Czajkowsky, D. M., Zhang, Y., and Shao, Z. (1995) High-resolution atomic-force microscopy of DNA: the pitch of the double helix. FEBS Lett. 371, 279–282. 19. Lyubchenko, Y. L., Gall, A. A., Shlyakhtenko, L., Harrington, R. E., Jacobs, B. L., Oden, P. I., and Lindsay, S. (1992) Atomic force microscopy imaging of doublestranded DNA and RNA. J. Biomol. Struct. Dynamics. 10, 589–606. 20. Yang, J., Takeyasu, K., and Shao, Z. F. (1992) Atomic force microscopy of DNA molecules. FEBS Lett. 301, 173–176. 21. Bezanilla, M., Manne, S., Laney, D. E., Lyubchenko, Y. L., and Hansma, H. G. (1995) Adsorption of DNA to mica, silylated mica and minerals: characterization by atomic force microscope. Langmuir 11, 655–659. 22. Yoshida, K., Yoshimoto, M., Sasaki, K., Ohnishi, T., Ushiki, T., Hitomi, J., Yamamoto, S., and Sigeno, M. (1998) Fabrication of a new substrate for atomic force microscopic observation of DNA molecules from an ultrasmooth sapphire plate. Biophys. J. 74, 1654–1657. 23. Furrer, P., Bednar, J., Stasiak, A. Z., Katritch, V., Michoud, D., Stasiak, A., and Dubochet, J. (1997) Opposite effect of counterions on the persistence length of nicked and non-nicked DNA. J. Mol. Biol. 266, 711–721. 24. Fang, Y. and Hoh, J. H. (1998) Early intermediates in sperimidine-induced DNA condensation on surfaces of mica. J. Am. Chem. Soc. 120, 8903–8909. 25. Lin, Z., Wang, C., Feng, X., Liu, J., and Bai, C. (1998) The observation of the local ordering characteristics of spermidine-condensed DNA: atomic force microscopy and polarizing microscopy studies. Nucleic Acids Res. 26, 3228–3234. 26. Allen, M. J., Bradburyl, E. M., and Balhorn, R. (1997) AFM analysis of DNAprotamine complexes bound to mica. Nucleic Acids Res. 25, 2221–2226. 27. Hansma, H. G., Golan, R., Hsieh, W., Lollo, C. P., Mullen-Ley, P., and Kwoh, D. (1998) DNA condensation for gene therapy as monitored by atomic force microscopy. Nucleic Acids Res. 26, 2481–2487. 28. Golan, R., Pietrasanta, L. I., Hsieh, W., and Hansma, H. G. (1999) DNA toroids: stages in condensation. Biochemistry 38, 14,069–14,076. 29. Dunlap, D. D., Maggi, A., Soria, M. R., and Monaco, L. (1997) Nanoscopic structure of DNA condensed for gene delivery. Nucleic Acids Res. 25, 2995–3101. 30. Fang, Y. and Hoh, J. H. (1999) Cationic silanes stabilize intermediates in DNA condensation. FEBS Lett. 459, 173–176. 31. Ono, M. Y. and Spain, E. M. (1999) Dynamics of DNA condensates at the solidliquid interface by atomic force microscopy. J. Am. Chem. Soc. 121, 7330–7334. 32. Fang, Y., Spisz, T. S., and Hoh, J. H. (1999) Ethanol-induced DNA structural transition on mica. Nucleic Acids Res. 27, 1943–1949. 33. Manning, G. S. (1978) Molecular theory of polyelectrolyte solutions with applications to electrostatic properties of polynucleotides. Q. Rev. Biophys. 11, 179–246. 34. Wilson, R. W. and Bloomfield, V. A. (1979) Counterion-induced condensation of deoxyribonucleic acid. A light-scattering study. Biochemistry 18, 2192–2196.

158

Fang and Hoh

35. Yoshikawa, Y., Yoshikawa, K., and Kanbe, T. (1999) Formation of a giant toroid from long duplex DNA. Langmuir 15, 4085–4088. 36. Porschke, D. (1984) Dynamics of DNA condensation. Biochemistry 23, 4821–4828. 37. Hansma, H. G. and Laney, D. E. (1996) DNA binding to mica correlates with cationic radius: assay by atomic force microscopy. Biophys. J. 70, 1933–1939. 38. Arkles, B., Steinmetz, J. R., Zazyczny, J., Mehta, P. (1991) Factors contributing to the stability of alkoxysilanes in aqueous solution, in Silicon Compounds: Register and Review (Anderson, R., Larson, G. L., and Smith, C., eds.), Huls America, Piscataway, NJ, pp. 65–73. 39. Plueddemann, E. P. (1991) Silane Coupling Reagents, 2nd ed., Plenum, New York.

BAC/PAC Transfer by HAEC Retrofitting

159

13 Rapid and Systematic Transfer and Recovery of Large BACs/PACs into Mammalian Cells by HAEC Retrofitting* Rona S. Scott and Jean-Michel H. Vos 1. Introduction Studies of the human genome have prompted the development of several cloning systems that can manipulate large fragments of human DNA as functional units. The discoveries in yeast of the sequences for replication (autonomously replicating sequence [ARS]), centromeres, and telomeres have allowed the creation of large linear molecules (yeast artificial chromosome [YAC]), which can replicate and segregate as chromosomes (1). YACs have been used to generate genomic libraries from different organisms with megabase insert size and have become an essential tool for physical and genetic mapping of various mammalian genomes (2). Nevertheless, the instabilities associated with YAC libraries (chimerism and deletions) have led to the development of alternative cloning systems (3). One alternative cloning system based on bacteria has reduced the problems of chimerism and instability of cloned inserts. The P1 derived artificial chromosome (PAC) system is based on the P1 plasmid replicon, uses kanamycin as a selectable marker and loxP sites to circularize or insert fragment DNAs of interest (4,5). Alternatively, the bacteria artificial chromosome (BAC) cloning system is based on the F factor replication origin from Escherichia coli (6). The F factor origin not only can maintain large plasmid DNA, but also restricts the copy number to 1–2 copies/cell, limiting recombination events between molecules. In addition, the BAC cloning system also takes advantage of the loxP recombinatorial site. Both systems have shown successful manipulation of human genomic DNA greater than 100 kb in bacterial cells (6,7). * In memory of Dr. Jean-Michel H. Vos. From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

159

160

Scott and Vos

Even though these systems have been helpful in various gene-mapping efforts, functional characterization of the genomic clones in mammalian cells has been difficult. Previously, the authors and others (8,9) have reported the development of human artificial episomal chromosomes (HAEC), which can stably maintain genomic inserts of up to 650 kb in human cells. These systems are dependent on the Epstein-Barr virus (EBV) latent origin of DNA replication, oriP, and the EBV nuclear antigen 1 (EBNA-1) to propagate and maintain these circular molecules as minichromosomes (10). The HAEC-based system has also been successful in the functional analysis of human cDNAs (11,12). However, screening and isolating of genomic clones from a HAEC library for functional analysis has proved difficult. Thus, methods are described for the development of a HAEC-based system that converts BAC and PAC clones for stable and episomal propagation in human cells.

1.2. Engineering HAEC Shuttle Vector The backbone vector used to create the HAEC shuttle vector is pBS246 (LifeTechnologies, LTI), which contains a multiple cloning site flanked by two loxP sites. To this vector, the EBV oriP, hygromycin (Hyg) resistance gene, and the tetracycline (Tet) resistance gene are added to create a HAEC shuttle vector dependent on the trans expression of EBNA-1 in cell lines. Alternatively, the EBNA-1 gene can be inserted into the HAEC shuttle, incorporating all elements needed for episomal maintenance. Since the authors are interested in identifying replicator sequences in the human genome, the following cloning strategy described for the HAEC shuttle will not rely on EBNA-1. A schematic diagram of the final HAEC shuttle vector is shown in Fig. 1. 2. Materials 1. pBS246 (LTI), pBR322 (LTI), pCMVEBNA (Invitrogen), p108L (6), and pH200 (8) vectors. 2. Various restriction enzymes and restriction buffers. 3. Nuclease-free purified water. 4. Calf-intestinal phosphatase (CIP, Promega). 5. Phenol/chloroform. 6. Ethanol. 7. Phosphorylated linker containing XhoI site (dCCCTCGAGGG, New England Biolabs). 8. T4 DNA ligase and ligase buffer (LTI). 9. JM109 LB agar plates containing 12.4 µg/mL chloramphenicol. 10. Various antibiotics: ampicillin, tetracycline, chloramphenicol, and kanamycin. 11. Luria broth (LB) media. 12. LB agar plates containing 50 µg/mL ampicillin. 13. LB agar plates containing 50 µg/mL ampicillin and 10 mg/mL tetracycline. 14. LB agar plates containing 5 µg/mL tetracycline and 12.5 µg/mL chloramphenicol (BAC) or 25 µg/mL kanamycin (PAC).

BAC/PAC Transfer by HAEC Retrofitting

161

Fig. 1. Schematic diagram of the HAEC shuttle vector. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37.

LB agar plates containing 12.4 µg/mL chloramphenicol. Low melting point seaplaque agarose (FMC). Tris-acetate-EDTA and Tris-borate EDTA buffers for electrophoresis. h HindIII, low range PFG, and h PFG ladders (New England Biolabs). 100 mM Deoxynucleotides (Pharmacia). Klenow and T4 DNA polymerases and buffers (Promega). Cre recombinase (Dupont). Tris-HCl buffer, pH 7.0 and 7.5. Sodium chloride. Spermidine. Bovine serum albumin. Qiagen Maxi kit plasmid purification columns. Cesium chloride. BAC or PAC clones of interest. D98-Raji (DR) or other EBV positive cell line. Eagle’s minimal essential media supplemented with fetal bovine serum, glutamine, streptomycin, and penicillin. Lipofectamine (LTI). Hygromycin B (Boeringer). EDTA. Sodium dodecylsulfate (SDS). Sodium hydroxide. Proteinase K (LTI). Nylon membrane (Micron Separations, Inc.).

162 38. 39. 40. 41. 42. 43.

Scott and Vos Random priming kit (Promega). COT-1 DNA (LTI). Hybridization buffer (Amersham). Glass slides dipped in 2% 3-aminopropyl triethoxysilane dissolved in acetone. Prime-It-Fluor kit (Stratagene). Sonicated herring testes DNA (Oncor).

3. Methods 3.1. Insertion of XhoI Site into pBS246 In order to clone the oriP and Hyg fragment, a XhoI site is incorporated between the loxP sites of the pBS246 vector by linker tailing. 1. pBS246 is digested with EcoRV, which leaves blunt ends. 2. To a 20-µL reaction, 1 µg DNA, 2 µL 10X compatible restriction buffer, and 5 U EcoRV enzyme are added. 3. The reaction mix is incubated for 1 h at 37°C. Directly to the restriction digest, 10 µL 10X CIP buffer, 1 µL CIP, and 69 µL nuclease-free H2O are added. 4. The dephosphorylation reaction is incubated for 30 min at 37°C, followed by heat inactivation of the enzyme for 15 min at 65°C (see Note 1). 5. Subsequently, the DNA is isolated by phenol/chloroform extraction and ethanol precipitation. 6. A phosphorylated linker containing a XhoI site (dCCCTCGAGGG) is ligated into the EcoRV site of pBS246. To 200 ng dephosphorylated vector, a 100-fold molar excess of linker, 2 µL 5X T4 ligase buffer (LTI), and 1 µL T4 ligase are added in a reaction volume of 20 µL. 7. The ligation mix is incubated for 15 min at room temperature and transformed into competent JM109 cells (Promega). For the transformation, 5 µL ligation mix is added to 50 µL JM109 competent cells. Cells are incubated on ice for 30 min, followed by heat shock for 45 s at 42°C. 8. The transformation mix is plated on LB agar plates containing 50 µg/mL ampicillin and allowed to grow overnight at 37°C. 9. The plasmid is isolated by alkaline lysis extraction. 10. Positive clones are identified by double digestion with XhoI and PstI. This clone is named pBS246X.

3.2. Insertion of oriP, Hyg, and Tet Fragments The oriP and Hyg sequences are isolated from pH200 (8) as a SalI fragment. After digestion, the DNA is gel-purified and isolated by `-agarase (New England Biolabs [NEB]) digestion, as suggested by the manufacturer. The purified SalI fragment is ligated to pBS246X, previously linearized with XhoI and dephosphorylated with CIP. The ligation reaction is set up as described in Subheading 3.1. A threefold molar excess of the SalI fragment is used in relation to the vector. Colonies are screened by restriction enzyme digestion, and orientation of the fragment is determined. This vector will be termed pHS1.

BAC/PAC Transfer by HAEC Retrofitting

163

The second bacterial resistance marker (Tet) is inserted by blunt ligation into the KpnI site of pHS1, which lies between the two loxP sites of pHS1. Insertion of the Tet gene at this location will allow selection for the BAC–PAC–HAEC recombinants (see below). The Tet gene is isolated as an EcoRI/AvaI fragment from pBR322 and the overhangs are blunt-ended by Klenow fill-in. 1. To the restriction digest, 33 µM each deoxynucleotide and 1 µL Klenow enzyme are added. 2. The reaction mix is incubated for 30 min at 30°C. 3. The enzyme is heat-inactivated by incubation for 15 min at 65°C. 4. The Tet fragment is isolated by phenol/chloroform extraction and ethanol precipitation.

KpnI digestion of pHS1 generates 3' overhangs, which are blunt-ended by employing the 3' to 5' exonuclease activity of the T4 DNA polymerase. 1. The restriction digest is phenol/chloroform extracted, and ethanol precipitated. 2. The precipitated DNA is resupended in 1X T4 DNA polymerase buffer containing 100 µM of each deoxynucleotide. 3. 5 U T4 DNA polymerase are added, and the reaction is incubated for 15 min at 37°C. 4. The DNA is subsequently dephosphorylated with CIP, and purified by phenol/ chloroform extraction, and ethanol precipitation. 5. The blunt-ended Tet fragment and pHS1 are ligated using a 3:1 molar ratio of insert to vector and transformed into JM109. 6. Transformants are plated onto LB agar plates containing ampicillin (50 µg/mL) and Tetracycline (10 µg/mL). 7. Clones are analyzed by BamHI restriction enzyme digestion, and the orientation of the Tet fragment is determined. This vector is named “pHS2” (HAEC shuttle), and a schematic diagram of this vector is depicted in Fig. 1.

3.3. Strategy for Conversion of BACs or PACs into HAECs Elements from the HAEC system that allow stable, episomal propagation in human cells are added to genomic clones isolated from a BAC or PAC library by Cre-loxP recombination. As outlined in Fig. 2, the strategy depends on two in vitro recombinations. The first recombination (“pop-out”) eliminates the bacterial origin of replication and ampicillin gene present on the pHS2 vector. It is necessary to delete the high-copy ColE1 bacterial origin of DNA replication because it will interfere with the low-copy BAC and PAC replication origins. The second recombination (“pop-in”) inserts into the BAC or PAC clone the necessary elements (oriP and Hyg) needed to generate episomal minichromosomes. The authors have successfully converted various BAC and PAC clones into HAECs. These include a ribosomal DNA (rDNA) clone containing 90 kb rDNA sequence, an anonymous BAC clone containing 140 kb human DNA, the BAC vector (p108L), the P1 vector, and an 180-kb PAC clone carrying a hypoxanthine-guanine phosphoribosyl transferase (HPRT) insert (GM00847).

164

Scott and Vos

Fig. 2. Schematic illustration of BAC/HAEC or PAC/HAEC shuttling strategy (popout/pop-in strategy). The mammalian sequences necessary for selection, replication, and maintenance in human cells are inserted into the BAC or PAC clone by Cre-loxP recombination. The BAC/HAEC or PAC/HAEC clone is electroporated in E. coli cells and selected with chloramphenicol and tetracycline or kanamycin and tetracycline. The DNA is purified by CsCl density centrifugation, and transfected into EBNA-1 (or EBV) positive human cells. Transfected cells are selected with hygromycin for 12–14 d. The BAC/HAEC or PAC/HAEC clones are pooled and episomal maintenance of the clones is determined.

BAC/PAC Transfer by HAEC Retrofitting

165

Fig. 3. Cre-loxP intramolecular recombination of the HAEC shuttle. Before the HAEC shuttle can be used to convert a BAC to a BAC/HAEC (or a PAC to a PAC/HAEC), the bacterial sequences are removed (pop-out) by Cre-loxP recombination. The DNA is digested with SsPI and the products analyzed by agarose gel electrophoresis (1% agarose in 1X TBE, 100 V). The supercoiled molecule contains the desired sequences for the HAEC shuttle. The linearized products represent the pop-out fragment (bacterial sequences) and the unrecombined full-size substrate molecule.

3.3.1. Pop-Out Cre-loxP Recombination 1. 1 µg pHS2 is incubated with 0.5 U Cre recombinase (Dupont) in 50 mM TrisHCl, pH 7.5, 33 mM NaCl, 5 mM spermidine, and 500 µg/mL bovine serum albumin for 15 min at 37°C (13). 2. The enzyme is heat inactivated at 70˚C for 10 min. 3. The reaction products are subsequently digested with SspI (100 mM Tris-HCl, pH 7.5, 10 mM MgCl2, 50 mM NaCl) for 1 h at 37°C to linearize the unwanted recombination products (Fig. 3). 4. The recombination products are phenol/chloroform extracted and ethanol precipitated.

3.3.2. Isolation of BAC and PAC Clones 1. BAC clones are grown in 12.5 µg/mL chloramphenicol LB media; PAC clones are grown in 25 µg/mL kanamycin LB media. 2. BAC and PAC clones are isolated using Qiagen plasmid purification columns (see Note 2).

3.4. Pop-In Cre-loxP Recombination A second recombination reaction mediated by Cre-loxP is performed to insert HAEC shuttle sequences into the BAC or PAC clone. 1. 250 ng Shuttle DNA is recombined with 200 ng BAC DNA in the reaction conditions described in Subheading 3.3.1.

166

Scott and Vos

2. The reaction mix is heat inactivated for 10 min at 70°C and dialyzed in ddH2O for 30 min. 3. The entire amount of recombination product is transformed into competent DH10B cells by electroporation (2.5 kV, 25 µF, and 200 1, Bio-Rad electroporator). 4. The transformants are plated onto agar plates containing 5 µg/mL tetracycline and 12.5 µg/mL chloramphenicol for BAC or 5 µg/mL tetracycline and 25 µg/mL kanamycin for PAC. Positive clones containing the HAEC shuttle vector will be designated with a “BH” (BAC–HAEC) or “PH” (PAC–HAEC). 5. Restriction digests are performed to check the recombined BH–PH clones by either field inversion electrophoresis (FIGE) or pulse-field gel electrophoresis (PFGE) using a Bio-Rad CHEF Mapper. The PFGE settings are set at an initial switch time of 0.22 s; final switch time of 8.55 s; linear ramp, 120 degree angle, and 15-h run time. The FIGE settings are set with a reverse voltage of 6 V/cm2, forward voltage of 9 V/cm2, switch times of 0.2 s initial and 8.5 s final, and run time of 14 h. As shown in Fig. 4, the restriction digestion pattern of the BH–PH clones after Cre-loxP recombination shows successful insertion of the HAEC shuttle vector with no change in the size of the large human genome insert.

3.5. Episomal Maintenance of BAC/HAECs in Human Cells The stability and extrachromosomal maintenance of the BAC/HAEC constructs is tested using a rDNA BAC as a model system. The rDNA BAC was chosen because it contains repetitive elements that may be prone to recombination.

3.6. Cell Culture and Transfection The 90-kb rDNA BH is stably transfected into D98-Raji hybrid cell line (DR) cells and selected with hygromycin. DR cell line (14) is grown in Eagle's minimum essential medium (EMEM) containing 10% fetal bovine serum, 2 mM glutamine, 100 µg/mL streptomycin, and 100 U/mL penicillin. Since the function of oriP is dependent on the EBNA-1 protein, the resident EBV in the Raji component of the DR cell hybrid provides the EBNA-1 protein needed for episomal maintenance. Cells are transfected by liposome-mediated transfer of the DNA. 1 × 105 cells are plated into a 6-well plate and allowed to attach overnight. 1. 2 µg DNA (purified by cesium chloride gradient centrifugation) and 10 µg lipofectamine reagent (LTI) are mixed and incubated with the cells for 8 h. 2. The cells are grown for 2 d in complete media prior to selection with 200 µg/mL hygromycin B (Boehringer). 3. Approximately 2 wk after selection, colonies are pooled and analyzed for episomal maintenance of the DNA.

3.7. Episomal Analysis 1. Episomal DNA is extracted as previously described (8,15). Approximately 1 × 107 cells are lysed with 2 mL lysis solution (50 mM NaCl, 8 mM EDTA, 1% sodium dodecyl sulfate, pH 12.5), vortexed for 2 min, and incubated at 30°C for 30 min.

BAC/PAC Transfer by HAEC Retrofitting

167

Fig. 4. Conversion of BACs to BAC/HAECs by Cre-loxP intermolecular recombination. The size of the various recombined BH and PH clones are analyzed by PFGE and compared to the unrecombined DNA. The BAC DNA recombinants are digested with EcoRV, which does not cleave within the human ribosomal DNA (2 DNA). Lane 1, p108L (BAC vector), lane 2, HAEC shuttle, lane 3, 90 kb rDNA BAC, lanes 4 and 5, recombined p108L–HAEC vector, lanes 6 and 7, recombined 90 kb rDNA BAC/HAEC (BH), lanes 8 and 10, P1 vector, lane 9, recombined P1–HAEC vector, lane 11, 180 kb HPRT PAC, lane 12, recombined 180 kb HPRT PAC–HAEC (PH). All recombined products show an increase in the size of the vector backbone sequence, without any visible change in the size of the large human insert DNA. 2. The lysate is neutralized by addition of 400 µL of 1 M Tris pH 7.0, and incubated with proteinase K (24 µL 10 mg/mL proteinase K and 264 µL 5 M NaCl). 3. The lysate is extracted 2–3× with equilibrated phenol and one time with chloroform. 4. The DNA is then precipitated with 2 vol of ethanol. 5. The supercoiled episomal DNA is separated on a 0.8% agarose gel in 1X TBE at 4°C (see Note 3). The electrophoresed DNA is transferred to nylon membrane by Southern blotting and hybridized with various probes. 6. The rDNA probe is labeled by random priming (kit from Promega) from the 90 kb rDNA BAC. The EBNA-1 probe is obtained from a HindIII digest of pCMVEBNA (InVitrogen). When the BAC/HAEC containing human inserts are used as probes, repetitive elements are suppressed from hybridization with 5 µg Cot-1 DNA (LTI). 7. The probe and Cot-1 DNA are boiled and preincubated in hybridization buffer for 15 min. This mixture is added to the hybridization. Figure 5A shows the episomal integrity of the supercoiled 90-kb rDNA–BH and vector(p108L)–BH clones after transfection.

168

Scott and Vos

BAC/PAC Transfer by HAEC Retrofitting

169

Recovery of these supercoiled molecules shows that the shuttled vectors are maintained extrachromosomally. By comparison of the migration at the expected patterns of EBV (170 kb) and mitochondrial DNA (17 kb), the size of the rDNA–BH is estimated at 100 kb (8). The size and structure of the episomes isolated from human cells can be further analyzed by restriction digestion and PFGE. As shown in Fig. 5B, the episomes are digested with EcoRI or EcoRV, and the banding patterns of the transfected DNA and the original DNA are compared. The structures and sizes of the 12.6-kb BH vector and of the 90-kb rDNA–BH clone appear to remain the same as those of the original DNA stock before passage in human cells.

3.8. Fluorescence In Situ Hybridization (FISH) To determine the copy number of episomes per cell, FISH analysis was performed on early passage cells, as previously described (12). For the rDNA BAC/HAEC FISH, clean microscope slides are dipped in 2% 3-aminopropyl triethoxysilane in acetone, and dried prior to applying the fixed cell suspension. Slides are aged in a desiccator for 5 d before hybridization. Chromosome spreads are probed with p108L (BAC vector, 7 kb) labeled with Prime-It-Fluor kit (Stratagene) to detect the BAC-containing clones. The FISH procedure is followed as outlined in the Prime-It-Fluor instructions. Final probe concentration is 10 ng/µL and final blocking DNA concentration (Oncor’s sonicated herring testes DNA) is 0.8 µg/µL. A total of 110 µg probe is used per hybridization and incubated with 20 µL propidium iodide–antifade solution (Oncor). Metaphase and interphase spreads are observed at ×100 magnification on an Olympus IMT-2 microscope equipped with epifluorescence and filters for fluorescein detection. Color pictures were taken using Kodak Ultra Gold film (ASA 400). Extrachromosomal and integration events are counted from 50 to 100 metaphase and interphase spreads (Fig. 6). For DR cells alone, an average of Fig. 5. (opposite page) Episomal analysis of supercoiled BAC/HAEC clones. (A) The BAC/HAEC clones are transfected into DR cells and the transfectants are analyzed as pools of cell clones. The size of the undigested episomal DNA isolated from DR cells is compared by agarose gel electrophoresis to the original DNA stock grown up in bacteria. The DNAs are probed with p108L or with the rDNA BAC as noted. An EBNA-1 probe is used to visualize the resident EBV in DR cells. The size of the 90-kb rDNA BH is determined by comparison to EBV (170 kb) and human mitochondrial DNA (17 kb). Lanes 1 and 2 correspond to the DNA before transfection into human cells; lanes 3–8 correspond to episomal DNA isolated from DR cells as described; lanes 9–11 correspond to probing for EBV episomes isolated from DR cells. (B) Structure of episomal DNA isolated from DR cells. FIGE and PFGE analysis is performed on the episomes isolated from DR cells and compared to the initial DNA isolated from bacterial cells. The DNA is digested with EcoRI (lanes 1–4) or EcoRV (lanes 5–8). The bacterial DNA are in the odd-numbered lanes; the episomal DNA isolated from DR cells are in the even numbered lanes.

170

Scott and Vos

Fig. 6. FISH analysis of rDNA BAC/HAEC in DR cells. FISH analysis is performed on DR cells transfected with BAC/HAEC vector or 90 kb rDNA BAC/HAEC. The chromosome spreads are probed with p108L, which hybridized to the BAC sequences. The number of extrachromosomal signals from 50 to 100 nuclei is counted and plotted as a histogram. The average number of signals for DR cells and the transfectants is also shown.

0.15 copies/spread is observed. An average of two copies per cell spread is counted for the 90 kb rDNA–BH. In contrast, the BH–vector is not detected above the background value obtained for DR cells alone (0.17). Since the Southern analysis is able to easily detect the vector–BH, the authors conclude that, because of its small size, it is probably lost during the FISH washes. Nevertheless, from the molecular analysis of the BAC/HAEC shuttling system, the authors conclude that large episomes containing human repetitive elements can be stably maintained by oriP–EBNA-1 replication in human cells.

3.9. Transfer of Episomes from Human to Bacteria The BH shuttling system also allows the recovery of human-transfected episomes back into bacteria. The ability to rescue transfected clones from human cells is useful for gene targeting and allows further analysis of the stability of the cloned inserts in human cells. To test the rescue of the BH shuttling system, a 140-kb anonymous human BH clone is transfected into DR cells, and the transfectants are grown as pools. 1. Episomal DNA is isolated by alkaline lysis as described in Subheading 3.7. 2. The DNA is dialyzed against 10 mM Tris and 1 mM EDTA, pH 8.0, for 30 min

BAC/PAC Transfer by HAEC Retrofitting

171

Fig. 7. Retrieval of human episomes into bacteria. (A) Episomal DNA isolated from DR cells is electroporated into DH10B cells. Individual bacterial clones are analyzed by NotI restriction digestion within flanks the insert on both sides. Lanes A and B correspond to h HindIII and h ladder markers; lanes C–T correspond to 18 individual bacterial clones electroporated with 140 BAC/HAEC episomes after passage through DR cells. (B) Lanes A and B correspond to h HindIII and h ladder markers; lane C corresponds to 140 BH DNA from the original DNA stock; lane D corresponds to 140 BH episomes retrieved from DR cells by electroporation into DH10B cells.

by filter dialysis. 3. 2 µL DNA is transformed into electrocompetent DH10B cells by electroporation (2.5 kV, 25 µF, 200 1). 4. Bacteria are plated on LB-chloramphenicol (12.5 µg/mL) agar plates. 5. Eighteen individual clones are analyzed by NotI digestion to release the large DNA insert and by PFGE (Fig. 7A,B). Thirteen clones show no changes in size after passage of the DNA through human cells. Four clones contain no DNA; one clone shows a change in size of the episomal DNA.

From these results, the HAEC/BAC shuttle system proves to be stable and allows retrieval of the BAC genomic clone after passage through human cells as large self-replicating episomes. 4. Notes 1. Dephosphorylation reduces background self-ligation of the vector. 2. When using the Qiagen columns, the volumes recommended by the manufacturer are doubled to improve the yield of DNA recovered. Alternatively, the BAC and PAC DNA can be isolated by standard alkaline lysis extraction, followed by purification through cesium chloride density gradients. 3. The gel is electrophoresed at 5.4 V/cm for 12–14 h. Alternatively, the episomal DNA is digested with the appropriate restriction enzymes and analyzed by PFGE or FIGE.

172

Scott and Vos

Acknowledgments The authors are grateful to Drs. Tian-Qiang Sun and Eva-Maria Westphal for helpful discussions, as well as contributing to some of the experiments. Help from L. Livanos in obtaining the fish data, and from D. Wilson in the manuscript preparation, is appreciated. This work was supported by the US Department of Energy. References 1. Burke, D. T., Carle, G. F., and Olson, M. V. (1987) Cloning of large segments of exogenous DNA into yeast by means of artificial chromosome vectors. Science 236, 806–812. 2. Schlessinger, D. (1990) Yeast artificial chromosomes: tools for mapping and analysis of complex genomes. Trends Genet. 6, 248–258. 3. Monaco, A. P. and Larin, Z. (1994) YACs, BACs, PACs, and MACs: artificial chromosomes as research tools. TIBTech 12, 280–286. 4. Ioannou, P., Amemiya, C. T., Garnes, J., Kroisel, P. M., Shizuya, H., Chen, C., Batzer, M. A., and de Jong, P. J. (1994) A new bacteriophage P1-derived vector for the propagation of large human DNA fragments. Nature Genet. 6, 84–89. 5. Hoess, R. and Abremski, K. (1984) Interaction of the bacteriophage P1 recombinase Cre with the recombining loxP. Proc. Natl. Acad. Sci. USA 81, 1026–1029. 6. Shizuya, H., Birren, B., Kim, U.-J., Mancino, V., Slepak, T., Tachiiri, Y., and Simon, M. (1992) Cloning and stable maintenance of 300-kilbase-pair fragments of human DNA in Escherichia coli using an F-factor based vector. Proc. Natl. Acad. of Sci. USA 89, 8794–8797. 7. Sternberg, N. (1990) Bacteriophage P1 cloning system for the isolation, amplification and recovery of DNA fragments as large as 100 kilobase pairs. Proc. Natl. Acad. Sci. USA 87, 103–107. 8. Sun, T.-Q., Fenstermacher, D., and Vos, J.-M. H. (1994) Human artificial episomal chromosomes for cloning large DNA fragments in human cells. Nature Genet. 8, 33–41. 9. Simpson, K., McGuigan, A., and Huxley, C. (1996) Stable episomal maintenance of yeast artificial chromosomes in human cells. Mol. Cell. Biol. 6, 5117–5126. 10. Reisman, D., Yates, J., and Sugden, B. (1985) A putative origin of replication of plasmids derived from Epstein-Barr virus is composed of two cis-acting components. Mol. Cell. Biol. 5, 1822–1832. 11. Banerjee, S., Livanos, L., and Vos, J.-M. H. (1995) Therapeutic gene delivery in human b-lymphocytes with nontransfoming Epstein-Barr virus. Nature Med. 1, 1303–1308. 12. Sun, T.-Q., Livanos, E., and Vos, J.-M. H. (1996) Engineering a mini-herpesvirus as a general strategy to transduce up to 180kb of functional self-replicating human minichromosomes. Gene Ther. 3, 1081–1088. 13. Abremski, K. and Hoess, R. (1984) Bacteriophage P1 site-specific recombination. J. Biol. Chem. 259, 1509–1514.

BAC/PAC Transfer by HAEC Retrofitting

173

14. Glasser, R. and Nonoyama, M. (1974) Host cell regulation of induction of EpsteinBarr virus. J. Virol. 14, 174–176. 15. Sun, T.-Q. and Vos, J.-H. M. (1996) Engineering 100- to 300-kb DNA as persisting extrachromosomal elements in human cells using human artificial episomal chromosome system, in Methods in Molecular Genetics (Adolf, K., ed.), Academic, San Diego, CA, pp. 167–188.

Intramuscular Injection of Plasma DNA

175

14 Systemic Delivery of Therapeutic Proteins by Intramuscular Injection of Plasmid DNA Holly M. Horton and Suezanne E. Parker 1. Introduction In vivo delivery of a cytokine gene to treat a tumor has usually involved either injection of ex vivo transfected cells around the tumor site or direct intratumoral injection of a virus or plasmid DNA (pDNA) vector encoding the cytokine gene (1,2). In this manner, transfected cells in or around the tumor site may secrete cytokine locally and stimulate an antitumor immune response (3,4). Recently, a new method of cytokine gene delivery for treating tumors was described. In this method, a naked pDNA encoding a cytokine, in this case, interferon-_ (IFN-_), was injected intramuscularly (im) into C57BL/6 mice bearing solid or metastatic B16F10 melanoma tumors (5). The mice treated in this manner had a striking inhibition of tumor growth. Although previous studies have shown that im injection of a cytokine-expressing pDNA could result in therapeutic effects in rodent models of autoimmunity or viral infectious disease (6–10), no previous studies had evaluated the ability of im cytokine pDNA to treat a distant tumor. The authors have found that im injection of IFN-_ encoding pDNA can lead to systemic circulating levels of IFN-_ (Horton, H. and Parker, S., unpublished data) and a significant antitumor effect in the aggressive B16F10 melanoma tumor model (5).The systemic IFN-_ pDNA therapy also markedly reduced the number of B16F10 lung metastases. The relevance of these findings is that one may be able to treat human cancer patients with a simple im injection of a naked pDNA expressing a cytokine gene, delivered on an infrequent basis. These results are supported by several recent studies (11,12) revealing that im injection of a pDNA encoding either IL-12 or the antiangiogenic factor, endostatin, can also reduce tumor formation in several mouse tumor models. From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

175

176

Horton and Parker

The im delivery of pDNA can also be used to administer a therapeutic protein for diseases other than cancer. Tripathy et al. (13) found that a single injection of a pDNA encoding erythropoietin in mice resulted in elevated hematocrits for up to 90 d. In other studies, transforming growth factor-`, delivered by im pDNA injection was beneficial for the treatment of lupus (6), diabetes (7), and arthritis (8); interleukin-10 delivered by im pDNA injection had therapeutic effects in diabetes (9). Furthermore, im injection of mIFN-_ pDNA was found to have antiviral effects in mice infected with murine cytomegalovirus (CMV) (10). Thus, im injection of pDNA encoding a therapeutic gene may be an effective strategy for numerous diseases. The delivery of a therapeutic gene using pDNA has several advantages compared to other types of gene delivery. The construction and purification of pDNA is simple and easy to scale-up (14). pDNA delivery of a therapeutic gene is distinguished from viral vectors, in that pDNA does not induce antivector antibodies that can shut down expression of the delivered gene and, in numerous studies (15,16) appears to be safe and well-tolerated. One criticism of pDNA-mediated gene delivery is that pDNA has a lower in vivo transfection efficiency compared to viral vectors. However, in recent years, improvements in the vector design have increased the degree of gene transfection by greater than 100-fold (17,18). Therefore, pDNA delivery of a therapeutic gene is a useful technique that may have broad applications for a variety of diseases. 2. Materials 2.1. Plasmid Construction The IFN-_ pDNA used in these studies, VR4111 (Vical, San Diego, CA), was constructed by cloning murine IFN-_ cDNA (gift from P. Pitha-Rowe, Johns Hopkins University, Baltimore) into the polylinker region of the eukaryotic expression vector, VR1055 (Vical) (5,17). VR1055 contains the cytomegalovirus immediate early (CMV IE) gene promoter/enhancer, the CMV IE 5' untranslated sequence, the CMV intron A sequence, a transcriptional terminator region derived from the rabbit `-globin gene, and a bacterial kanamycin resistance gene. The IFN-t pDNA used in these studies, VR4151 (Vical) was constructed by amplification of IFN-t from human genomic DNA and insertion into the polylinker region of VR1055 (5). The backbone pDNA, VR1055, served as the control pDNA for all studies.

2.2. Plasmid Purification pDNA is produced by bacterial fermentation and may be purified by standard double cesium chloride-ethidium bromide gradient ultracentrifugation followed by ethanol precipitation and dialysis (14). Alternatively, the pDNA may be purified using endotoxin-free column purification (Qiagen, Valencia, CA).

Intramuscular Injection of Plasma DNA

177

All plasmid preparations should be free of detectable RNA and have endotoxin levels of less than 0.06 endotoxin U/µg plasmid DNA. The spectrophotometric A260:A280 ratios should be between 1.75 and 2.0.

2.3. Cell Lines Cell lines were grown in culture medium obtained from Life Technologies (Gaithersburg, MD) and serum obtained from HyClone (Logan, Utah). The murine melanoma cell line, VM92, was provided by Dr. G. Nabel (University of Michigan, Ann Arbor), and grown in RPMI 1640 with 10% fetal calf serum (FCS). The murine B16F10 line was provided by Dr. F. Suzuki (University of Texas, Galveston) and grown in RPMI 1640 with 5% fetal bovine serum. 2.4. Other Materials 1. Sodium phosphate buffer (0.15 M, NaH2PO4/Na2HPO4, pH 7.2). 2. Sterile 0.3-cc tuberculin syringes fitted with a 28G 1/2 needle (Becton Dickinson, San Jose, CA) modified with a plastic collar cut from a 200-µL micropipet tip. 3. IFN-t enzyme-linked immunosorbent assay (ELISA) kit (Alexis, San Diego, CA).

3. Methods 3.1. In Vitro Analysis of pDNA Gene Expression Prior to initiating in vivo studies, the cytokine-encoding pDNA should be assayed for the ability to express the encoded gene: This can be done using standard in vitro transfection procedures. For VR4111 (mIFN-_ pDNA), VM92 cells are plated at a concentration of 2 × 105 cells/well in a 6-well plate, and incubated for 24 h. Medium is removed from the cells, which are then washed with phosphate-buffered saline followed by addition of the VR4111 pDNA complexed with the cationic lipid (±)-N-(2-hydroxyethyl)-N,N-dimethyl2,3-bis(tetradecyloxy)-1-propanaminium bromide and the neutral lipid dioleoylphosphatidylethanolamine (Vical) (DMRIE–DOPE) (1 µg of each, 1 mL/ well) in Optimem medium (Life Technologies). The cationic lipid DMRIE–DOPE and has been shown to be effective for in vitro (19) and in vivo (3) transfection. Control wells are treated with control pDNA(VR1055):DMRIE–DOPE in the same manner. After incubation of 4–5 h at 37°C, 1 mL Optimem with 30% FCS is added to each well, followed by addition of 1 mL Optimem with 10% FCS the next day. Tissue culture supernatants are collected 48 h after the start of the in vitro transfection and assayed for expression. In the case of mIFN-_, an enzyme-linked immunosorbent assay (ELISA) kit is not commercially available so expression can be monitored in a standard in vitro antiviral assay (5). Cytokines for which a commercial ELISA is available may be assayed using ELISA. The readings for the supernatants from the cells transfected with the control pDNA:DMRIE–DOPE should be subtracted from the readings for the cells transfected with the cytokine pDNA:DMRIE–DOPE to control for nonspecific background levels.

178

Horton and Parker

3.2. Rodent Tumor Models B16F10 subcutaneous (sc) tumors are established by sc injection of 104 B16F10 cells into 7–10-wk-old C57BL/6 mice (San Diego, CA). Tumors are measured in 3 dimensions (length [L] × width [W] × height [H]), 3×/wk, using calipers. The tumor volume (mm3) is determined using the formula: Tumor volume (mm3) = 0.52 (L × W × H) (20).

3.3. IM Delivery of pDNA to Treat Tumor pDNA (50 µg/50 µL) is diluted in sterile 0.15 M sodium phosphate buffer (NaH2PO4/Na2HPO4, pH 7.2) and injected into the rectus femoris muscle of each hind leg for a total pDNA dose of 100 µg. The muscle injections are performed using a sterile 0.3-cc tuberculin syringe fitted with a 28G 1/2 needle (Becton Dickinson), modified with a plastic collar cut from a 200-µL micropipet tip. The collar length is adjusted to limit the needle from penetrating beyond 2 mm into the muscle. 3.3.1. Optimization of pDNA Injection Regimen For the initial in vivo evaluation of a cytokine pDNA in a tumor model, mice are injected im with 100 µg cytokine pDNA 2/wk for 3 wk, beginning 4 d after tumor cell injection (Fig. 1). In the B16F10 melanoma tumor model, palpable tumors are apparent by d 12 (50–200 mm3) after tumor cells, which is approx 1 wk after the start of pDNA injections. If the tumor model is less aggressive, it may be necessary to start the pDNA at a later date after the cell injection, so that the pDNA begins approx 1 wk prior to the development of palpable tumors. Follow-on studies can evaluate other treatment regimens whereby the pDNA injections begin after palpable tumor growth. The regimen described here is optimized for the best efficacy in an aggressive tumor model (i.e., 100 µg pDNA, 2/wk for 3 wk, beginning 4 d after tumor cell injection) (see Note 1). The control groups should consist of mice injected with control pDNA and sodium phosphate vehicle alone using the same injection regimen. Tumor growth of the cytokine pDNA-treated mice is compared to the tumor growth of the control pDNA-treated mice to determine whether the cytokine pDNA im therapy has a significant antitumor effect. The tumor growth of the vehicle-treated mice should also be compared with the control pDNA-treated mice to determine whether im injection of the control pDNA has a nonspecific effect on tumor growth. If a strong nonspecific effect is apparent, so that it is difficult to distinguish the efficacy of the cytokine pDNA treatment from nonspecific effects of the control pDNA, either the dose of pDNA or the number of pDNA injections should be reduced. After beginning the im pDNA injections, palpable tumors are measured in 3 dimensions (L × W × H), 3×/wk using calipers. The tumor volume is calculated using the formula: Tumor volume (mm3) = 0.52 (L × W × H) (20).

Intramuscular Injection of Plasma DNA

179

Fig. 1. Systemic IFN-_ pDNA therapy in B16F10 melanoma. C57BL/6 mice were injected sc with 104 B16F10 cells. Beginning on d 4 after tumor cell injection, mice were injected im 2×/wk for 3 wk with 100 µg VR4111 (mIFN-_) or VR1055 (control) pDNA. n = 10 mice/group.

In order to achieve statistically significant results, each group should consist of a minimum of 10 mice. If a significant antitumor effect is found after injection of 100 µg cytokine pDNA, delivered 2/wk for 3 wk, then regimens involving fewer injections should be evaluated. Suggested regimens include 100 µg pDNA injected im 1/wk for 3 wk or once every other week for 4 wk, with all pDNA injections beginning one week before palpable tumors. Each treatment regimen should include a control pDNA injected im according to the same time course.

3.3.2. Dose-Response Once the optimal pDNA treatment regimen (i.e., number of injections and frequency of injections) is determined as described in Subheading 3.3.1., one may determine the lowest dose at which a significant antitumor effect still occurs. For the dose-response study, mice are injected im with 100, 50, 25, and 10 µg pDNA in the optimal treatment regimen. Each dose of cytokine pDNA should have its own control pDNA delivered at the same dose for comparison. The pilot studies should consist of a minimum of 10 mice/group to increase the chance of significant findings. Determination of the lowest dose of pDNA that still yields a significant anti-

180

Horton and Parker

tumor effect provides information on the potency of the pDNA therapy. For treatment of sc B16F10 melanoma, the authors found that 100 and 50 µg VR4111 (mIFN-_) had similar antitumor effects and could be delivered as infrequently as once every other week (5).

3.4. IM Delivery of pDNA to Treat Infectious Disease Therapeutic proteins delivered by im injection of pDNA may also be used to prevent infectious disease. For this type of study, involving the delivery of a cytokine such as IFN-_, a representative protocol is as follows: Mice are injected im with 100 µg cytokine pDNA or control pDNA, beginning 5 d prior to viral infection (Fig. 2). An example of a virus that is sensitive to IFN-_ therapy is encephalomyocarditis virus (EMCV). Five days after the first pDNA im injection, mice are challenged with virus using a dose that is approximately the LD90. Two days after viral challenge, and 1/wk for 2 wk more, mice are injected with 100 µg cytokine pDNA (for a total of four injections of pDNA). Follow-on studies may evaluate other injection regimens, and these would be designed based on the results of the initial study. Mice are monitored for survival on a daily basis after viral challenge, and the survival of the mice treated with cytokine pDNA is compared to the survival of mice treated with control pDNA. Initial studies should include a vehicle control for comparison with control pDNA to identify nonspecific effects of the pDNA therapy. In order to achieve statistically significant results, each group should consist of a minimum of 10–15 mice.

3.5. Statistical Analyses Since the dependent variable (e.g., tumor size, weight, and so on) may not be normally distributed, the data should be analyzed using a nonparametric test, such as the Mann-Whitney U statistical test. Mouse survival may be analyzed using a Kaplan-Meier survival plot followed by a Logrank (Mantel-Cox) test. Differences are considered statistically significant when p value ) 0.05.

3.6. Serum Pharmacokinetics In evaluating im therapy with a cytokine-encoding pDNA, it is important to determine whether im injection of the cytokine pDNA results in circulating levels of the cytokine. Other important factors are the concentration of the cytokine in the serum, as well as the duration of expression after a single pDNA injection. Therefore, the initial pharmacokinetics study should determine the duration and level of expression of the cytokine after a single im injection of cytokine-encoding pDNA. For this study, mice are injected im once with 100 µg of either the cytokine pDNA or control pDNA (50 µg/50 µL/rectus femoris muscle in 0.15 M sodium phosphate buffer, pH 7.2) (Fig. 3). Serum is collected every 3 d for 2 wk after injection (5 mice/d), and analyzed in a cytokine ELISA. More frequent time-

Intramuscular Injection of Plasma DNA

181

Fig. 2. Systemic IFN-_ pDNA therapy of EMCV infection. C57BL/6 mice were injected im with 100 µg VR4111 (mIFN-_) pDNA, followed 5 d later by ip injection of 100 pfu EMCV. Mice then received three more im injections of 100 µg VR4111 1/wk. Control mice were injected im with control pDNA and challenged with EMCV according to the same schedule. n = 15 mice/group.

Fig. 3. A single im injection of IFN-t pDNA results in pg/mL serum levels of IFN-t over 2 wk. C57BL/6 mice were injected im with 100 µg VR4151 (hIFN-t) in 100 µL sodium phosphate buffer. Serum was collected every 3 d from five mice per time-point and assayed for levels of hIFNt using ELISA. Mice injected im with VR1012 (control) pDNA had no detectable levels of IFN-t.

182

Horton and Parker

points can be taken, but this should be done on cohorts of mice to avoid bleeding the same mice on consecutive days. If no ELISA is available for the particular cytokine, as is the case for mIFN-_, other assays may be employed such as in vitro bioassays. For determination of mIFN-_ serum levels, the samples were assayed using an in vitro antiviral assay (5). 4. Notes 1. Since different tumor types may vary in the optimal injection regimen, pilot studies to determine the optimal regimen must be conducted for each tumor model.

References 1. Kay, M. A., Liu, D., and Hoogerbrugge, P. M. (1997) Gene therapy. Proc. Natl. Acad. Sci. USA 94, 12,744–12,746. 2. Roth, J. A. and Cristiano, R. J. (1997) Gene therapy for cancer: What have we done and where are we going? J. Natl. Cancer Inst. 89, 21–39. 3. Saffran, D. C., Horton, H. M., Yankauckas, M. A., Anderson, D., Barnhart, K. M., Abai, A., et al. (1998) Immunotherapy of established tumors in mice by intratumoral injection of interleukin-2 plasmid DNA: Induction of CD8+ T-cell immunity. Cancer Gene Ther. 5, 321–330. 4. Coleman, M., Muller, S., Quezada, A., Mendiratta, S. K., Wang, J., Thull, N. M., et al. (1998) Nonviral interferon _ gene therapy inhibits growth of established tumors by eliciting a systemic immune response. Hum. Gene Ther. 9, 2223–2230. 5. Horton, H. M., Anderson, D., Hernandez, P., Barnhart, K. M., Norman, J. A., and Parker, S. E. (1999) Gene therapy for cancer using intramuscular injection of plasmid DNA encoding interferon _. Proc. Natl. Acad. Sci. USA 96, 1553–1558. 6. Raz, E., Dudler, J., Lotz, M., Baird, S. M., Berry, C. C., Eisenberg, R. A., and Carson C. A. (1995) Modulation of disease activity in murine systemic lupus erythematosus by cytokine gene delivery. Lupus 4, 286–292. 7. Piccirillo, C. A., Chang, Y., and Prud’homme, G. J. (1998) TGF-B1 somatic gene therapy prevents autoimmune disease in nonobese mice. J. Immunol. 161, 3950–3956. 8. Song, X.-Y., Gu, M., Jin, W.-W., Klinman, D. M., and Wahl, S. M. (1998) Plasmid DNA encoding transforming growth factor-B1 supresses chronic disease in a streptococcal cell wall-induced arthritis model. J. Clin. Invest. 101, 2615–2621. 9. Nitta, Y., Tashiro, F., Tokui, M., Shimada, A., Takei, I., Tabayashi, K., and Miyazaki, J. I. (1998) Systemic delivery of interleukin 10 by intramuscular injection of expression plasmid DNA prevents autoimmmune diabetes in nonobese diabetic mice. Hum. Gene Ther. 9, 1701–1707. 10. Yeow, W.-S., Lawson, C. M., and Beilharz, M. W. (1998) Antiviral activities of individual murine IFN-_ subtypes in vivo: Intramuscular injection of IFN expression constructs reduces cytommegalovirus replication. J. Immunol. 160, 2932–2939. 11. Schultz, J., Pavlovic, J., Strack, B., Nawrath, M., and Moelling, K. (1999) Longlasting anti-metastatic efficiency of interleukin 12-encoding plasmid DNA. Hum. Gene Ther. 10, 407–417.

Intramuscular Injection of Plasma DNA

183

12. Blezinger, P., Wang, J., Gondo, M., Quezada, A., Mehrens, D., French, M., et al. (1999) Systemic inhibition of tumor growth and tumor metastases by intramuscular administration of the endostatin gene. Nature Biotechnol. 17, 343–348. 13. Tripathy, S. K., Svensson, E. C., Black, H. B., Goldwasser, E., Margalith, M., Hobart, P., and Leiden, J. M. (1996) Long-term expression of erythropoietin in the systemic circulation of mice after intramuscular injection of a plasmid DNA vector. Proc. Natl. Acad. Sci. USA 93, 10,876–10,880. 14. Horn, N., Meek, J., Budahazi, G., and Marquet, M. (1995) Cancer gene therapy using plasmid DNA: purification of DNA for human clinical trials. Hum. Gene Ther. 6, 565–573. 15. Nabel G. J., Nabel, E. G., Yang, Z.-Y., Fox, B. A., Plautz, G. E., Gao, X., et al. (1993) Direct gene transfer with DNA-liposome complexes in melanoma: Expression, biologic activity, and lack of toxicity in humans. Proc. Natl. Acad. Sci. 90, 11,307–11,311. 16. Parker S. E., Borellini F., Wenk M. L., Hobart, P., Hoffman, S. L., Hestrom, R., Le, T., and Norman, J. A. (1999) Plasmid DNA malaria vaccine: tissue distribution and safety studies in mice and rabbits. Hum. Gene Ther. 10, 741–758. 17. Hartikka, J., Sawdey M., Cornefert-Jensen F., Margalith, M., Barnhart, K., Nolasco, M., et al. (1996) An improved plasmid DNA expression vector for direct injection into skeletal muscle. Hum. Gene Ther. 7, 1205–1217. 18. Manthrope, M., Cornefert-Jensen F., Hartikka, J., Felgner, J., Rundell, A., Margalith, M., and Dwarki, V. (1993) Gene therapy by intramuscular injection of plasmid DNA: studies on firefly luciferase gene expression in mice. Hum. Gene Ther. 4, 419–431. 19. Felgner, J. H., Kumar, R., Sridhar, C. N., Wheeler, C. J., Tsai, Y. J., Border, R., et al. (1994) Enhanced gene delivery and mechanism studies with a novel series of cationic lipid formulations. J. Biol. Chem. 269, 2550–2561. 20. Tomayko, M. M. and Reynolds, C. P. (1989) Determination of subcutaneous tumor size in athymic (nude) mice. Cancer Chemother. Pharmacol. 24, 148–154.

Intratumoral Plasmid DNA–Lipid

185

15 Local Delivery of Therapeutic Proteins by Intratumoral Injection of Plasmid DNA–Lipid Complexes Holly M. Horton and Suezanne E. Parker 1. Introduction There are several strategies by which one may deliver a plasmid DNA (pDNA) encoding a therapeutic gene to a tumor. One may transfect cells ex vivo, single cell clone, expand the clone in vitro, and reinject the cells at the tumor site. This is a labor-intensive process and is especially impractical for human tumor therapy. Another method is intramuscular (im) injection of the therapeutic pDNA to achieve circulating levels of the protein (discussed in Chapter 14 by Horton and Parker). A third method is to directly inject the therapeutic pDNA into the tumor. For accessible neoplasms, this is a simple procedure, and can be useful for delivery of a therapeutic gene, such as a cytokine gene, to the tumor site. Using this technique, one may achieve high local levels of a therapeutic protein, yet have low systemic levels, thereby reducing side effects (1,2). In addition, producing a cytokine locally may attract immune cells to the tumor site and promote an antitumor immune response (1–3). Furthermore, certain cytokines may be more effective when delivered locally, rather than systemically (Horton, unpublished results). pDNA–lipid may be used to deliver a variety of immunotherapeutic molecules to the tumor site. In several phase I human trials, direct in vivo injection of a pDNA–lipid complex expressing the major histocompatibility complex class I gene, HLA-B7 (Allovectin-7®), was found to result in clinical responses in HLA-B7-negative melanoma patients. Some patients had regression of the injected tumor nodule and, in a few cases, rejection of a distant uninjected nodule (4,5). Other phase I trials of Allovectin-7® in patients with squamous cell carcinoma, colorectal carcinoma, and melanoma have found pDNA– lipid therapy to have little toxicity, and to induce some partial responses (6–8). From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

185

186

Horton and Parker

Recently, an interleukin-2 (IL-2)-expressing pDNA–lipid complex (Leuvectin™) was evaluated in a phase I/II trial of patients with renal cell carcinoma, melanoma, or sarcoma (9). Two patients with renal cell carcinoma and one patient with melanoma had partial responses and the treatment was found to be safe and well-tolerated. Thus, direct in vivo delivery of a cytokine pDNA–lipid complex may be a promising strategy in human cancer patients. By delivering the gene encoding a particular cytokine as a direct in vivo injection at the tumor site, transfected cells may continue to express the therapeutic gene over an extended period of time. This may enhance the antitumor effect of the therapy, while reducing the side effects common after systemic recombinant cytokine treatment. 2. Materials 2.1. Plasmid Construction The IL-2 pDNA used in these studies, VR1110 (Vical, San Diego, CA), was constructed by cloning murine IL-2 cDNA into the eukaryotic expression vector, VR1012 (Vical) (1). VR1012 contains the cytomegalovirus immediate early (CMV IE) gene promoter/enhancer, the CMV IE 5' untranslated (UT) sequence, the 3' UT sequence from the bovine growth hormone gene for polyadenylation and transcriptional termination, and a bacterial kanamycin resistance gene. Prior to insertion into VR1012, the 5'UT sequence of the IL-2 cDNA and the first two amino acids of the leader peptide were removed and replaced with the rat insulin II gene 5'UT sequence and the coding region of the first six amino acids of the rat preproinsulin leader peptide. The modified IL-2 cDNA was then cloned into the BamHI site of VR1012, to create the pDNA VR1110. The backbone pDNA, VR1012, served as the control pDNA for all studies.

2.2. pDNA Purification pDNA is produced by bacterial fermentation and may be purified by standard double cesium chloride-ethidium bromide gradient ultracentrifugation, followed by ethanol precipitation and dialysis (10). Alternatively, the pDNA may be purified using endotoxin-free column purification (Qiagen, Valencia, CA). All plasmid preparations should be free of detectable RNA, and have endotoxin levels of less than 0.06 endotoxin U/µg plasmid DNA. The spectrophotometric A260:A280 ratios should be between 1.75 and 2.0. 2.3. Cationic Lipid For intratumoral therapy, pDNA may be complexed with the cationic lipid, (±)-N-(2-hydroxyethyl)-N,N-dimethyl-2,3-bis(tetradecyloxy)-1-propanaminium bromide and the neutral lipid dioleoylphosphatidylethanolamine at a 1:1 mol:mol ratio. (DMRIE:DOPE) (Vical) (11), which has been shown to be effective for in vivo transfection of established tumors (1,8).

Intratumoral Plasmid DNA–Lipid

187

2.4. Cell Lines Cell lines were grown in culture medium obtained from Life Technologies (Gaithersburg, MD) and serum obtained from HyClone (Logan, Utah). The human melanoma cell line, VM92, was provided by Dr. G. Nabel (University of Michigan, Ann Arbor), and grown in RPMI 1640 with 10% fetal calf serum. The murine renal cell carcinoma (Renca) line was provided by Dr. D. Pardoll (Johns Hopkins University, Baltimore, MD) and grown in RPMI 1640 with 10% fetal bovine serum. Murine ovarian teratocarcinoma (MOT) cells were obtained from Dr. Robert Knapp and Dr. Robert C. Bast at the Dana-Farber Cancer Center (Boston, MA) and are grown by serial intraperitoneal (ip) transplantation of 105 cells in C3H/ HeN mice followed by collection of tumor ascites 14 d later. The tumor cells are collected from ascites by centrifugation and resuspension of the cell pellet in RPMI 1640 medium (Life Technologies) with 40% fetal bovine serum (HyClone) and 10% DMSO (Sigma, St. Louis, MO). The MOT cells are then cryopreserved in liquid nitrogen at a concentration of 107 cells/mL.

2.5. Additional Materials 1. mIL-2 enzyme-linked immunosorbent assay (ELISA) kit (R&D Systems, Minneapolis, MN).

3. Methods 3.1. In Vitro Analysis of pDNA Gene Expression Prior to initiating in vivo studies, the cytokine-encoding pDNA should be assayed for the ability to express the encoded gene. This can be readily accomplished using a standard in vitro transfection protocol. For VR1110 (mIL-2 pDNA), VM92 cells are plated at a concentration of 2 × 105 cells/well in a 6-well plate, and incubated for 24 h. Medium is removed from the cells, which are then washed with phosphate-buffered saline followed by addition of the VR1110 pDNA complexed with the cationic lipid DMRIE:DOPE (1 µg of each, 1 mL/well) in Optimem medium (Life Technologies). Control wells are treated with control pDNA(VR1055):DMRIE:DOPE in the same manner. After incubation of 4–5 h at 37°C, 1 mL Optimem with 30% fetal calf serum is added to each well, followed by addition of 1 mL Optimem with 10% fetal calf serum the next day. Tissue culture supernatants are collected 48 h after the start of the in vitro transfection and assayed for mIL-2 using a mIL-2 ELISA (R&D Systems). The readings for the supernatants from the cells transfected with the control pDNA–DMRIE:DOPE should be subtracted from the readings for the cells transfected with the cytokine pDNA–DMRIE/DOPE to control for background levels of cytokine expression.

188

Horton and Parker

3.2. Rodent Tumor Models Renca intradermal (id) tumors are established by id injection of 105 Renca cells into 7–10-wk-old Balb/c mice. Tumors are measured in three dimensions (length [L] × width [W] × height [H]), 3×/wk using calipers. The tumor volume (mm3) is determined using the formula: Tumor vol (mm3) = 0.52 (L × W × H) (12).

MOT ovarian tumors are established by intraperitoneal (ip) injection of 105 MOT cells in 7–10-wk-old C3H/HeN mice. The production of tumor ascites in mice injected ip with MOT tumor cells results in rapid weight gain (13). For this reason, the mice are monitored for tumor growth by determining body wt on 3–5 d/wk.

