Skip to main content
NCBI home page
Nucleic Acids Research logoLink to Nucleic Acids Research
. 2002 Apr 15;30(8):1808–1816. doi: 10.1093/nar/30.8.1808

Synergy between cationic lipid and co-lipid determines the macroscopic structure and transfection activity of lipoplexes

Marilyn E Ferrari 1, Denis Rusalov 1, Joel Enas 1, Carl J Wheeler 1,a
PMCID: PMC113211  PMID: 11937635

Abstract

The large number of cytofectin and co-lipid combinations currently used for lipoplex-mediated gene delivery reflects the fact that the optimal cytofectin/co-lipid combination varies with the application. The effects of structural changes in both cytofectin and co-lipid were systematically examined to identify structure–activity relationships. Specifically, alkyl chain length, degree of unsaturation and the head group to which the alkyl side chain was attached were examined to determine their effect on lipoplex structure and biological activity. The macroscopic lipoplex structure was assessed using a dye-binding assay and the biological activity was examined using in vitro transfection in three diverse cell lines. Lipoplexes were formulated in three different vehicles currently in use for in vivo delivery of naked plasmid DNA (pDNA) and lipoplex formulations. The changes in dye accessibility were consistent with structural changes in the lipoplex, which correlated with alterations in the formulation. In contrast, transfection activity of different lipoplexes was cell type and vehicle dependent and did not correlate with dye accessibility. Overall, the results show a correlation between transfection and enhanced membrane fluidity in both the lipoplex and cellular membranes.

INTRODUCTION

Lipoplexes form when liposomes containing positively charged cationic lipids (cytofectins) are mixed with negatively charged polynucleic acids. Lipoplexes have been widely used as a safe non-viral method of delivery of nucleic acids both in vitro and in vivo and transgenes have been delivered to a variety of tissues. There have also been several human clinical trials using lipoplexes (1–6). Lipoplexes have been used to enhance immune responses against several plasmid DNA (pDNA) encoded antigens (7–14). The large number of cytofectins currently in use for lipoplex-mediated gene delivery reflects the fact that the optimal cytofectin/co-lipid combination usually varies with the application. For example, (±)-N-(2-hydroxyethyl)-N,N-dimethyl-2,3-bis(tetradecyloxy)-1-propanaminium bromide (DMRIE):1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE) has been shown to be very effective for in vivo transfection of a variety of tumor types, while (±)-N-(3-aminopropyl)-N,N-dimethyl-2,3-bis(dodecyloxy)-1-propanaminium bromide (GAP-DLRIE):DOPE is better for functional gene delivery to the lung following intra-nasal administration (15) and to salivary epithelial cells (16). Vaxfectin, a new cytofectin formulation, has recently shown significant enhancement of humoral immune responses against pDNA encoded antigens compared with naked pDNA (7). Vaxfectin is composed of (±)-N-(3-aminopropyl)-N,N-dimethyl-2,3-bis(cis-9-tetradecenyloxy)-1-propanaminium bromide (GAP-DMORIE), structurally related to DMRIE, and the co-lipid 1,2-diphytanoyl-sn-glycero-3-phosphoethanolamine (DPyPE), a phosphatidylethanolamine lipid composed of polyisoprenoid alkyl chains, in a 1:1 ratio.

The basis for the observed tissue and application specificity of lipoplex-mediated gene delivery is complex and not well understood. Multiple factors influence the delivery process and the physical structure of the lipoplex, as well as its interaction with the specific tissue environment, are both important (17,18). The physical nature of lipoplexes is highly dependent on the formulation procedure and components. Each cytofectin, co-lipid and vehicle combination results in unique liposome properties and lipoplex structure.

The purpose of this work was to extend a previous study in which the effects of cytofectin head group, vehicle constituent and preparation technique on the pDNA binding properties and in vitro transfection of lipoplexes were examined. In the current study, the alkyl side chains of both the cytofectin and co-lipid were varied, either individually or in combination, and certain cytofectin/co-lipid-related structure–activity relationships were identified.

MATERIALS AND METHODS

Reagents

All chemicals were USP grade. All solutions were prepared using sterile water for injection (Baxter Health Care, Deerfield, IL) and sterile filtered through 0.2 µm nylon chamber filters (Nalgene, Rochester, NY).

Plasmid DNA

The plasmid VR1412 used for these studies contains the cytoplasmic β-galactosidase gene and was constructed as previously described (19) and purified using EndoFree Giga columns from Qiagen (Valencia, CA) according to the manufacturer’s instructions.

Cytofectin components

Synthesis of the (±)-N-(3-aminopropyl)-N,N-dimethyl-2,3-bis(alkyloxy)-1-propanaminium bromides. Racemic 1-dimethylamino-2,3-propanediol (0.96 g) (Janssen Chimica, Geel, Belgium) was converted to the disodium salt in situ by treatment with sodium hydride (60% in oil, 0.8 g) in tetrahydrofuran (70 ml). Condensation with the appropriate alkyl methane sulfonate (2.2 equiv.) (NuChek Prep, Elysian, MN) afforded the crude (±)-N,N-dimethyl-2,3-bis(alkyl)propylamine. This material was purified to homogeneity by filtration through celite followed by silica gel chromatography employing a step gradient of ether in hexane (from 10 to 50%) and, finally, neat ether as the eluents. The structure and purity of the product were confirmed by 1H NMR and IR spectroscopy.

The N,N-dimethyl-2,3-bis(alkyloxy)-1-propanamine thus prepared (2.4 g) was then treated with N-(3-bromopropyl)phthalimide (2 equiv.) in dimethylformamide (15 ml) at 105°C to effect quaternization of the amine. Removal of the dimethylformamide in vacuo followed by silica gel chromatography using chloroform/methanol/aqueous ammonia (90/10/0.5) as the eluent yielded the (±)-N-(3-phthalimido) compound. The purified (±)-N-(3-phthalimido) intermediate (2.1 g) was treated with anhydrous hydrazine (20 equiv.) in absolute ethanol (40 ml) to effect deprotection of the primary amine. Filtration, evaporative removal of the ethanol and basic extraction afforded the (±)-N-(3-aminopropyl) derivatives. The saturated derivatives were recrystallized from hexane to afford white solids. The structure and purity of the cationic lipids were confirmed by 1H NMR, IR spectroscopy and TLC.

