|Home | About | Journals | Submit | Contact Us | Français|
Human immunodeficiency virus type 1 (HIV) replicates only in certain primate cells. In murine cells expressing cyclin T1, a posttranscriptional block exists such that small amounts of capsid and little infectious virus are released. This block is relieved in part by fusion with human cells. Here we have tested a panel of mouse-human somatic cell hybrids for production of infectious virus. Only those containing human chromosome 2 were permissive, which correlated with capsid production. The effect was specific to HIV in that release of murine leukemia virus was minimally affected by the presence of chromosome 2. Although expression of Vpu markedly increased capsid production in the absence of chromosome 2, it did not result in a corresponding increase in infectious HIV. The presence of chromosome 2 did not have consistent effects on the amount of unspliced viral RNA, whereas the amount of cell-associated Gag p55 was increased a fewfold. These results suggest that processing of HIV Gag can be corrected by one or more genes present on human chromosome 2 to allow production of infectious HIV from murine cells.
Human immunodeficiency virus type 1 (HIV) vaccine development and pathophysiology study have been hampered by the lack of a small-animal model permissive for HIV replication. Even after the viral entry determinants (i.e., CD4 and the chemokine coreceptors) were identified, it was clear there were additional blocks to viral replication in the mouse. Expression of human cyclin T1 (cycT1) improved transcriptional elongation from the viral long terminal repeat (LTR) (14, 39) and resulted in greater amounts of viral gene products in murine cells. However, infected mouse cells that express human CD4, CCR5, and cyclin T1 release only trace amounts of capsid protein (CA) p24 and little infectious virus (2, 14, 23).
Mouse-human cell fusions have been shown to be more permissive to viral replication, suggesting that mouse cells lack one or more critical factors that permit completion of the viral life cycle (2). Because mouse cells express both unspliced and spliced viral RNAs (23), the block appears to be posttranscriptional in nature. Additionally, murine cells accumulate amorphous collections of Gag as observed by electron microscopy, suggesting that the observed replicative block is at an early, not late, assembly stage (23).
Tsg101 and other protein members of the cellular budding machinery have been shown to be directly involved in HIV assembly and budding. Tsg101 was shown to interact biochemically with the late domain of Gag (37), and inhibition of Tsg101 function resulted in extracellular tethered particles of decreased infectivity (15, 24). Reduced expression or interference with function of other proteins involved in the multivesicular body and vacuolar protein sorting pathway has resulted in a similar cellular phenotype (38). Those structures are not observed in mouse cells infected with HIV. Instead, electron-dense cytoplasmic vesicular structures collect, and extracellular or budding particles are rare (23). These observations suggest that although a late block to viral budding in mouse cells cannot be excluded, the defect appears to be earlier.
A chimeric murine leukemia virus (MLV)/HIV matrix protein (MA) p17 gene partially overcame the block to virus production in mouse cells, although viral replication was poor compared to the replication of wild-type HIV in human cells (6, 29). This suggests that in mouse cells HIV MA function may be either inefficient or inhibitory. Recently, Perez-Caballero and colleagues have demonstrated that the globular head of MA inhibits a plasma membrane targeting signal located at the amino terminus of MA (27). This inhibition was relieved by Gag multimerization, which was concentration dependent, and imparted a high degree of cooperativity to Gag-membrane association (27), suggesting that slight differences in Gag expression may have profound consequences for CA production and infectious virus release. Deletions in portions of MA increased Gag expression in both human and murine cells and also production of infectious virus (17).
In the absence of Vif, in human cells members of the APOBEC3 protein family deaminate first-strand cDNA synthesis, resulting in hypermutation (typically G-to-A changes) and catastrophe such that the level of infectious virus released is reduced by several orders of magnitude (3, 16, 20, 41). Vif promotes the degradation of APOBEC3G via interaction with cullin 5 and the proteosomal pathway (40). Murine APOBEC3 activity, however, is not neutralized by Vif (3, 22). The fact that mouse-human cell fusions partially restored infectivity might suggest that a human factor counteracts mouse APOBEC3, which has not been demonstrated. In addition, inhibition of mouse APOBEC3 would not explain the absence of CA production/release in mouse cells, since APOBEC proteins do not interfere with Gag processing.
