INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Historically, the identification and characterization of host cell gene products that are essential for retroviral replication has been considerably facilitated by genetic complementation in cell lines derived from species that are unable to support the replication of a given retrovirus. The inability of CD4-expressing nonhuman cells to support human immunodeficiency virus (HIV) and/or simian immunodeficiency virus entry is a particularly noteworthy example (11, 30) and was crucial for the initial identification and, in part, subsequent characterization of chemokine receptors as entry cofactors (2, 3, 10, 14-18, 26, 37, 39). The rodent cell-specific defect in HIV-1 Tat-mediated transactivation provides a second example. In this case, the defect was alleviated by gene products encoded on the human chromosome 12 (1, 21, 35). It later became clear that the mechanistic explanation for the defect was the inability of the murine form of the essential Tat cofactor, CycT1, to support efficient interaction with the TAR element when bound to Tat (5, 19, 20, 27, 46). Therefore, the poor activity of a variety of primate lentiviral Tat proteins in murine cells can be rescued by expression of the human CycT1 protein, which is encoded on human chromosome 12 (46). Alternatively, substitution of a single amino acid in the murine CycT1 protein, tyrosine 261, with its human CycT1 counterpart, cysteine, results in a CycT1 protein that can be efficiently recruited to TAR by Tat proteins and, hence, support transactivation (4, 5, 19, 20, 27).

While these studies illustrate how the complementation of defects in nonpermissive rodent cells has aided the identification and characterization of molecules that are critical in the HIV-1 life cycle, this work has also raised the possibility that rodents could be engineered to support HIV infection (7, 40). This would provide an inexpensive in vivo model for the evaluation of therapeutic agents and could aid molecular studies of host-virus interactions at an "organism" level, since both virus and host could be genetically manipulated. However, while expression of CD4, a coreceptor, and a functional CycT1 protein would be predicted to render murine cells permissive for HIV replication, previous work has demonstrated that this is not the case. Murine NIH 3T3 fibroblasts expressing CD4, CCR5, and hCycT1 were found not to be permissive for HIV-1 replication, although these cells were able to support at least some level of gene expression from an integrated provirus (20, 32). Conversely, CHO cells transiently expressing these molecules have been reported to support a single cycle of replication following cocultivation with HIV-1-infected human cells (47), although in this case, it is likely that infected human cells would fuse with the hamster cells to form heterokaryons.

Previously, few examples of species-specific defects in retroviral assembly and egress have been described. Simple mammalian retroviruses, such as murine leukemia virus, can be readily assembled and released by a variety of nonmurine cells (13). Similarly, the foamy retroviruses possess an extremely broad tropism and can be cultivated in a wide range of mammalian avian and even reptilian cells (22). However, an insect cell-specific defect in simian D-type retroviral particle assembly has recently been reported, as has a defect in HIV-1 assembly in mouse NIH 3T3 cells (32, 36). In this study, we extend this work and describe defects in HIV-1 replication in rodent cell lines derived from multiple lineages and species. Cumulatively, rodent cell-specific properties can result in an up to 10⁵-fold-reduced competence to support a single cycle of HIV-1 replication. Interestingly, specific defects in infectious HIV-1 particle assembly and egress are not universal among rodents and, importantly, can be substantially rescued by fusion with human cells. Taken together, these observations suggest that one or more species-specific cellular cofactors exist that dramatically enhance infectious HIV-1 particle production.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmids. A single-amino-acid point mutant of the murine cyclin T1 protein (mCycT1) in which tyrosine 261 has been replaced by cysteine (mCycT1Y261C) has been previously described (5). A cDNA encoding this protein was inserted into the retroviral expression vectors LXSN (33) and MSCVNeo (Clontech) to generate LXSN/YC and MSCV/N/YC, respectively. LXSH/CD4, pBABE/CXCR4, and pBABE/CCR5 were generated by insertion of cDNAs encoding CD4, CXCR4 and CCR5 (obtained by digestion of pBC12/CMV-based plasmids) (3) into the retroviral expression vectors LXSH (33) and pBABE Puro (34), respectively.