Co-lipids

The following 1,2-diacyl-sn-glycero-3-phosphoethanolamine co-lipids were used in this study: dilauroyl (12:0), dimyristoyl (14:0), dipalmitoyl (16:0), diphytanoyl (16:0[(CH3)4]), distearoyl (18:0), dioleoyl (C18:1, DOPE). All co-lipids were purchased as chloroform solutions from Avanti Polar Lipids (Alabaster, AL). The co-lipids were of >99% purity and were used without further purification.

Cytofectin/co-lipid mixtures were prepared by mixing chloroform solutions of cationic lipid and co-lipid. Dried films were prepared in 2 ml sterile glass vials by evaporating the chloroform under a stream of argon and placing the vials under vacuum overnight to remove solvent traces. Each lipid film contained 1.5 µmol each cytofectin and co-lipid. Figure 1 shows the structures of the cytofectins (Fig. 1A) and co-lipids (Fig. 1B) used in this study.

Figure 1.

Figure 1

Figure 1

Open in a new tab

(A) Structures of the cytofectins used in this study (from top to bottom): GAP-DLRIE (C12:0); GAP-DLORIE (C12:1); GAP-DMRIE (C14:0); GAP-DMORIE (C14:1); GAP-DPRIE (C16:0); GAP-DPyRIE (16:0); GAP-DSRIE (C18:0); GAP-DORIE (C18:1). (B) Structures of co-lipids used in this study (top to bottom): DLPE (C12:0); DMPE (C14:0); DPPE (C16:0); DPyPE (C16:0); DSPE (C18:0); DOPE (C18:1).

Lipoplex formulation

Liposomes were prepared by adding 1 ml of SWFI (sterile water for injection, VWR International) to a vial containing 1.5 µmol cytofectin:co-lipid (1:1), followed by continuous vortexing for 5 min on the highest setting of a Genie Vortex Mixer (Fisher Scientific). The resulting liposome solution contained 1.5 mM cytofectin.

Liposomes and pDNA were prepared at twice the final formulation concentration. Lipoplexes were formed by adding an equal volume of liposomes in SWFI to pDNA in twice the final vehicle concentration (e.g. 20 mM sodium phosphate, 1.8% NaCl for 2× PBS) using a syringe and a 28 gauge needle. Liposomes were added in a steady stream, followed by brief, gentle vortexing (a few seconds on setting 4 of a Fisher Genie vortex mixer).

Formulations were prepared at final cytofectin/pDNA(phosphate) molar ratios of 1:8, 1:6, 1:4 and 1:2 and a final pDNA concentration of 0.5 mg/ml in PBS pH 7.2, 150 mM sodium phosphate pH 7.2 or SWFI. Lipoplexes prepared at a 1:4 cytofectin/pDNA molar ratio composed of different cytofectin/co-lipid combinations were examined for in vitro transfection in three different cell lines. Lipoplexes were prepared in sterile water (SWFI), PBS or 150 mM sodium phosphate, pH 7.2.

The molar concentration of pDNA phosphate was calculated by dividing the pDNA concentration (in mg/ml) by 330, the average nucleotide molecular mass.

Accessible pDNA (PicoGreen)

The accessible pDNA in lipoplex formulations was determined using the PicoGreen reagent (Molecular Probes, Eugene, OR) as previously described (20). The PicoGreen fluorescence signal was low and unstable in the absence of added salt. Therefore, for formulations prepared in SWFI, assays were performed in 10 mM sodium phosphate pH 7.2. The percentage of accessible pDNA was calculated by normalizing the fluorescence signal of the formulation to the corresponding pDNA control: (Fformulation × 100)/Fcontrol.

Absorbance assay for total pDNA recovery

In formulations where aggregation was observed, the total pDNA was determined from the A260 of the formulation in the presence of Zwittergent. Failure to account for the loss of pDNA resulting from aggregation will lead to an underestimate of the accessible pDNA in lipoplex formulations. Formulations prepared in SWFI generally resulted in the highest degree of aggregation, followed by PBS. Aggregation was typically observed only in the 4:1 and 2:1 pDNA molar ratios and was rarely observed when 150 mM sodium phosphate was used as the formulation vehicle. Forty microliters of pDNA formulation or control at 0.5 mg/ml were added to 960 µl of 2% Zwittergent (Zwittergent 3-14; Calbiochem, La Jolla, CA) in PBS, followed by heating at 50°C, then cooling to room temperature (30 min for each step). Zwittergent disrupts the complex, allowing release of DNA from the cationic liposomes and elimination of light scattering in the 240–350 nm wavelength range. Total pDNA concentration can then be determined by measuring the absorbance at 260 nm (21). Samples were transferred to quartz cuvettes and scanned from 240 to 360 nm in 0.2 nm increments using a Shimadzu Biospec 1601 spectrophotometer. The DNA concentration in mg/ml is determined using the conversion 1 OD260 = 50 µg/ml (22). The accessible pDNA obtained from the PicoGreen assay on aggregated formulations was corrected as follows: accessible pDNA (%) = [(Fformulation × 100)/Fcontrol]/corrected total pDNA where corrected total pDNA = (A260, formulation × 100)/A260, DNA control.