3.3. In Vivo Transfection of Tumors 3.3.1. Determination of Optimal In Vivo pDNA–Lipid Ratio Tumors may be transfected in vivo with pDNA, using cationic lipid-based delivery. Because in vitro transfection efficiencies for a particular cell line do not correlate with in vivo transfection efficiencies (14), initial studies must determine the optimal in vivo pDNA–lipid ratio (see Note 1). Typically, the authors compare 1:1, 2.5:1, and 5:1 pDNA–DMRIE mass ratios, as well as a “naked” pDNA, i.e., no lipid added. The pDNA encoding the cytokine gene is formulated at the various pDNA–lipid ratios and injected intratumorally. Control pDNA (lacking the cytokine gene) is complexed with lipid at each of the same pDNA– lipid ratios. Tumor growth of the cytokine pDNA–lipid-treated mice, is compared to the tumor growth of the control pDNA–lipid-treated mice to determine whether the cytokine pDNA–lipid therapy has a significant antitumor effect. A saline-injected group should also be included to compare the tumor growth of the vehicle-treated mice with the control pDNA–lipid-treated mice, to determine whether intratumoral injection of the control pDNA–lipid complex has a nonspecific effect on tumor growth. If a strong nonspecific effect is apparent, so that it is difficult to distinguish the efficacy of the cytokine pDNA–lipid treatment from nonspecific effects of the control pDNA–lipid, either the dose of pDNA–lipid or the number of pDNA–lipid injections should be reduced. An example of a typical 1:1 pDNA–lipid formulation for intratumoral injection is as follows: 100 µg pDNA is diluted in 50 µL 0.9% saline (Radix Labs, Eau Claire, WI). In a second vial, DMRIE–DOPE lipid containing 100 µg DMRIE (DMRIE–DOPE 1:1 mol:mol) is diluted in 50 µL 0.9% saline. pDNA and DMRIE–DOPE are combined, and the 100 µL pDNA–lipid complex is injected into the tumor shortly thereafter. A second example is a 5:1 pDNA–lipid complex, which would consist of 100 µg pDNA in 50 µL saline and 20 µg DMRIE (DMRIE:DOPE 1:1 mol:mol) in 50 µL saline, followed by mixing to

Intratumoral Plasmid DNA–Lipid

189

Fig. 1. Intratumoral IL-2 pDNA–lipid significantly reduces the growth of intradermal (id) Renca tumors. Balb/c mice were injected id with 105 Renca cells. Once palpable tumors were established (d 14, 30–80 mm3), mice were injected intratumorally with 50 µg VR1110 (mIL-2) or VR1012 (control) complexed with DMRIE–DOPE at a 5:1 pDNA:lipid mass ratio. The intratumoral injections were administered for four consecutive days beginning on d 14 after tumor cell injection. n = 10 mice/group.

yield a final volume of 100 µL 5:1 pDNA–lipid complex. In all cases, the pDNA and lipid are prepared in such a manner that equal volumes of pDNA and lipid are combined to yield the final pDNA–lipid complex. The examples given above are for the formulation of a pDNA–lipid complex for intratumoral injection of a subcutaneous (sc) tumor, i.e., injection of 100 µL pDNA– lipid/sc tumor of 100–300 mm3. For injection of an id tumor of 30–80 mm3, 50 µL pDNA–lipid complex containing 50 µg pDNA is injected (one-half the amount injected into an sc tumor) (Fig. 1). For injection of an ip tumor, 1 mL pDNA–lipid complex containing 100 µg pDNA is injected (Fig. 2). The formulations should be scaled up for the number of mice injected, i.e., formulate 1.2 mL pDNA–lipid complex for 10 mice and inject 100 µL/sc tumor. For the pilot study comparing different pDNA–lipid ratios for treatment of id or sc tumors, palpable id (30–80 mm3) or sc (100–300 mm3) tumors are injected over 6 consecutive days. The tumors are measured in three dimensions (L × W × H) 3×/wk, using calipers, and the tumor volume is calculated using the formula (12): Tumor volume (mm3) = 0.52 (L × W × H).

In order to achieve statistically significant results, each group should consist of a minimum of 10 mice.

190

Horton and Parker

Fig. 2. Intraperitoneal (ip) IL-2 pDNA–lipid significantly reduces the growth of ip MOT ovarian tumor ascites. C3H/HeN mice were injected ip with 105 MOT cells. On d 5, 8, and 11 after tumor cell injection, mice were injected ip with 100 µg VR1110 (mIL-2) or VR1012 (control) complexed with DMRIE–DOPE at a 1:1 pDNA:lipid mass ratio. n = 10 mice/group.

For the pilot study comparing different pDNA–lipid ratios for treatment of ip tumors, tumors are injected for 6 consecutive days beginning approx 5 d after tumor cell injection. The mice are weighed daily to monitor tumor growth. Each group should consist of a minimum of 10 mice.

3.3.2. Determination of Optimal Number of Intratumoral Injections The number of cytokine pDNA–lipid injections required for optimal tumor therapy must also be determined in a pilot study. The optimal cytokine pDNA– lipid formulation is initially determined as described in Subheading 3.3.1. This formulation is then delivered according to various schedules to identify the optimal treatment regimen. Typically, the authors compare 6 vs 4 vs 2 intratumoral injections delivered over consecutive days, beginning when sc tumors are 100–300 mm3, id tumors are 30–80 mm3, or beginning on d 5 after ip tumor cell injection. Additionally, one should compare 6 vs 4 vs 2 intratumoral injections delivered every other day. Each cytokine pDNA–lipid treatment regimen should have its own control pDNA–lipid delivered according to the same schedule. The studies should consist of 10 mice/group to increase the chance of significant findings.

Intratumoral Plasmid DNA–Lipid

191

3.3.3. Determination of Dose-Response Once the optimal pDNA–lipid injection regimen is determined as described in Subheading 3.3.2. (i.e., the number of injections and the frequency of the injections), one may determine the lowest dose at which a significant antitumor effect still occurs. The optimal pDNA–lipid ratio determined in Subheading 3.3.1., and the optimal treatment regimen determined in Subheading 3.3.2., are used for the dose-response study. For sc or ip tumors, mice are injected intratumorally with 100, 50, 25, and 10 µg pDNA; and for id tumors, mice are injected intratumorally with 50, 25, 10, and 5 µg pDNA, according to the optimal pDNA–lipid ratio and optimal injection regimen. Each dose of cytokine pDNA–lipid should have its own control pDNA–lipid delivered at the same dose for comparison and each treatment group should consist of a minimum of 10 mice to increase the chance of significant findings. By determining the lowest pDNA–lipid dose that still yields a significant antitumor effect, one can get a sense of the potency of the therapy. For the treatment of ip MOT tumor ascites, the authors found that 100 and 50 µg VR1110 (mIL-2) had similar antitumor effects, when delivered on d 5, 8, and 11 after tumor cell injection at a 1:1 pDNA–lipid mass ratio (15).

3.4. Statistical Analyses Since the dependent variable (e.g., tumor size, weight, and so on) may not be normally distributed, the data should be analyzed using a nonparametric test such as the Mann-Whitney U statistical test. Mouse survival may be analyzed using a Kaplan-Meier survival plot, followed by a Logrank (Mantel-Cox) test. Differences are considered statistically significant when p value ) 0.05.

3.5. Assay of In Vivo Expression of Cytokine by Tumor Cells To determine whether tumor cells are expressing cytokine after intratumoral injection of pDNA–lipid complex, the following procedure may be conducted. For solid tumors, intratumoral cytokine pDNA–lipid injections are performed for four consecutive days. A control set of mice are injected intratumorally with control pDNA–lipid for four consecutive days. One d after the final pDNA–lipid injection, mice are euthanized and tumors removed. The tumors are disaggregated and put into tissue culture. One day later, supernatants are collected and assayed for levels of the cytokine using ELISA. Background levels obtained from the tumors injected with control pDNA–lipid are subtracted from the final readings, to determine the amount of cytokine released by the tumor cells in vivo after intratumoral injection of a cytokine pDNA– lipid complex. Five to ten mice should be used in each treatment group. To determine the expression of cytokine after ip injection of cytokine pDNA–lipid in mice bearing ip ascites, mice are injected for two consecutive

192

Horton and Parker

days with the cytokine pDNA–lipid, beginning 5 d after ip tumor cell injection. Control tumor-bearing mice are injected ip with control pDNA–lipid, according to the same schedule. Five mice from each group are sacrificed at an early time-point (< 24 h), d 1, and then every 2–4 d post-DNA injection for 2 wk. Ascites is collected from the sacrificed mice, the samples are spun to pellet the cells and the supernatant is harvested. Cytokine concentration in the ascites is determined using a cytokine ELISA. Background levels obtained from the tumors injected with control pDNA–lipid are subtracted from the final readings to determine the amount of cytokine released by the tumor cells in vivo after intratumoral injection of a cytokine pDNA–lipid complex. Since the volume of tumor ascites increases over time, the volume of ascites is also determined for each mouse. The total amount of cytokine in the total volume of ascites may then be determined. 4. Notes 1. Since different tumor types vary in the optimal in vivo pDNA–lipid ratio, as well as the optimal pDNA injection regimen, pilot studies to determine these must be conducted for each tumor model.

References 1. Saffran, D. C., Horton, H. M., Yankauckas, M. A., Anderson, D., Barnhart, K. M., Abai, A., et al. (1998) Immunotherapy of established tumors in mice by intratumoral injection of interleukin-2 plasmid DNA: induction of CD8+ T-cell immunity. Cancer Gene Ther. 5, 321–330. 2. Coleman, M., Muller, S., Quezada, A., Mendiratta, S. K., Wang, J., Thull, N. M., et al. (1998) Nonviral interferon _ gene therapy inhibits growth of established tumors by eliciting a systemic immune response. Hum. Gene Ther. 9, 2223–2230. 3. Addison, C. L., Braciak, T., Ralston, R., Muller, W. J., Gauldie, J., and Graham, F. (1995) Intratumoral injection of an adenovirus expressing interleukin 2 induces regression and immunity in a murine breast cancer model. Proc. Natl. Acad. Sci. USA 92, 8522–8526. 4. Nabel, G. J., Nabel, E. G., Yang, Z. Y., Fox, B. A., Plautz, G. E., Gao, X., et al. (1993) Direct gene transfer with DNA-liposome compelxes in melanoma: expression, biologic activity, and lack of toxicity in humans. Proc. Natl. Acad. Sci. USA 90, 11,307–11,311. 5. Nabel, G. J., Gordon, D., Bishop, D. K., Nickoloff, B. J., Yang, Z. Y., Aruga, A., et al. (1996) Immune response in human melanoma after transfer of an allogeneic class I major histocompatibility complex gene with DNA-liposome complexes. Proc. Natl. Acad. Sci. USA 93, 15,388–15,393. 6. Gleich, L. L., Gluckman, J. L., Armstrong, S., Biddinger, P. W., Miller, M. A., Balakrishnan, K., et al. (1998) Alloantigen gene therapy for squamous cell carcinoma of the head and neck: results of a phase-I trial. Arch. Otolaryngol. Head Neck Surg. 124, 1097–1104.

Intratumoral Plasmid DNA–Lipid

193

7. Rubin, J., Galanis, E., Pitot, H. C., Richardson, R. L., Burch, P. A., Charboneau, J. W., et al. (1997) Phase I study of immunotherapy of hepatic metastases of colorectal carcinoma by direct gene transfer of an allogeneic histocompatibility antigen, HLA-B7. Gene Ther. 4, 419–425. 8. Stopeck, A. T., Hersh, E. M., Akporiaye, E. T., Harris, D. T., Grogan, T., Unger, E., et al. (1997) Phase I study of direct gene transfer of an allogeneic histocompatibility antigen, HLA-B7, in patients with metastatic melanoma. J. Clin. Oncol. 15, 341–349. 9. Galanis, E., Hersh, E. M., Stopeck, A. T., Gonzalez, R., Burch, P., Spier, C., et al. (1999) Immunotherapy of advanced malignancy by direct gene transfer of an IL2 DNA/DMRIE/DOPE lipid complex (Leuvectin): phase I/II experience. J. Clin. Oncol, in press. 10. Horn, N., Meek, J., Budahazi, G., and Marquet, M. (1995) Cancer gene therapy using plasmid DNA– purification of DNA for human clinical trials. Hum. Gene Ther. 6, 565–573. 11. Felgner, J. H., Kumar, R., Sridhar, C. N., Wheeler, C. J., Tsai, Y. J., Border, R., et al. (1994) Enhanced gene delivery and mechanism studies with a novel series of cationic lipid formulations, J. Biol. Chem. 269, 2550–2561. 12. Tomayko, M. M. and Reynolds, C. P. (1989) Determination of subcutaneous tumor size in athymic (nude) mice. Cancer Chemother. Pharmacol. 24, 148–154. 13. Berek, J. S., Cantrell, J. L., Lichtenstein, A. K., Hacker, N. F., Knox, R. M., Nielberg, R. K., et al. (1984) Immunotherapy with biochemically dissociated fractions of propionibacterium acnes in a murine ovarian cancer model. Cancer Res. 44, 1871–1875. 14. Barnhart, K. M., Hartikka, J., Manthorpe, M., Norman, J., and Hobart, P. (1998) Enhancer and promoter chimeras in plasmids designed for intramuscular injection: a comparative in vivo and in vitro study. Hum. Gene Ther. 9, 2545–2553. 15. Horton, H. M., Dorigo, O., Hernandez, P., Anderson, D., Berek, J. S., and Parker S. (1999) IL-2 plasmid therapy of murine ovarian carcinoma inhibits the growth of tumor ascites and alters its cytokine profile. J. Immunol. 163, 6378–6385.

Polymeric Delivery of DNA

195

16 Nonviral DNA Delivery from Polymeric Systems Lonnie D. Shea and David J. Mooney 1. Introduction Gene therapy holds great promise for the treatment of disease by delivering genes encoding for therapeutic proteins. Although it was originally devised for the treatment of inherited genetic disorders, such as cystic fibrosis, recent work has expanded the applications of gene therapy to develop strategies for HIV, cancer (1), and wound healing applications (2). The challenge of gene therapy is to develop safe and efficient gene delivery systems (1). Most studies have focused on the use of viral vectors because of their potentially high efficiencies; however, the safety and ease of manufacturing of nonviral vectors may make them the preferred choice in the future. A major limitation for the nonviral approach to gene therapy is the inability to effectively deliver the DNA. The barrier to efficient delivery is frequently the lack of cellular uptake, rapid degradation of the DNA, or clearance from the body (3,4). Strategies to overcome these obstacles include the use of cationic lipids or polycations to condense the DNA and facilitate uptake, or the gene gun to directly deliver the DNA into the cell nucleus. Recently, however, polymer-based drug delivery systems have been examined for their abilities to efficiently deliver nonviral DNA (5). Polymer-based systems provide a number of potential advantages for the therapeutic delivery of drugs (6). Drug encapsulation within the polymer can protect against degradation until release. Injection or implantation of the polymer into the body can be used to target a particular cell type or tissue. Drug release from the polymer and into the tissue can be designed to occur rapidly, as in a bolus delivery, or to occur over an extended period of time; thus, a delivery system can be tailored to a particular application. For bolus delivery, drug levels quickly rise and decline as the drug is cleared or degraded. From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

195

196

Shea and Mooney

For sustained delivery, the drug concentration is maintained within a therapeutic window by adjusting the release rate (e.g., through the polymer choice). A result of sustained delivery is a possible reduction in the number of dosages or in the required cumulative dose (7). Sustained delivery can be particularly effective for drugs with unacceptable pharmacokinetics (e.g., short half-lives). Polymer delivery provides one approach to overcome the obstacles of rapid clearance, ineffective targeting, and limited stability associated with DNA delivery (6). Intravenous delivery of supercoiled DNA leads to rapid clearance from the circulation, because of extensive uptake by liver nonparenchymal cells via the scavenger receptors; although it does not extravasate to any organs or tissues (8). Condensation of DNA; however, does allow extravasation to highly permeable tissues, such as liver and tumor. DNA injected directly into a tissue drains rapidly into the lymphatics (9,10). Endonucleases, which are present in the extracellular space, can degrade nonviral DNA within 30 min (11). Polymeric delivery systems can function to delay clearance from the desired tissue, protect the DNA from degradation, and provide a sustained delivery to maintain DNA at high levels within the target tissue. Current DNA delivery systems can generally be categorized into two different forms: microspheres or pellets, and polymer matrices. Microspheres or pellets loaded with nonviral DNA can be fabricated from nondegradable and degradable polymers in sizes ranging from 1 to 100 µm (12). One of the main advantages for delivery vehicles of this size is that they can be administered in a minimally invasive manner (e.g., direct injection, oral delivery). The use of biodegradable polymers provides the additional advantage of not having to retrieve the implant following DNA release. Unfortunately, these two qualities make removal of the devices difficult, should the therapy need to be terminated prematurely. Formulations of microspheres or pellets have been developed to utilize nonviral DNA for vaccines and systemic protein delivery. DNA vaccines are being formulated using microspheres with diameters of approx 5 µm (13–16). These microspheres are too large to enter cells by endocytosis, but they can be phagocytosed by macrophages, which are the antigen-presenting cells for the immune system. Application of DNA for systemic protein delivery is being developed using injectable (17) and ingestible vehicles (18). These formulations are delivered to a specific anatomic location, and are designed to deliver DNA to surrounding cells for the production of a therapeutic protein. Alternatively, DNA delivery from polymer matrices is being examined for applications in tissue repair and wound healing (19,20). These matrices are typically implanted at a specific anatomic location where they serve multiple roles (2). Initially, the matrix functions to create and maintain a space in vivo. However, the matrix also acts as a scaffold to support cell migration,

Polymeric Delivery of DNA

197

proliferation, and differentiation of healthy cells from the surrounding tissue. As cells invade the matrix, they encounter DNA that is either released from or entrapped within the matrix. These transfected cells subsequently act as bioreactors for protein production. For wound-healing applications, these bioreactors may be producing a tissue-inductive factor, such as platelet-derived growth factor or parathyroid hormone (PTH), to stimulate new tissue formation. The controlled delivery of proteins is being widely investigated for the treatment of cancers and other chronic, life-threatening conditions (6). Some of the most promising systems for controlled protein release involve encapsulation or entrapment in biocompatible polymeric devices. Delivery of nonviral DNA from polymeric systems is being investigated as a means to effect local or systemic protein delivery for a wide range of potential applications. The polymers, both natural and synthetic, utilized during fabrication and the techniques for incorporation and in vivo delivery are described in the following pages. 2. Materials Recently, the strategies developed to deliver biologically active proteins have been adapted to deliver nonviral DNA. In terms of stability, a delivery vehicle releasing DNA has some distinct advantages, compared to proteinbased systems. Proteins are susceptible to loss of function based on small changes in tertiary or quaternary structure; however, aside from precipitation or aggregation, DNA must be chemically modified to lose biological activity (21). Control over the delivery of the nonviral DNA is achieved by fabricating both synthetic and natural polymers into a variety of geometries and configurations (e.g., reservoirs, matrices, and microspheres). Novel strategies for gene transfer and therapeutic application are being developed based on the particular advantages of each delivery system. A vast number of materials are being developed and utilized in the fabrication of drug delivery systems; however, this review focuses on polymeric systems that are biocompatible, have an established history in either drug delivery or cell transplantation, and have been studied for their ability to incorporate and deliver DNA.

2.1. Collagen Collagens are the major component of mammalian connective tissue. Although there are at least 14 types of collagen, the most abundant is type I collagen. Type I collagen is found in high concentrations in tendon, skin, bone, and fascia, which thus serve as a source for isolation. Following isolation, typically from bovine or porcine sources, collagen is preserved using techniques such as fixation with aldehydes or other chemical preservatives, `-irradiation, or lyophilization (22,23). The natural abundance of collagen and its unique physical and biological properties have led to investigations of its use as a

198

Shea and Mooney

biomaterial for drug delivery and cell transplantation. Collagen matrices are attractive because they are a natural component of the extracellular matrix, can be easily processed into a variety of shapes, have the ability to achieve high loading of DNA, and are relatively stable following implantation. The matrices are relatively resistant to enzymatic degradation, except by type-specific collagenases that are produced by invading fibroblasts as they attempt to remodel the collagen matrix. Much of the drug is maintained within the matrix; however, remodeling of the matrix by the fibroblasts or other cell types can result in release from the matrix. A concern when implanting collagen is its potential for causing immunogenic responses in the patient. It is important to preserve the native collagen structure when processing materials for use as implantable devices. Immunogenicity can be reduced through chemical crosslinking or protease treatment of the collagen, which removes telopeptides to form atelocollagen.

2.2. Poly(lactide-co-glycolide) Polymers composed of lactic and glycolic acid are perhaps the most widely used and recognized biodegradable synthetic polymers, because of their use as sutures (24). These polymers are FDA-approved and are generally considered to be biocompatible. Hydrolytic degradation of the polymer produces lactic acid and glycolic acid, two substances naturally involved in metabolic pathways of the body. The degradation rate can be controlled through the composition of the polymer and molecular weight of the chains. These polymers exhibit even polymer chain scission throughout their bulk after being placed in vivo (24). These properties have led to the use of these materials for drug delivery and as scaffolds for tissue engineering (12). For applications in drug delivery, drug release from these systems typically occurs by a combination of polymer degradation and drug diffusion from the polymer. For applications in tissue engineering, these polymers can readily be fabricated into three-dimensional (3-D) porous scaffolds that provide a temporary support for tissue formation.

2.3. Polyanhydride Biodegradable polyanhydride polymers were originally developed as vehicles for drug delivery (6), which has subsequently led to their use in other biomaterial applications. These polymers are synthesized from hydrophobic monomers connected by water-labile anhydride bonds. Through appropriate choice of monomers (which controls the hydrophobicity), degradation of these materials can be designed to occur over periods ranging from 1 d to many months. However, the feature that makes these materials desirable for various applications is the degradation mechanism. These polymers exhibit chain scission mainly at the surface of the polymer matrix and the matrix degrades

Polymeric Delivery of DNA

199

and loses mass from the exterior to the interior (25), analogous to a bar of soap dissolving. In drug delivery applications, drug is released continuously as the matrix degrades and is particularly attractive because it prevents dose-dumping. The observation that the material maintains its structural integrity as it degrades has motivated its use in other applications, including structure replacement in a load-bearing environment (26).

2.4. Polyethylene Vinyl Coacetate Polyethylene vinyl coacetate (EVAc) has been used as a model system for the fabrication of nondegrading controlled-release systems (12). Since the polymer is nondegradable, the release of high-molecular weight drugs is dependent on the presence of an interconnecting network of channels leading to the external environment, which is a function of the drug loading. At low drug loading, complete entrapment of isolated drug particles within the polymer prevents release. Higher drug loading or the addition of carrier proteins increases the possibility of an interconnecting network of drug particles with a path to the surroundings. As drug adjacent to the surface dissolves and diffuses out of the matrix, open channels remain, which permit drug located deeper within the matrix to be solubilized and released. Modifications that affect the network of pores, such as drug loading, polymer molecular weight, and presence of codispersants, can be used to control the release rate of the drug. 3. Methods 3.1. Delivery Vehicle Fabrication Development of technologies for drug delivery has been a long-standing focus of the biotechnology and pharmaceutical industry. Nonviral DNA release from these vehicles can be designed to occur over times ranging from hours to several weeks. Alternatively, the delivery system may be designed to never release the DNA. The choice of polymer and its physical form determine the time-scale of release. The choice of polymer affects not only the degradation rate, but also the mechanism of degradation, both of which can be designed to control DNA release. The physical form of the polymer, through parameters such as the polymer size, tortuosity of the pores, and interconnectedness of the pores, affect diffusion of DNA from the vehicle, and hence the release rate. Alternative mechanisms of release, such as chemical or enzymatic cleavage, are also being investigated for protein release and may eventually be applied to DNA (6). In developing a delivery vehicle for DNA, maintaining the stability of the incorporated DNA, not only during encapsulation and storage, but also for weeks or months after administration, is crucial to the efficacy of the delivery system (3,21). Stability of DNA is frequently quantified as the relative amounts of supercoiled and open-circular DNA using gel electrophoresis. However,

200

Shea and Mooney

both supercoiled and open-circular DNA have biological activity that causes difficulties in creating a precise definition of DNA stability (21). To maintain DNA integrity, many strategies have been investigated to stabilize DNA during polymer processing. EDTA is typically present because it chelates divalent metal cations and inhibits the activity of DNase. Various polymer-processing conditions (i.e., organic solvents, high temperatures) or techniques (e.g., lyophilization, sonication) are utilized during polymer fabrication and can cause DNA to become nicked or linear. Saccharides are known protectants for proteins during these polymer processing techniques and are also being used to perform a similar function for DNA (27,28). For stability during lyophilization, lactose, maltose, trehalose, cellobiose, and mannitol have been used. The use of lactose has been shown to increase the ratio of supercoiled to nicked DNA. Sonication, commonly used in microsphere fabrication, is a step that can degrade DNA. Protection against sonication can be provided by the use of cationic peptides (29). The delivery systems are fabricated to stably incorporate and release nonviral DNA in the form of microspheres or pellets and matrices. These forms can be delivered in a minimally invasive manner, or by implantation to a specific anatomic location to achieve either local or systemic protein production for applications such as DNA vaccines or wound healing. The following subheadings describe the fabrication procedure for these polymers and the results obtained using the delivery system.

3.2. Poly(lactide-co-glycolide) Microspheres For applications as DNA vaccines, poly(lactide-co-glycolide) (PLG) microspheres are being fabricated with diameters of approx 5 µm, to specifically target macrophages for internalization of the microspheres (13,14). Aqueous solutions of nonviral DNA, both supercoiled and complexed with poly-L -lysine, have been encapsulated into polymer microspheres using a double-emulsion process (30,31). In this process, the nonviral DNA, in an aqueous solution, is dispersed within a solution of polymer, such as PLG in methylene chloride, to form aqueous droplets within a continuous organic phase. The aqueous solution typically contains stabilizers, such as EDTA and lactose, to enhance DNA stability. The organic phase is subsequently emulsified into an aqueous phase. The resulting solution is a double emulsion in which a continuous aqueous phase contains droplets of an organic polymer phase, which contains droplets of the aqueous DNA solution. The continuous aqueous phase also contains stabilizers, such as poly(vinyl alcohol) (PVA), to stabilize the double emulsion. Microspheres are formed from this aqueous-organic-aqueous double emulsion by extraction and evaporation of the organic solvent, which causes the polymer to precipitate around the DNA, and results in

Polymeric Delivery of DNA

201

Fig. 1. PLG microsphere fabricated using a double-emulsion process. Microsphere shown was fabricated with incorporated protein, but nonviral DNA incorporation should produce a similar microsphere. Original magnification was ×2500.

hardened microspheres (Fig. 1). This procedure forms microspheres that are approx 3–5 µm in diameter. When supercoiled DNA is incorporated using this procedure, the incorporated DNA is structurally intact with 39% of the DNA in the supercoiled conformation. Typical incorporation efficiencies are approx 20–30%. Condensed DNA has been utilized to protect DNA during the emulsion step, which can involve high shear stresses caused by sonication. Condensation increases the DNA present in the supercoiled form to 85%. A plasmid encoding for firefly luciferase was encapsulated into PLG microspheres as a model plasmid for development of a DNA vaccine. Delivery of supercoiled DNA by intraperitoneal injection, or orally by gavage, led to the presence of specific antibodies in serum 3 and 6 wk after administration (15). To increase the relatively low incorporation efficiency of the process, Ando et al. (32) developed a novel cryopreparation method that is a modification of the standard double-emulsion procedure. Following formation of the primary emulsion, the temperature is lowered below the freezing point of the inner aqueous phase by immersion in liquid nitrogen. The inner aqueous phase is selectively frozen, which has the dual effect of increasing DNA integrity in addition to the incorporation efficiency. Supercoiled DNA was encapsulated with 89% efficiency, while maintaining the percentage of supercoiled DNA at 88%.

202

Shea and Mooney

3.3. Polyanhydride Microspheres Polyanhydride microspheres containing nonviral DNA was developed for oral delivery to the cells lining the gastrointestinal tract (18). The biological adhesivity of these microspheres, which is attributed to the surface carboxylic acids, delays passage of these microspheres through the gastrointestinal system. Polyanhydride polymers were fabricated from polymerization of fumaric acid and sebacic acid in a ratio of 20:80. The microspheres were subsequently fabricated from the polymers using a phase inversion process. Nonviral DNA is added to a polymer solution, such as polyfumaric acid:sebacic acid in methylene chloride, which is subsequently poured rapidly into an unstirred bath of petroleum ether, a nonsolvent, at a ratio of 1:100. Combining these two mixtures results in the spontaneous formation of nano- and microspheres (0.1–5.0 µm in diameter). The migration pattern of released DNA indicates that processing did not alter its structural integrity. Oral delivery of these microspheres led to transfection of cells in the muscularis mucosae and adventitia below the Peyer’s patches.

3.4. Injectable Collagen Pellets Intramuscular injection of a collagen pellet containing a plasmid encoding for fibroblast growth factor-4 (FGF-4) was investigated to achieve physiologically useful concentrations of the protein in the systemic circulation (17). A collagen–DNA pellet was fabricated into a size that could be delivered using a syringe into the desired anatomic location. To fabricate the pellet, an aqueous solution of plasmid DNA is mixed with an aqueous collagen solution (250 mg atellocollagen, 110 mg glucose), and subsequently lyophilized. This product is rehydrated and pushed through an 18-gage needle to form a rod that is then airdried. The resulting rod is cut into the appropriate lengths, then injected intramuscularly. Intramuscular injection of the collagen pellet led to pg/mL concentrations of FGF-4 in serum and an increased platelet counts.

3.5. EVAc Rods A porous EVAc matrix containing encapsulated DNA, was fabricated using an emulsion consisting of DNA/water and EVAc/methylene chloride (33). Upon emulsion formation, the mixture was quickly frozen in liquid nitrogen to prevent phase separation. Following overnight lyophilization, the polymer was heatextruded into a rod (1 mm diameter), using a custom apparatus operated at 48°C. The rods were subsequently cut into 1- or 5-mm lengths. Encapsulation efficiencies of at least 85% were obtained. Release profiles showed that a majority of the DNA was released during the first 7 d; however, release for more than 1 mo was observed. The released DNA was structurally intact (but found primarily in the open conformation after 2 d of release) and transfection competent.

Polymeric Delivery of DNA

203

3.6. Collagen Matrices Collagen-based systems for plasmid delivery have been used for applications in bone regeneration (20,34). In these applications, the collagen gel functions to entrap the nonviral DNA, and also provides a scaffold that supports cell attachment and cell migration. The nonviral DNA is incapable of diffusing through the collagen carrier. Thus, the matrix serves to hold DNA in situ until endogenous fibroblasts arrive. As repair fibroblasts grow into the wound, they are transfected by the plasmid and become bioreactors for the local production of a cytokine involved in tissue formation. The gel is designed for implantation at a wound site. For studies performed with bone regeneration, collagen matrices, loaded with a plasmid encoding for a 34-amino-acid segment of parathyroid hormone (PTH1–34), were implanted into a canine osteotomy model. Matrices were fabricated by thorough mixing of nonviral DNA with bovine collagen, which is subsequently frozen at –86°C, then lyophilized. The delivery of 100 mg PTH plasmid from collagen matrices resulted in bone regeneration across a 1.6-cm defect over a 6-mo time period.

3.7. PLG Matrices Interconnected open-pore matrices fabricated from PLGA copolymers have been utilized as scaffolds for the formation of a number of tissues such as bone, liver, and cartilage. The scaffolds are created containing pores with diameters greater than a cell diameter to allow for cell invasion into the polymer matrix. The surface of the polymer provides a support for cell migration, proliferation, and differentiation. Additionally, these scaffolds are biodegradable, so that as the tissue develops, the matrix degrades leaving behind a completely natural tissue. Recently, a technique was developed to fabricate similar 3-D porous polymer matrices that also serve as vehicles for the delivery of nonviral DNA (Fig. 2; 19). The fabrication technique is a gas-foaming/particulate-leaching process (35). Polymer, a porogen (e.g., sodium chloride), and lyophilized DNA are mixed and compression molded into a disk. The disk is subsequently allowed to equilibrate within a high-pressure CO2 environment. After equilibration, the pressure is rapidly released, which causes the individual polymer particles to expand and fuse together to form an intact 3-D polymer matrix. Placement of the resulting disk in an aqueous environment removes the porogen, thus creating an interconnected, open-pore matrix with entrapped DNA. The incorporation efficiency for this process was determined to be at least 50%. Release studies demonstrated a sustained release of DNA for more than 30 d. The released DNA is structurally intact, but tends toward the open conformation at later times of release. Implantation of these matrices into a subcutaneous pocket led to transfected cells throughout the polymer. Release of a plasmid encoding for platelet-derived growth factor resulted in physiological responses at 2 and 4 wk in vivo.

204

Shea and Mooney

Fig. 2. Porous PLG matrix, which functions as a scaffold for cell growth and as a vehicle for delivery of nonviral DNA to cells on the scaffold. Scanning electron photomicrograph of a cross-section of the PLG scaffold. Original magnification was ×200.

3.8. Summary and Future Directions Nonviral DNA delivery from polymeric systems may enhance transfection by limiting degradation, increasing retention in the tissue, and producing high local concentrations. Sustained delivery of nonviral DNA, by providing a continual source of DNA in the tissue, may enhance gene uptake and expression and provide a means to prolong gene expression. The polymeric system may also function as a vehicle to deliver cells to a specific location, and ultimately degrade or resorb, so that it does not have to be retrieved. Current formulations typically involve microspheres (or pellets), which can be delivered in a minimally invasive manner, or matrices that are implanted at the appropriate site. Future studies will undoubtedly develop novel polymeric delivery systems (e.g., nanospheres), which can be administered in novel ways (e.g., aerosol). These systems may be designed to deliver the DNA by mechanisms other than polymer degradation and DNA diffusion. In addition, studies need to be performed to develop a quantitative assessment of the factors involved in the success of these implants. An understanding of plasmid availability and uptake, the cell types involved and their properties governing successful transfection, and protein production levels and requirements for therapeutic effectiveness will lead to a better understanding of the fundamental design parameters, and

Polymeric Delivery of DNA

205

thus the development of optimal delivery systems. Delivery of nonviral DNA represents a powerful strategy for the development of novel therapies with the potential to treat a range of disorders. References 1. Anderson, W. F. (1998) Human gene therapy. Nature 392, 25–30. 2. Bonadio, J., Goldstein, S. A., and Levy, R. J. (1998) Gene therapy for tissue repair and regeneration. Adv. Drug Delivery Rev. 33, 53–69. 3. Ledley, F. D. (1996) Pharmaceutical approach to somatic gene therapy. Pharm. Res. 13, 1595–1614. 4. Ledley, F. D. (1995) Nonviral gene therapy: the promise of genes as pharmaceutical products. Hum. Gene Ther. 6, 1129–1144. 5. Luo, D. and Saltzman, W. M. (2000) Synthetic DNA delivery systems. Nat. Biotechnol. 18, 33–37. 6. Langer, R. (1998) Drug delivery and targeting. Nature 392, 5–10. 7. Sanders, L. M., Kell, B. A., McRae, G. I., and Whitehead, G. W. (1986) Prolonged controlled-release of nafarelin, a luteinizing hormone-releasing hormone analogue, from biodegradable polymeric implants: influence of composition and molecular weight of polymer. J. Pharm. Sci. 75, 356–360. 8. Takakura, Y., Mahato, R. I., and Hashida, M. (1998) Extravasation of macromolecules. Adv. Drug Delivery Rev. 34, 93–108. 9. Choate, K. A. and Khavari, P. A. (1997) Direct cutaneous gene delivery in a human genetic skin disease. Hum. Gene Ther. 8, 1659–1665. 10. Levy, M. Y., Barron, L. G., Meyer, K. B., and Szoka, F. C., Jr. (1996) Characterization of plasmid DNA transfer into mouse skeletal muscle: evaluation of uptake mechanism, expression and secretion of gene products into blood. Gene Ther. 3, 201–211. 11. Kawabata, K., Takakura, Y., and Hashida, M. (1995) Fate of plasmid DNA after intravenous injection in mice: involvement of scavenger receptors in its hepatic uptake. Pharm. Res. 12, 825–830. 12. Baldwin, S. P. and Saltzman, W. M. (1998) Materials for protein delivery in tissue engineering. Adv. Drug Delivery Rev. 33, 71–86. 13. Hedley, M. L., Curley, J., and Urban, R. (1998) Microspheres containing plasmid-encoded antigens elicit cytotoxic T-cell responses. Nat. Med. 4, 365–368. 14. Wang, D., Robinson, D. R., Kwon, G. S., and Samuel, J. (1999) Encapsulation of plasmid DNA in biodegradable poly(D,L-lactic-co-glycolic acid) microspheres as a novel approach for immunogene delivery. J. Controlled Release 57, 9–18. 15. Jones, D. H., Corris, S., McDonald, S., Clegg, J. C., and Farrar, G. H. (1997) Poly(D,Llactide-co-glycolide)-encapsulated plasmid DNA elicits systemic and mucosal antibody responses to encoded protein after oral administration. Vaccine 15, 814–817. 16. Luo, D., Woodrow-Mumford, K., Belcheva, N., and Saltzman, W. M. (1999) Controlled DNA delivery systems. Pharm. Res. 16, 1300–1308. 17. Ochiya, T., Takahama, Y., Nagahara, S., Sumita, Y., Hisada, A., Itoh, H., Nagai, Y., and Terada, M. (1999) New delivery system for plasmid DNA in vivo using atelocollagen as a carrier material: the Minipellet. Nat. Med. 5, 707–710.

206

Shea and Mooney

18. Mathiowitz, E., Jacob, J. S., Jong, Y. S., Carino, G. P., Chickering, D. E., Chaturvedi, P., et al. (1997) Biologically erodable microspheres as potential oral drug delivery systems. Nature 386, 410–414. 19. Shea, L. D., Smiley, E., Bonadio, J., and Mooney, D. J. (1999) DNA delivery from polymer matrices for tissue engineering. Nat. Biotechnol. 17, 551–554. 20. Bonadio, J., Smiley, E., Patil, P., and Goldstein, S. (1999) Localized, direct plasmid gene delivery in vivo: prolonged therapy results in reproducible tissue regeneration. Nat. Med. 5, 753–759. 21. Middaugh, C. R., Evans, R. K., Montgomery, D. L., and Casimiro, D. R. (1998) Analysis of plasmid DNA from a pharmaceutical perspective. J. Pharm. Sci. 87, 130–146. 22. Pachence, J. M. (1996) Collagen-based devices for soft tissue repair. J. Biomed. Mater. Res. 33, 35–40. 23. Schoen, F. J. and Levy, R. J. (1999) Tissue heart valves: current challenges and future research perspectives. J. Biomed. Mater. Res. 47, 439–465. 24. Wong, W. H. and Mooney, D. J. (1997) Synthesis and properties of biodegradable polymers used as synthetic matrices for tissue engineering, in Synthetic Biodegradable Polymer Scaffolds (Atala, A. and Mooney, D. J., eds.), Birkhauser, Boston, pp. 49–80. 25. Peppas, N. A. and Langer, R. (1994) New challenges in biomaterials. Science 263, 1715–1720. 26. Muggli, D. S., Burkoth, A. K., and Anseth, K. S. (1999) Crosslinked polyanhydrides for use in orthopedic applications: degradation behavior and mechanics. J. Biomed. Mater. Res. 46, 271–278. 27. Cleland, J. L., Powell, M. F., and Shire, S. J. (1993) The development of stable protein formulations: a close look at protein aggregation, deamidation, and oxidation [published erratum appears in Crit. Rev. Ther. Drug Carrier Syst. 1994;11:60]. Crit. Rev. Ther. Drug Carrier Syst. 10, 307–377. 28. Putney, S. D. and Burke, P. A. (1998) Improving protein therapeutics with sustained-release formulations [published erratum appears in Nat. Biotechnol. 1998 16:478]. Nat. Biotechnol. 16, 153–157. 29. Adami, R. C., Collard, W. T., Gupta, S. A., Kwok, K. Y., Bonadio, J., and Rice, K. G. (1998) Stability of peptide-condensed plasmid DNA formulations. J. Pharm. Sci. 87, 678–683. 30. Capan, Y., Woo, B. H., Gebrekidan, S., Ahmed, S., and DeLuca, P. P. (1999) Preparation and characterization of poly (D,L-lactide-co-glycolide) microspheres for controlled release of poly(L-lysine) complexed plasmid DNA. Pharm. Res. 16, 509–513. 31. Capan, Y., Woo, B. H., Gebrekidan, S., Ahmed, S., and DeLuca, P. P. (1999) Influence of formulation parameters on the characteristics of poly(D,L-lactideco-glycolide) microspheres containing poly( L -lysine) complexed plasmid DNA. J. Controlled Release 60, 279–286. 32. Ando, S., Putnam, D., Pack, D. W., and Langer, R. (1999) PLGA microspheres containing plasmid DNA: preservation of supercoiled DNA via cryopreparation and carbohydrate stabilization. J. Pharm. Sci. 88, 126–130.

Polymeric Delivery of DNA

207

33. Jong, Y. S., Jacob, J. S., Yip, K., Gardner, G., Seitelman, E., Whitney, M., Montgomery, S., and Mathiowitz, E. (1997) Controlled release of plasmid DNA. J. Controlled Release 47, 123–134. 34. Fang, J., Zhu, Y. Y., Smiley, E., Bonadio, J., Rouleau, J. P., Goldstein, S. A., et al. (1996) Stimulation of new bone formation by direct transfer of osteogenic plasmid genes. Proc. Natl. Acad. Sci. USA 93, 5753–5758. 35. Harris, L. D., Kim, B. S., and Mooney, D. J. (1998) Open pore biodegradable matrices formed with gas foaming. J. Biomed. Mater. Res. 42, 396–402.

Duplex and Triplex DNA Stabilization

209

17 Promotion of Duplex and Triplex DNA Formation by Polycation Comb-Type Copolymers Hidetaka Torigoe and Atsushi Maruyama 1. Introduction Triplex DNA has attracted considerable interest recently because of its possible biological functions in vivo and its wide variety of potential applications, such as regulation of gene expression, site-specific cleavage of duplex DNA, mapping of genomic DNA, and gene-targeted mutagenesis (1–3). A triplex is usually formed through the sequence-specific interaction of a single-stranded homopyrimidine or homopurine triplex-forming oligonucleotide (TFO) with the major groove of the homopurine–homopyrimidine stretch in duplex DNA (1–5). In the pyrimidine motif triplex, a homopyrimidine TFO binds parallel to the homopurine strand of the target duplex by Hoogsteen hydrogen bonding to form T•A:T and C+•G:C triplets (• and : represent Hoogsteen hydrogen bonding and Watson Crick base pairing, respectively). (1–5). Because the cytosine bases in a homopyrimidine TFO are to be protonated to bind with the guanine bases of the G:C duplex, the formation of the pyrimidine motif triplex needs an acidic pH condition, and is thus unstable at physiological pH (6–8). On the other hand, in the purine motif triplex, a homopurine TFO binds antiparallel to the homopurine strand of the target duplex by reverse Hoogsteen hydrogen bonding to form A•A:T (or T•A:T) and G•G:C triplets (1–5). Although the purine motif triplex is pH-independent, triplexes involving guanine-rich TFOs are inhibited by physiological concentrations of certain monovalent cations (M+), especially K+ (9,10). The instability of triplex under physiological conditions limits its use for artificial control of gene expression in vivo. Stabilization of the triplex under physiological conditions is therefore important in improving its therapeutic potential. Numerous efforts, such as the replacement of cytosine bases in a homopyrimidine TFO with 5–methylcytosine (7,11–13) or other chemically modified bases (14–18), the conjugation of different DNA intercalators to TFO From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

209

210

Torigoe and Maruyama

(19,20), and/or the use of polyamines, such as spermine or spermidine, as triplex stabilizers (21) have been made to improve the stability of the pyrimidine motif triplex at physiological pH. However, in some cases, modification strategies lessened the overall binding affinity of the TFO or increased its nonspecific interaction with DNA (2,18). On the other hand, a chemical modification strategy with guanine-rich TFOs has partially overcome the inhibitory effect of K+ on the purine motif triplex formation (22–27). Despite extensive efforts, significant stabilization of triplex under physiological conditions is, however, yet to be achieved. This chapter describes the promotion of triplex DNA formation under physiological conditions by polycationic comb-type copolymer (Fig. 1), consisting of a poly(L-lysine) (PLL) backbone and hydrophilic graft side chains of dextran (Dex), especially focusing on methods for studying triplex formation. The polycationic comb-type copolymer (Fig. 1) also has the ability to promote duplex DNA formation, which is covered in a recent paper by the authors (28). 2. Materials 2.1. Preparation of Oligonucleotides Complementary oligonucleotides for duplex DNA, and a homopyrimidine or homopurine TFO, were synthesized on an ABI DNA synthesizer using a solidphase cyanoethyl phosphoramidite method (29), and purified with reverse-phase high performance liquid chromatography on a Wakosil DNA column. 5'Biotinylated oligonucleotide was prepared from biotin phosphoramidite for kinetic analyses by resonant mirror measurement, described in Subheading 3.5. The concentration of oligonucleotides was determined by UV absorbance. The reported extinction coefficients for poly(dA) [¡257 = 8600/cm/(mol of base/L)] and poly (dT) [¡265 = 8700/cm/(mol of base/L)] were used for a homopurine TFO and a homopyrimidine TFO, respectively (30,31). Complementary strands for duplex DNA were annealed by heating at up to 90°C, followed by a gradual cooling to room temperature. When the removal of unpaired single strands was necessary, the annealed sample was applied on a hydroxyapatite column (Koken). The concentration of duplex DNA was determined by UV absorbance, considering DNA concentration ratio of 1 optical density = 50 µg/mL.

2.2. Preparation of Polycationic Comb-Type Poly(L-lysine)-graftDextran (PLL-g-Dex) Copolymer The copolymer, poly[(L-lysine)-graft-(_-1,6-D-glucopyranoside)] (PLL-g-Dex) (Fig. 1), was synthesized by a reductive amination reaction between PLL and dextran T-10, either in borate buffer or dimethylsulfoxide as a solvent for the reaction (28,32). The resulting graft copolymer, free from unreacted dextran, was isolated by gel permeation chromatography. Finally, the copolymer was lyophilized.

Duplex and Triplex DNA Stabilization

211

Fig. 1. Structural formula and schematic illustration of the PLL-g-Dex copolymer. Preparation and characterization of the copolymer has been described previously in detail (28,32). Degree of substitution (n) is 0.2. Number averaged polymerization degrees of PLL and Dex are 200 and 36.5 (m), respectively.

3. Methods 3.1. UV Melting of Triplex (32–34) Heating of triplex usually results in biphasic strand dissociation, according to the transitions between the three states: triplexA duplex + single strandA3 single strands. Base stacking interactions in the free strands are weaker than those in the bound strands, resulting in a hyperchromic increase in UV absorbance upon heating. UV melting monitors the process of triplex melting by the temperature-dependent change in UV absorbance. First-derivative plot of UV absorbance (dA/dT vs T) can be calculated from the UV melting curve (A vs T). Peak temperatures in the first derivative plot correspond to the melting temperature, Tm, of the transition. Figure 2 shows UV melting curves of a poly(dA)•2poly(dT) triplex (32). The triplex showed two-step melting in the absence of the copolymer. The first transition at lower temperature (37°C) was the melting of the triplex to a duplex and a single strand, and the second transition at higher temperature (72°C) was that of the duplex. In the presence of excess copolymers over DNAs, only one transition was observed at higher temperature (89°C). Because the magnitude in UV absorbance change (6A) at Tm in the presence of the copolymer was equal to the sum of those at Tm1 and Tm2 in the absence of the copolymer, the transition is indicated to be a direct melting of the triplex to its constituting single-stranded

212

Torigoe and Maruyama

Fig. 2. UV melting profile of poly(dA)•2poly(dT) triplex with or without PLL-gDex copolymer in 10 mM sodium phosphate buffer (pH 7.2) containing 150 mM NaCl and 0.1 mM EDTA. Reprinted with permission from ref. 32.

DNAs. The UV melting profile in the cooling process demonstrated reversibility of the transition even in the presence of the copolymer. These results indicate that the PLL-g-Dex copolymer thermally stabilized the triplex but did not disturb triplex formation from its constituting single-stranded DNAs.

3.2. Circular Dichroism Spectroscopy of Triplex (32,33) Circular dichroism spectroscopy is sensitive to interactions of adjacent bases that are vertically stacked in strands. Stacking interactions depend on the conformational details of nucleic acid structure. CD spectra suggest the overall conformation of strands in triplex. Formation of the poly(dA)•2poly(dT) triplex was evaluated by CD spectroscopy (32). As shown in Fig. 3, although the CD spectrum of the poly(dA)•poly(dT) duplex had a strong positive band near 220 nm, the poly(dA)•2poly(dT) triplex had no positive band near this wavelength. The mixture of poly(dA)•2poly(dT) and the copolymer showed almost the same signal as the triplex alone. The authors conclude that the PLL-g-Dex copolymer does not alter the conformation of the triplex.

3.3. Electrophoretic Mobility Shift Assay (EMSA) of Pyrimidine Motif Triplex DNA Formation (33,35,36) Triplexes migrate slower than duplexes in native polyacrylamide gel electrophoresis, so that the formation of triplexes retards gel migration of radiolabeled duplexes. The percentage of the formed triplex was calculated using the following equation: % triplex = [Striplex / (Striplex + Sduplex)] × 100

Duplex and Triplex DNA Stabilization

213

Fig. 3. CD spectra of poly(dA)•poly(dT) duplex (dotted line), poly(dA)•2poly(dT) triplex (broken line) and the 2:1 mixture of the PLL-g-Dex copolymer and the triplex (solid line) in the same buffer as described in Fig. 2. Reprinted with permission from ref. 32.

where Striplex and Sduplex represent the radioactive signal for triplex and duplex bands, respectively. The dissociation constant, Kd, of triplex formation is determined from the concentration of the TFO, which causes one-half of the target duplex to form triplex. Pyrimidine motif triplex formation between the 23-bp target duplex, Pur23A•Pyr23T (Fig. 4A), and its specific 15-mer homopyrimidine TFO, Pyr15T (Fig. 4A), was investigated in the absence or presence of the triplex stabilizer at pH 7.0 by EMSA (36). Figure 4B shows that the triplex was formed only at higher concentrations of Pyr15T, in the absence of the triplex stabilizer (None). The addition of 1.0 mM spermine (Spermine) with Pyr15T slightly increases (~2×) the binding affinity (see lane 5 for None and Spermine). The addition of 1.9 µM PLL-g-Dex copolymer (Polymer), along with Pyr15T, significantly increased the triplex formation. The copolymer, at 25× less charge ratio ([amino groups]stabilizer to [phosphate groups]DNA) than spermine showed higher stabilization effect. The Ka (= 1/Kd) of pyrimidine motif triplex formation in the presence of the PLL-g-Dex copolymer, was about 100× higher than that in the absence of the triplex stabilizer or even in the presence of spermine. The addition of 5 µM nonspecific oligonucleotides (Pyr15NS-2 [Fig. 4A] [see lane 1 for None] or Pyr15NS-1 [Fig. 4A] [data not shown]) failed to form a triplex, suggesting a sequence-specific interaction of Pyr15T with Pur23A•Pyr23T to form a triplex.

214

Torigoe and Maruyama

Fig. 4. (A) Oligonucleotide sequences of the target duplex, Pur23A•Pyr23T, the specific TFO, Pyr15T, and the nonspecific TFOs, Pyr15NS-1, and Pyr15NS-2. (B) EMSA of the pyrimidine motif triplex formation at neutral pH, either in the absence or presence of the triplex stabilizer. Reprinted with permission from ref. 36.

3.4. Thermodynamic Analyses of Pyrimidine Motif Triplex DNA Formation by Isothermal Titration Calorimetry (36–40) The isothermal titration calorimetry (ITC) relies on the accurate measurement of heat changes caused by the interaction of molecules in solution and possesses the advantage of not requiring labeling or immobilization of the components (37). ITC provides much thermodynamic information about the binding process, from only a single experiment. This information includes the binding stoichiometry (n), the binding equilibrium constant (Ka), the enthalpy change (6H) of binding, the entropy change (6S) of binding, and the Gibbs free-energy change (6G) of the binding process (37). A syringe containing a solution of one element (in the case of triplex formation, TFO) is incrementally titrated into a cell containing a solution of the second element (in the case of triplex formation, duplex) (36,38–40). As the TFO is added to the duplex, heat is released upon triplex formation. The

Duplex and Triplex DNA Stabilization

215

heat for each injection is measured by the ITC instrument and is plotted as a function of time over the injection series. The heat signal from each injection is determined by the area underneath the injection peak. The heat is plotted against the molar ratio of TFO added to the duplex in the cell. The titration plot provides the thermodynamic information of the triplex formation. The thermodynamics of the pyrimidine motif triplex formation between the 23-bp target duplex, Pur23A•Pyr23T (Fig. 4A), and its specific 15-mer homopyrimidine TFO, Pyr15T (Fig. 4A), were examined at 25°C and pH 6.8 by ITC under three different conditions: buffer A (10 mM sodium cacodylatecacodylic acid, at pH 6.8, containing 200 mM NaCl and 20 mM MgCl2); buffer A + 0.84 mM spermine, and buffer A + 0.038 mM PLL-g-Dex copolymer (36). The concentration of amino groups in 0.84 mM spermine is equivalent to that in 0.038 mM PLL-g-Dex copolymer. Figure 5A compares the ITC profiles of the initial three injections for triplex formation between Pyr15T and Pur23A•Pyr23T, at 25°C and pH 6.8 with TFO alone or in the presence of spermine or the PLL-g-Dex copolymer. The magnitudes of the exothermic peaks in the presence of 0.038 mM PLL-g-Dex copolymer were much larger than those observed with TFO alone. Triplex formation in the presence of 0.038 mM PLL-g-Dex copolymer reached to equilibrium within 10 min after each injection of Pyr15T. On the other hand, the magnitudes of the exothermic peaks in the presence of 0.84 mM spermine were indistinguishable from those observed with TFO alone. Figure 5B shows a 200-min ITC profile for triplex formation in the presence of the PLL-g-Dex copolymer. An exothermic heat pulse was observed after each injection of Pyr15T into Pur23A•Pyr23T. The magnitude of each peak decreased gradually with each new injection, and a small peak was still observed at a molar ratio of [Pyr15T]/[Pur23A•Pyr23T] = 2. The area of the small peak was equal to the heat of dilution measured in a separate experiment by injecting Pyr15T into the same buffer (data not shown). The area under each peak was integrated, and the heat of dilution of Pyr15T was subtracted from the integrated values. The corrected heat was divided by the number of moles injected, and the resulting values were plotted as a function of the molar ratio of [Pyr15T]/[Pur23A•Pyr23T], as shown in Fig. 5C. The resulting titration plot was fitted to a sigmoidal curve by a nonlinear least-squares method. Ka and 6H were obtained from the fitted curve (37); 6G and 6S were calculated from the equation 6G = –RTlnKa =6H–T6S (37). The titration plots, with TFO alone or in the presence of spermine are also shown in Fig. 5C. The thermodynamic parameters under these conditions were obtained from these plots in the same way. Table 1 summarizes the thermodynamic parameters of triplex formation at 25˚C and pH 6.8, obtained from ITC under the three different conditions. The signs of both 6H and 6S were negative under all reaction conditions. Because an observed

216

Torigoe and Maruyama

Fig. 5. Thermodynamic analyses of the pyrimidine motif triplex formation at neutral pH by ITC. (A) ITC profiles of the initial three injections for the triplex formation between Pyr15T and Pur23A•Pyr23T in the absence or presence of the triplex stabilizer (0.84 mM spermine or 0.038 mM PLL-g-Dex copolymer). The profile in the presence of spermine is indistinguishable from that observed in the absence of the triplex stabilizer. (B) Total ITC profile for the triplex formation between Pyr15T and Pur23A•Pyr23T in the presence of the PLL-g-Dex copolymer. (C) Titration plots in the absence or presence of the triplex stabilizer against the molar ratio of [Pyr15T]:[Pur23A•Pyr23T]. Data were fitted by a nonlinear least-squares method. Reprinted with permission from ref. 36.

negative 6S was unfavorable for triplex formation, triplex formation was driven by a large negative 6H under each condition. The magnitude of the 6H of pyrimidine motif triplex formation in the presence of 0.038 mM PLL-g-Dex copolymer was 2.5× larger than those observed with TFO alone or in the presence of 0.84 mM

Triplex stabilizer

217

None 0.84 mM spermine 0.038 mM PLL-g-Dex copolymer

Ka (M–1) (1.97 ± 0.43) × 105 (5.15 ± 0.29) × 105 (1.89 ± 0.32) × 107

Ka (relative) 1.0 2.6 95.9

Duplex and Triplex DNA Stabilization

Table 1 Thermodynamic Parameters for Triplex Formation Between 15-mer TFO, Pyr15T, and a 23-bp Duplex, Pur23A•Pyr23T, Obtained from ITC Measurements 6G (kcal/mol)

6H (kcal/mol)

6S (cal/mol/K)

–7.22 ± 0.15 –7.79 ± 0.03 –9.93 ± 0.11

–34.9 ± 2.2 –34.1 ± 3.2 –87.9 ± 2.3

–92.7 ± 8.0 –88.2 ± 10.8 –262 ± 8.1

Concentration of Pyr15T is 120 µM in the syringe. Pyr15T is injected 20 times in 5-µL increments into Pur23A•Pyr23T. Concentration of Pur23A•Pyr23T is 5 µM in the cell. The obtained values are the average of at least three ITC experiments carried out at at 25°C and pH 6.8 in 10 mM sodium cacodylate-cacodylic acid, 20 mM sodium chloride, and 20 mM magnesium chloride with or without triplex stabilizer. Reprinted with permission from ref. 36.

217

218

Torigoe and Maruyama

spermine, consistent with the ITC profiles in Fig. 5A. The Ka of pyrimidine motif triplex formation was increased 2.6 or 95.9× by the addition of 0.84 mM spermine or 0.038 mM PLL-g-Dex copolymer, respectively. Although the concentration of amino groups in 0.84 mM spermine and 0.038 mM PLL-g-Dex copolymer was equivalent, an increase in Ka by the addition of 0.038 mM PLL-g-Dex copolymer was higher than that by the addition of 0.84 mM spermine.

3.5. Kinetic Analyses of Pyrimidine Motif Triplex DNA Formation by Resonant Mirror Measurement (36,39–43) The resonant mirror is one example of a class of optical biosensors that can be used to determine the kinetic binding parameters of molecular interactions, such as association rate constant (kassoc), dissociation rate constant (kdissoc), and binding constant (Ka) (41,42). Resonant mirror measurements are usually performed with one partner immobilized (in the case of triplex formation, duplex) on a porous hydrogel to which the second component (in the case of triplex formation, TFO) then binds (36,39,40,43). Changes in the mass loading at the sensor surface on the triplex formation cause a shift in the resonance angle of light propagated through the wave-guiding structure immediately adjacent to the hydrogel. The time-dependence of the change in resonance angle yields the kinetic information of the triplex formation. The kinetic parameters for the association and dissociation of the 15-mer homopyrimidine TFO, Pyr15T (Fig. 4A), with the 23-bp target duplex, Pur23A•Pyr23T (Fig. 4A), were assessed at 25˚C and pH 6.8 by resonant mirror measurement under the three different conditions described in the ITC section (36). Figure 6A compares the resonant mirror sensorgrams representing the triplex formation and dissociation involving 2.0 µM of the specific (Pyr15T) or nonspecific (Pyr15NS-1) TFOs (Fig. 4A) and the immobilized biotinylated target duplex (Bt-Pyr23T•Pur23A) at 25˚C and pH 6.8, in either the absence or presence of the triplex stabilizer. The injection of Pyr15T alone (Pyr15T in no stabilizer) over the immobilized Bt-Pyr23T•Pur23A caused an increased response, and the injection of Pyr15T and spermine (Pyr15T in 0.84 mM spermine) slightly increased the response. However, the response was substantially changed when Pyr15T and the PLL-g-Dex copolymer were injected (Pyr15T in 0.038 mM copolymer). The negligible response observed when Pyr15NS-1 was injected with the PLL-gDex copolymer (Pyr15NS-1 in 0.038 mM copolymer) indicates that the sequence specificity of triplex formation was preserved in the presence of the copolymer. Taken together, it unambiguously indicates that the PLL-g-Dex copolymer significantly increased the association rate constant of triplex formation, and its triplex-promoting efficacy was higher than that of spermine. The authors have measured a series of association and dissociation curves at various concentrations of Pyr15T in the presence of the PLL-g-Dex copolymer

Duplex and Triplex DNA Stabilization

219

Fig. 6. Kinetic analyses of the pyrimidine motif triplex formation at neutral pH by resonant mirror measurement. (A) Typical resonant mirror sensorgrams for the triplex formation. (B) Series of resonant mirror sensorgrams for the triplex formation. (C) Measured on-rate constants, kon, of the triplex formation in (B), were plotted against the respective concentrations of Pyr15T. The plot was fitted to a straight line (r2 = 0.98) by a linear least-squares method. Reprinted with permission from ref. 36.

to obtain the kinetic parameters. An increase in the concentration of Pyr15T led to a gradual change in the response of the association curves as shown in Fig. 6B. The on-rate constant (kon) was obtained from the analysis of each association curve. Figure 6C shows a plot of kon against the concentrations of

220 Table 2 Kinetic Parameters for Triplex Formation Between 15-mer TFO, Pyr15T, and a 23-bp Duplex, Pur23A•Pyr23T with or Without Triplex Stabilizer, Obtained from IAsys Measurements Triplex stabilizer

kassoc (M–1s–1)

None (6.31 ± 0.18) × 102 0.84 mM spermine (2.29 ± 0.22) × 103 0.038 mM PLL-g-Dex copolymer (2.88 ± 0.23) × 104

kassoc (relative) 1.0 3.6 45.6

kassoc (s–1) (1.17 ± 0.14) × 10-2 (1.06 ± 0.13) × 10-2 (7.8 ± 3.4) × 10-3

kdissoc (relative) 1.0 0.91 0.66

Ka (M–1) (5.41 ± 0.91) × 104 (2.15 ± 0.54) × 105 (3.69 ± 1.32) × 106

Ka (relative) 1.0 4.0 68.2

Torigoe and Maruyama

Several different concentrations of Pyr15T are injected over the immobilized Bt-Pyr23T•Pur23A at 25°C and pH 6.8 in 10 mM sodium cacodylatecacodylic acid, 200 mM sodium chloride, and 20 mM magnesium chloride. Reprinted with permission from ref. 36.

