Cationic Liposomes Enhance the Rate of Transduction by a Recombinant Retroviral Vector In Vitro and In Vivo

Abstract

Cationic liposomes enhanced the rate of transduction of target cells with retroviral vectors. The greatest effect was seen with the formulation DC-Chol/DOPE, which gave a 20-fold increase in initial transduction rate. This allowed an efficiency of transduction after brief exposure of target cells to virus plus liposome that could be achieved only after extensive exposure to virus alone. Enhancement with DC-Chol/DOPE was optimal when stable virion-liposome complexes were preformed. The transduction rate for complexed virus, as for virus used alone or with the polycation Polybrene, showed first-order dependence on virus concentration. Cationic liposomes, but not Polybrene, were able to mediate envelope-independent transduction, but optimal efficiency required envelope-receptor interaction. When virus complexed with DC-Chol/DOPE was used to transduce human mesothelioma xenografts, transduction was enhanced four- to fivefold compared to that for virus alone. Since the efficacy of gene therapy is dependent on the number of cells modified, which is in turn dependent upon the balance between transduction and biological clearance of the vector, the ability of cationic liposomes to form stable complexes with retroviral vectors and enhance their rate of infection is likely to be important for in vivo application.

While retroviruses derived from murine leukemia virus (MLV) have been widely used as recombinant vectors for gene transfer, including clinical gene therapy, parameters which control the kinetics of infection are still poorly understood. Infection is dependent upon viral envelope interaction with a specific cell surface receptor protein to trigger envelope-mediated fusion (34). However, other factors also control the rate of infection. Wang et al. (33) described inefficient depletion of Moloney ecotropic MLV (MLV-E) from target cell medium and reported that the multiplicity of infection (MOI) was dependent on virus concentration but independent of target cell number. They concluded that virion adsorption was the limiting step of the infection process. Transduction with an amphotropic MLV (MLV-A) vector was also dependent on virus concentration rather than virion-cell multiplicity (26). It has been argued that the Brownian motion of retroviral particles in the medium imposes a significant limitation for infection of adherent NIH 3T3 target cells (7). In contrast, the binding kinetics of MLV-E vectors to NIH 3T3 cells in suspension has been shown to fit a bimolecular, noncooperative model with rapid attainment of equilibrium at 37°C when virus was in excess (36). The association rate constant was significantly lower than the calculated limitation imposed by viral diffusion in these experiments, suggesting that binding rather than encounter is rate limiting. The use of polycations, such as Polybrene, is standard in many retroviral infection protocols owing to early observations of improved infection efficiency (32). Its mechanism of action is thought to involve neutralization of electrostatic repulsion between virion and cell membranes allowing enhanced attachment.

In addition to cell adsorption, the postbinding efficiency of retroviral infection is poor. A careful study of binding, internalization, and degradation of C57 MLV-E at low virus particle-to-cell ratios showed that these processes were identical for physical particles and infectious virions although only 1 to 2% of bound particles were infectious (3). All steps were first-order with respect to virus concentration and showed half-times of 2 to 3 h. The low efficiency was partly accounted for by the retention of 15% of virions at the cell surface and the degradation of 75% of internalized virions. The use of lysosomotropic agents, such as chloroquine, demonstrated that passage through an acidic compartment following internalization was necessary for infection as well as responsible for degradation. For MLV, this pH dependence of infectivity has since been shown to be peculiar to MLV-E (24). Therefore, the current belief is that MLV-E infection involves internalization of intact virions by endocytosis, followed by a membrane fusion event releasing virion cores to the cytoplasm, with degradation of virions that fail to fuse (1–3, 28).

Apart from intrinsic inefficiencies of retroviral infection, gene delivery to target cells by recombinant retroviral vectors in vivo may be subject to further limitations. Factors likely to be important include vector stability (31), removal of vector from the target site due to blood flow, nonspecific adsorption to inappropriate cells, and clearance by cells of the immune system. We reasoned that efficient in vivo gene transfer would require rapid transduction since the time of exposure of target cells to virus would be limited. We therefore considered the use of cationic liposomes in combination with retroviral vectors to enhance their rate of infection. Previous studies have shown that cationic liposomes can enhance the infectivity of human immunodeficiency virus (HIV) (22, 23) and increase MLV vector titers after extensive exposure of cells in vitro (18). However, neither the influence of such agents on the kinetics of infection nor their application to in vivo transduction has been reported.