Viral protein U (Vpu) is a phosphoprotein expressed from an intron-containing RNA late in the viral life cycle (18). Vpu interacts with CD4, targeting it for degradation via interactions with h-βTrCP (21). Vpu is also capable of enhancing particle assembly/release through an undefined mechanism, which is cell type dependent. Recently, human cells have been demonstrated to express a restrictive factor that modestly blocks particle release, which was relieved by expression of Vpu (36). It is not presently known whether mouse cells express this factor; in the absence of Vpu, however, mouse-human cell fusions allowed CA and infectious virus release (23).
Proper assembly of HIV requires nuclear export and cytoplasmic translation of fully spliced, partially spliced, and unspliced viral mRNAs. In human cells, intron-containing HIV RNAs are exported from the nucleus using the Rev/Rev response element (RRE) pathway in which Crm1 is active (8). Other retroviruses, such as Mason-Pfizer monkey virus (MPMV), rely on a separate pathway for export of unspliced mRNAs (5). In this pathway, the constitutive transport element (CTE) of MPMV binds to NFX1 (also known as Tap), and RNA export occurs via normal transport mechanisms. Swanson and colleagues recently showed that when the RRE was replaced with a multimerized CTE from MPMV, Gag trafficking to cellular membranes and CA production were both restored in mouse cells (35). This suggests that Rev/RRE-containing transcripts may be incorrectly trafficked in the cytoplasm in mouse cells, perhaps because one or more factors that interact with those “marked” RNAs are functionally deficient in some manner.
Here we have used a panel of mouse-human somatic cell hybrids (SCHs) to demonstrate that the presence of chromosome 2 allowed efficient CA and infectious virus production from murine cells, which was specific to HIV. This effect did not appear to be due to alterations in cyclin T1 function, and the increase was out of proportion to the amounts of proviral mRNA and Gag protein precursor. Although Vpu increased CA production in the absence of chromosome 2, it did not correspondingly augment infectious virus release. These results suggest that HIV Gag processing can be corrected by one or more genes present on human chromosome 2 to allow both CA and infectious virus production.
HIV-cycT1-IRES-bsd (pHIV-CIB) has been described elsewhere (7), as has pHIV-IRES-eYFP (31). pHIV-cycT1-IRES-eYFP was constructed by inserting a hemagglutinin epitope-tagged, truncated version of human cyclin T1 (residues 1 to 303, kind gift of Kathy Jones of the Salk Institute) into the unique SmaI site just upstream of the internal ribosome entry site (IRES) element of pHIV-IRES-eYFP. M-tropic HIV envelope ADA expression construct was a gift from Dan Littman, and plasmid containing full-length infectious R5 isolate JR-CSF was obtained from the NIH AIDS Research and Reference Reagent Program. pBABE-IY was derived from pBABE-MN-IRES-Blasti (7) by exchanging enhanced yellow fluorescent protein (eYFP) for bsd. pHIT60 encodes MLV Gag-Pol, driven by the cytomegalovirus (CMV) immediate-early enhancer promoter (32), and pME VSV G encodes vesicular stomatitis virus (VSV) glycoprotein, driven by the SRα promoter. CMV-driven codon-optimized gag-pol expression plasmid was obtained from Oxford Biomedica (19). The helper-dependent adenovirus encoding VSV G (HDAd-VSV G) was constructed by cleaving HDΔ28E4lacZ (26) with NheI and inserting a 2.7-kb cassette of VSV G driven by the SRα promoter with compatible cohesive ends. This plasmid has the cis-acting inverted terminal repeats and packaging sequence of adenovirus, along with a CMV-lacZ reporter, but lacks all other adenoviral genes and sequences (26).
All cells in this study were maintained in Dulbecco's modified Eagle's high-glucose medium supplemented with 10% fetal calf serum, penicillin, streptomycin, and 20 μg/ml of ciprofloxacin (complete DMEM), with other antibiotics as indicated in the text. Cells were grown in 5% CO2, 37°C water-jacketed incubators and passaged every 3 to 5 days using trypsin-EDTA. Somatic cell hybrids were obtained from the Coriell Cell Repository unless otherwise stated (exceptions given in Acknowledgments). The cells were typically maintained under antibiotic selection as indicated in Table Table1.1. 293T and HOS TK− cells were originally obtained from the American Type Culture Collection, as were A9, 3T6, and LMTK− cells. GHOST HI5 cells, which have a stably integrated HIV LTR driving enhanced green fluorescent protein (eGFP) and also express human CD4 and CCR5 so that they are highly susceptible to R5 viral isolates, were obtained from the AIDS Research and Reference Reagent Program.