Duplex and Triplex DNA Stabilization

221

Pyr15T. The resultant plot was fitted to a straight line (r2 = 0.98) by a linear least-squares method. The association rate constant (kassoc) was determined from the slope of the fitted line (41,42). On the other hand, the off-rate constant (koff) was obtained from the analysis of each dissociation curve (Fig. 6A, and data not shown). Because koff is usually independent of the concentration of the injected solution, the dissociation rate constant (kdissoc) was determined by averaging koff for several concentrations (41,42). Ka was calculated from the equation, Ka = kassoc/kdissoc (41,42). The kinetic parameters with TFO alone and in the presence of spermine were obtained in the same way. Table 2 summarizes the kinetic parameters of triplex formation at 25˚C and pH 6.8, obtained from resonant mirror measurement under the three different conditions. The kassoc of pyrimidine motif triplex formation increased 3.6 or 45.6¥ by the addition of 0.84 mM spermine or 0.038 mM PLL-g-Dex copolymer, respectively. The authors have found a more significant increase in kassoc by adding 0.038 mM PLL-g-Dex copolymer, rather than 0.84 mM spermine, although the concentration of their amino groups is almost equal. In contrast, when the kdissoc of pyrimidine motif triplex formation was compared, a 1.1 or 1.5¥ lower kdissoc was obtained by the addition of 0.84 mM spermine or 0.038 mM PLL-g-Dex copolymer, respectively. Thus, the much larger Ka, in the presence of the PLL-g-Dex copolymer, resulted mainly from the increase in kassoc, rather than the decrease in kdissoc.

3.6. Conclusion The authors have shown that PLL-g-Dex copolymer (Fig. 1) not only significantly increases the thermal stability of triplex, but also promotes triplex formation under physiological conditions, especially focusing on methods for studying triplex formation. In addition to methods described here, DNase I footprinting is useful in revealing DNA sequences where a triplex is formed. Infrared spectroscopy and nuclear magnetic resonance are sensitive to the higher-order structure of a triplex. The ability of the PLL-g-Dex copolymer to promote triplex formation under physiological conditions supports further progress in therapeutic applications of the antigene strategy in vivo. References 1. Plum, G. E., Pilch, D. S., Singleton, S. F., and Breslauer, K. J. (1995) Nucleic acid hybridization: triplex stability and energetics. Annu. Rev. Biophys. Biomol. Struct. 24, 319–350. 2. Frank-Kamenetskii, M. D. and Mirkin, S. M. (1995) Triplex DNA structures. Annu. Rev. Biochem. 64, 65–95. 3. Soyfer, V. N. and Potaman, V. N. (1996) Triple-helical Nucleic Acids. SpringerVerlag, New York. 4. Sun, J.-S. and Helene, C. (1993) Oligonucleotide-directed triple-helix formation. Curr. Opin. Struct. Biol. 3, 345–356.

222

Torigoe and Maruyama

5. Sun, J.-S., Garestier, T., and Helene, C. (1996) Oligonucleotide directed triple helix formation. Curr. Opin. Struct. Biol. 6, 327–333. 6. Frank-Kamenetskii, M. D. (1992) Protonated DNA structures. Methods Enzymol. 211, 180–191. 7. Singleton, S. F. and Dervan, P. B. (1992) Influence of pH on the equilibrium association constants for oligodeoxyribonucleotide-directed triple helix formation at single DNA sites. Biochemistry 31, 10,995–11,003. 8. Shindo, H., Torigoe, H., and Sarai, A. (1993) Thermodynamic and kinetic studies of DNA triplex formation of an oligohomopyrimidine and a matched duplex by filter binding assay. Biochemistry 32, 8963–8969. 9. Milligan, J. F., Krawczyk, S. H., Wadwani, S., and Matteucci, M. D. (1993) An antiparallel triple helix motif with oligodeoxynucleotides containing 2'deoxyguanosine, and 7-deaza-2'-deoxyxanthosine. Nucleic Acids Res. 21, 327–333. 10. Cheng, A.-J. and Van Dyke, M. W. (1993) Monovalent cation effects on intermolecular purine-purine-pyrimidine triple-helix formation. Nucleic Acids Res. 21, 5630–5635. 11. Lee, J. S., Woodsworth, M. L., Latimer, L. J. P., and Morgan, A. R. (1984) Poly(pyrimidine)•poly(purine) synthetic DNAs containing 5-methylcytosine form stable triplexes at neutral pH. Nucleic Acids Res. 12, 6603–6614. 12. Povsic, T. J. and Dervan, P. B. (1989) Triple helix formation by oligonucleotides on DNA extended to the physiological pH range. J. Am. Chem. Soc. 111, 3059–3061. 13. Xodo, L. E., Manzini, G., Quadrifoglio, F., van der Marel, G. A., and van Boom, J. H. (1991) Effect of 5-methylcytosine on the stability of triple-stranded DNA: a thermodynamic study. Nucleic Acids Res. 19, 5625–5631. 14. Ono, A., Ts’o, P. O. P., and Kan, L.-S. (1991) Triplex formation of oligonucleotides containing 2'-O-methyl-pseudoisocytidine in substitution for 2'deoxycytidine. J. Am. Chem. Soc. 113, 4032–4033. 15. Krawczyk, S. H., Milligan, J. F., Wadwani, S., Moulds, C., Froehler, B. C., and Matteucci, M. D. (1992) Oligonucleotide-mediated triple helix formation using an N3-protonated deoxycytidine analog exhibiting pH-independent binding within the physiological range. Proc. Natl. Acad. Sci. USA 89, 3761–3764. 16. Koh, J. S. and Dervan, P. B. (1992) Design of a nonnatural deoxyribonucleoside for recognition of GC base pairs by oligonucleotide-directed triple helix formation. J. Am. Chem. Soc. 114, 1470–1478. 17. Jetter, M. C. and Hobbs, F. W. (1993) 7,8-dihydro-8-oxoadenine as a replacement for cytosine in the third strand of triple helices. Triplex formation without hypochromicity. Biochemistry 32, 3249–3254. 18. Ueno, Y., Mikawa, M., and Matsuda, A. (1998) Nucleosides and nucleotides 170. Synthesis and properties of oligodeoxynucleotides containing 5-[N-[2[N, N-bis(2-aminoethyl)–amino]ethyl]carbamoyl]-2'-deoxyuridine and 5-[N-[3[N, N-bis(3-aminopropyl)-amino]propyl]carbamoyl]-2'-deoxyuridine. Bioconjugate Chem. 9, 33–39.

Duplex and Triplex DNA Stabilization

223

19. Sun, J. S., Giovannangeli, C., Francois, J. C., Kurfurst, R., Montenay-Garestier, T., Asseline, U., et al. (1991) Triple-helix formation by _ oligodeoxynucleotides, and _ oligodeoxynucleotide-intercalator conjugates. Proc. Natl. Acad. Sci. USA 88, 6023–6027. 20. Mouscadet, J.-F., Ketterle, C., Goulaouic, H., Carteau, S., Subra, F., Le Bret, M., and Auclair, C. (1994) Triple helix formation with short oligonucleotideintercalator conjugates matching the HIV-1 U3 LTR end sequence. Biochemistry 33, 4187–4196. 21. Hampel, K. J., Crosson, P., and Lee, J. S. (1991) Polyamines favor DNA triplex formation at neutral pH. Biochemistry 30, 4455–4459. 22. Olivas, W. M. and Maher, L. J., III (1995) Overcoming potassium-mediated triplex inhibition. Nucleic Acids Res. 23, 1936–1941. 23. Gee, J. E., Revankar, G. R., Rao, T. S., and Hogan, M. E. (1995) Triplex formation at the rat neu gene utilizing imidazole and 2'-deoxy-6-thioguanosine base substitutions. Biochemistry 34, 2042–2048. 24. Vasquez, K. M., Wensel, T. G., Hogan, M. E., and Wilson, J. H. (1995) Highaffinity triple helix formation by synthetic oligonucleotides at a site within a selectable mammalian gene. Biochemistry 34, 7243–7251. 25. Dagle, J. M. and Weeks, D. L. (1996) Positively charged oligonucleotides overcome potassium-mediated inhibition of triplex DNA formation. Nucleic Acids Res. 24, 2143–2149. 26. Faruqi, A. F., Krawczyk, S. H., Matteucci, M. D., and Glazer, P. M. (1997) Potassium-resistant triple helix formation and improved intracellular gene targeting by oligodeoxyribonucleotides containing 7-deazaxanthine. Nucleic Acids Res. 25, 633–640. 27. Joseph, J., Kandala, J. C., Veerapanane, D., Weber, K. T., and Guntaka, R. V. (1997) Antiparallel polypurine phosphorothioate oligonucleotides form stable triplexes with the rat _1(I) collagen gene promoter and inhibit transcription in cultured rat fibroblasts. Nucleic Acids Res. 25, 2182–2188. 28. Maruyama, A., Watanabe, H., Ferdous, A., Katoh, M., Ishihara, T., and Akaike, T. (1998) Characterization of interpolyelectrolyte complexes between doublestranded DNA and polylysine comb-type copolymers having hydrophilic side chains. Bioconjugate Chem. 9, 292–299. 29. Sinha, N. D., Biernat, J., and Koster, H. (1983) Beta-cyanoethyl N, N-dialkylamino/ N-morpholinomonochlorophosphoamidites new phosphitylating agents facilitating ease of deprotection and work-up of synthesized oligonucleotides. Tetrahedron Lett. 24, 5843–5846. 30. Chamberlin, M. J. (1965) Comparative properties of DNA, RNA and hybrid homopolymer pairs. Proc. Fed. Am. Soc. Exp. Biol. 24, 1446–1457. 31. Riley, M., Mailing, B., and Chamberlin, M. J. (1966) Physical and chemical characterization of two-, and three-stranded adenine-thymine, and adenine-uracil homopolymer complexes. J. Mol. Biol. 20, 359–389.

224

Torigoe and Maruyama

32. Maruyama, A., Katoh, M., Ishihara, T., and Akaike, T. (1997) Comb-type polycations effectively stabilize DNA triplex. Bioconjugate Chem. 8, 3–6. 33. Ferdous, A., Watanabe, H., Akaike, T., and Maruyama, A. (1998) Comb-type copolymer: stabilization of triplex DNA and possible application in antigene strategy. J. Pharm. Sci. 87, 1400–1405. 34. Maruyama, A., Ohnishi, Y., Watanabe, H., Torigoe, H., Ferdous, A., and Akaike, T. (1999) Polycation comb-type copolymer reduces counterion condensation effect to stabilize DNA duplex and triplex formation. Colloids Surfaces B 16, 273–280. 35. Ferdous, A., Watanabe, H., Akaike, T., and Maruyama, A. (1998) Poly (L-lysine)graft-dextran copolymer: amazing effects on triplex stabilization under physiological pH and ionic conditions (in vitro). Nucleic Acids Res. 26, 3949–3954. 36. Torigoe, H, Ferdous, A., Watanabe, H., Akaike, T., and Maruyama, A. (1999) Poly (L-lysine)-graft-dextran copolymer promotes pyrimidine-motif triplex DNA formation at physiological pH: thermodynamic and kinetic studies. J. Biol. Chem. 274, 6161–6167. 37. Wiseman, T., Williston, S., Brandts, J. F., and Lin, L.-N. (1989) Rapid measurement of binding constants and heats of binding using a new titration calorimeter. Anal. Biochem. 179, 131–137. 38. Kamiya, M., Torigoe, H., Shindo, H., and Sarai, A. (1996) Temperature dependence and sequence specificity of DNA triplex formation: an analysis using isothermal titration calorimetry. J. Am. Chem. Soc. 118, 4532–4538. 39. Torigoe, H., Shimizume, R., Sarai, A., and Shindo, H. (1999) Triplex formation of chemically modified homopyrimidine oligonucleotides: thermodynamic and kinetic studies. Biochemistry 38, 14,653–14,659. 40. Torigoe, H., Ferdous, A., Watanabe, H., Akaike, T., and Maruyama, A. (1999) Poly (L-lysine)-graft-dextran copolymer remarkably promotes pyrimidine-motif triplex formation at neutral pH: thermodynamic and kinetic studies. Nucleosides Nucleotides 18, 1655–1656. 41. Cush, R., Cronin, J. M., Stewart, W. J., Maule, C. H., Molloy, J., and Goddard, N. J. (1993) Resonant mirror: a novel optical biosensor for direct sensing of biomolecular interactions. Part I: Principle of operation and associated instrumentation. Biosens. Bioelectron. 8, 347–353. 42. Edwards, P. R., Gill, A., Pollard-Knight, D. V., Hoare, M., Buckle, P. E., Lowe, P. A., and Leatherbarrow, R. J. (1995) Kinetics of protein-protein interactions at the surface of an optical biosensor. Anal. Biochem. 231, 210–217. 43. Bates, P. J., Dosanjh, H. S., Kumar, S., Jenkins, T. C., Laughton, C. A., and Neidle, S. (1995) Detection and kinetic studies of triplex formation by oligodeoxynucleotides using real-time biomolecular interaction analysis (BIA). Nucleic Acids Res. 23, 3627–3632.

Lyophilization of Gene Delivery Systems

225

18 Lyophilization of Nonviral Gene Delivery Systems S. Dean Allison and Thomas J. Anchordoquy 1. Introduction This chapter presents a qualitative description of the freeze-drying process as it pertains to the development of stable, dry polycation–DNA complex formulations. It is not intended to be a comprehensive treatise on freeze-drying. Readers are referred to a series of excellent papers by Pikal (1–5) for more detailed, quantitative explanations of the freeze-drying process. There is considerable interplay between the physical effects of process parameters and properties of formulation constituents, which mandates that both formulation and drying process be developed together. Although a freeze-dried formulation may offer the potential for long-term stability at room temperature, the lyophilization process itself is damaging to complexes (6–12). Loss of activity resulting from in-process damage can be prevented with the addition of appropriate additives, such as sucrose. Sucrose is known to preserve the structure and activity of isolated biomolecules, such as proteins and liposomes, in frozen and dried states (13–17). The mechanisms by which additives protect polycation–DNA complexes from freezing and drying stresses are not clear, but appear to be unlike those of simpler systems. The first nonviral gene delivery vectors were cationic liposomes (18), which continue to be used clinically, despite low in vivo transfection efficiency compared to viruses. Recent attempts to improve transfection efficiency have resulted in the development of complexes that include multiple components designed to overcome barriers to transfection, such as cationic polymers (19–27) and nuclear localization peptide sequences (28), in addition to cationic and neutral lipids. The added complexity is likely to further increase the difficulty of developing stable, dry formulations. Nonviral DNA vectors have several advantages compared to viral vectors that make their development worthwhile. Nonviral vectors do not elicit a From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

225

226

Allison and Anchordoquy

specific immune response, and can therefore be administered repeatedly (29); viral vectors have been implicated in at least one death, causing the suspension of clinical trials (30). Recombination events leading to production of a replicating virus are also a remote danger (31–33). In addition, the size of the plasmid to be delivered is restricted by the size of the viral capsid (31–33), which is not a limitation with nonviral vectors. The major disadvantage in using nonviral DNA delivery vectors is low transfection efficiency compared to their viral counterparts, which has resulted in low levels of usage in clinical trials. As of September 1999, only about 20% of gene therapy clinical trials used nonviral delivery systems. Consequently, the vast majority of research in nonviral gene therapy has been to develop more efficient vectors. Although this narrow focus may be justified by clinical results, critical issues surrounding the reproducible preparation and stability of nonviral delivery vehicles have been mostly ignored. Clearly, vectors possessing consistent physical and chemical properties will need to be manufactured in large batches, shipped, and stored in order to become a marketable pharmaceutical product. Production of stable, reproducible batches of polycation–DNA complexes is a major challenge. Typically, separate solutions of cationic agents and DNA are mixed together, and complexation occurs because of electrostatic charge minimization. However, steric and/or kinetic barriers prevent the homogeneous complexation of large polyelectrolytes. The resulting suspension contains a heterogeneous population of particles possessing a wide range of ± charge ratios and sizes (34,35). For example, studies on cationic liposome–DNA preparations have demonstrated that free (uncomplexed) liposomes and DNA coexist within a single suspension (36,37). Furthermore, the biophysical characteristics and efficacy of nonviral vectors are dependent on many factors including mixing protocol, buffer composition, and time allowed for complex formation (38–40). Controlled mixing methods that may be amenable to large-scale production are being explored to facilitate the manufacture of more reproducible populations (41). Since the biophysical characteristics of the most active complexes within a population are not presently known (42), the strategy has been to preserve the measurable physical characteristics of maximally active populations, such as mean particle diameter, c potential, and so on. Considering that complex formation results from electrostatic interactions between the negatively charged DNA and cationic agents, it is inevitable that some charges on these large structures are not completely neutralized by counterions. As a result, complexes possess local regions of charge imbalance that promote interactions between complexes and lead to aggregation (34,43,44). Van der Waals interactions are a major contributor to aggregation of electrostatically neutral complexes (45,46). Furthermore, aggregation in aqueous suspensions of both polymer– and liposome–DNA complexes correlates with loss of transfection activity.

Lyophilization of Gene Delivery Systems

227

The tendency to aggregate is further exacerbated in the highly concentrated suspensions prepared for clinical studies (41). Early clinical trials employing nonviral vectors circumvented the instability of lipid–DNA suspensions by preparing complexes at the bedside, immediately prior to administration (29,47,48). Although separate solutions of the different vector components are more stable during storage than preformed complexes, the potential exists for great variability in the population of complexes formed in individual preparations. Similar variability in the efficacy of the treatment would also be anticipated, and therefore this strategy is not acceptable. The physical and chemical properties of complexes relevant to biological activity must be maintained during storage and distribution. Although aqueous suspensions have been shown to possess storage stability, on the order of months, under controlled, refrigerated conditions, this level of stability falls short of the 2-yr benchmark sought for pharmaceutical products. In addition, agitation, which is virtually unavoidable during shipping, damages complexes, further reducing the practicality of aqueous formulations (12,49). Although aggregation is a significant problem, structural alterations within complexes, which do not increase particle size, can also affect biological activity (10,49). In addition to changes in their physical characteristics, complexes must be formulated to resist chemical degradation, i.e., attack by free radicals and/or singlet oxygen (50–57). Stability for 1 yr has been demonstrated for a frozen formulation. However, stability depends in part on the temperature to which the complexes are cooled (41). Maintenance of sufficiently low subzero temperatures for long periods is expensive and logistically difficult in some parts of the world. A single thawing or warming event can ruin a frozen product (7,12,41,49,58). Dried formulations offer the potential for long-term stability at ambient temperatures. A variety of methods may possibly be used to dry complexes, including air-drying, spray-drying, and lyophilization. The air-drying method with which the authors have experimented simply involves direct evaporation of solvent water under a stream of nitrogen gas, under controlled temperatures. After the suspension forms a thin film on the walls of the sample tube, a brief period of vacuum-drying is employed to facilitate water removal. This method of direct evaporation has been shown to stabilize proteins in sugar glasses (59,60). Initial attempts to dry cationic lipid-DNA complexes using direct evaporation have been unsuccessful. Although it is possible to spray dry cationic lipid-DNA complexes (US Patent no. 5,994,314), the relative processing efficiency of this method, compared to lyophilization, is questionable. Uncomplexed DNA is rapidly degraded by the high shear forces encountered during spray-drying (61,62). However, DNA complexed with lipid is more resistant to aerosolization-induced strand breakage than naked DNA (63). High shear forces encountered during aerosolization may promote unacceptable levels of aggregation (64).

228

Allison and Anchordoquy

Freeze-drying, or lyophilization, may be the least-destructive way to generate dehydrated polycation–DNA complex formulations, while maintaining the biophysical characteristics of the freshly made, optimized complex suspension. Moisture content of formulations can be reduced to less than 1% (w/w) without exceeding 30°C (10). Also, there is less chance of particulate contamination during freeze-drying, compared to spray-drying or direct evaporation (5). In addition, headspace gas of defined composition and pressure may be added to lyophilized vials prior to sealing. This feature may help reduce oxidative damage to dried preparations (52,54,57). A typical freeze-dryer consists of a sample drying chamber with temperaturecontrolled shelves connected to a condenser with refrigerated elements capable of maintaining low (~–70°C) temperatures. A vacuum pump is connected to the system to reduce chamber pressure, and one of several possible mechanisms is included to control pressure during drying. In order to remove water from frozen samples, pressures as low as approx 50 mTorr must be maintained. The freezedrying cycle converts much of the water in a sample to ice during freezing, leaving solutes, which still contain appreciable amounts of water (typically, 15–30% [5]), in the unfrozen fraction. Ice is removed at low temperature and pressure by sublimation during primary drying. Additional heat must then be added to the sample by increasing shelf temperature to desorb unfrozen water from the freezeconcentrated solutes, which is termed “secondary drying.” The goal of process optimization is to minimize lyophilization cycle times by maximizing drying temperatures and pressures, without impacting product quality. A necessary, but not sufficient, condition, for the successful lyophilization of macromolecules, is to generate a highly viscous, glassy matrix containing the sensitive molecule(s) and additives during the freezing step, and maintain that glass during subsequent drying. This process is also referred to as “vitrification” (59,65,66), which prevents translational mobility that fosters aggregation. An added claim is that intramolecular mobility, leading to first-order degradation mechanisms, is reduced. Different potentially protective additives have widely varying glass-forming properties (67,68). Therefore, precise control of sample temperature during freeze-drying is essential to the stabilization of biomolecules, such as proteins and liposomes. Hence, common manifold freeze-dryers are not adequate for formulation development of freezing- and dehydration-sensitive molecules. For cationic lipid–DNA complexes, vitrification appears not to be necessary during the lyophilization process, which has a time-scale on the order of days (10). However, vitrification may affect the long-term storage stability (on the order of years) of complexes (7,8). Freeze-drying may be the best way to stabilize DNA complexes for longterm storage. However, complexes must survive the freezing and drying stresses encountered during processing, before storage stability can be

Lyophilization of Gene Delivery Systems

229

addressed. The contribution of freezing and drying stresses to lyophilizationinduced damage to nonviral vectors may be assessed separately by comparing the results of freeze-thaw experiments with those of freeze-drying–rehydration experiments. Additional damage observed after freeze–drying, if any, is taken to result from the effects of dehydration–rehydration on complexes. Unprotected complexes can survive freeze–thawing when cooling rates are sufficiently high and freezing temperatures are sufficiently low (41). This observation suggests that process parameters, such as freezing rate, may be chosen to minimize damage to complexes. However, even the best equipment has practical limitations such as heat transfer rates, that restrict the choice of process parameters. In contrast to rapid freeze-thawing, the transfection activity of both cationic lipid–DNA complexes and polymer–DNA complexes is lost after lyophilization and rehydration without additives (7,9,10). The addition of stabilizing additives is therefore necessary, but the choice of additive is restricted both by regulatory approval and route of delivery. The remainder of this chapter is divided into subheadings focusing on each step in the lyophilization cycle: freezing, primary, and secondary drying. Data is reviewed pertaining to the impact of freezing and dehydration stresses on the physical characteristics and transfection activity of lipid– and polymer–DNA complexes. Hypotheses are presented to account for the mechanisms by which stabilizers protect complexes during freezing and drying. And finally, the authors discuss how parameters chosen for the lyophilization cycle affect the protective capacity of stabilizing additives. The aim is to illustrate the interrelated nature of the effect of drying cycle and additives on the stability of polycation–DNA complexes during the drying process, and the implications for long-term storage stability. 2. Freezing 2.1. Freezing Process Freezing constitutes the process by which most of the dehydration of polycation– DNA complexes takes place. The freezing process includes some degree of supercooling, followed by the sudden crystallization of water, with the temperature of the sample increasing toward that of the equilibrium crystallization temperature (Fig. 1). Solutes become highly concentrated in the unfrozen phase. As the heat of fusion is removed from the sample, and the temperature progressively decreases, ice continues to form. When the sample temperature decreases to the glass transition temperature, Tg', the rate of ice formation becomes negligible, because of the high viscosity of the glass. At the Tg', the nonfrozen product has a defined composition, i.e., all solutes will be maximally freeze-concentrated. Thus, a solution containing 1% total solids will have the same freeze-concentrated composition as the same mixture containing 10% solids prior to freezing. However, the volume of the freeze-concentrate would be increased 10-fold.

230

Allison and Anchordoquy

Fig. 1. Freezing profile of a typical lyophilization run showing the time delay in achieving shelf and sample temperature and the difference between shelf temperature and sample temperature at steady state.

The glass transition is a general property of liquids. It marks the boundary between amorphous (noncrystalline) solid-like and fluid-like behavior. During cooling, a pure liquid can be vitrified by cooling to a characteristic temperature at a rate sufficiently high to avoid crystallization. As a result, the disordered structure of the liquid state is immobilized (66,69). Glass transitions, which are characterized by a sudden increase in heat capacity during heating, are commonly measured using differential scanning calorimetry (inset, Fig. 2). Changes in electrical and mechanical properties of frozen samples, which accompany the onset of translational mobility with temperature can also be used (67,70). The significance of the Tg' is that it differentiates temperature regimes in which diffusion of vector particles is likely to be sufficient to allow collisions leading to aggregate formation. Thus, the goal of freezing is to reduce vector diffusion, thereby minimizing the potential for aggregation.

2.2. Effects of Freezing on Complexes Freezing dramatically increases the concentration of solutes. Cationic lipid– DNA complexes are especially susceptible to freeze-concentration-induced aggregation, possibly mediated by liposome fusion, in addition to colloidal mechanisms involving electrostatic and/or van der Waals interactions. The fact that vector populations are heterogeneous dictates that some aggregation will

Lyophilization of Gene Delivery Systems

231

Fig. 2. DSC thermogram showing separate glass transitions (arrows) for a frozen DMRIE-C–DNA complex–HES mixture. Vertical bar equals 0.1 W/g. Inset: A closeup of the glass transition of the cationic lipid–DNA complexes at approx –42°C. Tg' of the complex is increased from its normal value (–47°C) because of partial mixing with HES. Vertical bar in the inset equals 0.01 W/g.

occur via electrostatic interactions, and some will occur between electrically neutral particles. Although freeze-concentration favors particle collisions leading to aggregation, reducing sample temperature decreases vector diffusion, and thereby reduces the potential for aggregation. In general, for a secondorder process, the increase in potential for aggregation resulting from an 80-fold freeze-concentration outweighs the reduction in aggregation potential caused by a decrease in temperature from the equilibrium freezing point to Tg', by more than an order of magnitude (5). Consequences of freeze-concentration of excipients vary, depending on the solute. Some solutes, such as sugars and large polymers, tend to remain amorphous during freezing. Smaller-molecular-weight solutes (e.g., mannitol) and many salts tend to crystallize as a result of freeze-concentration. The degree of crystallization of mannitol is dependent on freezing rate (71,72). Freezing rates of 2.5°C/min, typical of shelf lyophilizers, do not result in the formation of amorphous mannitol. As a consequence, mannitol does not prevent freezinginduced aggregation of cationic lipid–DNA complexes (Allison and Anchordoquy, unpublished), and after thawing, transfection activity is reduced compared to fresh complexes. Mannitol crystallization also depends on formulation conditions. Crystallization of mannitol and other solutes can be

232

Allison and Anchordoquy

prevented by the presence of other solutes, e.g., glycine does not crystallize during freeze-drying if sucrose is present at a weight fraction of 0.25 (sucrose/ glycine) or greater (73). Although polycation–DNA complexes are typically manufactured in low-salt conditions, with nonelectrolyte compounds such as dextrose or sucrose added both to achieve isotonicity and stabilize the formulation, buffer salts, if used, can crystallize, resulting in dramatic pH shifts during freezing (74). Trace contaminants in excipients, such as metals, are also concentrated many-fold, which can increase the potential for oxidative damage to DNA (75). Recent studies on supercoiled plasmid DNA complexed with cationic liposomes have shown that transfection rates are maintained during a rapid cooling protocol, but not during slow cooling (41,49). Although DNA structure was not monitored during these studies, extensive strand breakage caused by cryolysis (76) would probably have resulted in reduced transfection rates. Possibly, DNA is protected by complexation with cationic agents, and that the level of strand breakage during freeze-thawing is too low to affect transfection rates, which is consistent with reports that DNA complexed with lipid is more resistant to structural damage resulting from aerosolization than naked DNA (63,77). Previous studies have also shown that DNA bound to proteins (e.g., in chromatin) is less susceptible to chemical degradation (50,78–80). Therefore, it seems more likely that reduced transfection rates after slow freeze-thawing result from structural alterations within the complexes (e.g., perturbation of the interaction between DNA and cationic agent) and/or aggregation, rather than chemical damage. In fact, several studies have shown that complex size increases dramatically after freeze-thawing in the absence of excipients, suggesting that aggregation is promoted during this stress (7,8,41,49). The increase in complex size correlates with the reductions in transfection rates (7,8,49). Interactions between polycation and DNA that affect transfection activity, but which are not detected as changes in gross physical properties of complex suspensions (e.g., mean particle size or c potential), can be altered by freezing. Freezing complexes with disaccharides can increase their in vitro transfection activity compared to freshly prepared complexes (10,49). One hypothesis that may explain this observation is that particle density increases after freeze-thawing with sugars, and transfection activity increases because vectors settle more efficiently onto tissue culture cells (20). However, the authors’ unpublished experiments show that 1,2-dimyristyloxypropyl-3-dimethyl-hydroxyethyl ammonium bromide:cholesterol (DMRIE-C)–DNA complexes, frozen and thawed with sucrose, have greater transfection activity than fresh complexes under conditions in which suspended cells and complexes are gently shaken during transfection, and therefore sedimentation (particle density) is not a factor. Two membrane-destabilizing processes occur during freezing: the liquid crystal to gel phase transition of lipids, and osmotic stress across the membrane,

Lyophilization of Gene Delivery Systems

233

caused by freeze concentration of solutes (81,82). Thus, physical stresses encountered during freezing, which destabilize bilayer structure, may alter the interaction between cationic lipids and DNA phosphate groups, possibly leading to an increase in the number of active particles in the population.

2.3. Mechanisms of Protection During Freezing 2.3.1. Preferential Exclusion Hypothesis Biomolecules, whose activity depends on higher-order structure (e.g., proteins), are protected during freezing by cryoprotective solutes that are preferentially excluded from the surface of the macromolecule (83,84). Preferential exclusion, by definition, results in a layer of solvent with a cryoprotectant concentration lower than that of the bulk solution being formed at the surface of the macromolecule. This situation, i.e., preferential exclusion of the cryoprotectant, is thermodynamically unfavorable from an entropic standpoint. Stresses that denature proteins (e.g., freezing, heating) cause an increase in the surface area of the macromolecule exposed to the solvent. This results in an increase in the excluded volume, which further increases the chemical potential of the system. Thus, the increase in free energy serves to counteract the denaturing force and to stabilize the native structure of the macromolecule (83,84). Solutes such as sucrose, which protect proteins in solution by preferential exclusion, also protect lipid- and polymer-based vectors (7,8,10,49). However, it is questionable whether the preferential exclusion mechanism applies to the stabilization of nonviral vectors, since their activity does not appear to be dependent on higher-order structure in the same way that it is for proteins. Furthermore, polycation–DNA complexes form a suspension that is phase-separated from aqueous cryoprotectant solutions. In addition to reducing the potential for interaction between complexes and excipients, phase-separation from the solution limits the validity of a mechanism based on solution thermodynamics (i.e., preferential exclusion).

2.3.2. Vitrification Hypothesis Franks (59) and Levine and Slade (65,66) have proposed that the structure and biological activity of macromolecules is preserved during freezing, by mechanical entrapment within a glassy cryoprotectant matrix, i.e., the vitrification hypothesis. In the case of nonviral vectors, the high viscosity of glasses prevents bimolecular collisions leading to aggregation. The vitrification hypothesis predicts that freezeconcentrated complexes must be cooled to temperatures below Tg' of the system in order to prevent damage caused by aggregation. Aggregation behavior of complexes in the absence of cryoprotectants depends on freezing rate and the temperature to which they are cooled. When DMRIE-C–DNA complexes at a 3:1 mass ratio (lipid:DNA) are quickly frozen

234

Allison and Anchordoquy

to –80°C, or immersed in liquid nitrogen, particle size measured by dynamic light scattering remains similar to that of fresh complexes, and in vitro transfection activity is preserved. However, if complexes are frozen to –40°C at 2.5oC/min, or placed in a –20°C freezer, particle size increases after thawing, and transfection activity is lost (41,49; Allison et al., unpublished). This behavior is consistent with the vitrification of DMRIE-C–DNA complexes (Tg ' = –47°C). However, excipients (e.g., sugars) are capable of preventing aggregation and maintaining transfection rates during freeze-thawing to temperatures greater than –47°C, suggesting that vitrification of complexes is not necessary when freezethawing takes place in the presence of cryoprotectants. Each of the disaccharides that have been tested to date (sucrose, trehalose, lactose, and maltose) possess Tg' values between –30 and –32°C, indicating that they form glasses when frozen to temperatures below –32°C. In their studies of cryoprotection by disaccharides, Cherng et al. (7) were careful to freeze complexes to temperatures below Tg'. Vitrification might have contributed to the observed protection by immobilizing complexes in a glassy excipient matrix and preventing aggregation. However, the mechanism of protection was not specifically addressed. Similarly, DMRIE:dioleoyl phosphatidyl choline–DNA complexes formulated in a sorbitol/sodium acetate buffer (Tg'~–43°C) were stable for 1 yr, when frozen rapidly to –70°C; complexes frozen to –20°C and thawed immediately, were damaged (41). However, in contrast to the observations noted above, complexes frozen quickly to –70°C, followed by storage at –20°C were also stable, indicating that vitrification may not be necessary for protection. To resolve this issue, the authors performed freeze-thawing experiments on cationic lipid-DNA complexes mixed with glucose (Tg' = –43°C), sucrose (Tg' = –32°C), and hydroxyethyl starch (HES) (T g' = –10°C) (Allison et al., unpublished). Samples were frozen at 2.5°C/min on the lyophilizer shelf to a final shelf temperature of –40°C, and held overnight prior to rapid thawing. Thermocouples placed in representative vials indicated that the sample temperatures of –38°C were reached. Sucrose samples, which were vitrified, retained particle size and transfection activity characteristics of fresh complexes. However, complexes containing glucose, which were not glassy, also retained particle size and activity. Furthermore, the authors found that HES was not capable of preventing particle aggregation at concentrations up to 10% (w/v), despite the fact that the samples were vitrified. Thus, vitrification is not necessary to prevent damage to nonviral vectors during freezing, at least on the time-scale of these experiments.

2.3.3. Particle Isolation Hypothesis DSC analysis of DMRIE-C–DNA complexes mixed with 1% sucrose shows two separate glass transitions (Fig. 3). Each glass transition corresponds to that

Lyophilization of Gene Delivery Systems

235

Fig. 3. DSC thermogram showing glass transitions for cationic lipid–DNA complexes and sucrose. The sample weighed approx 18 mg and was composed of 0.33 mg/mL DNA and 1 mg/mL DMRIE-C in 1% (w/v) sucrose. Noise in the trace is indicative of the close approach to the detection limit of the instrument, but illustrates the practical value of thermal analysis in formulation development. Vertical bar equals 0.002 W/g.

of each component of the mixture, complexes (Tg' = –47°C) and sucrose (T g' = –32°C), indicating that sucrose and complexes exist as separate amorphous phases in the freeze concentrate. Phase separation has also been observed for hydroxyethyl starch (Fig. 2). This observation is not surprising, because the formation of complexes results in a turbid suspension of particles. However, when the mixture of complexes and 1% sucrose is frozen to –40°C at 2.5°C/min and rapidly thawed, sucrose prevents particle aggregation and preserves transfection activity despite separation into two amorphous phases during freezing. Although complexes and additives are in separate phases, as judged by both visual appearance prior to freezing and thermal analysis of frozen samples, the degree to which complexes are protected by additives such as sucrose increases with the sucrose:complex mass ratio. For truly phase-separated systems, each component has a finite, fixed solubility in the other. Since Tg' of a single-phase mixture is proportional to the Tg' of each component (67,85), a truly phaseseparated mixture of sucrose and complexes would have two glass transitions, representing sucrose saturated in complexes, and complexes saturated in sucrose. Because the mixture in each phase would have a fixed composition, values for

236

Allison and Anchordoquy Table 1 Effect of Excipient:Complex Weight Ratio on Excipient Glass Transition Temperatures Excipient HES HES HES Sucrose Sucrose Sucrose Sucrose Glucose Glucose Glucose Glucose

Excipient/DNA (w/w) 300 600 ' 60 300 600 ' 60 300 600 '

Tg' (°C)a –10.9 ± 0.2 –10.2 ± 0.2 –9.4 ± 0.7 –37.2 ± 0.2 –32.4 ± 0.7 –31.6 ± 0.2 –31.3 ± 0.5 –47.2 ± 0.3 –44.0 ± 0.4 –43.1 ± 0.2 –42.6 ± 0.2

represent the mean ± 1 SD of thermal scans on triplicate samples. HES, hydroxyethyl starch. a Values

each glass transition would remain constant when the sugar:complex mass ratio was varied. However, this is not the case. Increasing the complex:sucrose ratio drives the Tg' values for sucrose toward that of complexes (Table 1), indicating a high degree of physical interaction between apparently separate phases. This interaction with complexes is also observed for HES and glucose. Aggregation of cationic liposome–DNA particles during freezing is probably mediated, at least in part, by liposome fusion events. Sucrose is thought to prevent liposome fusion during freezing, by hydrogen bonding to polar lipid headgroups at the surface of the liposome (81,86,87). Increasing the sucrose:complex ratio increases the degree of binding to lipid headgroups. Complex size would decrease to minimal levels, as sucrose binding to lipid headgroups becomes saturated, which is consistent with the author's experimental observations. In addition to liposome fusion, electrostatic and colloidal mechanisms also promote aggregation of nonviral vectors (34,43–46). In fact, vectors consisting of nonlipid components also aggregate during freeze-thawing (7). Therefore, specific binding of cryoprotectants to lipids during freezing is not likely to be the predominant mechanism of protection during freezing. The authors propose an alternative explanation for the protective effect of excipients on lipid–DNA complexes during freezing, i.e., the particle isolation hypothesis. According to this hypothesis, particle collisions leading to vector aggregation are inhibited by dilution in the cryoprotectant matrix. With increasing initial excipient concentration, vector particles become more dilute

Lyophilization of Gene Delivery Systems

237

within the freeze-concentrate. During cooling, viscosity of the matrix increases with decreasing temperature, with the largest increase arising from the sudden increase in concentration upon freezing. The authors propose that the increasing viscosity of the freeze-concentrated “syrup” retards vector diffusion. In order to prevent aggregation, viscosity of the freeze-concentrated matrix need only be high enough to limit vector diffusion on relevant time-scales. The typical vector particle has a diameter of 100–300 nm. As a result, diffusion coefficients are low compared to proteins, for which vitrification is a necessary condition for protection. Therefore, matrix vitrification during freezing may not be necessary, at least on the time-scale of a typical lyophilization protocol. It follows from the above argument that any excipient should protect nonviral vectors during freezing, if concentrations are added to sufficiently dilute vector particles. However, HES shows a limited capacity to prevent aggregation at concentrations at which other carbohydrates fully protect size and transfection activity. The question then arises: What physical property of HES could account for this observation? The authors find that the surface tension of HES is significantly higher than that of either glucose or sucrose at concentrations found in frozen solutions (Allison et al., unpublished), and propose that, during freezing, the surface tension of the highly concentrated HES becomes sufficiently high to force vector particles out of the excipient matrix. The resulting high local vector concentration favors aggregation and prevents HES from fully preserving complex size. According to this suggestion, the surface tension of the freeze-concentrated excipient matrix must be low enough to allow complexes to remain dispersed during freezing. Additional studies are needed to definitively determine the role of surface tension in the cryoprotection of nonviral gene delivery systems.

2.4. Impact of Freezing Parameters on Complex Stability As mentioned previously, complexes frozen to sufficiently low temperatures (at correspondingly high rates) retain transfection activity, regardless of the presence of stabilizers. Thus, for freezing, the fastest possible rate of cooling would appear to be optimal. Conventional pilot-scale freeze-drying equipment is capable of cooling rates of about 2.5°C/min (FTS Systems, Stone Ridge, NY), which specifies the cooling rate of the shelves. The actual product temperature lags behind the shelf temperature, because of the finite rate of heat transfer from vial to shelf (Fig. 1). Although it is possible to design lyophilization cycles using fast freezing methods, such as immersion in cold baths or liquid nitrogen, followed by transfer to precooled shelves (shelf temperatures of –70°C are possible on pilot-scale lyophilizers), such methods tend not to be practical on a large scale and can lead to batch variability. The physical behavior of excipients during freezing can be analyzed by differential scanning calorimetry, in order to determine freezing parameters for

238

Allison and Anchordoquy

the lyophilization cycle. Tg' values are typically measured during heating to avoid the interference of ice crystallization signals in supercooled samples. Because excipient behavior (e.g., crystallization of mannitol) may vary, depending on freezing rate, the freezing protocol of the DSC experiment must be designed to mimic freezing rates attainable by the lyophilizer. Thermal analysis of samples using multiple cooling rates, should be used to determine optimal rates for lyophilization cycles. Tg' values measured by DSC also depend on temperature and scanning rate. Increasing the heating rate increases both the amplitude (change in heat capacity) and the temperature of the glass transition (88). Thus, the actual Tg', and corresponding collapse temperature of formulations, can be significantly different from values measured by DSC. These properties also apply to samples during drying, as will be discussed in Subheading 3.2. The choice of freezing parameters, cooling rate and freezing temperature, affects the structure of the nonfrozen solute fraction, which in turn affects the drying characteristics of the sample. The degree of supercooling is determined by cooling rate. A high cooling rate increases the degree of supercooling in the sample prior to ice crystallization (1). More supercooling increases the number of ice nuclei, and upon ice formation, a greater number (of smaller) ice crystals are formed, compared to samples cooled at lower rates (1). Smaller crystals form smaller pores in the sample cake, resulting in higher resistance to primary drying (1), necessitating longer drying times. Thus, potential reductions in cost of production, resulting from decreased freezing times may be more than offset by the resulting necessity to increase the length of primary drying. On the other hand, cooling rates that are too slow may lead to long residence times at temperatures above Tg' of the excipient matrix, which can lead to aggregation. 3. Primary Drying 3.1. Sublimation Freezing separates water into two populations; ice crystals, and unfrozen water associated with the freeze-concentrated solution. Removal of water from the sample vials takes place in two distinct steps: sublimation of ice crystals (primary drying), followed by desorption of water from the freeze-concentrate (secondary drying). After the sample is frozen, sublimation begins by lowering the chamber pressure to a point less than the partial pressure of water vapor at the sample temperature (Fig. 4). Sublimation begins at the top of the vial and proceeds downward through the sample. Water vapor is released from ice into the drying chamber; the latent heat of sublimation reduces product temperature to a level below that of the shelf setting. Now, rather than sample cooling, the function of the shelf is to provide heat energy for sublimation from shelves to product. The degree to which sample temperature is lowered depends on the rate of sublimation. After the sublimation front passes, the sample temperature

Lyophilization of Gene Delivery Systems

239

Fig. 4. Relationship between sample temperature and the vapor pressure over ice.

increases. Water vapor is removed from the drying chamber by deposition onto refrigerated condenser coils. Thus, the temperature of the condenser must be set so that the vapor pressure of the ice formed on the coils is lower than the chamber pressure. This vapor pressure gradient allows the mass transfer of water from sample vials to the condenser coils. Economic considerations require that the fastest possible drying cycle be developed. Thus, it would appear to be desirable to maximize sublimation rates by using the highest possible shelf temperatures in combination with the lowest attainable chamber pressure and condenser temperature. Sublimation rate increases approximately twofold for every five-degree increase in temperature (5). However, the rate of heat transfer from shelf to product is adversely affected by chamber pressures below about 100 mTorr (4). Thus, drying performance can actually decrease at low chamber pressures. Conversely, chamber pressures that are too high decrease the rate of sublimation and increase drying times. Condenser temperatures much lower than necessary to control chamber pressure do not improve the rate of mass transfer (4,5) and may result in unnecessary equipment operation costs, especially for long freeze-drying cycles. Heat and mass transfer also depend on sample geometry. Large surface areas increase rates of heat flux, facilitating rapid sublimation. Minimization of fill depth also decreases thermal gradients generated in the sample during freezing and drying. A vial with a small shelf contact area will have a less efficient heat

240

Allison and Anchordoquy

transfer coefficient and a given volume of product will have higher fill depth. Therefore, the temptation to maximize the number of vials dried during a lyophilization run by minimizing vial footprint can affect product quality, i.e., a larger thermal gradient will exist from the bottom to the top of the smaller vial during freezing. Significant thermal gradients existing within vials during sublimation can lead to sample collapse.

3.2. Collapse Sample temperature is controlled by sublimation rate and chamber pressure during primary drying. Some evaporation of unfrozen water from the freeze concentrate occurs after the sublimation front passes (2,68). Since water has a low Tg (–135°C) (67), it acts as a plasticizer, reducing the Tg of vector-excipient formulations (Fig. 5). Reducing water content by evaporation from the freeze-concentrate increases the collapse temperature of the sample (Tc) above Tg'. The degree of evaporation that takes place after sublimation, and the resulting increase in Tc, depends on sample temperature during drying, and is on the order of a few degrees (2). Because Tc is difficult to predict, Tg' is often used as a conservative estimate of the maximum-allowable sample temperature during primary drying. Maintaining sample temperatures much below Tg' during primary drying does not improve product quality, but merely results in unnecessarily long drying times. If the sample temperature exceeds Tc during primary drying, the porous solute matrix collapses, because of surface tension and gravitational forces (2,68), forming a barrier to water vapor transport. The appearance of collapsed samples can vary, depending on the extent to which samples have dried when the collapse temperature was reached. If an effort is made to determine Tg' of the formulation, the most common problem is that the primary drying step is not long enough to completely sublimate all the ice in each sample. If this occurs, the sublimation front will not have passed completely through some of the vials before the drying program increases the shelf temperature for the secondary drying step. The top part of the samples may appear to be normal, but the part of the sample that still contains ice will have melted. Over time, water in the sample will equilibrate and further change the appearance of the collapsed samples. Considerations of product appearance dictate that collapse should be avoided, irrespective of vector stability. The authors have lyophilized cationic lipid–DNA complexes with glucose (Tg' = –43°C), using a drying cycle that maintained product temperature at –35°C. Samples appeared as a thin film on the bottom of the vial, indicative of collapse during drying. However, the vectors dried in glucose retained size and transfection characteristics comparable to both freshly made complexes and complexes dried during the same run containing sucrose (Tg' = –32°C), when rehydrated immediately after drying (10).

Lyophilization of Gene Delivery Systems

241

Fig. 5. Plasticizing effect of water on sucrose glasses. The glass transition curve was calculated from the simplified Fox equation described in ref. 110. Glass transition temperatures were taken to be 138 K for water (67), and 338 K for sucrose (111). The symbol indicates the approximate composition of the maximally freeze-concentrated sucrose glass at Tg' = –32°C.

3.3. Monitoring Primary Drying Sample temperature is most commonly monitored using thermocouples inserted directly into several representative vials. The end point of primary drying is defined using thermocouples placed at the bottom center of representative vials as the point at which sample temperature sharply increases to match the shelf temperature. However, the presence of the probe in the sample can affect the freezing behavior of the product (5). Typically, the level of supercooling is decreased relative to the rest of the vials in the chamber, with a subsequent increase in drying rate as described previously. Furthermore, sample temperature varies among vials in different positions on drying shelves. Vials adjacent to chamber walls behave differently than vials in the middle of

242

Allison and Anchordoquy

the shelf, because of heat transfer from the door and walls of the drying chamber. Thus, common practice adds time to the primary drying step to provide a safety margin based on this variability when thermocouples are used. Processing efficiency can be improved by minimizing unnecessary time spent at each drying step. Therefore, it is important to accurately determine the end point of primary drying. Manometric temperature measurement is a method developed to accurately determine the average product temperature among all vials in the chamber without affecting drying behavior (89). It is based on analysis of the rate of pressure increase in the sample chamber when a valve separating the drying and condenser chamber is periodically closed for a short time. Upon chamber isolation from the condenser, pressure increases, because of the production of water vapor from ice during sublimation (89). When sublimation is complete, the rate of pressure increase is sharply reduced when the chamber is isolated compared to the rate of pressure rise during sublimation. Using this technique, the end of the primary drying step can be determined accurately among all vials without affecting drying behavior. A less expensive option is an electronic moisture sensor, which has sufficient sensitivity to detect residual ice in less than 1% of the vials (90). A sharp decrease in the relative humidity in the drying chamber indicates the end of sublimation. One potential weakness of manometric temperature measurement is that product temperature can increase when the drying chamber is isolated, risking product collapse. Using a moisture sensor avoids the potential for sample collapse, which exists with manometric temperature measurement, since sample temperature does not increase with humidity measurement. 4. Secondary Drying 4.1. Secondary Drying Process Water evaporates from the freeze-concentrate during secondary drying. Moisture levels decrease from approx 30 to 1% (5). As a result, the glass transition temperature of the product continually increases from Tc. The purpose of secondary drying is to increase the Tg of the product to a level that will permit long-term storage at a convenient temperature. Exceeding Tg of dried formulations does not affect the capacity of sugars to protect the size and transfection activity of vectors over a period of days (10). However, just as vectors lose activity because of aggregation in aqueous suspension, they will probably also aggregate in dried formulations over longer time scales (years), if the dried matrices have sufficient fluidity (i.e., if storage temperature exceeds Tg). Furthermore, vitrified sugar matrices exhibit mobility at temperatures as low as 50°C below Tg, which is linked to degradation of model pharmaceutical compounds (85,91,92). Thus, it is desirable to maximize Tg of dried vector formulations.

Lyophilization of Gene Delivery Systems

243

Processing parameters impact product quality during secondary drying. Typically, chamber pressure is reduced to tens of mTorr, with the idea being to maximize the rate of water vapor mass transfer (68). However, desorption rates are limited by diffusion of liquid water through the glass and evaporation at the surface (3,93). Secondary drying rate is insensitive to pressures of 0–200 mTorr (3). Drying to moisture levels on the order of 1% requires an input of thermal energy that is achieved by increasing the shelf temperature. Although T g continually increases with dehydration, the rate at which shelf and product temperature increases must ensure (more for considerations of product appearance than stability) that sample temperature does not exceed Tg.

4.2. Effect of Dehydration on Complexes Dry DNA is susceptible to oxidative damage, especially when complexed with unsaturated lipids (52,54,55). Condensation of plasmid DNA into densely packed structures, with histone (94) or with cationic polymer (20,23–26), has been shown to increase resistance to chemical degradation, which also seems to be correlated with the exclusion of intercalating dyes (27,95). Although complexation with monotonically charged cationic liposomes does not condense plasmid DNA (34,45,96), cationic lipids may serve as a barrier to damaging agents, since complexes of DNA with DMRIE or 1,2-dioleoyl-3-trimethylammonium propane inhibit ethidium bromide intercalation (33,97). Cationic lipid bilayers have also been reported to induce a B to C conformational change in plasmid DNA, which may contribute to protection (98). Complexation and condensation may help protect dry DNA from chemical degradation, but charge neutralization coupled with increased particle concentration, increases the potential for aggregation. Dehydration is apparently a more physically destructive stress to vectors than is freezing (7,9,10), given that transfection activity can increase to levels greater than that of fresh preparations after freezing. Therefore, even in samples in which 100% activity is recovered after freeze-drying and rehydration, there can be significant loss of activity during dehydration (10).

4.3. Mechanisms of Protection During Dehydration As was noted for freezing, product vitrification during the drying process is not necessary to preserve the size and activity of cationic lipid–DNA complexes during acute lyophilization stress. Work from our group (10) shows that cationic lipid–DNA complexes containing glucose, which collapsed during freeze-drying, were capable of maintaining both particle size and transfection activity, when reconstituted immediately after drying. It is not known how long-term storage in fluid glucose (Tg = 7°C) would affect particle size and biological activity, apart from the fact that, as a reducing sugar, glucose could

244

Allison and Anchordoquy

possibly form adducts with complexes (99). This result suggests that aggregation of complexes is inhibited in fluid sugar matrices, at least on the time-scale of the drying process. Therefore, apart from considerations of product appearance, maintaining the glassy state of sugar matrices in which polycation–DNA particles are suspended during dehydration, is not a necessary condition for the retention of vector activity during freeze-drying. Although vitrification is not required for vector stability during acute freezing or dehydration (on time-scales of days), minimizing particle mobility via storage well below Tg of the dried product appears to be necessary for longterm stability. Cherng et al. (8) reported that dried poly([2-dimethylamino]ethyl methacrylate)–DNA complexes (Tg = 50°C), stored at 4 or 26°C, were stable for 10 mo. However, samples stored at 40°C, which were still glassy, lost 25% of transfection activity by aggregation during that same period. This behavior is consistent with demonstrations of mobility within glassy sucrose matrices at sufficient levels to permit damage to small molecules and proteins (74,85). During the final stages of secondary drying, when residual moisture levels decrease to around 1%, water that is directly associated with macromolecules (i.e., water that forms the primary hydration shell) is removed (100–102). Polyhydroxyl compounds, such as disaccharides, have been shown to bind directly to proteins and lipid membranes in place of this water (103–106). The replacement of a water-like hydrogen-bonding network at the surface of proteins and lipid membranes maintains a higher-order structure on which biological activity depends. Water replacement appears to be consistent with the observed capacity of several different types of excipients to protect vector size and transfection activity. As previously mentioned, sucrose stabilizes both cationic polymer– and cationic lipid–DNA complexes. Other disaccharides, trehalose and lactose, as well as monosaccharides, such as glucose, also protect cationic lipid–DNA complexes during lyophilization followed by immediate rehydration (9,10). However, mannitol does not protect freeze-dried complexes, because it crystallizes during the freezing step of lyophilization (10). Crystallization drastically reduces the number of hydroxyl groups available to hydrogen bond with complexes in the amorphous phase. Furthermore, large polymeric sugars such as dextran and hydroxyethyl starch do not protect dried proteins or liposomes, because these polymers are sterically prevented from associating with biomolecules during dehydration (104,106–109). As a result, polymers do not provide adequate levels of protection to dried cationic lipid–DNA complexes (9). Studies to date suggest that the water replacement hypothesis cannot entirely explain the mechanism by which cationic lipid–DNA complexes are protected during dehydration. Polyethylene glycol, a compound that remains amorphous

Lyophilization of Gene Delivery Systems

245

during freezing, but crystallizes during dehydration, provides levels of protection to complexes that are comparable to those of trehalose and lactose (10). Clearly, crystalline PEG is not likely to hydrogen-bond to amorphous lipid–DNA complexes. In contrast, the particle-isolation hypothesis proposed for the stabilization of vectors during freezing is consistent with the protection offered to lipid– DNA complexes by PEG. Instead of water replacement, vector particles could be dispersed in freeze-concentrated PEG solution. Rather than the high viscosity of an amorphous matrix, as proposed for sugars, vector aggregation may be inhibited mechanically by a network of PEG crystals in the dehydrated state. 5. Conclusions The requirements for protecting the biophysical properties and transfection activity of nonviral gene vectors during freeze-drying and long-term storage are unlike those for proteins or liposomes. Maintaining the dispersion of vector particles within the unfrozen excipient solution during freezing appears to be critical for the successful freeze-drying of cationic lipid- or cationic polymer– DNA complexes. Even with adequate protection during the freezing step, much activity can be lost during drying. This chapter focused on aggregation as the means by which vectors are damaged during lyophilization. However, other mechanisms involving perturbation of the interaction between polycation and DNA also contribute to loss of transfection activity. Although vitrification is not a necessary condition for the protection of complex suspensions during freezing or drying, it does appear to be necessary for long-term storage. Although water replacement may play a role in protecting complexes, recent work questions the ability of sugars to fully prevent damage during drying (10). Nonetheless, lyophilization parameters can be chosen, based on the physicochemical properties of formulation constituents, to minimize processinduced damage. Clearly, more work is needed to fully understand the mechanisms by which nonviral gene delivery vectors are stabilized during lyophilization and long-term storage. Progress in the rational development of stable dehydrated vector formulations will overcome a significant barrier to the production of nonviral gene therapy pharmaceuticals. Acknowledgments The authors would like to thank Wyeth-Ayerst Research (Malvern, PA), Valentis (Burlingame, CA), FTS Systems (Stone Ridge, NY), and Amgen (Boulder, CO), for providing materials and instrumentation. S. D. A. is financially supported by National Institutes of Health (NIH)-National Cancer Institute training grant no. 1T32-CA-79446-01. Additional support for this work was provided by NIH-National Institute of General Medical Sciences grant no. 1 RO1 GM60587-01 to T. J. A.

246

Allison and Anchordoquy

References 1. Pikal, M. J., Shah, S., Senior, D., and Lang, J. E. (1983) Physical chemistry of freeze-drying: measurement of sublimation rates for frozen aqueous solutions by a microbalance technique. J. Pharm. Sci. 72, 635–650. 2. Pikal, M. J. and Shah, S. (1990) Collapse temperature in freeze drying: dependence on measurement methodology and rate of water removal from the glass phase. Int. J. Pharm. 62, 165–186. 3. Pikal, M. J., Shah, S., Roy, M. L., and Putman, R. (1990) The secondary drying stage of freeze-drying: drying kinetics as a function of temperature and chamber pressure. Int. J. Pharm. 60, 203–217. 4. Pikal, M. J., Roy, M. L., and Shah, S. (1984) Mass and heat transfer in vial freeze-drying of pharmaceuticals: role of the vial. J. Pharm. Sci. 73, 1224–1237. 5. Pikal, M. J. (1994) Freeze-drying of proteins. ACS Symp. Ser. 567, 120–133. 6. Talsma, H., Cherng, J. Y., Lehrmann, H., Kursa, M., Ogris, W., Hennink, W. E., Cotten, M., and Wagner, E. (1997) Stabilization of gene delivery systems by freeze-drying. Int. J. Pharm. 157, 233–238. 7. Cherng, J., Wetering, P. v. d., Talsma, H., Crommelin, D. J. A., and Hennink, W. E. (1997) Freeze-drying of poly([2-dimethylamino]ethyl methacrylate)based gene delivery systems. Pharm. Res. 14, 1838–1841. 8. Cherng, J.-Y., Talsma, H., Crommelin, D. J. A., and Hennink, W. E. (1999) Long term stability of poly([2-dimethylamino]ethyl methacrylate)-based gene delivery systems. Pharm. Res. 16, 1417–1423. 9. Anchordoquy, T. J., Carpenter, J. F., and Kroll, D. J. (1997) Maintenance of transfection rates and physical characterization of lipid/DNA complexes after freeze-drying and rehydration. Arch. Biochem. Biophys. 348, 199–206. 10. Allison, S. D. and Anchordoquy, T. J. (2000) Mechanisms of protection of cationic lipid-DNA complexes during lyophilization. J. Pharm. Sci. 89, 682–691. 11. Anchordoquy, T. J. (1999) Nonviral gene delivery systems: physical stability. BioPharm 12, 42–48. 12. Anchordoquy, T. J. and Koe, G. S. (2000) Physical stability of nonviral plasmidbased therapeutics. J. Pharm. Sci. 89, 289–296. 13. Brosteaux, J. and Eriksson-Quensel, I. (1935) Etude sur la dessication des proteines. Arch. Phys. Biol. 12, 209–226. 14. Arakawa, T. and Timasheff, S. N. (1982) Stabilization of protein structure by sugars. Biochemistry 21, 6536–6544. 15. Carpenter, J. F. and Crowe, J. H. (1988) Modes of stabilization of a protein by organic solutes during dessication. Cryobiology 25, 459–470. 16. Carpenter, J. F. and Crowe, J. H. (1988) The mechanism of cryoprotection of proteins by solutes. Cryobiology 25, 244–255. 17. Crowe, J. H., Crowe, L. M., Carpenter, J. F., and Wistrom, C. A. (1987) Stabilization of dry phospholipid bilayers and proteins by sugars. Biochem. J. 242, 1–10. 18. Felgner, P. L., Gadek, T. R., Holm, M., Roman, R., Chan, H. W., Wenz, M., et al. (1987) Lipofection: a highly efficient, lipid-mediated DNA transfection procedure. Proc. Natl. Acad. Sci. USA 87, 7413–7417.

