INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Retroviruses have been intensively studied for use in gene therapy, in part because the DNA form of the virus becomes integrated into the host cell genome. In addition, some retroviruses exhibit a broad host range that allows for infection of many types of cells. In particular, amphotropic murine leukemia virus (MuLV) and gibbon ape leukemia virus (GALV) have a broad tropism that reflects fairly ubiquitous expression of the phosphate transporter molecules that serve as receptors for these viruses. GALV uses the phosphate transporter molecule Pit1 (24), whereas amphotropic MuLV uses Pit2, which is a related phosphate transporter protein that shares ~60% homology with Pit1 (21, 37). While these transport proteins are widely expressed, the levels of expression vary considerably from tissue to tissue (16, 27). For this reason, viruses that could use both receptors may be optimal for infecting the largest number of cell types and tissues. Recently, we showed that specific, engineered forms of feline leukemia virus subgroup B (FeLV-B) could use both the human Pit1 (HuPit1) and Hu Pit2 receptors, suggesting that vectors pseudotyped with these envelopes may be useful for gene delivery (6).

FeLV was historically categorized into three subgroups (FeLV-A, -B, and -C) on the basis of superinfection interference analyses; a fourth interference group (FeLV-T) was later identified (23, 31, 32). FeLV-A is the form of FeLV that is transmitted from cat to cat, but the receptor for this virus has not been isolated. FeLV-Bs evolve from FeLV-A by recombination with portions of endogenous FeLV-like (enFeLV) envelope sequences (26, 33). The acquisition of enFeLV envelope sequences results in a change to a Pit1 receptor specificity (6, 35). The recombinant forms of FeLV-B differ in the amount of envelope surface unit (SU) that is derived from enFeLV (26, 30), and this may affect whether or not the virus can also use HuPit2 as a receptor (6). For example, when chimeric envelope genes were engineered between an FeLV-B-90Z envelope and a prototype FeLV-A-61E envelope, some chimeric envelope protein had acquired the ability to recognize HuPit2 as well as HuPit1 as a receptor. Only one other virus, 10A1 MuLV, that can utilize both Pit1 and Pit2 for entry has been identified (22, 38). However, while 10A1 MuLV can efficiently infect cells using the murine Pit1 homolog, it cannot infect cells as efficiently using HuPit1(20). Thus, 10A1 MuLV is not useful for gene transfer to human cells expressing Pit1, and GALV has been the virus of choice for these purposes.

FeLV is related in sequence to MuLV. There have been numerous studies of the MuLV envelope domains that determine host cell or receptor specificity, in many cases by relying on differences in host range between the subgroups of MuLV. Collectively, these studies suggest that the major determinant for receptor specificity resides in the N-terminal half of SU, and specifically within the first variable region A (VRA), although additional domains of SU, including variable region B (VRB) and a downstream proline-rich region (PRR), have been implicated as secondary determinants for some SU-receptor interactions (2-5, 8, 9, 15, 18, 19, 22, 25, 29). In addition, truncated forms of MuLV SU that lack C-terminal sequences can specifically bind to their cognate receptor, which further supports a key role for VRA and VRB in determining receptor specificity (2, 3, 5, 8, 15, 19). Thus, sequences encompassing VRA and VRB are often collectively referred to as the receptor binding domain (RBD).

