The X/PMV viruses of mice represent a highly polymorphic group. While most isolates have either XMV or PMV host range, several have been described with atypical species tropism 14 18 . To characterize host range variation within the X/PMVs, we screened a panel of X/PMVs along with amphotropic MLV (A-MLV) (Table 1) for infectivity in rodent cells with different XPR1 receptors (Fig. 1). In addition to 6 laboratory mouse virus isolates and 3 previously described wild mouse isolates, this panel included a novel isolate from the eastern European wild-mouse derived strain, CZECH/EiJ, and XMRV, a xenotropic-like virus isolated from human prostate cancer patients 16 17 . LacZ pseudotypes were generated for these viruses and tested for infectivity on mouse cells carrying the 4 known Mus Xpr1 variants, on rat and hamster cells, and on nonrestrictive mink lung cells.

The two PMV isolates showed the same pattern of infectivity on mouse cells carrying the 4 variants of Xpr1 (Table 2). Both viruses infected NIH 3T3 (Xpr1 ⁿ ) and cells carrying Xpr1 ^sxv , but did not infect cells of M. pahari (Xpr1 ^p ) or cells carrying Xpr1 ^c . Chinese hamster cells were resistant to both viruses. Rat2 cells, however, were efficiently infected by HIX PMV, but were very resistant to FrMCF (Table 2). The resistance to FrMCF was observed only with this particular Friend PMV isolate as Rat2 cells were efficiently infected by three other Friend MCF PMVs as well as by MCF 247 (not shown). Resistance to this FrMCF was also observed in rat XC cells (not shown) indicating that this resistance is not limited to the Rat2 cell line.

Env sequence comparisons identified scattered substitutions that distinguish FrMCF and other PMVs and the presence of a 9 codon deletion unique to FrMCF (Fig. 2). This deletion has been identified in few replication competent PMVs 19 20 , although it is a hallmark of modified PMV-related endogenous env genes (Mpmvs) 21 . This deletion is outside the Env receptor binding domain (RBD) 22 , and lies in the proline-rich domain (PRD), a region that is thought to mediate conformational changes in Env during infection and to influence membrane fusion 23 .

In an attempt to recover novel PMV-type recombinant viruses, we inoculated mice of different taxonomic groups with MoMLV. Using this approach, we previously described a set of replication competent recombinant PMVs isolated from MoMLV inoculated M. spretus 24 . In the present study, we inoculated 11 CZECHII/EiJ mice, an inbred line of M. m. musculus. These mice carry dozens of XMV env genes, but few PMV copies 25 , unlike the common strains of laboratory mice which carry multiple XMV and PMV endogenous env genes 21 . Spleen or thymus cells from 2 month old inoculated mice were plated on M. dunni and/or mink cells, and media collected from one of these M. dunni cultures induced MCF-type foci on mink cells (not shown). Southern blotting of virus infected cells with ~ 120 bp env-specific probes identified sequences related to XMVs, but no PMV env-related fragments (not shown). The virus was biologically cloned by limiting dilution, and its env gene was cloned and sequenced.

The sequenced Cz524 env was not an env recombinant derived from the inoculated MoMLV; no segments identical to MoMLV were identified although the breakpoint positions identified in other MoMLV recombinants cluster in an env region just downstream of PRD 19 . Consistent with the Southern blot analysis, the env sequence of Cz524 MLV showed closest homology to XMVs (Fig. 2). Of the 33 RBD amino acid residues that distinguish Cz524 from MCF 247 PMV or CAST-X XMV, Cz524 resembled the prototype XMV at 26 sites, the prototype PMV at 4 sites, and had novel residues at 3 sites. The major difference between Cz524 and XMV viruses is in VRA, the first variable domain in SUenv, where PMVs have a 4 codon deletion relative to XMVs. Cz524 has a 3 codon deletion relative to XMVs at this same position, and there is a novel substitution at the 4^th site typically deleted in PMVs.

