Streamlined design of a self-inactivating feline immunodeficiency virus vector for transducing ex vivodendritic cells and T lymphocytes

Prototype vectors and packaging construct were developed from pΔ00 (Fig. 1A and 2A), a replication-competent molecular clone of the Petaluma strain of FIV (FIV-Pet), derived from plasmid p34TF10 [GenBank: NC_001482] and produced in our laboratory by substituting a tryptophan codon for the stop codon in the accessory gene ORF-A [17]. As reported [18], an intact ORF-A is essential for optimal FIV replication in lymphoid cells. Nucleotide (nt) positions of packaging and vector constructs are referred to the NC_001482 sequence. The sequence of primers used in polymerase chain reaction (PCR) is available by e-mail on request. All the intermediate and final constructs described below were checked for proper insertions and absence of unwanted mutations by cycle sequencing using an automated DNA sequencer (GE Healthcare, Milan, Italy). All vectors were tested using enhanced green fluorescent protein (GFP) as reporter molecule.

The Rev provided in trans to the pGP-RRE was obtained by amplifying the corresponding mRNA from RNA of FIV-Pet infected Crandell feline kidney fibroblast (CrFK) cells. The cDNA was then cloned into the pcDNA3.1 plasmid (pcDNA-Rev). Constructs pcDNA-Rev and pGP-RRE were cotransfected at equimolecular ratio. FIV particles were pseudotyped with the vesicular stomatitis virus glycoprotein envelope (Env) VSV-G (495 amino acids [aa]) or the chimeric retrovirus glycoprotein RD114/TR (546 aa) [24]. VSV-G was encoded by pCMV-VSV-G derived from the pcDNA3.1 plasmid and RD114/TR by the phCMV-RD114/TR plasmid (kind gift of Dr. François-Loic Cosset, Ecole Normale Supérieure, Lyon, France).

The cells used included the CrFK, highly permissive to FIV-Pet, and the human epithelial 293T and murine fibroblast NIH-3T3, two nonfeline lines that do not permit FIV replication. All cells were propagated in Dulbecco's modified Eagle medium (D-MEM, Sigma-Aldrich, Milan, Italy) supplemented with 10% fetal calf serum (FCS), 100 U/ml penicillin, 100 μg/ml streptomycin and 2 mM L-glutamine (Sigma-Aldrich), at 37°C in 5% CO₂. Murine DCs were generated from the bone marrow (BM) of 6- to 10-week old BALB/c mice. Briefly, BM cells were flushed from the femurs, filtered through a 200 μm mesh to remove fibrous tissues, and cleared from erythrocytes with ammonium chloride. Residual cells were cultured at 2 × 10⁶ cells/ml in complete RPMI 1640 medium (Sigma-Aldrich), supplemented with 10% FCS, penicillin-streptomycin and glutamine, and induced to differentiate into DCs with 20 ng/ml recombinant mouse granulocyte macrophage colony-stimulating factor (GM-CSF) and 5 ng/ml recombinant mouse interleukin (IL)-4. Floating cells were removed, and fresh GM-CSF/IL-4 enriched medium was added at days 3 and 5 of culture. On day 7, non-adherent and loosely adherent DCs were collected and analyzed by flow cytometry. While control cells cultured with no GM-CSF and IL-4 were essentially negative, they were 55% CD11c-positive, 70% CD80-positive and 15% CD40-positive, thus exhibiting the expression profile of immature DCs [25]. Microscopic examination of the cultures also revealed that they were rich in cells with DC morphology. Murine T lymphoblasts were produced by concanavalin A stimulation of spleen cells from the same mice. Briefly, spleen cells were cultured at 2 × 10⁶/ml in 6-well plates for 5 days in complete RPMI 1640 medium supplemented with 50 unit/ml IL-2, 10% FCS, penicillin-streptomycin, and glutamine. These cells were 30% CD4 positive and 12% CD8 positive.