Lyophilization of Gene Delivery Systems

247

19. Boussif, O., Lezoualc’h, F., Zanta, M. A., Mergny, M. D., Scherman, D., Demeneix, B., and Behr, J. P. (1995) A versatile vector for gene and oligonucleotide transfer into cells in culture and in vivo: polyethyleneimine. Proc. Natl. Acad. Sci. USA 92, 7292–7301. 20. Boussif, O., Zanta, M. A., and Behr, J.-P. (1996) Optimized galenics improve in vitro gene transfer with cationic molecules up to 1000-fold. Gene Ther. 3, 1074–1080. 21. Fischer, D., Bieber, T., Li, Y., Elsasser, H. P., and Kissel, T. (1999) Novel nonviral vector for DNA delivery based on low molecular weight, branched polyethyleneimine: effect of molecular weight on transfection efficiency and cytotoxicity. Pharm. Res. 16, 1273–1279. 22. Godbey, W. T., Wu, K. K., and Mikos, A. G. (1999) Poly(ethyleneimine) and its role in gene delivery. J. Controlled Release 60, 149–160. 23. Goldman, C. K., Soroceanu, L., Smith, N., Gillespie, G. Y., Shaw, W., Burgess, S., Bilbao, G., and Curiel, D. T. (1997) In vitro and in vivo gene delivery mediated by a synthetic polycationic amino polymer. Nature Biotechnol. 15, 462–466. 24. Adami, R. C., Collard, W. T., Gupta, S. A., Kwok, K. Y., Bonadio, J., and Rice, K. G. (1998) Stability of peptide-condensed plasmid DNA formulations. J. Pharm. Sci. 87, 678–683. 25. Gao, X. and Huang, L. (1996) Potentiation of cationic liposome-mediated gene delivery by polycations. Biochemistry 35, 1027–1036. 26. Katayose, S. and Kataoka, K. (1998) Remarkable increase in nuclease resistance of plasmid DNA through supramolecular assembly with poly(ethylene glycol)poly(L-lysine) block copolymer. J. Pharm. Sci. 87, 160–163. 27. Kwoh, D. Y., Coffin, C. C., Lollo, C. P., Jovenal, J., Banaszczyk, M. G., Mullen, P., et al. (1999) Stabilization of poly-L-lysine/DNA complexes for in vivo gene delivery to the liver. Biochim. Biophys. Acta 1444, 171–190. 28. Zanta, M. A., Valladier, P. B., and Behr, J. P. (1999) Gene delivery: a single nuclear localization signal peptide is sufficient to carry DNA to the cell nucleus. Proc. Natl. Acad. Sci. USA 96, 91–96. 29. Goddard, C. A., Ratcliff, R., Anderson, J. R., Glenn, E., Brown, S., Gill, D. R., et al. (1997) Second dose of a CFTR cDNA-liposome complex is as effective as the first dose in restoring cAMP-dependent chloride secretion to null CF mice trachea. Gene Ther. 4, 1231–1236. 30. Verma, I. M. (2000) A tumultuous year for gene therapy. Mol. Ther. 2, 415–416. 31. Felgner, P. L. (1997) Nonviral strategies for gene therapy. Sci. Am. 276, 102–106. 32. Lehn, P., Fabrega, S., Oudrhiri, N., and Navarro, J. (1998) Gene delivery systems: bridging the gap between recombinant viruses and artificial vectors. Adv. Drug Delivery Rev. 30, 5–11. 33. Pouton, C. W. and Seymour, L. W. (1998) Key issues in nonviral gene delivery. Adv. Drug Delivery Rev. 34, 3–19. 34. Gustafsson, J., Arvidson, G., Karlsson, G., and Almgren, M. (1995) Complexes between cationic liposomes and DNA visualized by cryo-TEM. Biochim. Biophys. Acta 1235, 305–312.

248

Allison and Anchordoquy

35. Dunlap, D. D., Maggi, A., Soria, M. R., and Monaco, L. (1997) Nanoscopic structure of DNA condensed for gene delivery. Nucleic Acids Res. 25, 3095–3101. 36. Zabner, J., Fasbender, A. J., Moninger, T., Poellinger, K. A., and Welsh, M. J. (1995) Cellular and molecular barriers to gene transfer by a cationic lipid. J. Biol. Chem. 270, 18,997–19,007. 37. Smith, J. G., Wedeking, T., Vernachio, J. H., Way, H., and Niven, R. W. (1998) Characterization and in vivo testing of a heterogeneous cationic lipid-DNA formulation. Pharm. Res. 15, 1356–1363. 38. Jaaskelainen, I., Monkkonen, J., and Urtti, A. (1994) Oligonucleotide-cationic liposome interactions: a physicochemical study. Biochim. Biophys. Acta 1195, 115–123. 39. Kichler, A., Zauner, W., Orgis, M., and Wagner, E. (1998) Influence of the DNA complexation medium on the transfection efficiency of lipospermine/DNA particles. Gene Ther. 5, 855–860. 40. Yang, J. P. and Huang, L. (1998) Time-dependent maturation of cationic lipsome-DNA complexes. Gene Ther. 5, 380–387. 41. Zelphati, O., Nguyen, C., Ferrari, M., Felgner, J., Tsai, Y., and Felgner, P. L. (1998) Stable and monodisperse lipoplex formulations for gene delivery. Gene Ther. 5, 1272–1282. 42. Anchordoquy, T. J. (1999) Nonviral gene delivery systems: physical characterization. BioPharm 12, 46–51. 43. Kay, M. A., Liu, D., and Hoogerbrugge, P. M. (1997) Gene therapy. Proc. Natl. Acad. Sci. USA 94, 12,744–12,746. 44. Tang, M. X. and Szoka, F. C. (1997) The influence of polymer structure on the interactions of cationic polymers with DNA and morphology of the resulting complexes. Gene Therapy 4, 823–832. 45. Radler, J. O., Koltover, I., Jamieson, A., Salditt, T., and Safinya, C. R. (1998) Structure and interfacial aspects of self-assembled cationic lipid-DNA gene carrier complexes. Langmuir 14, 4272–4283. 46. Birchall, J. C., Kellaway, I. W., and Mills, S. N. (1999) Physico-chemical characterization and transfection efficiency of lipid-based gene delivery complexes. Int. J. Pharm. 183, 195–207. 47. Nabel, G. J., Chang, A., Nabel, E. G., Plautz, G., Fox, B. A., Huang, L., and Shu, S. (1992) Imunotherapy of malignancy by in vivo gene-transfer into tumors. Hum. Gene Ther. 3, 399–410. 48. Nabel, G. J., Chang, A., Nabel, E. G., Yang, Z., Fox, B. A., Plautz, G. E., et al. (1993) Immunotherapy of malignancy by in vivo gene-transfer into tumors. Proc. Natl. Acad. Sci. USA 90, 11,307–11,311. 49. Anchordoquy, T. J., Girouard, L. G., Carpenter, J. F., and Kroll, D. J. (1998) Stability of lipid/DNA complexes during agitation and freeze-thawing. J. Pharm. Sci. 87, 1046–1051. 50. Lindahl, T. (1993) Instability and decay of the primary structure of DNA. Nature 362, 709–715. 51. Middaugh, C. R., Evans, R. K., Montgomery, D. L., and Casimiro, D. R. (1998) Analysis of DNA from a pharmaceutical perspective. J. Pharm. Sci. 87, 130–146.

Lyophilization of Gene Delivery Systems

249

52. Gedik, C. M., Wood, S. G., and Collins, A. R. (1998) Measuring oxidative damage to DNA; HPLC and the comet assay compared. Free Rad. Res. 29, 609–615. 53. Evans, R. K., Xu, Z., Bohannon, K. E., Wang, B., Brunner, M. W., and Volkin, D. B. (1999) Evaluation of degradation pathways for plasmid DNA in pharmaceutical formulations via accelerated stability studies. J. Pharm. Sci. 89, 76–87. 54. Weiland, B. and Hutterman, J. (1998) Free radicals from X-irradiated dry and hydrated lyophilized DNA as studied by electron spin resonance spectroscopy: analysis of spectral components between 77K and room temperature. Int. J. Radiat. Biol. 74, 341–358. 55. Pietronigro, D. D., Jones, W. B. G., Kalty, K., and Demopoulos, H. B. (1977) Interaction of DNA and liposomes as a model for membrane-mediated DNA damage. Nature 267, 78–79. 56. Li, D., Wang, M., Lehr, J. G., and Randerath, K. (1995) DNA adducts induced by lipids and lipid peroxidation products: possible relationships to I-compounds. Mutat. Res. 344, 117–126. 57. Wheeler, C. J., Sukhu, L., Yang, G., Tsai, Y., Bustamente, C., Felgner, P., Norman, J., and Manthorpe, M. (1996) Converting an alcohol to an amine in a cationic lipid dramatically alters the co-lipid requirement, cellular transfection activity and the ultrastructure of the DNA-cytofectin complexes. Biochim. Biophys. Acta. 1280, 1–11. 58. Hofland, H. E. J., Shephard, L., and Sullivan, S. M. (1996) Formation of stable cationic lipid/DNA complexes for gene transfer. Proc. Natl. Acad. Sci. USA 93, 7305–7309. 59. Franks, F., Hatley, R. H. M., and Mathias, S. F. (1991) Materials science and the production of shelf-stable biologicals. Pharm. Technol. Int. 3, 40–44. 60. Allison, S. D., Randolph, T. W., Manning, M. C., Middleton, K., Davis, A. S., and Carpenter, J. F. (1998) Effects of drying methods and additives on structure and function of actin: mechanisms of dehydration-induced damage and its inhibition. Arch. Biochem. Biophys. 358, 171–181. 61. Freeman, D. J. and Niven, R. W. (1996) The influence of sodium glycocholate and other additives on the in vivo transfection of plasmid DNA in the lungs. Pharm. Res. 13, 202–209. 62. Walter, E., Moelling, K., Pavlovic, J., and Merkle, H. P. (1999) Microencapsulation of DNA using poly (DL-lactide-co-glycolide): stability issues and release characteristics. J. Controlled Release 61, 361–374. 63. Mahato, R. I., Rolland, A., and Tomlinson, E. (1997) Cationic lipid-based gene delivery systems: pharmaceutical perspectives. Pharm. Res. 14, 853–859. 64. Bruno, M. S., Singhal, A., Mumper, R. J., Barron, M. K., Nichol, F., Szoka, F. C., and Rolland, A. P. (1995) Cryoprotection of cationic lipid-DNA complexes for gene delivery. Pharm. Res. 12, S–79. 65. Levine, H. and Slade, L. (1988) Principles of cryostabilization technology from structure/property relationships of carbohydrate/water systems: a review. CryoLetters 9, 21–63. 66. Slade, L. and Levine, H. (1993) The Glassy State in Foods (Blanshard, J. M. V. and Lillford, P. J., eds.), Nottingham University Press, Nottingham, pp. 35–101.

250

Allison and Anchordoquy

67. Chang, B. and Randall, C. S. (1992) Use of subambient thermal analysis to optimize protein lyophilization. Cryobiology 29, 632–656. 68. Willemer, H. (1999) Freeze-Drying/Lyophilization of Pharmaceutical and Biological Products (Rey, L. and May, J. C., eds.), Marcel Dekker, New York, pp. 79–121. 69. Levine, H. and Slade, L. (1988) Thermomechanical properties of small-carbohydrate water glasses and ‘rubbers’: kinetically metastable systems at sub-zero temperatures. J. Chem. Soc. Faraday Trans. 184, 2619–2633. 70. Her, L.-M., Jeffries, R. P., Gatlin, L. A., Braxton, B., and Nail, S. L. (1994) Measurement of glass transition temperatures in freeze concentrated solutions of non-electrolytes by electrical thermal analysis. Pharm. Res. 11, 1023–1028. 71. Kim, A. I., Akers, M. J., and Nail, S. L. (1998) The physical state of mannitol after freeze-drying: effects of mannitol concentration, freezing rate, and noncrystallizing cosolute. J. Pharm. Sci. 87, 931–935. 72. Talsma, H., Steenbergen, M. J. V., Salemink, P. J. M., and Crommelin, D. J. A. (1991) Cryopreservation of liposomes. 1. A differential scanning calorimetry study of the thermal behavior of a liposome dispersion containing mannitol during freezing/thawing. Pharm. Res. 8, 1021–1026. 73. Shalaev, E. Y. and Kanev, A. N. (1994) Study of the solid-liquid state diagram of the water-glycine-sucrose system. Cryobiology 31, 374–382. 74. Anchordoquy, T. J. and Carpenter, J. F. (1996) Polymers protect lactate dehydrogenase during freeze-drying by inhibiting dissociation in the frozen state. Arch. Biochem. Biophys. 332, 231–238. 75. Spear, N. and Aust, S. D. (1995) Effects of glutathione on fenton reagent-dependent radical production and DNA oxidation. Arch. Biochem. Biophys. 324, 111–116. 76. Lyscov, V. N. and Moshkovsky, Y. S. H. (1969) DNA cryolysis. Biochim. Biophys. Acta 190, 101–110. 77. Crook, K., McLachlan, G., Stevenson, B. J., and Porteous, D. J. (1996) Plasmid DNA molecules complexed with cationic liposomes are protected from degradation by nucleases and shearing by aerosolization. Gene Ther. 3, 834–839. 78. Setlow, P. (1992) I will survive: protecting and repairing spore DNA. J. Bacteriol. 174, 2737–2741. 79. Ljungman, M. and Hanawalt, P. C. (1992) Efficient protection against oxidative DNA damage in chromatin. Mol. Carcinog. 5, 264–269. 80. Breen, A. P. and Murphy, J. A. (1995) Reactions of oxyl radicals with DNA. Free Rad. Biol. Med. 18, 1033–1077. 81. Strauss, G. and Hauser, H. (1986) Stabilization of lipid bilayer vesicles by sucrose during freezing. Proc. Natl. Acad. Sci. USA 83, 2422–2426. 82. MacDonald, R. C., Jones, F. D., and Qiu, R. (1994) Fragmentation into small vesicles of dioleoylphosphatidylcholine bilayers during freezing and thawing. Biochim. Biophys. Acta 1191, 362–370. 83. Lee, J. C. and Timasheff, S. N. (1981) Stabilization of proteins by sucrose. J. Biol. Chem. 256, 7193–7201. 84. Timasheff, S. N. (1995) Protein-Solvent Interactions (Gregory, R. B., ed.), Marcel Dekker, New York, pp. 445–482.

Lyophilization of Gene Delivery Systems

251

85. Hancock, B. C., Shamblin, S. L., and Zografi, G. (1995) Molecular mobility of amorphous pharmaceutical solids below their glass transition temperatures. Pharm. Res. 12, 799–806. 86. Strauss, G., Schurtenberger, P., and Hauser, H. (1986) The interaction of saccharides with lipid bilayer vesicles: stabilization during freeze-thawing and freeze-drying. Biochim. Biophys. Acta 858, 169–180. 87. Anchordoquy, T. J., Rudolph, A. S., Carpenter, J. F., and Crowe, J. H. (1987) Modes of interaction of cryoprotectants with membrane phospholipids during freezing. Cryobiology 24, 324–331. 88. Her, L.-M. and Nail, S. L. (1994) Measurement of glass transition temperatures of freeze-concentrated solutes by differential scanning calorimetry. Pharm. Res. 11, 54–59. 89. Milton, N., Pikal, M. J., Roy, M. L., and Nail, S. L. (1997) Evaluation of manometric temperature measurement as a method of monitoring product temperature during lyophilization. PDA J. Pharm. Sci. Tech. 51, 7–16. 90. Bardat, A., Biguet, J., Chatenet, E., and Courteille, F. (1993) Moisture measurement: a new method for monitoring freeze-drying cycles. J. Parenter. Sci. Technol. 47, 293–299. 91. Duddu, S. P., Zhang, G., and Monte, P. R. D. (1997) The relationship between protein aggregation and molecular mobility below the glass transition temperature of lyophilized formulations containing a monoclonal antibody. Pharm. Res. 14, 596–600. 92. Duddu, S. P. and Monte, P. R. D. (1997) Effect of glass transition temperature on the stability of lyophilized formulations containing a chimeric therapeutic monoclonal antibody. Pharm. Res. 14, 591–595. 93. Aldous, B. J., Franks, F., and Greer, A. L. (1997) Diffusion of water within an amorphous carbohydrate. J. Mater. Sci. 32, 301–308. 94. Bloomfield, V. A. (1996) DNA condensation. Curr. Opin. Struct. Biol. 6, 334–341. 95. Li, S., Tseng, W. C., Stolz, D. B., Wu, S. P., Watkins, S. C., and Huang, L. (1999) Dynamic changes in the characteristics of cationic lipidic vectors after exposure to mouse serum: implications for intravenous lipofection. Gene Ther. 6, 585–594. 96. Radler, J. O., Koltover, I., Salditt, T., and Safinya, C. R. (1997) Structure of DNA–cationic liposome complexes: DNA intercalation in multilamellar membranes in distinct interhelical packing regimes. Science 275, 810–814. 97. Eastman, S. J., Siegel, C., Tousignant, J., Smith, A. E., Cheng, S. H., and Scheule, R. K. (1997) Biophysical characterization of cationic lipid:DNA complexes. Biochim. Biophys. Acta 1325, 41–62. 98. Akao, T., Fukumoto, T., Ihara, H., and Ito, A. (1996) Conformational change in DNA induced by cationic bilayer membranes. FEBS Lett. 391, 215–218. 99. Colaco, C. A. L. S., Smith, C. J. S., Sen, S., Roser, D. H., Newman, Y., Ring, S., and Roser, B. J. (1994) Chemistry of protein stabilization by trehalose. ACS Symp. Ser. 567, 222–240. 100. Crowe, J. H., Crowe, L. M., Carpenter, J. F., Prestrelski, S. J., and Hoekstra, F. A. (1997) Handbook of Physiology, vol. II (Dantzler, W. H., ed.), Oxford University Press, Oxford, pp. 1445–1478.

252

Allison and Anchordoquy

101. Rupley, J. A. and Careri, G. (1991) Protein hydration and function. Adv. Protein Chem. 41, 37–172. 102. Careri, G., Giansanti, A., and Gratton, E. (1979) Lysozyme film hydration events: an IR and gravimetric study. Biopolymers 18, 1187–1203. 103. Carpenter, J. F. and Crowe, J. H. (1989) An infrared spectroscopic study of the interactions of carbohydrates with dried proteins. Biochemistry 28, 3916–3922. 104. Allison, S. D., Chang, B., Randolph, T. W., and Carpenter, J. F. (1999) Hydrogen bonding between sugar and protein is responsible for inhibition of dehydrationinduced protein unfolding. Arch. Biochem. Biophys. 365, 289–298. 105. Crowe, L. M., Womersley, C., Crowe, J. H., Reid, D., Appel, L., and Rudolph, A. (1986) Prevention of fusion and leakage in freeze-dried liposomes by carbohydrates. Biochim. Biophys. Acta 861, 131–140. 106. Sun, W. Q., Leopold, A. C., Crowe, L. M., and Crowe, J. H. (1996) Stability of dry liposomes in sugar glasses. Biophys. J. 70, 1769–1776. 107. Crowe, J. H., Leslie, S. B., and Crowe, L. M. (1994) Is vitrification sufficient to preserve liposomes during freeze-drying? Cryobiology 31, 355–366. 108. Crowe, J. H., Hoekstra, F. A., Nguyen, K. H. N., and Crowe, L. M. (1996) Is vitrification involved in depression of the phase transition temperature in dry phospholipids? Biochim. Biophys. Acta 1280, 187–196. 109. Crowe, J. H., Oliver, A. E., Hoekstra, F. A., and Crowe, L. M. (1997) Stabilization of dry membranes by mixtures of HES and glucose: the role of vitrification. Cryobiology 35, 20–30. 110. Hancock, B. C. and Zografi, G. (1994) The relationship between the glass transition temperature and the water content of amorphous pharmaceutical solids. Pharm. Res. 11, 471–477. 111. Roos, Y. and Karel, M. (1990) Differential scanning calorimetry study of phase transitions affecting the quality of dehydrated materials. Biotechnol. Prog. 6, 159–163.

UV Absorption and CD Spectroscopy

253

19 Ultraviolet Absorption and Circular Dichroism Spectroscopy of Nonviral Gene Delivery Complexes Chad S. Braun, Lisa A. Kueltzo, and C. Russell Middaugh 1. Introduction A powerful, yet often underutilized tool available to probe the structure of macromolecules is absorption spectroscopy. The first of the various spectroscopic techniques to be widely developed, data from absorption spectroscopy is commonly viewed as low in overall information content compared to modern forms of X-ray diffraction, nuclear magnetic resonance, fourier-transformed infrared, and circular dichroism (CD) spectroscopies. With the development of diode array spectrophotometers and analysis software capable of quickly and simply producing high-definition derivative spectra, however, ultraviolet-visible spectroscopy remains a useful analytical technique, especially when applied to nucleic acids. This approach offers significant advantages in terms of flexibility, accuracy, speed, low expense, and instrumental availability. Thus, the technique can be productively applied to the analysis of nonviral gene delivery vectors, focusing primarily on the oligonucleotide component. Here are presented applications of derivative absorbance spectroscopy to the analysis of nucleic acid–cationic polymer complexes. Additionally, a brief overview of the theory, instrumentation, and general applications of zero and second-derivative absorption spectroscopy is presented, followed by a discussion of the complementary absorptive technique of circular dichroism in the same context. 2. Ultraviolet Absorption Spectroscopy 2.1. Theory and Instrumentation Before considering the physical origins of the absorbance spectra of nucleic acids, a brief general review of electronic absorption is warranted. Molecules generally reside in ground-state energy levels at room temperature. Various From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

253

254

Braun, Kueltzo, and Middaugh

excited states are available to molecules, and can be reached upon absorption of energy equivalent to the energy difference between the ground and excited states. Therefore, the excited states that molecules can reach are dependent on the wavelength of light used to irradiate the sample. In contrast to infrared absorption, in which a molecule is excited from a ground state to a higher vibrational energy level, the frequencies employed in conventional absorption spectroscopy lie within the UV- and visible range. The higher energy associated with light at these wavelengths excites the molecule to a higher electronic energy level, from which it subsequently relaxes and returns to the ground state. Although molecules can return to lower energy states by the emission of photons (fluorescence and phosphorescence), absorption spectroscopy is concerned with nonemissive processes. Within each electronic energy level exist multiple vibrational and rotational levels. Therefore, multiple transitions between the ground and excited states are possible, each of which can be induced by the light source. This multiplicity is reflected in the broad absorption peaks observed with molecules in solution. Each broad peak is actually composed of many distinct vibrationalelectronic bands in a distribution based on the probability of each individual transition. Collisions between molecules smear these individual signals into one broad peak, giving rise to the observed absorbance spectrum (1). Absorbance spectra are usually collected by employing commercial UVvisible spectrophotometers. Currently, both conventional grating and diode array instruments are available, which differ in both the physical arrangements of their components and their mechanism of light detection. In a conventional instrument, the light is usually generated by a lamp, then is passed through a monochromator. Deuterium arc lamps are commonly used for the UV range; tungsten-halogen lamps are employed for the visible. Broad-spectrum xenon lamps can also be used, but, because of the significantly higher noise of the xenon source, the deuterium-tungsten combination is more often employed. The monochromator consists of both a light-spectrum-generating (splitting) element, usually a diffraction grating, but occasionally a quartz prism, and an adjustable slit, permitting the selection of a defined band of light, which controls the instrument’s resolution. The light is then passed through the sample contained in a transparent cuvet (e.g., quartz for UV measurements) and the transmitted light is detected either by a photomultiplier or photodiode detector. In contrast, in a diode-array instrument, the polychromatic light from the source is passed through the sample, with the transmitted light dispersed by a polychromator prism directly onto the detector. The detector consists of an array of photodiodes, each capable of detecting a narrow bandwidth of light (2). This allows a full spectrum to be collected simultaneously without mechanical scanning, which increases the sampling speed compared to conventional instruments, although rapid-scanning grating instruments are now available that approach the spectral acquisition time of the

UV Absorption and CD Spectroscopy

255

diode-array format. A disadvantage is that the resolution of the spectrum is only as great as the spacing between the individual photodiodes. In conventional instruments, however, the arrays are often spaced as close as 0.5 nm, which is more than adequate for most routine measurements. Furthermore, interpolation between data points using continuous functions permits peaks to be defined to within ±0.01 nm, effectively offsetting any real problem. Dual-beam instruments are also available, allowing spectra of both sample and reference materials to be collected simultaneously, instead of consecutively, as performed by single-beam models. This design was originally intended to compensate for fluctuations in lamp intensity that may occur in the interval between collecting sample and blank spectra and retains some advantages in this regard. Over the past few years, however, advances in lamp and electronics design have greatly increased the stability of single-beam measurements, bringing their accuracy and precision well within the range of dual-beam instruments (2).

2.2. Zero-Order Absorbance Spectra of Nucleic Acids The intrinsic chromophores of greatest interest in nucleic acids are the purine and pyrimidine bases. The sugars and phosphates that make up the backbone do possess high energy transitions, but these are outside the range of commercial instrumentation. The bases possess extensive / electron systems, which produce intense //* transitions. The presence of nonbonding pairs of electrons on oxygen and some nitrogen atoms gives rise to additional n/* transitions, but these have rarely been observed, and are presumed to be of such weak intensity that they are hidden by the broad //* absorbance peak (1). The absorbance spectra of the individual bases obtained with a diode-array instrument in this laboratory are illustrated in Fig. 1. The bases absorb in the range of 180–290 nm, producing complex spectra with multiple underlying components, which arise from the multiple transition dipoles present in each base (5–8 transitions/base, with dipole moments ranging from ~0.03 to 0.09). Readers are referred to Table 9.2 in van Holde et al. (1) for more detail. Note that the spectra illustrated here are of mononucleotides. As indicated, the sugar and phosphate moieties also absorb, but their contribution minimally affects the base absorption spectra, with the spectra of the nucleoside and phosphate derivatives possessing only a small increase in intensity and a slight shift to longer wavelengths, compared to the individual bases (3). Thus, the absorbance spectra of nucleic acids should theoretically consist of a weighted average of the component bases. This is not the case, however, since the intensity observed is significantly less than that estimated based on composition. This hypochromism is common to all nucleic acids and reflects interaction between the bases, as discussed in Subheading 2.2.2.

256

Braun, Kueltzo, and Middaugh

Fig. 1. Absorption spectra of the five major deoxyribonucleotides. Samples were prepared in 10 mM Tris buffer, pH 7.4, at 25 µg/mL, and spectra were recorded with a Hewlett-Packard HP 8453 diode array spectrophotometer at 25°C. (A) Zero-order spectra. (B) Second-derivative spectra. (C) Second-derivative spectra of a 1:1:1:1 mixture of dGMP, dCMP, dAMP, and dTMP (6.25 µg/mL each).

2.2.1. Assessing Concentration and Purity The conventional measure of nucleic acid concentration is the absorbance intensity at 260 nm (A260), corresponding to the longest wavelength maxima of the absorbance spectrum. If base composition or sequence and size are known, a close approximation of the extinction coefficient can be calculated (see work by Borer for more detail; 4). The concentration can then be calculated employing the Beer-Lambert law: A(h) > log (I/I0) = C¡(h)l

(1)

where A(h) is the absorbance at the specified wavelength, I and Io are the final and initial intensities of light transmitted, C represents sample concentration, ¡(h) the molar extinction coefficient at the absorbance wavelength, and l the pathlength of the sample cell (5). In cases in which ¡(h) is not known, the following approximations for natural nucleic acids are usually adequate: 1 A260 U double-stranded DNA (dsDNA) = 50 µg/mL; 1 A260 U single-stranded DNA (ssDNA) = 37 µg/mL; 1 A260 U ssRNA = 40 µg/mL.

UV Absorption and CD Spectroscopy

257

Note that 1 A260 U is defined here as the quantity of polynucleotide having an absorbance of 1.0 at 260 nm, when dissolved in 1 mL buffer and measured in a 1 cm cuvet at 20°C. A typical buffer might contain 20 mM sodium phosphate at pH 7.0 and 0.1 M NaCl. However, the actual absorbance of nucleic acids is dependent on both pH and ionic strength. As the pH changes, the bases undergo protonation and deprotonation reactions, producing shifts and alterations in their absorbance spectra (5), and the extinction coefficient of DNA is decreased in the presence of salts, compared to pure water (6). For example, in 10 mM Tris and 1 mM EDTA (pH 7.4), 1 A260 corresponds to a dsDNA concentration of 45 µg/mL; in highly purified water 1 A260 U corresponds to only 38 µg/mL of the same polynucleotide (7). This ionic strength dependence necessitates careful consideration in any quantitative work. Nucleic acid preparations may contain trace amounts of protein impurities after isolation and extraction from culture, which must be quantitated and/or removed from the samples. The common method of quantitation is to measure the ratio of absorbance values at 260 and 280 nm (A260:A280). A sample that is considered free of protein contaminants has a typical ratio of 1.8 (DNA) to 2.0 (RNA) (8,9). Additionally, a measure of phenol contamination has been established using A260:A270, with a ratio of *1.2 indicating a phenol-free preparation (10). There are problems with this approach, however, as pointed out by Glasel and others (11–13), because the extinction coefficients originally used in determining acceptable ratios have been recalculated in the intervening years, leading to a possible over- or underestimation of purity. Furthermore, as described above, solution conditions such as pH and ionic strength may also affect the absorbance values (9). Nevertheless, when used with care, the A260:A280 ratio is still of some value under certain conditions. For example, ratios of less than 1.7 are indicative of impurities in the preparation, and, as the ratio decreases, the amount of UV-absorbing contaminants (i.e., proteins or phenol) are presumably increasing (13). Another approach used to estimate the degree of purity of nucleic acid preparations is the use of second-derivative spectroscopy, which is described below. In either method, purity estimates should be supplemented by electrophoretic and chromatographic analysis.

2.2.2. Application: Analysis of Thermal Disruption of Polynucleotide Secondary Structure (DNA “Melting”) Double stranded DNA undergoes a highly cooperative helix-to-coil transition at increasing temperatures. These melting curves arise from the large amount of hypochromism present in the absorption spectrum of dsDNA. The absorbance of double-stranded nucleic acids is approx 40% lower in intensity than that predicted from the base composition (3; see Subheading 2.2.1.). This hypochromic effect results from the presence of nondegenerate interactions

258

Braun, Kueltzo, and Middaugh

between the bases, leading to a loss in intensity. Degenerate interactions between the bases are also present, but these result in a splitting and redistribution of intensities, with no net intensity loss. As the strands begin to unwind and separate, the interactions between the bases decrease, increasing UV absorption to a point at which the helix is fully unwound. At high temperatures when the strands behave like random coils, however, there are still some interactions between the bases, and as a result the maximum absorbance reached is still only ~88% of the weighted average of the monomers (1). From a plot of A260 (or similar wavelengths) vs temperature, the midpoint of the melting transition can be determined and used to define the melting temperature (T m). Derivative curves of the thermal transitions (plots of dnA/dTn vs T) often exhibit fine structure indicative of a stepwise melting process (14). Differences in the stability of regions of the helix, caused by base pair composition or chemical modification of the bases (e.g., methylation or crosslinking) are reflected in the fine structure. Changes in the derivative profile may also provide information regarding the origin of these more complex structural transitions. The wavelength dependence of the individual derivative transitions, as well as the midpoint of the general melting curve, can be used to obtain information about the base composition of the regions being perturbed, or of the polynucleotide as a whole. By following the melting of DNA simultaneously at 260 and 282 nm, the fractional GC content of the melting region, and of the entire molecule, can be obtained, since the change in absorbance at 282 nm for a G-C bp greatly exceeds that of an A-T bp; the opposite is observed at 260 nm (15). Additionally, a linear correlation between base composition and Tm has been established (3,15–17). The deconvolution approach may also be employed to determine the effects of specific base modifications and/or solvent perturbation on the temperature stability of nucleic acids (18). DNA tertiary structural features, such as triplexes, hairpins, and pseudoknots, as well as properties of DNA–drug complexes (17), are all potentially detectable by absorbance-monitored melting (19). The formation of cruciform structures from inverted repeat sequences has been detected through the comparison of predicted and experimental Tm values (20). Transitions that are not accompanied by hyperchromism, as shown by Davis et al. (19), may be undetectable by derivative UV spectroscopy. CD, however, may be able to detect such transitions, and is often used as a complementary technique (see Subheading 3. for a detailed discussion). Changes in the thermal melting curve of the DNA, especially the finer transitions observed in the derivative, may provide useful information for the structural characterization of polymer:DNA complexes used in gene therapy. Melting curves have been used extensively in the evaluation of drug-nucleic acid interactions, enabling determination of 6H°, 6S°, 6G°, as well as association constants and stoichiometries of the binding reactions (14,17).

UV Absorption and CD Spectroscopy

259

The same technique has been applied to the interaction of cationic polymers with DNA. For example, the interaction of linear DNA with polylysine (PL) has been extensively characterized by this approach (21,22). More recently, Katayose and Kataoka were able to distinguish between free linear DNA and DNA complexed to a polyethylene glycol-poly(L)lysine block copolymer using the biphasic character of the unfolding transition (23). Unfortunately, the high intrinsic melting point of supercoiled plasmid DNA (Tm > 90°C) may limit the application of such analyses, since many instruments are unable to collect absorbance spectra over the entire Tm range, making the analysis incomplete or even impossible. However, open-circular and linear DNA melt at lower temperatures and are readily analyzed in complexed form (see the discussion of differential scanning calorimetry in Chapter 21 by Lobo et al.). Additionally, even though the Tm of the helix–coil transition is linearly dependent on the log of salt concentration, at high salt concentrations, the Tm actually begins to decrease (17). Thus, the Tm of supercoiled DNA can be lowered with high salt concentration or by the addition of chaotropic agents, such as NaClO4 (7.2 M), to permit thermal transitions to be detected (14,24). One must take into consideration, however, the effect of changes in ionic strength on the interaction of the DNA with the polymer when employing such methods. Changes from high to low (<10 mM) salt concentration have also been shown to produce changes in the fine structure of the melting profile (14), which can further complicate analysis. A review by Wada et al. (14) provides more comprehensive information regarding analysis of fine structure of polynucleotide thermal unfolding profiles.

2.3. Second-Derivative Absorbance Studies of Nucleic Acids As illustrated in Fig. 1, the absorption spectra of mononucleotides (and therefore nucleic acids) are composed of broad, poorly defined peaks. However, second-derivative spectra of these peaks, although often complicated, can provide more detailed information. Analysis of derivative spectra, whether involving changes in peak intensity or position, requires data of high signal:noise ratio. Often, peak shifts of interest are only of tenths or even hundredths of a nanometer in magnitude. This is often lower than the resolution of the instrument itself. Fortunately, current instrumentation is usually equipped with software that allows calculation of precise derivatives from raw data. When performing such calculations, spectra should be collected and averaged over a longer period of time (i.e., 20 s or more for diode-array instruments). The derivative spectra are then calculated, and the curves fitted to a spline function, interpolating between the raw data points, which permits derivatives to be defined to a precision of 0.01 nm, in many cases (25). The second derivative of the individual mononucleotides are complex, as a consequence of the multiple //* and n/* transitions occurring in each base

260

Braun, Kueltzo, and Middaugh

(Fig. 1). Further resolution can be achieved by employing higher (e.g., fourth) derivatives, although the increasing noise in higher derivatives can make interpretation more difficult (not illustrated). The second-derivative spectrum of DNA is dependent not only on the underlying base composition of the polynucleotide, but is also partially dependent on the secondary structure. Well-defined negative peaks at 271 and 282 nm have been observed in the second-derivative spectrum of DNA, which are absent in the spectra of both the individual mononucleotides and unfolded DNA (26). The effects of various perturbants or complexation on these peaks have received only limited attention, but they have the potential to offer some insight into DNA structural changes occurring upon various types of complexation (e.g., see Subheading 2.3.2.) (26).

2.3.1. Detection of Contaminants Second-derivative spectra of nucleic acids may contain a large number of components (Fig. 1), which could be an advantage, since they may differ between individual polynucleotides. This may enable spectral differentiation between different nucleic acid samples, as well as provide a qualitative measure of sample contamination. For example, the second-derivative spectrum of transfer RNA possesses a strong negative peak at 258 nm, with small secondary peaks near 265 nm (25). In contrast, dsDNA exhibits a series of peaks between 220 and 290 nm. Other common contaminants, such as trace proteins and phenol, which at times cannot be accurately detected using the A260:A280 ratio, can also be quantitatively analyzed through examination of the second derivative. Readers are referred to the work of Mach et al. (26) for an in depth analysis of this approach.

2.3.2. Analysis and Quantitation of DNA in DNA–Cationic Polymer Complexes A major problem in the analysis of DNA–cationic polymer complexes by UV-absorbance spectroscopy is the presence of a significant light scattering component, as the size of the particles nears the wavelength of the measuring light (250–300 nm). Absorbance spectra of complexes at typical formulation concentrations include a new broad spectral component that significantly distorts the normal absorbance spectrum. This can be seen in Fig. 2, which illustrates the absorption spectra of DNA–cationic lipid (CL) complexes of different charge ratios. An increase in baseline is clearly seen in the complexes at ratios of 0.8 and above. This reflects the presence of a Raleigh scattering component that typically displays a h–4 dependence. This spectral component can be at least partially removed by plotting log A vs logh in absorbance-free regions of the spectrum (e.g., 300–400 nm), extrapolating the resultant straight line into absorbing regions and subtraction of the data from the total spectrum (see Chapter 22 by Wiethoff and Middaugh on light scattering for more detail

UV Absorption and CD Spectroscopy

261

Fig. 2. Absorption spectra of DNA and DNA:DOTAP complexes at ± charge ratios of 0.2–5.0.

on the effects of charge ratio on complex size). Current instrumentation can often provide automatic correction of this type. Unfortunately, as particles get larger, another phenomenon known as absorption flattening, which results from the obscuration of one particle by another can further distort the spectra, (this is dealt with in detail later in Subheading 3.2.2.). A common method that can be used to reduce this problem is the reduction of particle size by methods such as sonication. Unfortunately, this is probably not a viable approach for gene delivery complexes, because alteration of the complex structure might also result. An alternate approach to this problem involves the use of second derivatives, since broad spectral bands, such as light scattering components, are not reflected in second-derivative spectra. Second-derivative peaks can also be directly related to the concentration of nucleic acids through the use of a simple calibration curve, which permits the quantitation of nucleic acids within complexes, despite the presence of significant light scattering (25). Possible effects of any complexing agent on the microenvironment of the nucleic acid bases and subsequent perturbation of their spectra also need to be considered. In addition to quantitation of DNA content in complexes, the second derivative of the UV-absorbance spectrum may be used to probe alterations in DNA structure as complexes are formed. The effect of different complexing agents may also be

262

Braun, Kueltzo, and Middaugh

examined. The potential of this approach is illustrated by examining the effect of cationic lipid upon the UV-absorption spectrum of DNA (see Subheading 3.3.2. for details of complex formation). A typical second-derivative spectrum of DNA (Fig. 3A) contains a series of peaks between 250 and 300 nm. The five major minima occur at approx 252.7, 260.2, 270.9, 282.2, and 290.2 nm. Changes in these peak positions may be used to distinguish between different types of complexes, or to provide information on the effect of complexation on the structure of the DNA. Analyses of plots of peak position vs charge ratio (Fig. 3B) demonstrate marked alterations in peak positions as complexes of lipid and DNA are formed. Determination of peak positions in this analysis was not as straightforward as seen for oligonucleotides or purified DNA plasmid. The heterogeneity of the complexes, coupled with the fact that the spectra of the uncoupled oligonucleotides may be affected in multiple ways by the complexing agent in different regions of the plasmid, introduces additional small peaks and noise into the spectra. Simple smoothing (using a seven-point Savitzgy-Golay function) removes much of the noise, however, allowing shifts in the five minima to be monitored. The peaks that behave most consistently, at least regarding peak shape and absence of new small shoulders, are those at 260, 282, and 290 nm. More studies employing this approach are needed, including establishing the systematic effects of solution environment on the absorption spectrum of plasmids, before any definite conclusion can be drawn about the structural basis of the changes observed, but these preliminary results suggest that such studies may prove to be of significant utility. 3. Circular Dichroism Historically, CD spectroscopy has been one of the preferred methods for obtaining structural information about optically active compounds. In recent years, this technique has been applied to the analysis of structural characteristics of macromolecules, both proteins and nucleic acids, with much success. In the analysis of dsDNA, CD can be used to distinguish among the different helical forms of DNA, as well as to monitor the unfolding of the double helix and perturbation of DNA secondary structure upon alteration of the environment by pH, salt, solvent, and so on. The sensitivity of CD to DNA conformation therefore makes it a powerful technique when employed in the analysis of DNA complexes, as exemplified by its application to the characterization of DNA–cationic lipid interactions as described in Subheading 3.3.

3.1. Theory and Instrumentation CD is primarily used to evaluate structural aspects of optically active compounds. A thorough discussion of the origin of the CD of these compounds, especially complex spectra such as that of nucleic acids, is beyond the scope of this chapter. For a more in-depth treatment, readers are referred to the work of Keller and Bustamante (27–29). However, there are some key concepts that

UV Absorption and CD Spectroscopy

263

Fig. 3. Second-derivative absorption spectra of (A) supercoiled DNA and (B) DNA:DOTAP complexes at increasing concentration of DOTAP. Complex charge ratios are indicated. X denotes points that were not acquired because of splitting of the minima into two peaks.

form the basis of this approach. CD arises from the differential absorption of right and left circularly polarized light by a chromophore. More formally, this differential type of measurement is given by: 6A = AL – AR = ¡L C l – ¡R C l

(2)

where AL and AR are the absorbances of left- and right-handed circularly polarized light, respectively, ¡L and ¡R represent the corresponding extinction

264

Braun, Kueltzo, and Middaugh

coefficients, C the molar concentration, and l the thickness of the absorbing layer. Values are usually reported as either ¡L – ¡R (6¡) or as molar ellipticity [e]. The two are directly related by: [e] = 4500 // [¡L – ¡R] ln 10 = 32986¡

(3)

To better understand this concept from a molecular perspective, one can consider CD from a quantum mechanical standpoint. The physical quantity of greatest interest in a CD measurement is the rotational strength, which is the imaginary part of the dot product of the electric and magnetic transition dipole moments. The latter quantity can simply be thought of as a light-induced current loop and the former as a light-induced oscillating dipole. The interaction between the two produces a helical charge circulation that is dramatically facilitated in self-helical structures, and accounts for the strong CD signals of helical macromolecules (Fig. 4). The overall rotational strength of polynucleotides is further influenced by the coupling of the individual transitions of the stacked base chromophores, as a consequence of their intimate proximity, which causes splitting of some of the observed peaks. A typical CD instrument (a spectropolarimeter) is usually capable of measurements over a UV-visible wavelength range of 180–700 nm, or, with an optional photomultiplier tube (PMT) into the near-infared region between 700 and 1000 nm. A typical optical system for such an instrument is shown in Fig. 5. UVand visible light produced by a xenon lamp is passed through a double monochromator and a series of prisms to produce monochromated and linearly polarized light. In modern instruments, a sinusoidally controlled, piezoelastic birefringence modulator circularly polarizes the plane polarized light. The now-alternating left and right circularly polarized light passes through the sample, and a PMT detects the transmitted light. The PMT converts the light into both an AC and DC signal (Fig. 6), with the former representative of the CD and the latter corresponding to the light intensity. Since the instrument measures light intensity (i.e., not ¡ directly), the equivalent form of Eq. 3 is: [e] = [4500 // l C] [ln 10] [log IR/IL]

(4)

where IL and IR are the light intensities in the corresponding directions of polarization. Practically, the difference between IL and IR is small and difficult to determine with much accuracy. Therefore, the mathematically equivalent expression: [e] = [4500 / / l C] [ln 10 (S / IA log e)]

(5)

is employed. In this case, the DC signal (IA) is kept constant by varying the PMT voltage; the AC signal (S) is representative of the CD (see Fig. 6). The DC signal is kept constant by a feedback loop between the PMT and power supply. The effect of this feedback loop is to increase the sensitivity of the PMT in response to any decrease of the intensity of light reaching the PMT.

UV Absorption and CD Spectroscopy

265

Fig. 4. Origin of CD signals. The electronic and magnetic transition dipoles of a molecule (A) combine to produce a helical charge circulation (B), a process greatly facilitated by optically active helical molecules. The wave functions of the ground and excited states of the asymmetric molecules are represented by s0 and sa, respectively; µ is the electric dipole moment, and m the magnetic dipole moment operator. The rotational strength (Roa, a measure of the CD intensity) is actually the imaginary part of the dot product of the electronic and magnetic transition dipoles. Adapted from ref. 5.

Fig. 5. Typical optical system for a CD instrument. Mirrors (M), light source (LS), slits (S), prisms (P), lens (L), filter (F), CD modulator (CDM), and photomultiplier tube (PMT) are indicated.

3.2. Circular Dichroism of Nucleic Acids 3.2.1. Spectral Characterization Investigations of gene delivery complexes by CD are primarily an evaluation of the changes that occur in the nucleic acid spectra. Therefore, a brief discussion is warranted of the salient features of nucleic acids that give rise to their CD. As in conventional absorption spectroscopy, the purine and pyrimidine bases are the major nucleic acid chromophores of interest. These bases contain a plane of symmetry, and therefore are not intrinsically optically active. However, the asymmetric sugars in the backbone of both RNA and DNA

266

Braun, Kueltzo, and Middaugh

Fig. 6. Relationship between the intensity of transmitted light and a CD signal. The intensity of left (IL) and right (IR) circularly polarized light, average light intensity (IA), and the CD signal (S) are shown.

induce a weak optical activity in their corresponding nucleotides. Furthermore, the base stacking inherent in the helical structure of both single- and doublestranded polynucleotides enhances the optical activity. Since the CD of nucleic acids is dependent on base–base interactions, both intrastrand stacking and interstrand pairing, the technique is particularly sensitive to changes in secondary structure. The solution helical structure of nucleic acids is polymorphic, varying in pitch, handedness, diameter, and the number of residues per turn (30). This polymorphism gives rise to various distinct secondary structure types, each with an accompanying unique CD spectrum. Knowledge of the spectrum associated with each secondary structure type is critical to interpreting the spectra of complexes. The spectral region of greatest interest is from 190 to 300 nm where the aromatic bases absorb. The CD spectra of most known secondary structures have been assigned based on synthetic polymers for which the structures were previously established by X-ray diffraction analysis. These major secondary structure types, which have been assigned the designations A, B, C, D, T, and Z, can be recognized from their unique spectra. The D and T forms, subsets of the B-form family, are rarely observed, and thus will not be discussed further (30). CD spectra of the more common A, B, and C forms, as well as the Z form, are illustrated in Figs. 7–9 (31,32). Nucleic acid secondary structures are generally characterized by the handedness of the helix, number of bp/turn and pitch of the bases. The pitch of the helix is the distance parallel to the helix axis over which the

UV Absorption and CD Spectroscopy

267

Fig. 7. The CD of calf thymus DNA at pH 7.0 in 10 mM sodium phosphate buffer, pH 7.0, in the B form with 10.4 bp/turn (_ _ _), in 6.0 M ammonium fluoride, 10 mM sodium phosphate buffer, pH 7.0, in the B form with 10.2 bp/turn (_____), and in 0.67 mM sodium phosphate buffer, 80% trifluoroethyl, pH 7.0, in the A form (__ • __). Sprecher, et al. Conformation and circular dichroism of DNA. Biopolymers 18, 1009–1019. Copyright 1979, reprinted with permission John Wiley and Sons, Inc.

helix makes one turn (33). Nucleic acid helices are either right- or left-handed, with the A, B, and C forms in the common right-handed configuration; the Z form is of opposite chirality. Changes in any of these parameters can affect the secondary structure and therefore the CD spectra of dsDNA. The B form is the most common solution conformation of DNA and can have two distinct states. One contains 10.4 bp/turn while the other is slightly more tightly wound with 10.2 bp/turn. This more tightly wound form is induced by alcoholic solvents or certain salts and is formed in the presence of histones (Fig. 7; 34). CD spectra of this form are characterized by a positive band at 275 nm, a negative band near 240 nm and a crossover point at about 258 nm. The two B forms have similar spectra with the exception of the 275 nm peak, which collapses in the 10.2 (bp/turn) form (Fig. 7; 35). Early work incorrectly attributed the 10.2 bp/turn collapse of the 275 nm peak to C-form DNA (34,35). Another common secondary structure is the A form (11 bp/turn), which is the natural form of RNA, but can also be induced in sodium DNA by dehydration

268

Braun, Kueltzo, and Middaugh

Fig. 8. CD spectra of calf thymus DNA in 1 mM lithium chloride at high concentrations of ethanol: 80% ethanol (_____), 85% ethanol (_ _ _), 90% ethanol (•••), and 95% ethanol (__ • __). The CD in 95% ethanol is presumably characteristic of C-form DNA. Bokma, et al. CD of the Li-salt of DNA in ethanol/water mixtures: evidence for the Bto C-form transition in solution. Biopolymers 26, 893–909. Copyright 1987, reprinted with permission John Wiley and Sons, Inc.

with 80% ethanol or 80% 2,2,2-trifluoroethanol (TFE) (34,35). The CD of the dehydrated DNA A form shown in Fig. 7 is characterized by a positive band near 260 nm, a moderately intense negative band at 210 nm, and an intense positive band near 190 nm. The C form (8.8-9.7 bp/turn) of DNA has been determined by X-ray diffraction studies of ethanol-induced dehydrated lithium DNA (Fig. 8; 34,35). As the ethanol concentration increases from 0 to 65%, a decrease of the 275-nm peak occurs (not illustrated). This trend reverses as the concentration rises from 65 to 90%, then decreases rather abruptly at 95% ethanol to produce a spectrum that is thought to be representative of the C form. This “C-form spectrum” has a reduced trough at 245 nm, but an intermediate magnitude at 275 nm, and is similar to that of the 10.2 bp/turn B form, but the 275 nm peak is not as dramatically reduced (see Fig. 7). Thus, since both the 10.2 bp/turn B and C DNA spectra show significant reductions in the 275-nm peak, probably the best diagnostic feature of the C form is the decrease in intensity of the 245 nm trough. The left-handed helical conformation of DNA is known as the Z form (12 bp/ turn). This quite different state of DNA is formed by GC-rich oligonucleotides. The Z-form CD spectrum (Fig. 9) has a negative band near 290 nm, a positive

UV Absorption and CD Spectroscopy

269

Fig. 9. CD of poly-d(GC) • poly-d(GC) in 10 mM sodium phosphate buffer, pH 7.0, in the B form (•••), in 0.67 mM sodium phosphate buffer, 80% trifluoroethyl, pH 7.0, in the A form (_ _ _), and in 2 M sodium perchlorate, 10 mM sodium phosphate buffer, pH 7.0, in the Z form (_____). Riazance, et al. Evidence for Z-form RNA by vacuum UV circular dichroism. Nucleic Acids Research 13, 4983–4989. Copyright 1989, reproduced by permission of Oxford University Press.

band at 260 nm, a negative band between 195 and 200 nm and a crossover point somewhere between 180 and 185 nm. The most characteristic features of a lefthanded helix are the band between 195 and 200 nm and the crossover point. By comparison, all of the right-handed nucleic acid forms have an intense peak near 190 nm and a crossover at approx 200 nm. Nucleic acid secondary structure is highly polymorphic with both base composition and solvent conditions having significant effects on conformation. Therefore, some variation in the actual spectra of natural polynucleotides is expected, compared to reference spectra of model polymers. Sometimes dramatically altered spectra are induced under conditions in which the charge on the DNA is extensively neutralized, resulting in condensation (a major reduction in volume of the DNA polymers). The spectra of highly condensed DNA are termed “psi”-type (an acronym for polymer and salt-induced). These

270

Braun, Kueltzo, and Middaugh

spectra are characterized by significant shifts in peak position and spectra of 10–100× the intensity of dispersed DNA. These unusual spectra are considered in more detail in Subheadings 3.4. and 3.5. Supercoiling of various types also influences nucleic acid CD spectra. In chromatin, the 275-nm peak is depressed to approx 70% of its noncollapsed B-form value. MacDermott and Drake (36) produced a variety of topoisomers of bacterial plasmids containing both positive and negative supercoils. The CD spectra of these topoisomers show that negative supercoiling (the natural state) increases the magnitude of the 275-nm peak, but that both no supercoiling and positive supercoiling reduce the magnitude at 275 nm with a greater reduction occurring in the latter (36). The 220-nm peak is also sensitive to superhelicity, but the magnitude of the trend is opposite to that seen in the 275-nm peak; the 245-nm trough is invariant.

3.2.2. Data Interpretation CD suffers from a number of artifacts that can dramatically perturb the true spectrum. This often results in confusion in determining the structural basis of any observed spectral changes, especially in the case of larger structures, such as those often encountered with nonviral gene delivery vehicles. The two most common of these artifacts, absorption flattening (Duysen’s flattening) and differential light scattering, need to be familiar to all workers investigating systems in which particles have a diameter greater than one-twentieth the wavelength of the incident light. Absorption flattening occurs as a result of an inhomogeneous arrangement of chromophores (37–39). Recall that the Beer-Lambert law relates the intensity of transmitted light to the extinction coefficient of the absorbing species (¡), pathlength (l), and concentration (C), and assumes a homogenous distribution of chromophores: Log (I /I0 ) = ¡ l C

(6)

In an inhomogeneous solution, the initial intensity (I0) of light is reduced when some particles shadow others. The resulting spectrum appears flattened. A red shift is also often produced by this phenomenon because of the wavelength dependence of any scattered light. Since the shorter wavelengths of light scatter more readily than the longer, the lower frequencies contribute less to the CD spectra, producing an apparent red shift. Since CD depends on the polarization of light, it should be recognized that left- and right-handed light are isotropically scattered in absorption flattening. Differential light scattering arises from a different phenomenon. In this case, right or left circularly polarized light is preferentially scattered distorting CD spectra (40–45). Differential scattering usually has its most significant effects at longer wavelengths, where the particle size is closer to the wavelength of the

UV Absorption and CD Spectroscopy

271

incident light. A distinguishing characteristic of differential light scattering is an increase in molar ellipticity at longer wavelengths, which results in “tails,” i.e., an increase in molar ellipticity outside of an absorption band. Reich et al. (43) have treated this phenomenon in some detail, evaluating the scattered light with various techniques (32–35). The simplest approach to the evaluation of differential scattering is to vary the distance of the sample cuvet from the photomultiplier. This technique allows one to control the amount of forward-scattered light collected, with variation in the intensity of CD signals evidence of differential scattering. As described by Reich et al. (43), however, forward-scattered radiation is a minor part of the total scattered light, with side scattering accounting for the majority. This method can therefore demonstrate that differential light scattering is significant, but can not rule out differential scattering, even if no detector distance dependent variation in spectra is observed. In contrast with absorption flattening, a complete correction of differential scattering can be obtained with fluorescence-detected CD, which involves adding an achiral fluorescent reagent to the sample. Any scattered light will excite the fluorophore, and the emitted light can be collected and measured (43).

3.3. Application: Study of DNA–Polymer Complexes 3.3.1. Materials Studies of nonviral gene delivery complexes by CD involve important choices in both instrument configuration and mechanism of complex formation. On the instrumentation side, an ideal cell-holder should have several properties. It should be temperature-controlled over a range of 0–100°C. Temperature control in conventional instruments is accomplished using either a liquid-jacketed cuvet holder-employing a circulating water bath, or a Peltier device, which is somewhat more convenient. Software is now available that allows programmable changes of temperature. Although this is frequently done in a continuous ramping mode, it is better to incubate for a fixed period at each temperature, to ensure equilibrium occurs between time-points and the sample and jacket temperatures. The option for controlled stirring of the sample is useful in some applications, as is the ability to vary the distance between the cell and PMT, to evaluate differential light scattering, as described in Subheading 3.2.2. Another important consideration is the choice of cuvet. CD cells come in a variety of configurations and pathlengths, but must be made of an appropriate material (e.g., quartz) to obtain transparency in the farUV range. The preferred configurations are generally either cylindrical or rectangular, with pathlengths of 0.01–1 cm. It is important to balance concentration and pathlength, to obtain an ideal signal. As defined by the Beer-Lambert law (Eq. 6), these two quantities are inversely related. Therefore, by adjusting concentration and pathlength, a light intensity compatible with the sensitivity of the PMT can be obtained.

272

Braun, Kueltzo, and Middaugh

Many instruments have a second channel, to monitor the power supply voltage that is used to increase the sensitivity of the PMT. This second channel can be used to detect saturation of the PMT, indicating that the signal:noise ratio has become too low to obtain useful information. This situation is usually obvious since noise begins to dominate the spectrum. It often occurs in the farUV region since solutes, oxygen and most molecules begin to significantly absorb at lower wavelengths (e.g., below 200 nm). Decreasing either the concentration or pathlength of the cell, as well as the concentration of absorbing solutes (which can also be replaced with less absorbing species), can often partially alleviate such problems, permitting spectra to be resolved into the farUV (e.g., down to 180–185 nm). One must be careful, however, not to overly attenuate the rest of the CD spectrum as a consequence of any such changes. The heart of any gene delivery complex, from a CD standpoint, is the nucleic acid component, although cationic polymers such as polylysine will also produce strong CD signals in the far UV region. The three most common types of DNA employed in structural studies are chromosomal, linear, and plasmid DNA. Common forms of chromosomal DNA studied are often derived from either calf thymus or salmon testes. Unless chemically or enzymatically modified, chromosomal DNA will be associated with histones and other strongly bound proteins, which will potentially contribute to the CD spectra. Plasmids are extrachromosomal circles of DNA, usually of bacterial origin (e.g., Escherichia coli); the most common form of linear DNA is derived from an E. coli phage (h – DNA). From a gene therapy perspective, plasmid DNA is by far the most common type of DNA used. Plasmids exist in three common topoisomers: supercoiled, open-circular, and linear. All three forms can be found in varying degrees in most preparations. They can be individually purified or enzymatically prepared, but it is most common to employ preparations that are >90% supercoiled and assume that any observed properties are primarily those of the biologically active supercoiled form. Determination of molar ellipticity (i.e., normalized or intrinsic CD spectra) requires an accurate measurement of nucleic acid concentration. The absorption of nucleotides is strongly dependent on the pH and ionic strength of the buffer (7,9), as described in Subheading 2.2.1.; therefore, one needs to carefully calibrate concentration measurements if CD spectra of nucleic acids are to be quantitatively comparable. Furthermore, one needs to be careful that spectral changes, produced directly by altering the solvent environment are not confused with structural changes. In the CD experiments described here, the solution concentrations of the nucleic acid solutions were determined by their absorbance at 260 nm (A260) and calibration curves were employed as needed. Investigators must know the molarity of nucleic acid solutions to prepare complexes at specified charge ratios, an important parameter in gene delivery

UV Absorption and CD Spectroscopy

273

Table 1 Molecular Masses of Nucleic Acids Deoxynucleotide base 324.5 g/mol Ribonucleotide base 340.5 g/mol 1 kb dsDNA (Na salt) 6.5 × 105 g/mol 1 kb ssDNA (Na salt) 3.3 × 105 g/mol 1 kb ssRNA (Na salt) 3.4 × 105 g/mol h DNA (48,502 bp) 3.1 × 107 g/mol qX-174 DNA (5386 b) 3.6 × 106 g/mol 6 E. coli DNA (4 × 10 bp) 3.1 × 109 g/mol

efficiency. Furthermore, to aid comparison among CD spectra the data are usually reported normalized as molar ellipticity. Therefore, estimated molecular weights of various commonly used nucleic acids are summarized in Table 1. These values are approximations based on the average weight of the four nucleotides in either RNA or DNA. A wide variety of cationic polymers and lipids are currently being used as vehicles for nonviral gene delivery. Some general classes of materials in use include peptide-based polymers (e.g., PL), peptoids (N-substituted polyglycines), branched-chain polymers (e.g., polyethylenimine and polyamidoamine dendrimers), and cationic lipids. These materials complex with DNA primarily by electrostatic interactions of positively charged functional groups with the negatively charged phosphates of the DNA backbone. Most of the materials used to enhance the delivery of DNA for gene therapy applications are not optically active, and therefore allow observation of the DNA CD without interfering spectral contributions. An important exception to this is peptide-and protein-based polymers. The bulk of their stronger CD contribution, however, occurs below 250 nm so that changes in the 275-nm band can often be detected with minimal interference. Although polypeptide aromatic side chains have CD peaks in this region, their intensity is usually low. If the CD spectra of these materials are of low intensity compared to the nucleic acids, their spectra can be subtracted from the complex spectra with standard software analysis tools, although this involves the tenuous assumption that their CD spectra are not altered during the complexation process.