Cell lines, viruses, and transfections. Retroviral stocks, generated by transient transfection of Phoenix Gagpol cells (25) with 10 µg of a retroviral expression vector and 1 µg of pHIT/G (42), were used to generate stable cell lines expressing mCycT1(Y261C), CD4, CXCR4, CCR5, or combinations thereof. The cell lines used in this study were derived from mice (NIH 3T3, Mus dunni tail fibroblasts [MDTF], BW5147), rats (Rat2, XC), hamsters (CHO, BHK-21), or humans (HeLa, HOS, CEMx174). Specifically, all rodent cells were engineered to express mCycT1(Y261C), and NIH 3T3, Rat2, and CHO cells were, in addition, transduced with CD4 and either CXCR4- or CCR5-expressing retroviruses. HOS-derived cells expressing CD4 and coreceptors were obtained from the NIH AIDS Reagent Program (GHOST cell lines). Expression of CD4, CXCR4, and CCR5 was verified by fluorescence-activated cell sorter (FACS) analysis using phycoerythrin (PE)-conjugated antibodies (Pharmingen), and where necessary, pure populations of receptor-expressing cells were obtained by sorting. Functional expression of mCycT1(Y261C) was verified by enhancement of proviral gene expression (see Results and Fig. 1).

Viral replication and entry assays. HIV-1 replication assays were performed with NL/IIIB for CD4⁺ CXCR4⁺ cells and NL/JRFL for CD4⁺ CCR5⁺ cells. Cells were inoculated with approximately 1,000 infectious units (IU) of virus (determined by using CD4⁺ CXCR4⁺ or CD4⁺ CCR5⁺ CHO cells and the focal immunoassay described below). The following day, cells were washed extensively, and p24 levels in the culture supernatant were monitored for the ensuing 10 days.

Comparative analysis of HIV-1 production in rodent and human cells during a single cycle of replication. For analysis of HIV-1 production during a single cycle of replication, cell lines lacking HIV-1 receptors were used. Human and rodent cell lines expressing mCycT1(Y261C) (approximately 2.5 × 10⁵ cells/35-mm-diameter dish) were infected with approximately 10⁶ infectious units of HIV-1IIIB pseudotyped with the VSV-G envelope protein. The following day, cells were washed extensively, and the growth medium was replaced. After a further 24 h, cells and supernatants were harvested, and viral RNA, proteins, and infectivity were analyzed by RNase protection assays, enzyme-linked immunosorbent assay (ELISA), Western blotting, and focal immunoassays as described below.

Viral RNA analysis. A PCR-generated fragment of pIIIB (nucleotides 78 to 340, relative to the start site of transcription) was inserted into the HindIII and SmaI sites of pBluescript KS+ to provide a template for the synthesis of an antisense RNA probe spanning the HIV-1 major 5' splice donor. This plasmid was linearized with HindIII, and the probe was generated by transcription with T7 polymerase in the presence of [³²P]CTP. Ten micrograms of total RNA, extracted from HIV-1IIIB(VSV-G)-infected cell lines was hybridized to the probe overnight and digested with an RNase A-RNase T1 mixture (RPAIII kit; Ambion). Protected fragments that corresponded to spliced and unspliced HIV-1 RNA were visualized by autoradiography and quantitated by PhosphorImager analysis after separation on a denaturing acrylamide gel.

Viral Gag protein analysis. Supernatants were harvested from infected cells and passed through a 0.45-µm-pore-size filter prior to infectivity or p24 assays. In some experiments, supernatants were layered above a 20% sucrose cushion and centrifuged at 35,000 rpm for 90 min in a Beckman SW50.1 rotor prior to quantitation of p24 in pelleted and nonpelleted fractions. Cell lysates were prepared by washing cells with phosphate-buffered saline (PBS) and resuspension in PBS-1% Triton X-100 (approximately 2 × 10⁶ cells/ml) followed by one cycle of freezing and thawing and centrifugation to remove insoluble debris. Viral antigen (p24) in cell lysates and supernatants was measured by ELISA (NEN/DuPont). Aliquots of cell lysates containing either 10 µg of total protein or 2.5 ng of p24 (as measured by ELISA) were separated on 4 to 15% acrylamide gradient gels and transferred to nitrocellulose. Western blots were probed sequentially with a monoclonal antibody to HIV-1 p24 (183 clone H12-C) (9) and an antimouse immunoglobulin G-peroxidase conjugate and were developed by using chemiluminescent detection reagents (Roche).

