Analysis of Cationic Liposome-mediated Interactions of Plasmid DNA with Murine and Human Melanoma Cells in Vitro *

Lipid-based DNA transfer formulations are typically selected on the basis of in vitro transfection studies where the activity of specific formulations is defined by transgene expression. It is unclear, however, whether expression is directly related to the efficiency of DNA transfer. In an attempt to correlate DNA transfer with transgene expression, we used a simple assay consisting of measuring DNA (3H-plasmid encoding for β-galactosidase) binding to murine (B16/BL6) and human (KZ) melanoma cells in vitro at 4 and 37 °C. The difference in cell association at these temperatures was assumed to be a consequence of DNA uptake, an assumption that was confirmed by protease removal of cell surface-associated DNA. DNA associated with B16/BL6 melanoma cells (up to 30 ng or 12% of the added DNA) following incubation with dioleoyldimethylammonium chloride/dioleoylphosphatidylethanolamine (DOPE) liposome-DNA aggregates was comparable to that achieved with 1,2-dioleoyloxypropyl-3-trimethylammonium bromide/DOPE or dimethyldioctadecylammonium bromide/DOPE liposomes; however, transgene expression was 2- and 5-fold less for the latter two formulations, respectively. Similarly, equivalent amounts of DNA delivery were achieved with B16/BL6 and KZ melanoma cells, yet the level of transgene expression in the KZ cells was undetectable. It was demonstrated that the lack of transgene expression was not a consequence of cell-specific differences in DNA degradation.

Although the understanding of the cellular and molecular mechanisms involved in cationic liposome-mediated gene transfer is limited, DNA delivered by this process must ultimately gain access to the nucleus if effective expression of the foreign gene is to take place (1). Experiments to date have focused on the development of various lipid formulations in an effort to identify an efficient carrier as measured by transgene expression (2)(3)(4). This approach is useful for comparing expression levels following plasmid DNA transfer using different lipid formulations, but it provides little information regarding the usefulness of such formulations to effectively deliver DNA to the cells. Binding of the liposome-DNA complex to the cell membrane and entry of the DNA into the cell are the initial steps of the transfection process, and these are important considerations in the development of effective cationic liposome-mediated gene transfer systems.
It is evident, however, that the liposome-DNA aggregates are heterogeneous with respect to size and shape (5,6) and generate highly complex fused structures as shown by electron microscopy (7,8). Transfection can be achieved using this heterogeneous population of aggregates, but the complex that is directly responsible for mediating the transfection process is still unknown. Furthermore, the mechanisms by which plasmid DNA in this heterogeneous mixture is delivered to a cell remains a mystery. It is currently thought that DNA-cationic liposome aggregates gain access to the cell by endocytosis (9). However, the liposomes used to prepare the complexes have the potential to fuse with membranes, thereby providing an alternative mechanism by which foreign DNA can be delivered into the cytoplasm of the cell (10 -15). The factors that govern this membrane fusion-dependent process have not been extensively investigated. The lipid composition of the liposome is, however, important (3,16). The presence of DOPE, 1 for example, appears to be necessary for optimal transfection efficiencies (17,18), and this lipid is known to adopt structures that can engender membrane fusion (19,20).
To identify the importance of delivering DNA to the cell, we evaluated the level of expression of the plasmid pCMV␤ in vitro comparing different cell lines that were transfected with DNA mixed with cationic liposomes of different lipid compositions. We designed a simple DNA delivery assay to complement these transfection studies. Delivery of intact DNA was determined as a function of the amount of DNA added and gene expression. Measurement of cell-associated DNA is not an accurate predictor of whether a specific cell line will be transfected or whether the cationic liposomes employed will be more or less effective than other cationic liposomes.

Materials
DDAB was purchased from Sigma, and DOPE was purchased from Avanti Polar Lipids (Alabaster, AL). DODAC was synthesized and supplied by Steven Ansell of Inex Pharmaceuticals Corp. (Vancouver, BC). [ 14 C]Cholesterylhexadecyl ether was purchased from Amersham Corp. (Oakville, ON). Lipofectin (DOTMA/DOPE, 50:50 mol %) was supplied by Life Technologies, Inc. Triton X-100 was purchased from Bio-Rad, and chlorophenol red ␤-galactopyranoside was obtained from Boehringer Mannheim (Laval, Quebec). All other chemicals used in this study were reagent grade and solvents used were high performance liquid chromatography grade.
