Cell Surface CD4 Inhibits HIV-1 Particle Release by Interfering with Vpu Activity*

One of the hallmarks of human immunodeficiency virus type I (HIV-1) infection is the rapid removal of the viral receptor CD4 from the cell surface. This remarkably efficient receptor interference requires the activity of three separate viral proteins: Env, Vpu, and Nef. We have investigated whether this unusually tight interference on cell surface CD4 expression had a more essential function during the viral life cycle than simply preventing superinfection. We now report that the removal of cell surface CD4 is required for optimal virus production by HIV-1. Indeed, maintenance of CD4 surface expression in infected cells lead to a 3–5-fold decrease in viral particle production. This effect was not due to the formation of intracellular complexes between CD4 and the gp160 viral envelope precursor but instead required the presence of CD4 at the cell surface and was specifically mediated by CD4 but not closely related plasma membrane receptors. The finding that CD4 had no significant effect on particle release by a Vpu-deficient variant indicates that CD4 acts by inhibiting the particle release-promoting activity of Vpu. Co-immunoprecipitation experiments further showed that CD4 and Vpu physically interact at the cell surface, suggesting that CD4 might inhibit Vpu activity by disrupting its oligomeric structure.

One of the hallmarks of human immunodeficiency virus type I (HIV-1) infection is the rapid removal of the viral receptor CD4 from the cell surface. This remarkably efficient receptor interference requires the activity of three separate viral proteins: Env, Vpu, and Nef. We have investigated whether this unusually tight interference on cell surface CD4 expression had a more essential function during the viral life cycle than simply preventing superinfection. We now report that the removal of cell surface CD4 is required for optimal virus production by HIV-1. Indeed, maintenance of CD4 surface expression in infected cells lead to a 3-5-fold decrease in viral particle production. This effect was not due to the formation of intracellular complexes between CD4 and the gp160 viral envelope precursor but instead required the presence of CD4 at the cell surface and was specifically mediated by CD4 but not closely related plasma membrane receptors. The finding that CD4 had no significant effect on particle release by a Vpu-deficient variant indicates that CD4 acts by inhibiting the particle release-promoting activity of Vpu. Co-immunoprecipitation experiments further showed that CD4 and Vpu physically interact at the cell surface, suggesting that CD4 might inhibit Vpu activity by disrupting its oligomeric structure.
One of the common features of retroviral infection is the selective removal of the cellular receptor from the cell surface, resulting in resistance to superinfection by viruses using the same receptor (1). The physiological significance of receptor down-regulation during the viral life cycle is still debated. There is in vitro evidence that viruses that fail to successfully inactivate their cellular receptor at the onset of infection display increased cytopathicity in the host. This correlates with massive reinfection leading to the accumulation of unintegrated viral DNA and cell death caused by toxic levels of viral protein expression and formation of syncytia (2)(3)(4)(5)(6). This suggests that the establishment of superinfection immunity is a key event involved in establishing a long term chronic rather than acute lytic infection. However, these data are not fully supported by in vivo evidence. Indeed, although it was reported that Friend leukemia virus can successfully establish a state of superinfection immunity in mice (7), other in vivo studies have shown that cells chronically infected by the human immunodeficiency virus type I (HIV-1) 1 can maintain expression of the CD4 receptor (8,9). This leaves open the question of the exact role and importance of receptor interference in the HIV-1 life cycle. For a majority of retroviruses, receptor interference is a direct consequence of high affinity interactions between the viral envelope glycoprotein and the cellular receptor (reviewed in Ref. 10). Such interactions can take place intracellularly and lead to the sequestration of the cellular receptor in the endoplasmic reticulum (ER) of cells (11). Alternatively, soluble forms of the viral envelope protein secreted from infected cells can interact with the cellular receptor at the cell surface resulting in the masking of the viral binding site (7). HIV-1 is unique among retroviruses in that it uses, aside from the function of receptor-Env complexes, two additional mechanisms to down-regulate cell surface expression of its cellular receptor. Indeed, both the nef and vpu genes encode activities that contribute to the down-modulation of cell surface CD4 (for review, see Ref. 10). Early in the viral life cycle, the Nef protein mediates the accelerated internalization and lysosomal degradation of cell surface CD4 by targeting it to an endocytic pathway involving clathrin-coated pits (for recent reviews, see Refs. 12 and 13). In later phases of the viral life cycle, the gp160 envelope glycoprotein precursor as well as the Vpu protein further contribute to a complete down-regulation of cell surface CD4 (5,14). gp160 is a major factor in CD4 down-modulation that can, in most instances, quantitatively block the bulk of newly synthesized CD4 in the endoplasmic reticulum (15)(16)(17)(18). This strategy has, however, an important shortcoming. Indeed, the formation of CD4-gp160 complexes in the ER blocks the transport and maturation of not only CD4 but of the envelope protein as well (17). Thus, if the levels of CD4 in the ER exceed those of gp160, Env trafficking to the cell surface and incorporation into virions are effectively blocked, resulting in production of noninfectious virus devoid of envelope protein (19,20).