Vectors were generated in CrFK or 293T cells. Briefly, 2.8 × 10⁶ cells were seeded in 10 cm Petri dishes and one day later co-transfected with one of the vector plasmids, pΔenv1, and either the VSV-G or the RD114/TR Env (4:5:1; 20 μg total DNA) using a modified calcium phosphate method [26]. Transfection efficiency was evaluated 48 h later by counting GFP-positive cells by flow cytometry with a FACScan and a CELLQuest Version 2 software (BD Biosciences, Milan, Italy). Vector content in the culture fluids collected on the same day was determined by measuring FIV p25 capsid protein and the number of FIV RNA genome copies as previously described [26, 27], following clarification at 1,500 rpm for 10 min and 0.45 μm filtration. Supernatants were aliquoted in 1 ml volume and stored at -80°C until use.

The day before transduction, 24-well plates were seeded with 7 × 10⁴ 293T cells, 5 × 10⁴ CrFK cells, 5 × 10⁴ NIH3T3 cells, 5 × 10⁵ DCs, or 2 × 10⁶ T lymphoblasts per well in 1 ml complete medium. Eighteen h later, the medium was replaced with the same volume of vector suspension. Transduction efficiency was evaluated by counting GFP positive cells by flow cytometry 2 days post-transduction (PT).

Vector titers were determined in 293T cells and expressed as number of transduction units (TU) per ml. Briefly, 7 × 10⁴ cells were transduced as described above with the vector preparation under test serially diluted 10-fold in culture medium. Two days later, the cells were harvested and analyzed for GFP fluorescence by flow cytometry. Each dilution was tested in triplicate.

Nucleic acids in supernatants, collected from transfected 293T cells and treated as described under "Vector production", were extracted using the QIAamp Viral RNA kit (Qiagen, Milan, Italy). Genomic DNA and RNA from 6 × 10⁴ transfected or transduced 293T cells were extracted with QIAamp DNA Blood Kit and RNeasy kit (Qiagen), respectively. Viral and genomic RNAs were treated with RNase-free DNase (Qiagen) to eliminate residual DNA. Presence of pΔenv1 plasmid and RNA transcripts in supernatants and transduced cells was investigated by PCR using 295s-296as primers targeting FIV p25 capsid protein (Fig. 3A). Translocation of the U3 deletion from the 3' to the 5'LTR in the vector provirus was examined by amplifying genomic DNA from transduced cells with U3s-R3as primers (Fig. 3B). Inactivation of LTR mediated transcription was ascertained by amplifying cDNA of transduced cells with INs and RREas primers (Fig. 3C). Reverse transcription was carried out with an avian myeloblastosis virus reverse trascriptase (RT) (Finnzyme, Celbio, Milan Italy) and the specific antisense. Amplification profiles were as follows: initial denaturation 94°C 2 minutes; cycling 94°C 30 seconds, 60°C (54°C for 295s-296as) 30 seconds, 72°C 30 seconds (40 seconds for 295s-296as), 35 cycles; extension 72°C 10 minutes. The 5'LTR amplicon was cycle sequenced using the automated ALF ExpressII DNA sequencer (GE Healthcare, Cologno Monzese, Italy).

Packaging and vector constructs were both derived from pΔ00, a replication competent clone of FIV-Pet (Fig. 1A and 2A).

The packaging constructs provided Gag and Pol and were under the control of the CMVp. Basically, they lacked the untranslated 5'LTR-gag region that, together with the initial part of gag, form the ψ signal required for viral RNA encapsidation into assembling virions and were deleted of part (pΔenv1 and pΔenv2 constructs) or the entire (pGP-CTE and pGP-RRE) env. The latter constructs also lacked Vif, ORF-A, and Rev. Since nuclear export of unspliced (i.e. Gag-Pol encoding mRNA) and singly spliced mRNAs occurs through Rev binding to the RRE motif, in pGP-CTE the Rev/RRE system was replaced by a CTE and in pGP-RRE RRE was maintained and Rev provided in trans by cotransfection of pcDNA-RRE plasmid (Fig. 1B–1D).

The SIN vector LA34 (Fig. 2B) was under the control of the CMVp and had few remnants of the FIV genome, namely 1) the ψ signal, 2) the RRE motif, placed downstream ψ and interacting with the Rev provided by the packaging construct or pcDNA-RRE, 3) the untranslated region between env and 3'LTR, and 4) the two LTRs, both deleted of the U3 domain to avoid the generation of functional LTRs during reverse transcription. Due to extensive rearrangement and deletions, LA34 was 2.2 Kb in size.