Pseudotyped retroviral and lentiviral particles were produced by calcium phosphate transfection of 293T cells with the appropriate viral envelope (here typically VSV G) and vector (33). For MLV, pHIT60 was included in the transfection mix. After 72 h, vector supernatant was harvested as previously described (33). Target cells were transduced in 18- or 35-mm dish format and 48 h later passaged into selective medium (typically complete DMEM supplemented with 10 μg/ml of blasticidin S [Invitrogen]).
HDAd-VSV G was produced by initially transiently transfecting 293 cells stably expressing Cre with the HDAd-VSV G plasmid and also infecting them with a first-generation, E3-deleted helper adenovirus in which the packaging sequence is flanked by LoxP sites (AdNG163R-2) (26). Supernatant from those cells was used to coinfect the 293-Cre cells along with AdNG163R-2) to cyclically amplify the amount of HDAd-VSV G as described previously (26). Resultant HDAd-VSV G was purified on two CsCl step gradients, and the amount of contaminating helper virus was estimated by Southern blotting to be ~0.02%. The titer, based upon spectrophotometric analysis, was >1012 PFU/ml.
In order to recover and determine the titer of integrated HIV vector encoding bsd from stable cell lines, typically cells were plated 1 day prior in 35-mm or 10-cm dish format, transfected the next day at 80 to 90% confluence with 1 to 2 μg each of a viral glycoprotein envelope plasmid (e.g., pME VSV G) along with an autofluorescent control (e.g., HIV-IRES-eYFP) using Lipofectamine 2000 reagent (Invitrogen), and 48 h later, culture supernatant was harvested, centrifuged at 2,000 × g for 10 min, and inoculated onto adherent human targets, typically HOS TK− or 293T cells. Transfected cells were immediately examined by epifluorescence microscopy or subjected to flow cytometry using a FACScan (Becton-Dickinson) to determine overall transfection efficiency. Alternatively, cells were infected with HDAd-VSV G at a high multiplicity of infection (MOI) and washed extensively the following day, and after 48 h, supernatant was recovered and added to target cells. Adenoviral transduction efficiency was estimated by staining the producers with 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal). After 48 h, target human cells were passaged into complete DMEM containing 10 μg/ml of blasticidin S. After 7 to 9 days, cells were fixed in methanol and acetic acid and stained with crystal violet, and colonies were enumerated.
The amount of CA produced by vector-transduced or infected cells was quantitated by using a commercial enzyme-linked immunosorbent assay (ELISA) kit (Beckman-Coulter) or by immunoblotting. For the latter, culture supernatants or radioimmunoprecipitation assay cell lysates were size separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and electrophoretically transferred to nitrocellulose. The membrane was preblocked using Tris-buffered saline supplemented with 0.02% Tween 20 and 2.5% nonfat dried milk, probed using primary human anti-HIV antibody from the AIDS Research and Reference Reagent Program (catalog no. 3957) at 1:1,000 in the same buffer, washed extensively in Tris-buffered saline containing 0.02% Tween 20, probed using rabbit anti-human immunoglobulin G antibody conjugated to horseradish peroxidase (Sigma) at 1:10,000, washed, and developed using enhanced chemiluminescence.
Genomic DNA was prepared from cell lines using DNAzol (Invitrogen) and resuspended in 8 mM NaOH. The DNA PCR primers used to amplify both human and mouse β-actin were 5′-TCATTGCTCCTCCTGAGCG-3′ and 5′-CTGCGCAAGTTAGGTTTTGTCA-3′ using 100 ng genomic DNA as the template, PCR conditions 94°C for 30 s, 52°C for 30 s, and 72°C for 30 s for 25 cycles, and Taq polymerase to give rise to a ~200-bp product, which was analyzed by 1.5% agarose gel electrophoresis and ethidium bromide (EtBr) staining. The HIV DNA primers for PCR amplification were NL4-3 831L (34) and NL4-3 569U (5′-TGTTGTGTGACTCTGGTAACTAGAGATC-3′), using conditions 94°C for 30 s, 62°C for 30 s, and 68°C for 30 s for 31 cycles and Expand Long DNA polymerase (Roche) to give rise to a ~260-bp product, which was similarly analyzed. The positive-control cell line had a single HIV vector integrant. Gels were Southern blotted at high stringency using the corresponding PCR product as a 32P-labeled probe amplified from the HIV vector plasmid template.
