INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Viral subgroups of the murine C-type viruses have been defined on the basis of host range, generally corresponding to distinct interference groups defined by receptor usage (40). Five distinct interference groups have been established for the well-studied isolates of Mus musculus, and the cellular receptors for these viruses have all been identified (Table 1). Three families of receptors have been characterized for these viruses, all of which share multiple transmembrane-spanning topology and are implicated in transporter activity: (i) cationic amino acid transport for the cationic amino acid transporter (CAT) family (20, 48), (ii) inorganic phosphate transport for the Pit family (18, 37), and (iii) a presumed phosphate transport activity for the newly identified Sgy1 family (44). Whereas some viruses, such as 10A1 murine leukemia virus (10A1 MuLV), recognize several members of the same transporter family (e.g., Pit1 and Pit2) as well as homologous receptors of different species (e.g., mouse, rat, and human) (34), other viruses recognize only one or at most two members of a receptor family and are unable to recognize homologous receptors of other species (e.g., ecotropic MuLV recognizes only the rodent CAT1 and, with less affinity, the CAT3 receptor, but not the CAT2 receptor or CAT homologues of other species) (19, 30). Interestingly, viruses of different interference groups may use the same receptor but show different affinities to homologues in other species (e.g., the xenotropic MuLV does not recognize all murine homologues of its human cellular receptor, all of which, however, are recognized by polytropic MuLVs [5, 28, 44, 50]). Determining the domains in the viral Env protein complex and the receptor molecule which dictate this specificity has been a subject of intense research over the last few years (11).

In addition to being isolated from M. musculus, type C retroviruses have also been isolated from Mus caroli (25), Mus cervicolor (7), and, more recently, Mus dunni (9). With the exception of the latter, the interference groups of these viruses have not been well defined. Earlier studies showed that two classes of type C retroviruses could be isolated from M. cervicolor. Type CI has a classical xenotropic host range (i.e., does not infect cells of inbred mouse strains but infects cells of wild mice and other mammalian species) and, based on immunological and nucleic acid hybridization criteria, is closely related to the virus isolated from M. caroli (7) and antigenically related to type C viruses from woolly monkeys (e.g., simian sarcoma-associated virus) and gibbon apes (gibbon ape leukemia virus [GALV]). Based on close sequence similarity to GALV, the M. dunni endogenous virus (MDEV) isolate is also related to this group (49), although it uses a virus receptor distinct from that of GALV.

The host range of the CII type isolates of M. cervicolor is limited to M. musculus. Based on antigenic determinants and RNA homology, these type CII viruses appear to be closely related to MuLVs isolated from M. musculus (7). Despite similarities to the ecotropic MuLVs, genomic mapping studies showed that CII isolates use a distinct cellular receptor (29, 39). Whereas the cellular receptor for the CII isolate M813 is located on mouse chromosome 2, the CAT1 gene encoding the ecotropic MuLV receptor is located on chromosome 5. The mapping of the gene encoding the Pit1 receptor to chromosome 2 led to the speculation that M813 may use the same receptor as MuLV 10A1 (1). As we have recently shown that viruses that use this family of phosphate transporters for cell entry induce a severe spongiform encephalomyelopathy in mice (35, 36), we were interested in establishing the interference group of M813.

The studies presented here show that the CII isolate M813 uses none of the known MuLV retroviral receptors for cell entry. Sequence analysis of a molecular clone of the env gene confirmed its homology with other MuLVs but showed considerable sequence diversity in the regions known to be important for receptor recognition. Replacement of the env gene of Moloney MuLV (Mo-MuLV) with that of M813 results in a replication-competent virus with a host range and receptor interference analogous to those of the original M813. Thus, M813 defines a novel receptor-interference group of the type C MuLVs.

(This work was done as part of the doctoral thesis of S. Hein in the Department of Biology at the University of Hamburg.)

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Virus-producing cell lines and cell culture. SC1 mouse fibroblasts were used as virus producers in all experiments. SC1 cells expressing M813 were obtained by infection of SC1 cells with supernatant of either M813(NIH) or M813/M813 cell lines, kindly provided by U. Rapp (39). No differences in virus titers or host interference between the virus expressed by these two independent sources were observed, so all experiments described here were performed with virus derived from M813(NIH). Molecular clones of ecotropic (Mo-MuLV, clone mov3 [15]), amphotropic (Mo-AmphoV, clone AMS [33]), polytropic (Moloney mink cell focus-forming virus [Mo-MCFV], clone 16I [10]), and 10A1 (10A1 and Mo-10A1V, clone RR1 [38]) MuLVs were transfected into SC1 fibroblasts to obtain virus producers of these different interference groups. All cells were grown in Dulbecco's modified Eagle's medium with 10% fetal bovine serum.