The murine B16/BL6 melanoma cell line was obtained from NCI Tumor Repository 12-102-54 (Bethesda, MD) and was maintained in RPMI 1640 with 5% FBS. KZ, a human melanoma cell line, was a gift from Dr. Wright (Fred Hutchinson Cancer Center, Seattle, WA) and was maintained in RPMI 1640 with 10% FBS. Cell lines were main-tained at 37°C in a 5% CO 2 atmosphere with no antibiotics.
The 7.2-kb plasmid pCMV␤ (GenBank accession number U02451) encoding the Escherichia coli ␤-galactosidase was obtained from CLON-TECH Laboratories Inc. (Palo Alto, CA) and was used for analyzing DNA delivery and transfection analyses. The 4.5-kb plasmid, pInex-CATv2.0, was constructed by Roger Graham of Inex Pharmaceuticals Corp (Vancouver, BC) and contains the E. coli chloramphenicol acetyltransferase gene under the control of the cytomegalovirus promoter and avian myeloblastosis virus enhancer. This plasmid was used for the assessment of intact DNA, since this plasmid can be separated from genomic DNA in a 0.8% agarose gel. Plasmid DNAs were isolated using standard molecular techniques (21) and purified using Qiagen Plasmid Purification Kit (Qiagen, Chatsworth, CA). Radiolabeled plasmids were generated by culturing E. Coli pCMV␤ or pInexCATv2.0 with [ 3 H]thymidine-5Ј-triphosphate (NEN Life Science Products) and purified using standard techniques. Specific activity of 3 H-pCMV␤ and 3 H-pInex-CATv2.0 were approximately 50,000 and 200,000 dpm/g, respectively.

Methods
Preparation of Liposomes and Liposome-DNA Complexes-Preformed DOTMA/DOPE liposomes (50:50 mol %) were supplied under the trade name Lipofectin. These liposomes were prepared by sonication and exhibited a mean diameter of 50 -70 nm as measured by quasi-elastic light scattering. DODAC/DOPE (50:50 mol %) and DDAB/ DOPE (50:50 mol %) liposomes were prepared according to the method of Hope et al. (22). Lipids were dissolved in chloroform (50 mg/ml) and radiolabeled at a specific activity of 1-2 mCi/50 mg with [ 14 C]cholesterylhexadecyl ether as a non-exchangeable liposomal lipid marker (23). The lipids were dried to a thin film under a stream of nitrogen gas and vacuum-dried at Ͼ76 mm Hg for at least 4 h. The films were hydrated in H 2 O. The hydrated lipid solutions were passed 10 times at 65°C through an extruder (Lipex Biomembranes, Vancouver, BC) containing three stacked 100-nm polycarbonate membranes. The resulting liposomes were evaluated for lipid concentration by measuring the levels of [ 14 C]cholesterylhexadecyl ether using a Packard TR 1900 scintillation counter. The size of the liposomes was measured by quasielastic light scattering using a Nicomp Submicron Particle Sizer (model 270, Pacific Scientific, Santa Barbara, CA) operating at a wavelength of 632.8 nm. All liposomes used exhibited a mean diameter of 100 -140 nm, and these were stored at 4°C until use. Liposomes were used within 2 months of preparation and liposome size did not change during storage.
Prior to use, liposomes were diluted into sterile distilled H 2 O to a final lipid concentration ranging from 100 to 1000 M. Plasmid DNA was diluted into sterile distilled H 2 O at concentrations ranging from 10 to 60 g/ml. Ten l of diluted liposomes were mixed with 10 l of DNA in a microcentrifuge tube at the desired concentrations and incubated at room temperature for 30 min. The DNA-liposome complexes were used immediately for transfection.
Transfection Analysis-B16/BL6 and KZ cells were trypsinized, and ϳ1700 cells were seeded per well in a 96-well flat bottom microtiter plate (Costar, Cambridge, MA) and incubated overnight. At the time of transfection, the media were removed, and 80 l of serum free RPMI 1640 was added to each well. Twenty l of prepared liposome-pCMV␤ complexes were added to each well and incubated at 37°C in 5% CO 2 for 4 h. The complex containing media was subsequently removed and replaced with 200 l of RPMI 1640 containing 5% FBS. The cells were incubated for a further 48 h prior to assessment for ␤-galactosidase activity as described below.
