- Email: [email protected]
Oligonucleotide lipoplexes: the influence of oligonucleotide composition on complexation
Biochimica et Biophysica Acta 1568 (2001) 177^182
www.bba-direct.com
Oligonucleotide lipoplexes: the in£uence of oligonucleotide composition on complexation Victor M. Meidan a , Judith Glezer a , Ninette Amariglio b , Jack S. Cohen Yechezkel Barenholz c
a;
*,
a
Advanced Technology Center, Chaim Sheba Medical Center, Tel Hashomer 52621, Israel Department of Hematology, Chaim Sheba Medical Center, Tel Hashomer 52621, Israel Laboratory of Membrane and Liposome Research, Department of Biochemistry, Hadassah Medical School, The Hebrew University, P.O. Box 12272, Jerusalem 91120, Israel b
c
Received 30 May 2001 ; received in revised form 11 September 2001; accepted 13 September 2001
Abstract Despite extensive investigations into oligonucleotide lipoplexes, virtually no work has addressed whether the physicochemical properties of these assemblies vary as a function of the constituent oligonucleotide (ODN) sequence and/or composition. The present study was aimed at answering this question. To this end, we complexed N-(1-(2,3-dioleoyloxy)propyl)-N,N,N-trimethylammonium chloride (DOTAP) liposomes, in dispersion, with either 18-mer phosphorothiote homo-oligonucleotides composed of either adenine, thymidine or cytosine; or one of three structurally related 18-mer phosphorothioate oligonucleotides (S-ODNs) (G3139, its reverse sequence and its two-base mismatch). After ODN addition to vesicles at different mole ratios, changes in pH and electrical surface potential at the lipid^water interface were analyzed by using the fluorophore heptadecyl-7-hydroxycoumarin while particle size distributions were analyzed by staticlight scattering. The results indicate that each homo-oligonucleotide does indeed exhibit different complexation behavior. In particular, the maximal level of DOTAP neutralization by the polyadenine S-ODN is much lower than that for the two other homo-oligonucleotides and hence its lipoplex is much more positively charged. Much smaller electrostatic differences are also apparent between lipoplexes formed from each of the G3139-related ODNs. This paper identifies nucleotide base selection and sequence as a variable that can affect the physicochemical properties of oligonucleotide lipoplexes and hence probably their transfection competency. ß 2001 Elsevier Science B.V. All rights reserved. Keywords : Base sequence; Cationic liposome ; Lipoplex; Oligonucleotide ; 4-Heptadecyl-7-hydroxycoumarin ; Fluorescence
1. Introduction Antisense oligonucleotides (ODNs) are now extensively employed in various therapeutic applications as well as in elucidating genetic modulation. However, ODN clinical use is still frequently limited by the poor penetration of the nucleic acids into the cytoplasm and nucleus of cells [1,2]. One promising clinical delivery strategy involves us-
Abbreviations : DOTAP, N-(1-(2,3-dioleoyloxy)propyl)-N,N,N-trimethylammonium chloride; HEPES, N-(2-hydroxyethyl)piperazine-NP(2-ethanesulfonic acid); HC, heptadecyl-7-hydroxycoumarin; LUV, large unilamellar vesicle ; MLV, multilamellar vesicle; ODN, oligonucleotide ; SLS, static-light scattering ; S-ODN, phosphorothioate oligonucleotide * Corresponding author. Fax: +972-3-530-3146. E-mail address : [email protected] (J.S. Cohen).
ing ODN^cationic lipid complexes, termed lipoplexes, in which the negatively charged nucleotides bind electrostatically to a cationic lipid such as DOTAP. Such lipoplexes, which form spontaneously when ODNs are mixed with cationic liposomes [3], have been shown to facilitate enhanced ODN transfection both in vitro [4,5] and in vivo [6]. As the ¢rst phase in designing the most transfectioncompetent complex, many physicochemical characterization studies have been performed on lipoplexes in recent years. In particular, the physicochemical properties of these assemblies have been quanti¢ed as a function of the; nucleotide to cationic lipid charge ratio [3,7^12], lipid composition [3,7^9,13,14], complexation mode [12,13], the presence of serum components [15,16] and ionic strength of the medium [10,11,17]. However, virtually no work has addressed whether the physicochemical properties of a lipoplex are dependent upon its ODN sequence and/or composition.
0304-4165 / 01 / $ ^ see front matter ß 2001 Elsevier Science B.V. All rights reserved. PII: S 0 3 0 4 - 4 1 6 5 ( 0 1 ) 0 0 2 1 6 - 1
BBAGEN 25254 19-12-01
178
V.M. Meidan et al. / Biochimica et Biophysica Acta 1568 (2001) 177^182
The aim of the present study was to shed light on this issue by investigating whether ODNs composed of di¡erent sequences of nucleotide bases subsequently formed lipoplexes exhibiting di¡erent properties. To this end, we complexed cationic LUVs, in aqueous dispersion, with 18mer phosphorothioate homo-oligonucleotides composed of either adenine, thymidine or cytosine and monitored the associated electrostatic and particle size changes. We did not examine polyguanine oligonucleotides as these are insoluble in aqueous solution. We also similarly compared complexes formed from an 18-mer antisense S-ODN termed G3139, and those formed from its two-base mismatch and reverse sequence control. We employed DOTAP throughout since this is one of the most frequently used cationic lipids [18]. We used S-ODNs since these are preferable to phosphodiesters for antisense applications due to their greater chemical and biological stability [19]. 2. Materials and methods 2.1. Materials The lipid 1,2-dioleoyl-3-trimethylammonium-propane (DOTAP) was purchased from Avanti Polar Lipids (Albaster, AL). HC was purchased from Molecular Probes (Eugene, OR). HEPES sodium salt was obtained from Sigma Chemical Co. (St Louis, MO). The DNA synthesis reagents were acquired from Glen Research (Sterling, VA) while the anhydrous solvents were obtained from Biolab Laboratories (Jerusalem, Israel). All these chemicals were of analytical grade or better. All solutions were prepared in water that had been de-ionized (Barnstead Easypure LF compact ultrapure water system). 