LacZ pseudotypes carrying the Cz524 Env were tested for infectivity on rodent and mink cells (Table 2). Cz524 shows a novel pattern of species tropism that differs from that of CasE#1 and all XMVs and PMVs tested. This virus infects mink cells and cells carrying Xpr1 ^sxv with high efficiency, shows very poor infectivity on cells carrying Xpr1 ^c and on Rat2 cells, and is restricted by hamster cells and cells carrying the mouse Xpr1 ⁿ and Xpr1 ^p variants.

Three of the four XPR1 variants of Mus supported replication of XMVs; only Xpr1 ⁿ of the laboratory mouse strains failed to mediate infection of any of these viruses (Table 2). Among the susceptible mouse cells, there was variation in infectivity by the 3 XMVs, and this could be due to receptor polymorphism or non-receptor factors. The pseudotypes that we used here carry the Gag proteins of their parental viruses, and studies on some XMVs 26 indicates that they may be subject to restriction by Fv1, a mouse gene responsible for post-entry virus resistance that targets specific capsid residues. The capsid sequence for one of the 3 XMVs used in this analysis, XMRV, has been determined 16 , and it carries the Fv1 ⁿ target residue E110 27 . The NXPR-S and NXPR-C cells carrying Xpr1 ^sxv and Xpr1 ^c have the restrictive Fv1 ⁿ allele. Therefore, to determine if our XMV pseudotypes are subject to Fv1 restriction, we examined infectivity in a second cell line carrying Xpr1 ^sxv , the Fv1-null M. dunni cell line (Table 2). We noted an Fv1-type 100-1000 fold reduction in infectivity of all 3 XMVs in NXPR-S relative to M. dunni. A similar 1000-fold reduction for CAST-X was observed in NFS/N cells carrying Xpr1 ^c , but infectivity with XMRV and AKR6 was further reduced in these cells, suggesting either that this XPR1 variant is not an efficient receptor for these particular XMV viruses, or that additional factors inhibit infection. These observations taken together indicate that while there are some infectivity differences that are consistent with Fv1 restriction, both Xpr1 ^sxv and Xpr1 ^c receptor variants function as XMV receptors for all 3 isolates.

AKR6 MLV shows typical xenotropic host range; it fails to infect mouse cells, but can infect cells of heterologous species 14 . When tested on mouse, rat, and mink cells, AKR6 showed the same general pattern of infectivity as the wild mouse CAST-X virus (Table 2) and NZB-IU-6 XMV (not shown). However, while other mouse XMVs showed low but reproducibly detectable infectivity in E36 Chinese hamster cells, AKR6 showed no such infectivity. Because infection of hamster cells with most gammaretroviruses is blocked by glycosylation 28 , we examined virus infectivity in E36 cells treated with inhibitors of glycosylation (Table 3), as well as in Lec8 cells, a hamster glycosylation mutant that lacks GlcNAc-transferase I (Table 4). The reduction of glycosylation in hamster cells by mutation or by exposure to inhibitors results in increased susceptibility to ecotropic MLVs (not shown) and XMVs (Tables 3, 4), but did not relieve resistance to PMVs as observed previously 28 , or to Cz524 or CasE#1. Unlike other viruses with XMV host range, however, AKR6 did not infect inhibitor-treated E36 cells or Lec8 cells. The human-derived XMV, XMRV, shows the PMV-like restriction of AKR6 in hamster cells; XMRV does not infect Lec8 cells or inhibitor-treated E36 cells (Tables 3, 4).

CasE#1 efficiently infected M. dunni cells (Xpr1 ^sxv ) and M. pahari cells (Xpr1 ^p ) as well as rat and mink cells, but failed to infect hamster cells, NIH 3T3 (Xpr1 ⁿ ) and cells carrying Xpr1 ^c (Table 2). Reduced infectivity of this virus in NXPR-S relative to M. dunni suggests it may be subject to Fv1 restriction. The overall pattern of CasE#1 infectivity is distinct from that of the XMVs, PMVs and Cz524.