The primary host range and receptor determinants of FeLV-B map to regions that correspond to the MuLV VRA and VRB. Our studies showed that the VRA of FeLV-B SU is sufficient to confer Pit1 receptor specificity to FeLV-A (6). However, sequences in both VRB and downstream, in the C-terminal half of SU, are secondary determinants for Pit2 receptor specificity (6). Interestingly, a subsequent study by Tailor and Kabat (34) using an independently derived FeLV-B isolate (Gardner Arnstein; FeLV-B-GA) suggested that Pit2 could not function as a receptor for this FeLV-B variant. There were several key differences between the two studies: our study analyzed infectivity using human and hamster Pit2, whereas Tailor and Kabat examined infection with rat Pit2. The two FeLV-B clones used for these studies differed in the amount of envelope that was derived from enFeLV sequences, as well as at several specific sequence positions in the envelope gene. Finally, there were major differences in the complementary envelope sequences used to construct the chimera: we generated envelope chimeras between FeLV-A and FeLV-B-90Z, whereas Tailor and Kabat engineered chimeras between amphotropic MuLV and FeLV-B-GA. Thus, it is unclear whether the differences observed between the two studies represented differences in Pit2 receptor specificity of the FeLV-B variant or chimera examined or differences in function among the different receptor homologs. Because of the potential utility of FeLV-B as a dualtropic virus for gene delivery, we undertook this study to determine the basis for these different findings. These studies demonstrated that an arginine at position 73 in VRA is critical for FeLV-B infection using the HuPit2 receptor.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cell lines. Mus dunni tail fibroblast (MDTF) cells and their derivatives and AH927 feline fibroblast cells were grown in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum, 100 U of penicillin per ml, 100 µg of streptomycin per ml, 0.25 mg of amphotericin fungicide per ml, and 2 mM L-glutamine (complete DMEM). MDTF-HuPit1 (39), MDTF-HuPit2 (11), and MDTF-HaPit2 (39) cells were grown in complete DMEM supplemented with 0.6% G418.

Construction of viral subclones. A packaging-defective genome encoding FeLV Gag and Pol proteins, 61E-LTR- gag-pol, was constructed from a full-length packaging-defective FeLV genome, 61E-. The 61E- clone was derived from an FeLV-A proviral clone, 61E, and was shown to express the mature forms of the Gag, Pol and Envelope proteins (7). 61E-LTR- gag-pol was generated by subcloning a region containing the 5 along terminal repeat (LTR) and Gag-Pol from 61E- into pZeoSV (Invitrogen). Briefly, p61E- was digested with AflIII, which cuts once, at position 6199, in the full-length 61E proviral genome, 218 bases into the envelope open reading frame. Blunt ends were created using Klenow polymerase, and the DNA was then digested with EcoRI, which cuts outside the FeLV genome in the polylinker. The resulting ~6.1-kb fragment was cloned into EcoRI/EcoRV-digested pZeoSV.

View larger version (31K): [in this window] [in a new window]   FIG. 1.   Schematic representation of the parental and chimeric envelope constructs. A 2.1-kb XhoI/BamHI restriction fragment encoding the FeLV-A-61E envelope protein was cloned into the mammalian expression vector pcDNA3.1 zeo() as indicated. Locations of the restrictions sites used for cloning are indicated, with sequence positions shown relative to the full-length FeLV-A-61E genome. The approximate locations of VRA, VRB, and PRR are indicated. White indicates sequences that are derived from FeLV-A-61E in the constructs. The sequences within the parental FeLV-B-90Z and FeLV-B-GA envelopes that were predicted to be derived from enFeLV sequences are indicated in black, whereas the regions within the FeLV-B clones that are highly related to FeLV-A are shown with light shading. The approximate locations of the recombination sites are indicated, although the latter cannot be defined to the precise nucleotide or amino acid due to the homology between FeLV-A and FeLV-B in those regions. The FeLV-B SU is predicted to contain 432 amino acids.