Virus assays. Marker rescue assays were employed to verify the host range of M813 and to determine the virus interference groups. For these assays, either MPEV or SFEV retroviral vector was used, both of which are based on the murine embryonal stem cell virus (14) with either LTRs from the myeloproliferative sarcoma virus (23) or the spleen focus-forming virus SFFVp (6), respectively. For the various assays, one of the following marker genes was incorporated in the retroviral backbone: (i) the neomycin resistance gene (neo) derived from the Tn5 transposon, which confers resistance to G418 in eukaryotes; (ii) the hph gene of Escherichia coli, conferring resistance to hygromycin; or (iii) the gene encoding the enhanced mutant of the green fluorescence protein (EGFP) from Aequorea victoria. SC1 fibroblasts expressing these retroviral vectors were established by DNA transfer by electroporation, followed either by selection in G418 or hygromycin or by cell sorting. The resulting cell lines were infected with supernatant from virus-producing cells and cultured for a minimum of 10 days to allow complete virus spread. Medium was replaced 24 h prior to harvesting of virus stocks from confluent monolayers and passed through a Millex-GP 0.22-µm-pore-size filter (Millipore).

Construction of human cell lines expressing the ecotropic CAT1 receptor. The murine CAT1 cDNA was amplified from the plasmid pJET (kindly provided by L. Albritton) by PCR using primers incorporating a NotI site upstream of the ATG (at position 199) and a HindIII site downstream of stop codon at position 2065 (2). The resulting fragment was cloned in the EcoRV site of pBluescript SK (Stratagene), excised with NotI and HindIII, and then cloned into the SFEV vector SF1-ires-Neo using the same sites. The resulting plasmid (SFEV-mCAT1-neo; R664) and MPEVneo, as a control, were transfected independently into Phoenix-ampho packaging cells together with a plasmid encoding the vesicular stomatitis virus G protein (47). After 48 h, supernatant was collected and used to infect cells of the human fibroblast line TE671. Cultures were selected for neomycin resistance in the presence of 800 to 1,000 µg of G418/ml (specific activity of 0.6). DNAs from the resulting cell lines, TE671neo and TE671neo-mCAT1, were subjected to Southern blot analysis to confirm the presence and intactness of the vector (data not shown).

Molecular cloning of env gene and sequence analysis. Virus particles collected from the supernatant of confluent monolayers of SC1-M813(NIH) cells were filtered through Millex-HA 0.45-µm-pore-size filters (Millipore) and pelleted over a 4-ml 20% sucrose cushion in SW28 ultracentrifuge tubes at 26,000 rpm for 2 h at 4°C in a Beckman LS-65 ultracentrifuge. Virus particles were resuspended in 500 µl of TNE+S buffer (50 mM Tris, pH 7.4; 150 mM NaCl; 50 mM EDTA; 2% sarcosyl), and viral poly(A)⁺ RNA was extracted using the Micro-Fast Track kit (Invitrogen). cDNA was synthesized using oligo(dT) primers. A fragment encompassing the 3' pol and 5' env sequences was amplified by PCR with Taq polymerase using the primers 5'-GGTCAGGTAGAAAGAATGAA-3' (cs252) and 5'-CACAGCCAACATTCTTG-3' (cs245) (corresponding to nucleotides [nt] 5259 and 5278 and nt 6226 to 6245 of the Mo-MuLV genome [GenBank accession no. J02255] or nt 1 to 20 and nt 1383 to 1364 of the M813 sequence compiled from this study [GenBank accession no. AF327437]). An identical fragment was also isolated from genomic DNA prepared from SC1-M813(NIH) cl5 cells.

