Vpu exerts a positive effect on HIV-1 infectivity by down-modulating CD4 receptor molecules at the surface of HIV-1-producing cells.

Human immunodeficiencey virus, type 1 (HIV-1) encodes three proteins, Nef, Vpu, and gp160, that down-modulate surface expression of the CD4 receptor during viral infection. In the present study, we have investigated the role of CD4 down-modulation in the HIV-1 infection cycle, primarily from the perspective of Vpu function. We report here that, like Nef, Vpu-mediated CD4 degradation modulates positively HIV-1 infectivity. Our data reveal that accumulation of CD4 at the cell surface of Vpu-deficient HIV-1-producing cells leads to an efficient recruitment of CD4 into virions and to an impairment of viral infectivity. This CD4-mediated inhibition of viral infectivity was not observed when a CD4 mutant unable to bind Env gp120 was used or when VSV-G glycoprotein was utilized to pseudotype viruses, suggesting that an interaction between CD4 and gp120 is required for interference. Indeed, protein analysis of Vpu-defective viral particles reveals that CD4 recruitment is associated with an increased formation of gp120-CD4 complexes at the virion surface. Interestingly, we did not detect any difference at the level of total virion-associated Env glycoproteins between wild-type and Vpu-defective virus, indicating that accumulation of CD4 at the cell surface and recruitment of CD4 into Vpu-defective HIV-1 particles exert a negative effect on viral infectivity, most likely by promoting the formation of nonfunctional gp120-CD4 complexes at the virion surface. Finally, we show that both Vpu- and Nef-induced CD4 down-modulation activities are required for production of fully infectious particles in CD4+ T cell lines and primary cells, an observation that has clear implications for viral spread in vivo.

CD4 is a type 1 integral membrane glycoprotein expressed primarily on the surface of thymocytes, helper T lymphocytes, and, to a lesser extent, on cells of the monocyte/macrophage lineage (1). Besides its role in T cell ontogeny and antigenspecific T-cell activation, CD4 also serves as the primary cellular co-receptor for human immunodeficiency virus, type 1 (HIV-1) 1 viral entry into cells (1). Infection of target cells is initiated by the high affinity interaction (Ͻ10 nM) occurring between CD4 and the viral envelope glycoprotein gp120 (2).
Although the CD4 receptor plays a critical role during HIV-1 entry into target cells, it is well established that HIV-1 downmodulates its own receptor at later stages of the infection cycle in productively infected cells. Indeed, HIV-1 encodes three proteins, Vpu, Env, and Nef, that have a profound effect on CD4 catabolism and are involved in the removal of CD4 from the surface of infected cells (3,4). Early in infection, the Nef protein down-regulates CD4 molecules that are already present at the cell surface by accelerating their endocytosis and subsequent degradation in lysosomes (5)(6)(7)(8)(9). At later stages of infection, the envelope gp160 precursor, through its high receptor binding affinity, retains newly synthesized CD4 in the endoplasmic reticulum (ER). These gp160-CD4 complexes not only prevent cell surface expression of CD4 but also block gp160 maturation and trafficking (10 -12). The accessory Vpu (viral protein U) regulates the half-life of CD4 molecules complexed to gp160 by selectively inducing their degradation in the ER, thus, presumably, releasing gp160 and permitting its maturation, trafficking, and incorporation into progeny virions (13).
Vpu is an 81-amino acid protein unique to HIV-1 and the closely related simian immunodeficiency virus isolated from chimpanzee (SIV cpz ) (14,15). The vpu gene product is a phosphorylated oligomeric type 1 integral membrane protein that is translated from a bicistronic mRNA that also encodes the Env glycoproteins (16). Vpu performs two main functions during the viral life cycle; it enhances the release of virions from infected cells, and it mediates the selective degradation of the CD4 receptor in the ER (17)(18)(19). Vpu-mediated CD4 degradation is a multistep process that is triggered by the direct physical binding of Vpu to the cytoplasmic tail of CD4 (20). Mutational studies have revealed that phosphorylation of seryl residues at positions 52 and 56 of Vpu is required to initiate CD4 proteolysis but is not essential for the initial binding of the protein to CD4 (20). Using genetic and biochemical assays, Margottin et al. have shown that Vpu connects CD4 to the ubiquitin-proteasome degradative pathway through a direct, phosphoserine-dependent interaction with the F box protein ␤-TrCP (21). This finding, combined with the demonstrated sensitivity of Vpumediated CD4 degradation to proteasome inhibitors, strongly suggests that Vpu targets CD4 to proteolysis by recruiting the ubiquitin-proteasome cytosolic machinery (22). * This work was supported by grants from the Canadian Institutes for Health Research (CIHR), the Canadian Foundation for Innovation, and the Fonds pour la Recherche en Santé du Québec (to E. A. C.