Transfection in vitro

C2C12 (mouse muscle myoblasts, CRL 1772), BHK-21 (Syrian hamster kidney, C-13) and CPAE (bovine pulmonary artery endothelium, CCL-209) cell lines were purchased from American Type Tissue Culture Collection. On day 0, 1 × 104 cells (100 µl) were plated into each well of a 96-well plate (Nunc no. 167008) and incubated overnight at 37°C. Lipoplexes prepared at a cytofectin/pDNA molar ratio of 1:4 were used in transfection. On day 1, serially diluted lipoplexes were further diluted with an equal volume of serum-free OptiMEM cell culture medium, reducing formulation vehicle concentration by half (e.g. lipoplexes formulated in PBS pH 7.2 are diluted to 0.5× PBS and 50% OptiMEM). Cell culture medium was aspirated and the cells were transfected with 100 µl/well diluted lipoplexes. Cells were incubated for 1.5 h at 37°C, followed by addition of 50 µl/well OptiMEM supplemented with 30% fetal calf serum. On day 2 (24 ± 0.5 h post-transfection), cells received 100 µl/well OptiMEM supplemented with 10% fetal calf serum. On day 3 (48 ± 2 h post-transfection), cell culture medium was removed by aspiration and 50 µl of lysis buffer (0.1% Triton X-100 in 250 mM Tris, pH 8.0) was added to each well. Plates were frozen at –80°C for at least 2 h and assayed for β-galactosidase. Expression levels were reported as the percentage of the β-galactosidase level obtained for cells transfected with Vaxfectin/pDNA lipoplexes formulated in PBS.

RESULTS

It is well known that lipoplex-mediated gene delivery needs to be optimized for different tissues and applications. The alkyl side chains of both cytofectin and co-lipid are key determinants of lipoplex structure and biological activity. The approach taken in this study was to maintain common features in the lipid head groups, but vary the side chains of the cytofectin and co-lipid, either individually or in combination. Cytofectins in the propanaminium class containing a γ-amine substituent (Fig. 1A) and phosphoethanolamine type co-lipids (Fig. 1B) were juxtaposed in the study. Lipoplexes were prepared in sterile water (SWFI), PBS or 150 mM sodium phosphate, pH 7.2, three commonly used vehicles for lipoplex delivery.

Alterations in lipoplex structure resulting from changes in cytofectin/co-lipid pairs were examined by determining the accessibility of pDNA to the PicoGreen dye (20,21). In vitro transfection experiments were performed in three cell lines to determine whether alkyl side chain alterations resulted in cell type-specific effects on biological activity and to allow a determination of whether pDNA accessibility, an indirect method of assessing macroscopic structure, correlated with biological activity.

Effect of co-lipid on GAP-DMORIE and GAP-DLRIE lipoplexes

Given the activity of GAP-DMORIE (7) and GAP-DLRIE in vivo (15,16), it was of interest to determine the common as well as the distinctive properties of the lipoplexes which result when pDNA was mixed with these cationic lipids. Both of these cytofectins are predicted to produce relatively fluid bilayer liposome structures due to the short alkyl side chain of GAP-DLRIE (C12:0) and the mono-unsaturated C14 side chain of GAP-DMORIE (C14:1). In addition, the co-lipid used in the lipoplex formulations is also important, and thus structural differences in either or both lipid components may effect biological activity.

The following co-lipids were examined alone or in combination with GAP-DMORIE and GAP-DLRIE: 1,2-dilauroyl-sn-glycero-3-phosphoethanolamine (DLPE) (C12:0), 1,2-dimyristoyl-sn-glycero-3-phosphoethanolamine (DMPE) (C14:0), 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine (DPPE) (C16:0), DPyPE (C16:0, branched) and DOPE (C18:1). No systematic trends were observed in the accessible pDNA with changes in co-lipid for either GAP-DLRIE or GAP-DMORIE lipoplexes (data not shown). Lower accessible pDNA was observed when formulations were prepared in SWFI or PBS, compared with sodium phosphate, indicative of tighter cytofectin–pDNA association in the lipoplex. pDNA accessibility varied linearly with cytofectin/pDNA molar ratio and was lower in formulations prepared in SWFI compared with PBS, as anticipated. Similarily, lipoplexes prepared in 150 mM sodium phosphate consistently resulted in ≥90% accessible pDNA, also consistent with previous results (20).

The results of the in vitro transfection experiments are shown in Figure 2. Lipoplexes prepared with GAP-DMORIE generally resulted in higher transfection levels than GAP-DLRIE formulations, with the optimal co-lipid depending on vehicle and cell line. For example, when lipoplexes were formulated in 150 mM sodium phosphate the highest expression in BHK cells was observed with GAP-DMORIE:DPPE, while GAP-DMORIE:DPyPE resulted in the highest level of transfection in C2C12 cells. BHK cells transfected using PBS were the only instance where GAP-DLRIE resulted in higher expression than GAP-DMORIE.

Figure 2.

Figure 2

Open in a new tab

The effect of co-lipid on β-galatosidase expression in three cell lines. Lipoplexes were prepared at a 0.25 cytofectin/pDNA molar ratio with GAP-DLRIE (C12:0) (A–C) or GAP-DMORIE (C14:1) (D–F) in the absence of co-lipids (dotted bars) or in the presence of the following co-lipids: DLPE (C12:0), horizontal striped bars; DMPE (C14:0), herringbone bars; DPPE (C16:0), scalloped bars; DPyPE (C16:0, branched), diamond bars; DOPE (C18:1), vertical striped bars. In vitro transfection experiments were performed as described in Materials and Methods. Error bars represent the standard deviation of triplicate measurements.

Effect of cytofectin and co-lipid containing side chains of equal length

Experiments were also performed using a series of cytofectins and co-lipids with saturated alkyl chains to map the effects of chain length. Lipoplexes were prepared with GAP-DLRIE:DLPE (C12:0/C12:0), (±)-N-(3-aminopropyl)-N,N-dimethyl-2,3-bis(tetradecyloxy)-1-propanaminium bromide (GAP-DMRIE):DMPE (C14:0/C14:0) and (±)-N-(3-aminopropyl)-N,N-dimethyl-2,3-bis(hexadecyloxy)-1-propanaminium bromide (GAP-DPRIE):DPPE (C16:0/C16:0) in three vehicles.