Media were removed from each well of the microtiter plate, and the adherent cells were washed 2 ϫ with 100 l of phosphate-buffered saline at room temperature. The cells were lysed by the addition of 50 l of 0.1% Triton X-100, 250 mM sodium phosphate (pH 8.0) to each well. Subsequently, 150 l of chlorophenol red ␤-galactopyranoside (1 mg/ml in 60 mM sodium phosphate buffer (pH 8.0) containing 1 mM MgSO 4 , 10 mM KCl, and 50 mM ␤-mercaptoethanol) was added to each well, and the samples were left to incubate at room temperature. Absorbances at 570 nm were measured at regular time intervals using a Titertek Multiscan Type 310C microtiter plate reader (Flow Laboratories, Mississauga, ON). A standard curve using a range of concentrations of ␤-galactosidase (Sigma) in phosphate-buffered saline with 0.5% bovine serum albumin was also generated. Enzyme activity in transfected cells was calculated by comparing A 570 obtained with the ␤-galactosidase standard concentration curve and were expressed as milliunits of ␤-galactosidase/well.
Analysis of DNA Delivery-B16/BL6 and KZ cells were seeded at 2 ϫ 10 4 cells/well in 48-well flat bottom plates (Falcon Labware, Missis-sauga, ON) and incubated overnight. At the time of transfection, media were removed, and 80 l of serum-free RPMI 1640 was added to each well. Twenty l of prepared liposome-DNA complexes were added to each well and incubated at 37 or 4°C for 1-4 h. At the appropriate time interval, the supernatant consisting of non-cell-associated DNA-liposome complexes was removed from each well and retained for scintillation counting. Each well was rinsed carefully with 200 l of serumfree RPMI 1640 and pooled with supernatant containing DNA-liposome complexes. Washed cells were lysed in 300 l of 10 mM Tris-HCl (pH 8.0), 10 mM NaCl, 3 mM MgCl 2 , 0.5% SDS for 5 min, and the lysate was retained for scintillation counting. Each well was rinsed with 100 l of lysis buffer and pooled with matched lysates. Pico-Fluor scintillant (Packard Instrument Co., Meriden, CT) was added to all samples, and the radioactivity ( 3 H-pCMV␤) was measured in a Canberra Packard TR 1900 Scintillation Counter (Packard Instrument Co., Meriden, CT).
In an attempt to evaluate whether DNA associated with the cells at 37°C was internalized, transfected B16/BL6 cells were proteasetreated prior to sample collection. B16/BL6 cells were transfected for 4 h at 4 and 37°C as described above using DODAC/DOPE liposomes complexed to 3 H-pInexCATv2.0. Adherent cells were removed from wells by trypsinization at 4°C for 2 min using 0.25% trypsin (Life Technologies, Inc.) and diluted with 400 l of Hanks' balanced salt solution. Subsequently, cell-associated DNA was separated from unassociated DNA-liposome complexes by fractionating cells on a gradient consisting of NycoPrep (Sigma) layered at densities of 1.077 and 1.022. Cells were collected at the 1.022/1.077 interface following centrifugation at 400 ϫ g for 30 min at room temperature, and the amount of plasmid DNA associated with the cells was evaluated by scintillation counting. Control studies revealed that free plasmid and the liposome-DNA complexes were consistently found above the 1.022 fraction.
Assessment of Intact DNA-B16/BL6 cells were seeded at 2 ϫ 10 4 cells/well, and KZ cells were seeded at 5 ϫ 10 4 cells/well in 48-well flat bottom plates (Falcon Labware, Mississauga, ON) and incubated overnight in a 37°C incubator. Liposome-DNA complexes were prepared using 3 H-pInexCATv2.0 as described previously. The 4.5-kb plasmid pInexCATv2.0 was used for these experiments because the migration of the three conformations of plasmid can readily be separated from genomic DNA in a 0.8% agarose gel. At the time of transfection, media were removed from all wells and replaced with 80 l of serum-free medium. Twenty l of prepared liposome-3 H-pInexCATv2.0 complexes were added to each well and incubated in a 37°C incubator for 4 h. Six hundred l of media (with 5% FBS for B16/BL6 and 10% FBS for KZ) was added to each well and incubated for a further 24 h.
