Retrovirus Targeting by Tropism Restriction to Melanoma Cells

Abstract

Targeted vectors will be necessary for many gene therapy applications. To target retroviruses to melanomas, we fused a single-chain variable fragment antibody (scFv) directed against the surface glycoprotein high-molecular-weight melanoma-associated antigen (HMW-MAA) to the amphotropic murine leukemia virus envelope. A proline-rich hinge and matrix metalloprotease (MMP) cleavage site linked the two proteins. The modified viruses bound only to HMW-MAA-expressing cells, as inclusion of the proline-rich hinge prevented viral binding to the amphotropic viral receptor. Following attachment to HMW-MAA, MMP cleavage of the envelope at the melanoma cell surface removed the scFv and proline-rich hinge, allowing infection. Complexing of targeted retroviruses with 2,3-dioleoyloxy-N-[2(spermine-carboxamido)ethyl]N,N-dimethyl-1-propanaminium trifluoroacetate–dioleoyl phosphatidylethanolamine liposomes greatly increased their efficiency without affecting their target cell specificity. In a cell mixture, 40% of HMW-MAA-positive cells but less than 0.01% of HMW-MAA-negative cells were infected. This approach can therefore produce efficient, targeted retroviruses suitable for in vivo gene delivery and should allow specific gene delivery to many human cell types by inclusion of different scFv and protease combinations.

Retroviral vectors have several features which make them attractive for clinical gene delivery. In particular, integration of the vector genome allows stable expression of the transduced gene in the infected cell and its progeny. Retroviral vectors can infect a wide range of cell types, including nondividing cells, following the development of vectors based on human immunodeficiency virus (20). Because viral coding regions are deleted from the vector, viral proteins are not expressed in infected cells, which avoids stimulation of an inappropriate antiviral immune response.

In gene therapy clinical trials, retroviral vectors have been used for in vitro infection, followed by transfer of modified cells to the patient. Such modification of cells is time consuming and costly and may not always be possible. It is therefore desirable to develop retroviruses suitable for in vivo gene delivery. Such vectors should efficiently infect specific target cells. Nontarget cells should not be infected, as gene delivery could be deleterious to their function and they would deplete the pool of viral particles.

The host range of retroviruses is partly determined by the surface domain (SU) of the envelope glycoprotein, which binds to a cell surface receptor (37). For murine leukemia viruses (MLVs), the receptor binding domain has been mapped to the N-terminal portion of SU (1). Previously described strategies for targeting of retroviruses have incorporated ligands (10, 29, 39) or single-chain variable-fragment antibodies (scFvs) (12, 15, 16, 31) recognizing targets on human cells into the SU protein of ecotropic MLV (MLV-E), which infects only rodent cells. Although viruses bound to human cells, infection was generally poor or not observed.

The aim of this study was to achieve efficient, specific targeting of human melanomas by limiting the tropism of amphotropic MLV (MLV-A), which can infect cells of many mammals, including humans and rodents. The receptor for MLV-A on human cells is RAM-1, a phosphate transporter expressed on most cell types (11). High-molecular-weight melanoma-associated antigen (HMW-MAA; also called melanoma-associated chondroitin sulfate proteoglycan) was selected as the target molecule. This integral membrane proteoglycan is expressed in more than 90% of human melanomas but not in most normal adult tissues (21, 25). Its expression by melanomas is associated with a poor prognosis (9). HMW-MAA has been used successfully as an in vivo target for radioimaging (6, 14, 19, 30) and immunotherapy of melanoma (2, 17, 18).

An scFv which recognizes HMW-MAA was linked to the extreme N terminus of MLV-A SU. A proline-rich spacer was used to link the scFv to SU in order to block MLV-A SU binding to RAM-1. Inclusion of this spacer in a chimeric envelope has been shown to block MLV-E binding to its receptor (35). A cleavage site for matrix metalloproteases (MMPs) was inserted after the proline-rich spacer. MMPs are highly expressed on the cancer cell surface (3) and are critical for tumor invasion of normal tissue (27, 33, 38). Cleavage of a chimeric retroviral envelope by exogenous factor X or cell surface MMP to remove an epidermal growth factor (EGF) domain has previously been reported (7, 22, 23). In these experiments, the EGF domain efficiently targeted virus to the EGF receptor, which destroyed its infectivity, so MLV-A SU binding to RAM-1 did not need to be blocked. The rationale for our approach was that attachment of viruses to HMW-MAA would lead to MMP removal of the scFv and proline spacer at the cell surface, allowing infection following MLV-A interaction with its receptor, RAM-1. Retroviruses carrying these targeted envelopes selectively infected HMW-MAA-positive cells in culture; their efficiency was increased by complexing with 2,3-dioleoyloxy-N-[2(spermine-carboxamido)ethyl]-N,N-di methyl-1-propanaminium trifluoroacetate (DOSPA)–dioleoyl phosphatidylethanolamine (DOPE) liposomes.