Infectivity assays. Infectious HIV-1 in supernatants was quantitated by using a focal immunoassay similar to those previously described (9, 26). Briefly, CHO cells expressing mCycT1(Y261C), CD4, and either CXCR4 or CCR5 were seeded in 24-well plates and inoculated with serial dilutions of HIV-1-containing supernatant. Forty-eight hours later, the cells were fixed, and foci of infection were enumerated microscopically after sequential incubations with the HIV-1 p24^Gag-specific 183 clone H12-C monoclonal antibody and an antimouse immunoglobulin G-fluorescein conjugate.

Viral production by heterokaryons. MDTF or Rat2 cells (2.5 × 10⁵ cells/35-mm-diameter dish) were infected with approximately 10⁶ IU of HIV-1IIIB pseudotyped with the VSV-G envelope protein. Twenty hours later, the cells were washed extensively, trypsinized, and plated with an equal number of MDTF, Rat2, or HOS cells. After 6 h, cells were washed once with serum-free medium and overlayed with a 50% solution of polyethylene glycol (PEG) for 2 min at 37°C. After washing, fresh growth medium was added for 16 to 20 h, and infectious virus production was measured with the focal immunoassay.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Mouse and rat cell lines that express functional CycT1, CD4, and coreceptor proteins do not support HIV-1 replication. Rodent cell lines that are capable of supporting Tat function were generated by transduction with retroviral vectors that express mCycT1(Y261C). To verify that expression of a functional CycT1 protein was sufficient to increase HIV-1 proviral gene expression to a level comparable to that observed in human cell lines, NIH 3T3, Rat2, and CHO cells expressing mCycT1(Y261C) were infected with VSV-G-pseudotyped NL4-3Eluc. For comparison, the human cell lines HeLa, HOS, Jurkat, and CEMx174 were identically infected. As can be seen in Fig. 1A, this analysis revealed that introduction of the mCycT1(Y261C) expression vector into rodent cells was sufficient to increase gene expression from an integrated provirus to a level that fell within the range observed in human cell lines. Surprisingly, NL4-3Eluc-infected, but otherwise unmodified, Rat-2 cells expressed high levels of luciferase, comparable to that seen in human cells. Luciferase expression was only modestly (approximately threefold) enhanced by mCycT1(Y261C). We speculate that Rat2 cells express a form of CycT1 that is able to support at least some level of Tat function. In contrast, mCycT1(Y261C) expression in NIH 3T3 and CHO resulted in 21- and 76-fold increases in proviral gene expression, respectively (Fig. 1A). All subsequent experiments were done with rodent cells that expressed mCycT1(Y261C).

View larger version (41K): [in this window] [in a new window]   FIG. 1.   Rodent cell lines that express human gene products required for HIV-1 replication. (A) NIH 3T3, Rat2, and CHO cells either unmodified (open bars) or transduced with retroviral vectors that express mCycT1(Y261C) (solid bars). These cell lines, as well as various human cell lines (hatched bars) were infected with NL4-3Eluc, pseudotyped with the VSV-G envelope glycoprotein. Forty-eight hours after infection, luciferase activity in cell lysates was determined. RLU, relative light units. (B) The cell lines in panel A were sequentially transduced first with an LXSH-derived retrovirus vector that expresses CD4 and subsequently with pBABE Puro-derived vectors that express either CXCR4 or CCR5. The expression level of each of these proteins was determined by using PE-conjugated antibodies. CD4+ CXCR4+ cell lines are displayed with dark grey lines, and CD4+ CCR5+ cell lines are displayed with pale grey lines. The parental, mCycT1(Y261C)-expressing cell lines were used as negative controls (black lines). For comparison, HOS cells and the CD4- and coreceptor-positive derivatives (GHOST cell lines) were simultaneously analyzed.

View larger version (29K): [in this window] [in a new window]   FIG. 2.   HIV-1 replication in mCycT1(Y261C)- and CD4-, and coreceptor-expressing rodent cell lines. (A) The CD4+ CXCR4+ NIH 3T3, Rat2, CHO, and HOS cell lines described in the legend to Fig. 1 were infected with NL/IIIB. Alternatively (B) CD4+ CCR5+ counterparts were infected with NL/JRFL. Viral replication was observed for the subsequent 10 days by monitoring the p24 concentration in the culture supernatants by ELISA.