Total RNA was prepared from cell lines using the RNeasy Protect kit from QIAGEN following the manufacturer's instructions and treated with RNase-free DNase. The first-strand reaction was performed using Superscript II reverse transcriptase (RT) (Invitrogen), 1 to 3 μg of total RNA, and oligo(dT) as the primer for 60 min at 42°C. DNA PCR for β-actin was performed as described above for 21 cycles using three different amounts of the first-strand reaction mixture. For HIV, primers spanning the tat-rev intron were 5′-CTGTGGCATTGAGCAAGCTAACAGCAC-3′ and 5′-ACAGCGACGAAGAGCTCATCAGAACAG-3′. Amplification conditions were 94°C for 30 s, 62°C for 30 s, and 68°C for 60 s for 35 cycles using Expand Long DNA polymerase and the equivalent of 15, 45, and 145 ng of total RNA from the first-strand reaction mixture. This resulted in DNA products of 1,250 and 350 bp, corresponding to unspliced and spliced viral mRNAs, respectively. The spliced 350-bp product was directly visualized by agarose gel electrophoresis and EtBr staining. For the unspliced product, Southern blotting was performed using a 0.7-kb HindIII fragment specific to the intron as a probe. For a control for contaminating DNA, Superscript II RT was omitted from the first-strand reaction mixture. Total RNA from 293T cells that had been transduced with an HIV vector carrying both luciferase and bsd genes was used as a positive control.
In order to examine why mouse cells do not produce infectious HIV, primary mouse embryo fibroblasts null for p53 (9) were stably transduced with a murine leukemia virus vector encoding a truncated form of human cyclin T1 (7). This version lacks carboxy-terminal PEST sequences and is more stable than full-length cycT1. These cells were infected with a replication-competent HIV that had its gp160 replaced with vesicular stomatitis virus G coupled to the fluorophore eYFP by an internal ribosome entry site (31). No viral spread occurred in these mouse embryo fibroblasts, and virus could not be recovered from culture supernatant, despite the fact that infected cells were easily visible by fluorescence microscopy (not shown). CA levels in the supernatant were quantitated by ELISAs and were approximately 0.5 to 1.0% of those of infected human cells. Western blotting of both supernatant and cell lysates using polyclonal anti-HIV antisera demonstrated only small amounts of CA but large amounts of Gag precursor in these mouse cells. Sucrose gradient centrifugation followed by Western blotting demonstrated that the reactive material in the cell lysates had a large S value, consistent with high-molecular-weight aggregates present in mouse cells. This work thus confirmed previously published results (2, 23).
Previously it had been shown that mouse-human cell fusions were more permissive than mouse cells to HIV replication. In order to map the responsible factor(s) or gene, a panel of monochromosomal mouse-human SCHs was obtained from the Coriell Cell Respository and other sources (Table (Table1).1). The parents of these hybrids were mouse A9, 3T6, or LMTK− cells, which are of fibroblast origin. Each of the cell lines of this panel was created by fusion with human cells containing a single marked chromosome and is essentially monochromosomal for that human chromosome. Each of these cell lines was separately transduced at a low MOI with VSV G-pseudotyped HIV vector (HIV-CIB) that encodes all cis-acting sequences, gag, pol, tat, rev, and a cassette that contains both cycT1 and blasticidin deaminase (bsd) genes. In human cells transduced with HIV-CIB, provision of a viral envelope glycoprotein is sufficient to mediate virus production. Transduced SCHs were maintained as a pool under blasticidin S selection. Each SCH pool was then transiently transfected with an HIV-eYFP reporter (to measure transfection efficiency and also confirm functionality of cyclin T1) and VSV G expression plasmid, and the titers of released virus were determined on 293T cells. Only the SCH harboring chromosome 2 (GM11712) reproducibly gave high virus titers, and the titer was occasionally greater than that obtained from similarly transfected 293T cells harboring an HIV vector carrying both luciferase and bsd genes (Table (Table1).1). GM11686, a second chromosome 2-containing cell line, gave very similar results (Table (Table11).