The pDNA accessibility for lipoplexes prepared with the matched cytofectin/co-lipid combinations shows a clear dependence on alkyl chain length (Fig. 3). The accessible pDNA in lipoplexes formulated in PBS or SWFI increases with increasing chain length, consistent with the overall trends previously observed (20). However, the effect of increasing chain length was greater when both cytofectin and co-lipid contained C16 alkyl side chains; >80% of the pDNA was accessible regardless of the formulation vehicle, which was not observed when DOPE (C18:1) was used as the co-lipid with GAP-DPRIE (data not shown). The results suggest that the pDNA may be more weakly associated with the cytofectins when the alkyl side chains are matched in length.

Figure 3.

Figure 3

Open in a new tab

The effect of matching the cytofectin and co-lipid alkyl side chain length on the accessible pDNA at different cytofectin/pDNA molar ratios. Lipoplexes were prepared with: (A) GAP-DLRIE:DLPE (C12:0/C12:0); (B) GAP-DMRIE:DMPE (C14:0/14:0); (C) GAP-DPRIE:DPPE (C16:0/C16:0). The lipoplexes were prepared in PBS (diamonds); water (squares); 150 mM sodium phosphate (triangles). pDNA accessibility was determined using PicoGreen assays as described in Materials and Methods. Error bars represent the standard deviation of triplicate measurements.

Figure 4 shows that lipoplexes prepared in PBS generally resulted in the highest levels of transfection, independent of the lipid components. Preparation in sodium phosphate resulted in negligible transfection levels in BHK and CPAE cell lines and low levels in C2C12 cells. A clear dependence of alkyl side chain length was observed in C2C12 and CPAE cell lines when PBS was used as the formulation vehicle, with β-galactosidase expression levels decreasing with increasing chain length. The results show that the effect of chain length is dependent on the formulation vehicle and cell type.

Figure 4.

Figure 4

Open in a new tab

The effect of matching the cytofectin and co-lipid alkyl side chain length on β-galactosidase expression in three cell lines: (A) BHK cells; (B) C2C12 cells; (C) CPAE cells. Lipoplexes were prepared at a 0.25 cytofectin/pDNA molar ratio with: GAP-DLRIE:DLPE (C12:0/C12:0), solid bars; GAP-DMRIE:DMPRE (C14:0/C14:0), dotted bars; GAP-DPRIE:DPPE (C16:0/C16:0), vertical striped bars. In vitro transfection experiments were performed as described in Materials and Methods. Error bars represent the standard deviation of triplicate measurements.

Effect of head group and alkyl side chain juxtaposition

The effects of exchanging the C12 and C18 alkyl side chain between cytofectins and co-lipids were examined to determine if a relationship exists between head group and alkyl chain juxtaposition. The choice of the chain lengths and substitution pattern was motivated by the in vivo results showing the biological activity of GAP-DLRIE:DOPE in transfecting lung and salivary tissue (15,16). Accessible pDNA in lipoplexes containing C18:0 alkyl side chains prepared in SWFI and PBS was more dependent on the cytofectin/pDNA molar ratio than when C18:1 side chains were used, suggesting that pDNA is more weakly bound in lipoplexes containing the C18:1 (DOPE) side chain (data not shown). The results in Figure 5 show that GAP-DLRIE:DOPE (C12:0/C18:1) and (±)-N-(3-aminopropyl)-N,N-dimethyl-2,3-bis(cis-9-octadecenyloxy)-1-propanaminium bromide (GAP-DORIE):DLPE (C18:1/C12:0) were the only combinations that resulted in significant transfection levels, regardless of the cell line. Thus, for the combinations examined, a short, saturated side chain and a long, unsaturated side chain resulted in the highest transfection levels.

Figure 5.

Figure 5

Open in a new tab

The effect of head group and alkyl side chain juxtapostion on β-galactosidase expression in three cell lines: (A) BHK cells; (B) C2C12 cells; (C) CPAE cells. Lipoplexes were prepared at a 0.25 cytofectin/pDNA molar ratio with cytofectins containing C12 alkyl side chains and co-lipids containing C18 alkyl side chains. GAP-DLRIE:DSPE (C12:0/C18:0), dark solid bars; GAP-DLRIE:DOPE (C12:0/C18:1), vertical striped bars; GAP-DLORIE:DOPE (C12:1/C18:1), diagonal striped bars; GAP-DLORIE:DSPE (C12:1/C18:1), light solid bars (no detectable β-galactosidase expression); GAP-DORIE:DLPE (18:1/C12:0), dotted bars. In vitro transfection experiments were performed as described in Materials and Methods. Error bars represent the standard deviation of triplicate measurements.

Based on the ability of Vaxfectin/pDNA formulations to enhance effective antibody responses against a variety of pDNA encoded antigens, it was of interest to address whether incorporating the diphytanoyl alkyl side chain into the cytofectin versus the co-lipid affected in vitro transection and pDNA accessibility. The following combinations were examined: (±)-N-(3-aminopropyl)-N,N-dimethyl-2,3-bis(3,7,11,15-tetramethylhexadecyloxy)-1-propanaminium bromide (GAP-DPyRIE):DLPE, GAP-DLRIE:DPyPE and GAP-DMORIE:DPyPE, and no differences in accessible pDNA were observed for any of the combinations examined (data not shown). In all but two instances, GAP-DMORIE:DPyPE (Vaxfectin) produced the highest level of transfection. Thus, it appears that in most cases the diphytanoyl side chain affords the best activity when incorporated into the co-lipid.

DISCUSSION

The results of the current study support the notion that lipoplex structure and transfection activity strongly depend on the specific cytofectin/co-lipid combination comprising the liposomes. It is therefore important to understand the relationship between the factors governing liposome structure.