2.2. Oligonucleotides The bcl-2 phosphorothioate antisense 18-mer G3139 (S-d-5P-(TCT CCC AGC GTG CGC CAT)), its reverse sequence control G3622 (S-d-5P-(TAC CGC GTG CGA CCC TCT)) and its two-base mismatch G4126 (S-d-5P(TCT CCC AGC ATG TGC CAT)) were obtained from Genta, Lexington, MA. ODN purity was assessed by HPLC (Merck Hitachi D-7000) incorporating an anion exchange column. A gradient of ammonium acetate in 50% isopropanol in a pH 8 bu¡er was employed. For each ODN, a single sharp peak was obtained, thus con¢rming that our supplies of both G3139 and G3622 were indeed pure. The ODNs S-d-5P-[A18 ], S-d-5P-[C18 ], S-d-5P-[T18 ] were synthesized by the phosphoramidite method on a 1-Wmol scale on commercial solid supports using an automated DNA/RNA synthesizer (Applied Biosystems, 392ABI). Phosphorothioate linkages were generated by using 0.05 M [3 H]1,2-benzodithiole-3-one 1,1-dioxide in acetonitrile. The ODNs were analyzed and isolated by HPLC (Merck
Hitachi D-7000) incorporating an anion exchange column. A gradient of ammonium acetate in 50% isopropanol in a pH 8 bu¡er was employed. The isopropanol was removed by evaporation and the solution was desalted by using cartridges (Poly-Pack RP, Glen Research). Following further evaporation, the ODNs were isolated. All of the ODNs were dissolved in 20 mM HEPES bu¡er (pH 7.4). ODN concentrations were determined by optical density measurements (Shimadzu UV-24101PC, Duisburg, Germany) made at 254 nm [20]. 2.3. Liposome preparation Large unilamellar vesicles (LUVs) labeled with HC (HC/DOTAP 1:400 mol/mol) were prepared as described by [21]. DOTAP powder was dissolved in tert-butanol to which HC in tert-butanol was added. The mixture was centrifuged under vacuum for 3 h in an Automatic Environmental Speedvac, Model AE2010 (Savant Instruments, Holbrook, NY). The residue was hydrated with 20 mM HEPES bu¡er (pH 7.4) to produce a multilamellar vesicle dispersion exhibiting a total lipid concentration of 31 mM. These vesicles were downsized and converted to LUVs exhibiting a diameter of approximately 100 nm by the use of a `Liposofast' extrusion system [22]. The LUVs were stored at 4³C. 2.4. HC titration experiments The aim of these experiments was to study the e¡ect of bulk pH on the £uorescence of HC-labeled DOTAP liposomes. To this end, 3.87-Wl aliquots, each containing 0.12 Wmol of HC-labeled DOTAP LUVs, were diluted in vials with 3 ml of 20 mM HEPES bu¡er (pH 7.4). The pH of each medium was adjusted to a desired value in the pH 3^ 9 range by the addition of an appropriate amount of concentrated hydrochloric acid. To abolish any intraliposome to medium pH gradient, the samples were sonicated for 5 s in an ultrasonic water bath (Branson B-12 ultrasonics). Fluorescence characterization was performed at ambient temperature on a spectrometer (Shimadzu RF 5301PC). Fluorescence of HC was measured at two excitation wavelengths. These were 330 nm (excitation at the pH-independent isosbestic wavelength of HC which quanti¢es the total amount of HC in the lipidic assembly) and 380 nm (excitation wavelength of the nonprotonated charged HC), using a single emission wavelength of 450 nm (excitation and emission bandwidths of 3 nm). For more details see Section 3.1 and [21]. 2.5. Fluorescence and static-light scattering (turbidity) experiments An aliquot of HC-labeled liposome dispersion, pre-prepared in 20 mM HEPES bu¡er (pH 7.4), was diluted in a cuvette with 3 ml of the same bu¡er, to a ¢nal DOTAP
BBAGEN 25254 19-12-01
V.M. Meidan et al. / Biochimica et Biophysica Acta 1568 (2001) 177^182
179
concentration of 4U1035 M. An appropriate volume of ODN was added such that the ODN3/L+ mole ratio reached either 0, 0.1, 0.2, 0.5, 1, 1.5 or 2. Both the £uorescence and turbidity of each ODN^lipid sample was recorded just before the addition of ODN as well as 17 min after ODN addition. Fluorescence of HC was measured as described above. Static-light scattering (SLS) at 90³ was measured concurrently on the above spectrophotometer, using both an excitation and an emission wavelength of 600 nm (excitation and emission bandwidths of 1.5 nm). Mixing of the cuvette contents was attained by pipetting; immediately after liposome addition, immediately after ODN addition, 8 min after ODN addition, as well as 16 min after ODN addition. All these measurements were performed at ambient temperature. Each individual experiment was performed in triplicate. 3. Results
Fig. 2. 380/330 Fluorescence ratio of homo-oligonucleotide lipoplexes as a function of the oligonucleotide to cationic lipid mole ratio. Each error bar represents a standard deviation of n = 3 experiments.
3.1. HC titration curve The £uorophore of the pH-sensitive, £uorescent probe HC is the hydroxycoumarin moiety, which is a weak acid [23]. When incorporated into liposomes or lipoplexes, the £uorophore resides at the water^lipid interface of the lipid assemblies. The molecule has the key property that its £uorescence intensity at the excitation wavelength of 380 nm is proportional to the amount of charged HC present while its £uorescence intensity at the excitation wavelength of 330 nm is proportional to the total amount of £uorophore present. Since aqueous HC does not contribute to £uorescence, the 380 nm to 330 nm £uorescence ratio essentially re£ects the extent of HC disassociation at the water^lipid interface of the lipid assemblies. Fig. 1 presents a graph relating the £uorescence ratio of HC in DOTAP LUVs as a function of bulk pH. This type of sigmoidal pro¢le has been obtained previously [21,24] and
Fig. 1. 380/330 Fluorescence ratio of HC-labeled (380/330 HC £uorescence ratio) DOTAP liposomes as a function of bulk pH.