Generation of mutants. Mutations encoding single amino acid changes were introduced into the FeLV-B-GA envelope sequence by using a PCR-based, site-directed mutagenesis strategy. For this purpose, the XhoI/SacII fragment was first cloned into pBluescript SK(+) (Stratagene Inc.) and used as a substrate for PCR. Mutagenic PCR was performed using a Quick-Change site-directed mutagenesis kit (Stratagene) according to the manufacturer's protocol and using the following primers. GAmut1 (sense; GCCAATCCTAGTCCGCACCAAATATATAATGTAACTTGG) and GAmut2 (antisense; GTCCAAGTTACATTATATATTTGGTGCGGACTAGGATTGGC) were used to generate a GTGATA codon change (underlined) leading to a valine-to-isoleucine change at position 8. Primers GAmut3 (sense; CCTATGAGGAGGTGGCGACAGAGAAACACACC) and GAmut4 (antisense; GGTGTGTTTCTCTGTCGCCACCTCCTCATAGG) were used to introduce a CAACGA codon change leading to a glutamine-to-arginine change at position 73. GAmut5 (sense; GCTGTTCACTCCTCGATAACGGGAGCTAGTGAAGGG) and GAmut6 (antisense; CCCTTCACTAGCTCCCGTTATCGAGGAGTGAACAGC) were used to introduce a ACAATA codon change leading to a change from threonine to isoleucine at position 152. Following mutagenic PCR, reaction mixtures were digested overnight with DpnI and transformed into Epicurean Coli XL-1 Blue cells by electroporation. Individual plasmid clones were then sequenced. Those envelopes containing the desired mutation and no other changes were digested with XhoI and SacII and ligated into similarly digested pcDNA3.1-61Eenv.

Generation of cell-free virus for single cycle infection assays. The pcDNA3.1-env constructs were contrasfected with 61E-LTR- gag-pol and an MuLV-based vector genome encoding -galactosidase (pRT43.2Tnlsgal 1 [36]) in 293T cells by a calcium phosphate method (Stratagene). Cells were washed twice the following day with phosphate-buffered saline and complete DMEM was added. Viral supernatants were harvested at 48 h posttransfection, filtered through a 0.22-µm-pore-size syringe, and stored at 70°C. Infections using these viruses, which were competent to undergo a single round of infection, were performed essentially as described previously (13). Briefly, MDTF cells were seeded into 24-well dishes at 2 × 10⁴ cells per well in complete DMEM. The following day, the medium was removed, and cell-free viral supernatants were diluted as needed into 1 ml of complete DMEM containing 4 µg of Polybrene per ml and added to the cells. Approximately 48 h following infection, cells were stained for -galactosidase activity, and foci of infected cells were counted by visual inspection as described previously (13).

Binding studies. Constructs encoding various FeLV surface units engineered to express the hemagglutinin (HA) epitope tag have been described elsewhere (A. Lauring, M. Anderson, and J. Overbaugh, submitted for publication). These constructs encode the open reading frame for the envelope protein, including the signal peptide, with a C-terminal deletion that removes the last 10 amino acids of the SU and the entire TM domain so that the SU should be shed from the cell. Two copies of the HA epitope are in frame at the C terminus of the SU to permit detection. The methods for immunoprecipitation and Western blot analyses to detect expression and flow cytometry to detect binding of the tagged SU are described elsewhere (Lauring et al., submitted). The HA.11 antibody used for these analyses was obtained from Covance, Berkeley, Calif.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Generation of a two-component FeLV packaging system to study receptor specificity. Previously, we defined a 107-base sequence in the untranslated region between the 5' LTR and gag coding sequences of FeLV that leads to a greater than 2-log reduction in specific encapsidation of the FeLV RNA genome (7). To facilitate the construction and testing of FeLV envelope variants, we constructed a plasmid expressing Gag and polymerase from an FeLV-61E genome lacking this 107-base sequence (61E- [7]). While the 107-base deletion renders the provirus defective in its ability to encapsidate viral RNA, Gag and Pol proteins are expressed and processed normally (7). Using the 61E- genome, we removed the majority of the envelope gene to make a plasmid expressing FeLV core and polymerase proteins but not envelope. The envelope genes of interest were then cloned into an expression plasmid behind a CMV promoter (Fig. 1) and introduced into cells in trans. MuLV-derived retroviral vectors encoding selectable marker genes are efficiently encapsidated into FeLV particles (7), and the various markers can be used to monitor a single cycle of infection. In this study, we used an MuLV-based vector encoding a -galactosidase gene (36) to monitor gene transfer, and viral titer was defined by the number of cell foci that expressed -galactosidase. We found that in transient transfection experiments using 293T cells, expression of 61E-LTR- gag-pol and the FeLV envelope from separate plasmids yielded viral titers comparable to that of the of a full-length proviral clone (approximately 10⁵ to 10⁶/ml [Fig. 2]). Parallel transfections in which the plasmid encoding envelope was not included did not yield detectable infectious virus (not shown). Thus, this system allows us to generate virus with different SU proteins but with the same structural and enzymatic proteins derived from FeLV-A, which is the parental virus from which FeLV-B variants evolve.