MATERIALS AND METHODS

Chimeric envelopes.

The scFv which recognized HMW-MAA was derived from monoclonal antibody LMH2. The scFv coding sequence was removed from the vector pCantab.5 (13) by digestion with SfiI and NotI. pEGFPRO4070A was constructed by inserting a proline spacer (isolated from the 4070A envelope by using the primers 5′-atcgaggtcaccgcggccgcgggaccccgagtccccatagggccc-3′ and 5′-ataatcggccgggggtggctgtgggac-3′ into the amphotropic chimera EA (4).

ScLPA was made by inserting the scFv coding sequence into pEGFPRO4070A after removal of the EGF coding fragment by digestion with SfiI and NotI. Insertion of the peptide Pro-Leu-Gly-Leu-Trp-Ala as an MMP cleavage site between the PRO linker and the env protein was made by PCR site-directed mutagenesis (Quikchange site-directed mutagenesis kit; Stratagene). The oligonucleotide primers used were 5′-cacagccacccccggccgcacccctgggcctgtgggccccccatcaggtctttaatgtaacctgg-3′ and 5′-ccaggttacattaaagacctgatggggggcccacaggcccaggggtgcggccgggggtggctgtg-3′, where the PLGLWA coding sequence is underlined. The 4070A envelope expression plasmid (ALF) were described previously (5).

Cell culture, virus production, and virus concentration.

TELCeB6 cells (5) are derived from the human rhabdomyosarcoma TE671 cell line (ATCC CRL-8805) and harbor the MFGnlslacZ vector genome and an MLV-Gag-Pol expression plasmid, CeB (5). BOWES (ATCC CRL-9607) and A375m (ATCC CRL-1619) are human melanoma cell lines. B-1 is a human melanoma cell line established in our laboratory. Ecv304 are a spontaneously transformed immortal human endothelial cell line (ATCC CRL-1998). All cells were grown in Dulbecco modified Eagle medium (DMEM; GIBCO-BRL) supplemented with 10% fetal calf serum (FCS) at 37°C and 10% CO₂.

Envelope expression plasmids scLPA, scLPMA, and ALF were transfected into TELCeB6 cells by using Lipofectamine (GIBCO-BRL). Transfected cells were selected with phleomycin (50 μg/ml), and pools of phleomycin-resistant clones were used for virus production. The TELCeB6-scLPMA bulk population was also cloned by serial dilution, and the clone which produced the highest virus titer was identified. To harvest viruses, producer cells were grown at 37°C until they became confluent and then cultured at 32°C for 4 to 7 days with feeding of fresh DMEM supplemented with 10% FCS every 2 days. The medium was then replaced with serum-free Optimem (GIBCO-BRL), and supernatant was collected 12 to 16 h later. The harvested virus was filtered through 0.45-μm-pore-size filters and, in some cases, concentrated by centrifugation at 2,500 × g and 4°C for 12 h. Concentrated virus was kept frozen at −70°C.

Envelope incorporation and gelatinase A cleavage.

Virus pellets were subjected to Western blot analysis using goat antisera against the Rauscher leukemia virus gp70 (SU) and p30 (CA) proteins as described previously (4). Accessibility of the MMP cleavage site to protease was demonstrated by treatment of pelleted viruses with activated gelatinase A (Boehringer Mannheim). scLPMA viruses were centrifuged at 100,000 × g for 1 h at 4°C, resuspended in 50 μl of 100 mM Tris (pH 7.5)–200 mM NaCl–1 U of activated gelatinase A, incubated for 6 h at 37°C, and then subjected to Western blot analysis using anti-RLV gp70 antibody.

Protease activity.

