Background

Retroviruses, including lentiviruses, integrate into the genome of host cells, and the expression of the transduced genes can persist throughout cell divisions. Hence, murine leukemia virus (MLV)- and lentivirus-based vectors are among the most commonly used tools for gene transfer in eukaryotic cells in the laboratory, and may one day become clinically important. Lentiviral vectors have also the additional advantage of transducing non-dividing cells, which broadens their application to both resting and terminally differentiated cells.

Despite continuous improvement of retroviral and lentiviral gene transfer over the past years [1–3], the current inability to target infection to cells of interest remains a severe limitation, preventing the development of efficient, safe and cost-effective clinical application. A number of reports have already been published to this end (for review, see [4–6]). The majority of these studies were attempts to redirect the tropism of the ecotropic envelope glycoprotein (GP) of MLVs by the addition of ligand motifs, which bind to specific molecules associated with the cell membrane. However, these approaches generally met with limited success. Although the engineered viruses usually did bind to the new receptors, infection titres were low. Inefficient transduction was mostly due to diminished fusion activity of the engineered GP, which consequently prevented infectious translocation of the viral capsids into cells [7–9].

Retroviral and lentiviral GPs are made of two parts, produced from the same precursor following proteolytic maturation. SU, or surface protein, recognises the viral receptor, and TM, the transmembrane protein, carries the fusion activity and tethers the GP to virions [4–6]. However, retroviruses and lentiviruses can be pseudotyped by a number of GPs from other viruses, such as the hemagglutinin (HA) of influenza virus, the envelope proteins (E1 and E2) of Sindbis virus and the G protein of vesicular stomatitis virus (VSV-G). These have all higher fusion activity than the native GPs and remain tightly attached to virions. HA has already been engineered for targeting purposes through N-terminal addition of various ligands, of which one successfully redirected MLV tropism towards human melanoma cells [10]. E2 has also been genetically modified to display the immunoglobulin-binding domain of Staphylococcus aureus protein A [14, 15]. VSV-G is a trimerised transmembrane molecule, although its exact structure is not fully known. Moreover, its ligand has not been identified [16], which hampers rational design of targeting strategies. Additionally, only a few permissible sites for short (2–10 amino acids) peptide insertions have been isolated [17–20]. Nevertheless, these studies all confirmed that VSV-G might be amenable to genetic engineering for targeting purposes. Guibinga et al inserted a 10 amino acid collagen-binding peptide close to the N-terminal of VSV-G, and could show specific attachment of MLV- and HIV-1-based vectors to collagen matrix [17]. To date, however, no redirected cell transduction has been reported. We therefore attempted to target infection by attaching a large ligand binding domain, an scFv against MHC-I, directly in the N-terminal of the protein, a site that Yu and Schaffer confirmed permissive. We show that the novel GP, with its large exogenous peptide, (i) is processed and transported to the cell surface, (ii) provides a new binding specificity but (iv) transduces target cells very inefficiently, although better than control scFv/VSV-G. We speculate that this is due to an inefficient fusion activity, and discuss potential improvements.

Results and discussion

As a model system, we decided to target MHC-I molecules on human cells, as these membrane receptors can mediate cell infection by retroviral and lentiviral vectors [

Materials and methods

Engineering of VSV-G and scFv/VSV-G expression plasmids

The 1.6 kB HindIII-BamHI VSV-G fragment (serotype Indiana) was transferred from pFB. VSV-G into pcDNA3 (InVitrogen). To introduce a HA tag in the C terminal of VSV-G, the cDNA was amplified with a T7-specific sense primer (5' TAATACGATCACTTTAGGG) and an antisense oligo, including the HA sequence (in miniscule), a stopcodon and an Xho site (5' CCCCTCGAG TTA agcgtaatcaggaacatcataaggata CTTTCCAAGTCGGTTCATCTC). The product was digested with HindIII and XhoI, and reinserted into pcDNA3.

To generate scFv/VSV-G molecules, the sequence for mature VSV-G (from amino acid 17) was amplified with a sense primer, also containing a NotI site and an additional nucleotide to retain the reading frame (5' CCCGCGGCCGC A AAGTTCACCATAGTTTTTCCACAC). The anti-sense primer hybridises to the HA sequence, contains a stopcodon and carries a Cla I site (5' CCCATCGAT TTA AGCGTAATCAGGAACATCATA). The NotI/ClaI restricted PCR product was ligated into NotI/ClaI-cleaved PM441 and PM442 plasmids [23]. These constructs originate from an MLV-derived plasmid (FBMOSALF [31]), modified to contain an scFv (αMHC and αHEL, [27]), upstream of the GP gene. Consequently, the resulting constructs (αMHC/VSV-G and αHEL/VSV-G) express the gene from the MLV LTR, with a MLV leader sequence and 6 additional amino acids from the virus (see Fig 1).

Restriction enzymes were purchase from Roche or Invitrogene and all oligonucleotides were obtained from Sigma.

Cells and culture conditions

HeLa [32], 293T [33], TelCeb6 [31], Cos-7 [34] and Mus Dunni cells [35] were grown at 37°C in Dulbecco's modified Eagle's medium (Sigma), supplemented with 10% heat inactivated foetal calf serum (Gibco), 100 units/ml streptomycin, 100 units/ml penicillin and 2 mM L-glutamine in a humified 5% CO2 incubator.