The packaging constructs (Fig. 1B–1E) were developed and tested for stability and safety (non-infectivity) in CrFK and 293T cells. Briefly, the packaging constructs were transfected into cells which were propagated for at least two weeks and monitored twice a week for FIV p25 and viral RNA release into supernatant. Further, cell-free supernatants were collected at 7 and 14 days post-transfection, and seeded into fresh CrFK cell that were cultivated for additional two weeks. Except for a low and transient production of p25 found two-three days post-transfection, neither FIV RNA nor infectious particles were found in supernatants (data not shown). The results indicated that the packaging constructs were free from residual pΔ00 plasmid molecules, were stable, and did not generate infectious virus.

Transduction efficiency, stability, and safety were tested in CrFK and 293T cells, using LA34 pseudotyped with VSV-G (LA34/VSV-G) generated in CrFK or 293T cells with pΔenv1 as packaging construct. The proportions of GFP-positive CrFK and 293T cells 2 days post-transfection were generally greater than 75%. As determined by measuring p25 and number of vector RNA copies in the supernatant from day 2 to 7 post-transfection, LA34/VSV-G production peaked at day 2 or 3 (data not shown). Vector particles collected at these times were pelleted by ultracentrifugation and analyzed by western blot for protein content. Regardless of whether produced in CrFK or 293T cells, the vector generated protein patterns that, with the obvious exception of Env, were identical to the one of whole FIV-Pet virus used as control, indicating that proper generation and maturation of virions had taken place (data not shown). As shown in Table 1, vector titers exceeded 10⁹ RNA copies/ml and 10⁶ TU/ml in 293T cells and were slightly higher when the vector was produced in CrFK rather than 293T cells.

Vector safety was evaluated by several approaches. First, the progeny particles were checked for pΔenv1 incorporation by testing them as well as transduced cells for a gag p25 sequence contained in the packaging construct only (Fig. 1B to 1E). The 674 bp amplicon generated by the PCR and RT-PCR assays used was readily detected in the DNA of transfected cells (positive control) but uniformly absent in the vector particles (not shown) and in transduced cells (Fig. 3A). Second, it was checked whether the U3 deletion, created in the 3'LTR of the vector construct, was indeed translocated to the 5'LTR of proviral DNA by using primers that generated amplicons of different sizes from the full-length (352 bp) and the U3 deleted LTRs (232 bp). The amplicon generated from the DNA of transduced cells was clearly smaller than the one generated from the pΔ00 plasmid used as a source of full-length LTR (Fig. 3B) and had the sequence expected for the deleted LTR (not shown). Third, functional inactivation of the 5' LTR was checked by examining transduced cells for LA34 RNA genomes, the transcription of which would have required a full-length 5'LTR. As shown by Fig. 3C, while the DNA of transduced cells was clearly positive for LA34 sequences, the RNA obtained from the same cells was uniformly negative, regardless of whether it was digested or not with RNase-free DNase. Collectively, these findings demonstrated that LA34 is indeed a SIN vector, that the pseudotyped particles it generates are safe, and that no vector RNA is produced by LA34 transduced cells.

The packaging constructs were tested for ability to produce LA34/VSV-G virions in 293T cells that were transfected with equimolecular amounts of packaging, vector, and VSV-G plasmids. Supernatants collected 2 days post-transfection were analyzed for virus release by measuring p25 and vector RNA genome copies and for competence for transduction in 293T cells. The efficiency of transfection was similar for all plasmid combinations and averaged 80%. As shown in Table 2, LA34/VSV-G production with pΔenv1 and pΔenv2 was very high, as shown by the high levels of p25 produced, 10⁹ vector RNA copies and 10⁶ TU/ml. In contrast, both vector RNA and TU titers of VSV-G pseudotyped particles produced by using pGP-RRE and pGP-CTE were one and two logs lower, respectively, suggesting that virus release, rather than infectivity, was less efficient. The pseudotyped particles generated with pΔenv1 and pΔenv2 were further tested for transduction in CrFK. Since the LA34/VSV-G generated with pΔenv1 performed slightly better in this cell subtype, this construct was selected as packaging for subsequent experiments (data not shown).