Medium containing 3 H-pInexCATv2.0-liposome complexes was removed and retained for scintillation counting. Each well was rinsed carefully with 200 l of serum-free media, pooled with supernatant, and radioactivity ( 3 H-pInexCATv2.0) evaluated by scintillation counting. Cells were lysed by the addition of 400 l of 0.6% SDS, 10 mM EDTA. Cell lysates were collected in microcentrifuge tubes, the wells rinsed with 100 l of lysis buffer, and pooled with respective cell lysates. Samples were further processed following the method of Hirt (24) with minor modifications. This procedure selectively precipitates genomic DNA leaving the smaller molecular weight plasmid DNA in the supernatant, thereby enhancing the detection of plasmid DNA following gel electrophoresis. NaCl was added to cell lysates at a final concentration of 1 M, and the samples were incubated at 4°C overnight. Following centrifugation at 10,000 ϫ g for 15 min at 4°C, the supernatant was recovered and further purified. DNA was extracted from the supernatant once with equal volumes Tris-HCl-buffered phenol/chloroform (1:1) and precipitated with 0.1 volumes 3 M sodium acetate and 0.7 volumes isopropyl alcohol. DNA was recovered by centrifugation at 10,000 ϫ g for 30 min at 4°C and the pellet resuspended in 10 l of 10 mM Tris-HCl (pH 8.0), 2 mM EDTA. Control samples were also evaluated by adding free DNA or DNA-liposome complexes to untransfected cell lysates and processed along with test samples. All experiments were performed in triplicate.
The integrity of the DNA was evaluated by two methods. First, the presence of intact plasmid DNA was assessed by agarose gel electrophoresis. Each DNA sample (containing DNA isolated from a single transfection from a 48-well plate) was loaded onto a 0.8% agarose gel and subjected to electrophoresis at 80 V in TBE buffer (89 mM Tris borate, 2 mM EDTA) for 45 min. The gel was stained with ethidium bromide (0.5 g/ml) for 20 min and photographed with UV transillumination (Ultra Violet Products, San Gabriel, CA). Second, the amount of 3 H-pInexCATv2.0 associated with the three plasmid bands was quantified by liquid scintillation counting. The area of the agarose gel encompassing the three 3 H-pInexCATv2.0 plasmid bands was excised and placed into a scintillation vial with 500 l of H 2 O. Each sample was boiled in a water bath for ϳ10 min until the gel slice was fully melted. Pico-Fluor scintillant was added to each sample and the radioactivity measured.
Statistical Analysis-Quantitative data generated for ␤-galactosidase activity, and the DNA uptake studies were statistically evaluated using regression analysis from the Microcal Origin version 3.5 computer program (Microcal Software, Inc., Northampton, MA). Analysis of variance was used to determine statistical significance of ␤-galactosidase activity and DNA uptake in different liposome formulations and in different cell lines using Statistica (Statistica Software Inc., Tulsa, OK).

RESULTS
Although the mechanisms of liposome-mediated transfection are not understood, the ratio of cationic lipid to DNA is known to be important for optimal efficiency in vitro (25). The optimal ratio occurs when the number of positive charges incorporated into the cationic liposome is greater than the number of negative charges on the DNA (2), thereby promoting association of the liposome-DNA aggregates with the negatively charged surface of the cell. It is believed that cationic liposomes provide a multivalent surface to bind plasmid DNA, thus protecting the DNA from degradation (26). In addition, these liposomes are thought to effectively neutralize the negative charges (1). The quantity of DNA required and the optimal DNA phosphate to cationic lipid ratio for efficient transfection, however, varies with cell type (27,28).