Construction of a Mo-MuLV/M813 env recombinant. To generate a Mo-MuLV recombinant containing M813 env sequences, a primer was designed to insert a ScaI restriction site immediately upstream of the sequence encoding the proposed cleavage site of the mature M813 gp85^env (5'-GTCAGTACTCAGGGAGGGAGTCC-3'; cs265, corresponding to nt 624 to 637 of the M813 sequences.) A ScaI site (underlined) is found in Mo-MuLV at this position. The second primer (5'-CGTTCCCGCCGGCGTCTGGGGTTGCTTGGT-3'; cs284, corresponding to nt 1278 to 1293 in M813) was designed to introduce an SgrA1 site (underlined) at the 5' end of the proline-rich region of the SU, a site found in Mo-MuLV at this location. The 5' half of M813 env was amplified using these primers with Pfx polymerase and, after digestion with ScaI and SgrA1, used to replace the corresponding fragment in a plasmid encoding the entire Mo-MuLV (pUC-MoMuLV; R686). The resulting plasmid (Mo-M813; R704) was transfected into SC1 cells, and supernatant was monitored for virus spread by reverse transcriptase activity.

Nucleotide sequence accession number. The nucleotide sequences of the 3' M813 viral genome, including partial sequences of pol, the complete env gene, and the U3 and R region of the long terminal repeat (LTR), have been deposited in GenBank under the accession no. AF327437.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

M813 MuLV efficiently infects cells of mouse but not rat or human origin. Using a reverse transcriptase assay, it has been previously shown that type CII viruses from M. cervicolor infect cells derived from M. musculus (e.g., SC1 and NIH 3T3 cells) but not cells derived from M. cervicolor or from other diverse species, including rat and human (7). To verify that the clone M813 belongs to this same virus type, we used a marker rescue assay to determine the virus titers on mouse, rat, and human cells. For these studies, SC1 cells expressing the MPEVneo retroviral vector were infected with either M813 or as controls, Mo-MuLV or Mo-10A1V. After complete virus spread (10 to 14 days), the supernatant of these cells was used as a source of MPEVneo pseudotypes with different Env proteins. The transfer efficiency of these pseudotypes was determined on two or three different cell types from all three species, and the number of GTU per milliliter was calculated.

M813 MuLV shows only slight or no interference with other MuLVs. Despite its different host range, it remains possible that M813 uses the same receptor for cell entry as other MuLVs. To test this, we performed interference assays using a neoR rescue assay. In the first set of experiments, MPEVneo(M813) pseudotype titers were determined on SC1 cells that either were uninfected or expressed virus of one of the four known interference groups, i.e., Mo-MuLV (ecotropic), Mo-AmphoV (amphotropic); 10A1V (10A1), and Mo-MCFV (polytropic). Xenotropic MuLV was not tested, as these viruses do not infect all M. musculus cells and use the same receptor as polytropic viruses (24). As a control for infectivity of the cell lines, MPEVneo(10A1) pseudotype infections were carried out in parallel. As shown in Table 3, expression of M813 in SC1 cells inhibits infection of M813 by more than 4 orders of magnitude, but no or only slight interference (at most fourfold) was observed in SC1 cells expressing any one of the other MuLVs.

Human TE671 cells expressing the mCAT1 protein are susceptible to Mo-MuLV infections but not M813 infections. To address the question of whether M813 can use mCAT1 as a receptor, a cDNA encoding mCAT1 was introduced into the human fibroblast line TE671 by the retroviral vector MPEVneo. After selection for G418 resistance, the resulting cell line, TE671neo-mCAT1, as well as the control cell line TE671neo, were assayed for expression of the mCAT1 gene by infection with Mo-MuLV. For this assay, we used an SFEV vector containing the EGFP cDNA, which was pseudotyped with either Mo-MuLV or, as a control, 10A1V. Forty-eight hours after infection, target cells were monitored for EGFP expression. As shown in Fig. 1, TE671 cells carrying the mCAT1 construct were susceptible to Mo-MuLV infections, in contrast to control TE671neo cells. The infection efficiency of Mo-MuLV was reduced by approximately 25% (average of two independent experiments) on TE671neo-CAT1 cells, compared to SC1 cells; however, a comparable decrease was also observed with 10A1V. We therefore conclude that the TE671neo-CAT1 cells express sufficient levels of CAT1 to allow infection by Mo-MuLV.

View larger version (38K): [in this window] [in a new window]   FIG. 1.   Susceptibility to Mo-MuLV infection conferred to human TE671 cells by expression of mCAT1 was confirmed by EGFP transfer. Target cells were incubated for 48 h with supernatant containing the SFEV-EGFP vector pseudotyped with either 10A1, Mo-MuLV, or M813 produced from SC1 fibroblasts. Infected cells were then monitored for expression of EGFP by flow cytometry.