One of the well characterized functions of Vpu is to induce the degradation of CD4 molecules trapped in intracellular complexes with Env, thus releasing gp160 for transport toward the cell surface (21). Vpu-mediated degradation of CD4 requires direct interactions between the cytoplasmic tails of the two molecules (22) and is dependent on the presence of two phosphorylated serine residues at positions 52 and 56 in the Vpu cytoplasmic domain (23). However, the phosphoserine residues do not participate in CD4-Vpu interactions (22). They rather serve to link Vpu-targeted CD4 to the ubiquitin-proteasome degradation machinery by providing a binding site for the E3 ubiquitin ligase complex component ␤TrCP (24). In addition to its destabilizing effect on CD4, Vpu is known to mediate the efficient release of viral particles from HIV-1-infected cells (25,26). This function of Vpu relies on the integrity of its transmembrane domain as well as the ability of Vpu to form homooligomeric complexes in a post-ER compartment (27)(28)(29). In contrast, the cytoplasmic domain of Vpu is necessary and sufficient to mediate CD4 degradation in the ER (22,30,31). Given these mechanistic and structural differences the question arises as to why such apparently unrelated activities have evolved within the Vpu protein.
The present work examines the physiological relevance of receptor down-modulation by HIV-1. By using a model system where cell surface CD4 was maintained despite the interfering activities of HIV-1, we were able to show that cell surface CD4 exerts a profound inhibitory effect on HIV particle production. We further identified the mechanism of CD4 action by showing that it directly interfered with the ability of Vpu to promote viral particle release, presumably by disrupting the oligomeric structure of Vpu. These data provide new insights into the physiological relevance of receptor interference and suggest that the inhibitory activity of CD4 on an essential step of the viral life cycle likely participated in the selection of three separate receptor interference factors in HIV-1. Our data also present for the first time a functional link between the two mechanistically independent activities of Vpu: namely, Vpumediated degradation of CD4 participates in the particle release function of Vpu by reducing the levels of inhibitory cell surface CD4.
pHIV-CD4⌬Bam, expressing full-length CD4; pHIV-CD4/Q421 encoding an ER retention variant of CD4; and pHIV-CD4/Nar bearing a C-terminal 22-residue deletion in the CD4 cytoplasmic tail were described previously (14,22). A CD4-deficient variant of pHIV-CD4⌬Bam, pHIV-CD4(Ϫ), was constructed by digestion with MfeI followed by fill-in of the 5Ј overhangs with the Klenow fragment of DNA polymerase I. Self-ligation of the modified vector introduced a frameshift in the CD4 ORF that truncates the molecule to the 130 N-terminal residues (approximately half of the ectodomain).
pHIV-CD8 expresses the full-length CD8 molecule under the control of the HIV-1 long terminal repeat. The construct was generated by subcloning a 1.5-kilobase EcoRI fragment from pT8F1 (obtained through the AIDS Research and Reference Reagent Program) into the BssHII-XhoI sites of pNL4-3. pHIV-CD8/4 encodes a chimeric molecule bearing the CD8 ectodomain fused to the CD4 transmembrane and cytoplasmic tail. This construct was obtained by cloning a 1968-base pair BssHII-XhoI fragment from the previously described pCMV-CD8/4 plasmid (31) into pHIV-CD8.