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Both authors contributed equally to this work. Although recent studies have provided new details on the molecular mechanism underlying the down-modulation of the CD4 receptor, it is still unclear why HIV-1 has evolved multiple mechanisms to remove its receptor from the cell surface. Given such functional convergence, it is tempting to speculate that cell surface CD4 regulation must be an important determinant of viral replication and pathogenesis in vivo. Indeed, recent reports have shown that down-regulation of cell surface CD4 by Nef and/or Vpu is important for viral spread (23)(24)(25)(26). All of these studies show that high levels of CD4 on the cell surface interfere with the production of infectious HIV-1 particles; however, they propose different mechanisms to account for this inhibition. Ross et al. showed that overexpression of CD4 in HIV-1-transfected human embryonic kidney cells (293T) leads to a reduction in the overall amounts of virus released but does not affect the absolute infectivity of released virus (26). This study further reported that this CD4-mediated inhibition of viral release is dependent on the specific interaction between CD4 and the viral envelope glycoproteins and is reversed by Nef expression (26). On the basis of these results, Ross et al. (26) proposed that cell surface CD4 interacts with the envelope protein present on budding HIV-1 virions to inhibit viral release. More recently, Bour et al. (23) reported that maintenance of CD4 surface expression in HIV-1-transfected HeLa cells inhibited HIV-1 particle release in an Env-independent manner. The finding that CD4 had no significant effect on particle release by a Vpu-deficient virus led these authors to suggest that CD4 acts by inhibiting the particle release-promoting activity of Vpu (23). However, the infectivity of the released particles was not analyzed in this study. In contrast to the latter studies, Lama et al. (25) found that expression of high levels of CD4 at the cell surface of 293T cells induced a drastic reduction in the infectivity of released virion by the sequestration of the viral envelope by CD4. Protein analysis of virions released from 293T cells transfected with proviral constructs isogenic except for the expression of Nef or/and Vpu revealed that CD4 could accumulate in viral particles while at the same time blocking incorporation of Env into virions (25). More recently, Cortes et al. (24) extended these studies to Jurkat T cell lines and found that physiological levels of surface CD4 interfere in an envelope-dependent manner with the infectivity of released HIV-1 particles.
In the present study, we have investigated the role of CD4 down-modulation in the virus life cycle, primarily from the perspective of Vpu function. We report here that Vpu, like Nef, exerts a positive effect on HIV-1 infectivity by down-regulating cell surface CD4. The effect of Vpu on viral infectivity correlates with the protein's ability to degrade CD4 in the ER and with levels of CD4 surface expression in virus-producing cells. Interestingly, accumulation of CD4 at the cell surface of Vpudeficient HIV-1 producing cells was found to lead to efficient recruitment of CD4 into virions while at the same time promoting the formation of nonfunctional gp120-CD4 complexes at the virion surface. Finally, both Vpu and Nef were found to be required for production of fully infectious HIV-1 particles in CD4 ϩ T cell lines and in peripheral blood mononuclear cells (PBMCs), thus suggesting that these accessory proteins, through their ability to down-regulate CD4, are therefore likely to facilitate the spread of HIV-1 in vivo.

EXPERIMENTAL PROCEDURES
Cell Lines-Human SV40-transformed 293-T fibroblasts and HeLa-CD4-LTR-␤-gal cells (27) were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum (FCS) and 1% antibiotics (penicillin and streptomycin) (Dulbecco's modified Eagle's medium plus 10% FCS). The human lymphoid T cell lines MT4, Jurkat low-CD4, and 1G5 were maintained in RPMI 1640 medium supplemented with 10% FCS and 1% antibiotics (RPMI plus 10% FCS). Jurkat CD4 Ϫ and Jurkat high-CD4 cell lines were maintained in RPMI plus 10% FCS containing Geneticin (500 g/ml) or Geneticin (500 g/ml) and zeocin (50 g/ml) (Invitrogen), respectively, as described (24). The CD4 ϩ 1G5 indicator cell line is derived from Jurkat cells and expresses a luciferase reporter gene under the control of the HIV-1 long terminal repeat (LTR) (28). Peripheral blood mononuclear cells (PBMCs) were isolated from volunteers by Ficoll-Paque centrifugation as recommended by the manufacturer (Amersham Biosciences) and washed thoroughly with PBS to remove platelets. Cells were then cultured in the presence of phytohemagglutinin (5 g/ml) for 72 h to activate lymphocytes. Following activation, cells were washed with complete media to remove lectins and maintained in RPMI plus 10% FCS supplemented with 20 units/ml interleukin-2 (Roche Applied Science). All cells were cultured at 37°C under a 5% CO 2 atmosphere.
The anti-CD4 (OKT4) and anti-p24 monoclonal antibodies were derived from ascitic fluids of Balb/c mice that were injected with the OKT4 or p24 hybridoma (catalog nos. ATCC CRL-8002 and ATCC HB9725, respectively; American Type Culture Collection, Manassas, VA). Rabbit anti-CD4 H-370 polyclonal antibodies were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA) (catalog no. sc-7219). The anti-HIV-1 serum (162) was obtained from an HIV-1 infected individual whose serum was tested positive for the presence of HIV-1 antibodies by ELISA. Finally, the anti-gp41 monoclonal antibodies were obtained from supernatant of cultured Chessie 8 hybridoma cells (34). The anti-gp120 mouse monoclonal antibody (1D6) (35) and Chessie 8 hybridoma cells were both obtained from the AIDS research and reference reagent program.