Since HIV does not use VSV G as its envelope, these experiments were repeated with M-tropic ADA envelope, with testing limited to a select number of hybrids. Again, infectious virus was recovered only from the chromosome 2 hybrids, as measured on GHOST HI5 cells (Fig. (Fig.1).1). As a third test, select cell lines were first transduced with an MLV-based vector encoding cyclin T1, and the functionality of cyclin T1 was confirmed by transducing those cells with HIV-eYFP(VSV G) (the vector encodes eYFP but not cyclin T1) (not shown). These lines were then transiently transfected with an R5 proviral DNA plasmid clone JR-CSF, and supernatant was used to infect GHOST HI5 cells, which were then subjected to flow cytometry. A substantial amount of R5 virus was released from GM11712 cells as judged by the activation of the integrated LTR-eGFP reporter in the target cells, almost equivalent to that from 293T cells (Fig. (Fig.22).
Several controls were performed both to demonstrate the specificity of the result obtained with the chromosome 2 SCHs and to eliminate potential sources of error. Select SCHs were transfected with an MLV vector encoding eGFP, CMV-driven MLV gag-pol expression construct, and VSV G plasmid. Differences observed in measurable infectious MLV release (based upon eGFP titer on naive targets) between the SCHs were only a fewfold and did not reflect those observed with HIV (Fig. (Fig.3A).3A). Similar results were observed after transfecting select SCHs with a replication-competent avian leukosis virus-based vector encoding eYFP (data not shown).
As a completely separate method of recovering integrated provirus from cells, select SCHs with integrated HIV-CIB were infected at a high MOI with HDAd-VSV G. This “gutted” adenoviral vector encodes only VSV G and LacZ. Adenoviral transduction efficiency, as judged by X-Gal/LacZ staining of the producers, was >95% and approximately equivalent for each of the SCHs (although 2 orders of magnitude less than that of 293T cells). At 48 h, the titers of virus from recovered supernatant were determined on naive human cells and again only the chromosome 2-containing SCH gave appreciable infectious HIV release (Fig. (Fig.3B3B).
It is conceivable that the observed results were due to differences in LTR activity secondary to variability in cyclin T1 or other Tat cofactor function. In order to test LTR transcriptional activation, select SCHs already expressing cycT1 from integrated HIV-CIB were transduced with increasing amounts of HIV-eYFP(VSV G) (the vector encodes eYFP but not cyclin T1). With the exception of the GM13259 cell line, which harbors intact chromosome 12 and the gene for cyclin T1, both transduction efficiency (as measured by eYFP) and fully spliced product expression level (as measured by mean fluorescence intensity [MFI]) were roughly equivalent for chromosome 2- and non-chromosome 2-containing SCHs (Table (Table2),2), suggesting approximately equivalent Tat cofactor function in the tested SCHs. For GM13259 cells, both transduction efficiency and MFI were greater by severalfold, consistent with the presence of chromosome 12 carrying the cyclin T1 gene.
For further confirmation of this result, we next directly transfected parental A9 cells and select SCHs with HIV-cycT1-IRES-eYFP along with VSV G, and measured by flow cytometry both transfection efficiency in the mouse cells and transduction efficiency of released virus on naïve human cells. Again, only the chromosome 2-containing cell lines gave appreciable virus release, despite similar transfection efficiencies and MFIs (Table (Table3).3). Recoverable viral titers from 293T cells were 10-fold higher, possibly due to differences in transfection efficiency or because in the mouse cells expression of cyclin T1 and all viral gene products were dependent upon expression of cyclin T1 from the transfected plasmid itself.
Despite the fact that we typically normalized infectious HIV release to transfection efficiency (especially when comparing murine cell lines), we were concerned that slight differences in transfection efficiency would yield markedly different amounts of released virus. To investigate this, the GM13259 cell line was transfected using increasing equimass amounts of pME VSV G, pHIT60, and pBABE-IY plasmid DNAs, and transfection efficiency was measured by both syncytium formation and flow cytometry. Transfection efficiency correlated with infectious MLV release from GM13259 in a linear manner, with r values ranging between 0.75 and 0.84, thereby justifying the simple normalization procedure.