Transient expression of αMHC/VSV-G and αHEL/VSV-G

HeLa or 293T cells were seeded on 6-well plate at 60% confluency. The following day, cells were transiently transfected using the classic CaPO4 co-precipitation method [36] with 5 μg DNA/well. The precipitate was removed and gene expression was confirmed 24–48 hours later by Western Blot, immunofluorescence or flow cytometry.

Production of VSV-G and scFv/VSV-G pseudotyped lentiviral particles

To express HIV-1 particles, 293T cells (75% density) in a 10-cm culture plate were transiently transfected with 5 μg of an LTR-driven EGFP vector (pHRCMV-EGFP) and 4 μg of a helper plasmid (pCMV△8.2) [29]. 5 μg VSV-G or 30 μg scFv/VSV-G plasmids were also included. DNA precipitate was removed after 16 hours, and the viral supernatants were collected 24–48 hours later.

Immunoblotting assays of VSV-G and scFv/VSV-G

For virion protein preparation, 1 ml of culture supernatant from virus producing cells were adjusted to 10 mM CaCl2 and left at room temperature for 30 minutes. Precipitated viruses were spun down at 13 k rpm at 4°C for 1 minute and resuspended in 50 μl of electrophoresis loading buffer. Cells were resuspended in triplex lysis buffer (50 mM Tris-HCl pH8.0, 150 mM NaCl, 0.2% NaN3, 0.1% SDS, 1% NP40, 0.5% Na-deoxycholate, 2 mg/ml leupeptin, 1 mM phenylmethyl sulfone fluoride) and left on ice for 30 minutes. Cell debris and nuclei were removed by centrifugation (13 k rpm at 4°C for 10 minutes). The samples were fractionated through SDS polyacrylamide (10%) gels (SDS-PAGE) and transferred to Protran nitrocellulose membranes (Schleicher and Schuell). VSV-G and scFv/VSV-G carry an HA tag, and were detected by a rat anti-HA antibody (Sigma), followed by a horseradish peroxidase (HPO) conjugated anti-RatIgG (Dako). p24Gag, detected by SF2 rabbit monoclonal antibody (NIH AIDS Research and Reference Reagent Program) and an anti-rabbit IgG/HPO (Santa Cruz), was used as an internal reference to normalise the virion proteins. The membranes were developed with Renaissance chemoluminescence kit (NEN Life Science Products), as recommended by the supplier.

Detection of scFv/VSV-G by immunofluorescence assay

Transfected HeLa or 293T cells were incubated with a conformation specific anti-VSV-G antibody (5G8F11, a generous gift by Dr Douglas Lyles, Winston-Salem) for 30 minutes, washed and revealed by a fluorescein isothiocyanate conjugated anti-mouse IgG antibody (FITC-anti-MuIg; Sigma). VSV-G expressing cells were detected under a fluorescence microscope (Zeiss).

Distribution of intracellular, HA-tagged VSV-G was assessed in paraformaldehyde-fixed, Triton-X permeabilised transfected HeLa cells, grown on cover slips. The proteins were visualised with a rat anti-HA antibody together with a FITC labelled anti-Rat IgG (both Sigma), and analysed with a confocal microscope (Leica).

Detection of scFv/VSV-G by flow cytometry

2 × 105 transfected 293T cells were collected in phosphate buffered saline (PBS), incubated in block buffer (BB: 10% bovines serum albumin, 0.1 M Glycine in PBS) for 30 minutes on ice, which was replaced by 200 μl of 5G8F11 hybridoma supernatant. After 1 hour on ice, the cells were washed twice with BB and resuspended in 100 μl FITC-anti-MuIg. The cells were rinsed again after 1 hour, fixed with 0.2 % formaldehyde and analysed on a FACScalibur fluorescence-activated cell sorter (Becton Dickinson).

VSV-G binding assays

5 × 105 human 293T and HeLa cells, and Mus Dunni cells were incubated with 1 ml pseudotyped HIV-1 particles from transiently transfected 293T cells for 30 minutes on ice. Cells were washed two times with PBS. scFv/VSV-Gs or VSV-G, attached to the cell surfaces, were detected as previously described.

Infection assays

Supernatant from HIV-1-producing 293T cells were passed through a 0.45 μm filter (Sarstedt). Some samples were concentrated 100 times by centrifugation (25 k rpm at 4°C for 2 hours in a BeckmanCoulter ultracentrifuge) and were carefully resuspended in 1% BSA. When required, the virus was stored at -80°C.

50% confluent target cells, either human 293T and HeLa cells, mouse Mus Dunni cells or monkey Cos-7 cells, were cultured with dilutions of virus for 16 hours in the presence of 5 mg/ml polybrene. 48 hours later, green fluorescent colonies were counted or cells were analysed by flow cytometry.

To block αMHC/VSV-G driven infection, target cells were preincubated with the B9.12.1 (< 1 μg/ml, Beckman Coulters) for 30 minutes before addition of the virus.