Initial experiments in this study evaluated the transfection efficiency of a mouse melanoma cell line (B16/BL6) using DO-DAC/DOPE liposomes and pCMV␤ plasmid DNA. To determine the optimal cationic liposome:DNA ratio required for this cell line, pre-formed liposomes were added to plasmid DNA at various lipid and DNA concentrations as described under "Methods." The results presented in Fig. 1 show that the transfection efficiency for this cell line was dependent on both factors. Regardless of the amount of DNA used, lipid concentrations of 50 M showed ␤-galactosidase activity of Ͻ1 milliunit/ well (Fig. 1A). The positive to negative charge ratio for these samples ranged from 1.62 to 16.23. The expression level increased more than 10-fold when 100 -300 M lipid was used for the transfection assays (up to 11 milliunits/well when the DNA concentration was 20 g/ml). Lipid concentrations of 500 M showed consistently lower levels of ␤-galactosidase activity (Ͻ7 milliunits/well) compared with 100 -300 M lipid. This was due to lipid-induced toxicity associated with high concentrations of lipid. The apparent maximum level of expression was obtained using 20 g/ml DNA and either 100 or 300 M lipid (Fig. 1B). The charge ratios for these samples were 0.81 and 2.40, respectively. It is important to note that ␤-galactosidase activity was linearly related to the amount of DNA added when using a constant liposome-DNA charge ratio of 1.62 (50 M lipid: 5 g/ml DNA; 100 M lipid: 10 g/ml DNA; and 300 M lipid: 30 g/ml DNA) giving a regression coefficient of 0.91 (data not shown).
The first steps of transfection are binding of the complex to the cell and entry of DNA into the cytoplasm. From Fig. 1, we propose that increased transfection should correlate with the amount of cell-associated DNA. DNA binding was measured under conditions comparable to those for transfection, where DODAC/DOPE liposomes were mixed with 3 H-pCMV␤ plasmid such that the final lipid and DNA concentrations were 100 M and 10 g/ml, respectively. This corresponds to a ϩ/Ϫ charge ratio of 1.62. This ratio was selected based on transfection data ( Fig. 1) that generated a transgene expression level that was midway between the minimum (Ͻ1 milliunits/well) and maximum (Ͼ10 milliunits/well) activity. This ratio also falls within the range of optimal transgene expression achieved in our laboratory using a variety of other cell lines 2 and is comparable to optimal charge ratios reported by other investigators (3). B16/BL6 cells, incubated at 37°C, were exposed to this complex in serum-free media as described under "Methods" for times ranging from 1 to 4 h. At the appropriate time interval, the complexes were removed from the cells, and the amount of 3 H-pCMV␤ in the cell lysate was measured. As a control, the amount of cell-associated DNA obtained when cells were incubated at 4°C was measured. Fig. 2A shows the amount of pCMV␤ associated with B16/BL6 cells over the 4-h transfection period at 37 and 4°C. For both temperatures, there was a time-dependent increase in the amount of DNA associated with these cells. However, this increase was less evident at 4°C than at 37°C. The difference in cell-associated DNA levels obtained at 37°C minus 4°C was defined as temperature-dependent cell-associated DNA. The data presented in Fig. 2B show that this value increased over a 3-h incubation period. the B16/BL6 cells in a temperature-dependent process. It can be suggested from these data that 60 -65% of the cell-associated DNA measured at 37°C is a consequence of binding events that occur in metabolically inactive cells maintained at 4°C. Similar results were obtained using the plasmid pInex-CATv2.0 (data not shown), which suggests that the binding of the complex to the cell is not plasmid-specific.
The assay described above is very simple and analogous to standard endocytosis and phagocytosis assays used to measure internalization of liposomes or latex beads. It is worth noting that the cationic-liposome-DNA aggregates used here are heterogeneous and unstable. It can be argued, for example, that the characteristics of aggregates may change as a consequence of incubation temperature. Studies in our laboratory, however, have demonstrated that once formed these aggregates are stable in terms of lipid components. Specifically, lipid mixing assays 3 demonstrated that significant lipid mixing occurred when DNA was added to DODAC/DOPE liposomes. Dilution in media resulted in additional lipid mixing and further aggregation. Regardless, the samples incubated with cells