To define receptor determinants for this panel of viruses, we tested 6 viruses for infectivity on E36 Chinese hamster cells expressing Xpr1 ⁿ or Xpr1 ^p as well as variants of the mouse XPR1 receptor (Fig. 3A). These transfectants included previously described chimeras between Xpr1 ^p and Xpr1 ⁿ and two Xpr1 ⁿ mutations that independently introduce sensitivity to XMVs 6 7 , namely E500K (mutant ECL3-1) and Δ582T (ECL4-1). We also generated a novel set of ECL3 substitutions made in Xpr1 ^p or Xpr1 ⁿ . Expression of the novel constructs in E36 cells was confirmed by western analysis (Fig. 3B).

Two Xpr1 variants reproduce the susceptibility pattern of M. pahari, that is, susceptibility to CasE#1 and all XMVs, but show resistance to PMVs and Cz524. Chimera Pah3/4 carries ECL3 and ECL4 of Xpr1 ^p in an Xpr1 ⁿ backbone demonstrating that the receptor determinants for XMVs and CasE#1 are in the ECL3 and ECL4 domains 7 (Fig. 3A). The same pattern of susceptibility is shown by the single ECL3 substitution P505S, although this change introduces an N-linked glycosylation site. The reciprocal change, S505P, made in Xpr1 ⁿ , abolishes an N-linked glycosylation site, but does not alter the Xpr1 ⁿ infectivity profile, that is, susceptibility to PMVs only. This suggests that residues at position 505 are not critical for PMV, XMV or CasE#1 entry. Western analysis shows that the P505S and S505P XPR1s show no obvious size differences suggesting that this glycosylation site is not utilized (Fig. 3B).

Reciprocal chimeras Pah4 and Pah3 contain, respectively, Xpr1 ^p ECL3 (Pah3) or ECL4 (Pah4) in an Xpr1 ⁿ backbone and are dramatically different receptors 7 . Pah3 is nonfunctional as a receptor for any of the tested viruses. Pah4 retains Xpr1 ⁿ susceptibility to PMVs, but the combination of Xpr1 ⁿ ECL3 and Xpr1 ^p ECL4 introduces susceptibility to Cz524, CasE#1 and XMVs, although all inefficiently infect these cells except Cast-X.

The difference between the Pah3 and Pah4 chimeras suggests that the PMV receptor determinants are in ECL3; so we introduced substitutions at codon sites that distinquish ECL3 of Xpr1 ⁿ and Xpr1 ^p (Fig. 1). Mutant ESTV has substitutions in the 4 most C-terminal of these 6 sites in Xpr1 ^p , and like Pah4, mediates susceptibility to PMVs. Making the reciprocal changes at these 4 sites in Xpr1 ⁿ (mutant KPYK) results in loss of PMV susceptibility. Thus, some combination of residues at these 4 sites specifies the PMV receptor. Substitutions at positions 500, 507 and 508, all resulted in changes in the pattern of PMV susceptibility. Reciprocal substitutions were made at ECL3 position 507 in Xpr1 ^p (Y507T) and Xpr1 ⁿ (T507Y), and a double Xpr1 ^p mutant carried K508V and Y507T. The two Xpr1 ^p mutants acquired susceptibility to Cz524 and limited susceptibility to PMVs. T507Y retained susceptibility to HIX although infectivity with FrMCF was barely detectable. Finally, Xpr1 ⁿ with E500K (mutant ECL3-1) is an efficient receptor for HIX, but a poor receptor for FrMCF. These results indicate that PMV infectivity is influenced by residues at the C-terminal end of ECL3, but that different PMVs rely on different residue combinations.

As shown previously, mutations E500K and Δ582T independently convert Xpr1 ⁿ into a receptor for XMV 6 . Only one of these changes, Δ582T (mutant ECL4-1), generates a receptor for CasE#1 7 . This mutation also produces a receptor for Cz524, results in reduced susceptibility to AKR6, but does not change susceptibility to PMV. In contrast, E500K (mutant ECL4-1) is efficiently infected by AKR6. Thus, K500 provides a more efficient receptor for AKR6 than does the T582 insertion.