The head groups and alkyl chains of the lipids are two important elements influencing the liposome structure and, thus, the biological activity of the final lipoplex. The aliphatic chain length and degree of saturation affect the phase transition temperature (Tm) and lipid packing within the liposomes. Longer, saturated side chains favor tight lipid packing and higher Tm, while a decrease in aliphatic chain length decreases the Tm. The Tm values for the lipids used in the current study were not determined, however, the values for the related cytofectins DMRIE, (±)-N-(2-hydroxyethyl)-N,N-dimethyl-2,3-bis(hexadecyloxy)-1-propanaminium bromide (DPRIE) and (±)-N-(2-hydroxyethyl)-N,N-dimethyl-2,3-bis(octadecyloxy)-1-propanaminium bromide (DSRIE) are 24, 42 and 56°C, respectively (23). An increase in side chain unsaturation and branching also disrupts lipid packing and decreases Tm. Generally, when hydrophobic interactions between side chains are decreased the lateral mobility of lipids within the liposome increases. For example, it has been shown that lipid bilayers composed of unmatched side chains (e.g. C14:1 and C18:1) can result in non-ideal lipid mixing (24). Short chain lipids tend to segregate within the bilayer and form clusters or ‘rafts’, creating a non-uniform distribution of lipids throughout the membrane, resulting in a more fluid, less stable bilayer (24,25). The effect is anticipated to be more pronounced as the degree of unsaturation, and hence disorder, increases. Non-uniform distributions of propanaminium type cationic lipids similar to those used here have been reported. Specifically, DSC measurements on 1,2-dimyristoyl-3-trimethylammonium propane (DMTAP):DOPE (C14:0/C18:1) liposomes showed a phase separation between a DOPE-rich phase and a DMTAP phase (26).

The results of the pDNA accessibility studies in which C18:1, C12:1, C18:0 and C12:0 alkyl chains were exchanged between cytofectin and co-lipid head groups are consistent with a fluidity-based rationale. Lipoplexes containing cytofectins or co-lipids with C18:1 side chains show a lower dependence on cytofectin/pDNA molar ratio in PBS and SWFI compared with lipoplexes prepared with lipids containing C18:0 side chains. Lipoplexes containing C18:1 side chains may be more fluid and less regular in structure, perhaps resulting in weaker binding of pDNA within the lipoplex. Conversely, the presence of the long, saturated side chains may serve to anchor the lipids in the membrane. As a result, the lipoplex structure may be less flexible, with a more regular placement of the positively charged cytofectin head groups on the lipid surface, allowing tighter and more uniform coverage of the pDNA. It is important to emphasize that the combination of side chain length and degree of saturation in both cytofectin and co-lipid is important. For example, when cytofectin and co-lipid contain saturated side chains of equal length, the dependence of the pDNA accessibility on cytofectin/pDNA molar ratio decreases with an increase in aliphatic chain length, as shown in Figure 3. Since the lengths of both lipid side chains are changed in concert, the membrane fluidity is functionally changed. As noted above, lipid Tm values increase with increasing aliphatic side chain length. Thus, a decrease in accessible pDNA with increasing chain length is consistent with previous work showing that cationic lipids interact with pDNA most effectively near the lipid Tm (27). The association between lipids becomes as important as pDNA–lipid associations, resulting in weaker or altered pDNA binding and thus greater dye accessibility.

The final lipoplex structure is also influenced by the surface charge on the liposome prior to complexation. This surface charge depends on several factors, including lipid structure and solution environment. For example, a charged primary amine head group would be predicted to extend further from the liposome surface than a head group containing an alcohol moiety (28). Ion pairing between the cytofectin and DOPE head groups may also occur at the liposome surface, thereby altering the surface properties (29,30). Surface charge was also found to be an important determinant in the tendency of bilayers to adopt lamellar or non-lamellar phases (31), being dependant not only on the lipid formal charges but also on pH, hydrogen bonding and the ionic strength of the formulation vehicle. The ionic strength of PBS is higher than water and thus should more effectively screen charges on the liposome surface. The results in Figure 3 show that pDNA accessibility is higher in PBS relative to SWFI and suggest that the pDNA is more loosely bound, consistent with such an interaction.

The nature of the ions, and not simply the ionic strength of the formulation vehicle, can also dramatically alter the charge interactions at the liposome surface. Although sodium phosphate (150 mM) is comparable in ionic strength to PBS, nearly all the pDNA is accessible to dye binding when high phosphate concentrations are used. The phosphate ion may form bridges between bilayer surfaces containing cytofectins with γ-amine head groups, thus limiting the number of binding sites available for pDNA (20). DNA binding induces further changes in liposome structure, such as changes in lipid packing or more subtle changes in surface interactions between lipid head groups. Results from fluorescence studies suggest that when lipoplexes were prepared with pDNA in stoichiometric excess over cationic lipid (as in the present study), dehydration occurred at the hydrophobic–hydrophilic interface of the lipid head group region at the bilayer surface. Dehydration of both pDNA and lipid are thought to be a major driving force in facilitating and stabilizing pDNA and lipid interactions within lipoplexes (32). Calorimetry studies have shown binding of 1,2-dioleoyl-3-trimethylammonium-propane (DOTAP):DOPE and N,N-didecyl-N,N-dimethylammonium bromide (DDAB):DOPE to pDNA to be exothermic and consistent with an uptake of protons by DOPE upon complex formation (33). Thus, the final lipoplex structure is a sum of many complex interactions which likely effect not only lipoplex structure, but biological activity as well.

The results of the current study suggest that the highest transfection levels are achieved when lipoplexes contain cytofectin/co-lipid combinations which promote fluid liposome structures, and may be more active in destabilizing cellular membranes, consistent with results of previous studies (30,34,35). Transfection involves interaction of the lipoplex with cellular and/or endosomal membranes and fusion between lipids in the lipoplex and cellular membranes allows release of pDNA into the cytoplasm (30,36–46). The ability to destabilize the bilayer membrane and promote HII lipid phase formation has been correlated with high transfection activity, rationalized based on the idea that the HII phase promotes liposome fusion (30,36,47). DOPE is thought to play a major role in release of pDNA from the endosome and in the absence of DOPE lipoplex uptake may occur by lipid exchange between lipoplex and cellular membranes (2). Studies have shown that cationic lipids may interfere with the ability of DOPE to adopt the HII phase, but addition of DNA promotes the HII phase (39,42,44). However, DOPE is not always essential to achieve high levels of transfection (48), and the results of the current study confirm this fact.