was shown to ¢t a modi¢ed Henderson^Hasselbach equation [25]. We can use this curve to convert our £uorescence ratio measurements to values of local pH at the liposome or lipoplex water^lipid interface. Such pH values can then be converted into corresponding electrical surface potential values by using the Boltzmann equation [21]. 3.2. Homo-oligonucleotide data The graph in Fig. 2 plots the £uorescence ratio as a function of the ODN to cationic lipid mole ratio for HC-labeled lipoplexes. It can be seen that in the absence of oligonucleotide, in 20 mM HEPES bu¡er (pH 7.4), the water^lipid interface exhibits a 380/330 £uorescence ratio of 3.5. From Fig. 1, it can be determined that this means its pH is equivalent to the bulk pH of 7.4. The addition of all three types of homo-oligonucleotides to the cationic vesicles produced a decrease in the disassociation degree of HC. This indicates that the S-ODNs neutralized the cationic lipid, inducing a concomitant reduction in pH at the water^lipid interface of the assemblies. For all three SODNs, the curves representing HC disassociation versus ODN to cationic lipid mole ratio resembled inverted sigmoids and this type of pro¢le can be predicted and calculated from theory by the Gouy^Chapman approximation [21]. Of considerable interest is the fact that there were crucial di¡erences between the curves described by each homo-oligonucleotide. It can be seen that at [ODN3]/ [L+]v1.5, the HC 380/330 £uorescence ratio was signi¢cantly higher in the A18 system (V1.5) than in the C18 (V0.6) or T18 system (V0.4). This indicates that the
BBAGEN 25254 19-12-01
180
V.M. Meidan et al. / Biochimica et Biophysica Acta 1568 (2001) 177^182
maximal level of DOTAP neutralization by the polyadenine S-ODN is much lower (V29%) than that for the two other homo-oligonucleotides (V54%) and hence the A18 / DOTAP lipoplex at [ODN3]/[L+]v1.5 is much more positively charged. In terms of pH at the lipid^water interface at maximal neutralization (saturation), the surface pH value was V6.2 for polyadenine lipoplexes and V5.2 for both polythymidine and polycytosine lipoplexes. It is noteworthy that at [ODN3]/[L+] = 1.5, the £uorescence ratio value exhibited by the polycytosine system was slightly yet signi¢cantly higher than that exhibited by the polythymidine system (P = 0.033, t-test). However, upon the addition of more ODN such that [ODN3]/ [L+] = 2, this di¡erence had become insigni¢cant (P = 0.106, t-test). The 90³ SLS re£ects the sum of all the changes in number of particles, size of particles and variations in the refractive index. It has been demonstrated that for assessing lipoplex size distributions, 90³ SLS is a more accurate measurement than the otherwise more commonly-used photon correlation spectroscopy, also termed dynamic light-scattering [24,26]. It can be seen from Fig. 3 that for the polyadenine complexes, SLS peaked when the mole ratio of ODN to cationic lipid was 0.5 and that at either side of this value, the SLS was much lower. The addition of C18 produced a di¡erent pattern in that SLS reached a plateau through the ODN to cationic lipid mole ratio range of 0.5^2. Furthermore, the absolute levels of speci¢c SLS (SLS per mole DOTAP) at this plateau were signi¢cantly lower than the peak level observed for A18 . When T18 was added to the liposomes, the derived pro¢le was somewhat intermediate to that observed for the A18 and C18 systems in that SLS slightly peaked at a value of
Fig. 3. 90³ Static-light scattering (SLS) of homo-oligonucleotide lipoplexes as a function of oligonucleotide to cationic lipid mole ratio. Each error bar represents a standard deviation of n = 3 experiments.
Fig. 4. 380/330 HC £uorescence ratio of G3139 and G3622 lipoplexes as a function of the oligonucleotide to cationic lipid mole ratio. Each error bar represents a standard deviation of n = 3 experiments.
0.5 but the speci¢c SLS values were close to those observed for C18 . 3.3. G3139 data Fig. 4 plots the variation in £uorescence 380/330 ratio for HC-labeled complexes as a function of increasing amounts of G3139 or reverse sequence control. In comparing the two ODN neutralization curves, it can be seen they are similar and that they both resemble inverted sigmoids
Fig. 5. 380/330 HC £uorescence ratio of G3139 and G4126 lipoplexes as a function of the oligonucleotide to cationic lipid mole ratio. Each error bar represents a standard deviation of n = 3 experiments.
BBAGEN 25254 19-12-01
V.M. Meidan et al. / Biochimica et Biophysica Acta 1568 (2001) 177^182
Fig. 6. 90³ Static-light scattering (SLS) of G3139-related lipoplexes as a function of oligonucleotide to cationic lipid mole ratio. Each error bar represents a standard deviation of n = 3 experiments.
as theoretically predicted by the Gouy^Chapman approximation. However, at [ODN3]/[L+] = 1.5, the lipid^water interface of the G3139 lipoplex exhibits a slight but signi¢cantly lower pH than the lipid^water interface of the reverse sequence lipoplex (P = 0.045, t-test). This result demonstrates that reversing the nucleotide sequence can a¡ect complexation. The graph in Fig. 5 presents the same 380/330 £uorescence ratio data but compares G3139 with its two base mismatch G4126. Again, the neutralization curves are sigmoidal and similar. Nevertheless, at one mole ratio value, [ODN3]/[L+] = 1, the lipid^water interface of the G3139 lipoplex exhibits a signi¢cantly lower pH than the lipid^ water interface of G4126 (P = 0.021, t-test). This indicates that even the substitution of two bases on an S-ODN can subtly alter its complexation with cationic lipid. Fig. 6 shows the turbidity data as a function of mole ratio for all three G3139-related ODNs. It can be seen that the SLS pro¢les in all three systems were similar with peak levels occurring at an ODN3/DOTAP+ mole (charge) ratio value of 0.5. 4. Discussion The electrostatics of lipoplex formation were followed through changes in surface potential and pH at the lipid^ water interface upon complexation. The most marked result was that at excess amounts of S-ODN [ODN3]/ [L+] v 1.5, A18 complexes exhibited a much more positively charged lipid^water interface (V170 mV) than C18 or T18 complexes (V110 mV). This indicates that when A18 was `crowded' (in terms of a high ODN3/DOTAP+ ratio) on the DOTAP surface, the A18 phosphate groups could not reach and/or neutralize the positive charges of the DOTAP as readily as the other two homo-oligonucleotides. A possible explanation for this observation is that
181
the approach of the A18 phosphates towards the cationic lipid was hindered by the bulkier purine rings to a greater extent than the approach of the C18 or T18 phosphates was by the smaller pyrimidine rings. There was also a small but signi¢cant di¡erence between the neutralization curves of C18 and T18 . This may be due to the fact that cytosine has an extra primary amino (positively charged) group attached to C4 (missing in thymidine) which may reduce the interaction between the pyrimidine negatively charged phosphate and the DOTAP quaternary amine, therefore lowering neutralization. An additional explanation stems from the fact that cytosine-rich ODNs, but not any other ODNs, readily form tetrameric structures in solution. Each such tetramer, also termed the i-motif, is characterized by two base-paired parallel-stranded duplexes associating together [27]. Such an association was shown to progressively develop in cytosine-rich ODNs as the pH was lowered below pH8 [28]. In our studies, performed at pH 7.4, T18 molecules existed as single strands but C18 molecules probably existed at least partially in the form of tetramers. Such variations may explain the di¡erences between neutralization curves. We also followed the complexation of the homo-oligonucleotides by SLS measurements. As we previously showed [3], SLS in lipoplex systems generally peaks at intermediate mole ratios [ODN3]/[L+]W0.5. This e¡ect occurs because when ODN partially neutralizes cationic lipids, lateral phase separation develops and this causes the formation of membrane defects in the upper part of the bilayer acyl chains. The defects cause the lipoplexes to exhibit increased size instability, which manifests as increased light scattering. In contrast, at ODN excess, there are less or no defects and the turbidity increase is smaller. There were di¡erences in turbidity pro¢les between each of the three homo-oligonucleotides and these were probably due to the structural and steric factors described above. All our data show that nucleotide composition a¡ects the ODN^DOTAP complexation process. The G3139 oligomer is complementary to the ¢rst six codons of the openreading frame of bcl-2 and has biological activity, and is being tested for clinical application [29]. Indeed, our results from the G3139-related ODN experiments indicate that even reversing the base sequence or substituting two bases will alter the pH at the lipid^water interface of the lipid assembly at a certain mole ratio. Such physicochemical changes in complex properties are likely to change the transfection e¤cacy of lipoplexes. Such a view ¢ts in with the ¢ndings published by Conrad and co-workers [30] who investigated whether molecular modi¢cations to ODNs a¡ected in vitro lipofection rates into avian embryonic cardiomyocytes by lipoplexes. Two S-ODNs that di¡ered in their nucleotide sequence exhibited di¡erent transfection rates. The group also examined 15 di¡erent cationic lipids and found that the best lipid for lipofection in this cell line was not the same for all ODNs but depended upon ODN composition.