Transduction efficiency and duration of transgene expression were assessed by inoculating LA34/VSV-G produced in 293T cells into CrFK, 293T and NIH-3T3 cultures at varying RNA copy numbers. The best results were obtained with 10⁹ copies, corresponding roughly to 10 TU/cell, which at the first readout, 4 days PT, transduced 66% and 52% of CrFK and 293T cells, respectively. In contrast, GFP-positive NIH-3T3 cells rarely exceeded 20%, indicating that these cells were largely refractory to transduction by this vector. Furthermore, during 25 days of observation, GFP-positive cell numbers remained essentially unchanged regardless of initial transduction level, indicating that transduction and transgene expression were stable, as a likely result of vector integration into the cell genomes (Fig. 4A). No infectious virus was detected in the transduced cultures at any time, not even after passaging the culture fluids in fresh CrFK cells repeatedly (data not shown).

Because the findings above were indicative of a preferential ability of LA34 to transduce feline CrFK cells relative to the non feline cells 293T and NIH-3T3, we made efforts to widen its breadth of action by improving nuclear translocation of PIC, stabilization of transgene mRNAs, and incorporation of vector RNA into pseudotyped particles that are major determinants of lentiviral cell transduction and transgene expression [28]. Two short single-stranded regions of the lentiviral genome DNA (flaps) are believed to optimize viral genome folding, thus enhancing its steric fit in the nuclear pore [29]. In FIV, one of these flaps is located upstream the 3' LTR (U3PPT) and the other close to the 3' end of pol (central PPT, cPPT) [21]. Since LA34 contains only the U3PPT, we inserted the cPPT between RRE and MCS (construct LA34-cPPT, Fig. 2C), similar to what already done in the FIV vectors developed in previous reports [28, 30]. However, the change had no appreciable effects on vector performance in any of the three cell lines (Fig. 3B).

When we inserted the RNA transport WPRE element downstream of MCS in LA34 (LAW34, Fig. 2D), so that it could be incorporated into transgene mRNA, the efficiency of transduction improved greatly. Compared to LA34, LAW34 showed similar RNA titers but the TUs per ml were approximately 1 log higher (Table 1). Most importantly, LAW34/VSV-G gave rates of NIH-3T3 cell transduction that were nearly twice as great. In fact, close to 50% of the latter cells expressed GFP and did so for at least 4 weeks (Fig. 4B and data not shown).

Recent reports have suggested that the main packaging domain in the gag of FIV is comprised in a 120 nt stretch [31]; however, previous studies had suggested the existence of additional encapsidation determinants in gag, bringing ψ to about 300 nt in size [23]. We, thus, also constructed versions of LA34 and LAW34 having a 311 nt-long ψ (LA34-L and LAW34-L). When compared to the respective vectors with 120 nt-long ψ, these versions showed no appreciably increased titers (data not shown) and, with the exception of LAW34-L for CrFK cells, had reduced transduction efficiencies (Fig. 4C). As a result of these studies, LAW34 was selected for the experiments below.

In these experiments, we compared the standard transduction method described under Methods with several protocols that have been shown to increase transduction efficiency by other vectors. These included: 1) vector ultracentrifugation, a procedure frequently used to concentrate VSV-G pseudotyped viruses that, unlike lentiviral Env-coated pseudoparticles, do not tend to shed Env [32]; 2) low-speed centrifugation following poly-L-lysine addition [33]; 3) addition of polybrene (PB), a polycation that forms large complexes with viral particles leaving out nonviral proteins and other factors that may be inhibitory (protocol PB) [34]; 4) treating the cells with the vector twice, four h apart (double transduction; protocol DT); and 5) combined use of 3 and 4 (protocol PB/DT). LAW34/VSV-G produced in 293T cells (5 × 10⁹ vector RNA copies/ml) was concentrated 10-fold by ultracentrifugation at 200,000g at 4°C for 2 h. In spite of a 4-fold increment in vector RNA titer, transduction efficiency was essentially unchanged relative to the standard method (Table 1). The same output was observed by using protocol 2 in which the virus was concentrated by aggregation with poly-L-lysine. Moreover, the use of poly-L-lysine often caused shrinking, granulation, and, occasional detachment of the cells (data not shown). Protocols 3 to 5 were instead beneficial. A dose-response study of PB in 293T using a vector RNA copy/cell ratio of 200/1 cells (approx. 10 TU/cell) yielded the highest transduction rates at 8 μg/ml (Fig. 5A). Importantly, at this dose PB more than doubled the proportion of transduced NIH-3T3 cells (Fig. 5B), suggesting that this protocol substantially increased virus infectivity on this cell type. At same conditions, protocol DT also increased transduction of all cell types (data not shown). However, protocol DT/PB was the most effective, since transduced cells exceeded 90% at day 4 (Fig 5B). Overall, the results also underlined the robustness of LAW34 under various test conditions.