Since the initial description of lipofection as a means of DNA transfection (12, 13), cationic liposomes have been widely used for delivery of nucleic acids, both in vitro and in vivo (11, 14, 15, 19, 37, 38). Such liposomes are unilammelar, typically of the order of 100 to 200 nm in diameter. Various complexes form when DNA is added, notably the so-called “meatball-and-spaghetti” structures, in which DNA is bound to the liposome surface or encased in a bilayer tubule (29, 37). The efficiency of transfection is dependent on the preincubation time and DNA-to-lipid ratio, suggesting that only a subset of these structures may be transfection competent (17, 29). Other polycations can be incorporated to increase condensation of the DNA (16). The overall “mass-action” process is inefficient, involving endocytosis as the major entry mechanism, endosome release, dissociation of DNA from lipid, and nuclear entry (37), and the delivered DNA is only transiently retained by the target cell. Membrane fusion or destabilization can be enhanced by the neutral helper lipid dioleoyl phosphatidylethanolamine (DOPE) (10, 14). Other cationic liposomes that do not contain DOPE are presumably intrinsically fusigenic (14). Nonlipid polycations have also been used for DNA transfection, including poly-l-lysine (PLL) (25) and Polybrene (coupled with dimethyl sulfoxide permeabilization) (4). Cell surface proteoglycans, particularly heparan sulfate, play a role in cation-mediated transfection with PLL and some liposomes (25). Here we demonstrate that cationic liposomes enhance stable target cell transduction by an amphotropic retroviral vector both in vitro and in vivo.

MATERIALS AND METHODS

Cell lines.

Murine NIH 3T3, human rhabdomyosarcoma TE671, and human mesothelioma H-Meso-1 (27) cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum (FCS).

Helper-free MFGnlsLacZ pseudotypes.

TElac2 is a clone of TE671 harboring the MFGnlsLacZ retroviral vector (30). TELCeB6 is a derivative of TElac2 expressing Moloney MLV (Mo-MLV) Gag-Pol proteins (8). Pseudotypes of MFGnlsLacZ were obtained from clonal TELCeB6-derived producer lines expressing the envelope protein for amphotropic strain 4070A MLV (MLV-A), feline endogenous virus RD114, or ecotropic Mo-MLV (MLV-E) from plasmids AF and RDF (8) or an equivalent construct expressing Mo-MLV env (7a). The helper-free status of these producer cells has been reported (8).

Serum-free culture supernatants containing pseudotyped MFGnlsLacZ virus were obtained after overnight incubation in OptiMEM, filtered (pore size, 0.45 μm), and frozen at −80°C before use. Virus titers, determined by counting colonies of histochemically positive cells following infection with viral dilutions in the absence or presence of 8 μg of Polybrene per ml, were 2 × 10⁶ and 1 × 10⁷ CFU/ml (absence and presence, respectively) for MLV-A and 4 × 10⁵ and 1 × 10⁶ CFU/ml for RD114 (both values obtained with TE671 cells), and 6 × 10⁶ and 1 × 10⁷ CFU/ml for MLV-E (obtained with NIH 3T3 cells).

Infection assays and determination of MOI.

Target cells were seeded at 2.5 × 10⁴ cells/well in 24-well plates. At 2 to 3 days after infection, the cells were histochemically stained for β-galactosidase expression by being washed with phosphate-buffered saline (PBS), fixed for 15 min at room temperature with 0.5% glutaraldehyde in PBS, washed again with PBS, and incubated for 4 h at 37°C in PBS–2 mM MgCl₂–0.01% sodium deoxycholate–0.02% Nonidet P-40–5 mM potassium ferricyanide–5 mM potassium ferrocyanide–0.1% X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside). The total number of cells and the number of blue cells were counted in at least five microscopic fields of view: for low levels of transduction, blue cells were counted from up to 100 fields to determine the proportion of positive cells.

The MOI was determined from the proportion of uninfected cells (P) according to the formula MOI = −(log P)/0.4343, derived from the following statistical argument. The number of viruses leading to successful transduction events (n_v) as a fraction of the number of target cells (n_c) is the multiplicity of infection (MOI = n_v/n_c). The probability of any cell being infected by one of these viruses is 1/n_c, and thus the probability of any cell remaining uninfected (P) is [1 − (1/n_c)] raised to the power n_v. Thus, log P = n_v · log [1 − (1/n_c)]. For n_c greater than 3 × 10³, n_c · log [1 − (1/n_c)] = −0.4343. Consideration of MOI corrects for the nonlinearity of a transduction assay based on a simple positive-negative scoring.

Determination of β-galactosidase specific activity.