We were concerned that perhaps the HIV-CIB vector copy number was much higher in the chromosome 2-containing SCHs, which would result in higher infectious virus release. We thus performed semiquantitative PCR using HIV-specific primers (β-actin serving as normalization control), and there were no discernible differences in genomic proviral copy number between the SCHs tested (not shown). To determine whether there were differences in the amounts or ratios of spliced and unspliced mRNAs, total RNA was prepared from select SCHs and subjected to RT-PCR using primers that spanned the truncated tat-rev intron. RT-PCR for β-actin again served as the normalization control. As shown in Fig. Fig.4,4, the amount of β-actin mRNA present was approximately the same for each of the SCHs tested. A9 and GM11686 cells appeared to have increased amounts of unspliced mRNA compared to the other SCHs, including GM11712 cells (Fig. (Fig.5).5). Using the same primers, however, there was little variability in the amount of fully spliced mRNA products in the SCHs tested (Fig. (Fig.4).4). 293T cells had a marked increase in unspliced viral mRNA, which had been previously noted. A similar result for unspliced viral mRNA was obtained using primers that span 2.0 kb of gag-pol, with GM11686 having at most a two- to fourfold increase in the amount of unspliced viral mRNA compared to the other SCHs tested (not shown).
Because of the known defect in Gag processing that occurs in mouse cells, we wished to see whether the defect was corrected in the chromosome 2-carrying cell line. Culture supernatants from several SCHs with integrated HIV-CIB vector were analyzed by immunoblotting for the presence of CA. Notably, GM11712 cells released substantially greater absolute amounts of CA compared to the other cell lines, similar to 293T cells (Fig. (Fig.5A,5A, top panel). In addition, the percentage of CA released in the supernatant was also much greater for GM11712 (Fig. (Fig.5A,5A, bottom panel). Similar results were obtained with GM11686 (not shown). To demonstrate that the released CA was actually part of a viral particle, the supernatant was run on a sucrose step gradient and the amount of CA was detected by an ELISA. The large amount of CA that was released from GM11712 had essentially the same buoyant density as viral particles released from 293T cells (Fig. (Fig.5B).5B). A9 and 9HIS cells released much less CA. Flow cytometric analysis of cells that had been fixed and then permeabilized demonstrated detectable amounts of intracellular CA in GM11712, whereas the parental A9 cells had very little (not shown).
There is some evidence that production of CA from Gag may not be a strictly linear phenomenon; in other words, Gag may cooperatively assemble or multimerize at the plasma membrane (27), leading to much greater CA processing and potentially infectious virus release in the presence of larger amounts of p55 precursor. To examine intracellular Gag levels, whole-cell protein lysates were prepared from select SCHs. These were denatured, electrophoresed using SDS-PAGE, transferred to nitrocellulose, and probed using HIV antisera. At most, there was a twofold increase in the amount of p55 in the chromosome 2-containing SCHs, whereas there was a marked increase in the amount of intracellular p24 in those cell lines (Fig. (Fig.5C5C).
Expression of Vpu is known to increase viral particle release up to 10-fold from certain cell types, perhaps by countering a restrictive host factor (36). A vector that is similar to HIV-CIB but that also carries the Vpu gene in its native location (within the tat-rev intron) was introduced at low MOIs into select SCHs, and culture supernatants and cell lysates were immunoblotted for CA. In the presence of Vpu, many of the SCHs released much greater amounts of CA, but the effect was more modest in the chromosome 2-containing cell line (Fig. (Fig.6A).6A). The CA yield (amount in culture supernatant divided by amount in cell lysate) also increased severalfold in the non-chromosome 2-carrying mouse cell lines, similar to the yield in HeLa cells, but increased less in the chromosome 2-carrying SCHs (Fig. (Fig.6B).6B). As a control, COS cells showed no difference in CA release in the presence of Vpu (Fig. (Fig.6A6A).