In the present study, GAP-DMORIE (C14:1) generally resulted in higher levels of transfection than GAP-DLRIE (C12:0) regardless of the co-lipid, with the GAP-DMORIE:DpyPE (1:1) combination being the best transfection agent (Fig. 2). The C14:1 side chain of GAP-DMORIE combined with the branched 16 carbon side chain of DPyPE may act in concert to provide an enhanced degree of membrane fluidity compared with the other combinations.

Further evidence of the relationship between fluid liposome structure and optimal transfection comes from the experiments where alkyl side chains were exchanged between cytofectin and co-lipid head groups (Fig. 5). GAP-DORIE:DLPE (C18:1/C12:0) resulted in the highest transgene expression of any combination examined; CPAE cells transfected using SWFI or sodium phosphate as the formulation vehicles being the exceptions. In SWFI, GAP-DLRIE:DOPE (C12:0:18:1) was the only combination which gave a detectable level of transfection. As mentioned above, the combination of the short and long side chains (asymmetrical in the hydrophobic domains) is predicted to result in a disruption of lipid packing and a more fluid liposome structure. The results of several studies conclude that the asymmetrical side chains may afford better intermembrane mixing and thus higher trasfection levels (30,34,35). In addition, there may be some relationship involving intramolecular juxtaposition of alkyl chain and head group, since the C18:1/C12:0 cytofectin/co-lipid combination generally resulted in higher levels of transfection compared with the C12:0/C18:1 cytofectin/co-lipid combination (Fig. 5). Further evidence illustrating the relationship between alkyl chain substituents and head group composition comes from the results of the experiments where the diphytanoyl side chain was part of the cytofectin or, alternatively, part of the co-lipid. Transfection levels were usually highest when GAP-DMORIE:DPyPE was used in preparing lipoplexes, although some vehicle-dependent differences were observed. For example, in BHK cells GAP-DMORIE:DPyPE gave the highest transfection of any combination when formulations were prepared in SWFI, but GAP-DLRIE:DPyPE and GAP-DPyRIE:DLPE were superior in PBS and sodium phosphate, respectively.

Several lines of evidence presented in the current study and in several other studies support the idea that fluid liposome structure is important for optimal transfection. However, one notable exception comes from the experiments where the alkyl side chains of the cytofectin and co-lipid were matched in length. The highest transfection in BHK cells was achieved when lipoplexes were prepared in PBS using GAP-DPRIE:DPPE (C16:0/C16:0), as shown in Figure 4. In this case, a lipoplex structure stabilized by hydrophobic interactions between the long side chains is predicted and supported by the pDNA accessibility results discussed above.

One possible rationalization for how a rigid lipoplex structure could result in high transfection levels may be differences in the membranes of different cell types. The cellular membrane composition varies with cell type and cell cycle (49). For example, the percentage of lipids with unsaturated side chains and the cholesterol composition vary with cell type (50,51). In general, interaction between lipids in the bilayer increases when the cholesterol or protein composition is increased, and decreases when the unsaturated lipid content increases (49). However, destabilization of the cellular membrane may depend on specific interactions between cellular membrane lipids and the lipids comprising the lipoplex. Differences in transfection activity of cytofectin formulations has been reported to be cell type dependent, consistent with the results of the current study (30,34,35).

In summary, three major conclusions can be drawn from the current study. First, the transfection activity of different lipoplexes was cell type and vehicle dependent and did not correlate with dye accessibility. Second, a correlation may exist between transfection and enhanced membrane fluidity in both the lipoplex and cellular membranes. Finally, the characteristic molecular features which give rise to fluidity may be incorporated into either the cytofectin, co-lipid or both. The last point suggests that there exists a ‘physical synergy’ between the structural features in the cytofectin and co-lipid which allows for design of optimal lipoplexes for a particular application.