In some experiments, the β-galactosidase activity in cell lysates was determined by a quantitative photometric assay (11). Cells were washed in PBS and lysed in 350 μl of 250 mM Tris.HCl (pH 8.0)–0.1% Triton X-100, and the lysates were stored at −80°C for at least 1 h. To determine amounts of β-galactosidase, 50 μl of lysates and dilutions were combined with 50 μl of PBS–0.5% bovine serum albumin and 150 μl of 60 mM Na₂HPO₄ (pH 8.0)–1 mM MgSO₄–10 mM KCl–50 mM β-mercaptoethanol–0.1% chlorophenol red galactopyranoside (Boehringer). After light-protected incubation at 37°C, absorption at 578 nm was measured and converted to picograms of β-galactosidase by using a standard curve obtained with purified enzyme (Sigma). These values were standardized against the total amount of protein in each lysate. A 10-μl volume of lysate was combined with 150 μl of water and 40 μl of Bio-Rad protein assay reagent. Absorption at 595 nm was converted to micrograms of protein by using a bovine serum albumin standard curve. Specific activity was expressed as picograms of β-galactosidase per microgram of protein. Statistical comparisons were performed by Student’s t test, and the null hypothesis was rejected at P < 0.05.

Cationic liposomes.

The following cationic liposomes were used in this study. N-[1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethylammonium methylsulfate (DOTAP) was obtained from Boehringer Mannheim. A 3:1 (wt/wt) formulation of 2,3-dioleoyloxy-N-[2(spermine-carboxamido)ethyl]-N,N-dimethyl-1-propanaminium trifluoroacetate (DOSPA) and neutral DOPE (Lipofectamine) was obtained from Gibco-BRL. A 3:2 (molar ratio) formulation of 3β[N-(N′,N′-dimethylaminoethane)=carbamoyl]cholesterol (DC-Chol) and DOPE (15) was a kind gift from L. Huang, Pittsburgh, Pa. All liposomes were stored at 4°C. DOTAP was supplied bottled under argon and was used fresh after opening. DC-Chol/DOPE was stable, yielding similar transduction rate enhancements after storage for 1 year.

Basic protocol for transduction with virus-liposome complexes.

For standard conditions, the liposome reagent was diluted to 50 μg/ml in OptiMEM. Equal volumes of virus and liposome were combined and incubated for 30 min at room temperature in polystyrene tubes or multiwell plates. A 100-μl volume of this mixture was added to target cells in 24-well plates, plated the day before at 2.5 × 10⁴ cells/well, after replacement of the medium with 250 μl of OptiMEM. After incubation at 37°C for various times, the reagents were removed by aspiration and replaced with 1 ml of DMEM–10% FCS. Thus, the final liposome concentration was 2.5 μg/350 μl; at this concentration, there was some toxicity for extended exposure to DOTAP. For comparison, the cationic nonliposomal reagent Polybrene (hexadimethrine bromide) was diluted to give a final concentration of 8 μg/ml and used by the same method. As a control, virus was mixed 1:1 with OptiMEM. The in vitro transduction data presented in Fig. 1 to 4 were confirmed by at least one other independent experiment.

FIG. 1.

Transduction rate enhancement with DC-Chol/DOPE. (A) MLV-A vector (diluted 1:10) was preincubated for 30 min at room temperature with an equal volume of OptiMEM containing no agent, 50 μg of DC-Chol/DOPE (DC-Chol) per ml, or Polybrene (final concentration, 8 μg/ml) and added to TE671 target cells. After the indicated exposure time at 37°C, the reagents were replaced with medium. Two days later, histochemically lacZ-positive cells were scored as a percentage of the total cells. (B) The data in panel A were converted to MOIs enabling a more quantitative assessment of transduction efficiencies.

FIG. 4.

Comparison of transduction rate enhancement by different cationic liposomes for different virus-target cell combinations, and their use for envelope-independent transduction. (A) MLV-A, RD114, and MLV-E vectors were serially diluted and preincubated with OptiMEM (No agent), 50 μg of DC-Chol/DOPE (DC-Chol), DOTAP, or Lipofectamine per ml, or Polybrene (final concentration, 8 μg/ml) before exposure of TE671 or NIH 3T3 target cells for 40 min, as in Fig. 1. (B) Nonenveloped vector (derived from TELCeB6 cells) and MLV-E or RD114 vectors were preincubated with the reagents described above and used to transduce TE671 and 3T3 target cells in nonpermissive combinations by exposure for 4 h.

In vivo transduction of the H-meso1 xenograft.