To determine protease activity on target cells, the 2,4-dinitrophenol (DNP)–Pro–Leu–Gly–Leu–Trp–Ala–D-Arg–NH₂ peptide (Bachem) was used. Target cells were grown to semiconfluency and then washed twice in buffer A (50 mM Tris [pH 7.5], 10 mM Ca₂Cl, 0.2 M NaCl). The DNP-peptide was diluted to 20 μM in buffer A, added to the different target cells, and incubated for 1 h at 37°C. Substrate hydrolysis was determined by monitoring the increase in fluorescence emission at 346 nm using an excitation wavelength of 280 nm.

HMW-MAA expression, envelope binding, and virus binding.

Expression of HMW-MAA on the target cells was determined by using LMH2 antibody (13) and CP/Me1.2 (Immune Systems Ltd.). For envelope binding, cells were incubated with viral supernatants and washed and envelope binding was then determined by use of a fluorescence-activated cell sorter (FACS) and goat anti-RLV gp70 antibody (16). For detection of virus binding, a fluorescence microscopy method was used (24). Briefly, cells were grown to confluence on glass coverslips and then incubated in concentrated virus for 30 min at 37°C. Cells and viruses were fixed in 4% paraformaldehyde for 5 min, washed once with phosphate-buffered saline (PBS), permeabilized in 0.02% Triton X-100 for 2 min, washed once with PBS, and then incubated with PBS containing 1% (wt/vol) bovine serum albumin (PBA) for 15 min at room temperature (RT). For detection of gp70 and p30, samples were incubated for 1 h with goat anti-RLV p30 polyclonal antibody and rat anti-gp70 83A25 monoclonal antibody, washed three times with PBA, incubated with a mixture of anti-rat immunoglobulin G (IgG)–tetramethyl rhodamine isocyanate (TRITC) and anti-goat IgG-fluorescein isothiocyanate (FITC) for 45 min, washed four times with PBS, mounted, and analyzed by confocal scanning microscopy (Bio-Rad).

Analysis of viral infection.

Target cells were seeded in 24-well plates at a density of 5 × 10⁴ cells/well 24 h before infection. Viruses were incubated with 4 μg of Polybrene (PB) per ml or 10 μg of Lipofectamine per ml, as indicated in Results, for 10 min at RT before being added to the target cells. Cells were incubated in the presence of the viruses for 1 h at 37°C, washed once in Optimem and once in DMEM–10% FCS, and then cultured for 24 to 48 h. 5-Bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) staining was performed as previously described (34). For infection of mixed target cells, HMW-MAA-negative and -positive cells were mixed at a 10:1 ratio and then seeded on glass coverslips. After 24 h, the mixed population was infected, washed once with PBS after a further 48 h, incubated with PBA at RT for 15 min, incubated with LMH2 antibody for 45 min at 4°C, washed three times with PBS, fixed with 4% paraformaldehyde, and permeabilized in 0.02% Triton X-100. To detect β-galactosidase, cells were washed once with PBS, incubated with PBA for 15 min at RT, and then incubated with rabbit anti-β-galactosidase antibody in PBA for 3 h. After three washes in PBA, samples were incubated with a combination of anti-rabbit IgG-TRITC and anti-mouse IgG-FITC. After five washes in PBS, cells were analyzed by using a confocal scanning microscope (Bio-Rad).

To inhibit targeted virus infection, cells were preincubated with 50 μg of LMH2 per ml for 5 min at 37°C and then incubated with virus in the presence of 50 μg of LMH2 per ml and 10 μg of Lipofectamine per ml for 30 min at 37°C. To inhibit MMP activity, cells were incubated with virus in the presence of 4 μg of TIMP-2 (Boehringer) per ml for 30 min at 37°C. After incubation, viruses were removed and the cells were washed, cultured for 24 to 48 h, and then X-Gal stained.

RESULTS

scLPA and scLPMA chimeric envelopes.

To construct the modified retroviral envelope scLPA (Fig. 1A, top), an scFv which recognizes HMW-MAA (13) was inserted at amino acid +5 of the MLV-A 4070A envelope with a proline-rich spacer (PRO) (36). In scLPMA, the MMP cleavage site PLGLWA (40) was introduced between the PRO linker and the envelope protein (Fig. 1A, bottom). Plasmids expressing scLPA and scLPMA were transfected into TELCeB6 cells, which express the MLV gag and pol proteins and carry a provirus encoding β-galactosidase. Virions from bulk populations of TELCeB6-scLPA and TELCeB6-scLPMA producer cells were pelleted and analyzed for the presence of chimeric envelopes by using a polyclonal anti-SU antibody. The 100-kDa scLPA and scLPMA envelope proteins were incorporated into virions, although at a lower level than the MLV-A 4070A envelope (Fig. 1B).