REFERENCES

  • 1.Caplen N.J., Alton,E.W., Middleton,P.G., Dorin,J.R., Stevenson,B.J., Gao,X., Durham,S.R., Jeffery,P.K., Hodson,M.E., Coutelle,C. et al. (1995) Liposome-mediated CFTR gene transfer to the nasal epithelium of patients with cystic fibrosis. Nature Med., 1, 39–46. [Erratum (1995) Nature Med., 1, 272.] [DOI] [PubMed] [Google Scholar]
  • 2.Gao X. and Huang,L. (1995) Cationic liposome-mediated gene transfer. Gene Ther., 2, 710–722. [PubMed] [Google Scholar]
  • 3.Felgner P.L., Zaugg,R.H. and Norman,J.A. (1995) Synthetic recombinant DNA delivery for cancer therapeutics. Cancer Gene Ther., 2, 61–65. [PubMed] [Google Scholar]
  • 4.Nabel G.J., Gordon,D., Bishop,D.K., Nickoloff,B.J., Yang,Z.Y., Aruga,A., Cameron,M.J., Nabel,E.G. and Chang,A.E. (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, 15388–15393. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Rubin J., Galanis,E., Pitot,H.C., Richardson,R.L., Burch,P.A., Charboneau,J.W., Reading,C.C., Lewis,B.D., Stahl,S., Akporiaye,E.T. and Harris,D.T. (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. [DOI] [PubMed] [Google Scholar]
  • 6.Stopeck A.T., Hersh,E.M., Akporiaye,E.T., Harris,D.T., Grogan,T., Unger,E., Warneke,J., Schluter,S.F. and Stahl,S. (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. [DOI] [PubMed] [Google Scholar]
  • 7.Hartikka J., Bozoukova,V., Ferrari.,M., Sukhu,L., Enas,J., Sawdey,M., Wloch,M.K., Tonsky,K., Norman,J., Manthorpe,M. and Wheeler,C.J. (2001) Vaxfectin enhances the humoral response to pDNA encoded antigens. Vaccine, 19, 1911–1923. [DOI] [PubMed] [Google Scholar]
  • 8.Hartikka J., Bozoukova,V., Jones,D., Mahajan,R., Wloch,M.K., Sawdey,M., Buchner,C., Sukhu,L., Barnhart,K.M., Abai,A.M., Meek,J., Shen,J. and Manthorpe,M. (2000) Sodium phosphate enhances plasmid DNA expression in vivo. Gene Ther., 7, 1171–1182. [DOI] [PubMed] [Google Scholar]
  • 9.Ishii N., Fukushima,J., Kaneko,T., Okada,E., Tani,K., Tanaka,S.I., Hamajima,K., Xin,K.Q., Kawamoto,S., Koff,W., Nishioka,K., Yasuda,T. and Okuda,K. (1997) Cationic liposomes are a strong adjuvant for a DNA vaccine of human immunodeficiency virus type 1. Aids Res. Hum. Retroviruses, 13, 1421–1428. [DOI] [PubMed] [Google Scholar]
  • 10.Reyes L., Hartikka,L., Bozoukova,V., Sukhu,L., Nishioka,W., Singh,G., Ferrari,M., Wheeler,C.J., Manthorpe,M. and Wloch,M.K. (2001) Vaxfectin enhances antigen specific antibody titers and maintains Th1type immune responses to plasmid DNA immunization. Vaccine, 19, 3778–3786. [DOI] [PubMed] [Google Scholar]
  • 11.Toda S., Ishii,N., Okada,E., Kusakabe,K.I., Arai,H., Hamajima,K., Gorai,I., Nishioka,K. and Okuda,K. (1997) HIV-1-specific cell-mediated immune responses induced by DNA vaccination were enhanced by mannan-coated liposomes and inhibited by anti-interferon-gamma antibody. Immunology, 92, 111–117. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Okada E., Sasaki,S., Ishii,N., Aoki,I., Yasuda,T., Nishioka,K., Fukushima,J., Miyazaki,J., Wahren,B. and Okuda,K. (1997) Intranasal immunization of a DNA vaccine with IL-12- and granulocyte-macrophage colony-stimulating factor (GM-CSF)-expressing plasmids in liposomes induces strong mucosal and cell-mediated immune responses against HIV-1 antigens. J. Immunol., 159, 3638–3647. [PubMed] [Google Scholar]
  • 13.Yokoyama M., Zhang,J. and Whitton,J.L. (1996) DNA immunization: effects of vehicle and route of administration on the induction of protective antiviral immunity. FEMS Immunol. Med. Microbiol., 14, 221–230. [DOI] [PubMed] [Google Scholar]
  • 14.Gregoriadis G., Saffie,R. and de Souza,J.B. (1997) Liposome-mediated DNA vaccination. FEBS Lett., 402, 107–110. [DOI] [PubMed] [Google Scholar]
  • 15.Wheeler C.J., Felgner,P.L., Tsai,Y.J., Marshall,J., Sukhu,L., Doh,S.G., Hartikka,J., Nietupski,J., Manthorpe,M., Nichols,M., Plewe,M., Liang,X., Norman,J., Smith,A. and Cheng,S.H. (1996) A novel cationic lipid greatly enhances plasmid DNA delivery and expression in mouse lung. Proc. Natl Acad. Sci. USA, 93, 11454–11459. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Baccaglini L., Hoque,A., Wellner,R., Goldsmith,C., Redman,R., Sankar,V., Kingman,A., Barnhart,K., Wheeler,C.J. and Baum,B. (2001) Cationic liposome-mediated gene transfer to rat salivary epithelial cells in vitro and in vivo. J. Gene Med., 3, 82–90. [DOI] [PubMed] [Google Scholar]
  • 17.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. GeneTher., 6, 585–594. [DOI] [PubMed] [Google Scholar]
  • 18.Monkkonen J. and Urtti,A. (1998) Lipid fusion in oligonucleotide and gene delivery with cationic lipids. Adv. Drug Delivery Rev., 34, 37–49. [DOI] [PubMed] [Google Scholar]
  • 19.Doh S.G., Vahlsing,H.L., Hartikka,J., Liang,X. and Manthorpe,M. (1997) Spatial-temporal patterns of gene expression in mouse skeletal muscle after injection of lacz plasmid DNA. Gene Ther., 4, 648–663. [DOI] [PubMed] [Google Scholar]
  • 20.Ferrari M.E., Rusalov,D., Enas,J. and Wheeler,C.J. (2001) Trends in lipoplex physical properties dependent on cationic lipid structure, vehicle and complexation procedure do not correlate with biological activity. Nucleic Acids Res., 29, 1539–1548. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Ferrari M.E., Nguyen,C.M., Zelphati,O., Tsai,Y. and Felgner,P.L. (1998) Analytical methods for the characterization of cationic lipid-nucleic acid complexes. Hum. Gene Ther., 9, 341–351. [DOI] [PubMed] [Google Scholar]
  • 22.Ausubel F., Brent,R., Kingston,R., Moore,D., Seidman,J., Smith,J. and Struhl,K. (1987) Current Protocols in Molecular Biology, Vol. 2. Green Publishing, New York, NY.
  • 23.Bennett C.F., Mirejovsky,D., Crooke,R.M., Tsai,Y.J., Felgner,J., Sridhar,C.N., Wheeler,C.J. and Felgner,P.L. (1998) Structural requirements for cationic lipid mediated phosphorothioate oligonucleotides delivery to cells in culture. J. Drug Target., 5, 149–162. [DOI] [PubMed] [Google Scholar]