H-Meso1 cells (27) were delivered by injection to the peritoneal cavity of 5- to 6-week-old female athymic nude (nu/nu) mice at 10⁷ cells per mouse in 500 μl of PBS. Undiluted MLV-A virus was preincubated with an equal volume of 50 μg of DC-Chol/DOPE per ml or OptiMEM control, as described above, and 500 μl of this mixture was similarly delivered 2 days after injection of the tumour cells. Five animals received MLV-A with DC-Chol/DOPE, and five received MLV-A alone; control mice received OptiMEM only. The animals were sacrificed on day 14, and ascites and tumor cells from each mouse were reestablished in vitro for determination of the proportion of transduced cells and the β-galactosidase specific activity. Ascites cells were flushed from the peritoneal cavity, washed in PBS, and cultured in DMEM–10% FCS for 2 days before being subjected to histochemical staining or lysis. Solid tumor deposits were explanted, washed in PBS, minced, and disaggregated by digestion with 1 μg of DNase I per ml–100 U of collagenase II per ml–100 U of collagenase IV per ml for 1 h at room temperature. Cells were decanted, washed, and cultured as for ascites. Additionally, pieces of solid tumor were lysed directly for determination of β-galactosidase specific activity.

RESULTS

Transduction rate enhancement with DC-Chol/DOPE.

The rate of transduction of human TE671 cells with an MLV-A enveloped viral vector carrying a lacZ gene is shown in Fig. 1A. Preincubation of vector with DC-Chol/DOPE or Polybrene enhanced the rate of infection. These primary data are presented as the proportion of lacZ-positive cells. While for practical applications this is a relevant measurement, to discuss infectivities quantitatively it is necessary to calculate the MOI, as described in Materials and Methods. At low levels of infection, the MOI is essentially the same as the positive fraction. However, the correspondence is not linear at high efficiencies due to the inability of the histochemical assay to distinguish singly and multiply infected cells. The data in Fig. 1A transformed to MOI are shown in Fig. 1B.

The rate of transduction under each condition was linear for the first 40 min. Comparison of these initial rates of transduction showed increases of approximately 10- and 20-fold for Polybrene and DC-Chol/DOPE, respectively. To enable a statistical assessment of this enhancement, cells were exposed for 40 min to virus preincubated as above in multiple replicate wells. Parallel replicates were used for determination of the MOI and β-galactosidase specific activity, analysis of genomic DNA to check proviral copy number and integrity, and passage to monitor the stability of transduction. Histochemical staining showed rate increases of 10- and 15-fold with Polybrene and DC-Chol/DOPE, respectively, with parallel increases in β-galactosidase specific activity (Table 1, expt. 1). The greater enhancement with DC-Chol/DOPE than with Polybrene was significant (P < 0.002) and has been consistently observed. A similar correspondence was also obtained in a second experiment (Table 1, expt. 2). The parallel enhancements of MOI and β-galactosidase specific activity suggest that DC-Chol/DOPE increases the number of cells infected by single virions. Southern blot analysis showed the expected proviral integrant fragments for virus used with Polybrene or DC-Chol/DOPE, with increased intensity for the latter, but no signal was detected for virus used alone (as expected for MOI < 0.05) (data not shown). Following extended culture of the transduced cells, the proportion of positive cells remained similar over 10 passages (data not shown). These data show that both Polybrene and DC-Chol/DOPE enhance the rate but do not alter the basic properties of retroviral transduction.

TABLE 1.

MOI and specific activity^a

Condition	MOIb		β-Galactosidase sp act (pg/μg)b
	Expt. 1 (n = 6)	Expt. 2 (n = 3)	Expt. 1 (n = 6)	Expt. 2 (n = 5)
Uninfected control	0	0	−2.8 ± 1.4	1.5 ± 1.2
Virus alone	0.044 ± 0.0021	0.038 ± 0.0047	19 ± 2.9	40 ± 2.1
Virus + Polybrene	0.46 ± 0.022	0.28 ± 0.013	250 ± 33	270 ± 17
Virus + DC-Chol/DOPE	0.65 ± 0.014	0.34 ± 0.015	490 ± 45	380 ± 43

^aTE671 target cells were exposed for 40 min to virus preincubated with OptiMEM, Polybrene, or DC-Chol/DOPE. Replicates were used for histochemical staining to derive MOIs or for determination of β-galactosidase specific activity. 

^bValues are mean ± SE of n replicates for two independent experiments. 

Dependence of transduction rates on virus and DC-Chol/DOPE concentrations.