View larger version (56K): [in this window] [in a new window]   FIG. 2.   Infection of MDTF cells expressing various Pit receptors. The indicated cells were infected with -galactosidase vector pseudotyped with the indicated viral SUs as described in Materials and Methods. Results are shown as log of the focus-forming units (ffu) per milliliter and are based on the results of duplicate experiments. These results are representative of at least four independent experiments.

FeLV-B-GA envelope does not permit entry using the HuPit2 or HaPit2 receptor. We have shown that HuPit2 and hamster Pit2 (HaPit2) can act as receptors for some FeLV-Bs (6). In the study of Tailor and Kabat, virus with an FeLV-B-GA envelope was unable to infect cells expressing rat Pit2, but other Pit2 proteins were not examined (34). To determine if FeLV-B-GA could infect cells using HuPit2 or HaPit2, we examined the ability of a virus bearing this envelope to infect MDTF cells expressing various Pit receptors. MDTF cells are not infectable by FeLV-B because the murine homologues of the Pit1 and Pit2 receptors do not function as a receptor for FeLV-B (38). We found that neither HuPit2 nor HaPit2 could confer susceptibility to infection by a virus carrying a FeLV-B-GA SU (Fig. 2). All viruses, including the virus with FeLV-B-GA SU, could readily infect both feline fibroblast cells and MDTF cells expressing HuPit1 (Fig. 2), demonstrating that the envelope proteins present on these vectors were functional.

A single mutation in VRA is a determinant for Pit2 receptor specificity. We determined the nucleotide sequence for the portion of the FeLV-B-90Z and FeLV-B-GA envelope gene encoding the SU. We noted several specific codon deviations from the previously reported sequences of these regions. An alignment of the two FeLVs shows that there are three residues that differ between the FeLV-B-90Z and FeLV-B-GA in the RBD of the envelope open reading frame (Fig. 3). One mutation was in the N terminus at position 8 of the mature SU the second was in VRA (position 73), and the third was in VRB (position 152) (Fig. 3). Of the three residues that differ between these two isolates of FeLV-B, only one represents a nonconservative change: the arginine versus glutamine in the VRA domain. We mutated the FeLV-B-GA_RBD envelope at each of these positions and examined whether vectors bearing these mutant envelopes could infect cells expressing the HuPit2 receptor. The vector containing envelope proteins encoding a glutamine-to arginine change at position 73 (FeLV-B-GA_RBD-73QR SU) was able to infect cells expressing the HuPit2 receptor, whereas the other vectors containing either of the other mutant envelopes failed to do so (10 focus-forming units/ml [Fig. 4]). As was observed for FeLV-B-90Z_RBD, infection with a virus pseudotyped with FeLV-B-GA_RBD-73QR SU was 1 to 2 log units lower in cells expressing HuPit2 versus HuPit1. Because the levels of receptor expression cannot be directly compared in these two cell lines, it is unclear if these differences reflect a lower level of expression of HuPit2 that may be suboptimal for FeLV-B infection or actual differences in how efficiently these viruses infect cells expressing HuPit1 and HuPit2. In either case, the studies comparing infection of viruses with FeLV-B-GA_RBD-73QR SU versus FeLV-B-GA_RBD SU in MDTF-HuPit2 cells show that the arginine at position 73 in VRA plays a key role in determining HuPit2 receptor specificity.