Virus Stock Production-HXBH10-vpu ϩ /vpu Ϫ and VSV-Gpseudotyped virus stocks were produced by transfecting 5 ϫ 10 6 293T cells with 5 g of proviral DNA constructs alone or, in the case of pseudotyping, with 10 g of proviral DNA constructs and 10 g of SVCMV-VSV-G DNA by the calcium-phosphate method (29). 48 -72 h post-transfection, supernatants were collected, clarified by centrifugation at 3,000 rpm for 30 min, and filtered through 0.45-m membrane filters. Virions were then sedimented onto a 20% sucrose cushion by ultracentrifugation at 45,000 rpm in a 50.4 rotor type (Beckman) for 2 h at 4°C. Viral production was then evaluated by performing a standard reverse transcriptase (RT) activity assay in 50 l as described previously (31).
Flow Cytometry Analysis-Jurkat T cell lines or transfected 293T cells were washed, harvested in PBS containing 5% FCS, and incubated with OKT4 anti-CD4 monoclonal antibody in 5% FCS-PBS. After 30 min at 4°C, cells were washed twice in PBS and incubated for another 30 min at 4°C with fluorescein isothiocyanate-conjugated goat antimouse antibody in 5% FCS-PBS. Cells were then washed twice in PBS and analyzed by flow cytometry on a FACSCalibur® cytometer (Becton Dickinson). As negative control, cells were stained using fluorescein isothiocyanate-conjugated goat anti-mouse antibody only (Becton Dickinson).
Replication Studies and Viral Infections-MT4 cells (10 6 ) or PBMCs (3 ϫ 10 6 ) were infected with equivalent amounts of virus (HxBH10vpu ϩ /vpu Ϫ ) as determined by RT activity (MT4, 2 ϫ 10 5 cpm of RT; PBMC, 7.5 ϫ 10 6 cpm of RT). Following a 4-h adsorption period, cells were washed extensively with PBS and resuspended in 10 ml of RPMI plus 10% FCS. At regular time intervals, cells were recovered by centrifugation and seeded in fresh medium at a concentration between 10 6 and 5 ϫ 10 5 viable cells/ml. Cell supernatants were analyzed for the presence of virus by performing RT assays as described previously (31). Jurkat T cell lines (CD4 Ϫ , low-CD4, or high-CD4) (3 ϫ 10 6 ) were infected with equivalent amounts (3 ϫ 10 7 cpm of RT) of VSV-Gpseudotyped HxBH10-vpu ϩ /vpu Ϫ virus. Following a 4-h adsorption period, cells were washed extensively with PBS and resuspended in 10 ml of RPMI plus 10% FCS. The infectivity of virions released in the supernatant was analyzed 48 h postinfection.
Infectivity Assays-Infectivity of virus present in clarified supernatants was assessed using the multinuclear activation of ␤-galactosidase infectivity (MAGI) assay (27) or the luciferase infectivity assay (28). Briefly, for the MAGI assay, equal amounts of virus (6 ϫ 10 5 to 6 ϫ 10 6 cpm of RT) was used to infect HeLa-CD4-LTR-␤-gal cells seeded the previous day in 24-well plates at 3 ϫ 10 4 cells/well. 48 h postinfection, cells were washed, fixed, and stained as described (27). The number of ␤-galactosidase-positive blue cells was determined using light microscopy. Each sample was analyzed in duplicate. For the luciferase assay, equal amounts of virus (1-5 ϫ 10 6 cpm of RT) was used to infect 2 ϫ 10 5 1G5 cells (LTR-luciferase indicator cell line). Immediately after adsorption, cells were thoroughly washed, and neutralizing amounts of anti-CD4 and anti-gp120 antibodies were added to cell supernatant to prevent syncytia formation. 24 h postinfection, AZT was added to a final concentration of 10 M to limit viral production to one infection cycle. 48 h postinfection, cells were harvested and lysed, and a luciferase assay was performed according to the manufacturer's recommendation (Promega).
Western Blot and Co-immunoprecipitation-Supernatants from HIV-1-transfected MT4 cells or from HIV-1/CD4 co-transfected 293T cells were collected and clarified by centrifugation at 3000 rpm. Virions were pelleted from cell-free supernatants by ultracentrifugation onto a 20% sucrose cushion as described above and lysed in radioimmune precipitation buffer (140 mM NaCl, 8 mM Na 2 HPO 4 , 2 mM NaH 2 PO 4 , 1.2 mM deoxycholic acid sodium salt, 0.5% SDS, and 1% Nonidet P-40). Similar amounts of virion, normalized by RT activity measurements, were lysed, divided in two equal fractions, and immunoprecipitated using either anti-CD4 (OKT4) monoclonal antibodies or anti-HIV-1 serum (162). Immune complexes were precipitated using protein A-Sepharose beads (Amersham Biosciences), washed extensively, resuspended in reducing sample buffer, and separated by 8 or 12.5% SDS-PAGE. Following electrophoresis, viral proteins were transferred onto a nitrocellulose membrane (0.45-m pore size; Bio-Rad) by electroblotting for 3 h at 30 V in a Bio-Rad Trans Blot Cell. The membrane was then incubated for 1 h in blocking solution buffer (Tris-buffered saline containing 1% Tween 20 and 5% nonfat dry milk) and incubated for 2 h with either rabbit anti-CD4 (1:500 dilution) or monoclonal antibodies directed against gp120 (1:1000 dilution) or p24 (1:4000 dilution). Bound antibodies were then probed with horseradish peroxidase-linked antimouse or anti-rabbit immunoglobulin (used at 1:7500), washed extensively, and revealed using a sensitive enhanced chemiluminescence detection system (ECL detection kit; Amersham Biosciences). For detection of virion-associated gp41 levels, serial dilution of virion lysates were directly separated by PAGE and analyzed by Western blot as described above using monoclonal antibodies directed against gp41 (undiluted culture fluid supernatant).