pHIV-CD4/8 and pHIV-CD4/3 encode chimeric proteins bearing the CD4 ectodomain and CD8 or CD3 transmembrane and cytoplasmic tail, respectively. The chimeric open reading frame was constructed by using a two-step PCR strategy as follows. In a first step, two PCR fragments overlapping at the chimera junction were generated using pHIV-CD4 and pHIV-CD8 or pXS-CD3␥ (a generous gift from Dr. David L. Wiest, The Fox Chase Cancer Center), respectively. The two PCR fragments were purified, mixed, and allowed to hybridize in a second round of PCR designed to amplify the joined fragments. The product of the second round of PCR was digested with MfeI and AgeI restriction enzymes (New England Biolabs, Beverly, MA) and cloned into the corresponding sites in pHIV-CD4⌬Bam.
All pHIV constructs are under the transcriptional control of the HIV-1 long terminal repeat and therefore require the presence of the HIV-1 Tat transactivator. This conditional expression system ensures that in co-transfection experiments CD4 and CD4-derived chimeras are expressed only in cells also expressing HIV-1 proteins.
Cells and Transfections-HeLa-T4 cells were maintained in Dulbecco's modified Eagle's medium/5% fetal bovine serum supplemented with 1 mg/ml G418. HeLa or HeLa-T4 cells were transfected by the calcium phosphate method as described (34). Briefly, 2 ϫ 10 6 HeLa cells were incubated for 4 h with 20 -30 g of calcium phosphate-precipitated plasmid DNA, washed in phosphate-buffered saline, and subjected to a 2.5-min glycerol shock. Cells were maintained in 5 ml of Dulbecco's modified Eagle's medium/fetal calf serum for 16 h prior to metabolic labeling or lysate preparation.
Antibodies-The TP human serum reacts against all major HIV-1 proteins and was obtained from an asymptomatic HIV-seropositive patient. A rabbit polyclonal antiserum directed against the cytoplasmic tail of Vpu (U2-3) was used for Western blot hybridization (35). The T4-4 rabbit polyclonal antibody, directed against the ectodomain of CD4, was obtained from the AIDS Research and Reference Reagent Program and was originally contributed by Dr. R. Sweet (36). The T4-Cy polyclonal antibody is directed against the cytoplasmic tail of the CD4 molecule and was generated by immunization of rabbits with a synthetic peptide representing residues 394 -422 of CD4 coupled to keyhole limpet hemocyanin.
Pulse-Chase and Immunoprecipitation-Measurement of HIV viral particle release efficiency was performed as described previously (33). Briefly, cells were pulse labeled for 30 min with 1 mCi/ml [ 35 S]methionine and chased at 37°C for up to 4 h. Aliquots of the cells and culture medium cleared of cell debris harvested at each time point were lysed by freeze-thaw in Nonidet P-40-DOC lysis buffer consisting of 10 mM Tris-HCl, pH 7.6, 150 mM NaCl, 20 mM EDTA, 1% Nonidet P-40, and 0.5% deoxycholate (DOC). Viral proteins were recovered by immunoprecipitation with TP serum and protein A-agarose beads and separated on 12.5% polyacrylamide gels.
HIV Gag proteins detected in the cell and supernatant fractions were quantified with a Fuji FLA2000 phosphoimager, and the particle release efficiency at each time point was calculated as the ratio of Gag proteins in the medium fraction over the total amount of Gag proteins present in the cell and medium fractions.
Cell Surface Biotinylation-Cells were washed with 5 ml of ice-cold phosphate-buffered saline and surface biotinylated with 1 ml of Sulfo-NHS Biotin (Pierce) at 0.25 mg/ml in phosphate-buffered saline for 30 min at 4°C. Cells were washed twice in phosphate-buffered saline and lysed by freeze-thaw in either Nonidet P-40-DOC buffer for analysis of CD4 surface expression or digitonin lysis buffer (10 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.5% digitonin) (22) for analysis of CD4-Vpu interactions. Biotinylated proteins were recovered from the lysate by immunoprecipitation with Neutravidin-agarose beads (Pierce).