None of the ECL3 mutations in Xpr1 ⁿ introduces susceptibility to CasE#1, confirming that its primary receptor determinant is in ECL4, although as for AKR6, substitutions in ECL3 residues influence the efficiency of infection.

Susceptibility to Cz524 is introduced into Xpr1 ⁿ by either of the XMV determinants, Δ582T or E500K, or by the Y507T and K508V substitutions in Xpr1 ^p that also introduce some susceptibility to PMVs. However, other mutant receptors carrying these residues are not susceptible to Cz524, suggesting that Cz524 has additional requirements for entry. The 4 most efficient Cz524 receptors are also efficient XMV receptors that also mediate PMV infection, suggesting that Cz524 virus utilizes receptor determinants required by both PMVs and XMVs.

The characterization of entry-based virus resistance factors has obvious importance for a broader understanding of how viruses spread and adapt to new hosts, and how natural populations adapt to retrovirus infections. Infectious XMVs and endogenous X/PMVs have been identified in Eurasian mice, and these mice have evolved two protective mechanisms that restrict infection at the level of entry. Receptors can be blocked by Env glycoprotein produced by endogenous retroviruses (ERVs), and ERVs with intact env genes have been linked to the resistance genes Fv4, Rmcf and Rmcf2 29 30 31 . More commonly, resistance to retrovirus entry is due to polymorphic mutations in the cell surface receptor. The present study indicates that the sequence variations that distinguish the rodent XPR1 receptors can result in subtle differences in the efficiency of virus infection or complete resistance to specific X/PMVs. Additional functional variants of XPR1 and determinants for X/PMV entry may be identified by expanding this analysis to non-rodent species exhibiting different virus susceptibility profiles [ 14 ; CAK, unpublished observations], as recently shown by a recent analysis of human/mouse XPR1 chimeras 15 .

Receptor-mediated virus restriction can result in the outgrowth of virus variants able to circumvent such blocks by adapting to receptor variation, by using alternative receptors or, as in the case of XMVs, using alternative receptor determinants on the same protein. The panel of variant viruses used in the present study were all the products of such adaptations and included naturally occurring mouse-derived isolates, the human-adapted XMRV, and HIX and FrMCF, variants adapted to cultured cell lines or laboratory-bred animals. These viruses differ from one another at multiple sites within env. Mutagenesis studies focusing on these RBD differences and other env regions implicated in receptor binding and/or fusion should provide further information on the critical residues involved in entry and the factors that limit or extend receptor usage.

Defining genetic factors that underlie resistance to mouse gammaretroviruses is important because retroviruses are capable of trans-species transmission, and retroviruses that cluster with mouse gammaretroviruses are widespread among vertebrates. Martin and colleagues 32 found MLV-related ERVs in approximately one-fourth of the vertebrate taxa and identified recent zoonotic transmissions from mammals to birds and from eutherians to metatherians. Infectious viruses resulting from transspecies transmissions have been isolated from koalas and gibbon apes 33 34 35 . One of the viruses used in the present study, XMRV, is an infectious MLV-related virus from human prostate cancer patients 16 17 , and it should be noted that similar viruses have also been reported in cell lines derived from other human tumors 36 . It would not be surprising to find more examples of interspecies transmissions involving MLVs, since mice have a worldwide geographic distribution and all mammalian species tested have functional XPR1 receptors [ 14 ; CAK, unpublished observation]. Thus, the examination of the co-evolution of the XPR1 receptor and the X/PMVs should contribute to an understanding of the natural history of infectious pathogenic gammaretroviruses in their murine hosts and provide a foundation for the study of epizoonotic infections.

CAST-X is a xenotropic MLV isolated in our laboratory from the spleen of a CAST/EiJ mouse 7 . The human xenotropic-related virus, XMRV 16 17 , was kindly provided by R. Silverman (Cleveland Clinic, Cleveland, OH). Cz524 is a novel MLV isolated from the spleen of a CZECHII/EiJ mouse 2 months after inoculation with MoMLV. Other viruses are listed in Table 1 and were originally obtained from Dr. J. Hartley (NIAID, Bethesda, MD) along with 3 additional Friend PMVs: Fr-MCF-1, FrMCF A1807 and MCF-Fr Nx.