View larger version (25K): [in this window] [in a new window]   FIG. 3.   Schematic of the sequence differences between FeLV-B-90Z SU and FeLV-B-GA SU. The sequence differences between the two constructs are shown, with shading applied as described in the legend to Fig. 1. The FeLV-B-GA SU constructs that were generated with individual sequence changes are shown below, with the sequence changes indicated. The cluster of lysine residues that are unique to the FeLV-B-90Z SU are highlighted with a box. The layout is otherwise similar to that in Fig. 1.

View larger version (52K): [in this window] [in a new window]   FIG. 4.   Infection of MDTF-HuPit2 cells with vectors pseudotyped with FeLV-B-GA mutant SUs. The SUs used in this study are shown in Fig. 3. For details, see the legend to Fig. 2.

73QR

Analyses of envelope binding to HuPit1 versus HuPit2 receptors. To assess the role of the FeLV-B VRA and C-terminal sequences in receptor binding, we used flow cytometric analyses to detect SU bound to MDTF cells expressing Pit receptors. A secreted form of SU with a C-terminal HA epitope tag was expressed in human 293T cells. Approximately equal levels of the different FeLV-B SUs were detected in the supernatant by Western blotting (Fig. 5). We could readily detect binding of each of the FeLV-B SUs to MDTF cells expressing Pit1, as judged by a shift in fluorescence resulting from antibody binding to the HA epitope-tagged SU protein (Fig. 6A). In contrast, there was reduced binding of all the FeLV-B SUs to MDTF cells expressing HuPit2 even when we used 10 times more supernatant in the binding experiments (Fig. 6B and C). This is consistent with the greater than 10-fold differences in FeLV-B infectivity observed in cells expressing HuPit1 versus HuPit2 (Fig. 4). We could not detect binding of the FeLV-B-GA or FeLV-B-GA_RBD SU to MDTF-HuPit2 cells in assays using similar amounts of supernatant above the level of nonspecific binding observed with MDTF cells alone (Fig. 6B and C and data not shown for FeLV-B-GA). However, we could detect a low but reproducible shift in fluorescence in MDTF-HuPit2 compared to MDTF cells in assays using similar amounts of supernatant with FeLV-B-GA_RBD-73QR, FeLV-B-90Z, and FeLV-B-90Z_RBD SUs, each of which encodes an arginine at position 73. Thus, envelopes encoding arginine 73 bind to cells expressing HuPit2, whereas envelope proteins encoding a glutamine at this position do not. These data indicate that an arginine at position 73 is important for FeLV-B SU binding to HuPit2. However, binding of these envelopes to HuPit2 is reduced relative to HuPit1.

View larger version (61K): [in this window] [in a new window]   FIG. 5.   Western blot analyses of HA epitope-tagged SU. The supernatants from 293T cells transfected with the constructs indicated at the top were immunoprecipitated with a monoclonal antibody that recognizes the HA epitope. The immunoprecipitates were resolved on a sodium dodecyl sulfate-10% polyacrylamide gel. Western blot analysis was performed with a polyclonal antibody to the HA epitope. The mock sample represents cells that did not receive any DNA in the transfection. The details of the method are described elsewhere (Lauring et al., submitted). Sizes are indicated in kilodaltons.