Vpu-mediated CD4 Degradation
Alleviates the Negative Effect of Cell Surface CD4 on Viral Infectivity-Given that Nefmediated CD4 down-modulation was previously reported to be an important regulator of HIV-1 infection cycle (24 -26), we first investigated the effect of cell surface CD4 expression on viral infectivity in 293T cells using proviral constructs isogenic except for Vpu expression. HxBH10-vpu ϩ or HxBH10-vpu Ϫ isogenic virions were produced in 293T cells following co-transfection of the corresponding proviral construct with increasing concentrations of SVCMV-CD4. These proviral constructs are derived from HxBc2 and as such contain a premature stop codon in Nef (36). 48 h post-transfection, cells and culture supernatants were collected for cell surface CD4 measurement and evaluation of viral infectivity, respectively. As previously reported, the expression of Vpu was found to down-regulate CD4 expression at the cell surface as revealed by FACS analysis ( Fig. 1A) (19). The infectivity potential of Vpu Ϫ and Vpu ϩ HIV-1 virus produced in each cell culture was then evaluated by MAGI assay using equivalent amounts of virus as measured by reverse transcriptase activity. The results of Fig. 1B clearly show that CD4 expression exerted an inhibitory effect on HIV-1 infectivity that was directly proportional to cell surface CD4 expression levels. Interestingly, the negative effect of CD4 on viral infectivity was more pronounced (2-7-fold) with Vpudefective HIV virus than with Vpu ϩ virus. In contrast, the infectivity potential of Vpu ϩ and Vpu Ϫ viruses produced from CD4 Ϫ 293T cells was found to be similar (Fig. 1, C and D). To further confirm the infectivity data obtained using the MAGI assay, we tested the infectivity of Vpu ϩ or Vpu Ϫ virus produced in mock-or CD4-transfected 293 T cells using the 1G5 indicator (luciferase) CD4 ϩ T cell line. As shown in Fig. 1C, we found that the infectivity potential of Vpu ϩ and Vpu Ϫ virus produced in CD4 Ϫ cells was almost identical, whereas it differed substantially when viruses were produced in the presence of CD4. In the latter case, Vpu-defective virus displayed an infectivity potential that was ϳ4-fold lower than Vpu ϩ virus. Furthermore, we extended these analyses to Jurkat T cell lines expressing different levels of cell surface CD4 (Fig. 1E). Data presented in Fig. 1F clearly confirm that Vpu alleviates the negative effect of cell surface CD4 on viral infectivity in Jurkat T cell lines expressing different levels of cell surface CD4.
To examine whether Vpu-mediated CD4 degradation was directly responsible for modulating positively viral infectivity in CD4-expressing cells, we tested mutant forms of Vpu or CD4 whose phenotypes were associated with a Vpu-mediated CD4 degradation impairment. The HxBH10-vpu 52/56 provirus encodes a mutant Vpu protein that harbors substitution mutations at both Ser-52 and Ser-56 phosphoacceptor sites. This Vpu mutant is unable to mediate CD4 proteolysis (31). CD4⌬32 is a CD4 cytoplasmic domain deletion mutant that is expressed at high levels on the cell surface and is not sensitive to Vpumediated CD4 degradation (37). HxBH10-vpu ϩ , HxBH10vpu Ϫ , or HXBH10-vpu 52/56 proviral constructs were co-transfected with SVCMV-CD4, SVCMV-CD4⌬32, or control plasmids in 293 T cells. 48 h post-transfection, viruses were collected, and MAGI assays were performed as described under "Experimental Procedures." Fig. 1D shows that Vpu Ϫ , Vpu ϩ , or Vpu 52/56 HIV-1 particles produced in the absence of CD4 all exhibit similar infectivity potential. Interestingly, in the presence of wild type CD4, the number of ␤-galactosidase-positive cells obtained upon infection with the Vpu 52/56 virus was reduced considerably as compared with the Vpu ϩ virus. Indeed, the infectivity potential of the the Vpu 52/56 was comparable with that of the Vpu Ϫ virus. Moreover, both Vpu Ϫ and Vpu ϩ HIV-1 virus produced in 293T cells expressing CD4⌬32 displayed a drastic decrease in infectivity (Fig. 1D), thus further supporting the importance of down-regulating CD4 to obtain infectious viral particles in CD4-expressing cells. Taken together, these results confirm that cell surface CD4 expression exerts a negative effect on the infectious potential of nascent HIV-1 virus produced from CD4-expressing cells and show that Vpu-mediated CD4 degradation alleviates the negative effect of cell-surface CD4 on viral infectivity.