To determine whether Vpu also increased infectious virus release, select SCHs were separately transduced at low MOIs with an HIV-CIB vector with or without Vpu intact. Semiquantitative PCR confirmed that the average proviral copy numbers were approximately the same for each SCH (not shown). These were then transfected with VSV G expression plasmid and a CMV-RFP reporter (to normalize for transfection efficiency), and the titers of infectious virus were determined on HOS cells. Inclusion of Vpu in the vector did not increase the amount of infectious virus from mouse cell lines that were very nonpermissive, such as A9 and GM11715, whereas it did have a slight effect on the other SCHs tested, including GM11686 (Fig. (Fig.6C).6C). These results suggest that the presence of chromosome 2 increases both Gag processing and infectious HIV particle release, separate from any effects Vpu may have on facilitating CA release.
Previously it had been shown that transfection of mouse cells with codon-optimized, CMV-driven gag-pol (CO GP) allowed Gag to reach the plasma membrane and also increased CA release from those cells (35). Transient transfection of murine cells with a plasmid encoding codon-optimized, CMV-driven gag-eGFP fusion resulted in fluorescence localized to the plasma membrane (not shown), consistent with previous work (35). We first confirmed that CO GP had approximately equivalent activity compared to our standard packaging vector pHIV-PV (34) in producing lentiviral vector supernatant from 293T cells. Transfection of parental A9.CIB cells with CO GP, however, did not rescue infectious virus, nor did it increase the amount of infectious virus release from the chromosome 2-containing cell lines that were cotransfected with HIV-eYFP (Fig. (Fig.7).7). This result was confirmed in other murine cell lines (not shown). This suggests that neither the plasma membrane localization of Gag nor overexpression of structural gene products (independent of Rev and the RRE) is sufficient to mediate virus release from mouse cells.
Here we utilized a panel of monochromosomal mouse-human SCHs to demonstrate that the presence of human chromosome 2 is sufficient to allow infectious HIV release. This correlated with CA production and occurred whether VSV G or M-tropic envelope was used as the complementing cell surface glycoprotein. The effect appeared specific to HIV in that production of other retroviruses was minimally affected. The effect was out of proportion to the increase seen for Gag protein and unspliced mRNA. While Vpu did increase CA production, especially in the absence of chromosome 2, its expression did not correlate well with infectious virus release. A parsimonious explanation for the above results is that one or more determinants or genes that is capable of stimulating virus release from mouse cells lies on human chromosome 2.
Because production of infectious virus may not correlate linearly with the amount of Gag due to cooperative assembly effects (27), we went through some effort to demonstrate the following in select SCHs. (i) The number of integrated proviruses was similar. (ii) Cyclin T1 function was approximately thesame (with the exception of the chromosome 12-carrying SCH). (iii) The amounts of spliced mRNA were roughly equivalent. (iv) The quantities of cell-associated precursor Gag were comparable. (v) Transfection of codon-optimized, CMV promoter-driven HIV gag-pol did not rescue the mouse cell defect. We also observed inconsistent amounts of unspliced viral RNA in the SCHs in that the level in nonpermissive parental A9 cells was similar to that in permissive GM11686 cells, whereas permissive GM11712 cells had levels lower than those of other SCHs (Fig. (Fig.4).4). We cannot exclude the possibility that the differences in the amount of unspliced RNA result in variations in the amount of p55 Gag. This in turn could result in much higher levels of infectious virus release, related to highly cooperative effects of Gag at the plasma membrane (27). Consequently, it is possible that one or more factors on chromosome 2 influences the amount of unspliced mRNA and Gag expression, either through transcriptional, antisplicing, translational, or other poorly defined effects.
We also tested whether expression of Vpu is capable of alleviating the block to infectious virus release from mouse cells. Vpu is known to assist in the down-regulation of cell surface CD4 through targeting to an adapter protein (21); its role in increasing CA release from certain human cell lines is more enigmatic. Here we also demonstrated that expression of Vpu allowed increased CA production/release from mouse cells, similar to nonpermissive human cells. That effect, however, did not translate into a marked increase in infectious virus release, suggesting that Vpu functions in a separate pathway to augment CA production, independent of the action of any factors present on chromosome 2.