Susceptibility to X/PMVs was tested in various cell lines including M. dunni 37 , NIH 3T3, mink Mv-1-Lu (ATCC CCL64), Rat2 (CRL-1764), Chinese hamster cells E36 38 and Lec8 (CRL-1737), a cell line from the Asian species M. pahari obtained from J. Rodgers (Baylor College of Medicine, Houston), rat XC cells (CCL-165), and E36 hamster cells transfected with Xpr1 variants. Embryo fibroblasts were prepared from the progeny of crosses between CAST/Rp and NFS/N mice that were homozygous for Xpr1 ^c ; these cells are termed NXPR-C. NXPR-S embryo fibroblast cells were prepared from NFS/N-Xpr1 ^sxv congenic mice 39 . CAST/Rp mice were obtained from R. Elliott (Roswell Park Cancer Institute, Buffalo, NY). CZECHII/EiJ mice were obtained from The Jackson Laboratory (Bar Harbor, Maine). NFS/N and congenic mice were bred in our laboratory.

LacZ pseudotypes were generated for all viruses by infection of the packaging cell line GP2-293 (Clontech, Mountain View, CA) that had been transfected with pCL-MFG-LacZ (Imgenex, SanDiego, CA) along with pMSCVpuro (Clontech) by J. Silver (NIAID, Bethesda, MD). Filtered media from the virus infected cultures contained a mixture of infectious virus and LacZ pseudotypes. Cells were infected with appropriate dilutions of these pseudotype virus stocks in the presence of 4-8 μg/ml polybrene. One day after infection, cells were fixed with 0.4% glutaraldehyde and assayed for β-galactosidase activity using as substrate 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-Gal, 2 mg/ml; ICN Biomedicals, Aurora, Ohio). Infectious titers were expressed as the number of blue cells per 100 microliters of virus supernatant.

Cells were treated by various inhibitors of N-linked glycosylation as follows: deoxymannojirimycin (DMM, 100 ug/ml); castanospermine (CST, 100 ug/ml), and 2-deoxy-D-glucose (2DG, 25 mM). All inhibitors were obtained from SIGMA (La Jolla, CA). Inhibitors were added to cultures that had been seeded the previous day and were not removed when pseudotype virus and polybrene were added 18-24 hours later.

Seven novel mutant variants of the Xpr1 gene were generated using previously described clones of Xpr1 ⁿ and Xpr1 ^p 7 . The mutants KPYK and ESTV were made by exchanging fragments of the 2 receptors using primers 1F, 1R, 2F, 2R (Table 5). All other mutants were generated by mutagenesis PCR using QuikChange II XL Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA). All mutants were confirmed by sequencing.

The recombinant plasmids were transfected into E36 Chinese hamster cells. Stable transfectants were selected with geneticin (830 μg/ml), and the expression of the Xpr1 variants was confirmed by western analysis. Proteins were extracted from transfected cells with M-PER Mammalian Protein Extraction Reagent (Pierce, Rockford, IL). The expression vector used for XPR1 inserts a V5 epitope at the C-terminus; XPR1 expression was detected in western blots using anti-V5 antibody (Invitrogen) followed by goat anti-mouse IgG conjugated with HRP (Invitrogen). The membrane was then stripped and incubated with mouse anti-α-tubulin (Sigma, St. Louis, Mo) and goat anti-mouse IgG conjugated with HRP (Invitrogen).

RNA was extracted from Cz524, AKR6 and FrMCF virus infected mink cells. The full-length 2.1 kb env gene of Cz524 and AKR6 and the 0.9 kb segment of the 5' end of the FrMCF env were amplified by RT-PCR, cloned into pCR2.1-TOPO and sequenced. Primer sequences available on request. One substitution in the leader sequence, P4S, distinguishes our AKR6 from GenBank No. DQ199948. The sequences of the env genes of Cz524 and FrMCF were deposited [GenBank:GQ375545 and GenBank:GQ420673].