Molecular cloning of the M813 env gene. To gain insight into the relationship of M813 MuLV with other MuLVs, as well as to define the molecular basis for its unique receptor usage, the env gene of M813 was molecularly cloned. Viral RNA was isolated from virus particles and subjected to cDNA synthesis. Primers designed to recognize conserved sequences in the pol and env genes of MuLVs were used to amplify the intervening sequences (see Materials and Methods). 3'-RACE was successfully employed to obtain 3' env sequences and LTR U3 and R regions. Compiled sequence data of the overlapping clones have been submitted to the public data banks.

View larger version (55K): [in this window] [in a new window]   FIG. 2.   Northern blot analysis confirms that the env gene is expressed within a full-length retrovirus expressed in the SC1-M813 virus producer. Total RNA from SC1 and SC1-M813 producer was isolated, separated by agarose gel electrophoresis, and transferred to a nylon membrane. The blot was hybridized under standard conditions with a probe extending from nt 519 to 1358, corresponding to the first 840 bp of the env coding region.

View larger version (69K): [in this window] [in a new window]   FIG. 3.   Southern blot analysis of SC1-M813 and SC1 cells confirms that M813 is not an endogenous M. musculus MuLV. DNA isolated from M813-infected and uninfected SC1 cells was digested with the indicated restriction enzymes and separated by agarose gel electrophoresis. After transfer to a nylon membrane, DNA was hybridized with an M813 env probe (nt 519 to 1358).

Replacement of the env gene of Mo-MuLV with that of M813 MuLV results in a replication-competent virus with an M813 interference profile. To functionally show that the isolated env sequences are derived from M813, the cloned sequences were replaced with those of Mo-MuLV to generate a replication-competent virus. Using the strategy described in Materials and Methods, a Mo-M813V recombinant was constructed in which the sequences encoding the first 225 amino-terminal residues of the mature gp70^env protein are derived from the M813 env clone. This includes the proposed receptor recognition domain (including variable regions A, C, and B), as well as the proline-rich domain of the SU. Plasmid DNA was transfected into SC1 cells containing the MPEVneo construct and then monitored for reverse transcriptase activity and marker rescue. After 10 days, the supernatant of SC1-Mo-M813 cells was positive for both reverse transcriptase activity (data not shown) and its ability to transfer G418 resistance to SC1 cells (Table 6), conclusively showing that the cloned env sequences encoded an Env protein that was able to bind to a cellular receptor and mediate cell penetration.

Sequence analysis of the M813 env gene confirms its close but distinct relationship to M. musculus MuLVs. The amino acid sequence of M813 Env was compared to that of several other MuLVs (see Materials and Methods for list of viruses used for comparison). The M813 Env protein shares an overall sequence identity of 51 to 58% with other MuLV Envs, the highest identity being with the Env protein of the C3H ecotropic virus. Sequence alignments with other mammalian C-type viruses showed relatively high identity with FeLV-A (50.4%) but less identity with the M. dunni multitropic MDEV-MuLV isolate, GALV, or the koala retrovirus (42.5, 39, and 42.4%, respectively), consistent with the hypothesis that FeLV-A originated from MuLV (8).

View larger version (88K): [in this window] [in a new window]   FIG. 4.   Multiple-sequence alignment (Clustal W) of the SU Env of several MuLVs. Two potential cleavage sites of the signal peptide in the M813 sequence are indicated by arrows (46). The variable regions and the proline-rich domain (PRD) are boxed and shaded. Identical amino acid residues are denoted with asterisks, whereas residues with highly similar or fairly similar physiochemical properties (e.g., polar, hydrophobic, or charged) are denoted with colons or dots, respectively.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Interest in the type C MuLVs stems from various disciplines. MuLVs have proven to be outstanding tools for studying oncogenesis, for establishing principles of viral persistence in host organisms, for tracking evolutionary relationships among mice and viruses, and finally as gene transfer vectors in both the laboratory and the clinical setting. MuLVs are generally classified into different host range or interference groups that reflect sequence homology in the env gene. These sequences encode the SU protein, which is incorporated into the virus envelope and mediates cell-specific binding to a transmembrane receptor. Receptor usage has several implications for both the pathogenicity of the virus (36, 38) and its usefulness as a vector for gene transfer (31, 32, 47). In addition, these unique env sequences have been used to facilitate phylogenetic analysis of mouse strains and viruses (17, 22, 43, 45). Although the best-studied MuLVs have been isolated from M. musculus, MuLVs have also been isolated from several other strains. In the present study, we have defined a novel receptor group of the MuLVs by characterizing the M813 MuLV isolate from M. cervicolor.