LAW34 was pseudotyped with RD114/TR in 293T cells, and the vector thus produced (LAW34/RD114/TR; Fig. 6A) was titrated for TU in the same cell substrate using the ultracentrifugation and the PB/DT protocols. Again, in spite of a three-fold increment of the vector RNA titer after supernatant ultracentrifugation, transduction efficiency was lower compared to the PB protocol, confirming the modest performances of ultracentrifugation in our hands (Table 1). LAW34/RD114/TR and LAW34/VSV-G had essentially same number of vector RNA copies/ml, yet the former exhibited a 4 fold higher 293T transduction titer when complexed with PB, confirming the efficacy of this protocol even in the case of easy-to-transduce cells.

LAW34/RD114/TR and LAW34/VSV-G were also compared for transduction efficiency using the standard protocol, i.e. with no further manipulations or additions. Supernatants of packaging 293T cells diluted to achieve a vector RNA copy/cell ratio 1/200 of either vector performed equally in 293T and CrFK cells even after prolonged propagation (Fig. 6B and data not shown). In contrast, in NIH-3T3 cells LAW34/RD114/TR transduced with an efficiency the VSV-G counterpart had exhibited only when used with the PB protocol (Fig. 5B and 6B). Thus, LAW34/RD114/TR proved effective at transducing non feline cell lines with no need for treatments known to boost transduction efficiency.

BM-derived DCs were transduced with VSV-G or RD114/TR pseudotyped LAW34 by using 200 vector RNA copies per cell and protocols PB, DT, and DT/PB. Cells were examined 2 days later for GFP expression. LAW34-RD114/TR transduced much more efficiently than LAW34/VSV-G, regardless of protocol used. This striking difference, already clearly evident by microscopic inspection, was confirmed by flow cytometry analysis of the cells: whereas the fluorescence signal of LAW34/VSV-G transduced cells was barely distinguishable from that of mock transduced cells, LAW34/RD114/TR produced a clearly defined peak at 10¹ FL1-H, indicating that most DCs were transduced and actively expressing GFP (Fig. 7A). In fact, LAW34/RD114/TR performed better with all 3 protocols, transducing up to 52% DCs when the DT/PB protocol was used versus 17% with LAW/VSV-G (Fig. 7B and 7C). LAW34/RD114/TR transduction was also examined for possible effects on markers expression by DCs. DCs transduced with LAW34/RD114/TR at 200 vector RNA copies per cell using the PB protocol were compared to similarly treated cells except that the vector was replaced by the supernatant of mock-transfected cells. Relative to mock treated DCs, transduced DCs showed no changes in CD11c and CD80 expression but underwent a substantial increase of CD40-positive cells which was already evident by day 2 PT (Fig. 7D) and lasted throughout the observation period of 10 days (not shown). Of note, GFP expression by transduced cells, monitored in parallel, increased slightly over time (data not shown).

LAW34/VSV-G or LAW34/RD114/TR were also compared for ability to transduce cultured murine T lymphoblasts using the PB or the DT protocol and same vector/cell ratios. As shown by Fig. 7E and 7F, which reports the readings at day 2 PT, GFP positive cells ranged around 60% and fluorescence peaks were superimposible regardless of the Env used for pseudotyping. Also, the use of PB greatly reduced the efficiency of T lymphoblast transduction by both LAW34/VSV-G and LAW34-RD114/TR, in spite that no obvious effects on cell viability were noted. Proportions of CD4 and CD8 positive cells in the cultures remained stable for at least 10 days, following transduction (data not shown).