Host cell-dependent variation in HIV-1 entry mediated by CD4 and coreceptors. It has been shown that expression of CD4 and an appropriate coreceptor on nonhuman cells is sufficient to permit HIV-1 envelope-induced fusion and entry (10, 14-17). Nevertheless, the low level of Tat function in rodent cells has precluded any quantitative comparison of the relative efficiency of the early stages of HIV-1 infection in this context. To examine whether the mouse, rat, and hamster cells could indeed support efficient HIV-1 entry, the cell lines described in Fig. 1 were infected with HIV-1 reporter viruses that express GFP in place of Nef. Virus entry mediated by the HIV-1IIIB (X4-tropic), YU2 (R5-tropic), and VSV-G envelopes was compared by counting the number of infected cells 36 to 40 h later by FACS. An example of these experiments is presented in Fig. 3A; the accumulated data for all of the cell lines examined are shown in Fig. 3B.

View larger version (44K): [in this window] [in a new window]   FIG. 3.   A target cell-specific, CD4- and coreceptor-dependent HIV-1 entry defect. (A) FACS analysis of GFP reporter virus infection of CD4+ CXCR4+ CHO and CD4+ CXCR4+ Rat2 cells, mediated by the VSV-G envelope (upper panels) or by the X4 tropic HIV-1IIIB envelope (lower panels). (B) Analyses identical to that shown in panel A were done with all of the CD4- and coreceptor-positive cell lines, using HIV-1IIIB enveloped GFP reporter virus for CD4+ CXCR4+ cells and a YU2 enveloped virus for CD4+ CCR5+ cell lines. The proportion of GFP-positive cells was determined by setting a gate at a fluorescent intensity of between 10 and 30, depending on the cell line, so that less than 5 positive events were observed in 50,000 mock-infected cells (0.01%).

Analysis of a single cycle of HIV-1 replication in rodent cells. Differences in entry efficiency could partly explain why HIV-1 failed to establish a spreading infection in mouse and rat cells. However, it remained likely that other blocks in the HIV-1 life cycle contributed to rendering viral replication undetectable in NIH 3T3 and Rat2 cells. In addition, HIV-1 enveloped viruses could enter CHO cells at least as efficiently as HOS cells, yet spreading infection was significantly attenuated in the former. To examine which, if any, of the subsequent steps of the HIV-1 life cycle are blocked in rodent cells, a single cycle of HIV-1 replication was examined in detail. In order to determine whether any potential differences were either species specific or cell line specific, additional cell lines from each species were examined. Fibroblasts from wild mice (Mus dunni) as well as a mouse T-cell line, BW5147, were included, as were additional rat (XC), hamster (BHK-21), and human (HeLa and CEMx174) cell lines. All of the nonhuman cell lines were transduced with recombinant LXSN- or MSCV-derived retroviral vectors expressing mCycT1(Y261C), and, after inoculation with VSV-G-pseudotyped reporter viruses, expressed luciferase and GFP at levels comparable to those of human cells (data not shown).

View larger version (35K): [in this window] [in a new window]   FIG. 4.   Posttranscriptional defects in infectious virus production from rodent cells during a single cycle of HIV-1 replication. (A) Rodent and human cells are similarly susceptible to HIV-1(VSV-G) infection. The indicated cell lines were infected with R7/E/GFP pseudotyped with VSV-G. The number of infected cells per well was determined by multiplying the percentage of GFP-positive cells by the total number of cells in each well. (B) Cell lines were infected with replication-competent, X4-tropic but VSV-G-pseudotyped HIV-1IIIB. After extensive washing, infectious virus production was analyzed by using CD4+ CXCR4+ CHO cells and a focal immunoassay, as described in Materials and Methods. FFU, focus-forming units.

Reduced levels of genomic HIV-1 RNA in infected rodent cells. Since events up to and including early gene expression proceeded efficiently when diverse cell lines were infected with HIV-1(VSV-G) (Fig. 4A), we undertook a systematic examination of subsequent steps in the HIV-1 life cycle to determine why rodent cells do not efficiently produce infectious HIV-1 particles. We first analyzed the level at which HIV-1 mRNA is present in spliced and unspliced forms, using ribonuclease protection assays and a probe spanning the 5' major splice donor. This analysis, shown in Fig. 5, revealed that spliced HIV-1 RNA was present at similar levels in all of the cells analyzed, as expected (less than 3.5-fold variation). In contrast, the full-length, unspliced HIV-1 transcript is present at lower levels (6- to 40-fold) in each of the nonhuman lines, as compared to the human cell lines. Nevertheless, the levels of the unspliced transcript (Fig. 5) clearly did not correlate with the level of infectious virus production by a given cell line (Fig. 4B). For example, the total amount of HIV-1 RNA and the degree to which it is spliced appear almost identical in MDTF, Rat2, and CHO cells (Fig. 5), but the latter produce an approximately 1,000-fold higher level of infectious virus (Fig. 4B). Thus, while the reduced abundance of the unspliced HIV-1 transcript likely contributes to the observed reduction in infectious HIV-1 production by rodent cells, it appears to play only a minor role in determining this phenotype.