3.3.2. Methods: DNA Complexes One of the most effective ways to use CD to investigate gene delivery complexes is comparison of CD spectra of complexes of different polymer compositions. This method is most simply performed as a titration of the gene delivery agent into a known amount of DNA. This results in a series of spectra illustrating the state of DNA in a known native secondary structure (e.g.,

274

Braun, Kueltzo, and Middaugh

solution B form) and spectra representative of DNA within the gene delivery complex at various ratios of cationic polymers to DNA. Most nonviral gene delivery vehicles are heterogeneous, noncovalent complexes between DNA and the cationic delivery agent. Therefore, consistency in the method used to form the complexes is critical for reproducibility of spectra. The following preparation procedure for cationic lipid–DNA complexes demonstrates some of the important parameters, but these will vary, depending on the polymer employed. Cationic lipids are effective delivery agents that have already been employed to deliver genes in a number of human clinical trials. CD studies of two cationic lipids, dimethyldioctadecylammonium bromide (DDAB) and dioleoyl-1,2-diacyl3-trimethylammonium propane (DOTAP), are presented here as examples. Cationic lipids interact at least partially with DNA through an electrostatic interaction between their positively charged headgroups and the negatively charged phosphate backbone of the DNA. Complexes are typically prepared at charge ratios (+/–) of 0.25–2. These charge ratios are calculated by assuming a single positive charge for each cationic lipid. Lipids were prepared as liposomes in 10 mM Tris-HCl buffer at pH 7.4. The lipid, solubilized in chloroform, was added to a glass vial, and dried with nitrogen to a thin film. At this point, weights were recorded, to provide a measure of the amount of lipid for future calculations. The lipids were then dispersed in buffer and extruded 10× through a 0.22-µm filter. Temperatures were maintained above the glass transition of the liposome. Supercoiled (>95%) DNA was prepared by diluting a stock solution to the required concentration, typically 10–100 µg/mL. An optimal signal:noise ratio can be achieved at 20–100 µg/mL in a 1-mm pathlength cell. The DNA solution is assayed by absorbance at 260 nm after dilution, and the experimental solution concentration calculated from the result. The DNA concentration is usually calculated in terms of the molar concentration of deoxynucleotide bases, employing an average molecular weight of 324.5 g/ mol (see Table 1). Complexes are individually prepared by adding equal volumes of liposome and DNA solution at the appropriate concentrations. The lower concentration solution is always added to the larger, to avoid passing through charge neutrality, which can result in precipitation of the complexes. At some charge ratios, the complexes are physically unstable, and are therefore prepared immediately prior to spectroscopic analysis.

3.3.3. Results The formation of lipoplexes, consisting of the cationic lipid DDAB, and a DNA plasmid, results in significant changes in the native CD spectra of DNA. As Fig. 10 illustrates, with increasing charge ratio, a peak intensity shift occurs in the 210, 220, 245, and 275 nm bands. In addition, shifts in peak position also occur. Up to a 0.75 charge ratio (+/–), the DNA manifests a continuous change in the 275-nm CD

UV Absorption and CD Spectroscopy

275

Fig. 10. Plasmid DNA (9.1 kb) complexed to increasing amounts of DDAB liposomes in 10 mM Tris buffer at pH 7.4 and at charge ratios of 0, 0.25, 0.50, 0.75, and 1.0.

peak with the magnitude decreasing as the charge ratio increases. The shifts occurring in the 210, 220, and 245 nm bands, however, do not occur directly in parallel with the 275-nm peak changes, although they do seem to follow a trend. With increasing charge ratio, the intensity decreases in the 245- and 210-nm troughs, and increases in the 220-nm peak. The exception is the sample that precipitated at a charge ratio of 1. In this case, the 245- and 275-nm peaks follow the established trends, but the 210- and 220-nm peaks decrease to near zero. Commonly, the spectra of precipitated materials dramatically decrease in intensity. That sample material is grossly aggregated or precipitated may not, however, be immediately apparent from the shape of the observed spectra. Another important observation is appearance of the red shift that occurs in the 245 and 275 nm bands, but is absent in the 210- and 220-nm peaks. As discussed previously, a characteristic feature of absorption flattening is a red shift. Experiments described in Subheading 3.5., however, suggest that absorption flattening does not significantly contribute to these spectra. The addition of neutral helper lipids often provides a significant enhancement in gene transfer efficiency of cationic lipids (47). Figure 11 illustrates the CD spectrum of lipoplexes consisting of DOTAP present in an equal molar ratio with the helper lipid L-_-dioleoyl phosphatidylethanolamine (DOPE). Note that the actual concentration of lipid is twice the previous example at any

276

Braun, Kueltzo, and Middaugh

Fig. 11. Plasmid DNA (9.1 kb) complexed to increasing amounts of DOTAP:DOPE liposomes in 10 mM Tris buffer at pH 7.4 and at charge ratios (+/–) of 0, 0.25, 0.50. and 0.75.

particular charge ratio. The spectra are similar to the previous lipoplexes with conservation of the native DNA bands at 210, 220, 245, and 275 nm. The spectral changes in the 275-nm peak are similar to those described in the DDAB lipoplexes with a decrease in magnitude correlating with increasing charge ratio. In fact, this feature seems conserved in the CD spectra of lipoplexes containing various combinations of cationic and helper lipids. A distinguishing feature, in contrast to the DDAB lipoplexes, is the 245-nm trough, which increases in magnitude with increasing charge ratio. This characteristic can be assigned primarily to the influence of the helper lipid, DOPE. The remaining bands, at 210 and 220 nm, manifest increasing and decreasing magnitudes, respectively, as described in the previous example. The utility of CD in such studies is evident from the finding that lipoplexes of similar charge ratios, but differing in lipid composition are both similar (i.e., 275-nm peak features) and different (i.e., 245 nm trough features) in their CD spectral features.

3.4. Applications: DNA–Cationic Polymer Complexes The other major class of nonviral gene delivery agents is the cationic polymers. This class contains a wide variety of compounds, but PEI and polylysine have been frequently used, and are discussed as representatives of this class. Polyethylenimine is a linear or branched-chain polymer containing both primary and secondary amines. The molecular weight may vary considerably, with nonviral vehicles being produced from PEI of 2–800 kDa, although a mol

UV Absorption and CD Spectroscopy

277

wt )2 kDa requires formulation with an endosomolytic agent, to ensure efficacy (48). The higher-mol-wt forms of PEI have been found to be more efficient delivery vehicles, but this highly branched polymer has been shown to have some cytotoxicity (49). In contrast, low-mol-wt PEI (e.g., ~12 kDa) has significantly diminished cytotoxicity. The generally accepted charge ratios used for most cationic nonviral delivery agents are less applicable to PEI because of its multiple pKa values, which are altered by nucleic acid binding (48). Complexes are therefore formed based on a PEI nitrogen:DNA phosphate ratio (N:P). Figure 12 illustrates the spectra of small (2 kDa) linear PEI complexed to plasmid DNA at various N:P ratios. Even at the lowest N:P ratio examined (0.5), the 275-nm native DNA peak decreases and red shifts (~280 nm), producing strong negative ellipticity at ~260 nm. This new trough reaches a maximum at 2.0 N:P ratio. At an N:P ratio of 6.7, the 260-nm trough is reversed, decreasing in magnitude, with a corresponding increase in the 285–290 nm peak. The CD spectrum of the 10- and 20-N:P ratio complexes overlay one another, implying the presence of a stable form. The large peak shifts and changes in intensity are more characteristic of psi-type CD spectra than those seen in cationic lipid complexes, although not as dramatic as those that arise in polymer- and salt-condensed DNA (29,44). The polylysines are another frequently used class of cationic polymer. These linear peptides have been extensively studied both as simple models of DNA:protein interactions and for their DNA condensing properties. The CD spectra of polylysine–DNA complexes (Fig. 13) show a dramatic effect on the bands of uncomplexed plasmid with addition of polylysine. The 275-nm peak decreases in magnitude, which seems to be a common observation as the molar ratio of various cationic delivery agents’ increase. The 258-nm crossover point of B-form DNA, however, red shifts, as seen in B A s transitions. The 210- and 220-nm bands both decrease in their negative ellipticity values, with increasing charge ratio. Weiskoff et al. (50) examined the CD of polylysine–DNA complexes by various preparation methods. At high salt concentration (1 M NaCl) and a 0.72 charge ratio, the complexes exhibited a psi-type CD spectra. At this charge ratio, the characteristic 275-nm peak of B form is lost, and a large negative psi-type trough is produced at 270 nm. At lower salt concentrations ()0.5 M NaCl), however, the polylysine–DNA complexes had more moderate spectral changes. At low salt, the major spectral feature was a decrease in the 275-nm peak similar to that seen here.

3.5. Discussion The nonviral gene delivery complexes discussed are generally large inhomogeneous structures. Therefore, at this time, it is difficult to correlate spectral changes to specific alterations in supramolecular structure. However, the following discussion may provide some insight into how the formation of

278

Braun, Kueltzo, and Middaugh

Fig. 12. Plasmid DNA (9.1 kb) complexed to PEI at N:P ratios of 0, 0.5, 1, 2, 6.7, 10, and 20.

Fig. 13. CD spectra of polylysine/DNA complexes in 2 mM sodium phosphate buffer at pH 7.4 and at charge ratios (+/–) of 0, 0.2, 0.4, 0.6, and 0.8.

supramolecular structures influences CD spectra. When DNA complexes to cationic polymers, aggregates in the size range of 50–200 nm are typically formed (see Chapter 22 by Wiethoff and Middaugh). The CD of DNA present

UV Absorption and CD Spectroscopy

279

in such structures will generally differ significantly from that of free polynucleotides. The fact that such spectra are significantly distorted has been known for some time, as the result of studies of entities such as viruses and chromosomes. In general, one finds that as the size of such complexes increases, tails begin to appear in regions outside absorption bands. Larger aggregates also often display CD bands with dramatically altered intensities (usually larger) and distorted peak shapes. As indicated previously, the latter are often referred to as psi-type effects. The origin of the tails is now well accepted to be caused by differential scattering of either right or left circularly polarized light. This phenomenon, however, is generally thought to contribute only a limited amount to the abnormal spectra seen in DNA–polymer complexes. Another possible origin of such spectral alterations could be Duysens’ flattening, which arises from a screening phenomenon among the dispersed particles. The authors have examined the various complexes discussed above over a range of concentrations and pathlengths. If absorption flattening is responsible for the spectral changes seen upon complex formation, variations in spectra should be observed as these parameters are varied, but this is not seen. The authors therefore conclude that a statistically uneven distribution of complexes does not play a major role in the CD spectra induced by polymer binding. Another possibility is that the polymers produce actual structural changes in the structure of the DNA itself (e.g., a B A C transition) (44). In general, however, the newly induced CD spectra cannot be clearly identified as any known structural form. Furthermore, X-ray diffraction studies of DNA complexes; which display typical distorted spectra, are seen to be present in the normal B form (50). Currently, the most attractive hypothesis is that large-scale chiral elements in the structure of the complex interact differently with differentially handed, circularly polarized light. These elements, or liquid crystalline phases, have been reported in both condensed DNA and cationic lipid–DNA complexes (47,51). The ordered packing is generally of two types, with nematic and hexagonal phases aligned along the long axis of neighboring molecules; second-order cholesteric phases align similarly, with the addition of a twist or helix perpendicular to the first-order alignment (52). Supramolecular helices of cholesteric phases propagate one circular polarization of light in preference to the other, in a wavelength-dependent fashion (52). The basic idea is that, when the size of any DNA-containing particle approaches that of the incident light, the overall chirality of the complex will have a significant effect in either increasing or reducing the CD signals. That chiral order within the complex can be responsible for these spectral perturbations has been directly demonstrated in a simple model system (41). When a highly concentrated solution of DNA is pressed between two quartz plates and rotated, a psi–like spectrum of

280

Braun, Kueltzo, and Middaugh

both intensity-enhanced and shape-distorted peaks is produced. If the rotation is reversed, the sign of the psi–like peak is reversed with little alteration in peak shape. If the film is twisted back and forth numerous times to randomize the large-scale chiral order, essentially natural spectra result. Thus, it is clear that in at least some cases, large-scale chiral order can produce major changes in CD spectra of DNA. These effects appear to primarily arise from electrostatic coupling between chromophores. Such long-range coupling appears to be possible when the complex is large (>h/4), has at least a certain density of chromophores (>1/nm3), and has a defined three-dimensional structure (29). The critical chromophoric interactions usually seem to be the result of a combination of static dipole, intermediate, and radiation coupling. As a consequence, structural factors, such as the size, pitch, handedness, density of chromphores, and internal order of the aggregate, all control the sign, shape, and magnitude of the altered CD spectra seen in DNA–polymer complexes. However, most cases in the literature find major increases in the intensity of the CD spectra induced by polymers. In systems containing agents as diverse as various cationic homopolypeptides (e.g., PL, polyarginine, polyhistidine), spermine, histones, polyethylene glycols, as well as high concentrations of hydroxylcontaining organic solvents and certain salts, psi-like spectra are seen in which the DNA CD peaks are dramatically enhanced. This, in fact, is not what is seen when gene delivery polymers are complexed to supercoiled DNA plasmids over the range of charge ratios used to prepare active delivery complexes. One possible explanation may be that the long-range chirality of the complexes formed is opposite to that usually observed, resulting in the observed decreases in spectral intensity. Another possibility is that the gene delivery complexes lack the high degree of chiral order seen in the other systems, thus permitting more subtle effects to dominate. For example, X-ray studies indicate that cationic lipid–linear DNA complexes exist as alternating layers of lipid bilayers and DNA (47,51). These supercoiled plasmids could in effect be stretched out (i.e., pseudo-linearized) by lipid binding, thus lowering electrostatic coupling, and thereby reducing dichroic effects. The observation that the unwinding of negatively supercoiled DNA produces decreases in the intensity of the 277-nm CD peak is consistent with this hypothesis (36). Furthermore, at the secondary structure level, in a series of papers by Chen and Hanlon (53–55), it was shown that a subtle increase in winding angle results in substantial decrease in the rotational strength above 260 nm. Clearly, DNA CD spectra are sensitive to subtle differences in the structure of the DNA component at both the tertiary and secondary level. However, the spectra observed can be considered to sensitively reflect the overall structural organization of the complex and thus provide a valuable measure of structural integrity.

UV Absorption and CD Spectroscopy

281

4. Conclusion Of the many techniques currently employed in the analysis of nucleic acids and their complexes, UV-absorption spectroscopy is rarely the first choice, except in the case of concentration determination. As argued here, however, there appear to be specific applications of the technique that can provide structural information with a few fast and simple experiments. Current instrumentation is inexpensive, user-friendly, and, most important, already present in most laboratories. It seems probable that this ready availability of equipment, accompanied by the considerations outlined above, will lead to an increased use of the technique, especially in the context of derivative analysis of gene delivery complexes. CD is also a sensitive technique for investigating nonviral gene delivery systems. CD spectral differences provide a method for differentiating between complexes constructed with different gene delivery agents. Although interpretation of the structural basis of spectral differences in these complex constructs is elusive, the spectral features are reproducible. The fact that CD seems to provide some insight into the supramolecular organization of these components is actually beneficial, since recent work suggests structural features are important in the delivery process. Thus, this technique is a potentially valuable tool in the characterization of gene delivery formulations and could be used to provide a measure of quality control. An intriguing possibility is that correlations between the CD of gene delivery complexes and their efficiency of transfection may be present, which could aid in the construction of more efficacious vehicles. References 1. van Holde, K. E., Johnson, W. C., and Ho, P. S. (1998) Principles of Physical Biochemistry. Prentice-Hall, Upper Saddle River, New Jersey. 2. Owen, T. (1996) Fundamentals of Modern UV-vVisible Spectroscopy: A Primer. Hewlett Packard Co., Germany. 3. Bloomfield, V. A., Crothers, D. M., and Ignacio Tinoco, J. (1974) Physical Chemistry of Nucleic Acids. Harper & Row, New York. 4. Fasman, G. D., ed. (1975) Handbook of Biochemistry and Molecular Biology. CRC, Cleveland, OH. 5. Cantor, C. R. and Schimmel, P. R. (1980) Biophysical Chemistry Part II: Techniques for the Study of Biological Structure and Function. W. H. Freeman, San Francisco, CA. 6. Beavan, G. H., Holiday, E. R., and Johnson, E. A. (1955) Optical properties of nucleic acids and their components, in The Nucleic Acids (Chargaff, E. and Davidson, J. N., eds.), Academic, New York, pp. 493–553. 7. McGown, E. L. (2000) UV Absorbance measurements of DNA in microplates. Biotechniques 28, 60–64. 8. Baumann, C. G. and Bloomfield, V. A. (1995) Large-scale purification of plasmid dna for biophysical and molecular biology studies. Biotechniques 19, 884–890.

282

Braun, Kueltzo, and Middaugh

9. Wilfinger, W. W. (1997) Effect of pH and ionic strength on the spectrophotometric assessment of nucleic acid purity. Biotechniques 22, 474–481. 10. Stulnig, T. M. and Amberger, A. (1994) Exposing contaminating phenol in nucleic acid preparations. Biotechniques 16, 403–404. 11. Glasel, J. A. (1995) Validity of Nucleic acid purities monitored by 260nm/280nm absorbance ratios. Biotechniques 18, 62–63. 12. Huberman, J. A. (1995) Importance of measuring nucleic acid absorbance at 240 nm as well as at 260 and 280 nm. Biotechniques 18, 636. 13. Manchester, K. L. (1995) Value of A260/A280 ratios for measurement of purity of nucleic acids. Biotechniques 19, 208–210. 14. Wada, A., Yabuki, S., and Husimi, Y. (1980) Fine structure in the thermal denaturation of DNA: high temperature-resolution spectrophotometric studies. Crit. Rev. Biochem. 9, 87–144. 15. Blake, R. D. and Hydorn, T. G. (1985) Spectral analysis for base composition of DNA undergoing melting. J. Biochem. Biophys. Meth. 11, 307–316. 16. Felsenfeld, G. (1971) Analysis of temperature-dependant absorption spectra of nucleic acids, in Procedures in Nucleic Acid Research (Cantoni, G. L. and Davies, D. R., eds.), Harper & Row, New York, pp. 233–261. 17. Wilson, W. D., Tanious, F. A., Fernandez-Saiz, M., and Rigl, C. T. (1997) Evaluation of drug-nucleic acid interactions by thermal melting curves, in Methods in Molecular Biology, vol. 90: Drug-DNA Interaction Protocols (Fox, K. R., ed.), Humana, Totowa, NJ, pp. 219–240. 18. Blackburn, G. M. and Gait, M. J., eds. (1996) Nucleic Acids in Chemistry and Biology. Oxford University Press, Oxford. 19. Davis, T. M., McFail-Isom, L., Keane, E., and Williams, L. D. (1998) Melting of a DNA hairpin without hyperchromism. Biochemistry 37, 6975–6978. 20. McCampbell, C. R., Wartell, R. M., and Plaskon, R. R. (1989) Inverted repeat sequences can influence the melting transitions of linear DNAs. Biopolymers 28, 1745–1758. 21. Carroll, D. (1972) Complexes of polylysine with polyuridylic acid and other polynucleotides. Biochemistry 11, 426–433. 22. Mandel, R. and Fasman, G. D. (1976) Chromatin models. interaction between dna and polypeptides containing L-lysine and L-valine: circular dichroism and thermal denaturation studies. Biochemistry 15, 3122–3130. 23. Katayose, S. and Kataoka, K. (1997) Water-soluble polyion complex associates of dna and poly(ethylene glycol)-poly(L-lysine) block copolymer. Bioconj. Chem. 8, 702–707. 24. Thumm, W., Seidl, A., and Hinz, H.-J. (1988) Energy–structure correlations of plasmid DNA in different topological forms. Nuc. Acids Res. 16, 11,737–11,757. 25. Mach, H., Sanyal, G., Volkin, D. B., and Middaugh, C. R. (1997) Applications of ultraviolet absorption spectroscopy to the analysis of biopharmaceuticals, in ACS Symposium Series. American Chemical Society, pp. 186–205. 26. Mach, H., Middaugh, C. R., and Lewis, R. V. (1992) Detection of proteins and phenol in dna samples with second-derivative absorption spectroscopy. Anal. Biochem. 200, 20–26.

UV Absorption and CD Spectroscopy

283

27. Kim, M.-H., Ulibarri, L., Keller, D., Maestre, M. F., and Bustamante, C. (1986) Psi-type circular dichroism of large molecular aggregates. III. Calculations. J. Chem. Phys. 84, 2981–2989. 28. Keller, D. and Bustamante, C. (1986) Theory of the interaction of light with large inhomogeneous molecular aggregates. I. Absorption. J. Chem. Phys. 84, 2961–2971. 29. Keller, D. and Bustamante, C. (1986) Theory of the interaction of light with large inhomogeneous molecular aggregates. II. Psi-type Circular Dichroism. J. Chem. Phys. 84, 2972–2980. 30. Neidle, S., ed. (1999) Oxford Handbook of Nucleic Acid Structure. Oxford University Press, New York. 31. Ivanov, V. I., Minchenkova, L. E., Schyolkina, A. K., and Poletayev, A. I. (1973) Different conformations of double-stranded nucleic acid in solution as revealed by circular dichroism. Biopolymers 12, 89–110. 32. Middaugh, C. R., Evans, R. K., Montgomery, D. L., and Casimiro, D. R. (1998) Analysis of plasmid DNA from a pharmaceutical perspective. J. Pharm. Sci. 87, 130–146. 33. Mathews, C. K. and van Holde, K. E. (1996) Biochemistry. Benjamin/Cummings, Menlo Park, California. 34. Johnson, W. C. (1994) CD of nucleic acids, in Circular Dichroism: Principles and Applications (Nakanishi, K., Berova, N. and Woody, R. W., eds.), VCH, New York, pp. 523–540. 35. Johnson, W. C. (1996) Determination of the conformation of nucleic acids by electronic CD, in Circular Dichroism and the Conformational Analysis of Biomolecules (Fasman, G. D., ed.) Plenum, New York, pp. 433–468. 36. MacDermott, A. J. and Drake, A. F. (1986) circular dichroism of postively and negatively supercoiled DNA. Stud. Biophys. 115, 59–67. 37. Glaeser, R. M. and Jap, B. K. (1985) Absorption Flattening in the circular dichroism spectra of small membrane fragments. Biochemistry 24, 6398–6401. 38. Mao, D. and Wallace, B. A. (1984) Differential light scattering and absorption flattening optical effects are minimal in the circular dichroism spectra of small unilamellar vesicles. Biochemistry 23, 2667–2673. 39. Wallace, B. A. and Teeters, C. L. (1987) Differential absorption flattening optical effects are significant in the circular dichroism spectra of large membrane fragments. Biochemistry 26, 65–70. 40. Bustamante, C., Tinoco, I., Jr., and Maestre, M. F. (1983) Circular differential scattering can be an important part of the circular dichroism of macromolecules. Proc. Natl. Acad. Sci. USA 80, 3568–3572. 41. Maestre, M. F. and Reich, C. (1980) Contribution of light scattering to the circular dichroism of deoxyribonucleic acid films, deoxyribonucleic acid-polylysine complexes, and deoxyribonucleic acid particles in ethanolic buffers. Biochemistry 19, 5214–5223. 42. Phillips, C. L., Mickols, W. E., Maestre, M. F., and Tinoco, I., Jr. (1986) Circular differential scattering and circular differential absorption of DNA-protein condensates and of dyes bound to DNA-protein condensates. Biochemistry 25, 7803–7811.

284

Braun, Kueltzo, and Middaugh

43. Reich, C., Maestre, M. F., Edmondson, S., and Gray, D. M. (1980) Circular dichroism and fluorescence-detected circular dichroism of deoxyribonucleic acid and poly[d(A-C)•d(G-T)] in ethanolic solutions: a new method for estimating circular intensity differential scattering. Biochemistry 79, 5208–5213. 44. Tinoco, I., Bustamante, C., and Maestre, M. (1980) Optical activity of nucleic acids and their aggregates. Ann. Rev. Biophys. Bioeng. 9, 107–141. 45. Tinoco, I. J. and Mickols, W. (1987) Absorption, scattering, and imaging of biomolecular structures with polarized light. Ann. Rev. Biophys. Biophys. Chem. 16, 319–349. 46. Ausubel, F. M., et al., eds. (1989) Current Protocols in Molecular Biology. Wiley, New York. 47. Koltover, I., Salditt, T., and Safinya, C. R. (1999) Phase diagram, stability, and overcharging of lamellar cationic lipid-DNA self-assembled complexes. Biophys. J. 77, 91–924. 48. Huang, L., Hung, M.-C., and Wagner, E. (1999) Non-Viral Vectors for Gene Therapy. Academic, San Diego, CA. 49. Fischer, D., Bieber, T., Li, Y., Elsasser, H.-P., and Kissel, T. (1999) Novel Nonviral vector for DNA delivery based on low molecular weight, branched polyethylenimine: effect of molecular weight on transfection efficiency and cytotoxicity. Pharm. Res. 16, 1273–1279. 50. Weiskoff, M. and Jei Li, H. (1977) Poly (L-lysine) -DNA interactions in NaCl solutions: B - C and B - psi transitions. Biopolymers 16, 669–684. 51. Koltover, I., Salditt, T., Rädler, J. O., and Safinya, C. R. (1998) An inverted hexagonal phase of cationic liposome-DNA complexes related to DNA release and delivery. Science 281, 78–81. 52. Gottarelli, G. and Spada, G. P. (1994) Application of CD to the study of some cholesteric mesophases, in Circular Dichroism: Principles and Applications (Nakanishi, K., Berova, N., and Woody, R. W., eds.), VCH, New York, pp. 105–119. 53. Chen, C., Kilkuskie, R., and Hanlon, S. (1981) Circular dichroism spectral properties of covalent complexes of deoxyribonucleic acid and n-butylamine. Biochemistry 17, 4987–4995. 54. Chen, C., Pheiffer, B. H., Zimmerman, S. B., and Hanlon, S. (1983) Conformational characteristics of deoxyribonucleic acid-butylamine complexes with C-type circular dichroism spectra 1. An X-ray fiber diffraction study. Biochemistry 22, 4746–4751. 55. Fish, S. R., Chen, C., Thomas, G. J., Jr., and Hanlon, S. (1983) Conformational characteristics of deoxyribonucleic acid-butylamine complexes with C-type circular dichroism spectra 2. A Roman spectroscopic study. Biochemistry 22, 4751–4756.

Synthetic Vectors by IR Spectroscopy

285

20 Characterization of Synthetic Gene Delivery Vectors by Infrared Spectroscopy Sirirat Choosakoonkriang, Christopher M. Wiethoff, Lisa A. Kueltzo, and C. Russell Middaugh 1. Introduction For many decades, infrared (IR) spectroscopy has been used to characterize the structure of molecules. In IR spectroscopy, absorption of light, corresponding to vibrational and rotational transitions of a molecule, is measured. For a transition to be IR-active, a change in the dipole moment of a particular bond must occur upon excitation. This vibrational energy is not only dependent on the chemical nature of the particular covalent bonds, but also on the environment of these coupled atoms and bonds. IR spectroscopy has been previously employed in the study of the structure of nucleic acids, producing not only information about the individual bases, sugars, and phosphate backbone, but also providing information about the helical conformation of polynucleotides (1–3). IR spectroscopy has also been successfully applied to the analysis of lipids, as well as to numerous other polymers (4). Thus, IR spectroscopy potentially possesses the ability to obtain structural information about all of the components of most synthetic gene delivery complexes, as well as changes in the structure of polymeric or lipid components upon complex formation. In addition to the ability to gather detailed structural information, there are also some practical advantages to the use of IR spectroscopy for the study of plasmid DNA and DNA complexes compared to other techniques, including the availability of a variety of sampling techniques, permitting the analysis of samples in a wide variety of physical states including solutions, solids, and gels. There is also no upper limit to the size of the sample molecule examined, allowing both short oligonucleotides and higher molecular weight DNA to be From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

285

286

Choosakoonkriang et al.

studied. IR spectroscopy is not a destructive technique, and requires only small amounts of material, making it ideal for the analysis of valuable samples. Here are reviewed IR instrumentation, common sampling geometries in relation to the physical states of samples, and different methods of data analysis. Examples of the use of Fourier-transformed infrared (FTIR) to characterize the interaction of DNA with several polycations used in gene delivery including cationic lipids and polymers, are presented and discussed. 2. Representative IR Vibrational Modes 2.1. Characteristic IR Absorption Bands of Polynucleotides The most useful vibrational bands of DNA are observed between 1800 and 700 cm–1, with approx 35 well-defined absorption bands occurring in this region of the IR spectrum (Fig. 1). Four general aspects of DNA structure are reflected in the vibrational modes in this region. Bands between 1750 and 1500 cm–1 primarily represent vibrations of the bases. Vibrations caused by DNA basesugar entities, heavily dependent on glycosidic torsion angles are observed between 1500 and 1250 cm–1. Strong absorption occurs for the phosphate groups and deoxyribose between 1250 and 1000 cm–1 and the region below 1000 cm–1 primarily contains vibrations of the phosphodiester bond coupled to vibrations of the deoxyribose. Table 1 provides a more complete listing of the IR vibrational modes of DNA. Vibrational modes in the spectral region from 1750 to 1500 cm–1 result from carbonyl stretching and N-H bending, as well as C = C and C = N stretching vibrations of the nucleic acid bases. Each of the four bases, when present as mononucleotides or single-stranded homopolymers, has characteristic bands in this region, and it is common to observe multiple bands in this region that result from the overlap of bands from the individual bases. Unfortunately, this region contains a strong water absorption band, making peak assignments difficult. Water subtraction algorithms have greatly reduced this problem, but studies in this region are often performed in deuterium oxide (D2O), to eliminate the water interference. Therefore, the following assignments in this region are reported for D2O solutions. Assignments in other regions of the polynucleotide spectra discussed below are for samples in water solutions. Guanine has characteristic bands at 1531, 1581, and 1668 cm–1 (5). Similarly, this region contains bands at 1506, 1524, 1619, and 1652 cm–1 (cytosine), 1626 cm–1 (adenine) and 1632, 1663, and 1695 cm–1 (thymine) (5). In many cases, these characteristic vibrations of the nucleic acid bases undergo abrupt changes when involved in intermolecular base-pairing in duplex DNA. Important spectral differences upon base-pairing, as seen for the interaction of complementary homopolymers, include a shift of the carbonyl stretching vibration of guanine from 1668 to 1689 cm–1, accompanied by a strong reduction in intensity of the

Synthetic Vectors by IR Spectroscopy

287

Fig. 1. FTIR absorbance spectra of DNA (plasmid) in H2O solution (top) (top to bottom: 4.0, 3.0, 2.0, 1.0, and 0.5 mg/mL), and on polyethylene cards (bottom) (top to bottom: 50 µL of 5.4, 4.0, 3.0, 2.0, 1.0, and 0.5 mg/mL).

peak at 1581 cm–1 (5). Cytosine shows little change in its carbonyl stretching vibration at 1652 cm–1, but the band at 1524 cm–1 can no longer be observed upon base-pairing. In the case of adenine, the band at 1626 cm–1 is shifted down to 1622 cm–1 (5). The carbonyl bands for thymine (1695 and 1663 cm–1) change only with respect to their relative intensities, while the 1632 cm–1 band shifts to 1641 cm–1.

288

Choosakoonkriang et al.

Table 1 Representative IR Vibrational Modes of DNA Nucleic acid base Adenine Cytosine

Wavenumber (cm–1)a,b 1626 (1622 upon base pairing) 1652 1619 1524 (not observed upon bp) 1506 1668 (1689 upon bp) 1581 1531 1695 1663 1632 (1641 upon bp)

Guanine

Thymine

Conformationally Sensitive Bands (cm–1) Ab

Bb

Cc

Zb

1705 1418 1375 1335

1715 1425 1375 1344 1328

1710 1425 1375 1344 1328

1695 1408 1355

1275

1281

1240 1188

1225

1320

970 (triplet)

970 (singlet)

Assignmentb,c Base carbonyl indicative of base pairing Deoxyribose dGdA dA dT dC dT

1265 1230–1217 1215

Antisymmetric phosphate stretch

1069

Deoxyribose

1065 1013

968

Deoxyribose 929

898 (triplet)

897 (singlet) 840

891 833

806

Deoxyribose Deoxyribose Deoxyribose

a Assignments

are for bases in D2O. are from refs. 2 and 5. cAssignments are from ref. 9. b Assignments

The helical conformation of DNA potentially exists in multiple forms depending on factors such as salt concentration, base sequence and composition, relative humidity, and pH. The most common antiparallel, double-stranded, helical geometries of DNA are the right-handed A- and B-form and the left-handed

Synthetic Vectors by IR Spectroscopy

289

Z-form families. Two important differences in DNA helical geometries allow one to spectroscopically distinguish one form from another. Differences include departures of the signature bands for base-pairing from that of the archetypal Bform geometry and differences in the conformation of the sugars in the helix (e.g., N-type, C3' endo/anti geometry for the A-form and S-type, C2' endo/anti geometry for the B form). The IR spectra of these various forms contain tell-tale bands that permit discrimination between each of the helical geometries. Of the two right-handed geometries, the B form prevails in aqueous solutions at moderate ionic strength and neutral pH. Thus, it is this form of DNA that is thought to most commonly occur in biological systems. The A form occurs primarily under conditions of low water activity (66–47% relative humidity) or in high concentrations of organic solvents, such as 80% EtOH (2,5). Between 1750 and 1500 cm–1, a band around 1715 cm–1, caused by the carbonyl stretching of the bases reflects the intermolecular hydrogen bonding of base pairs for the B form of DNA in H2O (5). This band is shifted down to 1705 cm–1 for the A form. In the spectral region between 1500 and 1250 cm–1, several vibrational modes are characteristic of B-form DNA. A vibration caused by the deoxyribose sugar (C2' endo) occurs at ~1425 cm–1, but is shifted to approx 1418 cm–1 for A form (5). Another characteristic sign of the right-handed geometry is the torsion angle of the glycosidic bond for purine bases. Both A and B forms manifest a band near 1375 cm–1, because of the anticonformation of deoxyguanylate and deoxyadenylate (6). In comparing poly(dA)·poly(dT) (B form) to poly(rA)·poly(dT) (A form), the B-form spectra include a doublet at 1344 and 1328 cm–1 from the adenine and thymine bases, respectively, as well as a single band at 1281 cm–1, arising from the N-3 H-bending vibration of thymine (2). For the A form, the doublet bands are merged into a single peak near 1335 cm–1 and the N-3 H-bending vibration is found at about 1275 cm–1. In the region between 1250 and 1000 cm–1, information regarding the antisymmetric and symmetric phosphate- stretching vibrations is contained. The B form exhibits a strong band at 1225 cm–1 due to the antisymmetric stretching vibration of the phosphate. This band is found ~15 cm–1 higher in A-form DNA near 1240 cm–1. The position of the symmetric-stretching vibration of the phosphate is essentially independent of the helical geometry of DNA and is found at about 1089 cm–1. As mentioned above, the spectral region between 1000 and 700 cm–1 contains information about the vibrations of the phosphate esters coupled to vibrations of the deoxyribose ring. A band at 970 cm–1 is seen for the B form; a poorly resolved triplet centered at this position is observed for A form. The B form also shows two bands at 894 and 840 cm–1; the A form displays a wellresolved triplet at 898, 882, and 864 cm–1 (2,5). The puckering of the ribose ring for the A- and B-form geometries gives rise to bands at 807 cm–1 (N type, C3'

290

Choosakoonkriang et al.

endo/anti) and 840 cm–1 (S type, C2' endo/anti) for the A and B forms, respectively (5). Thus, distinct spectral bands for the A and B forms allow the two major helical geometries of DNA to be unambiguously distinguished by IR spectroscopy. Several variants of B-form DNA have been identified, including the C and D forms. The C form has been detected in the presence of specific counterions, such as Li+ and Cs+ (7), and has been shown to occur with reduced H2O activity (lower relative humidity), while specific sequences (polyd(A-T)·polyd[AT]) have been found to be present in the D form (5). Conflicting reports have indicated that the antisymmetric phosphate stretching vibration of C-form DNA occurs anywhere from 1230 (7) to 1217 cm–1 (8). Additional differences between the C and B form may be observed by a shift in the base carbonyl stretching peak from 1715 cm–1 (B form) to ~1710 cm–1 (C form) (8,9). Perhaps the most consistent difference observed between the IR spectra of B- and C-form DNA is the appearance of a more prominent band around 1069 cm–1 in the C-form spectra, compared to a weak shoulder at 1071 cm–1 seen in B-form spectra (8,9). Initial studies using circular dichroism have suggested the existence of C-DNA, when DNA is complexed with either cationic lipids (10,11) or a cationic polymer, such as polylysine (12). These results are complicated by distortions in the CD spectra of large complexes, making interpretation difficult (see Chapter 19 by Braun et al.). The IR spectrum of the left-handed double helical Z-form DNA contains several distinct differences compared to its right-handed counterparts. In the case of Z-form DNA, the band reflecting the presence of base-pairing seen at 1715 cm–1 for B form is shifted to around 1695 cm–1 (5). The phosphate antisymmetric-stretching vibration is also present at lower frequency, near 1215 cm –1. Additionally, new absorption bands around 1320 cm–1 (cytosine), 1265, 1123, 1067, 1013, and 928 cm–1 are indicative of the Z form (13). Although no information regarding IR vibrational frequencies have been reported for ^-form DNA, this highly ordered state has been proposed to exist for several polycation–DNA complexes (11,14). The many conformationally sensitive bands of DNA provide an excellent basis with which to examine complex formation in gene delivery vehicles. In particular, analysis of the shifts in position of these peaks, as polycations bind to DNA, should permit the structure of complexed DNA to be probed.

2.2. Major IR Bands of Cationic Lipids and Select Polycations Cationic lipids have been extensively used in nonviral gene therapy as vectors for the delivery of DNA to cells because of their low toxicity, reduced immunogenicity, and established efficacy, providing a simple system for DNA delivery. Cationic lipid systems have been shown to produce significant levels of gene expression in tissues when administered intravenously (15), with

Synthetic Vectors by IR Spectroscopy

291

Fig. 2. FTIR absorbance spectra for 1,2-dioleoyl-3-trimethylammonium-propane (DOTAP) (top, 1.2 mg/mL) and dimethyldioctadecylammonium bromide (DDAB) (bottom, 2.5 mg/mL) liposomes in solution.

activity demonstrated in multiple tissues (16–22), increasing the attractiveness of the system. A variety of such vectors is currently in human clinical trials. Based on the potential of these systems, the authors concentrate here on the IR characterization of examples of cationic lipid-DNA complexes, although, as is shown, extension to other polymeric-based delivery systems is straightforward. Cationic lipid molecules usually have several IR-active groups (Fig. 2). The main vibrational modes observed in lipids originate from the molecular vibrations of the acyl or alkyl chains and the head groups. There are two major

292

Choosakoonkriang et al.

types of acyl chain vibrational modes, the C–H vibration characteristic of the hydrophobic hydrocarbon region, and the C–O vibration characteristic of the glycerol–acyl-chain interface. These bands are easily identified (Table 2). C–H stretching bands appear in the region of 3200–2700 cm–1, primarily because of the antisymmetric (2925 cm –1 ) and symmetric (2852 cm–1 ) vibrations of the methylene groups (23). Bending vibrations are observed for methyl and methylene groups, including the symmetric CH3 deformation (umbrella-type) vibration at 1375 cm–1, a CH2 scissoring band between 1474 and 1468 cm–1, a series of CH2 wagging bands between 1190 and 1345 cm–1, and CH2 twisting/rocking peaks in the region of 720–1150 cm–1 (23). The ester carbonyl-stretching mode has a strong absorption band that appears at 1700–1750 cm–1. The position of this band is sensitive to the geometry of the packing of the acyl chains and glycerol moiety, as well as the hydration state of the lipid headgroup (23). For the cationic lipid, 1,2-dioleoyl-3-trimethylammoniumpropane (DOTAP), the ester C = O stretch appears ~1739 cm–1 in solution (24). The headgroups of membrane lipids also produce a number of characteristic IR bands. In phospholipids, for example, the antisymmetric PO2– stretching vibrations result in a strong IR band in the range of 1220–1240 cm–1, with a symmetric stretch occurring at ~1085 cm–1 (23). Unfortunately, these features significantly overlap the related vibrations in DNA, complicating spectral assignments of complexes. The trimethylammonium headgroup of many lipids has several vibrational modes. Bending modes of the C–H group appear as medium to weak bands at 1485 (antisymmetric) and 1405 (symmetric) cm–1. The N-C-bending vibration can be followed at 970 (antisymmetric) and 920 (symmetric) cm–1. These bands are weak to medium in intensity (25). Polycations such as polyethylenimine (PEI) or poly-L-lysine have IR spectra with a number of distinct bands. Polylysine possesses an NH3+ group, which has symmetric and antisymmetric bending modes at ~1520 and 1630 cm–1 (see Table 2). PEI manifests -NH2+- deformations in the frequency range of 1560–1620 cm–1 (26). Polylysine also possesses the Amide I and Amide II regions, which overlap with much of the in-plane double bond vibrations of the DNA bases and the conformationally sensitive DNA base-sugar coupled vibrations in the range of 1700–1450 cm–1. However, polylysine and PEI do not possess major absorption bands in the spectral region dominated by the deoxyribose-phosphate backbone (1250–950 cm–1), nor do they absorb in the region just above 1700 cm–1, where conformationally sensitive bands, indicative of DNA base-pairing occur. This allows one to follow changes in the double stranded helical geometries caused by changes seen in base pairing and the sugar-phosphate backbone of DNA upon formation of complexes with many polycations. A more detailed list of vibrational bands observed in lipids and polycations is shown in Table 2.

Synthetic Vectors by IR Spectroscopy

293

Table 2 IR Absorption Bands of Common Lipids and Polycations Wavenumber (cm–1) 3000–3500 3200–3500 3038 2956 2920 2870 2850 1720–1750 1630 1623 1520–1650 1520 1485 1473 1468 1460 1405 1370–1380 1200–1400 1228 1170 1085 1070 1060 970 920 720–730 aAssignments

Assignmenta N-H Stretch O-H Stretch (R-OH) CH3 Antisymmetric stretch (Choline) CH3 Antisymmetric stretch CH2 Antisymmetric stretch CH3 Symmetric stretch CH2 Symmetric stretch C=O Stretch of ester NH3+ Antisymmetric bend COO– Antisymmetric stretch N-H Bend NH3+ Symmetric bend (RNH3+) N(CH3)3 C-H Antisymmetric bend CH2 Scissoring (triclinic) CH2 Scissoring (hexagonal) CH3 Antisymmetric bend +N(CH ) C-H Symmetric bend 3 3 CH3 Symmetric bend CH2 Wagging band progression PO2– Antisymmetric stretch CO-O-C Antisymmetric stretch PO2– Symmetric stretch CO-O-C Symmetric stretch C-O-P-O-C Stretch +N(CH ) N-C Antisymmetric bending 3 3 +N(CH ) N-C Symmetric bending 3 3 CH2 Rocking

are from refs. 27, 28, and 31.

3. Instrumentation, Sampling Geometries, and Sample Preparation Over the past decade, development of IR instrumentation has focused on FTIR spectrometers. The newer FTIR instruments provide a number of advantages compared to the older grating models, including faster sampling times (typically 1 vs 20 min for a 2 cm–1 resolution spectrum), high resolution (as low as <0.001 cm–1), simpler and more comprehensive data analysis with current software, and increased sensitivity, resulting from the newer mercury– cadmium–telluride (MCT) detectors. Deuterated triglycine sulfate (DTGS) detectors are also used, but have slower response times than the MCT detectors

294

Choosakoonkriang et al.

(3), and can extend sampling time. Nevertheless, they are often sufficient for the purposes described here. The broad bands seen in solution spectra, however, require resolution of only 2–4 cm–1 for usable spectra to be obtained. Most FTIR instruments available today offer a wide variety of sampling geometries through easily attachable accessories, thus accommodating a wide range of sample types. Unlike UV absorption, circular dichroism, and fluorescence spectroscopies, sample states such as solutions, suspensions, gels, solids, and films can all be examined with IR spectroscopy with minimal artifacts. There are many sample geometries available, and each offers unique advantages and disadvantages. This chapter concentrates on the three sampling geometries most often employed in the study of macromolecules: transmission, attenuated total reflectance, and diffuse reflectance. 3.1. Transmission Transmission spectroscopy (Fig. 3) is commonly used to examine macromolecular samples such as proteins, DNA, and their complexes with various ligands. Liquids, IR-transparent compressed salt pellets, thin films, cast films, and gases can all be analyzed by this method. However, the pathlength and the construction material of the transmission window are important and should be carefully chosen. Cells constructed with water-insoluble materials that are transparent in the spectral region of interest are required. Crystals such as calcium fluoride (CaF2) or related halide salts are often used for the transmission window. The use of CaF2 windows to study DNA, however, will limit the amount of information that can be obtained because of its poor transmission properties below 1100 cm–1. Most materials also have specific temperature and pH ranges over which they can be used, as well as solvent specific limitations, all of which must be considered. Both adjustable and fixed pathlength cells are available, with the required pathlength of the transmission cell depending on the concentration of the sample and solvent employed (H2O vs D2O). A typical DNA concentration (for solution sampling) is on the order of 1–5 mg/mL or higher. Associated pathlengths employed are 50–100 µm for D2O and 2–10 µm for H2O. Modern subtractive methods do allow longer pathlength cells to be used in the presence of H2O if care is taken. Higher concentrations lead to better signal:noise (S:N) ratios, but at the expense of ease of sample handling, since the viscosity of the solution increases, as the result of the high molecular weight of plasmid DNA. Recall that the Beer-Lambert law states that absorbance is directly proportional to both concentration and pathlength, which are inversely proportional to each other. Thus, for lower concentrations, an increased pathlength may be used to increase the S:N ratio; for more concentrated samples, a shorter pathlength, and smaller sample volume, may be used. However, as pathlength increases, so does solvent absorption. Ultimately, a compromise must be reached to optimize the quality of the spectra.

Synthetic Vectors by IR Spectroscopy

295

Fig. 3. Transmission geometry. Black arrows represent the path of the light beam.

Another major factor in transmission spectroscopy of solution samples is solvent correction. Most biological samples contain buffer species in aqueous solution, which must then be subtracted from the spectra. The most difficult spectrum to subtract, however, is that of water itself. Although the molar absorptivity of the major vibrational bands of water is not significantly different from most biological macromolecules, the fact that it is present in such high concentrations, relative to the molecule of interest, results in a spectrum dominated by water. The IR spectrum of water contains a broad, intense band in the region of 3700–3100 cm–1, because of O–H stretching, in addition to an O-H bending band at 1640 cm–1, and a broad liberation peak below 800 cm–1. The association band of water, which is a linear combination of the O–H bending frequencies, occurs at around 2200 cm–1. These bands will usually mask the weaker signals of the sample. The use of D2O for solution samples can help by opening different “windows” in the spectra since the strong O–D stretching bands of the solvent then shift to between 2800 and 2200 cm–1 and 1200 cm–1 for O–D bending vibrations, with the association band occurring at 1570 cm–1 (26). Transmission experiments can also be performed on solid samples, often in the form of thin films. In such methods, the sample is dried on a polymer membrane (e.g., polyethylene) or suitable IR-transparent substrate (e.g., a silver chloride window). This technique is advantageous, in that there are no broad solvent peaks to subtract, it is easy to use, and requires small sample volumes (25–50 µL). However, interpretation of the final spectrum must take into consideration the possibility of changes in the sample that may occur upon drying. This is especially critical for macromolecular samples, in which hydrogen bonding of water to the macromolecule plays an important role in its structure. Finally, samples can be compressed into transparent pellets, typically using KBr as a matrix. This is a simple, well-established method, but suffers from the presence of the chaotropic bromide anion, which may dissolve in any residual solvent, potentially producing structural disruption.

296

Choosakoonkriang et al.

3.2. Attenuated Total Reflectance Another commonly used sampling geometry is attenuated total-internal reflectance (ATR), in which the sample is placed on a crystal comprised of an IR transparent material with a high index of refraction, such as zinc selenide (ZnSe) or germanium. These crystals, or internal reflection elements (IREs), vary in properties such as angle of reflection, usable temperature and pH range, and solvent sensitivity, and, as such, the selection of the proper crystal is important. Alternative crystal geometries, such as cylinders, in which the solution completely surrounds the IRE, are also available. Once the sample is placed on the crystal, the IR beam is passed through the crystal at a specific angle (commonly 45 degrees), and reflected through the crystal, penetrating slightly into the sample (typically, 1–2 µm) with each reflection (Fig. 4). This penetration, known as an evanescent wave, allows the sample to absorb light as in the transmission method, but in a cumulative fashion. Thus, the effective pathlength is the sum of the penetration of each bounce which contacts the sample. The nonabsorbed radiation is then collected by the detector and the spectrum is subsequently extracted (27). The resulting spectrum can be affected by several factors, including the angle of reflection, sample size (both thickness and area), the wavelength of the radiated light, the total number of reflections, and the refractive index of the sample (27). The spectral quality also depends on the effective “penetration depth” of the evanescent wave into the sample. This depth can be calculated at any wavelength using the following equation: dp =

h 2/ nsp (sin2e – nsp2) 1/2

(1)

where h is the wavelength of the radiation in the IRE, e is the angle of incidence, np is the refractive index of the IRE, and nsp is the ratio of the refractive indices of the sample to the IRE (27). The important points to note here are first, that a greater depth of penetration should result in better overall signal, and second, that the depth of penetration is dependent on the wavelength. This last feature requires that the spectrum be corrected, because the spectrum will appear to have a sloped baseline, resulting from the wavelength dependent penetration depth. Automated correction for this effect is routinely available through the software of modern instruments. The ATR geometry can be used for any sample that can be placed in close contact with the crystal surface, and has been used historically for solids, gels, films, and more recently, solutions. It is especially useful for the analysis of samples that are too thick or viscous for transmission methods. ATR has also been used to study amyloids, inclusion bodies, and aggregated samples, since it is not necessary to pass the radiated light completely through the sample to obtain the spectrum. This advantage also presents one of the major drawbacks

Synthetic Vectors by IR Spectroscopy

297

Fig. 4. ATR geometry. IR beam path through a multi-bounce accessory. Black arrows represent the path of the light beam.

to ATR methods. The spectrum obtained from an ATR sample will, at lowpenetration-depth values, reflect only a population of sample close to the ATR surface, as opposed to the overall properties of the sample. At the surface of the ATR element, the sample may in fact be adsorbed onto the crystal, and thus the spectrum obtained is partially that of surface-bound species. Oberg and Fink have shown that protein samples in particular can interact with certain ATR crystals (28). This appears to be only a minor problem with DNA and DNA complexes, however, since conventional transmittance spectra give similar results (Choosakoonkriang, et al., unpublished results). The advantages of the ATR technique, especially the ability to analyze samples in solution quickly and accurately, appear to outweigh the limited drawbacks in most cases.

3.3. Diffuse Reflectance Diffuse reflectance (DRIFTS when used in conjugation with an FTIR instrument) is a sampling geometry for use in the analysis of solid samples or molecules adsorbed to solids. In principle, lyophilized DNA and DNA–lipid complexes can be studied by this technique, although, to the authors’ knowledge, no such studies have yet been reported. Therefore, the theory, advantages, and major disadvantages of this technique are only briefly discussed. Readers are referred to Culler (29) for a more in-depth treatment of the subject. DRIFTS experiments involve the collection of incident radiation scattered from the surface of a solid sample. Light is reflected onto the surface of a solid sample, then randomly scattered. This scattered radiation is less intense at frequencies at which the sample absorbs and thus a spectrum can be collected (Fig. 5). The reflectance spectrum can be defined by the Kubelka-Munk equation: f(R) =

(1 – R)2 2R

= K s

(2)

where f(R) represents the reflectance spectrum, s is the scattering coefficient, K is the molar absorption coefficient, and R is the ratio of the reflectance of the

298

Choosakoonkriang et al.

Fig. 5. DRIFTS single-mirror sampling geometry. Black arrows represent the path of the light beam.

sample to a standard, typically KBr. This equation is analogous to the BeerLambert law, as applied to the diffuse reflectance technique. From this equation, a linear relationship between the maximum reflectance at each peak in the absorption spectrum (f[R]max) and K is inferred, assuming the scattering coefficient for the sample is constant for all frequencies (29). This relationship has been used in the quantitative analysis of certain samples. However, the range of linearity is dependent of the particle size and molar absorptivity of the sample, and is often nonlinear at high concentrations (29). The DRIFTS method has two main advantages, besides the ease of use with solid samples. First, the sample does not need to be compressed into any type of solid pellet, allowing the analysis of loose powders, and no matrix is necessary, although one is sometimes used for reference purposes. Additionally, any internal reflectance of the radiation within the sample greatly magnifies the pathlength, and therefore the strength of the signal (3). The technique also possesses some major disadvantages. In areas of high absorptivity, the reflected radiation can be close to zero. In these regions, specular reflectance (reflection of part of the radiated light that does not penetrate the sample) becomes a noticeable problem. This additional reflectance from the surface of the solid can produce so-called Reststrahlen bands, which can lead to distortion or even inversion of the absorption bands of the sample (29). Correction of peak distortions and weak signals commonly observed with this technique is usually performed through application of the Kubelka-Munk equation (30) and is frequently automated in instrument software. 4. Data Analysis FTIR data can be used for both the quantitative and qualitative analysis of molecular structure. Qualitatively, the existence of a peak in the spectra indicates the presence of a particular functional group in the sample in a certain

Synthetic Vectors by IR Spectroscopy

299

environment and the spectra can be used for identification purposes. For example, the oxidation of a lipid in a given formulation can be followed. Quantitatively, the peak height or area for a given peak can be used to determine the relative or absolute concentration of a functional group in the sample. Perhaps the most useful information for studies of the interaction of DNA with various polycations, however, is the relative position of the peak in the spectrum. For instance, changes in the vibrational energy upon interaction of DNA with a polycation could potentially show differences in the hydrogen bonding of the DNA bases or environmental perturbations of the lipid esters upon complexation with DNA. This approach will be considered in more depth in Subheading 5.4.

4.1. Zero and Second-Derivative Spectral Analysis Traditionally, IR spectra of small molecules have been presented as transmittance spectra. In contrast, biochemists usually display IR data of macromolecules such as DNA and proteins as absorbance spectra (zero order spectra). The simplest way to determine the peak position of bands of interest in a zero-order spectrum is through the use of the peak finding option built into most common software packages. These programs allow the user to set both the sensitivity and absorbance detection limit of the assigned peaks. These functions do not, however, always detect all of the peaks within a given spectrum. In such cases, peak positions must be determined manually. Although a cursor can be used for this purpose, the use of the first or second derivative of the spectrum is less ambiguous. Second-derivative analysis is a band-narrowing technique that can be used to resolve overlapping peaks, and aid in more accurate determination of peak position. For this technique to be useful, however, the quality of the raw data in terms of its S:N ratio must be high (>1000:1). It is, in fact, often the case that the second derivative of high-resolution spectra of macromolecular samples having a low S:N ratio is too noisy to give accurate peak positions, and “phantom” peaks may be introduced. Switching to a slightly lower resolution often eliminates this problem. However, probably the major advantage of this technique is its ability to determine with great accuracy the positions of not only well-resolved peaks, but also weak shoulders and underlying peak components. This application is especially important when studying complex systems, such as those of interest here, which consist of a large number of overlapping peaks.

4.2. Fourier Self-Deconvolution Often, several peaks of interest lie in the same region of the spectrum and are present as a single broad peak. In some cases, such as the analysis of the complex Amide I bands of protein molecules, the underlying components

300

Choosakoonkriang et al.

cannot be detected by eye. The second-derivative spectrum can be used as indicated, with the intensities of the individual second-derivative peaks used to estimate the relative contribution of each component. Another band-narrowing technique, Fourier self-deconvolution (FSD), can also be used to deconvolute overlapping spectral features. The FSD technique is especially useful from a quantitative perspective because the relative integral areas of the peaks are conserved upon deconvolution. This type of deconvolution is a built in function in most modern data analysis programs and is quite useful. The proper use of this technique requires some care, however. The operator is required to set a single postulated bandwidth at half height for all underlying components and an enhancement parameter representing the degree to which features are resolved. Preliminary results are then calculated, and one iteratively searches for optimal parameters. This method can be subjective and a good rule of thumb is to establish the bandwidth and enhancement parameters with one sample, and, if possible, to remain consistent with them throughout all subsequent deconvolutions of a sample set. Overdeconvolution of the peak can easily occur, introducing spurious peaks. This problem can be identified by the introduction of oscillations into the baseline of empty spectral regions near the peak of interest. Iterative alteration of the FSD parameters involving increases in the bandwidth and decreases in the amplification factor can then be used to minimize the problem. Again, high S:N ratios are necessary for the use of this method, but these are usually obtainable with modern FTIR instruments. It should also be noted that zero-order spectra can be smoothed prior to secondderivative or FSD analysis, but loss of critical information may result. A good example of the application of FSD is in the analysis of the carbonyl region of phosphatidylethanolamine (PE), the IR spectrum of which contains a broad peak centered around 1738 cm–1 (23). This peak has been resolved into two separate components at 1743 and 1726 cm–1 using FSD. It was originally thought that the presence of two peaks resulted from differences in the sn-1 and sn-2 esters of the lipid. However, 13C-isotope editing of one of the ester groups has revealed that both esters contribute to each of the two peaks. It is thought that differences in hydrogen bonding of the carbonyl results in a splitting of the peak with the lower frequency being the result of a species that has more interaction with the solvent water (23). Changes in the relative contributions of the resolved peaks to the zeroorder spectra were followed as a function of temperature, and shifts in the peak position in the zero-order spectra were found to be caused more by changes in their relative intensity than changes in their vibrational energy (peak position).

4.3. Curve Fitting The quantitative analysis of poorly resolved peaks, such as those seen in macromolecules, can present many problems. Although FSD can help identify the presence and location of such bands, it cannot be used to quantitate the

Synthetic Vectors by IR Spectroscopy

301

relative area of each peak. Curve-fitting methods are therefore generally used to determine the relative amounts of previously detected underlying components. In most curve-fitting programs, Gaussian and/or Lorentzian functions are used to represent characteristic component absorption bands. Initial positions and size of the underlying peaks can be found either automatically by the software, or manually determined by the user. The second-derivative and FSD spectra are best used as a guideline for manual peak selection. Once the initial parameters are set, a nonlinear least-squares method is used to get bestfit values of peak position, height and width for each band. In the analysis of DNA and DNA complexes, this technique has not yet been fully explored, although isolated reports have appeared for DNA (9). 5. Application to Gene Delivery Vehicles: Analysis of Cationic Lipid–DNA Complexes Previously, IR spectroscopy has been employed in the characterization of lipids, carbohydrates, nucleic acids (31), and DNA–metal ion binding (32–35). Moreover, IR has been used to detect conformational changes and the interaction of DNA with drugs (36,37), natural or synthetic diamines (5), and peptides (38), as well as changes occurring under various sample conditions, including altered salt concentration (39) and humidity levels (40). Based on these previous studies, the authors have developed methodology for the study of the interaction of DNA with cationic lipids. Here is discussed the analysis of cationic lipid–DNA complexes from spectra obtained by both ATR and thinfilm transmission methods, although similar results have been obtained by the less convenient transmission methods but are not presented.

5.1. Preparation of Liposomes and Complexes The cationic lipid DOTAP was deposited from chloroform solutions on the sides of a glass vial by evaporating the solvent under a stream of nitrogen gas. The resulting film was then placed under vacuum for several hours to remove residual solvent. The lipid was dispersed in 10 mM Tris buffer, pH 7.4, at temperatures above that of the lipid gel-liquid crystalline phase transition for 0.5–1 h. Unilamellar vesicles were prepared by extruding the suspension 10× through a 100 nm pore polycarbonate membrane. All liposomes were used within 3 d of preparation. Cationic lipid–DNA complexes were prepared using various weight ratios of lipid:DNA. Aliquots of lipid and DNA to be mixed were always of equal volume, and were combined by adding the less concentrated component to the more concentrated one. A 9.1-kb plasmid containing >95% supercoiled DNA was used in all studies. The samples were stirred continuously for 10 min, and equilibrated at room temperature for at least 20 min before measurement. All complexes were used the same day as preparation and the final DNA concentration was 1 mg/mL in all cases.

302

Choosakoonkriang et al.

5.2. IR Measurements FTIR spectra were collected using a Nicolet Magna-IR 560 spectrometer equipped with a MCT detector. Spectra were obtained at 4 cm–1 resolution, by co-addition of 256 interferograms. The instrument was continuously purged with dry air to minimize water vapor and CO2 absorption. Samples were measured either by attenuated total reflectance (ATR) or a thinfilm method involving the deposition of the sample on a PE film. Background scans with minimal water vapor were collected before sampling. For ATR measurements, ~1 mL of each sample was placed in a trough ATR accessory (Thermal A.R.K., SpectraTech) equipped with a ZnSe crystal (45 degree angle of reflection, 12 bounces). The samples covered the entire crystal. The sample compartment was purged with dry air for at least 20 min before the spectrum was collected. For the transmission study, sampling cards with a PE membrane were used (3 M IR card, type 61, 19-mm aperture). For each sample, 50 µL was deposited onto the center of the PE card and dried under mild vacuum overnight, prior to measurement. After each card was placed in the transmission accessory, the sample compartment was again allowed to purge for at least 20 min before collecting spectra. All samples were prepared and analyzed in triplicate at a minimum. Because the spectral shifts of interest are often small (<1–2 cm–1), replicates of experiments and accompanying error analysis are critical. 5.3. Data Analysis All analysis was performed using OmnicTM software (Nicolet, Madison, WI). For aqueous solutions, the buffer spectrum was subtracted from the sample spectrum, using the association band of water as a guide. The goal is to achieve a flat baseline in the region of 2200 cm–1 (41). For film samples, the subtraction was based on the double peak of polyethylene observed at 1472 and 1462 cm–1. Baseline correction (1804–904 cm–1) and seven-point Savitsky-Golay smoothing were then applied to the spectra. Peak positions were determined using the peak-finding function in the OmnicTM program. Second-derivative analysis was employed to locate underlying peaks in the zero order spectra. The resulting second-derivative curves were splined by changing data spacing from 2.0 to 0.0625 cm–1 to provide accurate determination of the peak position. 5.4. Results and Discussion A shift in the frequency of a specific peak in the FTIR spectrum of a molecule or a change in the peak shape can represent a change in the vibrational energy of a particular functional group. Such changes may result when two species physically interact with one another. These changes may represent structural changes in the molecules as a consequence of complexation. In this study, several vibrational modes were monitored upon titration of DNA with cationic lipids to probe potential structural alterations.