View larger version (41K): [in this window] [in a new window]   FIG. 6.   Envelope binding to MDTF-HuPit1 and MDTF-HuPit2 cells. MDTF cells expressing the Pit receptors were incubated with conditioned medium containing HA epitope-tagged FeLV-B SU (shown in Fig. 5, using methods described elsewhere [Lauring et al., submitted]). Bound SU was detected by staining with a monoclonal antibody, HA.11, which is directed against the HA epitope. For each panel, the x axis is fluorescence intensity (log scale) and the y axis is the event count. Cells incubated with medium only are indicated with thin black lines. The histogram with the gray fill indicates background levels of binding of SU to naive MDTF cells. Results of SU binding to MDTF cells expressing HuPit1 (A) and HuPit2 (B and C) are shown with bold black lines. For experiments in panel A, 500 µl of supernatant was added to MDTF-HuPit1 cells. There was no detectable shift in fluorescence when 500 µl of supernatant was incubated with MDTF cells; thus, this control is superimposed on the black line in panel A (not shown). For experiments in panels B and C, 5 ml of supernatant was used because we could not detect any binding to HuPit2 when 500 µl of supernatant was used (not shown). Two representative experiments with supernatants from independent transfections are shown in panels B and C. The supernatants used for panels A and B correspond to those shown in Fig. 5.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The phosphate transport protein Pit1 can function as a receptor for all FeLV-B variants examined to date (6, 28, 34, 35). Previous studies indicated that some variants of FeLV-B can infect cells by using HuPit2 as a receptor (6). Domains critical for FeLV-B-HuPit2 interaction include VRA and VRB, both of which are within the putative RBD. Comparative analyses of two independently derived FeLV-B envelopes allowed us to identify an arginine at position 73 in VRA that is critical for binding and receptor specificity for HuPit2 but not HuPit1. Our studies also defined a domain in the C terminus of FeLV SU, outside the RBD, that affects entry via HuPit2 but not HuPit1. Collectively, this information may be useful in designing envelope proteins for gene therapy that permit a retrovirus to efficiently infect cells that express either HuPit1 or HuPit2.

The structure of the FeLV-B envelope, including the location of the predicted RBD, is largely inferred based on its relatedness to the better-studied MuLVs. The structure of the ecotropic MuLV RBD has been resolved (12), and even though FeLV-B and ecotropic MuLV RBD share very limited homology, there are some conserved regions. One of these is a highly conserved FYVCP motif found at the base of the VRA helix in MuLV (Fig. 7). The arginine at position 73 in FeLV-B is nine amino acids before this conserved cysteine, suggesting that the arginine residue we identified as important for HuPit2 binding may lie in a similar helical structure in FeLV-B envelope. Interestingly, several residues important for both amphotrophic and ecotropic MuLV receptor binding map in this region of SU, or just upstream in the first small helix or in loop 2 of VRA (1, 2, 9, 19). Two other positions in VRA of FeLV-B, at residues 60 and 61, have been identified as important for Pit1 receptor specificity (34). Collectively, these studies imply that among highly divergent retroviral envelopes, key receptor contact residues may reside in a similar structural domain in VRA.

View larger version (23K): [in this window] [in a new window]   FIG. 7.   Sequence alignment of FeLV, MuLV, and GALV VRA. The sequences within the predicted VRA are aligned, with gaps indicated by dots. For GALV and Moloney MuLV (MoMuLV), the alignments to the first four sequences were made largely using the cysteine at the beginning of the VRA helix and the conserved FYVCP motif C terminal to the helix. There was insufficient homology to accurately align sequences N terminal of the VRA helix for these two viruses. Thus, this portion of the alignment is arbitrary and is included to show the regions of MoMuLV implicated in receptor specificity. The sequences that form the helical domains and loop 2 in VRA of ecotropic MoMuLV are shown below the sequences. The position of the arginine at residue 73 is indicated by an arrow at the top of the sequence. Amino acids that have been identified as important in determining receptor specificity in other studies are shown in bold (for FeLV-B-GA [34]; for MoMuLV [1, 9, 19]; for amphotropic MuLV [A-MuLV] [2]; for 10A1 MuLV [14]).

The arginine at position 73 is found in all of the known viruses that use HuPit2 as a receptor, which includes amphotrophic and 10A1 MuLVs (Fig. 7), providing further evidence that this residue may play a critical role in interaction with this receptor. Indeed, we could reproducibly detect a low level of binding to HuPit2 with all of the FeLV-B SUs that encode arginine 73, but we did not detect evidence of binding with any of the FeLV-B SUs that encoded a glutamine at position 73. The fact that FeLV-B-GA, which encodes a glutamine at position 73, can efficiently bind and infect cells expressing HuPit1 suggests that arginine at position 73 may not be required for efficient interaction between FeLV-B SU and HuPit1. Moreover, binding of all of these FeLV-B SUs to HuPit2 is much weaker than it is to HuPit1. Thus, residues that contact HuPit1 and HuPit2 may differ, even among a virus competent to use both receptors.