BBAGEN 25254 19-12-01
182
V.M. Meidan et al. / Biochimica et Biophysica Acta 1568 (2001) 177^182
5. Conclusions The data presented here indicate that nucleotide base selection and sequence are parameters that can a¡ect the electrostatics and extent of aggregation that develop during oligonucleotide^cationic lipid complexation. Both electrostatics and aggregation state are important parameters in the lipofection process [12,18,31]. Hence, it follows that ODN composition can almost certainly a¡ect transfection e¤cacy from lipoplexes. Clearly, further work is required in order to elucidate more precisely how modi¢cations in nucleotide sequence in£uence both the electrostatics and macrostructure of ODN lipoplexes, as well as their transfection competency. Acknowledgements This study was supported in part by Israel Science Foundation Grant ISF 30/98-16.1 to J.S.C. and Y.B. We would like to thank Genta Inc. for the supplies of G3139, G3622 and G4126. References [1] R.L. Juliano, S. Alahari, H. Yoo, R. Kole, M. Cho, Pharmacol. Res. 16 (1999) 494^502. [2] S. Akhtar, M.D. Hughes, A. Khan, M. Bibby, M. Hussain, Q. Nawaz, J. Double, P. Sayyed, Adv. Drug Deliv. Rev. 44 (2000) 3^ 21. [3] V.M. Meidan, J.S. Cohen, N. Amariglio, D. Hirsch-Lerner, Y. Barenholz, Biochim. Biophys. Acta 1464 (2000) 251^261. [4] M. Rodriguez, V. Noe, C. Alemany, A. Miralles, V. Bemi, I. Caragol, C.J. Ciudad, Int. J. Cancer 81 (1999) 785^792. [5] K.N. Chi, A.E. Wallis, C.H. Lee, D. Lopez de Menezes, J. Sartor, W.H. Dragowska, L.D. Mayer, Breast Cancer Res. Treat. 63 (2000) 199^212. [6] Y. Clare Zhang, B. Kimura, L. Shen, M.I. Phillips, Hypertension 35 (2000) 219^224. [7] J.C. Birchall, I.W. Kellaway, S.N. Mills, Int. J. Pharm. 183 (1999) 195^207.
[8] S.J. Eastman, C. Siegel, J. Tousignant, A.E. Smith, S.H. Cheng, R.K. Scheule, Biochim. Biophys. Acta 1325 (1997) 41^62. [9] P.C. Ross, S.W. Hui, Gene Ther. 6 (1999) 651^659. [10] J. Turek, C. Dubertret, G. Jaslin, K. Antonakis, D. Scherman, B. Pitard, J. Gene Med. 2 (2000) 32^40. [11] E.K. Wasan, P. Harvie, K. Edwards, G. Karlsson, M.B. Bally, Biochim. Biophys. Acta 1461 (1999) 27^46. [12] N.J. Zuidam, D. Hirsch-Lerner, S. Marguiles, Y. Barenholz, Biochim. Biophys. Acta 1419 (1999) 207^220. [13] M.E. Ferrari, D. Rusalov, J. Enas, C.J. Wheeler, Nucleic Acids Res. 29 (2001) 1539^1548. [14] A.E. Regelin, S. Fankhaenel, L. Gurtesch, C. Prinz, G. von Kiedrowski, U. Massing, Biochim. Biophys. Acta 1464 (2000) 151^164. [15] O. Zelphati, L.S. Uyechi, L.G. Barron, F.C. Szoka, Biochim. Biophys. Acta 1390 (1998) 119^133. [16] S. Audouy, G. Molema, L. de Leij, D. Hoekstra, J. Gene Med. 2 (2000) 465^476. [17] I. Blank, M. Bueno Da Costa, J. Bolard, M. Saint-Pierre Chazalet, Biochim. Biophys. Acta 1464 (2000) 299^308. [18] D.D. Lasic, Genosomes (DNA^lipid complexes), in: D.D. Lasic (Ed.), Liposomes in Gene Delivery, CRC Press, Boca Raton, FL, 1997. [19] J.S. Cohen, Adv. Pharmacol. 24 (1994) 319^339. [20] T. Brown, D.J.S. Brown, Modern machine-aided methods of oligonucleotide synthesis, in: F. Eckstein (Ed.), Oligonucleotides and Analogues, A Practical Approach, Oxford University Press, New York, NY, 1991, p. 20. [21] N.J. Zuidam, Y. Barenholz, Biochim. Biophys. Acta 1329 (1997) 211^ 222. [22] R.C. MacDonald, R.I. Macdonald, B.P.M. Menco, K. Takeshita, N.K. Subbarao, L. Hu, Biochim. Biophys. Acta 1061 (1991) 297^303. [23] V. Borenstain, Y. Barenholz, Chem. Phys. Lipids 64 (1993) 117^127. [24] N.J. Zuidam, Y. Barenholz, Biochim. Biophys. Acta 1419 (1998) 207^ 220. [25] J.E. Whitaker, R.P. Haugland, D. Ryan, P.C. Hewitt, R.P. Haugland, F.G. Prendergast, Anal. Biochem. 207 (1992) 267^279. [26] D. Hirsch-Lerner, Y. Barenholz, Biochim. Biophys. Acta 1370 (1998) 17^30. [27] K. Gehring, J.L. Leroy, M. Gueron, Nature 363 (1993) 561^565. [28] G. Manzini, N. Yathindra, L.E. Xodo, Nucleic Acids Res. 22 (1998) 4634^4640. [29] B. Jansen, V. Wacheck, E. Heere-Ress, H. Schlagbauer-Wadl, C. Hoeller, T. Lucas, M. Hoermann, U. Hollenstein, K. Wol¡, H. Pehamberger, Lancet 356 (2000) 1728^1733. [30] A.H. Conrad, M.A. Behlke, T. Ja¡redo, G.W. Conrad, Antisense Nucleic Acid Drug Dev. 8 (1998) 427^434. [31] M. Kerner, D. Hirsch-Lerner, Y. Barenholz, Biochim. Biophys. Acta (2001) in press.