Vpu Modulates Positively the Infectivity of HIV-1 Particles Produced in T Cell Lines and PBMCs-To investigate whether Vpu can modulate the infectivity of HIV-1 particles released from cells expressing physiological levels of surface CD4, we analyzed the infectious potential of Vpu ϩ and Vpu Ϫ isogenic viruses produced upon infection of the human MT4 CD4 ϩ T cell line and PBMCs isolated from two different donors. The viruses used to perform the replication kinetics in MT4 cells and PB-MCs were produced following transfection of 293T cells with HxBH10-vpu ϩ and HxBH10-vpu Ϫ . As previously reported, MT4 cells and PBMCs, infected with HIV-1 Vpu ϩ viruses released 2-8 times more virus than HIV-1 Vpu Ϫ -infected cell cultures at the peak of viral production (Fig. 2, A-C) (38). At various time intervals during the replication kinetics, cell-free supernatants were collected from each infected culture, and the infectivity potential of the containing virus was determined by MAGI assay using the same amount of virus as determined by RT activity. We consistently observed that Vpu Ϫ virus produced in MT4 (Fig. 2D) and Jurkat (data not shown) CD4 ϩ T cell lines as well as in PBMCs (Fig. 2, E and F) exhibited a substantial impairment of their infectious potential as compared with their Vpu ϩ isogenic counterparts. This impairment of viral infectivity resulting from the lack of Vpu expression ranged approximately from 2-to 10-fold at the peak of viral replication depending on the donor or the tested T cell line. From these results, we conclude that Vpu modulates positively the infectious potential of HIV-1 virions produced from both CD4 ϩ T cell lines and primary PBMCs.
Incorporation of CD4 into Virions Interferes with the Level of Functional Envelope Associated with HIV-1 Viral Particles-To understand the molecular mechanism underlying cell surface CD4-mediated inhibition of HIV-1 infectivity, we performed a detailed protein analysis of viral particles released from CD4expressing 293T cells and investigated whether (i) CD4 molecules were incorporated into HIV-1 virions and (ii) modulation of cell surface CD4 by Vpu influenced the levels of CD4 molecules and HIV-1 Env glycoproteins incorporated into virions. 293T cells were co-transfected with SVCMV-CD4 or control plasmids and HxBH10-vpu Ϫ or HxBH10-vpu ϩ proviral constructs. 48 h post-transfection, cell-free supernatants were collected for viral infectivity determination by MAGI assay or sedimented by ultracentrifugation to isolate HIV-1 virions. Pelleted virions were lysed in radioimmune precipitation buffer and divided into two equal fractions that were immunoprecipitated with either anti-CD4 monoclonal antibodies (OKT4) or anti-HIV-1 human serum. Immunocomplexes were then separated on SDS-PAGE and transferred to nitrocellulose membrane for Western blot analysis using specific antibodies including anti-CD4, anti-gp120 monoclonal antibodies, or anti-p24. The results of Fig. 3A clearly show that CD4 molecules are incorporated into viral particles released from CD4-expressing 293 T cells (Fig. 3A, lanes 5-6, total CD4). Lack of CD4 detection in the supernatant of CD4-expressing 293 T cells rules out the possibility that CD4 associated with viral particles might originate from cell microvesicle contamination (Fig. 3A, lane 2, Total CD4). Substantial amounts of CD4 molecules were incorporated into HIV-1 Vpu-defective virions, whereas levels of CD4 molecules associated with HIV-1 particles were drastically reduced in the presence of Vpu (Fig. 3A, compare lanes 5 and 6, Total CD4). Similar results were obtained when virus produced from MT4 CD4 ϩ T cell lines were analyzed for endogenous CD4 virion incorporation (Fig. 4A, compare lanes 2 and  3). Interestingly, CD4 recruitment into virion was found to be independent of CD4-gp120 interaction, since CD4 was packaged into viral particles produced from an Env-defective proviral construct, and CD4-⌬CDR2, a CD4 mutant unable to bind gp120 (30), was still found to be incorporated into viral particle when coexpressed with Vpu Ϫ proviral constructs in 293T cells (data not shown). Interestingly, Vpu-mediated reduction of virion-associated CD4 molecules was shown to correlate with an enhancement of viral infectivity (Fig. 3C, compare Vpu Ϫ / CD4 ϩ with Vpu ϩ /CD4 ϩ ). Moreover, Fig. 3C reveals that the viral infectivity impairment observed in the absence of Vpu results from CD4 molecules that have the ability to bind gp120. Indeed, the gp120-binding CD4 mutant, CD4⌬CDR2 (30), does not impair the infectivity of Vpu Ϫ virus although CD4⌬CDR2 is expressed at the cell surface and is incorporated in substantial amounts in Vpu Ϫ virion (data not shown). This observation is further supported by the data of Fig. 3D, which shows that VSV-G-pseudotyped Vpu-defective HIV-1 virions produced in the presence or absence of wild type CD4 exhibit comparable viral infectivity, although CD4 incorporation in VSV-G pseudotyped virus or wild type virus was found to be comparable (data not shown). These results suggest that the negative effect of CD4 on viral infectivity involves recruitment of CD4 molecules into virions and requires an interaction between CD4 and gp120.