Synthetic Vectors by IR Spectroscopy

303

To study changes in the environment of the lipid–DNA interface upon interaction, the authors followed the change in peak position of the ester C = O stretching vibration of DOTAP around 1739 cm–1, the C-H bending modes of the trimethylammonium group of DOTAP at 1485 cm–1, and the antisymmetric and symmetric stretching vibrations of the DNA PO2– at 1225 and 1086 cm–1, respectively. The DNA carbonyl stretching, arising from the G and T bases (1717 cm–1), reflects the hydrogen bonding of the DNA base pairs, and was analyzed to determine the effect of the cationic lipid on the conformation of DNA double helix. Finally, the CH3 umbrella-type deformation at 1377 cm–1 was monitored to follow changes in the hydrophobic core of the lipid bilayer upon complexation with DNA. The cationic lipid DOTAP has a gelAliquid crystalline phase-transition temperature below room temperature. Therefore, the pure liposome suspension was presumably in a liquid crystalline phase for this study. With the exception of some film results, all samples appeared to maintain DNA in a B form (Fig. 6A). No clear change from B to any of the other forms discussed could be definitively observed. Although small shifts in spectral features were observed as described below, no changes in the helix-sensitive bands such as those observed at 1425, 1375, 1281, and 970 cm–1 occurred (see Table 1). This is not to say that structural transitions from B form to some other alternate helical geometries do not occur, but just that under the solution conditions used in this study (e.g., moderate concentration, H2O instead of D2O) no clear indications of changes in helical geometries were observed. More work is needed to unambiguously address this issue, however. When DNA and DNA complexes are dried on polyethylene cards, the exact form that the DNA adopts is less certain. This ambiguity is partially the result of the inability to effectively subtract the Tris buffer spectra from that of the complex itself. This probably reflects the fact that the spectral properties of the Tris cation change upon binding to the negative charge of the DNA phosphate. The presence of bands from the buffer at 1630, 1470–1370, 1296, ~1050, and 1039 cm–1, makes structural assignments difficult, but changes in some of the more distinctive DNA signals not obscured by the buffer spectra indicate a conversion to some state other than the canonical B form. This is especially evident for the in-plane C = O stretching of the bases, which is now observed at 1705 cm–1, suggesting the presence of more dehydrated A-form DNA. Nevertheless, the antisymmetric and symmetric stretching vibrations of the phosphate groups maintain the peak positions corresponding to the B form at 1225 cm–1 (Fig. 6B), as opposed to the higher frequency of 1240 cm–1 that is commonly seen with A-form DNA. Conversion from the B to the A, C, or Z form is known to occur upon variation of the water content of the sample (5). One possible explanation for these results involves an incomplete drying of the sample, leaving some water associated with the DNA, which could be responsible for partial conservation of the B form in this thin film

304

Choosakoonkriang et al.

Fig. 6. FTIR absorbance spectra for DOTAP–DNA complexes in solution (top) and on polyethylene cards (bottom). The spectra were collected and processed as described in the text. The weight ratio of cationic lipid:DNA is indicated on the left. 1 mg/mL DNA was present in all cases.

state. Additionally, because the size of the plasmid is large, the coexistence of different forms of DNA within a single molecule may be possible, each with different sensitivity to water activity. The carbonyl absorption of DOTAP occurs at about 1739 cm–1 in the solution state. As indicated previously, the carbonyl-stretching mode of lipids, such as DOTAP has separate but overlapping bands, which are not resolved. Mantsch et al. (23) argue that the two peaks are caused by the presence of carbonyl species with different hydrogen bonding interactions. In complexes where DNA is present in large excess compared to DOTAP (e.g., 1:5 DOTAP:DNA [w/w]), the ester C = O stretching vibration occurs at 1740 cm–1 (Fig. 7A). As

Synthetic Vectors by IR Spectroscopy

305

Fig. 7. Effect of DOTAP:DNA weight ratio on DNA and lipid vibrational modes in solution. The peak positions were plotted vs the weight ratio of lipid to DNA. (A) Lipid carbonyl stretch. (B) DNA base in-plane double-bond stretch. (C) Lipid headgroup C-N stretch and in-plane vibration of cytosine. (D) Methyl region of lipid and in-plane base vibration of DNA. (E) DNA antisymmetric phosphate stretch.

the relative amount of DOTAP is increased in the complexes, the frequency shifts to 1738 cm–1, reaching a plateau at a 1:1 DOTAP:DNA weight ratio. This suggests that when DNA is in excess, the ability of the DOTAP ester

306

Choosakoonkriang et al.

carbonyls to interact with the solvent is decreased (i.e., dehydration of the lipid headgroups occurs). This idea is supported by a recent study by Hirsch-Lerner and Barenholz (42). Using differential scanning calorimetry and fluorescence measurements, it was determined that the lipid headgroup–solvent interface is dehydrated upon complexation with DNA. The band at ~1717 cm–1 reflects the base pairing of B form DNA. As the amount of the DOTAP is increased, this peak shifts to a higher frequency (Fig. 7B). This change to higher vibrational energy suggests a stronger interaction between the hydrogen-bonded bases. This change is corroborated by differential scanning calorimetry, which demonstrates an increase in the helix–coil transition temperature of the dsDNA upon complexation with cationic lipid (see Chapter 21 by Lobo et al.). With smaller amounts of lipid, the peak shift goes through a small maximum, suggesting some type of structural rearrangement at lower stoichiometries. The antisymmetric bending vibration of the +N(CH3)3 headgroup of DOTAP is seen at ~1485 cm–1; the symmetric bend is near 1405 cm–1. Both of these vibrations are of weak intensity. Unfortunately, DNA has a peak at 1492 cm–1 produced by the NH/CH in-plane base deformation (C,G) (5) that obscures the DOTAP peak at 1485 cm–1. As shown in Fig. 7C, however, a shift in this peak position is observed upon DNA complexation. The nature of this shift is difficult to ascertain due to this peak overlap. At low amounts of lipid, the DNA band is expected to dominate; at the higher lipid:DNA weight ratios, the lipid peak should dominate, producing an apparent shift in overall peak position. However, in a study of the effects of acetic acid on the solution structure of calf thymus DNA, a shift down of the DNA in-plane vibrations of C and G from 1492 to 1485 cm–1 was observed (43). This change in peak position was attributed to the protonation states of the C and G bases. Based on this, the authors' results may indicate that the shift of the peak at 1492 cm–1 to a lower frequency is the result of an interaction of the cationic lipid with C and G bases. These results were also observed in dimethyldioctadecylammonium bromide (DDAB)–DNA complexes (data not shown). The +N(CH3)3 symmetric vibration band at 1405 cm–1 could not be clearly resolved under these experimental conditions, since it is of weak intensity. The +N(CH3)3 antisymmetric stretch at 972 cm–1 overlaps with the 970 cm–1 band of the deoxyribose vibration. Upon plotting the frequency of this peak vs weight ratio, a peak shift was not observed (data not shown). This is not surprising, considering the large overlap of the two peaks, and illustrates the potential problem of the lack of detection of shifts in overlapping peaks when IR peak position is the only method employed in the analysis of such situations. Figure 7D shows the effect of DNA binding on the CH3 symmetric bending vibration of DOTAP. In this region, DNA exhibits a peak at ~1374 cm–1 (sugarglycosidic linkage of guanosine and adenosine, C2' endo/anti) (5), which gradually

Synthetic Vectors by IR Spectroscopy

307

shifts to 1377 cm–1, upon lipid binding. As in many of the examples described here, saturation occurs substoichiometrically. Again, the shift in this peak position could result from relative intensities of the DOTAP peak vs the DNA peak at different weight ratios rather than arising from a true structural change. An increase in the antisymmetric phosphate stretching vibration peak from 1224 to 1226 cm–1 was also observed (Fig. 7E). This increase in vibrational frequency can be interpreted as either being caused by an ionic interaction with the lipid headgroup or decreased hydrogen bonding with the solvent (32). This trend seems to plateau at a 1:1.5 lipid to DNA weight ratio, which corresponds to a 0.35:1 lipid+/DNA– charge ratio, suggesting that the interaction saturates well before charge neutrality is reached. This further suggests that the shift may not merely result from an ionic interaction of the cationic lipid headgroup with the DNA phosphates, but involves some type of structural rearrangement. No change is observed in the symmetric phosphate stretching vibration at 1089 cm–1. The same regions of the lipid–DNA spectra were also studied using thin-film samples dried on PE cards (Fig. 6B). Between a 1:2.5 and 1:2 lipid–DNA weight ratio, a distinct change occurs in the spectra from 1750 to 800 cm–1. Again, these changes are observed at substoichiometric ratios of lipid:DNA phosphate. Figure 8A shows the lipid ester carbonyl stretching vibration as a function of lipid:DNA weight ratio for samples deposited on a polyethylene film. The results obtained are similar to those obtained for complexes in solution. The only difference is that for samples below a 1:3 lipid:DNA weight ratio the peak is seen at the frequency of the lipid itself. With the exception of the two points at low weight ratios, the trend again seems to indicate that when the lipid is bound to the DNA it is less able to interact with the solvent water, as evidenced by the shift to higher vibrational frequencies. Why the low weight ratio complexes do not behave the same in a film as they do in solution is unknown. The lipid-induced change in the carbonyl stretching region of the DNA G and T bases also shows a trend similar to that seen in solution (Fig. 8B), with the main difference being an overall shift in the peak positions from approx 1716–1720 cm–1 to 1705–1708 cm–1. This difference in peak position from the film compared to the solution result presumably reflects the different water activities of the solution and film states. This is also consistent with the idea that the conformation of DNA undergoes a transition from the B to other forms upon lowering of the relative humidity (5). Thus, distinct changes in the doublehelix hydrogen bonding are observed upon addition of DOTAP. It should also be noted that the shapes of the titration curves in Figs. 7B and 8B are similar, consistent with some type of specific structural alteration. In contrast, the antisymmetric phosphate stretching vibration shows the exact opposite trend in the film that is seen in the solution results (Fig. 8C). Although the initial frequency is the same in the film and solution, the frequency of this

308

Choosakoonkriang et al.

Fig. 8. Effect DOTAP:DNA weight ratio on DNA and lipid vibrational modes in thin films. The peak positions are plotted against the weight ratio of cationic lipid:DNA. (A) Lipid carbonyl stretch. (B) DNA base in-plane double-bond. (C) DNA antisymmetric phosphate stretch.

vibration decreases as the relative amount of lipid is increased. This trend begins to saturate at a 1:1 lipid:DNA weight ratio, significantly higher than that seen in solution. A shift in frequency upon addition of lipid to the DNA is perhaps not surprising and presumably reflects an altered environment of the phosphate on complexation with the lipid. No bands that would indicate formation of Z-form DNA, which exhibits an antisymmetric phosphate stretching band at 1215 cm–1, were observed under any conditions. The remaining vibrations studied in solution, namely, the C-H bending modes of the trimethylammonium group of the lipid and the CH3 umbrellatype deformation, showed no significant change in frequency in the range of 1:2 to 3:1 lipid:DNA weight ratios. These peaks, however, could not be

Synthetic Vectors by IR Spectroscopy

309

accurately determined at weight ratios below 1:2. Whether the loss of these peaks is significant at these small weight ratios or the S:N ratio was too low for these peaks to be resolved is unknown. At weight ratios above 1.5:1 DOTAP:DNA, gross aggregation of the samples occurred, and lack of homogeneity may have influenced the relative intensities of overlapping modes. FTIR requires a high sample concentration compared to many other techniques used to characterize polycation:DNA complexes, and it is difficult to prepare the complexes at high concentrations of DNA without precipitation. However, other techniques such as fluorescence, CD (44), and small-angle X-ray scattering (45), have shown that the nanoscale properties of cationic lipid–DNA complexes do not necessarily depend on the colloidal properties of the complexes, suggesting that a rigorous interpretation of these spectral changes even in the presence of aggregation may be feasible. 6. Application to Polymer–DNA Complexes (PEI–DNA) Most nonviral gene delivery systems are formed by the condensation of DNA into nanometer-scale particles by the electrostatic interaction between the polyanionic DNA and a cationic carrier. In this condensed form, the DNA is protected from degradation. Positively charged particles are thought to then bind to anionic cell-surface proteoglycans and enter cells through a conventional endocytotic process. Currently, a number of synthetic cationic polymers are being investigated for this purpose in addition to cationic lipids. One such polymer, PEI, has shown significant potential for efficient gene delivery (46). The toxicity of PEI is a major concern, but the lower molecular weight and more linear forms of PEI have been shown to be less toxic than the branched and higher molecular weight forms (47,48). Therefore, the authors performed an initial characterization of PEI–DNA complexes using FTIR as a further example of this approach. It was found that similar FTIR analyses can be performed using polylysine, amino dendrimers, and peptoids (N-substituted polyglycines) as well.

6.1. Methods The methods used in this study are the same as those used in the solution studies of the cationic lipid–DNA complexes described in Subheading 5., with the following modifications. Equal amounts of DNA were added dropwise to appropriate concentrations of PEI solution (mol wt 25,000) with constant stirring to give nitrogen:phosphorous molar ratios (N:P) of 0.5:1, 1:1, 2:1, 7:1, and 10:1. Representative spectra are shown in Fig. 9. Samples precipitated from solution at ratios from 3:1 to 5:1, and data were not collected. PEI does not have an IR absorption band in the region of the DNA sugar-phosphate backbone, or in the region of C = O stretching above 1700 cm–1 (49). Therefore, for the analysis presented here, the

310

Choosakoonkriang et al.

Fig. 9. FTIR absorbance spectra of PEI–DNA complexes in solution. The spectra were collected and processed as described in the text. The ratio of PEI N:DNA phosphate is indicated on the left, along with both DNA and PEI alone. 1 mg/mL DNA was present in all cases.

authors examined the PO2- antisymmetric and symmetric vibrations, as well as the C = O stretching vibration of the DNA. The PEI N-H bending vibration peak at 1565 cm–1 was also examined.

6.2. Results and Discussion Changes in the selected vibrational frequencies on addition of PEI to DNA are shown in Fig. 10. In panel A, a continuous increase in peak frequency initially at 1713 cm–1, assigned to the in-plane double-bond vibration of the bases, occurs from about 0.5:1 to 2:1 N:P. At this point, the peak plateaus at frequency values approx 1715 cm–1, even when the amount of PEI increases through 10:1 N:P. A similar shift to higher frequency is observed in the antisymmetric phosphate region of DNA (Fig. 10C). In this case, a small initial decrease in the frequency of ~1 cm–1 is observed from 0.5:1 to 1:1 N:P, then a gradual increase through 2:1 N:P, followed again by a plateau region. At excessive PEI:DNA ratios (7:1 and above), the peak once again begins to gradually shift to higher frequencies. For the symmetric phosphate vibration, no change in peak position was observed, as seen with the lipid–DNA complexes. Note that in all cases, data at intermediate PEI:DNA ratios are not available due to precipitation of the complexes. The peak at 1374 cm–1 (Fig. 10B) corresponds to the in-plane vibration of the base pairs, specifically the dAdG antisymmetric vibration. Changes in frequency at low N:P values are difficult to quantify, because of high error

Synthetic Vectors by IR Spectroscopy

311

Fig. 10. Effect of the molar ratio of PEI:DNA complexes on DNA vibrational modes in solution. The peak positions are plotted against the N:P ratio of PEI:DNA. The region of colloidal instability is indicated by ppt. (A) DNA base in-plane double bond (B) In-plane base vibration of DNA. (C) DNA antisymmetric phosphate stretch. (D) Deoxyribose C-C main vibration.

values, but a definite increase of ~1 cm–1 is observed from 1:1 to 2:1 N:P, with the higher N:P ratio complexes exhibiting peaks at frequencies about 0.5 cm–1 lower than DNA itself. The phosphodiester vibrations, coupled to vibrations from the deoxyribose, produce the peak at about 969 cm–1. The change in the vibrational frequency of

312

Choosakoonkriang et al.

this mode occurs in a pattern almost the inverse of that seen in the 1224 cm–1 peak (Fig. 10C). This peak is used in many papers as an internal standard, since it has been demonstrated to be constant upon the interaction of DNA with aspirin (36) or carbohydrates (31). In light of this, the PEI-induced change in this peak is intriguing, and further studies are merited to determine the origin of this variation. The PEI peak corresponding to the +NH2 symmetric bending is readily observed in samples containing PEI alone at high concentrations. However, upon complex formation, this peak was undetectable. Whether this results from an interaction with the DNA that alters this peak, or merely from the relatively low concentration of the PEI compared to the standard sample, is unclear. The data presented here is too preliminary to draw any definite conclusions regarding the nature of the interaction between the two components. However, it is obvious that PEI and DNA do interact, and that this interaction is dependant on the relative concentration of the two components in the complex. A study of DNA–PEI interactions in multilayer films by Sukhorukov et al. (49) indicates that the interaction of PEI and DNA may be based on both electrostatic interactions and hydrogen bonding between DNA phosphates and the amino groups of the PEI (49). These interactions may be the cause of the changes in both the DNA carbonyl and antisymmetric phosphate stretching modes seen here. 7. Conclusion IR spectroscopy has long been established as a powerful technique for the analysis of small molecule samples. Qualitative and quantitative analysis can be conducted easily, organic as well as inorganic samples can be studied, and it can be applied to samples in multiple physical states. The development of FTIR spectroscopy has allowed this approach to be further extended to the analysis of complex macromolecular systems. The ready availability of easy-to-use sampling methods for a variety of sample types has also expanded the use of this technique. The characterization of DNA and DNA–polycation complexes is a topic of great interest to research labs, pharmaceutical and biotechnology companies, and regulatory agencies. FTIR has the potential to detect structural changes in DNA, as well as in bound delivery polymers, and thus be of general utility to these general groups. Distinct changes are seen in certain peak positions at substoichiometric weight ratios of lipid:DNA. The involvement of both the antisymmetric phosphate vibration and carbonyl region of both DNA and lipid in complexation is evident on the basis of the preliminary studies described above. FTIR studies of such complexes are in their infancy, but the potential for the acquisition of unique information is compelling. A comprehensive characterization of DNA and DNA–polymer complexes will not be accomplished by a single technique. A combination of calorimetric

Synthetic Vectors by IR Spectroscopy

313

and other spectroscopic methods, discussed elsewhere in this text, will be necessary. Although not the subject of this review, another powerful technique exists that examines the vibrational transitions of molecules. In Raman spectroscopy, a change in the polarizability of the molecule must occur, as opposed to the change in the dipole moment that produces an IR-active transition. Thus, some transitions can be detected by Raman that are not seen in IR-spectroscopy, and vice versa. Some transitions are, of course, detectable by both methods. In modern Raman spectroscopy, a sample is illuminated with intense laser of UV, visible, or near-IR wavelength, and the light scattered by the sample is examined. Most scattered light is unchanged by the interaction (i.e., Rayleigh scattered light), but a small portion is of higher or lower energy, because of energetic interactions with the vibrational transitions. The energy difference between these weak peaks and the Rayleigh scattered light corresponds to the Raman spectrum of the target molecule. Raman spectroscopy has both distinct advantages and disadvantages compared to FTIR spectroscopy. Because a frequency of light can be used that is removed from the major solvent absorption bands, spectral overlap of the molecular vibrational bands of interest with those of the solvent tends to be much less of a problem. Water bands are only weakly Raman-active, making aqueous studies much simpler. Polynucleotides and complexes of polynucleotides in chromatin, viruses, and related structures have already been extensively studied by Raman spectroscopy, and detailed assignments now exist (50). A major disadvantage to this technique is that high concentrations of macromolecules are usually necessary to obtain quality spectra. Typically, solutions of greater than 10 g/L are employed, with concentrations above 30 g/L often used. Only small volumes of sample (a few microliters), however, are usually necessary since experiments are performed in small capillary tubes. Recent advances in UV resonance Raman spectroscopy, however, do permit lower concentrations to be examined (51). In this approach, the molecule is excited within an electronic absorption band (e.g., purine or pyrimidine base), which results through a coupling process in vibrational band intensity enhancement of many orders of magnitudes. In general, Raman instrumentation is not widely available and is more expensive than FTIR spectrometers, which currently relegates this method to specialists. Future studies of gene delivery complexes using Raman approaches should, however, result in increased interest in this method. References 1. Tsuboi, M. (1969) Application of infrared spectroscopy to structure studies of nucleic acids. Appl. Spectrosc. Rev. 3, 45–90. 2. Taillandier, E. and Liquier, J. (1992) Infrared spectroscopy of DNA, in Methods in Enzymology. Academic, New York, pp. 307–335.

314

Choosakoonkriang et al.

3. Fritzsche, H. (1991) Infrared spectroscopy and infrared linear dichroism of nucleic acids. J. Mol. Struct. 242, 245–261. 4. Lewis, R. N. A. H. and McElhaney, R. N. (1996) Infrared Spectroscopy of Biomolecules (Mantsch, H. H. and Chapman, D., eds.), Wiley-Liss, New York, pp. 159–202. 5. Liquier, J. and Taillandier, E. (1996) Infrared spectroscopy of nucleic acids, in Infrared Spectroscopy of Biomolecules (Mantsch, H. H. and Chapman, D., eds.),Wiley-Liss, New York, pp. 131–158. 6. Ghomi, M., Letellier, R., Liquier, J., and Taillandier, E. (1990) Interpretation of DNA vibrational spectra by normal coordinate analysis. Int. J. Biochem. 22, 691–699. 7. Brahms, J., Pilet, J., Phuong Lan, T. T., and Hill, L. R. (1973) Direct evidence of the C-like form of sodium deoxyribonucleate. Proc. Nat. Acad. Sci. USA 70, 3352–3355. 8. Loprete, D. M. and Hartman, K. A. (1989) Existence of C structure in poly (dAdC) poly (dG-dT). J. Biomol. Struct. Dynam. 7, 347–362. 9. Kang, H. and W. C. Johnson, J. (1994) Infrared linear dichroism reveals that A-, B-, and C-DNAs in films have bases highly inclined from perpendicular to the helix axis. Biochemistry 33, 8330–8338. 10. Akao, T., Fukumoto, T., Ihara, H., and Ito, A. (1996) Conformational change in DNA induced by cationic bilayer membranes. FEBS Lett. 391, 215–218. 11. Zuidam, N. J., Barenholz, Y., and Minsky, A. (1999) Chiral DNA packing in DNA-cationic lipid assemblies. FEBS Lett. 457, 419–422. 12. Maestre, M. F. and Reich, C. (1980) Contribution of light scattering to circular dichroism of DNA films, DNA-polylysine complexes, and DNA particles in ethanolic solutions. Biochemistry 19, 5214–5223. 13. Middaugh, C. R., Evans, R. K., Montgomery, D. L., and Casimiro, D. R. (1998) Analysis of plasmid DNA from a pharmaceutical perspective. J. Pharm. Sci. 87, 130–146. 14. Bailly, F., Bailly, C., Colson, P., Houssier, C., and Henichart, J. P. (1993) A tandem repeat of the SPKK peptide motif induces psi-type DNA structures at alternating AT sequences. FEBS Lett. 324, 181–184. 15. Zhu, N., Liggitt, D., Liu, Y., and Debs, R. (1993) Systemic gene expression after intravenous DNA delivery into adult mice. Science 261, 209–211. 16. Liu, Y., Liggitts, D., Zhong, W., Tu, G., Gaensler, K., and Debs, R. (1995) Cationic liposome-mediated intravenous gene delivery. J. Biol. Chem. 270, 24,864–24,870. 17. Gorman, C. M., Aikawa, M., Fox, B., Fox, E., Lapuz, C., Michaud, B., et al. (1997) Efficient in vivo delivery of DNA to pulmonary cells using the novel lipid EDMPC. Gene Ther. 4, 983–992. 18. Mounkes, L. C., Zhong, W., Cipres-Palacin, G., Heath, T. D., and Debs, R. J. (1998) Proteoglycans mediate cationic liposome-DNA complex-based gene delivery in vitro and in vivo. J. Biol. Chem. 273, 26,164–26,170. 19. Stephan, D. J., Yang, Z.-Y., San, H., Simari, R. D., Wheeler, C. J., Felgner, P. L., et al. (1996) A new cationic liposome DNA complex enhances the efficiency of arterial gene transfer in vivo. Human Gene Ther. 7, 1803–1812. 20. Pitard, B., Oudrhiri, N., Vigneron, J.-P., Hauchecorne, M., Aguerre, O., Tour, R., et al. (1999) Structural characteristics of supramolecular assemblies formed by

Synthetic Vectors by IR Spectroscopy

21.

22.

23.

24.

25. 26. 27. 28.

29.

30. 31.

32.

33.

34.

315

guanidinium-cholesterol reagents for gene transfection. Proc. Natl. Acad. Sci. USA 96, 2621–2626. Thierry, A. R., Lunardi-Iskandar, Y., Bryant, J. L., Rabinovich, P., Gallo, R. C., and Mahan, L. C. (1995) Systemic gene therapy: biodistribution and long-term expression of a transgene in mice. Proc Natl Acad Sci USA 92, 9742–9746. Parker, S. E., Khatibi, S., Margalith, M., Anderson, D., Yankauckas, M., Gromkowski, S. H., et al. (1996) Plasmid DNA gene therapy: studies with the human interleukin-2 gene in tumor cells in vitro and in the murine B16 melanoma model in vivo. Cancer Gene Ther. 3, 175–185. Mantsch, H. H. and McElhaney, R. N. (1991) Phospholipid phase transitions in model and biological membranes as studied by infrared spectroscopy. Chem. Phys. Lipids 57, 213–226. Choosakoonkriang, S., Wiethoff, C. M., and Middaugh, C. R. (2001) Analysis of cationic lipid/DNA complexes by infrared spectroscopy. J. Biol. Chem. 276, 8037–8043. Tamm, L. K. and Tatulian, S. A. (1997) Infrared spectroscopy of proteins and peptides in lipid bilayers. Q. Rev. Biophys. 30, 365–429. Colthup, N. B., Daly, L. H., and Wiberley, S. E. (1975) Introduction to Infrared and Raman Spectroscopy, 2nd ed. Academic, New York. Griffiths, P. R. and Haseth, J. A. D. (1986) Fourier transform infrared spectroscopy. Wiley, New York. Oberg, K. A. and Fink, A. L. (1998) A new attenuated total reflectance fourier transform infrared spectroscopy method for the study of proteins in solution. Anal. Biochem. 256, 92–106. Culler, S. R. (1993) Diffuse reflectance infrared spectroscopy: sampling techniques for qualitative/quanitative analysis of solids, in Practical Sampling Techniques for Infrared Analysis (Coleman, P. B., ed.), CRC Press, London, pp. 93–106. Kasbauer, M., Junglas, M., and Bayerl, T. M. (1999) Effect of cationic lipids in the formulation of asymmetries in supported bilayers. Biophys. J. 76, 2600–2605. Tajmir-Riahi, H. A., Naoui, M., and Diamantoglou, S. (1994) DNA-carbohydrate interaction. The effects of mono- and disaccharides on the solution structure of calf-thymus DNA. J. Biomol. Struct. Dynam. 12, 217–234. Heidar-Ali, Tajmir-Riahi, H. A., and Messaoudi, S. (1992) The effects of monovalent cations Li+, Na+, K+, NH4+, Rb +, and Cs+ on the solid and solution structures of the nucleic components. Metal ion binding and sugar conformation. J. Biomol. Struct. Dynam. 10, 345–365. Tajmir-Riahi, H. A., Naoui, M., and Ahmad, R. (1993) Effects of Cu2+ and Pb2+ on the solution structure of calf thymus DNA: DNA condensation and denaturation studied by fourier transform IR difference spectroscopy. Biopolymers 33, 1819–1827. Tajmir-Riahi, H. A., Naoui, M., and Ahmad, R. (1993) The effects of cobalthexa-amine and cobalt-penta-amine cations on the solution structure of calf-thymus DNA. DNA condensation and structural features studied by FTIR difference spectroscopy. J. Biomol. Struct. Dynam. 11, 83–93.

316

Choosakoonkriang et al.

35. Ahmad, R., Naoui, M., Neault, J. F., Diamantoglou, S., and Tajmir-Riahi, H. A. (1996) An FTIR spectroscopic study of calf-thymus DNA complexation with Al(III) and Ga(III) cations. J. Biomol. Struct. Dynam. 13, 795–802. 36. Neault, J. F., Naoui, M., Manfait, M., and Tajmir-Riahi, H. A. (1996) AspirinDNA interaction studied by FTIR and laser raman difference spectroscopy. FEBS Lett. 382, 26–30. 37. Medina, M. A., Ramirez, F. J., Ruiz-Chica, J., Chavarria, T., Lopez-Navarrete, J. T., and Sanchez-Jimenez, F. (1998) DNA-Chlopheniramine interaction studied by spectroscopic techniques. Biochim. Biophys. Acta 1379, 129–133. 38. Walters, L. and Dev, S. B. (1989) Conformational analysis of a peptide-DNA interaction by fourier transform infrared spectroscopy (FTIR), in Peptides (Rivier, J. E. and Marshall, G. R., eds.), ESCOM Science, Leiden, pp. 702–703. 39. Kim, J. S., Lee, S. A., Carter, B. J., and Rupprecht, A. (1997) Stabilization of the B conformation in unoriented films of calf thymus DNA by NaCl: a raman and IR study. Biopolymers 41, 233–238. 40. White, A. P. and Powell, J. W. (1995) Observation of hydration-dependent conformation of the (dG)20.(dG)20(dC)20 oligonucleotide triplex using FTIR spectroscopy. Biochemistry 34, 1137–1142. 41. Alex, S. and Dupuis, P. (1989) FT-IR investigation of cadmium binding by DNA. Inorg. Chim. Acta. 157, 271–281. 42. Hirsch-Lerner, D. and Barenholz, Y. (1999) Hydration of lipoplexes commonly used in gene delivery: follow-up by laurdan fluorescence changes and quantification by differential scanning calorimetry. Biochim. Biophys. Acta 1461, 47–57. 43. Tajmir-Riahi, H. A., Neault, J. F., and Naoui, M. (1995) Does DNA acid fixation produce left-handed Z Structure? FEBS Lett. 370, 105–108. 44. Zuidam, N., Hirsch-Lerner, D., Margulies, S., and Barenholz, Y. (1999) Lamellarity of cationic liposomes and mode of preparation of lipoplexes affect transfection efficiency. Biochim. Biophys. Acta 1419, 207–220. 45. Koltover, I., Salditt, T., and Safinya, C. (1999) Phase diagram, stability, and overcharging of lamellar cationic lipid-DNA self-assembled complexes. Biophys. J. 77, 915–924. 46. Kichler, A., Behr, J. P., and Erbacher, P. (1999) Polyethyleneimines: A family of potent polymers for nucleic acid delivery, in Nonviral Vectors for Gene Therapy (Huang, L., Hung, M.-C., and Wagner, E., eds.), Academic, San Diego, pp. 191–206. 47. Godbey, W. T., Wu, K. K., and Mikos, A. G. (1999) Tracking the intracellular path of poly(ethylenimine)/DNA complexes for gene delivery. Proc. Natl. Acad. Sci. USA 96, 5177–5181. 48. Fischer, D., Bieber, T., Li, Y., Elsasser, H. P., and Kissel, T. (1999) A novel nonviral vector for DNA delivery based on low molecular weight, branched polyethylenimine: effect of molecular weight on transfection efficiency and cytotoxicity. Pharm. Res. 16, 1273–1279. 49. Sukhorukov, G. B., Montrel, M. M., Petrov, A. I., Shabarchina, L. I., and Sukhorukov, B. I. (1996) Multilayer films containing immobilized nucleic acids. Their structure and possibilities in biosensor applications. Biosensor Bioelectronics 11, 913–922.

Synthetic Vectors by IR Spectroscopy

317

50. Thomas, G. J., Jr. and Tsuboi, M. (1993) Raman spectroscopy of nucleic acids and their complexes. Adv. Biophys. Chem. 3, 1–70. 51. Wen, Z. Q., Armstrong, A., and Thomas, G. J., Jr.(1999) Demonstration by ultraviolet resonance Raman spectroscopy of differences in DNA organization and interactions in filamentous viruses Pf1 and fd. Biochemistry 38, 3148–3156.

Cationic Vectors by ITC and DSC

319

21 Characterization of Cationic Vector-Based Gene Delivery Vehicles Using Isothermal Titration and Differential Scanning Calorimetry Brian A. Lobo, Sheila A. Rogers, Christopher M. Wiethoff, Sirirat Choosakoonkriang, Susan Bogdanowich-Knipp, and C. Russell Middaugh 1. Introduction Within the past 10 years, major advances in the design and development of differential scanning calorimeters (DSC) (1) and isothermal titration calorimeters (ITC) (2) have resulted in an unparalleled level of sensitivity, stability, and reproducibility in calorimetric measurements of large molecules. These improvements have allowed the thermal stability and ligand binding processes of biological macromolecules to be thermodynamically characterized with speed, accuracy, and convenience. With their increasing commercial availability, experiments that were previously limited to specialist calorimetry laboratories can now be routinely performed by most investigators. The fundamental advantage of DSC and ITC remains their ability to directly measure a thermodynamic parameter during a thermal transition or binding event: heat capacity (6CP) for DSC and enthalpy (6H) for ITC. Under ideal conditions (e.g., equilibrium, no aggregation, and so on) additional parameters (free energy [6G], transition enthalpy [6S]) can be extracted from a single well-designed experiment to obtain a complete thermodynamic description of the process under investigation. In pace with these technological improvements, advances in the theory of biological calorimetry have also been made, especially with respect to the partitioning of heat capacity changes upon ligand binding and the deconvolution of the unfolding of macromolecules into individual events (3). From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

319

320

Lobo et al.

These advancements in DSC and ITC instrumentation and theory have led to an elegant description of the thermodynamic factors involved in the complex processes of protein folding and unfolding and ligand binding (3–6). Such approaches also have the potential to provide new information concerning thermodynamics of processes that influence the stability and formation of more complex physical systems, such as nonviral gene delivery vectors. The development of these vectors over the past 15 years has generally been empirical, with maximized transfection efficiency the major experimental endpoint. Only recently has the importance of their physical characterization been realized and given increased attention. Even as the microscopic and macroscopic properties of these vectors are being elucidated (7–12), a fundamental thermodynamic description of their formation and stability remains unavailable. These vectors are known to self-assemble in what is generally thought to be a nonspecific manner and are significantly affected by their colloidal properties. Therefore, ITC and DSC are ideal techniques to monitor the effects of kinetic and solution conditions upon their formation and thermal stability. Solution concentration and ionic strength can be controlled during each measurement and kinetic processes can be perturbed via changes in titration program and scan rate with ITC and DSC, respectively. As is emphasized throughout this chapter, the collection of useful data from these techniques is not normally limited by the instruments themselves (which have sophisticated automation programs), but rather by the ability of the scientist to prepare these vectors and their components with acceptable levels of quality and reproducibility and by problems in analysis induced by vector aggregation during the experimental process. Since the experimental methods and theory for ITC and DSC have been well-established for the characterization of protein–protein and DNA–protein interactions (13–14), they are useful as foundations for the analysis of representative nonviral gene delivery vectors. ITC and DSC are discussed individually, beginning with a brief review of instrumentation and theory in each case. Experimental procedures and selective data are then presented as a guide to representative approaches that are particularly relevant to DNA–polymer complexes. The authors conclude with a discussion of the limitations of both methods and suggest additional experimental approaches that may be necessary to better understand the basis of these limitations. 2. Isothermal Titration Calorimetry 2.1. Instrumentation Because ITC instrumentation is only briefly described here, the reader is referred to an excellent review by Freire et al. (15) for additional information on ITC design and theory. A description of the MicroCal Omega calorimeter and its operation procedures is also available (13).

Cationic Vectors by ITC and DSC

321

The basic design of an isothermal titration calorimeter consists of sample and reference cells that are contained in a heat-sink block maintained at the desired temperature by either a large water bath or some type of Peltier device. In a typical experiment, the sample cell contains a solution of the macromolecule (titrate) and the reference cell an equivalent volume of a solution with a similar heat capacity (buffer or water for the usual dilute solution measurements) to offset the effects of random electrical and thermal noise. The heating block contains thermoelectric sensors to convert the difference in heat between the two cells into a voltage difference, which is recorded over time, as the experiment progresses. Various methods are used to improve instrument sensitivity and response, including power compensation (2,15) and mathematical deconvolution of the decaying heat signal (16). The ligand solution (titrant) is contained in a syringe outside the bath. Aliquots of ligand are periodically injected into the receptor solution in the sample cell via a motor-driven plunger. Dual-injection calorimeters contain a second ligand injection system and titrant is injected simultaneously into both the sample and reference cells (15). This design directly compensates for heats of dilution of the ligand solution and heat effects caused by the mechanical injection process, obviating the need for separate blank titrations. The contents of the sample cell are rapidly stirred in an attempt to maintain the contents at equilibrium throughout the experiment.

2.2. Models of Ligand Binding Theoretical aspects of ITC have been presented in numerous reviews (2,13,15), and only the most popular approaches are summarized here. The measured experimental variable in an ITC experiment is the heat evolved or absorbed upon binding of the ligand to the macromolecule. In the absence of other thermal effects (e.g., aggregation, conformational changes and so on), the heat/injection is directly proportional to the amount of ligand bound/injection: q = V 6H 6[LB]

(1)

where q is the heat released or absorbed/injection, V is the sample cell volume, 6H is the molar reaction enthalpy (heat/mol ligand) and 6[LB] is the change in concentration of bound ligand/injection. Alternatively, the heat can be expressed as the total cumulative heat Q, which is proportional to the total amount of bound ligand present [LB] after each injection: Q = V 6H Y 6[LB] = V6H [LB]

(2)

For ITC data analysis the binding equilibria must be written in terms of the total ligand concentration since this is the known independent variable. The simplest model describes the binding of a ligand to a receptor (macromolecule) of molar concentration [M] to a single type of binding site,

322

Lobo et al.

characterized by a stoichiometry n, an association constant K and a binding enthalpy 6H. Expressed in terms of the total heat, such a binding equilibria takes the familiar form: [LB] = [M] n K [L] 1 + K [L] Q = V 6H [LB] = V [M] n6H K [L] 1 + K [L]

(3)

(4)

Substituting Eq. 3 into Eq. 4 and applying mass balance of ligand [LT] = [LB] + ]L] (where [LT], [LB], and [L] represent the total, bound, and free ligand concentrations, respectively), one arrives at a closed-form solution of the experimental total heat as a function of the total ligand concentration, with n, K, 6H, and V as unknown parameters: Q2( K ) + Q(–1 – [M] n K – K [LT]) + ([M] V 6H n K [LT]) = 0 V6H

(5)

Q = (1 + [M]nK + K[LT]) – [(1 + [M] n K + K[LT])2 – 4[M] n K2 [LT]]1/2 (2K/ V6H)

(6)

A similar binding model can be derived for a receptor containing two types of independent ligand-binding sites, each with a characteristic value of n, K, and 6H. In this model, the cumulative heat is the sum of the heats of binding of ligand to each independent site: Q = Q1 + Q2 = (6H1[LB1] + 6H2[LB2]) V

(7)

Q = V [M] (n16H1K1[L] + n26H2K2[L]) 1 + K1[L] 1 + K2[L]

(8)

Although a closed-form solution can be derived that equates the total cumulative heat to the total ligand concentration, this derivation is much more complex than that for the single class of binding sites and is not described here. The reader should see ref. 15 for a complete mathematical solution. An additional theoretical description of ITC-binding isotherms was introduced by Wiseman (2) and is shown below as applied to an interaction of 1:1 stoichiometry, where 1/r = c = [M]Ka and XR = [LT]/[M]: dQ/d[LT] = V6H (1/2 + [1 - XR – r ]/2 ([XR + r + 1]2 – 4XR)1/2)

(9)

This equation is the derivative of Eq. 1 and relates the incremental change in heat/injection to the parameter, c, which describes the curvature obtained within the binding saturation region. A complete derivation is reported by Indyk and Fisher (17). Titrations should be set up so that the product of the

Cationic Vectors by ITC and DSC

323

receptor concentration in mol/L and the affinity constant (the c value) should be within the range of 10–1000. This will allow sufficient curvature to exist in the saturation region of the binding isotherm for an accurate determination of the affinity constant. One consequence of this is that a high-affinity interaction (large Ka) will require a low macromolecule concentration and either a large 6H or a sensitive calorimeter for the detection of binding heats. The suitability of ITC for any binding interaction is therefore dependent on the relative magnitudes of Ka and 6H. Using analyses of the slope of isotherms at the equivalence point, Faergeman et al. (6) have extracted binding constants with c values as high as 5000. They determined that the change in slope of the isotherm for c values between 1000 and 10,000 was still significant, but much less than the change from 100 to 1000. Because the binding constant is determined from the slope of the isotherm at the equivalence point, binding constants with high c values are expected to be determined with lower accuracy. When multiple binding sites exist on a macromolecule, there exists the possibility of interaction between sites, leading to either positive cooperativity (ligand binding to the first site increases the affinity of ligand binding to remaining sites) or negative cooperativity (ligand binding to the first site decreases the affinity of ligand binding to remaining sites). Binding models are available to describe such cooperative interactions between sites (18). These models fit curvature in the experimental isotherm to an interaction between multiple binding sites, each characterized by individual values of K and 6H, with the addition of a cooperativity parameter to reflect the degree of interaction between sites. Modern ITCs inevitably include software to perform a variety of different types of fits to various binding behavior. There is insufficient space to address the issue here, but caution is urged in the naive use of such fitting algorithms, as good fits can often be obtained to data from a variety of different models. Obviously, a good fit does not directly imply the correctness of the model. Additional data from other sources is necessary to this end. Furthermore, assumptions involved in each model must be carefully accessed.

2.3. Experiment Design A properly designed ITC experiment should show initial binding heats representing the complete binding of ligand to the receptor, a gradual decrease in heats as saturation is approached and an acceptable signal:noise ratio. As mentioned in Subheading 2.2., the critical parameter for the proper setup of an ITC experiment is the c value, which defines the curvature of the binding isotherm in the saturation region. This should ideally be in the range of 10–1000. A suitable test to illustrate the proper design of an ITC experiment is the well-characterized interaction of BaCl2 with 18-crown-6 ether, a reaction also frequently used to

324

Lobo et al.

calibrate cell volumes of ITC instruments. Typical data obtained from this experiment are displayed in Fig. 1. Thermograms (raw data) obtained from the actual titration and a blank titration of BaCl2 into water are shown in Fig. 1A. The blank titration serves to correct the actual titration for the heats of dilution of ligand and for any heat effects from the mechanical injection process. The integrated heats for the blank titration should be subtracted from the actual titration using the mean dilution heat of the ligand if it is constant with injection number, or on a injection-per-injection basis if the ligand dilution heat varies with injection number. A plot of the corrected integrated heats per injection vs total ligand concentration results in a binding isotherm to which appropriate binding models are fitted, and is shown in Fig. 1B, using the best fit for a single-site-binding model. From the values of the affinity constant and 6H, the interaction is seen to be weak (Ka = 5900 ± 200/[M]), but evolves a significant amount of heat/injection (6H = –31.42 ± 0.20 kJ/mol) (16). The gradual loss of binding heats in the saturation region is reflected in the c parameter of 59. For convenience, many ITC software programs include an experiment design module, which can simulate the binding isotherm that will be obtained when estimated parameters and experimental conditions are inputted, to reduce method development time. Unfortunately, as shown in Subheading 2.5., the direct measurement of an association constant is generally not obtainable by ITC for a typical DNA– cationic polymer interaction, rendering the c parameter incalculable. This problem primarily results from aggregation during complex formation during the ITC experiment. Because complex formation involves kinetic factors such as reactant concentration and complex formation rate, however, these factors can be varied by the titration program to investigate such effects on the binding isotherm. Generally, titrations should be performed at the lowest concentrations of polymer (e.g., lipid) and DNA (50–100 µg/mL) that will give an acceptable heat signal to minimize the aggregation phenomenon.

2.4. Interaction of Cationic Lipids and Plasmid DNA It is generally accepted that the interaction of cationic lipids with DNA occurs at least partially as a consequence of electrostatic interactions between positively charged lipid headgroups and the negatively charged DNA phosphate backbone (19). This process is thought to be primarily driven entropically by the release of bound counterions into solution. In addition to this ionic interaction, variation in lipid–lipid and DNA–DNA ionic repulsive forces, attractive lipid–lipid apolar interactions, hydration forces and other structural properties of the liposome and plasmid DNA can cause a timedependent rearrangement of the complex on a macromolecular scale to form a wide variety of supramolecular structures. These structures have been

Cationic Vectors by ITC and DSC

325

Fig. 1. ITC thermogram (A) and binding isotherm (B) from a titration of BaCl2 (0.1 M) into 18-Cr-6 ether (0.01 M) using 25 injections of 10 µL at 25.4°C. A blank titration of BaCl2 into purified water is shown below the 18-Cr-6 titration.

characterized as lipid-coated DNA strands with intact liposomes (spaghetti and meatballs) (9), an ordered lattice of DNA with lamellar lipid stacks (8) and an ordered inverse hexagonal lipid phase with lipid-coated DNA (11,12). The final resulting structures of cationic lipid–DNA complexes are also dependent on the chemical nature of the cationic and any secondary (helper) lipids present, their ratio and the charge on the complex (20).

326

Lobo et al.

The complexity of this interaction introduces numerous difficulties and limitations into the determination and interpretation of ITC thermodynamic binding parameters. Fortunately, the use of full reverse titrations (titration of DNA into the cationic lipid as opposed to the usual forward titration of lipid into DNA), which would normally be considered superfluous in equilibrium protein–ligand titrations can reveal differences in the thermodynamics of binding due to the alterations in complex structure when the lipid is in excess. Therefore, a thorough, interpretable calorimetric investigation should include the effects of concentration, rate of complex formation, ionic strength, temperature, and titration direction.

2.5. Results 2.5.1. Forward Titration A forward titration is arbitrarily defined here as the addition of a cationic lipid or other polymers (ligand) to the plasmid DNA (receptor). Figure 2 shows three representative titrations (thermograms) of the cationic lipid dioctadecyldimethyl ammonium bromide (DDAB) into plasmid DNA (~9 kb) at 25.0°C in 10 mM Tris buffer. Titrations shown in Fig. 2A,B were performed with a titration program of 25 injections of 10 µL each, with 5-min incubations between injections, although the titration in Fig. 2B utilized solutions at one-half the DNA and lipid concentrations that were used in Fig. 2A. Fig. 2C used DNA and lipid concentrations identical to those of Fig. 2A, but the titration program consisted of 40 injections of 5 µL, to reduce the rate of complex formation. As seen in the similarity between these three titrations, the effects of solution concentrations (within this range) and complex formation rate did not have a significant effect on the interaction. In all three titrations, the binding of lipid to DNA is endothermic and low in enthalpy (<2 kcal/mol), which is consistent with the involvement of ionic and/or apolar interactions (21). Negative heats are observed up to a charge ratio near unity after which the complexes precipitate from solution. This aggregation event is manifested as an increase in the slope of the baseline, which presumably results from an increase in the heat of mixing of the aggregated complexes in the cell. This aggregation process near charge neutrality has been confirmed with identical forward titrations performed with dynamic light scattering and phase analysis light scattering. These measurements show that at a 1:1 charge ratio, complexes attain a low (±20 mV) c potential, and rapidly and completely aggregate from solution (not illustrated). This aggregation is not a saturation event nor an equilibrium transition and therefore should not fit to any type of equilibrium binding site model. This is especially important, because the authors have found that this type of data does fit well to simple two-independent-site models. The lack of saturation with forward titrations implies that an affinity constant (and 6G)

Cationic Vectors by ITC and DSC

327

Fig. 2. ITC thermograms of titrations of DDAB liposomes into plasmid DNA at 25.0°C in 10 mM Tris, pH 7.4. (A) 3.7 mM DDAB into 0.31 mM DNA, using 25 injections of 10 µL (B) 1.9 mM DDAB into 0.16 mM DNA, using 25 injections of 10 µL (C) 3.7 mM DDAB into 0.31 mM DNA, using 40 injections of 5 µL. A blank titration of lipid into buffer is shown below each DNA titration.

327

328

Lobo et al.

cannot be directly calculated for the interaction based on the ITC data. If the titration manifests a variation in binding heats prior to aggregation, this could reflect multiple sites or cooperativity (see Subheading 2.2.), but additional data would be necessary to evaluate this possibility. Nevertheless, an enthalpy of binding is directly obtained from these measurements.

2.5.2. Reverse Titration Titrations can also be performed in the reverse direction by titrating the plasmid into a cationic lipid solution, which permits the thermodynamics of complex formation in the region of high (+/–) charge ratios to be examined. Reverse titrations are often used in receptor–ligand interaction studies to independently determine the binding enthalpy so that this parameter need not be floated in model fitting of forward titrations (22). This strategy assumes that binding enthalpy is independent of titration direction that is often the case in the absence of self-association or aggregation processes. The first few injections result in complete binding of ligand to the macromolecule from which the 6H of binding is determined. The authors have observed that a complete titration of plasmid DNA into cationic lipid solution produces dramatically different thermograms and slightly lower enthalpies (calculated from initial binding heats) than those seen with forward titrations. This may result from the formation of complexes with different structures and colloidal properties when either species is in excess. Reverse titrations of plasmid DNA into 1,2-dioleoyl 3-trimethyl ammoniumpropane (DOTAP) and DDAB are shown in Fig. 3A,B, respectively. The binding of DNA to lipid is still endothermic, but two immediate differences from the forward titrations are evident. The aggregation event is manifested as a small deflection in the baseline and is preceded by an increase in the heats of binding, especially evident for the DNA into DOTAP titration. This increase in binding heat is probably caused by gross aggregation and not some type of a cooperative binding event, since this effect coincides with a rise in baseline (mixing heat) and is abolished in the presence of salt. Again, one cannot extract a true binding constant from these titrations because aggregation will contribute to the heat signal. The (+/–) charge ratio of the complex at the onset of colloidal instability with reverse titrations is different than that observed with forward titrations for both lipids. Aggregation occurs at a (+/–) charge ratio near unity in the forward direction, but at 3.1 ± 0.2 in the reverse direction for DOTAP and 3.5 ± 0.1 for DDAB (23). This asymmetry of interaction is confirmed when these reverse titrations are conducted using dynamic light scattering and phase analysis light scattering to monitor changes in complex size and c-potential, respectively. Complexes show a gradual increase in size up to the point of aggre-

Cationic Vectors by ITC and DSC

329

Fig. 3. Reverse ITC titrations of 7.1 mM plasmid DNA into 1.1 mM DOTAP (A) and 6.3 mM plasmid DNA into 1.6 mM DDAB (B), using 40 injections of 2.4 µL in 10 mM Tris, pH 7.4, with no added NaCl. Blank titrations of DNA into buffer produced negligible dilution heats.

gation that occurs at low c potentials (±20 mV) and a (+/–) charge ratio between 2 and 4. Because the heat signal will be subject to contributions from the aggregation processes, the use of additional methods to monitor changes in size and c potential of the complexes under conditions similar or identical to the ITC titrations can help to distinguish between colloidal instability and multiple-site or cooperative binding events.

330

Lobo et al.

2.5.3. Ionic Strength The strength of ionic interactions and the magnitude of polyelectrolyte effects can be described by the salt dependence of the affinity constant as measured by the parameter SaKobs (dlogKobs/dlog a± where a±, is the activity of a univalent salt), as defined by Record et al. (24,25). Since binding constants are not readily obtainable from ITC titrations in complex systems such as gene delivery vehicles, it may seem at first that related ITC studies have little merit. The effects of added salt, however, can manifest themselves as a salt dependence of the initial binding heat and as an effect on the entire binding isotherm, whatever its complex origin. Such data may still prove to be informative of the interaction as a whole. One need also be aware of the potentially detrimental effects of high salt concentration on the physical stability of DNA and liposomes themselves. Such effects must not be confused with direct perturbations of complex stability. In accordance with the Derjaguin, Landau, Verwey, and Overbeek theory of colloid stability, a high enough increase in ionic strength will also reduce the repulsive forces that stabilize colloidal particles, resulting in an increase in complex aggregation (20). The instability of liposome preparations in high-ionic-strength media may lead to erroneous interpretation of results if complexes are formed from aggregated liposomes. This limit of salt concentration is generally high. For example, preparations of DOTAP liposomes are stable up to a salt concentration of 400 mM NaCl. Binding isotherms from forward titrations of DOTAP (3.72 mM) into DNA (0.31 mM) in the presence of increasing NaCl concentrations are shown in Fig. 4A. The addition of salt causes a progressive reduction in initial and total binding heat and an increase in the amount of lipid required to bring about complex aggregation. The point of aggregation is less reproducible with increasing ionic strength, possibly because of a lack of gross aggregation of the complexes as observed visually after samples were removed from the reaction cell. The addition of salt had no effect on the slope of the isotherm at the aggregation point, suggesting that binding is limited by aggregation itself and not by the saltdependence of the binding processes. When the ionic strength is varied during reverse titrations a similar effect is observed. Figure 4B shows reverse titrations of DNA (7.1 mM) into DOTAP (1.1 and 4.3 mM) performed in the presence of increasing concentrations of salt. An elimination of the rapid increase in binding heats prior to aggregation is immediately apparent in the presence of added salt reflecting the lack of gross aggregation. A decrease in the initial heats of binding is observed, as is dramatic increase in the (+/–) charge ratio at which colloidal instability initiates, from 3.1 (no added salt) to 6.7 (400 mM NaCl). As shown in Fig. 4, the addition of salt can reveal portions of the isotherm that are most sensitive to colloidal effects, but additional information about

Cationic Vectors by ITC and DSC

331

Fig. 4. (A) ITC titrations of DOTAP into plasmid DNA in the presence of increasing concentrations of added NaCl. Titrations in 0 and 150 mM added salt were performed with 25 injections of 10 µL 3.7 mM DOTAP into 0.31 mM plasmid DNA. Titrations in 250 and 400 mM added salt were performed with 50 injections of 5 µL 7.4 mM DOTAP into 0.31 mM plasmid DNA. (B) ITC reverse titrations of 7.1 mM plasmid DNA into 1.1 mM DOTAP (0 mM NaCl) and 4.3 mM DOTAP (250, 400 mM NaCl) using 40 injections of 2.4 µL in 10 mM Tris pH 7.4, in the presence of the indicated concentrations of NaCl. Titrations performed in increasing NaCl concentrations show a progressive decrease in the heats of binding and an increase in the charge ratio at which the onset of aggregation occurs.

331

332

Lobo et al.

complex size and c potential are needed to draw more definite conclusions about the asymmetry of the interaction under these solution conditions. These results are complementary to the findings of Eastman et al. (26), and further support the external model of complex assembly, in which salt is postulated to reduce the contact area between lipid and DNA.

2.5.4. Heat Capacity Since the only thermodynamic parameter that can be directly obtained from most DNA–lipid calorimetric titrations is the binding enthalpy, a determination of the temperature dependence of the binding enthalpy can provide additional thermodynamic information in the form of the change in heat capacity upon binding (6CP). A negative 6CP is considered characteristic of apolar (hydrophobic) interactions that are driven by the reduction of solventaccessible apolar surface area upon binding. The magnitude of 6CP has been used to estimate polar and nonpolar surface areas of binding for many protein interactions and has correlated well with structural data (4). The temperature range selected for the determination of 6CP must be chosen carefully, especially when the 6H of binding is low as seen with DNA–lipid titrations. A negative 6CP will reduce the endothermic heats of binding at higher temperatures, while increasing the heats at lower temperatures. Therefore, an experimental protocol (DNA and lipid concentration) should be chosen to maintain an adequate signal:noise ratio throughout the temperature range. Consideration should also be given to the possible presence of any lipid phase transition (e.g., gel to liquid crystalline) that may occur within the temperature range. In addition, the effect of temperature on the stability of the liposome preparation (at lower temperatures, lipids with a high melting temperature (TM) will be prone to aggregation), any possible change in lipid critical micellar concentration with temperature and temperature effects on DNA structure itself must all be evaluated. Regarding the latter, although supercoiled DNA does not melt until 80–100°C, melting of open-circular and linear DNA components will occur at much lower temperatures (e.g., 50–60°C) (see below). The presence of open-circular and linear DNA as impurities in plasmid DNA preparation may thus introduce anomalies at these temperatures.

2.5.5. Proton-Linked Binding If the binding interaction is coupled to the release or uptake of protons from the buffer media, the experimentally observed binding enthalpy will include a contribution from the enthalpy of ionization of the buffer. When titrations are performed in buffers of different ionization enthalpy, the slope of a plot of observed binding enthalpy vs buffer ionization enthalpy will be equivalent to the number of protons released or absorbed/mole of bound ligand (27). This

Cationic Vectors by ITC and DSC

333

phenomenon might be expected to manifest itself with titrations of cationic species containing ionizable amines, such as polyethylenimine (PEI) and dioleoyl-phosphatidylethanolamine, but not with cationic lipids containing a quaternary ammonium group (DOTAP, DDAB), which carry a permanent positive charge.

2.6. Aggregation The tendency of DNA–lipid complexes to aggregate is dependent on a number of factors, including lipid composition, concentration, ionic strength, complex size and charge ratio and temperature (20). As described in Subheading 2.5.2., aggregation processes can significantly contribute to the heat signal in an ITC experiment and prevent the observation of binding saturation between cationic lipid or other polymers and DNA. Indications of aggregation are evident in thermograms as pronounced tailing of peaks, as a shift in the slope of the baseline near a zero c potential and as an increase in binding heats followed by a complete loss of binding. These signs may be observed to varying degrees depending on the factors previously mentioned and the titration direction. Increasing the concentrations of DNA and lipid will usually initiate the aggregation process sooner, which will be observed as an earlier rise in the baseline slope. Titration protocols with a low number of injections will show the aggregation event to be completed between successive injections, resulting in a precipitous loss of binding heat. The aggregation event can be spread out over multiple injections by lowering the amount of lipid added/injection and/or increasing the number of injections, resulting in a more gradual decrease in binding heats. This does not, however, allow aggregation events to be treated as equilibrium saturation phenomenon and this region of the isotherm should not be fitted to a binding-site model. Unfortunately, as indicated in Subheading 2.5.1., such thermograms are often well fit by simple two-site models. Attempts should be made to minimize aggregation by performing titrations at the lowest possible concentration of reactants.

2.7. Modeling ITC-Based Binding Data Data points that include altered binding heats as a consequence of any aggregation process during a forward titration should not be included in the binding isotherm during any attempted fitting to models. This limitation excludes the use of a single independent-site binding model because heats/injection are forced to zero in the binding saturation region in this approach. If a titration displays variations in binding heats during the injection process that are not caused by aggregation or some other nonbinding process, two options are available. First, a two (or more) independent binding-site model can fit the isotherm. Because this model can freely float up to six variables for two sites

334

Lobo et al.

(n, K, and 6H for each site), it will often provide an excellent fit to most isotherms. This has led to some skepticism of its applicability to protein–ligand interactions because nonintegral stoichiometries are sometimes obtained (17). It may, however, have some direct relevance to modeling the less-specific binding of cationic lipids and polymers to DNA, which may well possess nonintegral stoichiometries. The second option is to fit the binding isotherm to a cooperative model. As described in Subheading 2.2., affinities between sites and an enthalpy for each site can be obtained. Such models are not as easily fitted to ITC data and may require a large number of data points for a statistically acceptable fit. At the very least, a basic quantitative analysis of the binding enthalpy can be made for comparisons between titrations. A realistic value of 6H for the initial interaction (heat/mol ligand) can be obtained by dividing the average heat per injection for the first few injections by the number of moles of titrant per injection. This, of course, assumes that any aggregation process is minimal during the first few injections. Alternatively, a more overall estimate of 6H can be obtained by dividing the total heat under the isotherm before aggregation by the number of mol of ligand added until aggregation is detected. This assumes that all the ligand binds during each injection. Nevertheless, errors may still be introduced by this simple approach if any other heat-producing or -absorbing processes are occurring during titration. An additional method of characterizing tight cooperative binding processes has been described by Bains and Friere (28). This method is referred to as “total association at partial saturation” and is used to uncouple cooperative and intrinsic binding phenomena at high macromolecule concentrations. The use of a potentiometric method to determine the concentration of free ligand, has been used in tandem with ITC to supplement the binding analysis of the quaternary ammonium surfactants cetyl trimethyl ammonium bromide (CTAB) and dodecyl trimethyl ammonium bromide (DTAB) to DNA. This permitted the construction of binding isotherms for the evaluation of cooperative behavior (29). In general, if values of equilibrium-binding constants can be obtained by a noncalorimetric approach (e.g., by a spectroscopic or dialysis method), then the resultant value of 6G can be used to calculate 6S, permitting a complete thermodynamic description. This, however, requires the use of a van’t Hoff analysis with all of its attendant problems and assumptions somewhat obviating the elegance and rigor of the calorimetric approach.