at 4 and 37°C are heterogeneous, and we cannot discount whether differences observed were a result of subtle changes in the attributes of the aggregates at the two temperatures. To establish whether the cell-associated DNA measured using the assay described in Fig. 2 was not a consequence of differences at 4°C, we completed a 37°C control study where the cells were treated with trypsin to dissociate liposome-DNA complexes bound through protein interactions at the cell surface. Following the 4-h transfection at 4 and 37°C, cells were washed extensively to remove unassociated DNA. Subsequently, the cells were trypsinized as outlined under "Methods." Trypsinized cells were then subjected to centrifugation on a NycoPrep gradient. This procedure was developed to separate nonspecifically bound DNA-liposome aggregates from those that are internalized. As a control for the efficiency of trypsin removal of surface-associated liposome-DNA aggregates, we evaluated cells that were measured at 4°C. Following trypsinization and separation on a NycoPrep gradient, 6.91 Ϯ 0.45 ng of 3 H-pInexCATv2.0 remained associated with the cells at 4°C, suggesting that 85% cell-associated DNA could be removed by trypsin. As shown in Fig. 3, at 37°C 23.20 Ϯ 3.51 ng of 3 H-pInexCATv2.0 was associated with B16/BL6 cells after trypsinization in comparison to 57.36 Ϯ 4.39 measured in the absence of trypsin. The level measured in the absence of trypsin was not significantly different than that shown in Fig.  2A. Furthermore, the level of cell-associated DNA recovered at 37°C after trypsin (23.20 Ϯ 3.51 ng) was not significantly different from the cell-associated DNA measured using the differences observed as a function of incubation temperature (30.90 Ϯ 8.21 ng). Although these data increase our confidence in suggesting that temperature-dependent DNA cell association is due to internalization, direct quantification of DNA uptake is difficult, and the results obtained following trypsinization of cells incubated with liposome-DNA complexes at 4°C suggest that it is difficult to achieve complete removal of surface bound liposome-DNA complexes. For this reason, we will define DNA associated with cells in these experiments as temperature-dependent cell-associated DNA.
To evaluate whether cell-associated DNA is an important indicator of optimal transfection efficiency, the amount of temperature-dependent cell-associated DNA was correlated with ␤-galactosidase activity measured following addition of different amounts of DNA to B16/BL6 cells (Fig. 4). Transfection was measured 48 h after DNA addition (Fig. 4A), and the amount of cell-associated DNA was measured at 4 h (Fig. 4B). The results demonstrate that under these conditions, the amount of ␤-galactosidase expression increased linearly (r ϭ 3 E. Wasan, unpublished observations. 0.99) with increased amounts of added DNA. As the amount of DNA added to cells increased, there was also a linear increase in temperature-dependent cell-associated DNA (r ϭ 0.88) (Fig.  4B). These results were encouraging and suggest that a simple assay measuring cell association may be used to predict transfection efficiency at least for a given cell line and plasmid construct.
The influence of liposomal lipid composition and variations in cell line were assessed to establish the general utility of this approach. Relative ␤-galactosidase activity was measured in the B16/BL6 cells 48 h following transfection with pCMV␤ complexed to three different pre-formed cationic liposome formulations as follows: DDAB/DOPE (50:50 mol %), DODAC/ DOPE (50:50 mol %), and DOTMA/DOPE (50:50 mol %). Fig.  5A shows that DODAC/DOPE liposomes complexed with pCMV␤ yielded the highest level of ␤-galactosidase activity. Expression levels achieved using DODAC/DOPE-DNA aggregates were significantly higher than achieved with either DOTMA/DOPE (p Ͻ 0.05) or DDAB/DOPE (p Ͻ 0.001) under the conditions used. Surprisingly, DNA accumulation by B16/ BL6 cells, evaluated 4 h following addition of the cationic liposome-DNA aggregates, demonstrated that the level of cellassociated DNA was not significantly different (p Ͼ 0.64) among the formulations tested (Fig. 5B). These results suggest that the different liposome formulations tested are equally effective at delivering DNA to the cells. Subsequent processing of the cationic liposome-DNA aggregate must be critical to achieve transgene expression. We believe that this intracellular processing and dissociation of the DNA-liposome complex will be dependent on liposomal lipid composition.