BBAGEN 25254 19-12-01
www.bba-direct.com
Oligonucleotide lipoplexes: the in£uence of oligonucleotide composition on complexation Victor M. Meidan a , Judith Glezer a , Ninette Amariglio b , Jack S. Cohen Yechezkel Barenholz c
a;
*,
a
Advanced Technology Center, Chaim Sheba Medical Center, Tel Hashomer 52621, Israel Department of Hematology, Chaim Sheba Medical Center, Tel Hashomer 52621, Israel Laboratory of Membrane and Liposome Research, Department of Biochemistry, Hadassah Medical School, The Hebrew University, P.O. Box 12272, Jerusalem 91120, Israel b
c
Received 30 May 2001 ; received in revised form 11 September 2001; accepted 13 September 2001
Abstract Despite extensive investigations into oligonucleotide lipoplexes, virtually no work has addressed whether the physicochemical properties of these assemblies vary as a function of the constituent oligonucleotide (ODN) sequence and/or composition. The present study was aimed at answering this question. To this end, we complexed N-(1-(2,3-dioleoyloxy)propyl)-N,N,N-trimethylammonium chloride (DOTAP) liposomes, in dispersion, with either 18-mer phosphorothiote homo-oligonucleotides composed of either adenine, thymidine or cytosine; or one of three structurally related 18-mer phosphorothioate oligonucleotides (S-ODNs) (G3139, its reverse sequence and its two-base mismatch). After ODN addition to vesicles at different mole ratios, changes in pH and electrical surface potential at the lipid^water interface were analyzed by using the fluorophore heptadecyl-7-hydroxycoumarin while particle size distributions were analyzed by staticlight scattering. The results indicate that each homo-oligonucleotide does indeed exhibit different complexation behavior. In particular, the maximal level of DOTAP neutralization by the polyadenine S-ODN is much lower than that for the two other homo-oligonucleotides and hence its lipoplex is much more positively charged. Much smaller electrostatic differences are also apparent between lipoplexes formed from each of the G3139-related ODNs. This paper identifies nucleotide base selection and sequence as a variable that can affect the physicochemical properties of oligonucleotide lipoplexes and hence probably their transfection competency. ß 2001 Elsevier Science B.V. All rights reserved. Keywords : Base sequence; Cationic liposome ; Lipoplex; Oligonucleotide ; 4-Heptadecyl-7-hydroxycoumarin ; Fluorescence
1. Introduction Antisense oligonucleotides (ODNs) are now extensively employed in various therapeutic applications as well as in elucidating genetic modulation. However, ODN clinical use is still frequently limited by the poor penetration of the nucleic acids into the cytoplasm and nucleus of cells [1,2]. One promising clinical delivery strategy involves us-
Abbreviations : DOTAP, N-(1-(2,3-dioleoyloxy)propyl)-N,N,N-trimethylammonium chloride; HEPES, N-(2-hydroxyethyl)piperazine-NP(2-ethanesulfonic acid); HC, heptadecyl-7-hydroxycoumarin; LUV, large unilamellar vesicle ; MLV, multilamellar vesicle; ODN, oligonucleotide ; SLS, static-light scattering ; S-ODN, phosphorothioate oligonucleotide * Corresponding author. Fax: +972-3-530-3146. E-mail address : [email protected] (J.S. Cohen).
ing ODN^cationic lipid complexes, termed lipoplexes, in which the negatively charged nucleotides bind electrostatically to a cationic lipid such as DOTAP. Such lipoplexes, which form spontaneously when ODNs are mixed with cationic liposomes [3], have been shown to facilitate enhanced ODN transfection both in vitro [4,5] and in vivo [6]. As the ¢rst phase in designing the most transfectioncompetent complex, many physicochemical characterization studies have been performed on lipoplexes in recent years. In particular, the physicochemical properties of these assemblies have been quanti¢ed as a function of the; nucleotide to cationic lipid charge ratio [3,7^12], lipid composition [3,7^9,13,14], complexation mode [12,13], the presence of serum components [15,16] and ionic strength of the medium [10,11,17]. However, virtually no work has addressed whether the physicochemical properties of a lipoplex are dependent upon its ODN sequence and/or composition.
0304-4165 / 01 / $ ^ see front matter ß 2001 Elsevier Science B.V. All rights reserved. PII: S 0 3 0 4 - 4 1 6 5 ( 0 1 ) 0 0 2 1 6 - 1
BBAGEN 25254 19-12-01
178
V.M. Meidan et al. / Biochimica et Biophysica Acta 1568 (2001) 177^182
The aim of the present study was to shed light on this issue by investigating whether ODNs composed of di¡erent sequences of nucleotide bases subsequently formed lipoplexes exhibiting di¡erent properties. To this end, we complexed cationic LUVs, in aqueous dispersion, with 18mer phosphorothioate homo-oligonucleotides composed of either adenine, thymidine or cytosine and monitored the associated electrostatic and particle size changes. We did not examine polyguanine oligonucleotides as these are insoluble in aqueous solution. We also similarly compared complexes formed from an 18-mer antisense S-ODN termed G3139, and those formed from its two-base mismatch and reverse sequence control. We employed DOTAP throughout since this is one of the most frequently used cationic lipids [18]. We used S-ODNs since these are preferable to phosphodiesters for antisense applications due to their greater chemical and biological stability [19]. 2. Materials and methods 2.1. Materials The lipid 1,2-dioleoyl-3-trimethylammonium-propane (DOTAP) was purchased from Avanti Polar Lipids (Albaster, AL). HC was purchased from Molecular Probes (Eugene, OR). HEPES sodium salt was obtained from Sigma Chemical Co. (St Louis, MO). The DNA synthesis reagents were acquired from Glen Research (Sterling, VA) while the anhydrous solvents were obtained from Biolab Laboratories (Jerusalem, Israel). All these chemicals were of analytical grade or better. All solutions were prepared in water that had been de-ionized (Barnstead Easypure LF compact ultrapure water system). 