Dilutions of virus were preincubated with DC-Chol/DOPE or Polybrene, before exposure to target cells for 40 min, to determine the initial transduction rate (MOI/40 min). This rate was proportional to the virus concentration for dilutions greater than 1:10, with some saturation evident for the 1:3.16 dilution (Fig. 2A). Rate enhancement with DC-Chol/DOPE was greater than that with Polybrene at all virus dilutions. The dose dependence for DC-Chol/DOPE of the transduction rate was also determined by preincubating virus with 2.5 μg of DC-Chol/DOPE (the standard condition) and various dilutions down to 2.5 ng (Fig. 2B). The rate was not enhanced over that of the OptiMEM control for this lowest dilution. The sigmoidal dose-response curve is a reflection of no absolute dependence on DC-Chol/DOPE for transduction and of saturation for the highest doses. However, the rate enhancement showed first-order dependence on the DC-Chol/DOPE concentration over the range 0.025 to 0.25 μg. A higher concentration of DC-Chol/DOPE was required to achieve the maximum rate enhancement for the most concentrated virus.

FIG. 2.

Dependence of transduction rate enhancement upon virus dilution and DC-Chol/DOPE dose. (A) Serially diluted virus was preincubated as in Fig. 1A, and target cells were exposed for 40 min to enable comparison of the initial transduction rates (MOI/40 min). (B) MLV-A vector (diluted 1:10, 1:100, or 1:1,000) was preincubated for 30 min at room temperature with serially diluted DC-Chol/DOPE (50 ng/ml to 50 μg/ml), and TE671 target cells were exposed for 40 min at 37°C to enable comparison of the initial transduction rates. No enhancement above that obtained with OptiMEM was observed with the lowest DC-Chol/DOPE dose used. For presentation purposes, the values obtained with 1:100 and 1:1,000 dilutions are multiplied by 10 and 100, respectively.

The linear dependence of transduction on virus concentration demonstrates that single virions, rather than aggregates, are responsible for DC-Chol/DOPE-enhanced infection. In agreement with this, the enhanced infectivity of virus combined with DC-Chol/DOPE was efficiently recovered following filtration through a 0.45-μm-pore-size filter. This conclusion is consistent with the correspondence of enhancements in terms of MOI and β-galactosidase specific activity (see above). The linear dependence of transduction on DC-Chol/DOPE concentration further suggests that a virion-liposome complex with a particular stoichiometry is required for optimal transduction.

Evidence for stable complex formation was also provided by the linear dependence of the transduction rate when the virus and DC-Chol/DOPE were mixed before dilution. The rate showed first-order dependence with respect to virus concentration, as with dilution prior to addition of DC-Chol/DOPE (Fig. 2A). In contrast, when the virus and DC-Chol/DOPE were independently diluted before being combined, the rate showed higher-order dependence and the enhancement diminished at dilutions of >1:100 (data not shown). Virion-liposome complex formation was rapid, with only a marginal (1.3-fold) improvement of transduction with preincubation of virus and liposome for times of 5 min up to 1 h, compared to immediate addition of the virus-liposome mixture to cells (data not shown). This is in contrast to the marked preincubation time dependence of transfection efficiency when DC-Chol/DOPE is used with plasmid DNA, which is likely to be a reflection of changes in the nature of the structures formed (17, 29). Additionally, enhancement was slightly improved (1.4-fold) by the presence of 10% FCS during preincubation with DC-Chol/DOPE and unaffected by FCS during exposure of target cells.

Order of addition of virus and liposome to target cells.

Next, the optimal order of combining virus, DC-Chol/DOPE or Polybrene, and target cells was determined. When cells were pretreated with DC-Chol/DOPE followed by medium replacement and addition of virus, the transduction rate was reduced fourfold compared to preincubation of virus and DC-Chol/DOPE (Fig. 3A). Addition of virus after pretreatment of cells but without removal of DC-Chol/DOPE gave only a twofold reduction. Thus, although the enhancement observed may be in part due to an effect of the liposomes on the target cells, the optimal conditions require the preformation of the virion-liposome complex. This result contrasts with that observed for Polybrene, for which pretreatment of cells was optimal (Fig. 3A). Enhancement by preincubation of virus with DC-Chol/DOPE and Polybrene was not additive but intermediate between that with either agent alone, which suggests competition for virion association.

FIG. 3.