  • 24.Otto S., Janout,V., DiGiorgio,A., Young,M. and Regen,S. (2000) Detection of nonideal mixing of phospholipids in fluid bilayers. J. Am. Chem. Soc., 122, 1200–1204. [Google Scholar]
  • 25.Peirini L.M. and Maxfield,F.R. (2001) Flotillas of lipid rafts fore and aft. Proc. Natl Acad. Sci. USA, 98, 9471–9473. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.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. [DOI] [PubMed] [Google Scholar]
  • 27.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. [DOI] [PubMed] [Google Scholar]
  • 28.Asuncion-Punzalan E. and London,E. (1995) Control of the depth of molecules within membranes by polar groups: determination of the location of anthracene-labeled probes in model membranes by parallax analysis of nitroxide-labeled phospholipid induced fluorescence quenching. Biochemistry, 34, 11460–11466. [DOI] [PubMed] [Google Scholar]
  • 29.Zuidam N.J. and Barenholz,Y. (1997) Electrostatic parameters of cationic liposomes commonly used for gene delivery as determined by 4-heptadecyl-7-hydroxycoumarin. Biochim. Biophys. Acta Biomembr., 1329, 211–222. [DOI] [PubMed] [Google Scholar]
  • 30.Felgner J.H., Kumar,R., Sridhar,C.N., Wheeler,C.J., Tsai,Y.J., Border,R., Ramsey,P., Martin,M. and Felgner,P.L. (1994) Enhanced gene delivery and mechanism studies with a novel series of cationic lipid formulations. J. Biol. Chem., 269, 2550–2561. [PubMed] [Google Scholar]
  • 31.Lewis R. and McElhaney,R. (2000) Surface charge markedly attenutates the non-lamellar phase-forming propensities of lipid bilayer membranes: calorimetric and (31)P-nuclear magnetic resonance studies of mixtures of cationic, anionic and zwitterionic lipids. Biophys. J., 79, 1455–1464. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Hirsch-Lerner D. and Barenholz,Y. (1998) Probing DNA-cationic lipid interactions with the fluorophore trimethylammonium diphenyl-hexatriene (TMADPH). Biochim. Biophys. Acta Biomembr., 1370, 17–30. [DOI] [PubMed] [Google Scholar]
  • 33.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. [DOI] [PubMed] [Google Scholar]
  • 34.Balasubramaniam R.P., Bennett,M.J., Aberle,A.M., Malone,J.G., Nantz,M.H. and Malone,R.W. (1996) Structural and functional analysis of cationic transfection lipids: the hydrophobic domain. Gene Ther., 3, 163–172. [PubMed] [Google Scholar]
  • 35.Heyes J.A., Niculescu-Duvaz,D., Cooper,R.G. and Springer,C.J. (2002) Synthesis of novel cationic lipids: effect of structural modification on the efficiency of gene transfer. J. Med. Chem., 45, 99–114. [DOI] [PubMed] [Google Scholar]
  • 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, 18997–19007. [DOI] [PubMed] [Google Scholar]
  • 37.Wrobel I. and Collins,D. (1995) Fusion of cationic liposomes with mammalian cells occurs after endocytosis. Biochim. Biophys. Acta, 1235, 296–304. [DOI] [PubMed] [Google Scholar]
  • 38. Fasbender,A., Marshall,J., Moninger,T.O., Grunst,T., Cheng,S. and Welsh,M.J. (1997) Effect of colipids in enhancing cationic lipid-mediated gene transfer in vitro and in vivo. Gene Ther., 4, 716–725. [DOI] [PubMed] [Google Scholar]
  • 39.Mok K.W. and Cullis,P.R. (1997) Structural and fusogenic properties of cationic liposomes in the presence of plasmid DNA. Biophys. J., 73, 2534–2545. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Ouahabi A., Thiry,M., Pector,V., Fuks,R., Ruysschaert,J. and Vandenbranden,M. (1997) The role of endosome destabilizing activity in the gene transfer process mediated by cationic lipids. FEBS Lett., 414, 187–192. [DOI] [PubMed] [Google Scholar]
  • 41.Harvie P., Wong,F.M. and Bally,M.B. (1998) Characterization of lipid DNA interactions. I. Destabilization of bound lipids and DNA dissociation. Biophys. J., 75, 1040–1051. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.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. [DOI] [PubMed] [Google Scholar]
  • 43.Noguchi A., Furuno,T., Kawaura,C. and Nakanishi,M. (1998) Membrane fusion plays an important role in gene transfection mediated by cationic liposomes. FEBS Lett., 433, 169–173. [DOI] [PubMed] [Google Scholar]
  • 44.Mitrakos P. and Macdonald,P.M. (2000) Nucleotide chain length and the morphology of complexes with cationic amphiphiles: (31)P-NMR observations. Biochim. Biophys. Acta, 1463, 355–373. [DOI] [PubMed] [Google Scholar]
  • 45.Angelova M.I., Hristova,N. and Tsoneva,I. (1999) DNA-induced endocytosis upon local microinjection to giant unilamellar cationic vesicles. Eur. Biophys. J., 28, 142–150. [DOI] [PubMed] [Google Scholar]
  • 46.Remy J.S., Sirlin,C., Vierling,P. and Behr,J.P. (1994) Gene transfer with a series of lipophilic DNA-binding molecules. Bioconjug. Chem., 5, 647–654. [DOI] [PubMed] [Google Scholar]
  • 47.Hafez I.M., Maurer,N. and Cullis,P.R. (2001) On the mechanism whereby cationic lipids promote intracellular delivery of polynucleic acids. Gene Ther., 8, 1188–1196. [DOI] [PubMed] [Google Scholar]
  • 48.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 DNA-cytofectin complexes. Biochim. Biophys. Acta, 1280, 1–11. [DOI] [PubMed] [Google Scholar]
  • 49.Shinitzky M. and Barenholz,Y. (1978) Fluidity parameters of lipid regions determined by fluorescence polarization. Biochim. Biophys. Acta, 515, 367–394. [DOI] [PubMed] [Google Scholar]
  • 50.Leray C., Andriamampandry,M., Gutbier,G., Cavadenti,J., Klein-Soyer,C., Gachet,C. and Cazenave,J.P. (1997) Quantitative analysis of vitamin E, cholesterol and phospholipid fatty acids in a single aliquot of human platelets and cultured endothelial cells. J. Chromatogr. B Biomed. Sci. Appl., 696, 33–42. [DOI] [PubMed] [Google Scholar]
  • 51.Borchman D., Tang,D. and Yappert,M.C. (1999) Lipid composition, membrane structure relationships in lens and muscle sarcoplasmic reticulum membranes. Biospectroscopy, 5, 151–167. [DOI] [PubMed] [Google Scholar]

Articles from Nucleic Acids Research are provided here courtesy of Oxford University Press

RESOURCES