2.2. Oligonucleotides The bcl-2 phosphorothioate antisense 18-mer G3139 (S-d-5P-(TCT CCC AGC GTG CGC CAT)), its reverse sequence control G3622 (S-d-5P-(TAC CGC GTG CGA CCC TCT)) and its two-base mismatch G4126 (S-d-5P(TCT CCC AGC ATG TGC CAT)) were obtained from Genta, Lexington, MA. ODN purity was assessed by HPLC (Merck Hitachi D-7000) incorporating an anion exchange column. A gradient of ammonium acetate in 50% isopropanol in a pH 8 bu¡er was employed. For each ODN, a single sharp peak was obtained, thus con¢rming that our supplies of both G3139 and G3622 were indeed pure. The ODNs S-d-5P-[A18 ], S-d-5P-[C18 ], S-d-5P-[T18 ] were synthesized by the phosphoramidite method on a 1-Wmol scale on commercial solid supports using an automated DNA/RNA synthesizer (Applied Biosystems, 392ABI). Phosphorothioate linkages were generated by using 0.05 M [3 H]1,2-benzodithiole-3-one 1,1-dioxide in acetonitrile. The ODNs were analyzed and isolated by HPLC (Merck
Hitachi D-7000) incorporating an anion exchange column. A gradient of ammonium acetate in 50% isopropanol in a pH 8 bu¡er was employed. The isopropanol was removed by evaporation and the solution was desalted by using cartridges (Poly-Pack RP, Glen Research). Following further evaporation, the ODNs were isolated. All of the ODNs were dissolved in 20 mM HEPES bu¡er (pH 7.4). ODN concentrations were determined by optical density measurements (Shimadzu UV-24101PC, Duisburg, Germany) made at 254 nm [20]. 2.3. Liposome preparation Large unilamellar vesicles (LUVs) labeled with HC (HC/DOTAP 1:400 mol/mol) were prepared as described by [21]. DOTAP powder was dissolved in tert-butanol to which HC in tert-butanol was added. The mixture was centrifuged under vacuum for 3 h in an Automatic Environmental Speedvac, Model AE2010 (Savant Instruments, Holbrook, NY). The residue was hydrated with 20 mM HEPES bu¡er (pH 7.4) to produce a multilamellar vesicle dispersion exhibiting a total lipid concentration of 31 mM. These vesicles were downsized and converted to LUVs exhibiting a diameter of approximately 100 nm by the use of a `Liposofast' extrusion system [22]. The LUVs were stored at 4³C. 2.4. HC titration experiments The aim of these experiments was to study the e¡ect of bulk pH on the £uorescence of HC-labeled DOTAP liposomes. To this end, 3.87-Wl aliquots, each containing 0.12 Wmol of HC-labeled DOTAP LUVs, were diluted in vials with 3 ml of 20 mM HEPES bu¡er (pH 7.4). The pH of each medium was adjusted to a desired value in the pH 3^ 9 range by the addition of an appropriate amount of concentrated hydrochloric acid. To abolish any intraliposome to medium pH gradient, the samples were sonicated for 5 s in an ultrasonic water bath (Branson B-12 ultrasonics). Fluorescence characterization was performed at ambient temperature on a spectrometer (Shimadzu RF 5301PC). Fluorescence of HC was measured at two excitation wavelengths. These were 330 nm (excitation at the pH-independent isosbestic wavelength of HC which quanti¢es the total amount of HC in the lipidic assembly) and 380 nm (excitation wavelength of the nonprotonated charged HC), using a single emission wavelength of 450 nm (excitation and emission bandwidths of 3 nm). For more details see Section 3.1 and [21]. 2.5. Fluorescence and static-light scattering (turbidity) experiments An aliquot of HC-labeled liposome dispersion, pre-prepared in 20 mM HEPES bu¡er (pH 7.4), was diluted in a cuvette with 3 ml of the same bu¡er, to a ¢nal DOTAP
BBAGEN 25254 19-12-01
V.M. Meidan et al. / Biochimica et Biophysica Acta 1568 (2001) 177^182
179
concentration of 4U1035 M. An appropriate volume of ODN was added such that the ODN3/L+ mole ratio reached either 0, 0.1, 0.2, 0.5, 1, 1.5 or 2. Both the £uorescence and turbidity of each ODN^lipid sample was recorded just before the addition of ODN as well as 17 min after ODN addition. Fluorescence of HC was measured as described above. Static-light scattering (SLS) at 90³ was measured concurrently on the above spectrophotometer, using both an excitation and an emission wavelength of 600 nm (excitation and emission bandwidths of 1.5 nm). Mixing of the cuvette contents was attained by pipetting; immediately after liposome addition, immediately after ODN addition, 8 min after ODN addition, as well as 16 min after ODN addition. All these measurements were performed at ambient temperature. Each individual experiment was performed in triplicate. 3. Results
Fig. 2. 380/330 Fluorescence ratio of homo-oligonucleotide lipoplexes as a function of the oligonucleotide to cationic lipid mole ratio. Each error bar represents a standard deviation of n = 3 experiments.
3.1. HC titration curve The £uorophore of the pH-sensitive, £uorescent probe HC is the hydroxycoumarin moiety, which is a weak acid [23]. When incorporated into liposomes or lipoplexes, the £uorophore resides at the water^lipid interface of the lipid assemblies. The molecule has the key property that its £uorescence intensity at the excitation wavelength of 380 nm is proportional to the amount of charged HC present while its £uorescence intensity at the excitation wavelength of 330 nm is proportional to the total amount of £uorophore present. Since aqueous HC does not contribute to £uorescence, the 380 nm to 330 nm £uorescence ratio essentially re£ects the extent of HC disassociation at the water^lipid interface of the lipid assemblies. Fig. 1 presents a graph relating the £uorescence ratio of HC in DOTAP LUVs as a function of bulk pH. This type of sigmoidal pro¢le has been obtained previously [21,24] and
Fig. 1. 380/330 Fluorescence ratio of HC-labeled (380/330 HC £uorescence ratio) DOTAP liposomes as a function of bulk pH.