FIG. 1.

(A) An scFv derived from the antibody LMH2, which recognizes HMW-MAA, was fused to the N terminus of MLV-A 4070A SU by using the proline-rich region from MLV-A 4070A env (nucleotides 751 to 927) (P) as a spacer to make scLPA. An MMP cleavage site (PLGLWA) was introduced between the P linker and the SU of the MLV-A envelope to make scLPMA. TM, transmembrane protein. (B) Detection of unmodified MLV-A 4070A envelope (lane 1A), scLPA, and scLPMA in viral pellets from producer cell lines. p30 gag was detected in the same pellets to quantitate viral particles. (C) Cleavage of scLPMA by gelatinase A (Gel A) as described in Materials and Methods. Equal amounts of unmodified MLV-A 4070A and scLPMA envelopes were loaded in each lane. kd, kilodaltons.

To assess whether the cleavage site was accessible to MMPs, scLPMA viruses were incubated with activated gelatinase A. This treatment reduced the size of the chimeric scLPMA envelope to the size of unmodified SU (70 kDa) (Fig. 1C), demonstrating that the MMP cleavage site was cleaved. Some cleavage of scLPMA was observed without gelatinase A, probably due to production of MMPs by the TELCeB6 cells (Fig. 1C).

scLPA and scLPMA enveloped virions bind only to HMW-MAA-positive cells.

Binding of scLPA, scLPMA, and unmodified MLV-A 4070A (A) viral envelopes to HMW-MAA-positive or -negative cell lines was measured. HMW-MAA expression was defined by using LMH2, the monoclonal antibody from which the scFv used for viral targeting was derived (13). Producer cell supernatants were incubated with target cells, and envelope bound to cells was then detected with a polyclonal anti-SU antibody using a FACS. The scLPA and scLPMA envelopes bound to HMW-MAA-positive cell lines A375m and BOWES but not to HMW-MAA-negative cell lines B1 and ECV304 (Fig. 2A, black and green lines), demonstrating that the RAM-1 binding domain of scLPA and scLPMA was indeed masked. HMW-MAA-positive cells bound approximately 10-fold more targeted envelopes than unmodified MLV-A 4070A (Fig. 2A, blue lines). scLPA and scLPMA binding was blocked by LMH2 antibody (Fig. 2B), demonstrating that binding of the scLPA and scLPMA envelopes was dependent on interaction with HMW-MAA.

FIG. 2.

Specific binding of scLPA- or scLPMA-enveloped virus to HMW-MAA-positive cells. (A) Concentrated supernatants from producer cells expressing the MLV-A 4070A (blue), scLPA (black), or scLPMA (green) envelope were incubated with each target cell line for 45 min at 37°C. TELCeB6 supernatant was used to define control fluorescence (red). Envelopes were detected as described in Materials and Methods. HMW-MAA status of each target cell is shown (HMW-MAA + VE or −VE). (B) HMW-MAA-positive BOWES cells were incubated in TELCeB6 (red), scLPA (black), or scLPMA (green) concentrated supernatant in the absence (top) or presence (bottom) of LMH2 antibody. (C) HMW-MAA-positive BOWES or HMW-MAA-negative B-1 cells were incubated with concentrated scLPMA or MLV-A 4070A enveloped virus. Samples were stained for p30 and gp70 by using a goat anti-RLV p30 polyclonal antibody and rat anti-gp70 monoclonal antibody 83A25 as described in Materials and Methods. Viral particles appear as yellow dots as a result of the colocalization of FITC-labelled anti-goat and TRITC-labelled anti-rat signals.

Producer cell supernatants contain both virus particles and shed SU proteins, and FACS measurement does not discriminate between virus binding and SU binding. To directly measure cell attachment of scLPMA-enveloped viruses, we stained viruses bound to cells with antibodies against the SU and p30 gag proteins. Virions were detected by confocal microscopy as spots of colocalized SU and p30 staining. scLPMA-enveloped viruses attached efficiently to HMW-MAA-positive cells and showed little binding to HMW-MAA-negative cells (Fig. 2C). Virions with the unmodified MLV-A 4070A envelope bound to both HMW-MAA-positive and -negative cells (Fig. 2C). Thus, scLPMA enveloped virions had lost the ability to bind to RAM-1 but could efficiently attach to HMW-MAA.

scLPMA enveloped virions infect HMW-MAA-positive cells.