Order of combination and effects on virus internalization. (A) Transduction rates (MOI/40 min) were determined for preincubation of 1:10-diluted MLV-A vector with OptiMEM containing no agent (−), DC-Chol/DOPE (DC), Polybrene (PB), or a mixture of the two (DC + PB). Additionally, to evaluate the optimal order of reagent addition, target cells were pretreated with the same final concentrations of DC-Chol/DOPE (2.5 μg/350 μl) or Polybrene (8 μg/ml) for 30 min at 37°C, which was subsequently replaced with OptiMEM. Virus was added 10 min later for a 40-min exposure time. Values are mean ± SE for triplicate determinations. (B) Undiluted MLV-A vector was added to target cells on ice for 80 min (Bind virus). The cells were washed with cold OptiMEM and incubated for a further 1 h on ice in OptiMEM before being washed again to remove nonspecifically bound virus. Warm OptiMEM containing no agent (−), DC-Chol/DOPE (2.5 μg/350 μl), or polybrene (8 μg/ml) was added, and the cells were incubated for 2 h at 37°C to enable internalization of bound virus. Additionally, undiluted virus was preincubated with DC-Chol/DOPE or Polybrene for the same final concentrations and the resulting complex was cooled and added to target cells on ice for 80 min (Bind complex). After being washed as above, the cells were incubated for 2 h at 37°C in OptiMEM to enable internalization of bound virus complexes. Values are mean ± SE for triplicate determinations. (C) Undiluted (neat) and 1:10-diluted MLV-A vector, preincubated with OptiMEM containing no agent, DC-Chol/DOPE (DC), or Polybrene (PB), was added to target cells. After incubation for 5 min at 37°C, the virus was removed and replaced with medium or 0.05% trypsin and 0.02% EDTA (trypsin). Cells treated with trypsin were incubated 10 min at 37°C, after which they were recovered, washed, and replated. Values are mean ± SE for triplicate determinations.

DC-Chol/DOPE had no effect when added to cells after virus binding, while addition of Polybrene gave only a twofold enhancement (Fig. 3B). When preformed virus complexes were bound to cells at 0°C, substantial transduction enhancement relative to the use of uncomplexed virus was seen after the cells were warmed to 37°C (Fig. 3B). This should be a reflection of the total amount of transduction-competent virus binding to the target cells. Enhancement was 6.7- and 8.6-fold for Polybrene and DC-Chol/DOPE, respectively, compared to overall rate increases at 37°C of 10- and 15-fold in parallel experiments (MOI/40 min; Table 1, expt. 1).

To measure viral internalization, cells were exposed to virus for 5 min and then treated with trypsin-EDTA for 10 min. Transduction should be a reflection of internalization of virus in the initial 5 min, since trypsin inactivates virions that have not internalized (3). Trypsinisation reduced the rate of infection 8- to 10-fold for virus alone or with Polybrene and 4-fold for virus with DC-Chol/DOPE (Fig. 3C). The internalization kinetics of virus bound to cells at 0°C with or without either agent, measured by trypsinization of target cells at time intervals following addition of warm OptiMEM, were not significantly different, with half-maximal transduction achieved within 30 min (data not shown). These data suggest that internalization at 37°C is not greatly affected by Polybrene or DC-Chol/DOPE.

Use of other cationic liposomes and other virus-target cell combinations.

To test the generality of transduction enhancement, TE671 and NIH 3T3 cells were infected by a retroviral vector with an MLV-A envelope, NIH 3T3 cells were infected by a retroviral vector with an MLV-E envelope, and TE671 cells were infected by a retroviral vector with an RD114 envelope. In each case, the virus was preincubated with 50 μg of DC-Chol/DOPE, DOTAP, or Lipofectamine per ml. Polybrene was included for comparison. This dose of DC-Chol/DOPE was determined previously to give maximal enhancement for MLV-A infection of TE671 cells (see above). For DOTAP, this dose was also close to the optimum for virus used at a 1:10 dilution. However, subsequent experiments showed that this concentration of Lipofectamine was suboptimal and that when an optimal dose was used, efficiencies similar to that for DC-Chol/DOPE could be obtained. Complexes were formed with serially diluted virus and then applied to target cells for 40 min to determine the rate of infection, which was linearly dependent on virus concentration in most cases (Fig. 4A). The magnitude of the enhancement for NIH 3T3 target cells was less than that for TE671: the greatest effect was with DC-Chol/DOPE, which gave enhancement of approximately fourfold for MLV-A and threefold for MLV-E. For RD114 infection of TE671 cells, both DC-Chol/DOPE and Polybrene gave a 10-fold enhancement. A higher-order dependence on virus concentration was shown by DOTAP and, to a lesser degree, Lipofectamine. This was most pronounced for NIH 3T3 target cells: DOTAP had no effect on transduction of these cells by undiluted MLV-E but significantly reduced transduction by diluted virus (Fig. 4A, panel 4).

Dependence of transduction enhancement on envelope-receptor interaction: cationic liposome-mediated envelope-independent transduction.