was shown to ¢t a modi¢ed Henderson^Hasselbach equation [25]. We can use this curve to convert our £uorescence ratio measurements to values of local pH at the liposome or lipoplex water^lipid interface. Such pH values can then be converted into corresponding electrical surface potential values by using the Boltzmann equation [21]. 3.2. Homo-oligonucleotide data The graph in Fig. 2 plots the £uorescence ratio as a function of the ODN to cationic lipid mole ratio for HC-labeled lipoplexes. It can be seen that in the absence of oligonucleotide, in 20 mM HEPES bu¡er (pH 7.4), the water^lipid interface exhibits a 380/330 £uorescence ratio of 3.5. From Fig. 1, it can be determined that this means its pH is equivalent to the bulk pH of 7.4. The addition of all three types of homo-oligonucleotides to the cationic vesicles produced a decrease in the disassociation degree of HC. This indicates that the S-ODNs neutralized the cationic lipid, inducing a concomitant reduction in pH at the water^lipid interface of the assemblies. For all three SODNs, the curves representing HC disassociation versus ODN to cationic lipid mole ratio resembled inverted sigmoids and this type of pro¢le can be predicted and calculated from theory by the Gouy^Chapman approximation [21]. Of considerable interest is the fact that there were crucial di¡erences between the curves described by each homo-oligonucleotide. It can be seen that at [ODN3]/ [L+]v1.5, the HC 380/330 £uorescence ratio was signi¢cantly higher in the A18 system (V1.5) than in the C18 (V0.6) or T18 system (V0.4). This indicates that the
BBAGEN 25254 19-12-01
180
V.M. Meidan et al. / Biochimica et Biophysica Acta 1568 (2001) 177^182
maximal level of DOTAP neutralization by the polyadenine S-ODN is much lower (V29%) than that for the two other homo-oligonucleotides (V54%) and hence the A18 / DOTAP lipoplex at [ODN3]/[L+]v1.5 is much more positively charged. In terms of pH at the lipid^water interface at maximal neutralization (saturation), the surface pH value was V6.2 for polyadenine lipoplexes and V5.2 for both polythymidine and polycytosine lipoplexes. It is noteworthy that at [ODN3]/[L+] = 1.5, the £uorescence ratio value exhibited by the polycytosine system was slightly yet signi¢cantly higher than that exhibited by the polythymidine system (P = 0.033, t-test). However, upon the addition of more ODN such that [ODN3]/ [L+] = 2, this di¡erence had become insigni¢cant (P = 0.106, t-test). The 90³ SLS re£ects the sum of all the changes in number of particles, size of particles and variations in the refractive index. It has been demonstrated that for assessing lipoplex size distributions, 90³ SLS is a more accurate measurement than the otherwise more commonly-used photon correlation spectroscopy, also termed dynamic light-scattering [24,26]. It can be seen from Fig. 3 that for the polyadenine complexes, SLS peaked when the mole ratio of ODN to cationic lipid was 0.5 and that at either side of this value, the SLS was much lower. The addition of C18 produced a di¡erent pattern in that SLS reached a plateau through the ODN to cationic lipid mole ratio range of 0.5^2. Furthermore, the absolute levels of speci¢c SLS (SLS per mole DOTAP) at this plateau were signi¢cantly lower than the peak level observed for A18 . When T18 was added to the liposomes, the derived pro¢le was somewhat intermediate to that observed for the A18 and C18 systems in that SLS slightly peaked at a value of
Fig. 3. 90³ Static-light scattering (SLS) of homo-oligonucleotide lipoplexes as a function of oligonucleotide to cationic lipid mole ratio. Each error bar represents a standard deviation of n = 3 experiments.
Fig. 4. 380/330 HC £uorescence ratio of G3139 and G3622 lipoplexes as a function of the oligonucleotide to cationic lipid mole ratio. Each error bar represents a standard deviation of n = 3 experiments.
0.5 but the speci¢c SLS values were close to those observed for C18 . 3.3. G3139 data Fig. 4 plots the variation in £uorescence 380/330 ratio for HC-labeled complexes as a function of increasing amounts of G3139 or reverse sequence control. In comparing the two ODN neutralization curves, it can be seen they are similar and that they both resemble inverted sigmoids
Fig. 5. 380/330 HC £uorescence ratio of G3139 and G4126 lipoplexes as a function of the oligonucleotide to cationic lipid mole ratio. Each error bar represents a standard deviation of n = 3 experiments.
BBAGEN 25254 19-12-01
V.M. Meidan et al. / Biochimica et Biophysica Acta 1568 (2001) 177^182
Fig. 6. 90³ Static-light scattering (SLS) of G3139-related lipoplexes as a function of oligonucleotide to cationic lipid mole ratio. Each error bar represents a standard deviation of n = 3 experiments.
as theoretically predicted by the Gouy^Chapman approximation. However, at [ODN3]/[L+] = 1.5, the lipid^water interface of the G3139 lipoplex exhibits a slight but signi¢cantly lower pH than the lipid^water interface of the reverse sequence lipoplex (P = 0.045, t-test). This result demonstrates that reversing the nucleotide sequence can a¡ect complexation. The graph in Fig. 5 presents the same 380/330 £uorescence ratio data but compares G3139 with its two base mismatch G4126. Again, the neutralization curves are sigmoidal and similar. Nevertheless, at one mole ratio value, [ODN3]/[L+] = 1, the lipid^water interface of the G3139 lipoplex exhibits a signi¢cantly lower pH than the lipid^ water interface of G4126 (P = 0.021, t-test). This indicates that even the substitution of two bases on an S-ODN can subtly alter its complexation with cationic lipid. Fig. 6 shows the turbidity data as a function of mole ratio for all three G3139-related ODNs. It can be seen that the SLS pro¢les in all three systems were similar with peak levels occurring at an ODN3/DOTAP+ mole (charge) ratio value of 0.5. 4. Discussion The electrostatics of lipoplex formation were followed through changes in surface potential and pH at the lipid^ water interface upon complexation. The most marked result was that at excess amounts of S-ODN [ODN3]/ [L+] v 1.5, A18 complexes exhibited a much more positively charged lipid^water interface (V170 mV) than C18 or T18 complexes (V110 mV). This indicates that when A18 was `crowded' (in terms of a high ODN3/DOTAP+ ratio) on the DOTAP surface, the A18 phosphate groups could not reach and/or neutralize the positive charges of the DOTAP as readily as the other two homo-oligonucleotides. A possible explanation for this observation is that
181
the approach of the A18 phosphates towards the cationic lipid was hindered by the bulkier purine rings to a greater extent than the approach of the C18 or T18 phosphates was by the smaller pyrimidine rings. There was also a small but signi¢cant di¡erence between the neutralization curves of C18 and T18 . This may be due to the fact that cytosine has an extra primary amino (positively charged) group attached to C4 (missing in thymidine) which may reduce the interaction between the pyrimidine negatively charged phosphate and the DOTAP quaternary amine, therefore lowering neutralization. An additional explanation stems from the fact that cytosine-rich ODNs, but not any other ODNs, readily form tetrameric structures in solution. Each such tetramer, also termed the i-motif, is characterized by two base-paired parallel-stranded duplexes associating together [27]. Such an association was shown to progressively develop in cytosine-rich ODNs as the pH was lowered below pH8 [28]. In our studies, performed at pH 7.4, T18 molecules existed as single strands but C18 molecules probably existed at least partially in the form of tetramers. Such variations may explain the di¡erences between neutralization curves. We also followed the complexation of the homo-oligonucleotides by SLS measurements. As we previously showed [3], SLS in lipoplex systems generally peaks at intermediate mole ratios [ODN3]/[L+]W0.5. This e¡ect occurs because when ODN partially neutralizes cationic lipids, lateral phase separation develops and this causes the formation of membrane defects in the upper part of the bilayer acyl chains. The defects cause the lipoplexes to exhibit increased size instability, which manifests as increased light scattering. In contrast, at ODN excess, there are less or no defects and the turbidity increase is smaller. There were di¡erences in turbidity pro¢les between each of the three homo-oligonucleotides and these were probably due to the structural and steric factors described above. All our data show that nucleotide composition a¡ects the ODN^DOTAP complexation process. The G3139 oligomer is complementary to the ¢rst six codons of the openreading frame of bcl-2 and has biological activity, and is being tested for clinical application [29]. Indeed, our results from the G3139-related ODN experiments indicate that even reversing the base sequence or substituting two bases will alter the pH at the lipid^water interface of the lipid assembly at a certain mole ratio. Such physicochemical changes in complex properties are likely to change the transfection e¤cacy of lipoplexes. Such a view ¢ts in with the ¢ndings published by Conrad and co-workers [30] who investigated whether molecular modi¢cations to ODNs a¡ected in vitro lipofection rates into avian embryonic cardiomyocytes by lipoplexes. Two S-ODNs that di¡ered in their nucleotide sequence exhibited di¡erent transfection rates. The group also examined 15 di¡erent cationic lipids and found that the best lipid for lipofection in this cell line was not the same for all ODNs but depended upon ODN composition.