To further explore the molecular mechanism underlying the effect of Vpu-mediated CD4 degradation on viral infectivity, we next determined whether Vpu expression influenced the levels of Env glycoproteins incorporated into virions released from CD4-expressing 293T cells or MT4 T cells. Figs. 3, A and B, and 4B reveal that Vpu does not influence the levels of Env glycoproteins associated with progeny HIV-1 particles, since the amount of virion gp41 or gp120 was similar in the presence or absence of Vpu (Fig. 3A, lanes 5 and 6 (Total Env), and Fig. 3B or 4B, compare lanes 4 -6 with lanes 1-3). As previously reported by Lama et al. (25), levels of virion-associated gp120 were consistently higher in virion produced from 293 T cells (CD4 Ϫ ) than in virus produced from CD4-expressing 293T cells (Fig. 3A, compare lanes 3 and 4 with lanes 5 and 6, Total Env). This effect was independent of Vpu expression and probably reflects the previously reported CD4-mediated gp160 retention that occurs in the ER, which is known to reduce Env trafficking and presumably incorporation in progeny viral particles (13).
These results indicate that Vpu-mediated CD4 degradation does not lead to more Env glycoproteins being incorporated into HIV-1 virions, and consequently, the negative effect of CD4 on viral infectivity, in our experimental conditions, cannot be explained by a reduction of envelope incorporation into virions. We next investigated whether virion-associated CD4 molecules might interfere with Env function by diminishing the ability of incorporated Env glycoproteins to bind CD4 molecules at the surface of target cells. Specifically, we evaluated whether a percentage of virion-associated gp120 or CD4 was complexed to each other and whether Vpu expression could modulate the amount of such gp120-CD4 complexes. Virionassociated gp120-CD4 complexes were isolated by co-immunoprecipitation using anti-HIV-1 or anti-CD4 (OKT4) antibodies and analyzed for the presence of bound-CD4 or bound-gp120 by Western blot using anti-CD4 or anti-gp120 monoclonal antibodies. The data of Fig. 3A reveal that in the absence of Vpu

FIG. 2. Vpu modulates positively the infectivity of HIV-1 virions produced in MT4 cells and primary PBMCs. MT4 cells (A and D) and
PBMCs isolated from two distinct donors (B and C, E and F) were mock-infected (f) or infected with equivalent amounts of HxBH10-vpu ϩ (OE) or HxBH10-vpu Ϫ (ࡗ) virus. At different time intervals following infection, culture supernatants were collected. Viral production and infectivity were evaluated in duplicates using RT activity measurements (A-C) or the MAGI assay (D-F). These data are representative of the results obtained in at least three independent experiments.
(Vpu Ϫ ), a fraction of gp120 molecules is complexed with CD4, and inversely, a fraction of CD4 molecules is found associated with gp120 (lane 6, Bound CD4, Bound gp120). Expression of Vpu was shown to reduce the formation of gp120-CD4 complexes given that upon overexposure of the autoradiogram, a clear difference in bound gp120 was detected in the presence and absence of Vpu (compare lanes 5 and 6, Bound gp120; results shown for bound Env and total Env were obtained using different film exposure and accordingly cannot be compared). Moreover, using autoradiogram exposure time similar to total CD4, we were unable to detect bound CD4 signal in presence of Vpu (compare lanes 5 and 6, bound CD4). These results suggest that recruitment of CD4 into the virion affects viral infectivity by forming complexes with gp120 at the virion surface, thereby saturating CD4 binding sites on gp120 molecules and decreasing the levels of functional Env glycoproteins at the virion surface. Thus, Vpu, by reducing CD4 cell surface expression and packaging into virion, inhibits this process and consequently permits the release of infectious particles.
Vpu-and Nef-induced CD4 Down-modulation Activities Are Required for Optimal HIV-1 Infectivity-Both Vpu and Nef down-modulate surface expression of the CD4 receptor during HIV-1 infection (6). To investigate their relative contributions to HIV-1 infectivity, we transfected CD4 ϩ MT4 T cells with proviral constructs isogenic except for the expression of Vpu or/and Nef. 3 days post-transfection, the infectivity of virions produced in the culture supernatants was evaluated by MAGI assay using equivalent amounts of virus as determined by RT activity. Results of Fig. 5 indicate that the absence of either Vpu or Nef decreases viral infectivity by 4-fold as compared with wild type virus. Interestingly, the effect of Vpu and Nef on viral infectivity appears to be synergistic rather than additive, since HIV-1 virus produced in the absence of both Vpu and Nef displayed a 20 -30-fold decrease in viral infectivity as compared with wild type virus. This finding indicates that cell surface CD4 down-modulation mediated by Vpu and Nef during HIV-1 viral infection is required for optimal viral infectivity. DISCUSSION HIV-1 Vpu is associated with two biological activities during HIV-1 infection; (i) it facilitates the release of nascent viral particles by a mechanism that is independent of CD4 and Env glycoprotein expression (17,39), and (ii) it mediates the specific degradation of the CD4 receptor in the ER (19). Although Vpu-mediated enhancement of viral release has an obvious role during HIV-1 infection (to increase viral spread), the role of Vpu-mediated CD4 degradation still remains unclear. This effect of Vpu on CD4 must confer a selective advantage to the virus, since HIV-1 encodes two other proteins, Nef and the Env precursor gp160, that also contribute to CD4 down-modulation by acting at distinct steps of the infection cycle. Several hypotheses have been proposed to explain the functional significance of HIV-1-induced CD4 down-regulation. First, elimination of the primary receptor from the cell surface might prevent cytopathic effects induced by multiple superinfections (40). Al-

FIG. 3. Down-regulation of CD4 cell surface expression and packaging into HIV-1 particles enhances viral infectivity by augmenting functional Env glycoproteins levels at the virion surface.