It is possible that Gag is mislocalized in mouse cells, which is corrected in the presence of chromosome 2. In permissive human cells, Gag may have a punctate cytosolic appearance or mainly be localized at the plasma membrane, depending upon the level of expression and tissue/cell of origin. When codon-optimized Gag was expressed as a GFP fusion protein in a CMV promoter-driven expression vector, it was localized at the plasma membrane in all cell types tested, regardless of origin (not shown). Expression of codon-optimized gag-pol was insufficient, however, to attain infectious virus release from mouse cells. When expression of codon-optimized Gag-GFP was dependent on both Rev and RRE, we observed a punctate cytosolic pattern in both permissive human cells and mouse cells with chromosome 2, whereas in other mouse cells it was more diffuse (not shown). Whether this is reflective of the trafficking of the unspliced mRNA or differences in the Gag expression level is unknown.
Cytosolic localization of proviral HIV mRNA has not been fully evaluated in mouse cells, let alone permissive human cells. Use of the ms2 phage protein binding system has suggested that MLV genomic RNA traffics along an endosomal pathway, which is dependent upon envelope and colocalizes with Gag (1). Precise determinants for mRNA entry into this pathway have not yet been fully elucidated. Because HIV has additional cis-acting nucleic acid sequences that function in RNA nuclear export, it is conceivable that these elements play a role in directing the unspliced RNA to a specific, postnuclear pathway, perhaps in concert with other viral proteins and cellular factors. Absence of one or more of these factors would abrogate infectious virus release from mouse cells. Provision of a multimerized CTE to the RNA partially corrected the defect observed in mouse cells in that Gag became localized to the plasma membrane and was processed, but it is unknown whether infectious virus was produced (35).
At least two genes lie on chromosome 2 that are involved in export of unspliced HIV RNA. Crm1, located on 2p, binds to Rev to allow export of RRE-containing mRNAs (11, 13). hRIP (4, 12) is located on 2q36. At steady state it is located in the nucleus and in the perinuclear (cytoplasmic) region. Recently by the use of dominant-negative forms and RNA interference, hRIP was shown to be specifically involved in the movement of Rev/RRE-dependent HIV RNAs from the nuclear periphery to the cytoplasm (30). Whether Crm1 or hRIP has a role in viral RNA cytoplasmic trafficking (or is involved in the “hand-off” to other as-yet uncharacterized factors) is unknown.
Both of these genes were introduced by transient transfection into murine A9.CIB cells, alone and in combination, and there was no appreciable effect on either CA production or infectious virus production (data not shown). Other cellular genes may cooperate with Rev to allow more efficient nuclear export of unspliced mRNAs, including SAM68 (25, 28). SAM68 is not present in the chromosome 2-carrying SCHs, and transient transfection of SAM68 into A9.CIB cells did not result in an increase in CA or infectious HIV release (not shown). Other genes implicated in Rev function or binding carried on chromosome 2 (10) have yet to be tested.
The results presented here do not exclude the possibility that other factors that are not encoded on chromosome 2 are critical for infectious HIV release from mouse cells. We have identified three microcell mouse-human hybrid clones extremely permissive to HIV release (not shown). Two of these clones are identical and have a complex chromosome 2-12 fusion; the third has only an intact chromosome 12. Efforts to identify a portion of chromosome 2 (or any chromosome other than 12) in this clone have thus far failed. It is therefore conceivable that non-chromosome 2-encded factors contribute to infectious virus release. For the first time, however, we have identified several murine-human hybrid cell lines that are permissive for infectious HIV release. These lines may serve as useful tools to mechanistically dissect the postintegration block to HIV production in the mouse and to identify the gene(s) responsible for this phenotype.
We thank Dan Littman (NYU School of Medicine), Larry Donehower (Baylor), and Oxford Biomedica for generous reagent gifts. We also thank Phil Ng (Baylor) for preparation of the HDAd-VSV G stocks; J. Carl Barrett (NCI) for SCHs carrying chromosome 6, 8, 9, 18, or 19; and Elton Stubblefield (MD Anderson Cancer Center) for the SCH carrying chromosome 17. The following reagents were obtained from the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH: HIV-immunoglobulin from NABI and NHLBI, GHOST HI5 cells from Vineet KewalRamani (NCI-Frederick) and Dan Littman, and JR-CSF proviral clone from Irvin Chen (UCLA).
This work was supported initially by the American Foundation for AIDS Research (AmfAR) and then by the National Institutes of Health. R.E.S. was an Edward Mallinckrodt, Jr., Foundation Scholar.