2.8. ITC of Interaction of Heparin with CL–DNA Complexes One mechanism by which cationic lipid/DNA complexes (lipoplexes) enter cells is thought to be through initial attachment to cell surface proteoglycans and subsequent endocytosis (30). The authors have therefore used ITC to study

Cationic Vectors by ITC and DSC

335

the interaction of lipoplexes with heparin employing the latter as a model heparan sulfate proteoglycan. Although this data is subject to the same aggregation problems described in Subheading 2.6., the results support the presence of a charge– charge interaction between the negatively charged heparin and DOTAP:DNA complexes of increasing positive charge. Reverse titrations were performed with heparin (mol wt 4000–20,000) at 15.4 mM concentration added to DOTAP liposomes and complexes containing 1.5 mM DOTAP with 1.5, 0.75, and 0.375 mM DNA (charge ratios of 1:1, 2:1, 4:1 [+/–]). The titration thermograms are shown in Fig. 5, and are qualitatively similar to reverse titrations of DNA into DOTAP described in Subheading 2.5.2.. The contribution of aggregation to the heat signal was observed to varying extents depending on the lipoplex charge ratio. Binding isotherms are presented in Fig. 6, and show a progressive decrease in the amount of heparin bound as the charge ratio of the complex decreases. No interaction was observed between heparin and DNA alone. A fit to a two-independent-site model indicated that heparin binds to the lipoplex in proportion to the amount of excess lipid present in the complex, which is again consistent with charge–charge interactions playing a key role in the binding process. 3. Differential Scanning Calorimetry 3.1. Instrumentation Differential scanning calorimetry provides a direct measure of the heat evolved or absorbed during thermal transitions of polymers. Gene delivery vehicles are expected to possess such transitions because of the presence of both DNA and cationic polymers. Thus, DSC provides a means to evaluate the stability of complexes as well as perturbations in the stability of their component entities after complex formation. As described historically by Privalov (1), the new fourth generation of differential scanning calorimeters incorporate significant improvements in DSC design and technology. A representative example of one of these calorimeters is the CSC 5100 Nano DSC that is employed in the studies outlined below. The cells of this instrument are capillary in design rather than the cylindrical shape formerly more common. This provides roughly an order-of-magnitude increase in baseline reproducibility because of their constant-volume filling capability. The sample and reference cells (capillary channels) are cut into a gold block, and are maintained at the same temperature using Peltier elements. The sample cell is filled with a buffered solution of the polymer or complex of interest; the reference cell contains the buffer solution alone. The temperature can typically be repeatedly scanned between 0 and 125°C, both up and down at a constant rate and pressure. Temperatures above 100°C can be obtained because experiments can be performed at elevated pressures that increase the boiling point of the solvent. When a thermal transition occurs in the sample cell, a temperature differential is created between the sample and the reference.

336

336 Lobo et al.

Fig. 5. ITC titrations of heparin (15.4 mM) (mol wt 4–20 K) into 1.5 mM DOTAP liposomes (A), and DOTAP:DNA complexes containing 1.5 mM DOTAP at 1:1 (B), 2:1 (C), 4:1 (D) (+/–) charge ratios. Titrations were performed at 25.0°C in 10 mM Tris-HCl buffer, pH 7.4, using 50 injections of 5 µL with 5 min between injections. No detectable binding heats were observed with heparin and DNA alone.

Cationic Vectors by ITC and DSC

337

Fig. 6. Binding isotherms of ITC titrations of 15.4 mM heparin into DOTAP and DOTAP:DNA complexes. 䉬, = 1.5 mM DOTAP alone, 䊉, DOTAP:DNA (4:1) (1.5 mM DOTAP), 䉱, DOTAP:DNA (2:1) (1.5 mM DOTAP), 䊏, DOTAP:DNA (1:1) (1.5 mM DOTAP).

This temperature difference is sensed by the calorimeter and additional electrical energy is added to or removed from the sample cell by the powercompensation heaters. This compensation energy is directly proportional to the heat absorbed/evolved by the thermal transition and is displayed as the experimental heat capacity (which by definition is the temperature derivative of the enthalpy [heat flow]) as a function of increasing temperature. This experimental heat capacity is converted to an excess heat capacity (CPex) using the molecular weight and concentration of the analyte, the volume of the cell, and the specific volume of the target macromolecule or complex.

3.2. Theory Unlike linear DNA, a formal thermodynamic analysis of plasmid DNA is not possible because the unfolding of the supercoiled component is not reversible (no transitions are observed upon rescanning). The Lumry-Eyring model (31) simply but realistically describes the processes involved. As shown in Eq. 10, the supercoiled DNA (A) is in apparent equilibrium with structurally disrupted DNA (B), but is converted to a species (C), which cannot be converted back to the native state. Thus, true equilibrium does not exist. ACB AC

(10)

Therefore, a complete thermodynamic characterization cannot generally be performed unless some assumptions are made. The 6CP, however, can be

338

Lobo et al.

obtained from the difference in the pretransition and posttransition baselines in thermograms, regardless of the reversibility of the unfolding but will reflect the presence of various ill-defined states. A flat baseline is needed to obtain the absolute value of the heat capacity and this is usually not observed in these complex systems. The 6H can be obtained by integration of the area under the Cpex vs T curve, as shown in Eq. 11, if reversibility can be demonstrated. 6H = 0 CPex · dT

(11)

The transition entropy (6S) and free energy (6G) can then be obtained indirectly from the DSC data using Eqs. 12 and 13 and the Tm, the midpoint of the melting transition. 6S = 0 CPex/T · dT

(12)

6G = 6H – T6S = 6HTm [(Tm –T)/ Tm]–6CP(Tm–T)+6CPT ln(Tm/T)

(13)

To reiterate, the application of these thermodynamic expressions to experimental DSC data requires that transitions of interest be fully reversible and occur under equilibrium conditions. Because supercoiled DNA undergoes irreversible unfolding and nonnative association reactions, equilibrium conditions do not exist during the DSC measurement. A major factor in the potential use of thermodynamic parameters to characterize transitions is the rate of formation of the species responsible for the irreversibility. Thus, attempts have been made to use variation in scanning rates to extract true thermodynamic information (32). Distinctive evidence for a two-state irreversible transition includes asymmetry at the front of the melting transition (33). The advantages of DSC compared to van’t Hoff analysis for obtaining thermodynamic information include a model-independent determination of 6H and 6S and a direct measurement of the 6CP (34). The van’t Hoff enthalpy (6HvH) can be determined calorimetrically, using the following equation, where Tm is the temperature at which 6CP is maximal and 6CPTm is the value of 6CP at Tm (3): 6HvH = 4RTm2 6CPTm/6Hcal

(14)

The ratio of 6Hcal:6HvH has commonly been used as an indication of the presence of intermediate states along the structure alteration pathway from the native to unfolded states. Such intermediates are considered to be present when values of 6Hcal/6HvH are significantly greater than unity. Conversely, 6Hcal/ 6HvH values equal to unity are taken as indicative of a simple, two-state transition. However, this approach may be overly simplistic and may not give an accurate representation of the number of intermediate states present during the transition, since 6HvH is calculated from only one point of the DSC scan (3). Even if thermodynamic parameters cannot be obtained from DSC data, thermal

Cationic Vectors by ITC and DSC

339

transitions can still be defined in terms of their T m values permitting an empirical characterization of thermal stability. It is this approach that will prove of most utility in the analysis of gene delivery vehicles as described below.

3.3. Methods A typical approach that can be followed to prepare DNA–polymer complexes for DSC analyses is outlined below. Unless otherwise indicated, these steps were followed in the collection of the DSC data described here. Although these steps are specific to the CSC 5100 Nano DSC instrument, the general procedure should be applicable to all DSC instruments.

3.3.1. Preparation of Acceptable DSC Buffers Buffers unsuitable for DSC measurements include those that show a strong temperature dependence to their protonic dissociation (i.e., a high temperature coefficient). This is usually quantified as 6pKa/°C. For example, Tris buffer has an unacceptably high value, –0.031 (35), which absolutely precludes its use in DSC. Acceptable buffers include phosphate, citrate, cacodylic acid, MES (2-[N-morpholino]ethanesulfonic acid) and MOPS (3-[N-morpholino]propanesulfonic acid). Buffer concentration should not be excessive, as the presence of metal impurities may have destabilizing effects on DNA (36). A large quantity of buffer (e.g., 10 L) should be prepared for the dialysis of polymer stock solutions and for use in further dilutions, baseline measurements and preparation of DNA–polymer complex solutions. This procedure will allow exact matching of the heat capacity contribution from the buffer throughout the experiments and baseline determinations, which is a necessity if absolute heat capacities are to be obtained. All solutions should be degassed for at least 10 min under vacuum before use to prevent the occurrence of noise from the decreased solubility of dissolved gases at higher temperatures.

3.3.2. Preparation of Gene Delivery Complexes for DSC Analysis The preparation of the DNA solution should include dialysis into the desired buffer with a recommended minimum of two buffer changes. The suitability of the dialysis material should be determined by a comparison of a DSC scan of the dialyzed DNA solution to that of the original DNA solution. The effect of various commercially available dialysis materials on the thermal stability of DNA is shown in Fig. 7. All of the evaluated dialysis materials were acceptable with the exception of the Dispozo-dialyzer from Fisher. From this data, Pierce cartridges were selected for DNA dialysis in the experiments described below. The concentration of DNA in the cationic vector–DNA complexes prepared for DSC analysis is based on the detection limit of the thermal transitions of DNA, which will be dependent on solution conditions and the sensitivity of the

340

Lobo et al.

Fig. 7. DSC thermograms of DNA in Tris-HCl buffer before dialysis (a), after dialysis using a Spectropor dialysis tube (b), after dialysis using a cartridge from Pierce (c) and after dialysis using a Dispozo-dialyzer from Fisher (d). Samples b–d were dialyzed into 2 mM phosphate buffer pH 7.4. The Disposodialyzer was considered unsuitable for use, since the melting transition of supercoiled DNA was lower than that seen from samples dialyzed with the Pierce and Spectropor units. Note that the large temperature coefficient of the TRIS buffer results in a significant change in pH in A, which is unacceptable for an interpretable DSC experiment.

instrument. Samples of cationic vector–DNA complexes should be prepared at a DNA concentration at least twice the limit of detection to discern any subtle changes in magnitude or shifts of the melting transitions. Based on this criteria, a DNA concentration of 0.5 mg/mL was selected for all the DSC analysis. DNA concentration can easily be determined after dialysis using UV absorption measurement at 260 nm and an extinction coefficient of 0.02 AU·mL/cm·µg (see Chapter 19 on circular dichroism for further elaboration of some of the considerations in DNA concentration determination). To increase the reproducibility of the physical characteristics of the cationic lipid–DNA complexes, cationic lipids were prepared as uniform, single unilamellar vesicle liposomes (SUVs) by a film-extrusion method. An appropriate volume of lipid solution in organic solvent was added to a small vial and the solvent removed with nitrogen gas. Traces of organic solvent were removed from the dried lipids using vacuum desiccation for 3–6 h. After addition of 2 mM sodium phosphate buffer, pH 7.4 and brief vortexing, the lipid film was hydrated for 30 min to form multilamellar vesicles. SUVs were then produced by repeated extrusion of multilamellar vesicles through a 0.1-µm Nucleopore® polycarbonate membrane using an Avanti miniextruder. Liposome hydration and extrusion steps should generally be performed above the

Cationic Vectors by ITC and DSC

341

gel-to-liquid crystalline-phase-transition temperature of the lipid to assure that SUV formation occurs when the lipid is in the liquid crystalline state, in which mobility within the bilayer is unrestricted. Furthermore, the authors have detected a significant effect of liposome lamellarity upon the thermal transitions of DNA and lipid in complexes as monitored by DSC (not shown). Sonication of liposomes is an alternative method of size reduction, although the reproducibility of DSC measurements was greater with liposomes formed by the extrusion method. If cationic polymer–DNA complexes are to be made for DSC analysis, the pH of solutions containing polymers such as PEI and poly-L-lysine (PLL), should be adjusted prior to complexation with DNA, because these polymers contain basic groups that will increase the solution pH upon preparation. Complexation was performed by DNA addition to liposomes or polymers in equal volumes followed by equilibration for 1 h and <30 min, respectively, before loading into the DSC cells. A review of the various techniques of complex preparation and the dramatic effects of the preparation technique on their physical properties has been described by Lasic (20) and this source is suggested to readers as an excellent guide to improvement of the reproducibility of complexes for DSC analysis.

3.4. Experimental Aspects of DSC Independent of sample cell configuration, the presence of air bubbles in a sample will seriously compromise the quality of the results and every attempt should be made to avoid their introduction into the cell or to purge them. A large-volume pipetter (between 1 and 2 mL) can be used to fill the cells and the final volume of solution should be large enough to completely fill the cell. For the CSC 5100 Nano DSC, the nominal cell volume of 0.9 mL required at least 1.2 mL final sample volume. After sample loading, the cells should be allowed to reach thermal equilibrium before pressurization, which can be achieved in approx 20 min, depending on the temperature of the solutions. If the output power reading changes significantly while the cells are being pressurized, the cells must be refilled because air bubbles have formed. The low and high temperature limits should be set so that all transitions of interest may be observed. The scan range of the DSC experiments in this work correspond to the minimum and maximum temperature limits of the CSC 5100 Nano DSC (0–125°C). A scan rate of 1°C/min is commonly employed and was utilized for all data presented. Alternative scan rates can be used to observe kinetic effects (32,37,38), and to minimize experimental time. A single upward scan of solutions containing supercoiled DNA will generally suffice because the unfolding of supercoiled DNA is not reversible, but this needs to be confirmed. At least two sample scans were collected in this work and are

342

Lobo et al.

usually sufficient provided the scans show good reproducibility. Ideally, a baseline should be obtained immediately before each sample scan, although baseline scans obtained within 2 d of the sample scan were found to be acceptable when instrument cleanliness was maintained in the interim. The calorimeter cells normally require extensive flushing with large volumes of water after the completion of each experiment. The thermally induced residuals from cationic polymer samples, especially with excess polymer present, will require more intense cleaning procedures. Additional flushing with hot water, 1 molar potassium hydroxide, and soap solutions is usually sufficient. The effectiveness, or lack thereof, of these procedures will be apparent in the next baseline scan. Data analysis consists of subtraction of the baseline from the sample scan and conversion of the data to molar heat capacity, using the molecular weight and concentration of the DNA, its partial specific volume, and cell volume. Ideally, the partial specific volume of the complex itself should be used. This can be calculated from the weighted composition of the complex and the partial specific volumes of the DNA and polymer components. This procedure is automated in most instruments’ software.

3.5. DSC Results As mentioned in Subheading 3.2., the irreversibility of the thermal denaturation of supercoiled DNA limits the interpretation of the experimental DSC data to an analysis of the thermal stability of DNA and the cationic vector as a function of various solution and composition variables. Furthermore, because baselines are frequently unstable (possibly reflecting the presence of aggregation processes), only the excess heat capacities can be productively compared. The melting transition(s) of supercoiled DNA (as well as those of linear and open-circular forms of DNA, if present in the DNA preparation) can be monitored in DSC scans of the lipoplexes, as well as any gel-to-liquid crystalline-phase transition of the lipid if present within the temperature range of the DSC measurements. These transitions are apparent in the data shown in Fig. 8, which illustrates the melting transitions of DNA alone and in complexes with the cationic lipid DDAB, up to a 0.8 (+/–) charge ratio. In the absence of lipid, the linear and open-circle forms of DNA (which are present as minor impurities) melt between 60 and 70°C in a series of discrete, well defined transitions while the supercoiled DNA component has a single broad transition above 80°C (39). The gel-to-liquid crystalline transition of DDAB in the complexes is present at 50°C, which is significantly higher than the transition temperature of DDAB alone (41 ± 1°C) (40). A progressive increase in the magnitude of the DDAB phase transition is apparent with increasing (+/–) charge ratio of the complexes, although the temperature at which this transition occurs is relatively invariant.

Cationic Vectors by ITC and DSC

343

Fig. 8. DSC thermograms of DNA alone (a) and DDAB–DNA complexes at (+/–) ratios of 0.08 (b), 0.2 (c), 0.4 (d), 0.6 (e), and 0.8 (f). All solutions contained DNA at 0.5 mg/mL in 2 mM phosphate buffer, pH 7.4, and were scanned at 1°C/min.

Also evident with increasing (+/–) charge ratio is a corresponding increase in the Tm of the supercoiled component of the DNA, which implies a stabilizing effect of the lipid on the supercoiled DNA. A reduction in magnitude of the minor linear and open-circle transitions of DNA is also seen, but the pattern of these transitions is little affected by the presence of lipid. The thermal stability of DNA alone and complexed to increasing amounts of the cationic polymer, PEI (25 kDa), as characterized by DSC is presented in Fig. 9. For display purposes, the thermograms are spaced equidistantly along the y axis with increasing N:P ratio from bottom to top. Thus, the vertical position of each scan is not representative of the absolute heat capacity of that sample. The plasmid used to form these complexes had a much lower fraction of linear and open-circular DNA than the plasmid used to produce the DDAB complexes, resulting in a much-decreased appearance of the linear and opencircular DNA transitions compared to the cationic lipid results (Fig. 8). Two regions of stability (above and below an N:P ratio of 3) can be discerned. Below an N:P ratio of 3 (where the complexes retain an excess of negative charge), the linear and open-circle transitions are shifted to higher temperature and the supercoiled transition is shifted to lower values, which suggests an initial destabilizing effect of the polymer on the supercoiled DNA. At an N:P ratio of 3 (close to charge neutrality), an additional peak is observed concomitant with sample aggregation. At N/P ratios of 4 and higher (when the DNA is fully complexed to the polymer), all DNA transitions are shifted to higher temperature and remain essentially

344

Lobo et al.

Fig. 9. DSC thermograms of DNA alone (A) and PEI–DNA complexes at N:P ratios of 0.5 (B), 1 (C), 2 (D), 3 (E), 4 (F), 5 (G), 6 (H), 7 (I), 8 (J), 9(K), and 10(L). The molecular weight of the PEI used is 25 kDa. All solutions contained 0.5 mg/mL DNA in 2 mM phosphate buffer, pH 7.4, and were scanned at 1°C/min.

constant up to a N:P ratio of 10, which was the highest value examined. Thus, with an excess of PEI present in the complex, a significant and consistent thermal-stabilizing effect is induced by the polymer on the DNA. From these results, the interaction of DNA with PEI can be seen as qualitatively different from that with DDAB. A comparison of thermograms of PEI–DNA complexes, at N:P ratios below 3 (negatively charged), to the DDAB–DNA calorimetric data (all below charge neutrality) reveals that PEI destabilizes the DNA, but that DDAB induces the opposite effect. Only at N:P ratios of 4 and higher (when the DNA is fully complexed) is the thermal stabilizing effect of PEI apparent. A preliminary DSC analysis of the interaction between plasmid DNA and another cationic polymer often used for transfection, poly-L-lysine (29.3 kDa), is shown in Fig. 10, and demonstrates qualitatively unique thermal properties of this type of complex. DSC scans were performed of PLL–DNA complexes at (+/–) ratios from 0.2 to 4. The thermograms are presented with increasing (+/–) ratios from top to bottom along the y-axis. There is an apparent thermal destabilization of the supercoiled DNA seen at all charge ratios of this polymer. The linear and open-circle forms of DNA, however, are stabilized at (+/–) charge ratios below 1:1 and are not detectable in complexes above this charge ratio. Aggregation as determined by dynamic light scattering (data not shown)

Cationic Vectors by ITC and DSC

345

Fig. 10. DSC thermograms of DNA alone (a) and poly-L-lysine /DNA complexes at (+/–) ratios of (b) 0.2, (c) 0.4, (d) 1.6, and (e) 4. All solutions contained 0.5 mg/mL DNA in 2 mM phosphate buffer, pH 7.4, and were scanned at 1°C/min.

occurred at ratios of 1.6 and 4. Note that the supercoiled component manifests at least two transitions in the presence of the complexed polylysine. Some evidence for multiple components is also seen in PEI–DNA complexes (Fig. 9). The origin of these effects remain to be elucidated, but clearly demonstrate the potential utility of DSC in the characterization of DNA–polymer complex stability. 4. Basic Limitations of DSC and ITC for Analysis of Gene Delivery Vehicles The experimental data presented here employing ITC and DSC for the analysis of gene delivery vehicles reflect the formidable restrictions imposed on these techniques by the colloidal properties of these vectors, specifically their tendency to aggregate under certain conditions of ionic strength and at certain charge ratios of DNA:cationic polymer. Aggregation of complexes during ITC experiments resulted in alteration of binding and an inability to determine the affinity of the interaction. Similarly, with DSC, the induction of aggregation during the scanning process prevents any type of rigorous thermodynamic analysis. The supplementation of DSC and ITC with measurements of c

346

Lobo et al.

potential and hydrodynamic size under equivalent or similar representative conditions is indispensable to alert the investigator to these potential problems. The simple fact is that gene delivery systems are highly complex macromolecular entities and as such, are not easily amenable to simple equilibrium thermodynamic approaches. Nevertheless, valuable information can be obtained from the application of ITC and DSC, as well as related techniques if these limitations are clearly understood. When used in conjunction with other experimental approaches, these thermodynamic approaches can add significantly to an overall picture of the structure and stability of nonviral delivery complexes. References 1. Privalov, G. K., Kavina, V., Freire, E., and Privalov, P. L. (1995) Precise scanning calorimeter for studying thermal properties of biological macromolecules in dilute solution. Anal. Biochem. 232, 79–85. 2. Wiseman, T., Williston, S., Brandts, J. F., and Lin, L.-N. (1989) Rapid measurement of binding constants and heats of binding using a new titration calorimeter. Anal. Biochem. 179, 131–137. 3. Freire, E. (1994) Statistical thermodynamic analysis of differential scanning calorimetry data: structural deconvolution of heat capacity function of proteins. Methods Enzymol. 240, 502–530. 4. Spolar, R. S. and Record, M. T. (1994) Coupling of local folding to site-specific binding of proteins to DNA. Science 263, 777–784. 5. Murphy, K. P., Freire, E., and Paterson, Y. (1995) Configurational effects in antibody-antigen interactions studied by microcalorimetry. Proteins Struct. Funct. Genet. 21, 83–90. 6. Faergeman, N. J., Sigurskjold, B. W., Kragelund, B. B., Andersen, K. V., and Knudsen. J. (1996) Thermodynamics of ligand binding to acyl-coenzyme A binding protein studied by titration calorimetry. Biochemistry 35, 14,118–14,126. 7. Zuidam, N. J. and Barenholtz, Y. (1998) Electrostatic and structural properties of complexes involving plasmid DNA and cationic lipids commonly used for gene delivery. Biochim. Biophys. Acta 1368, 115–128. 8. Radler, J. O., Koltover, I., Salditt, T., and Safinya, C. R. (1997) Structure of DNA– cationic liposome complexes: DNA intercalation in multilamellar membranes in distinct interhelical packing regimes. Science 275, 810–814. 9. Sternberg, B., Sorgi, F. L., and Huang, L. (1994) New structures in complex formation between DNA and cationic liposomes visualized by freeze-fracture electron microscopy. FEBS Lett. 356, 361–366. 10. Gustafsson, J., Arvidson, G., Karlsson, G., and Almgren, M. (1995) Complexes between cationic liposomes and DNA visualized by cryo-TEM. Biochim. Biophys. Acta 1235, 305–312. 11. Mel’nikova, Y. S., Mel’nikov, S. M., and Lofroth, J-E. (1999) Physico-chemical aspects of the interaction between DNA and oppositely charged mixed liposomes. Biochim. Biophys. Acta 81, 125–141.

Cationic Vectors by ITC and DSC

347

12. Koltover, I., Salditt, T., Radler, J. O., and Safinya, C. R. (1998) An inverted hexagonal phase of cationic liposome-DNA complexes related to DNA release and delivery. Science 281, 78–81. 13. Fisher, H. and Singh, N. (1995) Calorimetric methods for interpreting proteinligand interactions. Methods Enzymol. 259, 194–221. 14. Freire, E. (1995) Differential scanning calorimetry, in Methods in Molecular Biology, vol. 40: Protein Stabilty and Folding: Theory and Practice (Shirley, B. A., ed.) Humana, Totowa, NJ, pp. 191–218. 15. Freire, E., Mayorga, O. L., and Straume M. (1990) Isothermal titration calorimetry: direct thermodynamic characterization of biological molecular interactions. Anal. Chem. 62, 950A–959A. 16. CSC 4200 Isothermal Titration Calorimeter User’s Manual. (1999) Revision 2.0, Calorimetry Sciences. 17. Indyk, L. and Fisher, H. (1998) Theoretical aspects of isothermal titration calorimetry. Methods Enzymol. 295, 350–364. 18. Wyman, J. and Gill, S. J. (1990) Binding and Linkage: The Functional Chemistry of Biological Macromolecules. University Science Books, Mill Valley, CA. 19. Zuidam, N. J., Hirsch-Lerner, D., Margulies, S., and Barenholtz, Y. (1999) Lamellarity of cationic liposomes and mode of preparation of lipoplexes affect transfection efficiency. Biochim. Biophys. Acta 1419, 207–220. 20. Lasic, D. D. (1997) Liposomes in Gene Delivery. CRC Press LLC, Boca Raton, FL. 21. Ross P. D. and Subramanian S. (1981) Thermodynamics of protein association reactions: forces contributing to stability. Biochemistry 20, 3096–3102. 22. Seelig J. (1997) Titration calorimetry of lipid-peptide interactions. Biochim. Biophys. Acta 1331, 103–116. 23. Lobo, B. A., Davis, A., Koe, G., Smith, J. G., and Middaugh, C. R. (2001) Isothermal titration calorimetric analysis of the interaction between cationic lipids and plasmid DNA. Arch. Biochem. Biophys. 386, 95–105. 24. Lohman, T. M. and Record, M.T. (1992) Thermodynamics of ligand-nucleic acid interactions. Methods Enzymol. 212, 400–424. 25. Mascotti, D. P. and Lohman, T. M. (1990) Thermodynamic extent of counterion release upon binding oligolysines to single-stranded nucleic acids. Proc. Natl. Acad. Sci. USA 87, 3142–3146. 26. Eastman, S. J., Siegel, C., Tousignant, J., Smith, A. E., Cheng, S. H., and Scheule, R. K. (1997) Biophysical characterization of cationic lipid:DNA complexes. Biochim. Biophys. Acta 1325, 41–62. 27. Wintrode, P. L. and Privalov, P. L. (1997) Energetics of target peptide recognition by calmodulin: a calorimetric study. J. Mol. Biol. 266, 1050–1062. 28. Bains, G. and Freire, E. (1991) Calorimetric determination of cooperative interactions in high affinity binding processes. Anal. Chem. 192, 203–206. 29. Spink, C. H. and Chaires, J. B. (1997) Thermodynamics of the binding of a cationic lipid to DNA. J. Am. Chem. Soc. 119, 10,920–10,928. 30. Mislick, K. A. and Baldeschwieler, J. D. (1996) Evidence for the role of proteoglycans in cation-mediated gene transfer. Proc. Natl. Acad. Sci. USA 93, 12,349–12,354.

348

Lobo et al.

31. Lumry, R. and Eyring, E. (1954) Conformational changes of proteins. J. Phys. Chem. 58, 110–120. 32. Galisteo, M. L., Conejero-Lara, F., Nunez, J., Sanchez-Ruiz, J. M., and Mateo, P. L. (1992) Calorimetric approach to the kinetics of the irreversible thermal denaturation of proteins. Thermochim. Acta 199, 147–157. 33. Carey, P. R. and Surewicz, W. K. (1996) Spectroscopic and Calorimetric methods for characterizing proteins and peptides, in Protein Engineering and Design (Carey, P. R., ed.), Academic, New York, p. 259. 34. Breslauer, K. J., Friere, E., and Straume, M. (1992) Calorimetry: a tool for DNA and ligand-DNA studies. Methods Enzymol. 211, 533–567. 35. Gueffroy, D. E., ed. (1978) Guide for the Preparation and Use of Buffers in Biological Systems. CalbioChem-Behring, La Jolla, CA. 36. Luck, G. and Zimmer, C. (1972) Conformational Aspects and Reactivity of DNA: Effects of manganese and magnesium ions on interaction with DNA. Eur. J. Biochem. 29, 528–536. 37. Galisteo, M. L., Mateo, P. L., and Sanchez-Ruiz, J. M. (1991) Kinetic study on the irreversible thermal denaturation of yeast phosphoglycerate kinase. Biochemistry 30, 2061–2066. 38. Conejero-Lara, F., Mateo, P. L., Aviles, F. X., and Sanchez-Ruiz, J. M. (1991) Effect of Zn+2 on the thermal denaturation of carboxypeptidase B. Biochemistry 30, 2067–2072. 39. Thumm, W., Seidl, A., and Hinz, H-J. (1998) Energy-structure correlations of plasmid DNA in different topological forms. Nucl. Acid Res. 16, 11,737–11,757. 40. Bhattacharya, S. and Mandal, S. S. (1998) Evidence of interlipidic ion-pairing in anion-induced DNA release from cationic amphiphile-DNA complexes. Mechanistic implications in transfection. Biochemistry 37, 7764–7777.

Synthetic Gene Therapy Vectors

349

22 Light-Scattering Techniques for Characterization of Synthetic Gene Therapy Vectors Christopher M. Wiethoff and C. Russell Middaugh 1. Introduction The colloidal properties of delivery systems currently being developed for nonviral gene therapy are extremely important. The physical stability of these systems on the shelf, as well as in the biological milieu, is mostly based on their size and interfacial properties. The size and surface charge of these systems can also have dramatic effects on their biological activity (1–4). With these facts in mind, it is apparent that adequate characterization of these properties is necessary for the development of any synthetic gene delivery system. The size and morphology of nonviral gene delivery systems have been characterized by both microscopic (5–7) and light scattering techniques (2–4,7–13). Electron and various scanning probe microscopic procedures (Atomic Force Microscopy, Scanning Tunneling Microscopy, and so on) provide detailed information about the size, shape, and molecular organization of various cationic polymer or lipid complexes with DNA. One drawback to these techniques is that the complexes are not imaged in the solution state, but rather on synthetic surfaces. The adsorption of complexes onto surfaces may alter both their structure and aggregation properties. In contrast, the scattering of light by such complexes in solution can provide direct information about the complex size distribution and particle shape (e.g., sphere, rod, or coil), as well as the nature of the interaction of the various components with each other. Furthermore, information about the electrokinetic (charge) character of the complexes can be obtained by certain light scattering based methods. All of these measurements provide key information that can aid in the characterization, formulation and stabilization of gene delivery complexes. From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

349

350

Wiethoff and Middaugh

The purpose of this chapter is to provide a background in various lightscattering techniques that can be used for the characterization of synthetic gene delivery vehicles. Information regarding theory, sample preparation, instrumentation, as well as data manipulation and interpretation, is presented. Examples of typical experiments in the context of gene delivery complexes of plasmid DNA with various cationic polymers is also described. 2. Dynamic Light Scattering When a solution of macromolecules is irradiated with light, some of the light is scattered as the result of interactions of the incident beam with the electrons of the particle. The oscillating electric field of the incident beam induces oscillations in the electrons of the macromolecule, which, in turn, re-emit radiation as scattered light. Most of the radiation that arises from the oscillating dipole moment is not changed in energy. Since these particles are undergoing random Brownian motion, however, the frequency of the scattered light will be slightly Doppler-shifted. If the power spectrum of the scattered light is analyzed using a spectrum analyzer, then a broadening of the spectrum is observed. This broadening is related to the diffusion coefficient of the particles in the sample. Unfortunately, these frequency shifts are often difficult to accurately detect, even when using a laser light source, without the use of a technique such as optical mixing (14). The more common method for measuring this phenomenon is to make measurements of the scattered light intensity fluctuations in the time domain. These fluctuations in the intensity of scattered light occur on a time-scale similar to the motions of the particles. Monitoring time-dependent fluctuations in the scattering intensity is done by autocorrelation analysis (14), which is based on the fact that Brownian motion is random. Therefore, the intensity fluctuations observed over increased time intervals are different than those observed initially. Consequently, later intensities are not correlated with initial values. Conversely, at shorter time delays, the particle positions are similar and therefore correlated. As the time interval is increased, the correlation is decreased. The decay in this correlation depends on the particle size, with smaller particles showing a more rapid decay, because of their more rapid velocity. Thus, from measuring the intensity (I) at time, t (It), and, at successive delay times, o, (It + o) (typically, between 5 µs and several ms), an autocorrelation function (g2, ACF) can be obtained. g2(t,t + o) = It • It +o

(1)

If Eq. 1 is corrected by g2(0), and normalized to g2('), it is related to the normalized scattered electric field ACF, g1(o), by: g2 (t,t + o) = 1 + `(g1 (o)2)

(2)

Synthetic Gene Therapy Vectors

351

where ` is an instrumental constant that accounts for nonideal deviations (ideally, ` = 1). For a monodisperse population of particles, g1(o) takes the single exponential form, g1(o) = e_Ko

(3)

where,

K = q2 D

(4)

and K is the relaxation decay constant, q is the scattering vector equal to 4/nsin(e/2)/l, where n is the refractive index of the solvent, and e is the angle of measurement. D is the translational diffusion coefficient and the major quantity of interest. Alternatively, for a sample with a distribution of particles differing in size, the electric field ACF takes the multiexponential form: '

g1 (o) = 0 0 G (K)e–KodK

(5)

where G(K) is the K distribution function of all the particles in the sample. g1(o) is the Laplace transform of G(K); therefore, by performing an inverse Laplace transformation of the electric field ACF, the distribution of K, and hence D, can be obtained. If one assumes that the particles in solution are spherical, the hydrodynamic radius can be approximated by using the Stokes-Einstein equation: kBT D=

(6) 6/dRh

where the translational diffusion coefficient, D, is inversely related to the hydrodynamic radius, Rh, by Boltzmann’s constant, the absolute temperature, T, and the viscosity of the solution, d.

2.1. Analysis of Size Distribution and Polydispersity If the particles of interest are truly uniform in size and the exclusion of dust or other particulate contamination is optimal, then fitting the ACF to a single exponential can easily provide the diffusion coefficient and therefore the hydrodynamic radius of the particle of interest. However, in the case of macromolecular complexes, such as those being developed for nonviral gene delivery, a homogeneous population of particles is very unlikely. Thus, some type of physically meaningful extraction of the diffusion coefficient from the multiexponential ACF is necessary to describe the size distribution of these colloidal species. Unfortunately, there are, in principle, an infinite number of solutions of G(K) that will satisfy Eq. 5. To address this problem, numerous algorithms have been developed for obtaining the distribution of the relaxation decay constants, G(K), from the ACF. Stock and Ray (15) have evaluated

352

Wiethoff and Middaugh

several of these methods using both simulated and experimental results with differing degrees of noise introduced, and a careful perusal of their results is highly recommended. The method of cumulants first described by Koppel (16) is the simplest and most common strategy employed to obtain the diffusion coefficient from the ACF. Essentially, this technique fits the natural logarithm of the ACF to a MacLaurin series: (7) The cumulants, kn, can be expressed in terms of the moment about the main values, as in Eq. 8. (8)

Typically, only the first two moments are used to analyze the fit because the usual experimental noise does not permit the higher-order terms to be accurately determined. This quadratic fit to the natural log of the ACF yields both the average relaxation rate K, and hence mean diffusion coefficient, as well as the variance of the population, µ2/2, which is usually referred to as the “polydispersity index.” If the polydispersity index is less than 0.02, then the sample is usually considered to be monodisperse (17). Major heterogeneity is usually assumed for values greater than 0.2. The method of cumulants most accurately describes a unimodal population of particles, but has frequently been used to provide population average values of the diffusion coefficient and polydispersity for mixtures of particles. The resultant values can be considered to accurately reflect relative sizes and heterogeneity and are especially valuable when comparing one sample to another. Despite the loss of information when this method is employed with heterogeneous samples, it is still usually considered the method of choice because of potential introduction of artifacts employing the more elaborate procedures described next. Other methods used to obtain G(K) from the ACF involve more sophisticated approaches to solve the inverse of the integral in Eq. 5. One common method is to use a nonnegative constraint on G(K), while performing a linear least-squares fit to the natural log of the ACF. An example of this is the Fortran NNLS routine (18). In addition to nonnegative constraints, a constrained

Synthetic Gene Therapy Vectors

353

regularization procedure has been employed by Provencher (19,20) in an algorithm known as “CONTIN.” Versions of these two programs are commonly available in the software of most commercial instrumentation. Both NNLS and CONTIN have correction factors for dust, so that any signal corresponding to the presence of large particles in the presence of smaller ones (e.g., 5 µm vs 50 nm) is assigned to a dust term in an effort to provide more accurate information regarding the smaller population. Both of these algorithms typically provide histogram plots of distributions in terms of size and some quantity parameter. NNLS methods work well for the description of bimodal distributions with narrow, widely spaced peaks. Although NNLS can give accurate estimates of the average values and variance of the sample, the distribution often contains false peaks and can overcompensate for dust (15). CONTIN also provides accurate descriptions of the average values and variance of the sample, but the population distributions derived are frequently oversmoothed if the data has significant noise (15). The CONTIN method, however, seems less prone to the generation of artifactual peaks. In addition to these two common algorithms for describing the population distribution of the diffusion coefficient, several other methods are often employed. One approach that can be used to increase confidence in multimodal results is to prepare synthetic mixtures of particles that correspond to the experimental results obtained. If the test mixtures are similarly resolved, the plausibility of the results of the unknown sample is enhanced. Readers are referred to Stock and Ray (15) or Chu (14) for an expanded review of this subject. The experimentally determined and deconvoluted distribution functions of the diffusion coefficients are weighted by the intensity of scattered light. This fact should be considered when interpreting the results obtained by NNLS, CONTIN, or any other algorithm that provides the distribution function. Subpopulations within multimodal populations of particles will scatter light differently, depending on their absolute size and refractive index. Frequently, software packages allow the presentation of the distribution function weighted by various other means, such as relative volume or number of particles. To perform this type of analysis, assumptions about the real and imaginary refractive index of the particles must be made. In general, the imaginary refractive index is 0 if the particles do not absorb light at the wavelength used (e.g., 0 for DNA and most polycations for wavelengths above 350 nm). It is therefore recommended that in the absence of this information, the intensity-weighted distribution be reported. However, a value of the real part of the refractive index of the sample is necessary. Ideally, values can be measured directly with a differential refractometer. More commonly, however, they are calculated from literature values and in the case of complexes, are estimated by weighted sums of the individual components. In such cases, the presence of large particles will severely distort the results since almost all of the scattering will be from these entities.

354

Wiethoff and Middaugh

3. Static Light Scattering In conventional reviews of light scattering, it is usual to present the static type of light scattering experiment first. The authors invert the presentation because of the greater importance in terms of frequency of use of DLS for complexes such as gene delivery vehicles. When a solution of particles is irradiated with light, the light is scattered in every direction. The relationship between the intensity of scattered light at a given angle and intensity of the incident light for particles much smaller than the wavelength of light used is: (9) where (10) Here, Re is the Rayleigh ratio which is the excess light scattering of the particles above that scattered by the solvent, Ie, divided by the intensity of the incident beam, I0. The distance the detector is from the center of the scattering volume is indicated by the value, r, and the angle between the detector and the incident beam is e. K is a constant that contains the refractive index increment dn/dc, the refractive index of the solvent, n0, Avogadro’s number, Na, and the wavelength of the light source in a vacuum, h0. Note the inverse fourth-power dependence of the wavelength, which accounts for much of the color-scattering phenomena observed in everyday life (e.g., blue sky, red sunsets, and so on). C is the concentration of the sample (g/mL) and M is the weight average molecular weight. This ratio can be obtained using an instrument specifically designed for light scattering experiments, such as one equipped with a laser light source and a photomultiplier detector mounted on a goniometer, which permits the measurement of scattered light at various angles. Some modern instruments simply mount a number of fixed detectors at various angles using fiber optics, eliminating the need for mechanical movement. Alternatively, a standard, L-format spectrofluorometer can be used to monitor the intensity of scattered light at 90 degrees. Even more simply, the Rayleigh ratio is related to the turbidity, o, a value which can be easily obtained using any ultravioletvisible spectrophotometer, by the relation o = 16/R90/3, where R90 is the Rayleigh ratio at 90 degrees. Thus, static light scattering data can almost always be obtained in one form or another in most laboratories. Eq. 7 is valid for particles with radii less than one-twentieth the wavelength of light (e.g., radii < 25 nm for 500 nm light). Because this requirement is

Synthetic Gene Therapy Vectors

355

typically not met when dealing with complexes of polycations with DNA, the Rayleigh-Gans-Debye approximation is employed:

(11) Or, rearranged, (12) where A2 is the second virial coefficient and reflects nonideal contributions, such as solution interparticle interactions and P(e) is the shape-dependent Debye factor, which accounts for intraparticle interference of light scattered from different regions of the particle. The Debye factor is dependent on the radius of gyration, Rg, the angle at which the scattered light is detected, relative to the incident beam and the wavelength of light in the relation: (13)

The radius of gyration is a parameter of particular interest in the characterization of macromolecules. This quantity is the root mean-square distance for any segment of the particle from its center of mass. P(e) A 1 as e A 0 and Rg A 0. With this in mind, it can be seen from Eqs. 12 and 13 that for a given angle (e.g., 90 degrees) as the radius of gyration decreases the scattering intensity increases. This phenomena explains the increase in scattering intensity seen when DNA is collapsed by spermine or other polycations from a more open form into more condensed toroidal and rod-like structures (21). If the concentrations employed in a light scattering experiment are sufficiently dilute to the extent that interparticle interactions can be ignored, the second term of Eq. 12 can be ignored. By rearrangement of Eq. 12 the following is obtained:

(14)

356

Wiethoff and Middaugh

where A is a constant and all other parameters are as previously defined. Therefore, if the scattering intensity is measured over a range of angles, the radius of gyration can be obtained by plotting the inverse of the intensity vs sin2(e/2). Given that the

and the

dividing the slope by the intercept permits Rg to be calculated. If the assumption of negligible interparticle interactions is not made, then obtaining useful results is more involved. In this case, both the angle of observation and concentration, of scatterer must be varied. By extrapolating to zero angle and zero concentration as originally described by Zimm (22), the weight-average molecular weight as well as the Rg can be calculated. Furthermore, the second virial coefficient, A2, is also obtained in this analysis. The second virial coefficient provides a convenient measure of solution nonideality, especially in terms of excluded volume effects. In the case of complexes of polycations with DNA, in which kinetic factors could potentially produce different structures at different concentrations, further insight into the concentration dependence of complex formation, either by dynamic light scattering or some other size-sensitive technique, is necessary for useful interpretation to be undertaken. 4. Electrophoretic Light Scattering Although static and dynamic light scattering techniques can provide information regarding the size and shape of a dispersion of particles in solution, electrophoretic light scattering provides information about the electrokinetic properties of the particles. Of particular interest is the determination of a particle’s c potential. For a particle that carries a given charge on its surface, Fig. 1 describes the conceptual basis of the c potential. Assuming the surface of a particle has a positive charge, negatively charged counterions in solution will become associated with the particle surface in what is known as the Stern layer. Just beyond this Stern layer, additional counterions (primarily negatively charged ions in this case) are also present with reduced mobility. These two ionic strata define the electric double layer of the particle. The outermost boundary of the double layer is known as the shear radius and defines the electrokinetic unit of the particle. The c potential is the potential difference between the shear radius and the bulk solution. The c potential is often used as a measure of colloidal

Synthetic Gene Therapy Vectors

357

Fig. 1. Schematic of a surface potential for a positively charged surface in solution. The Stern layer contains counterions that are intimately associated with the surface (surface to boundary a). The boundary b is known as the shear plane of the electrokinetic unit. Together the two boundaries form the electric double layer of the surface. Region c represents the bulk media, which exhibits a random distribution of ions.

stability since the greater the c potential of a population of particles, the less likely they are to aggregate because of increased repulsive forces. Unfortunately, no technique exists to directly measure the c potential of a particle. It can, however, be estimated from the electrophoretic mobility of the particle, which is commonly obtained by some form of electrophoresis. The electrophoretic mobility is the velocity of a particle in a given electric field. For a spherical particle, the electrophoretic mobility, µ, is defined as: 2¡c f (grh,c)

µ=

(15) 3d

where ¡ is the permittivity of the solvent, c is the c potential, and d is the viscosity. f(grh, c) is a model-dependent function of grh, the product of the inverse of the shear radius, g, and the particle radius, rh, as well as the c potential. The most common model used for estimating the c potential is that of Smoluchowski (23). In the limit of the Smoluchowski approximation, the electrophoretic mobility takes the form: µ=

¡c (16) d

Thus, by measuring the velocity of particles in an electric field, the c potential can be calculated.

358

Wiethoff and Middaugh

Alternatively, for a particle of charge, q, in an applied electric field, the electrophoretic mobility can be expressed: ze µ=

(17) 6/dRh

where ze is the product of the valency and the electron charge and Rh is the hydrodynamic radius of the particle. This alternative expression of electrophoretic mobility illustrates its dependence on the hydrodynamic size as well as z potential. Conventional electrophoretic light scattering measurements are performed using laser Doppler electrophoresis (LDE), which monitors the velocity of particles in a manner similar to the measurement of the Brownian motion of particles in dynamic light scattering experiments. Again, when particles are irradiated with light of a certain frequency, scattered light is radiated at frequencies both slightly above and below that of the incident light, as a result of the random movement of the particles (i.e., it is Doppler-shifted). If an electric field is applied, the particles will move based on their relative surface charge density. Conventional LDE instruments work by mixing the light scattered from a dispersion of particles moving in an electric field with unshifted light from a reference beam (24). This mixing results in a beating frequency that is dependent on the velocity of the particles. An additional phase modulation can be applied to the reference beam, resulting in an additional shift in the frequency. This permits either a positive or negative shift to be detected based on the direction in which the particles are moving, since the two shifts are additive. Because the direction of the electric field is known, the sign as well as the magnitude of the particle velocity, and hence, the electrophoretic mobility, can be ascertained. To prevent polarization of the electrodes, the electric field is periodically inverted. In cases in which the particles have a relatively low electrophoretic mobility, the only option available to improve measurement accuracy is to either increase the electric field strength above the typical 10 V/cm, which leads to excessive Joule heating, or to decrease the frequency at which the polarity of the electrodes is reversed, which can lead to complete polarization of the electrodes. If the displacement of the particle in the electric field over a given time is less than the inverse of the scattering vector, q, then the signal will not complete a full cycle and can not be accurately measured by spectral analysis. Recently, a technique >1000× more sensitive than LDE, known as phase analysis light scattering, has become commercially available (24). This technique applies a phase modulation so that the Doppler frequency of the scattered light for a particle with zero mobility is equal to the modulation

Synthetic Gene Therapy Vectors

359

frequency. The deviation of this frequency of the scattered light from the modulation frequency for a given sample can be measured by a phase comparator. If a small mobility exists, the relative phase of the scattered light will be shifted and can be used to calculate the electrophoretic mobility and subsequently the c potential. This technique has increased accuracy, compared to LDE because phase comparison takes place over many cycles, but the spectral analysis used in LDE is sensitive to a period of only one cycle (24). 5. Sample Preparation The major problem associated with obtaining interpretable and reproducible results from light scattering measurements is the purity of the sample. The presence of dust or other large particulate contamination is the most common obstacle to making these measurements. For example, particles of >200 nm diameter are larger than most DNA plasmids and associated complexes. The contribution of even a small quantity of such larger particles to the intensity of scattered light will be greater than that of the smaller population of interest because it is the weight average molecular weight (<MW>), featuring a molecular weight-squared dependence to which scattering is proportional. It is therefore necessary to minimize irrelevant particulate components in samples. One particularly problematic contaminant sometimes encountered in DNA solutions is agarose, which may be present if the DNA is isolated by gel electrophoresis. If the DNA is purified from bacterial cell culture, care should also be taken to minimize contamination from RNA or bacterial polysaccharides. Langowski et al. (25) recommend the use of ion exchange HPLC to isolate supercoiled plasmid DNA for light scattering studies because of these types of problems. Purity of the plasmid to be used in light scattering studies is of obvious importance with respect to the presence of linear and open-circular topoisomers. The plasmid DNA used in the data presented in this chapter was >95% supercoiled. In general, whatever the topological form of the DNA, it should be as homogeneous as possible. Glassware should be exhaustively rinsed with distilled and deionized water that has been filtered through a 0.22-µm or smaller filter. Additionally, it may be required that the glassware be cleaned with chromic acid solutions to ensure adequate cleanliness. Regarding solution preparation, glassware with groundglass connectors such as volumetric flasks should be avoided to eliminate glass particulates. All buffers should be filtered to remove dust material. In addition to simply using a syringe-based filter system, filtration devices can be produced using peristaltic pumps to cycle buffer and samples repeatedly through a filter or series of filters. The use of these continuous filtration devices is strongly recommended for weakly scattering samples and in any situation in which reproducibility problems are encountered. If solutions containing DNA, complexing agent, or a complex of these molecules are to be filtered, however,

360

Wiethoff and Middaugh

verification of a lack of extensive filter binding of particles and an absence of structural changes induced by filtration should be performed, if possible, before exhaustive characterization of the filtered samples. Concentration can be an important parameter when making light scattering measurements. Depending on the size and shape of complexes, concentration is crucial to obtaining accurate estimates of the molecular size in static experiments. To measure the radius of gyration, the assumption of negligible interparticle interactions is made. Thus, dilute solutions are best employed (see Eqs. 9 and 12). The configuration of the instrument may dictate the range of concentrations applicable because of the power of the light source, the ability to attenuate the light source with neutral density filters and the ability to adjust the cross-sectional area of the light seen by the photomultiplier tube, relative to the coherence area by means of an aperture. In general, a useful DNA concentration range for initial evaluation is between 10 and 100 µg/mL. The DNA concentration may need to be higher if the DNA is to be evaluated in the absence of complexing agents. Ideally, samples should be prepared in a concentration range in which the intensity of scattered light is linearly dependent on the concentration, assuming the physical properties (e.g., size or shape) of the complex do not change over this concentration range. If a sample is too concentrated in DLS experiments, the results may suggest a hydrodynamic radius that is smaller than the true value. This artifact is caused by deviation from the assumption that a single scattering event occurs. In concentrated solutions, multiple scattering events (i.e., a given photon is scattered from one particle to another before it is detected) can occur, producing an apparent increase in the fluctuations in the intensity of scattered light with time. To minimize this problem, it has been suggested that the optical density (i.e., turbidity) for a 1-cm pathlength cell of <0.04 be employed at the wavelength used (26). 6. Instrumentation A variety of different types of instruments can be used to obtain interpretable light scattering data. For dynamic light scattering, instruments specifically designed for such measurements must be employed. Typically, these instruments are equipped with a laser light source (15–5000 mW), a photomultiplier tube located 90 degrees to the incident beam (which may by movable) and a digital correlator card in a computer, which analyzes the intensity fluctuations as dictated by a user interface. These instruments will usually accommodate a variety of sample cells, ranging from 0.5 to 20 mL. A recent report described the use of glass capillaries similar to those used for the determination of melting points as sample cells (~1.2 mm outer diameter) (27). With these tubes, the concentrations that can be studied are much higher than previously employed

Synthetic Gene Therapy Vectors

361

because the cross-sectional area of the tube is closer to the coherence area, minimizing secondary scattering events. The use of such cells in the study of polycation–DNA complexes has yet to be formally reported, but the authors have found them especially useful for the examination of complexes at higher concentrations. Static light scattering measurements may be made with a wider range of instrumentation. Although dedicated multiangle light scattering instrumentation is now readily available, UV/Vis spectrophotometers and spectrofluorometers can also be used to obtain light scattering information at fixed angles. Many instruments used in light scattering studies employ laser light sources, permitting use of a single wavelength of light, but conventional spectrometers and spectrofluorometers offer a choice of wavelengths from the UV to the near infrared. This permits one to take advantage of the fact that the intensity of scattered light is roughly inversely related to the fourth power of the wavelength of incident light, to obtain greater scattering signal at lower wavelengths (higher frequency). In choosing a wavelength to monitor light scattering, regions of electronic absorption (e.g., 260 nm for DNA) should be avoided. For most polycation–DNA complexes, light between 400 and 700 nm is suitable. 7. Applications 7.1. Plasmid Characterization The various light scattering approaches described above can be used to characterize the properties of nonviral gene delivery vehicles and their macromolecular components. These methods are useful in this application since it is clear that the size and surface charge of these particulate systems can influence their efficiency of gene delivery both in vitro (1,2) and in vivo (28,29). In general, superhelical plasmid DNA is employed for nonviral gene therapy applications. These molecules are relatively large, making efficient delivery into cells a difficult task. The colloidal properties of the DNA itself have been shown to strongly influence the overall properties of cationic lipid/DNA complexes (30). Therefore, adequate characterization of the plasmid alone should be performed for the design of a well-defined gene delivery system. Superhelical plasmids have been well characterized by both dynamic (25,31–34) and static (35,36) light scattering. Langowski et al. (25,34) have used dynamic light scattering to extensively characterize differences in plasmids based on molecular weight and superhelical density. Using dynamic light scattering, they have shown that several DNA relaxation events can be detected. These processes include the translational and end-over-end rotational diffusion events of interest, as well as internal motions of the plasmid. Each of these components contributes to the ACF in an angle-dependent fashion. At higher scattering angles, (e.g., 90 degrees for a 488-nm laser), the ACF is dominated by the

362

Wiethoff and Middaugh

Fig. 2. Variation of the apparent hydrodynamic diameter of a 4.7-kbp supercoiled plasmid with the square of the scattering vector. A solution containing 100 µg/mL DNA in 10 mM Tris buffer, pH 7.4, was used. The effective diameter was obtained from the ACF by the method of cumulants. The scattering vectors shown correspond to angles between 20 and 140 degrees.

internal motions of the plasmid. Based on the diffusion coefficients measured at these high angles, the internal motions correspond to a segment of DNA roughly the size of its persistence length (~50 nm) (32). At low scattering angles, the translational and rotational diffusion processes are the major contributors to the ACF. By making DLS measurements at low scattering angles, a strong correlation was found between the molecular weight of the plasmid and the translational diffusion coefficient (34). The translational diffusion coefficient also showed a linear dependence on the superhelical density for a given plasmid (25). In contrast, the rotational diffusion of the plasmid remained relatively constant with scattering angle if the ACF was fit to two exponentials (25). An example of DLS measurements of the apparent hydrodynamic size of a 4.7-kbp supercoiled plasmid as the scattering vector is varied is shown in Fig. 2. To obtain high quality data, the glassware and buffers were carefully prepared as described above. The final DNA solution (100 µg/mL DNA in 10 mM Tris pH 7.4) was slowly passed several times through a 0.22-µm polysulfone syringe filter using two syringes and a three-way stopcock to isolate the

Synthetic Gene Therapy Vectors

363

solution from the environment during filtration. The final pass was emptied into to 12 × 75-mm cylindrical sample cell and the cell was sealed. A Brookhaven instrument equipped with a 50-mW HeNe diode laser (532 nm), a BI-200SM goniometer, a Model EMI 9863 PMT with high-voltage power supply, and a BI-9000AT digital correlator was used to obtain these results. The ACF was obtained by collection and averaging of nine 20-s intervals. Any interval in which the results obtained were significantly different from the average was removed to minimize the influence of dust on the results. Additionally, an electronic dust filter was used which removed data at 2-s intervals if the average scattering intensity over that interval differed by 30% from the overall average intensity. The effective diameter was obtained using the method of cumulants. It can be seen from Fig. 2 that, for lower angles (q2 between 3 × 1013 and 3 × 1014 m–2), the hydrodynamic diameter remains constant. As the square of the scattering vector is increased above 3 × 1014, the apparent size decreases as reported by Langowski (32). This observation illustrates the necessity of adequately defining the measurement parameters when studying the interaction of DNA with various complexing agents. If the scattering vector is sufficiently high, the DNA itself appears smaller, because of the detection of internal motions rather than the desired translational diffusion of the plasmid. Thus, simply making a naive measurement of a polymer–DNA complex without comparing the results to free DNA could lead to the erroneous conclusion that the DNA is condensed when, in actuality, the complexing agent had little effect on the hydrodynamic properties of the plasmid. Similar problems can be encountered when the ACF is deconvoluted by certain algorithms, as discussed in Subheading 2.2. Static light scattering experiments have also been used to characterize supercoiled plasmids. Fishman and Patterson (36) have accurately determined the molecular weight of a 3.5 kb plasmid as 2.5 × 106 Daltons (compared to a calculated value of 2.45 × 106 Daltons). They have also determined the radius of gyration (Rg) and the second virial coefficient (A2) by using this technique. The radius of gyration obtained in this study (82 nm) was in agreement with predicted estimates based on the behavior of a worm-like ring. Additionally, the positive value of the second virial coefficient (1.2 × 10–3 mol mL/g2) is indicative of an extended polymer exhibiting strong interpolymer-repulsive forces.

7.2. Polycation Characterization In addition to characterizing the colloidal properties of the plasmid, similar methods can be used to characterize polycations, including micellar and liposomal structures. Regarding the use of cationic lipids as nonviral gene delivery vehicles, the vesicle morphology has been shown to influence transfection efficiency when complexed with DNA (29). Upon initial formation of the vesicles, typically by hydration of a solid powder or thin film of lipid with

364

Wiethoff and Middaugh

Fig. 3. The effect of extrusion on the size distribution of DOTAP liposomes. Dynamic light scattering was used to obtain the size distribution of DOTAP liposomes. Data presented were obtained using a NNLS algorithm. The distributions represented are intensity-weighted. Data for unextruded liposomes (solid black line) and extruded liposomes (broken black line) are shown.

buffer, a large, heterogeneous population of multilamellar vesicles (MLVs) is formed. This initial lipid dispersion is often treated in such a way as to reduce the mean size and population distribution. Sonication or extrusion through polycarbonate membranes are the most common methods used for the generation of small unilamellar vesicles (SUVs). Figure 3 shows a comparison of the size distribution of an extruded and unextruded lipid preparation containing 1,2-dioleoyl-3-trimethylammonium-propane (DOTAP). The results were obtained for vesicles before and after extrusion of the vesicles 10× through a 100-nm pore size polycarbonate membrane. Dynamic light scattering measurements were taken at 90 degrees as described above. The distributions shown in Fig. 3 were obtained using a NNLS algorithm. The distribution obtained for the unextruded vesicles clearly shows two distinct populations around 300 nm and above 1000 nm. The mean diameter corresponding to this sample population is 950 nm. Results obtained using the method of cumulants gives an average effective diameter of 540 nm and a polydispersity index of 0.305. In comparison, the extruded vesicles exhibit a narrow distribution centered around 130 nm, as determined by the NNLS algorithm. Using the method of cumulants, an effective diameter of 130 nm was also obtained. In agreement with the narrow distribution obtained by NNLS, the cumulant polydispersity index was 0.04, suggesting a relatively unimodal distribution of vesicles.

Synthetic Gene Therapy Vectors

365

Important differences, such as those observed in Fig. 3, have been shown to affect the transfection efficiency of cationic lipid/DNA complexes (28,29). These authors found enhanced transfection efficiency when the DNA was complexed with MLVs vs SUVs prepared by sonication for a formulation containing the cationic lipid DOTIM. The average size determined by dynamic light scattering was 150 nm for multilamellar vesicle complexes vs 60 nm for SUV complexes (28). Additionally, when vesicles containing DOTAP and cholesterol were prepared using a series of extrusion steps at different temperatures, invaginated vesicles with large surface areas were formed. When complexed to DNA, a population of complexes between 200 and 450 nm were found to be responsible for increased systemic delivery and gene expression in mice (28).