The high transduction efficiencies attainable with virus complexed with DC-Chol/DOPE were dependent upon a cognate envelope-receptor interaction. However, reduced but significant transduction independent of such an interaction was also possible. Transduction of TE671 target cells, lacking a functional ecotropic virus receptor could be achieved with MLV-E or with nonenveloped viral particles derived from TELCeB6 (Fig. 4B, panel 1). This effect was also shown by particles complexed with DOTAP and, much less efficiently, with Lipofectamine. Polybrene did not enable transduction in this envelope-independent fashion, indicating a different mechanism of action compared to cationic liposome-mediated transduction. NIH 3T3 target cells, at lower efficiency, could similarly be transduced with RD114 or nonenveloped particles, although in this case DC-Chol/DOPE was comparatively ineffective (Fig. 4B, panel 2). Envelope-independent transduction was shown to require virus particles, since supernatant from TElac2 cells, lacking retroviral gag-pol expression, did not transduce either target cell in the presence of DOTAP.

Treatment of TE671 cells with nonenveloped particles com plexed with DOTAP resulted in stable transduction, with the proportion of positive cells following 20 passages being the same as that without passage (data not shown). The transduction rate with this combination (MOI = 0.05 in 40 min) was comparable to a 1:10 dilution (2 × 10⁵ CFU/ml) of uncomplexed virus (Fig. 4A, panel 1) and was 50- to 200-fold lower than that obtained with liposome-complexed MLV-A. Evidence of stable complex formation between particles and DOTAP was obtained in that infectivity was retained following sucrose density gradient centrifugation (data not shown).

DC-Chol/DOPE-mediated in vivo transduction of xenografted H-Meso1 cells.

The ability of DC-Chol/DOPE to enhance transduction of target cells in vivo was tested in a nude-mouse xenograft model of mesothelioma following intraperitoneal delivery of human H-Meso1 cells. These cells grew both as ascitic cells and as solid tumors. MLV-A virus, alone or complexed with DC-Chol/DOPE, was administered to the peritoneal cavity 2 days following tumor establishment. Virus delivery had no effect on the growth rate of the tumors. Ascitic and solid tumor cells were reestablished in vitro. Histochemical staining of ascitic and tumor cells revealed transduction efficiencies of 40% ± 3% and 44% ± 9%, respectively, with liposome, compared to 12% ± 3% and 17% ± 5% without (mean ± standard error). Comparing MOI, this reflects increases of 3.9- and 3.3-fold (Fig. 5A). The β-galactosidase specific activity of solid tumor deposits removed after sacrifice on day 14 was enhanced 5.0-fold (Fig. 5B), which was highly significant (P < 0.0007). Additionally, β-galactosidase specific activity for the reestablished ascitic and tumor cells showed increases of 4.6- and 3.4-fold, respectively. Control tumors showed no detectable enzyme activity.

FIG. 5.

In vivo transduction efficiency of xenografted H-Meso1 cells is enhanced with DC-Chol/DOPE. MLV-A vector was preincubated with OptiMEM (No agent) or DC-Chol/DOPE and delivered intraperitoneally to nude mice bearing intraperitoneal H-Meso1 xenografts. (A) Transduction efficiencies were determined by staining cells reestablished at sacrifice in vitro from both ascites and solid tumor. (B) Lysates (five replicates for each mouse) of solid tumor deposits were used for determination of β-galactosidase specific activity. The specific activity was also determined for lysates of ascites and solid tumor cells reestablished in vitro at sacrifice. Significance levels for differences with and without liposomes are given below. All values are mean ± SE for five mice.

DISCUSSION

The data presented here show that both Polybrene and DC-Chol/DOPE enhanced the rate of retroviral transduction. The initial rate enhancement was approximately 10- and 20-fold, respectively. The correlation of MOI with β-galactosidase enzyme activity and the first-order dependence of infection on virus concentration demonstrated that enhancement was due to an increase in the number of cells infected by single virions. Both Polybrene and DC-Chol/DOPE associated with virions. However, Polybrene was also effective when target cells were pretreated. Thus, Polybrene acted primarily on the target cells (and was equally competent whether added alone or in association with virions), while DC-Chol/DOPE acted by stable-complex formation with virions. These data are in agreement with the observations of Toyoshima and Vogt, who studied the influence of polycations and polyanions on infection by avian sarcoma viruses (32). Polybrene proved to be the most effective enhancer of virus titer, acting mainly on the target cells. Polycation-mediated enhancement was neutralized by polyanions, suggesting an electrostatic mechanism. We similarly observed reduction of Polybrene-enhanced transduction in the presence of 40 μg of heparan sulfate per ml; however, this had no effect on DC-Chol/DOPE-mediated enhancement (data not shown), suggesting that the mechanism of action of these liposomes was not a simple electrostatic effect. These data also indicate that liposome enhancement was not mediated by heparan sulfate, as is the case for DNA transfection with PLL and some cationic liposomes (25).