Efficiency of gene transfection is also dependent on the cell line used (27,28). To establish whether barriers to efficient transfection include DNA delivery, particularly in cell lines that are difficult to transfect, we evaluated ␤-galactosidase activity in cell lines derived from malignant melanomas in mouse (B16/BL6) and human (KZ). As shown in Fig. 6A, the ␤-galactosidase activity following transfection of the KZ cell line with DODAC/DOPE-pCMV␤ aggregates was not detectable. In contrast, the ␤-galactosidase activity in the cell line derived from mouse melanoma peaked around 5 milliunits/well when the lipid concentration was between 100 and 200 M and the DNA was 10 g/ml. To determine whether the poor transfection of KZ cells was a consequence of inefficient DNA delivery, we measured temperature-dependent cell-associated DNA levels using these two cell lines. Fig. 6B shows that there was no significant difference in cell-associated DNA levels between the KZ cells and B16/BL6 cells (p Ͼ 0.62) when measured at the 4-h time point. These results suggest that 8% of the added pCMV␤ was accumulated in the KZ cells. This is equivalent to All data are averaged from three replicates and expressed Ϯ S.E. Statistical analysis of ␤-galactosidase activity was evaluated by analysis of variance and showed that DODAC/DOPE exhibit a significantly higher level than DDAB/DOPE (p Ͻ 0.001) and DOTMA/DOPE (p Ͻ 0.05). The charge ratio (Ϯ) for each of these liposome-DNA aggregates is 1.62. Statistical analysis of cell-associated DNA among the liposome formulations was evaluated by analysis of variance and showed no significant differences (p Ͼ 0.64). approximately 10 6 plasmids associated per cell compared with 15% (approximately 2 ϫ 10 6 plasmids) for B16/BL6 cells.
The DNA delivery assay used here measures the quantity of 3 H-plasmid DNA associated with cells in vitro. It gives no information, however, about whether the DNA is intact, and this could account for the differences in transfection efficiencies observed for the KZ and B16/BL6 cells lines. To determine this, the integrity of the isolated, cell-associated pInexCATv2.0 DNA was evaluated by agarose gel electrophoresis. The pInex-CATv2.0 DNA was used instead of pCMV␤ because this plasmid could easily be resolved from genomic DNA on an agarose gel. DNA-liposome complexes were prepared at a lipid:DNA charge ratio of 1.62:1 and a DNA concentration of 10 g/ml. The pInexCATv2.0 DNA has a size of 4.5 kb and is small enough such that the three DNA conformations (linear, relaxed, and supercoiled) can be readily separated from genomic DNA on an agarose gel. Furthermore, the amount of DNA in these bands could be quantified using radiolabeled DNA as a tracer. Transfections were executed as described under "Methods," and 24 h later the cells were washed, lysed, and the DNA extracted. DNA obtained from cells transfected with cationic liposome-DNA aggregates or DNA that had not been complexed to liposomes (used as a control) were subjected to electrophoresis on a standard 0.8% agarose gel and stained with ethidium bromide. Fig. 7 shows that a portion of the DNA recovered from B16/BL6 and KZ cells transfected with DODAC/DOPE-pInexCATv2.0 liposome aggregates contained intact plasmid 24 h following transfection as evidenced by the appearance of the three bands on the gel (lanes 1 and 3). DNA that was cell-associated following addition of free plasmid, however, did not reveal the presence of any bands, suggesting that associated plasmid was not intact.
The amount of intact DNA was semi-quantified by excising the plasmid bands from the gel and measuring the radioactivity by liquid scintillation counting. This analysis was completed following transfection of the B16/BL6 and KZ melanoma cell lines, and the results are shown in Table I. These data suggest for cells incubated with DODAC/DOPE liposome-DNA aggregates that greater than 60% of the 3 H counts (650 -750 dpm) recovered from B16/BL6 and KZ were associated with the plasmid bands. The remainder of the counts were distributed throughout the lane and were found in association with the genomic DNA, small degraded fragments, and as DNA bound to protein retained in the wells. In contrast, as much as 90% of the radioactivity associated with B16/BL6 and KZ cells incubated with free DNA was degraded. It is important to point out that in comparison to the number of counts loaded onto the gel following transfection with liposome-DNA aggregates (1700 dpm), 400 dpm was loaded onto the gel when using cells transfected with free DNA. These data suggest that in the two cell lines that differ in ability to be transfected, DNA uptake is similar, and the proportion of intact DNA 24 h after transfection is comparable. The assay system used does not, however, allow discrimination between DNA bound at the cell surface and that internalized by the cell. It is possible, therefore, that the measure of DNA integrity reflects the stability of surfaceassociated DNA rather than internalized DNA. DISCUSSION If optimal non-viral DNA transfer formulations are going to prove to be therapeutically useful, then the important parameters that control DNA delivery and transgene expression must be elucidated. This is, however, an extremely complicated proc- ess, and studies attempting to define the mechanism(s) will likely work using a process of elimination. Transgene expression will be dependent on (i) the plasmid used and the gene selected, (ii) DNA delivery, (iii) rates and extent of intracellular DNA delivery to the nucleus, (iv) differences in cationic liposome-DNA dissociation rates, and (v) differences in DNA stability. These factors do not encompass concerns about mRNA stability, post-translational modification of the transgene product, or stability of the protein being expressed. Obviously, this is not a simple technology. The studies described here focus on DNA delivery, the first in a series of critical events that ultimately define transgene expression levels. It can be argued that DNA delivery is the event most likely controlled by the DNA transfer formulations, and provided that the plasmid and transgene product can be processed and expressed in the target cell, the liposome-DNA transfer formulations should be optimized on the basis of delivery characteristics. The results reported here were, therefore, surprising in that efficient delivery did not necessarily predict transfection activity. In this context, cell delivery must be defined in more specific terms, and these include delivery of the plasmid DNA in a form that is functionally active.