A, CD4 molecules are incorporated into HIV-1 particles during viral morphogenesis. HxBH10-vpu ϩ (Vpu ϩ ) or HxBH10-vpu Ϫ (Vpu Ϫ ) was cotransfected with SVCMV-CD4wt (CD4wt) or control plasmids (CD4 Ϫ ) in 293T cells. Virions were pelleted down from cell free supernatants by ultracentrifugation and normalized for RT activity. Virions were then lysed, divided into two equal fractions, and immunoprecipitated (IP) with either anti-CD4 monoclonal antibody (␣OKT4) or anti-HIV-1 serum (␣HIV-1). Immunocomplexes were separated by gel electrophoresis, transferred to nitrocellulose, and analyzed by Western blot (WB) using anti-CD4 (␣CD4) polyclonal antibodies or anti-gp120 (␣gp120) monoclonal antibodies or anti-p24 monoclonal antibodies (␣p24) as indicated. Results shown for bound Env and total Env were obtained using different film exposition and accordingly cannot be compared. B, effect of CD4 cell surface expression on gp41 incorporation into viral particles. Virion prepared from transfected 293T cells were lysed in radioimmune precipitation buffer. Serial dilutions of viral lysates were separated by SDS-PAGE, transferred to nitrocellulose, and analyzed by Western blot, using anti-gp41 monoclonal antibodies or anti-p24 serum. UD, undiluted. C, effect of CD4 cell surface expression on viral infectivity. Cell-free supernatants prepared as described above were collected, and virus production was determined by RT activity measurements. The infectivity of virus produced in each cell culture was evaluated by MAGI assay. These data are representative of the results obtained in at least four independent experiments. D, effect of CD4 cell surface expression on the infectivity of VSV-G-pseudotyped HIV-1 Vpu Ϫ particles. HxBH10-vpu Ϫ and SVCMV-CD4 (CD4 ϩ ) or control plasmid (CD4 Ϫ ) were cotransfected with or without SVCMV-VSV-G in 293T cells. Supernatants were collected, and virus production was determined by RT activity measurements 48 h post-transfection. The infectivity of virus produced in each cell culture was evaluated by MAGI assay. These data are representative of the results obtained in at least three independent experiments. ternatively, CD4 down-modulation might prevent the trapping and aggregation of nascent progeny virion at the cell surface, an effect reminiscent of the function of the influenza virus neuraminidase protein (41).
In the present study, we provide evidence showing that Vpumediated CD4 degradation allows the release of fully infectious viral particles, since excessive levels of CD4 molecules on the cell surface of virus-producing cells result in incorporation of these molecules into nascent viral particles. Accumulation of CD4 molecules at the cell surface and packaging into Vpudefective HIV-1 viral particles is shown to have a deleterious effect on viral infectivity as a result of increased formation of gp120-CD4 complexes at the virion surface, thus leading to virion with reduced levels of functional Env glycoproteins. This study provides a functional link between Vpu-mediated CD4 degradation and the HIV-1 infection cycle and pathogenesis, since the positive effect of Vpu on viral infectivity directly correlates with its ability to initiate CD4 degradation.
Much evidence indicates that the effect of Vpu on viral infectivity is intimately correlated with cell surface expression of the CD4 receptor. First, Vpu-defective viruses do not display an impairment of their infectious potential in cells lacking CD4 expression (Fig. 1, C, D, and F). Furthermore, the inhibitory effect of CD4 on Vpu-defective HIV-1 infectivity is dose-dependent and is directly proportional to the levels of cell surface CD4 (as shown in Fig. 1, A-B and E-F). Moreover, we found that a mutant form of CD4 that is not sensitive to Vpu-mediated degradation did inhibit HIV-1 infectivity. Likewise, a Vpu mutant that is unable to degrade CD4 could not restore viral infectivity to wild type levels. These results directly link the effect of Vpu on viral infectivity with the ability of the protein to mediate CD4 degradation (Fig. 1D). Importantly, our results provide evidence that this effect is not limited to cells overex-pressing CD4, such as CD4-transfected 293T cells, since it can be observed in MT4 and Jurkat CD4 ϩ T cell lines as well as in primary PBMCs, thus confirming the in vivo relevance of this phenomenon at CD4 expression levels that accurately reflect the physiological level found on HIV-1-infected host cells. Surprisingly, although the infectivity potential of Vpu-defective virus was significantly reduced (Fig. 2) as compared with Vpu ϩ virus, a delay in the apparition of peak of virus production during replication kinetic studies could not be observed. Indeed, if the role of Vpu is to enhance viral infectivity, one would expect to observe a replication kinetic delay between Vpu Ϫ and Vpu ϩ viruses. The failure of Vpu-defective virus to replicate more slowly than Vpu ϩ virus may be due to the fact that their viral propagation occurs primarily by cell-to-cell contact rather than through free virus in tissue culture systems. Interestingly, a recent study has shown that Vpu-defective virus propagates more efficiently by cell-to-cell transmission than virus expressing a wild-type Vpu (42). This enhanced transmission of HIV-1 Vpu-defective virus by cell to cell transfer might compensate for their infectivity impairment and consequently result in replication kinetics very similar to wild-type HIV-1. Consequently, given that the effect of Vpu on viral infectivity was observed in different experimental settings (transfected cells, T cell lines, and primary cells), it is tempting to speculate that the role of Vpu in HIV-1 infection is to enhance the release of fully infectious progeny virions, thus facilitating not only viral spread within a host but also horizontal transmission.