View larger version (114K): [in this window] [in a new window]   FIG. 5.   Analysis of HIV-1 RNA in infected cells. The indicated HIV receptor-negative cell lines were infected with VSV-G-pseudotyped HIV-1 as described in the legend to Fig. 4B. Forty-eight hours later, total RNA was extracted from infected cells and analyzed by ribonuclease protection assay. Arrows indicate the predicted migration of the 317-nucleotide undigested probe (P) which spans the major 5' splice donor site, resulting in two protected fragments of 262 and 213 nucleotides that correspond to unspliced (U) or spliced (S) HIV-1 RNA, respectively. The numbers below the lanes indicate the levels of each of the RNA species, determined by PhosphorImager analysis and expressed in arbitrary units. M, molecular size markers.

Reduced levels of Gag synthesis, processing, and assembly into infectious viral particles in rodent cells. The reduced levels of unspliced HIV-1 RNA might be predicted to result in a commensurate reduction in the levels of Gag and Gag-Pol polyprotein expression, since both are translated from unspliced HIV-1 mRNA. Indeed, as shown in Fig. 6, both Western blotting (Fig. 6A) and ELISA (Fig. 6C) analysis of infected cell lysates revealed a significant reduction in the level of Gag proteins that could be detected by using antibodies to the viral capsid protein (p24). When equivalent quantities of cell lysates were examined by Western blotting, however, qualitative as well as quantitative differences in Gag expression among this panel of cell lines were noted (Fig. 6A). While readily detectable but modest differences in the level of the p55^Gag and p160^Gag-Pol precursor could be observed, much larger differences in the level of the processed p24^CA protein were evident. Clearly, therefore, the level of intracellular processing of the Gag precursor varied both between and within species. Specifically, the p55^Gag precursor was inefficiently processed in each of the mouse cell lines. Similarly, Rat2 and hamster BHK-21 cells contained relatively low levels of the processed Gag proteins compared to p55^Gag. In contrast, p55^Gag appeared to be quite efficiently processed in the rat cell line, XC, and the hamster cell line, CHO.

View larger version (29K): [in this window] [in a new window]   FIG. 6.   Reduced levels of Gag expression, processing, and assembly into infectious particles in rodent cell lines. The same cells that generated the infectious virus measured in Fig. 4B as well as the corresponding supernatant samples were analyzed for viral Gag protein content. (A) Equivalent quantities (10 µg) of cell lysate were separated on acrylamide gels and subjected to Western blot analysis with a p24-specific monoclonal antibody. Note that the upper panel, which displays the p160Gag-Pol precursor, represents a longer exposure than the lower panel. (B) Aliquots of cell lysates containing equivalent amounts of p24 (2.5 ng) as measured by p24 ELISA were analyzed by Western blotting as in panel A. (C) Viral p24 in cell lysates and supernatants as measured with a commercial ELISA kit. (D) The quantity of infectious virus produced relative to the total amount of supernatant p24 was calculated for each cell line. (E) Relative proportions of the total supernatant p24 that could be pelleted by ultracentrifugation through a sucrose cushion. Numbers refer to the percentage of the starting material that was recovered in each fraction and their sum.