Although cationic liposome-DNA aggregates are used for gene transfer, the physical and chemical characteristics of these transfecting vectors are not well characterized. Freeze fracture microscopy of these aggregates for example reveals a plethora of structures (8). These include "spaghetti-meatball" structures that are thought to consist of DNA connected by semi-fused liposomes and tubule-like structures that may represent lipid-coated DNA. However, it is unclear which form of the complex is required for gene transfer. The first steps toward achieving useful plasmid delivery are association of the DNA-liposome aggregate with the cell membrane and entry of the DNA and, perhaps, associated lipid into the cell. The results generated from this study (Fig. 2) show that the amount of cell-associated DNA at 37°C is greater than that observed at 4°C. The significant binding at 4°C suggests that these liposome-DNA aggregates are bound efficiently to cells in a nonspecific manner, and this must contribute substantially to the DNA delivery process. Furthermore, we were unable to completely remove associated DNA by treatment with trypsin or extensive washing suggesting that this association is reasonably strong.
The role of liposomal lipid composition in influencing transfection efficiency may be related to promoting effective DNA/ cell membrane interaction. This may, in turn, be an important factor in designing effective lipid-based DNA delivery systems (3). However, the degree to which aggregation occurs following addition of DNA to cationic liposomes varies with different liposomal lipid composition (6), and the lipid structures generated as a consequence of the aggregation events are also dependent on the cationic liposomes used. Differences in lipid composition did not, however, appear to influence the level of DNA delivery in vitro. These results suggest that the lipid composition may play a more important role in terms of effecting the ability of DNA to be processed following binding and entry into the cell. It has been suggested that the strength of the ionic/hydrophobic interactions of the lipid to the DNA will affect DNA stability as well as the manner in which DNA is directed to the nucleus and ultimately expressed (1,4,9,29). With regard to cationic liposome-mediated protection of DNA, the sensitivity of internalized liposome-DNA aggregates to intracellular degradation may be dependent on lipid composition as well as cell line. In these studies, DNA degradation following addition of free plasmid to cells was significant in the cell lines tested. DNA delivered using DODAC/DOPE-DNA aggregates was shown to be significantly more stable (Fig. 7); however, we were unable to assess whether the cell-associated DNA was internalized or bound to the cell surface. Whether the stability of these aggregates is a function of lipid composition remains to be elucidated. Although the protection against enzymatic digestion is important, we believe a related parameter will involve dissociation of the lipid from DNA. It may, therefore, be necessary to derive lipid-based carriers that maintain strong ionic/hydrophobic interactions yielding complexes that are stable against circulating enzymes, yet allow dissociation of the lipid from the DNA after entry into the cell. We propose that evaluating DNA uptake as well as assessing complex stability will aid in developing formulations that exhibit such attributes.
Although lipid composition may play a role in efficient expression of the transgene, it is not unreasonable to suggest that different cell types will vary in the machinery used for expression of the transgene. For example, some cells may have higher levels of enzyme/proteins important in regulating DNA transport through the cell, transcription, stability of the message, and processing of the transgene product. Clearly, transfection in vitro and in vivo may not depend entirely on cellular delivery but on the inefficient expression of foreign DNA. Researchers have begun to investigate the role of specific promoters and enhancers in plasmid DNA on levels of transgene expression (30,31). How these new constructs influence the expression of the transgene remains to be elucidated and may still prove to be cell type-dependent.