It is unclear in what way the virus is complexed with the DC-Chol/DOPE liposomes and in what form it is deposited on the cell surface. Incubation of HIV-1 labelled with octadecylrhodamine (R₁₈) and DOTMA-containing cationic liposomes led to rapid fusion with the virus membrane, indicated by fluorescence dequenching (22). Fusion of virus membrane and liposome lipid bilayers is thus a possibility. Since there is apparently little effect on postbinding internalization and dissociation of adsorbed virions, enhanced transduction could be due to increased binding or more efficient fusion: our data do not allow a distinction to be made. By using an indirect fluorescence-activated cell sorter-based assay for virus binding (21), the shift in fluorescence seen for virus complexed with liposome, but not Polybrene, was slightly enhanced compared to that of virus alone (data not shown). However, this must be interpreted with caution since there is no distinction between intact virion and shed envelope binding.

It is evident from our data that the liposomes are able to mediate binding and subsequent fusion of virus and target cell membranes independent of any envelope-receptor interaction. Endocytosis is the primary route of cationic liposome-mediated DNA entry (37); depending on the liposome used, endosome escape of DNA may occur from late- or early-stage endocytic vesicles (11). Cationic liposome-mediated envelope-independent fusion therefore probably also occurs within endosomes after internalization, rather than at the cell surface. However, a great advantage of the use of nonenveloped retroviral particles rather than naked DNA is that the vector integrates following release of viral cores into the cytoplasm, resulting in stable transduction. It is also possible that such liposome-nonenveloped virion complexes will prove amenable to vector targeting. Noncationic liposomes have previously been used to bypass a specific receptor requirement by encapsulation of poliovirus (35) or murine sarcoma virus (9). Lipofectin and Lipofectamine could similarly facilitate the entry of rotavirus (5) and hepatitis delta virus (6), respectively. Lipofectin was able to mediate receptor-independent infection of nonpermissive cells by ecotropic and amphotropic MLV vectors (20): the liposome had no effect on receptor-dependent titers but enabled significant transduction of nonpermissive cells (titers approximately 1% of those on permissive cells). In agreement with our studies, no such transduction was observed with Polybrene.

In contrast, we also observed DC-Chol/DOPE enhancement of the transduction rate with viruses bearing relevant envelope proteins. Titers were approximately 100-fold higher than with nonenveloped viruses, indicating that envelope-mediated processes are still necessary for optimal infection. It is likely, therefore, that the liposomes serve to enhance virion adsorption independent of receptor binding but such that subsequent envelope-receptor interaction is facilitated. In the absence of such interaction, entry is mediated by the liposomes directly, but the efficiency of this process is poor. While MLV-E infection is endosomal pH dependent, suggesting that endocytosis is necessary for envelope-triggered fusion, infection by MLV-A is pH independent (24). Thus, depending upon the relative rates of endocytosis and fusion, liposome-enhanced enveloped virions may fuse at the cell surface or from within an endosomal compartment. Since the entry of fluorescently labelled DNA is slow (37), it is likely that fusion, at least for MLV-A, is at the cell surface. A previous study with HIV-1 showed that preincubation with a cationic liposome composed of DOTMA-cholesterol could enhance infectivity (22, 23). Optimal enhancement required pretreatment of the cells or the continued presence of the liposome during virus exposure. The mechanism appeared to be largely electrostatic and required envelope-CD4 interaction. More recently, receptor-dependent transduction enhancement by Lipofectamine (DOSPA/DOPE), Lipofectin (DOTMA/DOPE), and DOTAP for an MLV vector has been reported (18). The rate of infection was not monitored in this study, and the enhancement, which was apparently due to charge neutralization, was measured after exposure for 24 h.

The practical significance of increased transduction rates is that short exposure times can lead to levels of gene transfer that otherwise require extensive exposure. This is likely to be particularly relevant for in vivo gene delivery, when a combination of factors is likely to lead to rapid vector clearance. As an initial test of the relevance of the enhanced in vitro performance for use in vivo, we determined the influence of DC-Chol/DOPE liposomes on transduction of mesothelioma cells in a human tumour xenograft. The liposomes were found to mediate a highly significant increase in gene transfer to the xenograft. In summary, the association of cationic liposomes with retrovirus vectors offers a means of enhancing the gene transfer achievable in a brief period of exposure of target cells, such as is likely to be the case for in vivo delivery, and is therefore likely to find application in gene therapy protocols involving such vectors.