The degradation of the CD4 receptor by HIV-1 Vpu was thought to enhance the release of Env glycoproteins from the ER and their subsequent maturation, trafficking, and incorporation into nascent viral particles (13). Our data show that the effect of Vpu on viral infectivity correlates with decreased levels of CD4 molecule expressed at the cell surface and packaged into viral particles rather than with increased levels of Env incorporated into virions. In fact, although Vpu ϩ or Vpu Ϫ viruses display differences in their infectivity potential, the levels of gp41 associated with both viruses remain comparable (Figs. 3B and 4B). These results do not rule out the possibility that Vpu-mediated CD4 degradation might enhance the release of the Env precursor from the ER as previously reported (13). However, this efficient release of Env that presumably takes place with Vpu ϩ virus does not translate into more Env glycoproteins being incorporated into nascent viral particles. Moreover, these data indicate that the infectivity impairment FIG. 4. Protein analysis of HIV-1 particles produced from MT4 T cells. HxBH10-vpu ϩ (Vpu ϩ ) or HxBH10-vpu Ϫ (Vpu Ϫ ) were transfected into MT4 CD4 ϩ T cell line. At day 2, virions were pelleted down from cell-free supernatants by ultracentrifugation, normalized by RT activity, and lysed in radioimmune precipitation buffer. A, virion lysates were divided in two equal fractions and immunoprecipitated (IP) with either anti-CD4 monoclonal antibody (␣OKT4) or anti-HIV-1 serum (␣HIV-1). Immunocomplexes were separated by SDS-PAGE, transferred to nitrocellulose, and analyzed by Western blot (WB) using anti-CD4 (Total CD4) or anti-p24 serum (p24) as indicated. B, serial dilutions of virion lysates were separated by SDS-PAGE, transferred to nitrocellulose, and analyzed by Western blot using anti-gp41 monoclonal antibodies or anti-p24 serum. UD, undiluted. observed with Vpu-defective virus does not result from fewer envelope glycoproteins being incorporated into virions.
Our results are in agreement with recent reports, which showed, using virus isogenic for the expression of Nef and/or Vpu, that high levels of CD4 on the cell surface interfere with the production of infectious HIV-1 particles from infected cells (24,25). However, the mechanism they propose to account for this inhibition differs from the one suggested by the present study. They show that in the presence of high levels of surface CD4, released virus incorporates CD4 and displays a substantial reduction of Env glycoprotein levels presumably by a sequestering effect of the viral Env glycoprotein by CD4. Whereas the effect of Nef on Env incorporation was obvious from these studies, the effect of Vpu appeared less pronounced. Using Nef-defective proviral constructs isogenic except for the expression of Vpu, we failed to observe such a block in Env incorporation in Vpudefective virion, suggesting that the mechanisms by which Nef and Vpu contribute to viral infectivity may be distinct but complementary. It has been reported that HIV-1 Env undergoes rapid endocytosis using clathrin adaptor molecules (43). The same endocytic machinery is targeted by Nef to mediate CD4 internalization (44). It is therefore possible that the higher level of virion-associated Env detected in the presence of Nef could be the result of Nef redirecting the endocytic machinery to downmodulate CD4 cell surface expression, thereby diminishing Env endocytosis and allowing more Env incorporation into HIV-1 particles. Experiments aimed at analyzing the specific contribution of Nef and Vpu mediated CD4 down-modulation on Env function are currently under way.
The effect of Vpu on viral infectivity is strongly correlated with the ability of CD4 molecules to bind gp120. Indeed, cell surface expression and incorporation of a mutant form of CD4, unable to bind gp120, do not impair the infectivity of Vpudefective virus (Fig. 3C). Incorporation of CD4 molecules per se into virion does not appear to impair viral infectivity by affecting the architecture of the virus, since pseudotyping of Vpudefective virus with VSV-G glycoproteins fully restores viral infectivity (Fig. 3D). The most obvious model emerging from our data would be that recruitment of CD4 into virion affects viral infectivity by forming complexes with gp120 at the virion surface, thereby saturating CD4 binding sites on gp120 molecules and decreasing the levels of functional Env glycoproteins at the virion surface. Vpu, by diminishing CD4 cell surface expression and packaging into virion, reduces this interference and allows release of viral particles harboring functional Env glycoprotein capable of binding CD4 molecules at the surface of target cells. Finally, although our data clearly indicate that cell surface CD4 down-modulation by Nef and Vpu during HIV-1 infection is required for optimal infectivity, we cannot rule out the possibility that the synergy in the infectious potential that we have observed in the presence of Nef (Fig. 5) is solely due to Nef-mediated CD4 endocytosis, since it has been shown that Nef can contribute to viral infectivity in a CD4-independent manner (45).
In conclusion, we confirm here that one important role of CD4 down-modulation is to permit the release of fully infectious virions, since excessive levels of CD4 molecules at the cell surface and in nascent viral particles has dramatic consequences on viral infectivity. Two HIV-1 proteins, Nef and Vpu, counteract this inhibition by down-modulating the cell surface expression of CD4 and, as a consequence, have synergistic effects on viral infectivity.