Alignment of FeLV-B and GALV VRA suggests that GALV also encodes an arginine at the position analogous to residue 73 (Fig. 7). However, GALV cannot infect cells expressing the Pit2 receptor, suggesting that arginine 73 is necessary but not sufficient for HuPit2 receptor usage. Our data with FeLV-B chimeras are consistent with this model because the FeLV-B-90Z SU, which encodes arginine at position 73, cannot infect cells by using HuPit2 as a receptor. We could detect a low level of FeLV-B-90Z SU binding to cells expressing HuPit2, similar to what was observed for FeLV-B-90Z_RBD. More sensitive methods for measuring binding will be needed to clarify these findings, which suggest that sequences in the C terminus of FeLV-B-90Z SU may inhibit postbinding stages in viral entry. Thus, it is possible that sequences in the C terminus of GALV sterically inhibit HuPit2 interactions in a similar manner. This could be tested by generating a chimera with the N terminus of GALV SU and the C terminus of FeLV-A-61E SU since the latter relieves postbinding restrictions in a similar chimera with FeLV-B-90Z SU.

Although the precise recombinational breakpoints between FeLV-A and enFeLV sequences can only be approximated for FeLV-B-GA, sequence alignments suggest that this variant derived enFeLV sequences that would encode through amino acid ~270 in SU (6) (Fig. 1). The major difference between FeLV-B-GA_RBD-73QR, which can use HuPit2 to enter cells, and FeLV-B-90Z, which cannot, is the their recombinational breakpoint, although there is one amino acid difference in the PRR (Fig. 3). This suggests that sequences downstream of PRR, in the C-terminal ~160 amino acids in FeLV-B-90Z SU, may be inhibiting interactions between the FeLV-B-90Z SU and HuPit2. The most notable difference between FeLV-B-GA and FeLV-B-90Z in this region is a cluster of eight amino acids that includes four lysine residues unique to FeLV-B-90Z. Thus, we speculate that this highly charged region in the C terminus of 90Z SU may play a role in the postbinding restrictions observed for FeLV-B-90Z infection of cells expressing HuPit2. This effect of the FeLV-B-90Z C-terminal domain was observed only with the HuPit2 receptor, which binds the SU much less efficiently than HuPit1. Thus, it is possible that a stable interaction between the RBD and the receptor permits a weak interaction of the C-terminal domain of SU with either the RBD or the receptor to promote the required postbinding stages in entry.

Recently, Lavillette et al. (17) have shown that the analogous domain in MuLV, which forms a disulfide loop, plays a critical role in fusion activation. In light of these findings, we would predict that a virus with an FeLV-B-90Z SU may be unable to initiate fusion when bound to HuPit2. Interestingly, the lysine cluster in FeLV-B-90Z is immediately adjacent to the site of an insertion in the FeLV-T variants that determines their cell tropism (10, 13), suggesting that this C-terminal domain could define a more general structural determinant for FeLV receptor specificity that resides outside the actual binding domain. On the basis of the MuLV studies (17), this receptor determinant may be predicted to form a fusion activation domain.

Pit1 and Pit2 are widely expressed, though at different levels in various tissues. For example, Pit1 is more highly expressed in bone marrow cells and other progenitor cells, leading to the suggestion that viruses that use the HuPit1 receptor may be more useful for gene therapy (16, 27). The only other virus that can efficiently enter cells using HuPit1 is GALV, but this virus does not infect cells using HuPit2. Thus, a virus like the engineered FeLV-B forms described here, which can enter cells expressing both HuPit1 and HuPit2, may allow for optimal targeting of stem or progenitor cells and at the same time allow for a very broad host cell specificity as a result of its ability to use either of two broadly expressed receptors.