HMW-MAA-positive and -negative cells were infected with scLPMA-, scLPA-, or unmodified MLV-A 4070A-enveloped viruses and with nonenveloped virions. Infections were performed either in the absence of any enhancing reagent or with PB or DOSPA-DOPE liposomes (Lipofectamine). Nonenveloped or scLPA-enveloped viruses failed to infect either cell line. Viruses with unmodified MLV-A 4070A envelopes infected both cell lines, and their efficiency was increased by PB or DOSPA-DOPE liposomes. The scLPMA-enveloped viruses infected the HMW-MAA-positive cell line A375m when mixed with PB, but their efficiency was greatly enhanced when they were complexed with DOSPA-DOPE liposomes. They did not infect the HMW-MAA-negative cell line B1 (Fig. 3). In this experiment, centrifugation at low speed was used to concentrate all of the viral preparations (see Materials and Methods). In the absence of concentration, the scLPMA-enveloped viruses complexed with DOSPA-DOPE liposomes (virosomes) could still infect 20% of the A375m cells (data not shown). The fact that scLPA-enveloped viruses were unable to infect HMW-MAA-positive cells demonstrated that the MMP cleavage site in scLPMA virosomes was necessary for infectivity.

FIG. 3.

Percentage of HMW-MAA-positive or -negative cells infected. A375m and B-1 cells were incubated with virus with no envelope (Non) or an scLPA, scLPMA, or unmodified MLV-A 4070A (A) envelope concentrated by centrifugation (see Materials and Methods) without any treatment (N) or after mixing with PB or Lipofectamine (Lip) (see Materials and Methods).

The titer of scLPMA virosomes on HMW-MAA-positive and -negative cell lines is shown in Fig. 4A (top). On the HMW-MAA-negative cell line B-1, the titer was similar to that of nonenveloped virosomes (50 IU/ml) while the titer on Ecv304 cells (also HMW-MAA negative) was slightly higher (300 IU/ml). The scLPMA virosomes infected HMW-MAA-positive cells efficiently with titers of up to 10⁶ IU/ml on A375m cells and 10⁵ IU/ml on BOWES cells. Titers of the targeted virosomes showed that they were only 10-fold less efficient than virosomes with unmodified MLV-A 4070A envelopes.

FIG. 4.

Infection by targeted virosomes (A) The graph at the top shows titers of scLPMA-enveloped virus (best clone virus; see Materials and Methods) and unmodified MLV-A 4070A on HMW-MAA-negative (B-1 and Ecv304) cells and HMW-MAA-positive melanoma (A375m and BOWES) cells. Infections were performed after complexing of unconcentrated viruses with DOSPA-DOPE liposomes to produce virosomes. Titers are expressed as the mean infectious units (iu) per milliliter (± the standard error) of triplicate determinations. The lower graph shows cleavage of the PLGLWA peptide by B-1, Ecv304, BOWES, and A375m cells (see Materials and Methods). (B) Inhibition of scLPMA virosome infection of BOWES cells by competition for HMW-MAA or inhibition of MMP activity. scLPMA-enveloped virus and MLV-A 4070A were used to infect BOWES cells with LMH2 antibody, TIMP-2, or a combination of both as described in Materials and Methods. Data are expressed as mean percentages of the titer of untreated viruses (triplicate determinations ± the standard error).

We used the dansylated peptide DNP-PLGLWADR-NH₂ (32) to measure the level of MMPs capable of cleaving scLPMA present at the surface of each cell line. The cleavage of the peptide at its MMP site separates the DNP group (which acts as a quencher) from the tryptophan, leading to a fluorescence increase with excitation at 280 nm. A DNP-peptide buffer was incubated with the different target cells for 1 h at 37°C, and the change in fluorescence was measured. All of the cell lines showed similar levels of DNP-peptide cleavage (Fig. 4A, bottom). No change in fluorescence was observed in the absence of cells or without incubation at 37°C (data not shown). Thus, the ability of cells to cleave scLPMA was not sufficient to permit infection; expression of HMW-MAA to allow virosome attachment was also necessary. This conclusion was supported by the blocking of scLPMA virosome infection by LMH2 (60%), the MMP inhibitor TIMP2 (50%), or a combination of LMH2 and TIMP2 (80%) (Fig. 4B).

scLPMA virosomes selectively infect HMW-MAA-positive cells in mixtures.