Due to its high homology with MuLVs from M. musculus and its limited host range, M813 was originally classified as a type CII MuLV with an ecotropic (i.e., M. musculus) host range (7, 29). In this study, we have presented data that support the unique receptor usage of M813. This was initially demonstrated by interference assaysM813 infection was not hampered in cells expressing any one of the tested MuLVs. Although there was slight interference of ecotropic Mo-MuLV infections in cells expressing M813, Mo-MuLV expression did not block M813 infection. Such a nonreciprocal interference pattern is suggestive of at least two receptors, one of which would be shared by both virus groups. However, expression of the ecotropic receptor mCAT1 cDNA in human cells did not impart susceptibility to M813 infection. Although we cannot completely rule out the possibility that this interference is due to expression and utilization of the mCAT3 receptor, we find this unlikely, as ecotropic MuLV infection using CAT3 is inefficient (30) and CAT3 expression is limited to the brain and thus probably not present in the SC1 fibroblasts used in this study. We find it more likely that the slight interference was directly due not to M813 expression but to other sequences present in the original M813-expressing cell line. In support of this, we found that expression of molecularly cloned M813 env sequences did not show any detectable interference with Mo-MuLV infections.

To characterize the sequences that imparted the unique receptor usage to M813, we molecularly cloned the entire env sequences. Due to relatively high homology with other known MuLVs, we were able to amplify the env sequences using oligonucleotide primers in conserved regions of the pol and env genes. Importantly, incorporation of the M813 env sequences into the Mo-MuLV genome imparted an altered host range that was restricted by M813 but not by the other tested MuLVs, verifying that these sequence were of M813 origin.

As would be expected for a virus that recognizes a different receptor, only 37% sequence identity at most was found with other MuLVs within the amino-terminal domain of the SU protein, known to be important for receptor binding (4, 13), whereas sequences encoding the carboxyl end of the SU and the entire TM were more highly conserved (76%). Sequence similarity is not always an accurate prediction of receptor usage, however, as demonstrated for 10A1 and GALV, which use the same receptor but show only 24% identity in the receptor-binding and proline-rich domains. Nevertheless, the distinctive structure of the VRA domain in M813 is striking evidence that it has a unique binding domain.

No sequence homology was detected between M813 5' env sequences and M. musculus sequences by hybridization. This not only verifies that our molecular clone is not derived from an endogenous M. musculus provirus but also has implications with regard to its evolution. The absence of M813 5' env sequences in the M. musculus genome is in striking contrast to the wide distribution of both polytropic and xenotropic MuLVs in different mouse species but is analogous to ecotropic MuLVs, whose distribution is limited (22). This suggests that both ecotropic MuLVs from M. musculus and M813 from M. cervicolor may have more recent origins. In agreement with the high divergence in the amino-terminal region, M813 may have arisen by replacement of a region in the env gene of an endogenous xenotropic MuLV by the analogous regions of an unknown virus (not fixed in the germ line) or with cellular sequences that bind to a novel receptor, as has been suggested for ecotropic MuLVs (42). Further studies, such as screening other mouse species for homologous M813 env sequences, as well as sequence comparison of other M813 structural gene with the different MuLVs are necessary for establishing definitive phylogenetic relationships.

Interestingly, the original characterization of type CII viruses (such as M813) showed infectivity for M. musculus- but not M. cervicolor-derived cells (7). This implies that the host must have developed a mechanism of resistance after the virus was introduced into the genome. As the provirus appears to be in an active state in the M. cervicolor genome (induction of virus release was not necessary to isolate the original CII variant), the resistance mechanism most likely involves blocking virus binding to the cell surface receptors. This could be accomplished by ectopic expression of env sequences, as has been described for the Fv4 resistance locus, or accumulation of point mutations in the cellular receptor that prevent virus binding without disruption of its normal function, as predicted for the xenotropic receptor in laboratory mouse strains and shown for the ecotropic receptor in M. dunni (12, 21, 24). In this light, it will be of interest to determine the number of M813 env-related sequences that are present in the M. cervicolor genome.

In conclusion, we have identified a unique MuLV interference group defined by the type CII M813 MuLV isolated from M. cervicolor tissue. We predict that this virus may have evolved through "domain swapping" of either cellular or viral sequences with more ancient endogenous MuLVs, yielding a retrovirus with altered receptor usage. Characterization of this novel receptor will help us better understand the mechanisms by which retroviruses gain entry into the cell.