Fusion with uninfected human cells enhances the production of infectious virions from infected rodent cells. The reduced ability of rodent cells to produce infectious HIV-1 particles could be due to a lack of essential cellular cofactors that are required for virus assembly and egress. Alternatively, rodent cells could express a dominant inhibitor of these processes. To address this question, HIV-1-infected rodent cells were generated by using HIV-1IIIB(VSV-G) and subsequently fused, using PEG, with uninfected human HOS cells. As controls, the infected rodent cells were also fused with uninfected autologous cells, or were cocultivated with human cells, but not fused. As can be seen in Fig. 7, fusion of infected MDTF or Rat2 cells with uninfected human HOS cells resulted in a dramatic increase (50- to 200-fold) in the level of infectious virus production compared to that of unfused cells or cells that were fused with MDTF or Rat2 controls. While fusion with human cells did not restore levels of virus production to those observed if the HOS cells had been infected directly, we estimated visually that approximately 10% of the cells in the mixed cultures actually fused after PEG treatment. This being true, the observed 50- to 200-fold increase in viral production is quite close (within 1 order of magnitude) to the result that would be predicted if the human cell phenotype was completely restored in rodent-human cell heterokaryons. Furthermore, and as can be seen in Table 1, the predominant effect of fusion with human cells was to increase the apparent infectivity of the supernatant p24^CA (approximately 35-fold) rather than to increase the total amount of p24^CA secreted (approximately 2- to 3-fold). Minor or no effects on either total p24^CA production or virus infectivity were observed when only rodent cells were fused. These data suggest that the small number of rodent or human heterokaryons efficiently produced highly infectious virus, and the simplest explanation for these results is that human cells express a factor or factors required for infectious HIV-1 production that are either lacking or nonfunctional in rodent cells.

View larger version (15K): [in this window] [in a new window]   FIG. 7.   Fusion of HIV-1-infected rodent cells with uninfected human cells enhances infectious particle release. MDTF (A) or Rat2 (B) cells were infected with VSV-G-pseudotyped HIV-1 as described in the legend to Fig. 4. The following day, the cells were washed, trypsinized, and replated, either alone (None) or along with an equal number of MDTF, Rat2, or HOS cells. Cocultivated cells were fused with PEG, except where indicated, and subsequent infectious virus production was analyzed after 16 h.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In this study, we have characterized the ability of a number of rodent cell lines to support HIV-1 replication. Clearly, the expression of functional CycT1, CD4, and coreceptor molecules in most rodent cells was insufficient to constitute a permissive environment for HIV-1 replication. In fact, multiple additional blocks to HIV replication were observed among rodent cells. In most cases, these blocks were sufficiently severe to render HIV-1 replication undetectable. Nevertheless, CHO cells were found to be exceptional among the rodent cell lines examined in that at least some degree of viral replication could be observed. Upon more detailed analysis, the differences observed between NIH 3T3, Rat2, CHO, and HOS cells during single-cycle virus replication assays (Fig. 4) correlated well with the viral growth phenotypes shown in Fig. 2. Cumulatively, viral entry and production defects in NIH 3T3 and Rat2 cells would be predicted to result in a roughly 10⁵-fold (per cycle) replication defect as compared to human cells. Given these findings, it is not surprising that little or no viral replication is detected in these cell lines (Fig. 2). Conversely, a single cycle of replication was completed approximately 10% as efficiently in CHO cells as in human cell lines (Fig. 4), which correlates well with the partial replication defect observed during spreading infection assays in these cells (Fig. 2).

Blocks to HIV-1 entry in rodent cells. Unexpectedly, expression of CD4 and of a coreceptor was found to be insufficient to render some rodent cells highly permissive for HIV entry (Fig. 3). This phenotype was especially dramatic in CD4⁺ CXCR4⁺ Rat2 cells, which supported an approximately 400-fold-lower level of X4 tropic viral entry than did human HOS cells, even though they expressed at least as much human CD4 and CXCR4 on their surface. This entry defect was less pronounced, but still significant, in NIH 3T3 cells and in both cell types when CCR5 was the coreceptor. In fact, the modest decrease in YU2 enveloped HIV-1 entry into CD4⁺ CCR5⁺ NIH 3T3 compared to GHOST cells is in good agreement with a previous result obtained with the JRFL envelope (32). At present, the molecular basis for this entry defect in mouse and rat fibroblasts is unclear. It is likely that the block is at the level of envelope-receptor binding or membrane fusion, since no such defects were observed when using VSV-G-pseudotyped virus stocks. In addition, CD4⁺ CXCR4⁺ Rat2 cells (and to a lesser extent NIH 3T3 cells) are substantially resistant to formation of syncytia when cocultivated with HIV-1IIIB-expressing cells (P.D.B., unpublished data). Interestingly, previous evidence has suggested that glycosylation might influence the activity of coreceptors. Most recently, the introduction of mutations at potential glycosylation sites in the CXCR4 molecule was shown to result in a coreceptor that could be used by HIV-1 strains that are ordinarily CCR5 tropic (8). Other reports have indicated that deglycosylation of cell surfaces by pretreatment with inhibitors resulted in higher levels of CD4-independent entry by a particular HIV-2 strain (38, 44). It is quite possible that species or cell-type-dependent differences in coreceptor glycosylation and/or multimerization (28) could determine the apparent differences in activity observed when CXCR4 and CCR5 are expressed on different cell lines. Alternatively, it is conceivable that additional species or cell-type-specific molecules play a role (either positive or negative) in mediating HIV-1 entry.