BBAGEN 25254 19-12-01
182
V.M. Meidan et al. / Biochimica et Biophysica Acta 1568 (2001) 177^182
5. Conclusions The data presented here indicate that nucleotide base selection and sequence are parameters that can a¡ect the electrostatics and extent of aggregation that develop during oligonucleotide^cationic lipid complexation. Both electrostatics and aggregation state are important parameters in the lipofection process [12,18,31]. Hence, it follows that ODN composition can almost certainly a¡ect transfection e¤cacy from lipoplexes. Clearly, further work is required in order to elucidate more precisely how modi¢cations in nucleotide sequence in£uence both the electrostatics and macrostructure of ODN lipoplexes, as well as their transfection competency. Acknowledgements This study was supported in part by Israel Science Foundation Grant ISF 30/98-16.1 to J.S.C. and Y.B. We would like to thank Genta Inc. for the supplies of G3139, G3622 and G4126. References [1] R.L. Juliano, S. Alahari, H. Yoo, R. Kole, M. Cho, Pharmacol. Res. 16 (1999) 494^502. [2] S. Akhtar, M.D. Hughes, A. Khan, M. Bibby, M. Hussain, Q. Nawaz, J. Double, P. Sayyed, Adv. Drug Deliv. Rev. 44 (2000) 3^ 21. [3] V.M. Meidan, J.S. Cohen, N. Amariglio, D. Hirsch-Lerner, Y. Barenholz, Biochim. Biophys. Acta 1464 (2000) 251^261. [4] M. Rodriguez, V. Noe, C. Alemany, A. Miralles, V. Bemi, I. Caragol, C.J. Ciudad, Int. J. Cancer 81 (1999) 785^792. [5] K.N. Chi, A.E. Wallis, C.H. Lee, D. Lopez de Menezes, J. Sartor, W.H. Dragowska, L.D. Mayer, Breast Cancer Res. Treat. 63 (2000) 199^212. [6] Y. Clare Zhang, B. Kimura, L. Shen, M.I. Phillips, Hypertension 35 (2000) 219^224. [7] J.C. Birchall, I.W. Kellaway, S.N. Mills, Int. J. Pharm. 183 (1999) 195^207.
[8] S.J. Eastman, C. Siegel, J. Tousignant, A.E. Smith, S.H. Cheng, R.K. Scheule, Biochim. Biophys. Acta 1325 (1997) 41^62. [9] P.C. Ross, S.W. Hui, Gene Ther. 6 (1999) 651^659. [10] J. Turek, C. Dubertret, G. Jaslin, K. Antonakis, D. Scherman, B. Pitard, J. Gene Med. 2 (2000) 32^40. [11] E.K. Wasan, P. Harvie, K. Edwards, G. Karlsson, M.B. Bally, Biochim. Biophys. Acta 1461 (1999) 27^46. [12] N.J. Zuidam, D. Hirsch-Lerner, S. Marguiles, Y. Barenholz, Biochim. Biophys. Acta 1419 (1999) 207^220. [13] M.E. Ferrari, D. Rusalov, J. Enas, C.J. Wheeler, Nucleic Acids Res. 29 (2001) 1539^1548. [14] A.E. Regelin, S. Fankhaenel, L. Gurtesch, C. Prinz, G. von Kiedrowski, U. Massing, Biochim. Biophys. Acta 1464 (2000) 151^164. [15] O. Zelphati, L.S. Uyechi, L.G. Barron, F.C. Szoka, Biochim. Biophys. Acta 1390 (1998) 119^133. [16] S. Audouy, G. Molema, L. de Leij, D. Hoekstra, J. Gene Med. 2 (2000) 465^476. [17] I. Blank, M. Bueno Da Costa, J. Bolard, M. Saint-Pierre Chazalet, Biochim. Biophys. Acta 1464 (2000) 299^308. [18] D.D. Lasic, Genosomes (DNA^lipid complexes), in: D.D. Lasic (Ed.), Liposomes in Gene Delivery, CRC Press, Boca Raton, FL, 1997. [19] J.S. Cohen, Adv. Pharmacol. 24 (1994) 319^339. [20] T. Brown, D.J.S. Brown, Modern machine-aided methods of oligonucleotide synthesis, in: F. Eckstein (Ed.), Oligonucleotides and Analogues, A Practical Approach, Oxford University Press, New York, NY, 1991, p. 20. [21] N.J. Zuidam, Y. Barenholz, Biochim. Biophys. Acta 1329 (1997) 211^ 222. [22] R.C. MacDonald, R.I. Macdonald, B.P.M. Menco, K. Takeshita, N.K. Subbarao, L. Hu, Biochim. Biophys. Acta 1061 (1991) 297^303. [23] V. Borenstain, Y. Barenholz, Chem. Phys. Lipids 64 (1993) 117^127. [24] N.J. Zuidam, Y. Barenholz, Biochim. Biophys. Acta 1419 (1998) 207^ 220. [25] J.E. Whitaker, R.P. Haugland, D. Ryan, P.C. Hewitt, R.P. Haugland, F.G. Prendergast, Anal. Biochem. 207 (1992) 267^279. [26] D. Hirsch-Lerner, Y. Barenholz, Biochim. Biophys. Acta 1370 (1998) 17^30. [27] K. Gehring, J.L. Leroy, M. Gueron, Nature 363 (1993) 561^565. [28] G. Manzini, N. Yathindra, L.E. Xodo, Nucleic Acids Res. 22 (1998) 4634^4640. [29] B. Jansen, V. Wacheck, E. Heere-Ress, H. Schlagbauer-Wadl, C. Hoeller, T. Lucas, M. Hoermann, U. Hollenstein, K. Wol¡, H. Pehamberger, Lancet 356 (2000) 1728^1733. [30] A.H. Conrad, M.A. Behlke, T. Ja¡redo, G.W. Conrad, Antisense Nucleic Acid Drug Dev. 8 (1998) 427^434. [31] M. Kerner, D. Hirsch-Lerner, Y. Barenholz, Biochim. Biophys. Acta (2001) in press.
BBAGEN 25254 19-12-01