7.3. Characterization of Polycation–DNA Complexes 7.3.1. Cationic Lipid/DNA Complexes Of greatest interest for gene therapy are light scattering studies of complexes of DNA with various polycations. To characterize the interaction of DNA with the cationic lipid, DOTAP, complexes were prepared at various ratios of cationic lipid to DNA phosphate (positive charge:negative charge, since DOTAP has one positive charge/molecule). A solution of 30 µg/mL DNA (9.1 kbp supercoiled plasmid) was prepared by diluting the DNA from a stock solution into 10 mM Tris buffer, pH 7.4 that had been exhaustively filtered through a 0.22-µm polysulfone syringe filter. Appropriate dilutions of the DOTAP stock solution into the filtered buffer were performed, then, depending on the ratio of lipid:DNA phosphate, the lipid or DNA solution of lesser amount was rapidly added to an equal volume of the other solution containing the greater amount with gentle mixing. This mixing was performed in the sample cells to minimize the introduction of dust. The samples, now containing 15 µg/mL DNA each, were capped and allowed to equilibrate for 30 min before measurements were initiated. This method did not introduce a significant amount of micronized air bubbles, as determined by performing the mixing with buffer alone. Dynamic and static light scattering measurements were performed simultaneously, using the same instrumental conditions described in Subheading 7.1. For all measurements, scattering at 30, 60, and 90 degrees was monitored. In addition, LDE measurements were performed on these same samples after the dynamic and static measurements. The results of these studies are summarized in Figs. 4–6. Figure 4A describes the hydrodynamic size of the complexes as the charge ratio (lipid:DNA phosphate) is increased. These results were obtained from the ACF by cumulant analysis. At (+/–) charge ratios <1, the size shows a gradual increase with increasing charge ratio when monitored at angles above

366

Wiethoff and Middaugh

Fig. 4. The effect of charge ratio on the hydrodynamic size distribution of DOTAP– DNA complexes. (A) Results show the hydrodynamic diameter of the complexes obtained by dynamic light scattering at various angles as a function of the ratio of cationic lipid to DNA phosphate. The diameter was obtained from the ACF by the method of cumulants. (B) The polydispersity index as a function of charge ratio was obtained using the method of cumulants. The DNA concentration was 15 µg/mL in all cases.

60 degrees. When monitored at 30 degrees, the size shows a minimum between a charge ratio of 0.4 and 0.6. The overall larger effective diameters observed at lower angles suggest the presence of a population of larger aggregates that do not contribute as much to the scattering intensity at higher angles. This is

Synthetic Gene Therapy Vectors

367

Fig. 5. The effect of charge ratio on the intensity of scattered light at various angles for DOTAP–DNA complexes. The intensity of scattered light was monitored at 30 degrees (closed triangles), 60 degrees (open squares), and 90 degrees (closed circles), as a function of charge ratio. Samples were those used for Fig. 4.

Fig. 6. The effect of charge ratio on the c potential of DOTAP–DNA complexes. Using the same samples prepared for DLS and SLS studies (Figs. 4 and 5), the c potential was monitored as a function of charge ratio by LDE.

368

Wiethoff and Middaugh

confirmed in Fig. 4B when the polydispersity index in this same range of charge ratios is examined. It appears that as a 1:1 (+/–) ratio is approached, the polydispersity decreases at all angles examined. In the region between charge ratios of 1 to 2, massive aggregation occurs with the inset data in Fig. 4A showing extensive aggregation for all angles used. Note that values of the diameter >1 µm are not representative of the actual size since the data analysis methods are not accurate at these large values. The polydispersity index (Fig. 4B) only markedly reflects this aggregation at angles of 60 degrees and below, again suggesting the coexistence of several populations of aggregates. This gross aggregation occurs around the calculated charge neutrality of the system. This phenomena is expected in light of the analysis developed by Derjaguin, Landau, Verwey, and Overbeek (DLVO theory) (37), which describes the interplay between electrostatic repulsive forces and van der Waals attractive forces in colloidal stability. Assuming the attractive forces stay constant, if the electrostatic repulsive forces are minimized (e.g., by neutralization of charges on the surface of the particle), then the balance shifts in favor of net attraction and aggregation results. Above a charge ratio of 2, the complexes are again reduced in size to ~150 nm in diameter, with the size again appearing smaller at angles above 60 degrees. The polydispersity of the complexes in this region appears slightly less than in the other two, presumably reflecting the presence of a more homogeneous population of complexes with an excess positive surface charge. In the absence of equipment capable of making dynamic measurements, static light scattering can provide equally valuable information regarding the colloidal properties of complexes. Figure 5 shows the intensity of scattered light monitored at the same three angles. While this data was collected using an instrument dedicated to light scattering measurements, similar results could be obtained using a spectrofluorometer or UV/Vis spectrophotometer, although such data would generally be limited to a single angle. Fluorescence instruments can, however, often be operated using a front-surface geometry, potentially permitting light scattering measurements of at least one additional angle (e.g., 40–60 degrees). Recall from Eqs. 9 and 10 that the scattering intensity is directly proportional to both the molecular weight of the colloid and its concentration. Because the concentration only increases by a factor of four in this experiment, the majority of the change seen with increasing charge ratio is related to changes in the molecular weight of the complex (assuming all other parameters comprising K of Eq. 10 to be constant). This can be verified by normalizing the intensity to the concentration and constructing a plot similar to that shown in the inset of Fig. 5. This plot supports the conclusion that the observed changes in scattering intensity result from changes in the molecular weight (since the effect of concentration is removed by normalization).

Synthetic Gene Therapy Vectors

369

Thus, as the ratio of DOTAP:DNA is increased, a gradual increase in the scattering intensity is observed at angles above 60 degrees, resulting in a maximum charge ratio between 1 and 2. This observation corresponds to the increase in hydrodynamic size observed by dynamic light scattering. At 30 degrees, the initial intensity increase is less linear than at the greater angles, but again reaches a maximum charge ratio between 1 and 2. The nonlinear increase presumably reflects the greater contribution to the scattering of larger particles at low angles. The intensity decreases when DOTAP is in charge excess, eventually reaching a plateau reflecting more-fixed particle sizes. Overall, the intensity of scattered light correlates well with the hydrodynamic size of the complexes. Different trends observed for measurements at the three different angles presumably reflect differences in the Debye factor, P(e), suggesting differing morphologies of the complexes at different charge ratios. Although not entirely quantitative, this type of study does provide useful information regarding the relative size of the complexes formed at differing ratios of lipid:DNA. It also provides information about the nature of the interaction. Specifically, the fact that the intensity reaches a maximum near calculated charge neutrality suggests that electrostatics dominate the interaction of DOTAP with DNA. Completing the picture are results obtained from the electrophoretic light scattering studies of these same complexes using LDE. Representative results are shown in Fig. 6. The c potential was obtained from the electrophoretic mobility in the limit of the Smoluchowski approximation (Eq. 16). Using a Brookhaven ZetaPALS equipped with a 30-mW laser (676 nm) and the capability of making both PALS and LDE measurements, the power spectrum was obtained for each sample. The applied electric field was automatically chosen by the software, based on the conductivity of the sample solution. This value was generally between 14 and 16 V/cm. The polarity of the voltage was cycled at 4 Hz. The resulting power spectra were fit to a single Lorentzian curve and the mobility extracted from the fitted curve. Virtually identical results were obtained with PALS. As lipid is initially added to DNA, the c potential displays a gradual increase in magnitude (e.g., becomes more negative). This phenomenon may be explained by the fact that, as the lipid interacts with the DNA, the DNA becomes more closely packed at the surface of the complexes, resulting in an increased negative-charge density. As lipid is added beyond charge neutrality, a rapid inversion of the c potential from negative to positive is observed. Above a 2:1 (+/–) ratio, the c potential remains constant at +27 mV. Under these buffer conditions, DNA itself has a c potential of –25 mV; the DOTAP liposomes possess a c potential of +45 mV. By combining these studies with dynamic light scattering results, it can be seen that the colloidal stability of these complexes is much less when the c potential is between –20 and +20 mV.

370

Wiethoff and Middaugh

The values of the c potential reported here are not absolute values, but reflect also the concentration at which the complexes were prepared, as well as solution parameters, such as pH, ionic strength, and buffer species present. Therefore, making identical measurements for DOTAP–DNA complexes under slightly different solution conditions may yield different results.

7.3.2. Cationic Polymer–DNA Complexes For comparison, similar studies were conducted to characterize the interaction of the cationic gene delivery polymer, polyethylenimine (PEI) with DNA. Phase analysis light scattering was used, in addition to static and dynamic measurements. Instrumental conditions were employed similar to those in the DOTAP studies. In this case, 10 mM Tris, pH 7.8, was used as a buffer. Complexes of a 25,000 molecular weight PEI with a 4.7-kbp plasmid were prepared as descibed for the DOTAP studies. The final DNA concentration was 10 µg/mL. Results are summarized in Figs. 7–9. The variation in the hydrodynamic size of the complexes, upon addition of PEI, is presented in Fig. 7A. The reported hydrodynamic diameter was obtained using the method of cumulants. The size appears to be constant at 80–90 nm until a 1.5:1 PEI nitrogen:DNA phosphate (N:P) ratio is reached. At this point, a rapid increase in the hydrodynamic size occurs. Between an N:P ratio of 2 and 3.5, the complexes appear to be larger than 1 µm in diameter. Again, this large increase in size results from the formation of charge-neutral complexes that no longer possess a strong electrostatic repulsion between one another. The exact diameters of the complexes in this size range are not accurately determined, since the Rayleigh-Gans-Debye or RGD theory of light scattering begins to break down and absorption of light begins to become a problem. Other methods are, of course, available to measure the size of these larger particles. As the complexes begin to contain an excess positive charge from the PEI, the size begins to decrease sharply, eventually reaching a relative constant value between 45 and 60 nm in diameter. The change in the polydispersity index with increasing amounts of PEI was also determined (Fig. 7B). For complexes with DNA in charge excess, the polydispersity initially decreases, manifesting a minimum at ~1.7 N:P. At higher N:P ratios, the polydispersity index increases to a maximum at an N:P ratio of 3 before finally decreasing to a relatively constant value of approx 0.05. These results suggest that in the presence of excess PEI, the DNA is collapsed into small particles with a narrow size distribution. Studies using electron microscopy to visualize similar complexes between PEI and DNA have reported the existence of a homogenous population of toroidal particles in this size range (38). Static light scattering studies again correlate well with the hydrodynamic size, with the exception of complexes containing PEI in charge excess (Fig. 8). In such cases, the intensity of scattered light is greater than that of complexes at low N:P ratios, even though the hydrodynamic size is smaller. Thus, it appears that the

Synthetic Gene Therapy Vectors

371

Fig. 7. The effect of the PEI nitrogen:DNA phosphate ratio on the hydrodynamic properties of PEI–DNA complexes. (A) The hydrodynamic diameter of the complexes obtained by dynamic light scattering at 90 degrees as a function of the ratio of PEI:DNA phosphate, is shown. The diameter was obtained from the ACF by the method of cumulants. (B) The polydispersity index as a function of charge ratio was obtained using the method of cumulants. The DNA concentration was 10 µg/mL in all cases.

molecular weight of the complex is not the sole determinant of the intensity of scattered light at 90 degrees. Both the Debye factor, P(e) and the refractive index increment probably change significantly, suggesting that the excess scattered light

372

Wiethoff and Middaugh

Fig. 8. The effect of charge ratio on the intensity of scattered light at 90 degrees for PEI–DNA complexes. Samples were those described in Fig. 7.

Fig. 9. The effect of charge ratio on the c potential of PEI–DNA complexes. Using the same samples prepared for DLS and SLS studies, the c potential was monitored as a function of charge ratio by phase analysis light scattering.

is caused by the collapse of the DNA from a more extended to a more condensed structure. This observation has been previously reported (21,39–42) for the collapse of DNA induced by polycations such as spermine and cobalt hexa-amine.

Synthetic Gene Therapy Vectors

373

Phase analysis light scattering shows a similar trend to that seen with DOTAP complexes by laser Doppler electrophoresis (Fig. 9). When PEI was added to excess DNA, the c potential was initially near –23 mV. As the amount of PEI increased, the c potential showed a slight increase in magnitude to –30 mV at an N:P ratio of 2:1. Between an N:P ratio of 2 and 3, a rapid inversion in the c potential is again observed with a maximum value of approx +25 mV seen for complexes at N:P ratios above 3. The use of these various light scattering techniques to determine the colloidal properties of complexes prepared at various ratios of DNA:delivery polymer, is particularly useful when using polycations containing several different sources of positive charge and when the degree of positive charge may depend on parameters such as pH or ionic strength. This is the case for PEI, which is known to contain both primary and secondary amines. By preparing complexes at various N:P ratios and monitoring the hydrodynamic size and electrokinetic parameters such as the c potential, a description of the protonation state of the amines may be obtained. Here, it appears that roughly 33% of the PEI amines are protonated at a pH of 7.8, assuming that all of the negative charges of the DNA are neutralized by the PEI amines when the c potential is effectively 0. 8. Conclusion The use of light scattering techniques can provide a wealth of information regarding the colloidal properties of complexes of polycations with DNA, as well as similar information about the components themselves. Although a significant effort must be made to ensure that the data are not influenced by particulate contaminants, a thorough understanding of the principles and practices involved in making and interpreting such measurements can aid significantly in the development of synthetic delivery systems for use in gene therapy. Correlating these properties with biological activity may also provide insight into the influence of the structure of nonviral gene delivery vectors on their overall biological and clinical efficacy (2). Another possible use of these techniques not discussed in this chapter is in the study of the interaction of polycation–DNA complexes with other entities commonly encountered in the process of transfection. Several studies have already used dynamic and static light scattering techniques to explore the interaction of these complexes with various serum components such as proteins and polyanions (12,43,44). Problems arise from the fact that these entities may also contribute to the ACF, making interpretation more complicated. Another use not discussed here involves the monitoring of various light scattering parameters as a function of time. This approach permits the kinetics of assembly and disassembly of complexes to be studied. Phenomena such as nucleation (via analysis of lag times) and stability are readily

374

Wiethoff and Middaugh

analyzed by such methods. Overall, however, these techniques provide a powerful tool for the evaluation and characterization of colloidal systems used in gene delivery. References 1. Ross, P. C. and Hui, S. W. (1999) Lipoplex size is a major determinant of in vitro lipofection efficiency. Gene Ther. 6, 651–659. 2. Duguid, J. G., Li, C., Shi, M., Logan, M. J., Alila, H., Rolland, A., et al. (1998) A physicochemical approach for predicting the effectiveness of peptide-based gene delivery systems for use in plasmid-based gene therapy. Biophys. J. 74, 2802–2814. 3. Godbey, W. T., Wu, K. K., and Mikos, A. G. (1999) Size matters: Molecular weight affects the efficiency of poly(ethylenimine) as a gene delivery vehicle. J. Biomed. Mater. Res. 45, 268–275. 4. Goula, D., Remy, J. S., Erbacher, P., Wasowicz, M., Levi, G., Abdallah, B., and Demeneix, B. A. (1998) Size, diffusibility and transfection performance of linear PEI/DNA complexes in the mouse central nervous system. Gene Ther. 5, 712–717. 5. Dunlap, D. D., Maggi, A., Soria, M. R., and Monaco, L. (1997) Nanoscopic structure of DNA condensed for gene delivery. Nucleic Acids Res. 25, 3095–3101. 6. Sternberg, B., Hong, K., Zheng, W., and Papahadjopoulos, D. (1998) Ultrastructural characterization of cationic liposome-DNA complexes showing enhanced stability in serum and high transfection activity in vivo. Biochim. Biophys. Acta 1375, 23–35. 7. Xu, Y., Hui, S. W., Frederik, P., and Szoka, F. C., Jr. (1999) Physicochemical characterization and purification of cationic lipoplexes. Biophys. J. 77, 341–353. 8. Cherng, J. Y., van de Wetering, P., Talsma, H., Crommelin, D. J., and Hennink, W. E. (1996) Effect of size and serum proteins on transfection efficiency of poly ([2-dimethylamino]ethyl methacrylate)-plasmid nanoparticles. Pharm. Res. 13, 1038–1042. 9. Plank, C., Tang, M. X., Wolfe, A. R., and Szoka, F. C., Jr. (1999) Branched cationic peptides for gene delivery: role of type and number of cationic residues in formation and in vitro activity of DNA polyplexes. Hum. Gene Ther. 10, 319–332. 10. Zuidam, N. J., Hirsch-Lerner, D., Margulies, S., and Barenholz, Y. (1999) Lamellarity of cationic liposomes and mode of preparation of lipoplexes affect transfection efficiency. Biochim. Biophys. Acta 1419, 207–220. 11. Oupicky, D., Konak, C., Dash, P. R., Seymour, L. W., and Ulbrich, K. (1999) Effect of albumin and polyanion on the structure of DNA complexes with polycation containing hydrophilic nonionic block. Bioconjugate Chem. 10, 764–772. 12. Li, S., Tseng, W. C., Stolz, D. B., Wu, S. P., Watkins, S. C., and Huang, L. (1999) Dynamic changes in the characteristics of cationic lipidic vectors after exposure to mouse serum: implications for intravenous lipofection. Gene Ther. 6, 585–594. 13. Eastman, S. J., Siegel, C., Tousignant, J., Smith, A. E., Cheng, S. H., and Scheule, R. K. (1997) Biophysical characterization of cationic lipid: DNA complexes. Biochim. Biophys. Acta 1325, 41–62. 14. Chu, B. (1974) Laser Light Scattering. Academic Press, New York.

Synthetic Gene Therapy Vectors

375

15. Stock, R. S. and Ray, W. H. (1985) Interpretation of photon correlation spectroscopy data: a comparison of analysis methods. J. Polymer Scie. Polymer Physics Edition 23, 1393–1447. 16. Koppel, D. E. (1972) Analysis of macromolecular polydispersity in intensity correlation spectroscopy: the method of cumulants. J. Chem. Phys. 57, 4814. 17. Takeo, M. (1999) Disperse Systems, Wiley-VCH, Weinheim. 18. Lawson, C. L. and Hanson, R. J. (1974) Solving Least Squares Problems, Prentice Hall, Englewood Cliffs, NJ. 19. Provencher, S. W. (1982) A constrained regularization method for inverting data represented by linear algebraic or integral equations. Comp. Phys. Commun. 27, 213–218. 20. Provencher, S. W. (1982) CONTIN: A general purpose constrained regularization program for inverting noisy linear algebraic and integral equations. Comp. Phys. Commun. 27, 219. 21. Wilson, R. W. and Bloomfield, V. A. (1979) Counterion-induced condesation of deoxyribonucleic acid. a light- scattering study. Biochemistry 18, 2192–2196. 22. Zimm, B. H. (1948) Apparatus and methods for measurement and interpretation of the angular variation of light scattering: preliminary results on polystyrene solutions. J. Chem. Phys. 16, 1093–1099. 23. Hunter, R. J. (1993) Introduction to Modern Colloid Science. Oxford University Press, New York. 24. McNeil-Watson, F., Tscharnuter, W., and Miller, J. (1998) A new instrument for the measurement of very small electrophoretic mobilities using phase analysis light scattering. Collids and Surfaces A. 140, 53–57. 25. Langowski, J., Kremer, W., and Kapp, U. (1992) Dynamic light scattering for study of solution conformation and dynamics of superhelical DNA. Methods Enzymol. 211, 430–448. 26. Kissa, E. (1999) Dispersions. Characterization, Testing, and Measurement. Marcel Dekker, New York, pp. 433–492. 27. Patapoff, T. W., Tani, T. H., and Cromwell, M. E. (1999) Low-volume, short-path length dynamic light scattering sample cell for highly turbid suspensions. Anal. Biochem. 270, 338–340. 28. Liu, Y., Mounkes, L. C., Liggitt, H. D., Brown, C. S., Solodin, I., Heath, T. D., and Debs, R. J. (1997) Factors influencing the efficiency of cationic liposomemediated intravenous gene delivery. Nat. Biotechnol. 15, 167–173. 29. Templeton, N. S., Lasic, D. D., Frederik, P. M., Strey, H. H., Roberts, D. D., and Pavlakis, G. N. (1997) Improved DNA: liposome complexes for increased systemic delivery and gene expression. Nat. Biotechnol. 15, 647–652. 30. Kreiss, P., Cameron, B., Rangara, R., Mailhe, P., Aguerre-Charriol, O., Airiau, M., et al. (1999) Plasmid DNA size does not affect the physicochemical properties of lipoplexes but modulates gene transfer efficiency. Nucleic Acids Res. 27, 3792–3798. 31. Langowski, J., Benight, A. S., Fujimoto, B. S., Schurr, J. M. and Schomburg, U. (1985) Change of conformation and internal dynamics of supercoiled DNA upon binding of Escherichia coli single-strand binding protein. Biochemistry 24, 4022–4028.

376

Wiethoff and Middaugh

32. Langowski, J., Giesen, U., and Lehmann, C. (1986) Dynamics of superhelical DNA studied by photon correlation spectroscopy. Biophys. Chem. 25, 191–200. 33. Langowski, J. (1987) Salt effects on internal motions of superhelical and linear pUC8 DNA. Dynamic light scattering studies. Biophys. Chem. 27, 263–271. 34. Langowski, J. and Giesen, U. (1989) Configurational and dynamic properties of different length superhelical DNAs measured by dynamic light scattering. Biophys. Chem. 34, 9–18. 35. Schaper, A., Urbanke, C., and Maass, G. (1991) Salt dependent changes in structure and dynamics of circular single stranded DNA of filamentous phages of Escherichia coli. J. Biomol. Struct. Dyn. 8, 1211–1232. 36. Fishman, D. M. and Patterson, G. D. (1996) Light scattering studies of supercoiled and nicked DNA. Biopolymers 38, 535–552. 37. Verwey, E. J. W. and Overbeek, J. T. G. (1948) Theory of the Stability of Lyophobic Colloids. Elsevier, Amsterdam. 38. Tang, M. X. and Szoka, F. C. (1997) The influence of polymer structure on the interactions of cationic polymers with DNA and morphology of the resulting complexes. Gene Ther. 4, 823–832. 39. Bloomfield, V. A., He, S., Li, A. Z., and Arscott, P. B. (1991) Light scattering studies on DNA condensation. Biochem. Soc. Trans. 19, 496. 40. Deng, H. and Bloomfield, V. A. (1999) Structural effects of cobalt-amine compounds on DNA condensation. Biophys. J. 77, 1556–1561. 41. Arscott, P. G., Li, A. Z., and Bloomfield, V. A. (1990) Condensation of DNA by trivalent cations. 1. Effects of DNA length and topology on the size and shape of condensed particles. Biopolymers 30, 619–630. 42. Arscott, P. G., Ma, C., Wenner, J. R., and Bloomfield, V. A. (1995) DNA condensation by cobalt hexaammine (III) in alcohol-water mixtures: dielectric constant and other solvent effects. Biopolymers 36, 345–364. 43. Zelphati, O., Uyechi, L. S., Barron, L. G., and Szoka, F. C., Jr. (1998) Effect of serum components on the physico-chemical properties of cationic lipid/oligonucleotide complexes and on their interactions with cells. Biochim. Biophys. Acta 1390, 119–133. 44. Oupicky, D., Konak, C., Dash, P. R., Seymour, L. W., and Ulbrich, K. (1999) Effect of albumin and polyanion on the structure of DNA complexes with polycation containing hydrophilic nonionic block. Bioconjug. Chem. 10, 764–772.

Renal Gene Therapy

377

23 Renal Gene Therapy Yeong-Hau H. Lien and Li-Wen Lai 1. Introduction Until now there is no renal gene therapy available for clinical use, however, gene therapy for several experimental renal diseases has been tested with promising results. The kidney is a well-differentiated organ with a variety of specialized compartments, i.e., vascular, glomerular, tubular, and interstitial. Many physiological factors such as cell turnover rate, blood flow, and urine flow, as well as anatomical factors such as glomerular basement membrane and nephron segment arrangement, may affect the specificity and efficacy of gene therapy in the kidney. On the other hand, the kidney has a major advantage over other solid organs, since it is accessible by many routes, including intrarenal artery infusion, retrograde delivery through the urinary tract, direct injection into renal parenchyma, and perfusion into the donor graft prior to transplantation (1). This chapter reviews nonviral gene transfer in the different compartments of the kidney (for review of viral vectormediated renal gene transfer, see refs. 2–5), and in skeletal muscle, for renal gene therapy, and potential applications and safety concerns for renal gene therapy. 2. Gene Transfer in Glomerulus Table 1 lists studies of gene transfer targeting the glomerulus. Two nonviral approaches have been used to achieve gene transfer in the glomerulus: the combination of hemagglutinating virus of Japan (HVJ) and liposome, and multilamellar vesicle (MLV) liposomes.

2.1. HVJ–Liposome The HVJ–liposome system is an efficient gene delivery system targeted to the glomerulus (4,6). HVJ itself does not carry the gene, but provides fusion proteins to enhance gene entry into the cell. The lipid contains phosphatidylserine, From: Methods in Molecular Medicine, vol. 65: Nonviral Vectors for Gene Therapy Edited by: M. A. Findeis © Humana Press Inc., Totowa, NJ

377

378

Lien and Lai

Table 1 Gene Transfer in Glomerulus Vector

Gene

Route

Animal model

HVJ–lipo

SV40 T–Ag

Renal artery

Rat

HVJ–lipo

CD59

Renal artery

Dog

Renal artery

Rat

Renal artery

Rat

Renal artery

Rat, nephritis

Renal artery

Rat, nephritis

HVJ–lipo

Renin Promoter–CAT HVJ–lipo TGF-` and PDGF HVJ–lipo Antisense vs TGF-` MLV–lipo E2F Decoy

Results

Ref.

T–Ag expressed in glomeruli CD59 expressed in glomeruli CAT expressed in kidney Glomerulosclerosis

7

Inhibition of mesangial expansion Inhibition of mesangial expansion

11

8 9 10

12

HVJ–lipo, hemagglutinating virus of Japan–liposome; MLV–lipo; multilamellar vesicle–liposome; CAT, chloramphenicol acetyltransferase; Ag, antigen; TGF-`, transforming growth factor `; PDGF, platelet–derived growth factor.

phosphatidylcholine, and cholesterol mixing in a weight ratio of 1:4.8:2. In the initial studies (7–10), a high-mobility group 1 nuclear protein was added to facilitate DNA transport into the nucleus. In the later study (11), this component was omitted without hampering the efficiency of gene transfer. The results of HVJ–liposome-mediated gene transfer or gene therapy, are shown in Table 1. Initial studies showed that this system could deliver genes specifically to renal glomeruli and that expression was transient (7,8). The system has been used to study the tissue specificity of renin promoters (9), and to confirm the role of transforming growth factor-` (TGF-`) and platelet-derived growth factor (PDGF) in the pathogenesis of glomerulosclerosis (10). The main focus, however, has been on the usage of this system for treating experimental glomerulonephritis, which is induced by anti-thymocyte (anti-Thy) 1.1 antibodies. Akagi et al. (11) reported the effects of antisense oligodeoxynucleotides (AS ODN) against TGF-`1 on anti-Thy 1.1 glomerulonephritis. The AS ODNs were given 2 d after induction of glomerulonephritis. The gene therapy inhibited TGF-`1 expression and suppressed extracellular matrix accumulation. However, glomerular cell count was not affected.

2.2. Liposomes MLV liposomes were selected to target the glomerulus because of their efficient encapsulation of the DNA, high affinity for mesangial cells because of the positively charged surface, and low cytotoxicity (12). MLV liposomes contain

Renal Gene Therapy

379

Table 2 Gene Transfer in Tubule Vector

Gene

Liposome

`-gal

Liposome

CA II

PEI

Luc and `-gal

None

`-gal

None

Antisense vs NaPi-2 Antisense vs D1AR Antisense vs iNOS

None None

Route

Animal model

Renal artery Mouse and retrograde Retrograde CA II-def. mouse Renal artery Rat

Results

Ref.

Expressed in tubules

15

Correction of renal tubular acidosis Expressed in proximal tubules Expressed in tubules

16 17

Renal parenchyma iv

Mouse

Renal parenchyma ic

Rat

Reduction of phosphate 20 uptake Antidiuresis 21

Rat, ischemic renal failure

Attenuated acute renal failure

Rat

18

22

`-gal, `-galactosidase; CA II, carbonic anhydrase II; def., deficient; PEI, polyethylenimine; NaPi-2, sodium; phosphate cotransporter-2; D1AR, dopamine 1A receptor; iNOS, inducible nitric oxide synthase; iv, intravenous injection; ic, intracardiac injection.

dioleoylphosphatidylethanolamine, dilauroylphosphatidylcholine, and N(_-trimethylammonioacetyl)-didodecyl-D -glutamate chloride at a molar ratio of 1:2:2. The E2F decoy oligonucleotide was used to inhibit mesangial cell proliferation in anti-Thy 1.1 glomerulonephritis. E2F is a transcription factor that activates several cell cycle regulatory proteins including dihydrofolate reductase, c-Myc, DNA polymerase A, cell division cycle 2 (cdc2) kinase, and proliferating cell nuclear antigen (PCNA). A short double-stranded ODN containing a cis-element for a particular transcription factor, competes with the endogenous cis-element for the transcription factor, and thus is named “decoy.” MLV liposome–E2F decoy complexes were injected into the rat renal artery 36 h after induction of anti-Thy 1.1 glomerulonephritis. The gene therapy suppressed the proliferation of mesangial cells by 72%, and inhibited the expression of PCNA. Using fluorescein isothiocyanate-labeled oligonucleotides, they have shown that, 2 h after injection, 60% of the glomeruli were stained with the fluorescence, detectable mainly in the nuclei of mesangial cells. Tubular and interstitial cells exhibited no fluorescence (12). 3. Gene Transfer in Tubule Table 2 lists studies of nonviral gene transfer in the renal tubules. Three approaches have been used: liposomes, polycations, and naked DNA.

380

Lien and Lai

3.1. Liposomes Cationic liposomes are an effective vector for gene transfer in vivo. They condense DNA and protect DNA from degradation, and therefore enhance gene transfection (13). Liposomes are particularly effective for gene transfer to the renal tubule (14). Lai et al. (15) reported gene transfer to the renal tubules when Lipofectin (Life Technology, Gaithersburg, MD)–DNA were either injected into mouse renal artery or retrogradely infused via the renal pelvis. Renal artery injection of the DNA–liposome complex resulted in `-galactosidase expression in cortical and outer medullary tubules, but not in the inner medulla, or glomerular, vascular, and interstitial compartments. Retrograde infusion produced similar patterns of expression. Intraparenchymal injection produced no expression. The expression of `-galactosidase lasted for 3–4 wk (15). Lai et al. (16) further examined the efficacy of gene therapy in an animal model of carbonic anhydrase (CA) II deficiency. CAII is a cytoplasmic enzyme, which catalyzes the reversible hydration of carbon dioxide. A chimeric gene, pCMVCAII, containing a human CAII coding sequence driven by a cytomegalovirus promoter, was complexed with Lipofectin and injected via the intrarenal-pelvic route into both kidneys of CAII-deficient mice. After gene therapy, CAII deficient mice were able to acidify urine following an acid load. Both CAII DNA and mRNA levels were highest on d 3 after injection, diminishing gradually, but remaining detectable after 1 mo. Immunohistochemistry study (16) using antibodies against CAII showed the expression of CAII mainly in the tubules in the area of the cortico-medullary junction.

3.2. Polycations Polycations, molecules that carry multiple positive charges, and thus interact with negatively charged DNA, have been used as a vector for gene transfer. Boletta et al. (17) reported that polyethylenimine (PEI), a polycation, delivered reporter genes efficiently to the renal tubules. PEI is added to DNA solution at a ratio of 10 equivalents of PEI nitrogen per DNA phosphate. After injection via the renal artery, the transgene expression was located almost exclusively in proximal tubules. The transfection efficiency of branched PEI 25 k was significantly better than linear PEI 22 k, PEI 800 k, dioctadecylamido glyclspermine (DOGS), and 1,2dioleoy-3-trimethylammonium-propane in the kidney (17).

3.3. Naked DNA Naked DNA can be transferred efficiently in skeletal muscle and other organs. In the kidney, gene transfer of naked DNA can be achieved by direct injection into the renal parenchyma. AS ODNs can be delivered to the renal tubule without specific vehicles. Oberbauer et al. (18) have demonstrated active

Renal Gene Therapy

381

accumulation of oligonucleotides in the proximal tubules after intravenous infusion. Besides the proximal tubules, some oligonucleotides were found in the glomerulus, parietal epithelial cells of Bowman’s space, and distal tubules (19). Lai et al. (15) reported that direct injection of plasmid DNA into renal parenchyma led to transgene expression only in the renal tubules. The expression was not as strong as plasmid–liposome complexes given via the renal artery or renal pelvis. Intravenous administration of AS ODNs against sodium phosphate cotransporter (NaPi-2) resulted in reduction of phosphate uptake into renal brush border membrane vesicles and the expression of NaPi-2 protein in these vesicles (20). Wang et al. (21) used the same approach to prove that treatment with AS ODNs selectively inhibited dopamine D1A receptor in the renal tubular epithelium and vasculature, and reduced urine output and urinary sodium excretion. Noiri et al. (22) injected rats intracardiacally with antisense oligonucleotides targeted to inducible nitric oxide synthase (iNOS) to achieve selective blockage of this enzyme in the kidney. The rats were protected from the effects of subsequently inflicted ischemic renal injury. This study provided direct evidence of the cytotoxic effects of nitric oxide via iNOS and offers a novel approach for prevention of ischemic acute renal failure (22). 4. Gene Transfer in Renal Vasculature Ischemia induced a marked increase in intracellular adhesion molecule-1 (ICAM-1) expression along the endothelial cells of renal vasculature. ICAM-1 plays an important role on the neutrophil-mediated ischemia-reperfusion injury. Table 3 lists several studies that demonstrated the effects of ICAM-1 AS ODNs on reperfusion injury and graft function in renal transplantation model (23–26). Haller et al. (23) have shown that intravenous injection of a mixture of AS ODNs against ICAM-1 and Lipofectin (at a ratio of Lipofectin 0.8 mg/mg DNA) 6 h before the ischemia insult, suppressed ICAM-1 expression along the endothelial cell lining of the blood vessels and in the peritubular area. The gene therapy reduced the ischemia-induced infiltration of granulocytes and macrophages and resulted in less cortical renal damage. The same group reported that similar treatment caused less tubular necrosis in a rat autotransplantation model, which includes a 30-min ischemia time and 60 min warm ischemia time. One day after transplantation, the AS ODN-treated rat had a normal kidney function; the controls were anuric (24). In another study, using an isotransplantation model, the investigators demonstrated that pretreatment of the donor kidney with AS ODNs against ICAM-1 prolonged the graft survival. At wk 20, 90% of the AS ODN-treated group survived; less that 30% of the control group remained alive. The histology at wk 20 also showed less fibrosis and tubular atrophy in the AS ODN-treated kidney (25). Using a similar approach, Stepkowski et al. (26) showed that AS ODNs could block rejection in allografts when given to donor, given during graft perfusion, or given after transplantation.

382

Lien and Lai

Table 3 Gene Transfer in Renal Vasculature Vector

Gene

Route

Liposome Antisense vs ICAM-1 Liposome Antisense vs ICAM-1 Liposome Antisense vs ICAM-1 None Antisense vs ICAM-1

iv

Animal model

Results

Ref.

Attenuate reperfusion injury

23

iv

Rat, ischemic Renal failure Rat, autotransplantation

Prevent reperfusion injury

24

iv

Rat, isotransplantation

Prolong graft survival

25

iv

Rat, allotransplantation

Block allograft rejection

26

ICAM-1, intracellular adhesion molecule-1; iv, intravenous injection.

5. Extrarenal Gene Transfer for Renal Gene Therapy Renal gene therapy can be accomplished by extrarenal gene transfer of a gene encoding a circulatory protein, which in turn targets at the renal tissues via a ligand–receptor interaction. Table 4 lists studies using this strategy to treat glomerular diseases. TGF-`1 is the key cytokine that promotes the production and deposition of extracellular matrices in the injured tissue. The sustained and excessive expression of TGF-`1 is considered to be causal in the pathogenesis of human and experimental glomerular diseases. Isaka et al. (27) reported that gene therapy by skeletal muscle expression of decorin prevents fibrotic disease in rat kidney. Decorin, a proteoglycan, is a natural inhibitor of TGF-`. After gene transfer, there was an increase in production of decorin in the skeletal muscle and decorin was found to be accumulated in the kidney. The glomerular extracellular matrix accumulation and proteinuria were significantly suppressed in transfected nephritic rats (27). In another approach, Isaka et al. (28) constructed a gene that encodes the TGF-` receptor-immunoglobulin, Fc chimeric protein (TGF-`R-Ig). This chimeric protein is a soluble protein, which competes off the binding of TGF-` to its receptor. Gene therapy by skeletal muscle expression of TGF-`-Ig suppressed the glomerular TGF-` mRNA with a comparable effect in the reduction of extracellular matrix accumulation. Using the same method, Isaka et al. (29) reported their preliminary data on treating streptozotocininduced diabetic nephropathy. It appears that skeletal expression of TGF-`R-Ig also reduces extracellular matrix accumulation in experimental diabetic nephropathy (29). 6. Potential Applications for Renal Diseases Renal diseases that potentially can be treated with gene therapy fall into three categories: hereditary diseases, acquired diseases, and diseases related to renal transplantation. Table 5 lists these diseases and their potential therapeutic genes. This list continues to grow, because new disease-linked genes and new pathogenic mechanisms of hereditary renal diseases are discovered at a rapid pace.

Renal Gene Therapy

383

Table 4 Extrarenal Gene Transfer for Renal Diseases Vector HVJ–lipo HVJ–lipo HVJ–lipo

Gene

Route

Decorin im TGF-`R-Ig im TGF-`R-Ig im

Animal model Rat, nephritis Rat, nephritis Rat, diabetes

Results Inhibition of mesangial expansion Inhibition of mesangial expansion Inhibition of mesangial expansion

Ref. 27 28 29

HVJ–lipo, hemagglutinating virus of Japan-liposome; TGF-`R-Ig, TGF-`-receptor-immunoglobulin; im, intramuscular injection.

Table 5 Potential Applications of Renal Gene Therapy: Renal Diseases and Potential Therapeutic Genes Hereditary renal diseases Glomerular diseases Alport’s syndrome: collagen type IV _5, _6, _3, or _4 Fabry’s disease: _-galactosidase A Congenital nephrotic syndrome of the Finnish type: Nephrin Tubular diseases Nephrogenic diabetes insipidus: Vasopressin V2 receptor or aquaporin-2 water channel Cystinuria type I: Apical cystein-dibasic amino acid transporter Dent’s disease: Chloride channel (ClC-5) Bartter syndrome: Na-K-2Cl transporter, adenosine triphosphate-regulated potassium channel, kidney-specific chloride channel Gitelman’s syndrome: Thiazide-sensitive NaCl transporter Carbonic anhydrase II deficiency: Carbonic anhydrase II Pseudohypoaldosteronism type I: Epithelial Na channel Neonatal severe hyperparathyroidism: Calcium-sensing receptor Oculocerebrorenal syndrome of Lowe: Inositol polyphosphate-5-phosphatase Acquired renal diseases Glomerulonephritis: E2F decoy, TGF-`R-Ig, antisense vs TGF-`1 Diabetic nephropathy: TGF-`R-Ig Acute tubular necrosis: Antisense vs ICAM-1 and iNOS Acute interstitial nephritis: INF-a Renal artery stenosis: VEGF, cNOS, E2F decoy, antisense vs PCNA and cdc2 kinase Renal transplantation Posttransplant ischemic injury: Antisense vs ICAM-1 and iNOS Rejection: Fas ligand, CTLA4-Ig, antisense vs. ICAM-1 TGF-`R-Ig, TGF-`-receptor-immunoglobulin; INF, interferon; ICAM-1, intracellular adhesion molecule-1; iNOS, inducible nitric oxide synthase; VEGF, vascular endothelial growth factor; PCNA, proliferating cell nuclear antigen; cNOS, constitutive nitric oxide synthase.

384

Lien and Lai

6.1. Hereditary Renal Diseases Current gene therapy strategies for hereditary diseases aim at providing normal genes for overcoming specific genetic defects. Therefore, the gene replacement therapy is limited to autosomal recessive or x-linked diseases. For autosomal dominant diseases, strategies other than gene replacement must be developed. For example, the addition of normal gene products, polycystin 1 or 2, is unlikely to stop cyst formation in autosomal dominant polycystic kidney disease (ADPKD). Therapeutic genes aiming at antagonizing downstream events, such as epithelial cell proliferation, may be more useful for treating ADPKD. Three hereditary glomerular diseases (Alport’s syndrome, Fabry’s disease, and congenital nephrotic syndrome of the Finnish type) can be treated with gene therapy since therapeutic genes have been identified (Table 5). The molecular defect of congenital nephrotic syndrome of the Finnish type has been identified recently in nephrin, a 1241-residue putative transmembrane protein of the Ig family of cell adhesion molecules (30). Most patients with this disease died of complications of nephrotic syndrome or renal failure within the first year of life. Patients with Alport’s syndrome or Fabry’s disease usually develop renal failure in their adult life. There are currently no treatments available to stop or slow the progression of renal failure. Most of the hereditary tubular diseases suitable for gene therapy are disorders of electrolyte transport. Recently, genetic and molecular approaches have elucidated the underlying molecular defects in many of these disorders (reviewed in ref. 31). The clinical features of these disorders vary widely, from potentially fatal disease during infancy to normal life-span. However, most of the patients have to be on a life-long treatment to prevent potentially lethal events caused by electrolyte imbalance. The development of gene therapy for hereditary glomerular and tubular diseases is highly beneficial, but it is a challenging task. The major limitation is that currently there are no delivery systems that can provide long-term transgene expression. New strategies, such as homologous recombinationbased gene therapy using chimeric RNA/DNA oligonucleotides or short DNA fragments, may be able to overcome this problem (32). The preliminary results of in vivo gene conversion of mutated CAII using chimeric RNA/DNA oligonucleotides are promising (33). 6.2. Acquired Renal Diseases The success of gene therapy in previously described animal models of glomerulonephritis, diabetic nephropathy, and acute tubular necrosis suggest that these acquired diseases may be ameliorated by gene therapy. However, because the effect of gene therapy is transient, its utility of gene therapy may be limited to the acute phase of diseases, unless the treatment can be repeatedly given. More studies are needed before those modalities can be applied to patients.

Renal Gene Therapy

385

Gene therapy may be useful for acute interstitial nephritis. Kelley’s group (5) has developed a gene transfer technique to elucidate the mechanisms of autoimmune interstitial nephritis. They injected renal tubular cells, which are genetically modified to secret cytokines or chemokines into the renal capsule of MRL-Faslpr mice to elicit renal interstitial injury and found that gene transfer of cytokines incited more severe interstitial nephritis in interferon-a receptordeficient, than in interferon-a receptor intact autoimmune MRL-Faslpr mice. They suggested that interferon-a may provide a negative regulatory pathway capable of limiting macrophage-mediated renal inflammation (34). Renal artery stenosis is another disease for which gene therapy may be beneficial. Morishita et al. (35) have shown that gene therapy with E2F decoy can inhibit smooth muscle proliferation in vivo. In addition, treatment with PCNA and cdc2 kinase AS ODNs, or c-NOS cDNA reduced neointima formation in the arterial walls after balloon injury (36) or in the interposition grafted aorta (37). Therefore, all these therapeutic genes can be used to prevent postangioplasty restenosis of renal artery. Furthermore, vascular endothelial growth factor gene therapy has been shown to be effective for treating both coronary artery and peripheral artery diseases by increasing angiogenesis (38,39). It is conceivable that similar therapy may prevent the progression of renal failure in patients with renal vascular disease.

6.3. Renal Transplantation Post-transplant ischemic injury is a common problem of renal transplantation. The incidence is about 20%, but it increases with increase in donor’s age and in cold ischemic time of the graft. A series of studies described in Subheading 4. elegantly demonstrated that posttransplant ischemic injury can be ameliorated by pretreating the graft with AS ODNs against ICAM-1 (24,25). AS ODN against iNOS may have a similar effect (22). It is possible that the combination of the two may have additive effects, because the target cell types are different: endothelial cells for ICAM-1 and tubular cells for iNOS. Another common problem of renal transplantation is rejection. There are three approaches utilizing gene therapy to block rejection: inhibition of ICAM-1 expression, expression of Fas ligand, which interacts with Fas and causes apoptosis of T-cells; and expression of circulatory fusion protein, CTLA4-Ig, which blocks costimulation pathway of T cells. As mentioned earlier, AS ODNs against ICAM-1 given before harvesting of the graft, during graft perfusion, or after transplantation, can prolong graft survival by blocking rejection (26). Swenson et al. (40) showed that rat renal allografts perfused with the adenoviral vector containing Fas ligand cDNA could survive longer in recipients. Olthoff et al. (41) demonstrated that adenovirus-mediated gene transfer of a gene encoding CTLA4-Ig fusion protein into cold-preserved liver allograft blocked the costimulation pathway and resulted into prolonging graft survival and induction of immunological unresponsiveness.

386

Lien and Lai

7. Safety Concerns 7.1. Plasmids Plasmid DNA used for gene therapy is prepared from bacteria. Unmethylated CpG motifs in bacteria have stimulatory effects on lymphocytes. Schwartz et al. (42) reported that intratracheal instillation of bacteria DNA in mice caused inflammation in the lower respiratory tract, with increased number of neutrophil and concentrations of several cytokines. Methylation of the bacterial DNA significantly reduced inflammation. This complication has not been observed in studies using intravenous administration of plasmid DNA–liposome complexes (43,44). The importance of this unmethylated CpG motif-induced toxicity remains to be determined.

7.2. Oligonucleotides Monteith et al. (45) injected monkeys with an escalating dose regimen of 3–80 mg/kg oligonucleotides, and evaluated the changes in renal function, urinary parameters, and renal histopathology. They found that proteinuria developed at 40 mg/kg and mild renal insufficiency developed at 80 mg/kg. Tubular degeneration was present at 40 and 80 mg/kg. Lower doses, equal or less than 10 mg/kg did not affect renal function or histology. Granulation in the cytoplasm of proximal tubular cells was present at all doses and increased with dose. Henry et al. (46) reported systemic toxicity of oligonucleotides in mice, which received iv injection at 0.8–100 mg/kg every other day for 4 wk. At 20 mg/kg or higher, there was mononuclear cell infiltration in a number of organs, splenomegaly, and lymphoid hyperplasia. At 100 mg/kg, elevated liver function tests and thrombocytopenia were observed. Those changes were reversed at the end of the 4-wk recovery period. Oligonucleotides are clearly nephrotoxic, hepatotoxic, and immune-stimulatory at a high dose. However, they are quite safe if given at 10 mg/kg or less, even for a prolonged period of time (4 wk).

7.3. Liposomes and Polycations Haller et al. (23) reported that high doses of liposomes (Lipofectin at >2 mg/mg DNA) are cytotoxic to endothelial cells in vitro. Brazeau et al. (47) used an in vitro system of cultured muscle tissue to assess myotoxicity of various cationic macromolecules. They found that cationic liposomes were less myotoxic compared to the dendrimers and polylysine. Myotoxicity was dependent on concentration and mol wt. The presence of plasmid DNA significantly reduced toxicity probably because of the decrease of overall positive charge of the complexes. When DNA– liposome complexes were injected into the renal pelvis of mice, there were no histological changes in the kidney or changes in renal functions (15,16). PEI appears to have low toxicity both in vivo and in vitro (17).

Renal Gene Therapy

387

Intravenous administration of some liposomal drugs can trigger immediate hypersensitivity reactions, including symptoms of cardiopulmonary distress. Szebeni et al. (48) reported that iv injection of 5-mg boluses of large multilamellar liposomes (dimyristoyl phosphatidylcholine, dimyristoyl phosphatidylglycerol, and cholesterol at 50:5:45 mol ratio) into pigs (32–48 kg) caused a significant rise in pulmonary arterial pressure, pulmonary and systemic vascular resistance, and a decline in cardiac output. These changes peaked between 1 and 5 min after injection, subsided within 10–20 min, and were lipid-dose-dependent. There was a close quantitative and temporal correlation between the hemodynamic changes and the elevations of plasma thromboxane B2 level. Those changes could be inhibited by an anti-C5a monoclonal antibody and soluble CR1. The results of that study (48) indicate that liposomes may activate complements and cause lifethreatening events. However, this complication has not been reported in the studies of liposome-mediated gene transfer by other investigators (43,44). Whether this phenomenon is associated with this particular type of liposome or particular animal model, remains to be determined.

7.4. Fusogenic Protein So far, no toxicity associated with HVJ–liposome complex in small animals has been reported. Repeated administration of the HVJ–liposome induced antibody formation, but the transfection efficiency is not significantly affected (4).

7.5. Gene Products For hereditary renal diseases, gene therapy will introduce a normal, but novel, protein to the host who has not been exposed to such a protein prior to treatment. Theoretically, after gene therapy, patients may develop antibodies against therapeutic gene products, particularly if they are circulatory or membranous proteins. Therefore, it may be advisable to measure those antibodies, and, if they are present at a high titer, immunosuppressive therapy may be added in conjunction of gene therapy at least at the initial period to inhibit antibody formation. 8. Conclusions Gene therapy for renal diseases may become a clinical reality in the near future. The most promising field is the application of gene therapy for renal transplantation. The use of gene therapy for acute tubular necrosis, acute glomerulonephritis, and renal vascular disease is also likely to be developed. The most difficult area is gene therapy for hereditary or chronic renal diseases, because long-term and cellspecific expression of transgenes is still far from achievable. New strategies are needed to overcome these obstacles in this area. Finally, as more is learned about the side effects of nonviral gene therapy, efforts should be made to explore the

388

Lien and Lai

mechanisms of these side effects, and to minimize them. With continuous progress in vector technologies and gene therapy strategies, the authors are confident that the era of gene therapy is coming soon. Acknowledgments This work was supported by NIH grant RO1DK52358 and a grant from the Dialysis Clinic, a nonprofit organization. References 1. Lien, Y. H. and Lai, L. (1997) Gene therapy for renal diseases. Kidney Int. 52(Suppl. 61), S85–S88. 2. Lipkowitz, M. S., Klotman, M. E., Bruggeman, L. A., Nicklin, P., Hanss, B., Rappaport, J., and Klotman P. E. (1996) Molecular therapy for renal diseases. Am. J. Kidney Dis. 28, 475–492. 3. Moullier, P., Salvetti, A., Champion-Arnaud, P., and Ronco, P. M. (1997) Gene transfer into the kidney: current status and limitations. Nephron 77, 139–151. 4. Imai, E. and Isaka, Y. (1998) Strategies of gene transfer to the kidney. Kidney Int. 53, 264–72. 5. Kelley, V. R. and Sukhatme, V. P. (1999) Gene transfer in the kidney. Am. J. Physiol. 276, F1–F9. 6. Isaka, Y., Akagi, Y., Kaneda, Y., and Imai, E. (1998) The HVJ liposome method. Exp. Nephrol. 6, 144–147. 7. Tomita, N., Higaki, J., Morishita, R., Kato, K., Mikami, H., Kaneda, Y., and Ogihara, T. (1992) Direct in vivo gene introduction into rat kidney. Biochem. Biophy. Res. Commun. 186, 129–134. 8. Akami, T., Arkawa, K., Okamoto, M., Akioka, K., Fujiware, I., Nakai, I., et al. (1994) Introduction and expression of human CD59 gene in the canine kidney. Transplant Proc. 26, 1315–1317. 9. Yamada, T., Horiuchi, M., Morishita, R., Zhang, L., Pratt, R. E., and Dzau, V. J. (1995) In vivo identification of a negative regulatory element in the mouse renin gene using direct gene transfer. J. Clin. Invest. 96, 1230–1237. 10. Isaka I., Fujiwara Y., Ueda, N., Kaneda, Y., Kamada, T., and Imai, E. (1993) Glomerulosclerosis induced by in vivo transfection by transforming growth factor-` or platelet-derived growth factor gene into the rat kidney. J. Clin. Invest. 92, 2597–2601. 11. Akagi, Y., Isaka, Y., Arai, M., Kaneko, T., Takenaka, M., Moriyama, T., et al. (1996) Inhibition of TGF-beta 1 expression by antisense oligonucleotides suppressed extracellular matrix accumulation in experimental glomerulonephritis. Kidney Int. 50, 148–155. 12. Maeshima Y., Kashihara, N., Yasuda, T., Sugiyama, H., Sekikawa, T., Okamoto, K., et al. (1998) Inhibition of mesangial cell proliferation by E2F decoy oligodeoxynucleotide in vitro and in vivo. J. Clin. Invest. 101, 2589–2597. 13. Ledley, F. D. (1995) Nonviral gene therapy: the promise of genes as pharmaceutical products. Hum. Gene Ther. 6, 1129–1144.

Renal Gene Therapy

389

14. Lien, Y. H. and Lai, L. (1997). Liposome-mediated gene transfer into the tubules. Exp. Nephrol. 5, 132–136. 15. Lai, L., Moeckel, G. W., and Lien, Y. H. (1997) Kidney-targeted liposome-mediated gene transfer in mice. Gene Ther. 4, 426–431. 16. Lai, L., Chan, D., Erickson, R. P., Hsu, S. J., and Lien, Y. H. (1998) Correction of renal tubular acidosis in carbonic anhydrase II deficient mice with gene therapy. J. Clin Invest. 101, 1320–1325 17. Boletta, A., Benigni, A., Lutz, J., Remuzzi, G., Soria, M. R., and Monaco, L. (1997) Nonviral gene delivery to the rat kidney by polyethylenimine. Hum. Gene Ther. 8, 1243–1251. 18. Oberbauer, R., Schreiner, G. F., and Meyer, T. W. (1995) Renal uptake of an 18mer phosphorothioate oligonucleotide. Kidney Int. 48, 1226–1232. 19. Carome, M. A., Kang, Y. H., Bohen, E. M., Nicholson, D. E., Carr, F. E., Kiandoli, L. C., et al. (1997) Distribution of the cellular uptake of phosphorothioate oligodeoxynucleotides in the rat kidney in vivo. Nephron 75, 82–87. 20. Oberbauer, R., Schreiner, G. F., Biber, J., Murer, H., and Meyer, T. W. (1996) In vivo suppression of the renal Na+/Pi cotransporter by antisense oligonucleotides. Proc. Natl. Acad. Sci. USA 93, 4903–4906. 21. Wang, Z. Q., Felder, R. A., and Carey, R. M. (1999) Selective inhibition of the renal dopamine subtype D1A receptor induces antinatriuresis in conscious rats. Hypertension 33, 504–510. 22. Noiri, E., Peresleni, T., Miller, F., and Goligorsky, M. S. (1996) In vivo targeting of inducible NO synthase with oligodeoxynucleotides protects rat kidney against ischemia. J. Clin. Invest. 97, 2377–2383. 23. Haller, H., Dragun, D., Miethke, A., Park, J. K., Weis, A., Lippoldt, A., Gross, V., and Luft, F.C. (1996) Antisense oligonucleotides for ICAM-1 attenuate reperfusion injury and renal failure in the rat. Kidney Int. 50, 473–480. 24. Dragun, D., Tullius, S. G., Park, J. K., Maasch, C., Lukitsch, I., Lippoldt, A., et al. (1998) ICAM-1 antisense oligodeoxynucleotides prevent reperfusion injury and enhance immediate graft function in renal transplantation. Kidney Int. 54, 590–602. 25. Dragun, D., Lukitsch, I., Tullius, S. G., Qun, Y., Park, J. K., Schneider, W., Luft, F. C., and Haller, H. (1998) Inhibition of intercellular adhesion molecule-1 with antisense deoxynucleotides prolongs renal isograft survival in the rat. Kidney Int. 54, 2113–2122. 26. Stepkowski, S. M., Wang, M. E., Condon, T. P., Cheng-Flournoy, S., Stecker, K., Graham, M., et al. (1998) Protection against allograft rejection with intercellular adhesion molecule-1 antisense oligodeoxynucleotides. Transplantation 66, 699–707. 27. Isaka, Y., Brees, D. K., Ikegaya, K., Kaneda, Y., Imai E., Noble, N. A., and Border, W. A. (1996) Gene therapy by skeletal muscle expression of decorin prevents fibrotic disease in rat kidney. Nature Med. 2, 418–423. 28. Isaka, Y., Akagi, Y., Ando, Y., Tsujie, M., Sudo, T., Ohno, N., et al. (1999) Gene therapy by transforming growth factor-beta receptor-IgG Fc chimera suppressed extracellular matrix accumulation in experimental glomerulonephritis. Kidney Int. 55, 465–475.

390

Lien and Lai

29. Isaka, Y., Akagi, Y., Ando, Y., and Imai, E. (1997) Application of gene therapy to diabetic nephropathy. Kidney Int. 52(Suppl. 60), S100–S103. 30. Lenkkeri, U., Mannikko, M., McCready, P., Lamerdin, J., Gribouval, O., Niaudet, P. M., et al. (1999) Structure of the gene for congenital nephrotic syndrome of the finnish type (NPHS1) and characterization of mutations. Am. J. Human Genet. 64, 51–61. 31. Scheiman, S. J., Guay-Woodford, L. M., Thakker, R. V., and Warnock, D. G. (1999) Genetic disorders of renal electrolyte transport. N. Engl. J. Med. 340, 1177–1187. 32. Lai, L. and Lien, Y. H. (1999) Homologous recombination-based gene therapy. Mini-review. Exp. Nephrol. 7, 11–14. 33. Lai, L., Chau, B., and Lien, Y. H. (1999) In vivo gene targeting in carbonic anhydrase II deficient mice by chimeric RNA/DNA oligonucleotides. Annual meeting of American Society of Gene Therapy, Washington, DC. Conference Proceedings, p. 236a. 34. Schwarting, A., Moore, K., Wada, T., Tesch, G., Yoon, H. J., and Kelley, V. R. (1998) IFN-gamma limits macrophage expansion in MRL-Fas(lpr) autoimmune interstitial nephritis: a negative regulatory pathway. J. Immunol. 160, 4074–4081. 35. Morishita, R., Gibbons, G. H., Horiuchi, M., Ellison, K. E., Nakama, M., Zhang, L., et al. (1995) A gene therapy strategy using a transcription factor decoy of the E2F binding site inhibits smooth muscle proliferation in vivo. Proc. Natl. Acad. Sci. USA 92, 5855–5859. 36. Kaneda, Y., Morishita, R., and Dzau, V. J. (1997) Prevention of restenosis by gene therapy. Ann. NY Acad. Sci. 811, 299–308. 37. Poston, R. S., Tran, K. P., Mann, M. J., Hoyt, E. G., Dzau, V. J., and Robbins, R. C. (1998) Prevention of ischemically induced neointimal hyperplasia using ex vivo antisense oligodeoxynucleotides. J. Heart Lung Transplant. 17, 349–355. 38. Isner, J. M. and Asahara, T. (1999) Angiogenesis and vasculogenesis as therapeutic strategies for postnatal neovascularization. J. Clin. Invest. 103, 1231–1236. 39. Lopez , J. J., Laham, R. J., Stamler, A., Pearlman, J. D., Bunting, S., Kaplan, A., et al. (1998) VEGF administration in chronic myocardial ischemia in pigs. Cardiovasc. Res. 40, 272–281. 40. Swenson, K. M., Ke, B., Wang, T., Markowitz, J. S., Maggard, M. A., Spear, G. S., et al. (1998) Fas ligand gene transfer to renal allografts in rats: effects on allograft survival. Transplantation 65, 155–160. 41. Olthoff, K. M., Judge, T. A., Gelman, A. E., da Shen, X., Hancock, W. W., Turka, L. A., and Shaked, A. (1998) Adenovirus-mediated gene transfer into cold-preserved liver allografts: survival pattern and unresponsiveness following transduction with CTLA4Ig. Nature Med. 4, 194–200. 42. Schwartz, D. A., Quinn, T. J., Thorne, P. S., Sayeed, S., Yi, A. K., and Krieg, A. M. (1997) CpG motifs in bacterial DNA cause inflammation in the lower respiratory tract. J. Clin. Invest. 100, 68–73. 43. Stewart, M. J., Plautz, G. E., Buono, L. D., Yang, Z. Y., Xu, L., Gao, X., et al. (1992) Gene transfer in vivo with DNA-liposome complexes: safety and acute toxicity in mice. Hum. Genet. Ther. 3, 267–275.

Renal Gene Therapy

391

44. Canonico, A. E., Plitman, J. D., Conary, J. T., Meyrick, B. O., and Brigham, K. L. (1994) No lung toxicity after repeated aerosol or intravenous delivery of plasmidcationic liposome complexes. J. Appl. Physiol. 77, 415–419. 45. Monteith, D. K., Horner, M. J., Gillett, N. A., Butler, M., Geary, R., Burckin, T., Ushiro-Watanabe, T., and Levin A. A. (1999) Evaluation of the renal effects of an antisense phosphorothioate oligodeoxynucleotide in monkeys. Toxicol. Pathol. 27, 307–317. 46. Henry, S. P., Taylor, J., Midgley, L., Levin, A. A., and Kornbrust, D. J. (1997) Evaluation of the toxicity of ISIS 2302, a phosphorothioate oligonucleotide, in a 4-week study in CD-1 mice. Antisense Nucleic Acid Drug Dev. 7, 473–481. 47. Brazeau, G. A., Attia, S., Poxon, S., and Hughes, J. A. (1998) In vitro myotoxicity of selected cationic macromolecules used in non-viral gene delivery. Pharm. Res. 15, 680–684. 48. Szebeni, J., Fontana, J. L., Wassef, N. M., Mongan, P. D., Morse, D. S., Dobbins, D. E., et al. (1999) Hemodynamic changes induced by liposomes and liposomeencapsulated hemoglobin in pigs: a model for pseudoallergic cardiopulmonary reactions to liposomes. Role of complement and inhibition by soluble CR1 and anti-C5a antibody. Circulation 99, 2302–2309.