Reduced abundance of unspliced HIV-1 RNA and Gag proteins in rodent cells. Following viral entry, the HIV-1 life cycle appeared to proceed efficiently, independently of the cell species or type, up to and including early gene expression. Thereafter, profound defects in the ability of many rodent cell lines to produce infectious HIV-1 were observed. While the levels of spliced HIV-1-specific transcripts were approximately equal in all cell lines examined, the abundance of the unspliced, genomic transcript was significantly reduced in rodent cells (Fig. 5). This phenotype was more profound than that recently reported (32) and is similar to the predicted consequence of a defect in Rev function (45). However, it has been clearly shown that Rev is functional in rodent and even avian cells (31). Thus, the reduced abundance of genomic HIV-1 RNA in rodent cell lines is likely due either to oversplicing or to a reduced stability of the unspliced transcript in rodent compared to human cells. In fact, HIV-1 transcripts have been previously shown to splice significantly more efficiently in mouse than in human cells (31).

Defects in infectious HIV-1 production by rodent cells. Quantitative comparison can be made, however, of the intracellular and extracellular levels of fully processed p24^CA. In fact, all rodent cells displayed a significant reduction in the proportion of the total p24^CA that was present in the culture supernatant. With the exception of the hamster cell line CHO, all rodent cells secreted uniform and low levels of p24 (<7 ng/ml) into culture supernatants, values that were up to 500-fold lower than those observed with human cells. This defect in p24 secretion was not universally linked to defects in intracellular Gag processing, since p55^Gag processing in the rat cell line XC was at least as efficient as that in the human cell line HOS, but XC cells nevertheless displayed a profound defect in mediating p24^CA egress. In marked contrast to the other rodent cell lines used in single-cycle replication assays, CHO cell cultures secreted levels of p24^CA that were only 10-fold lower than that which was typical of human cell lines. The reduced quantity of p24^CA in CHO cell supernatants compared to that in human cells was partly due to a small reduction in overall Gag protein expression, possibly due to the reduced abundance of unspliced HIV-1 RNA. However, the generalized rodent cell defect in p24^CA egress was at least partly conserved in CHO cells, because p24 release was reduced approximately sevenfold compared to that in human cells.

Defective infectious virus production in rodent cells is rescued by factors present in human cells. Importantly, the block in infectious virus production in mouse and rat cells could be substantially relieved by fusion with uninfected HOS cells (Fig. 7). Multiple and complex hypotheses can be devised to account for this result. It is possible, for example, that human cells could express a factor that blocks the function or expression of a rodent cell-specific inhibitor of infectious HIV-1 production. However, the simplest, and most likely scenario, is that human cells provide one or more cofactors that enhance the assembly and/or release of infectious HIV-1 virions. This is the first evidence of which we are aware of a species-restricted positive cofactor(s) involved in the egress of infectious retrovirus particles. At present, we can only speculate as to its identity and mechanism of action. It has been reported that infected HIV-1 NIH 3T3 cells contain electron-dense vesicle-associated material (32), which may be aggregates of the p55^Gag precursor, and very few budding virions. Conceivably, therefore, cellular factors could be required for specific targeting of viral proteins to the plasma membrane prior to assembly and budding. Alternatively, mistargeting of the Gag precursor could be entirely secondary to a particle assembly defect. Clearly, the phenotypes described herein are not readily explained by a lack of Vif target proteins in HIV-1-producing rodent cells. The functional targets of Vif proteins are likely to be dominant-negative inhibitors of viral infectivity (29, 41) and do not affect virion egress. Similarly, the potential lack of a functional cyclophilin in rodent cells is not an adequate explanation for this phenotype. A genetically or pharmacologically induced lack of cyclophilin in human cell-derived HIV-1 particles results in a phenotype that is only manifested in subsequent target cells and is dramatically different from that described here (6, 43).