Cocultures of HMW-MAA-positive and -negative cells were infected with scLPMA or unmodified MLV-A 4070A virosomes. Cells were then stained for expression of HMW-MAA (green) and for nuclear β-galactosidase (red) and analyzed by confocal microscopy (Fig. 5A). In both cocultures, almost all of the cells infected with scLPMA virosomes were HMW-MAA positive (Fig. 5A), as shown by the colocalization of red nuclei and green surface fluorescence. Unmodified MLV-A 4070A virosomes infected all of the cells (Fig. 5A). In the B-1–A375m mixed population, scLPMA virosomes infected 39% of the A375m cells but only 0.008% of the B-1 cells. In the Ecv-A375m mixed population, scLPMA virosomes infected 37.5% of the A375m and 0.01% of the Ecv304 cells (Fig. 5B). In both mixtures, the HMW-MAA-negative cells were present in excess (40 B-1 cells to 1 A375m cell and 60 Ecv304 cells to 1 A375m cell). This shows that scLPMA virosomes became activated at the HMW-MAA–MMP-positive target cells membrane, resulting in infection of these cells but not neighboring cells.

FIG. 5.

Infection of mixed target populations. (A) Mixed populations were infected with unconcentrated MLV-A 4070A or scLPMA-enveloped virus complexed with DOSPA-DOPE liposomes. Two days after infection, cells were stained for β-galactosidase (red) and HMW-MAA (green) expression as described in Materials and Methods. (B) Percentage of HMW-MAA-positive and -negative cells infected in cocultures using unmodified MLV-A 4070A or scLPMA-enveloped virus. Final cell ratios at confluency were 40:1 Ecv304 to A375m cells and 60:1 B-1 to A375m cells. Data are means (± the standard errors) from two separate experiments in which 20 randomized fields (∼3,000 cells) were counted for each infection. To determine the infection of B-1 or Ecv304 cells by scLPMA-enveloped virus, 200 fields (∼30,000 cells) were counted.

DISCUSSION

Previous attempts to target retroviruses have tried to extend the tropism of MLV-E to specific human cells by incorporation of ligands or scFvs recognizing various human cell surface proteins into SU (10, 12, 15, 16, 29, 31, 39). Although retargeted binding was achieved, infection of target cells tended to be inefficient. In contrast, insertion of the RAM-1 binding domain of MLV-A at the N terminus of MLV-E SU allowed efficient infection of human cells (4). Thus, attachment to the natural retrovirus receptor RAM-1 allowed infection whereas attachment to a variety of other cell surface proteins did not. Probably only a very limited number of human cell surface proteins can function as retrovirus receptors and can therefore be used for MLV-E targeting.

The present paper describes a strategy to achieve retrovirus targeting by tropism restriction of MLV-A. The insertion of a PRO spacer (35) between the scFv and the envelope protein blocked RAM-1 binding but allowed the displayed scFv to bind HMW-MAA. After virus attachment to HMW-MAA, cell surface MMPs removed the scFv and the PRO spacer, allowing MLV-A interaction with its receptor RAM-1 and efficient infection. This report describes the first retroviral targeting strategy which produces sufficiently high-titer virus for in vivo gene delivery.

While several reports have described the use of liposomes to enhance retroviral infection (8, 26), this is also the first to describe enhancement of targeted infection by liposomes with no loss of specificity. The use of DOSPA-DOPE liposomes is critical for this maintenance of specificity. We have previously shown that complexing of nonenveloped retroviral particles with N-[1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethylammonium methylsulfonate allows infection while 3β[N-(N′,N′-dimethylaminoethane)⩵carbamoyl]cholesterol–DOPE or DOSPA-DOPE enhancement is dependent on retroviral envelope-receptor interaction (26).

Our targeting approach has several features which make it attractive for clinical gene delivery. Firstly, the target cell must express both a given surface antigen and a given surface protease. This double requirement creates an extra degree of specificity. Secondly, the targeted viruses do not attach to cells without the surface antigen. This prevents uptake of virus by nontarget cells, which remains a problem with approaches such as transcriptional targeting. Furthermore, envelopes based on MLV-A can be efficiently incorporated into lentiviral vectors, such as those based on human immunodeficiency virus (28). This targeting method could therefore also be used for specific gene delivery to nondividing cells.