Western Blotting-Proteins separated by SDS-polyacrylamide gel electrophoresis were transferred to Immobilon membranes, blocked with 5% skim milk, and probed with the indicated antisera at 1:1500 dilution. Proteins were revealed by chemoluminescence (ECL, Amersham Pharmacia Biotech) using a 1:4000 dilution of an horseradish peroxidase-conjugated anti-rabbit antiserum (Amersham Pharmacia Biotech). Bands were quantified on a Macintosh computer using the public domain NIH Image program (developed at the U. S. National Institutes of Health).

Viral Particle Release Is Impaired in HeLa-T4 Cells-We
first tested the effect of CD4 on viral particle release in HeLa-T4 cells expressing the human CD4 in a stable and constitutive manner (37). HeLa and HeLa-T4 cells were transfected with plasmid expressing either the NL4-3 molecular clone or its Vpu-deficient counterpart, NL4-3/Udel. 24 h posttransfection, the cell and supernatant fractions were separated, and viral particles secreted in the culture medium were recovered by centrifugation. Viral lysates obtained from each fraction were loaded onto polyacrylamide gels, transferred to nitrocellulose, and probed in Western blot using an HIV-positive human serum (data not shown). Viral proteins corresponding to the pr55 gag and p24 capsid components were quantified, and the efficiency of particle release was calculated as described under "Experimental Procedures." The result of a representative experiment shows that, in normal HeLa cells, the presence of Vpu enhanced viral particle release more than 3-fold (Fig. 1). In contrast, the rate of particle release of wildtype NL4-3 in HeLa-T4 cells was dramatically reduced and comparable with that observed for the Vpu-deficient (NL4-3/ Udel) virus in HeLa cells, suggesting that CD4 exerts a negative effect on viral production. In addition, the particle release activity of the Vpu-deficient NL4-3/Udel was similar in HeLa and HeLa-T4 cells (Fig. 1), suggesting that the inhibitory effect of CD4 likely targets the particle release activity of Vpu.

CD4 Inhibits the Particle Release Activity of Vpu in HeLa
Cells-To determine whether the low level of virus release observed in HeLa-T4 cells was indeed the result of impaired Vpu activity, we performed a detailed pulse/chase analysis of viral particle release. For that purpose, HeLa cells transfected with pNL4-3 in the presence or absence of the CD4 expression vector pHIV-CD4⌬Bam were labeled for 30 min and chased for up to 4 h. Aliquots of the cell and virus fractions harvested at the indicated chase times were immunoprecipitated with an HIV-positive patient serum and separated on polyacrylamide gels ( Fig. 2A). As a control, virus release by the Vpu-deficient pNL4-3/Udel variant was determined in parallel. As previously reported, the absence of Vpu led to a significant reduction in the amounts of virus-associated p24 capsid protein released in the supernatant over time ( Fig. 2A, compare NL4-3 and NL4-3/Udel). This was accompanied by a characteristic intracellular accumulation of Gag proteins but no apparent effect on the kinetics of the pr55 gag and gp160 Env precursor cleavage. In the presence of CD4, the amount of p24 in the supernatant was reduced as compared with NL4-3 alone, although no differences in the kinetics of Gag maturation were observed ( Fig. 2A,  NL4-3 ϩ CD4). This correlated with accumulation of p24 in the cell fraction, similar to that observed for NL4-3/Udel. Bands corresponding to pr55 gag and p24 in Fig. 2A were quantified, and particle release efficiency was calculated as described under "Experimental Procedures" and plotted as a function of the chase time (Fig. 2B). In agreement with previous studies (33), more than 40% of Gag proteins synthesized by NL4-3 were released as viral particles within the 4 h of chase (Fig. 2B,  NL4-3). In the presence of CD4, however, the particle release of NL4-3 dropped 4-fold (Fig. 2B, NL4-3 ϩ CD4) to levels similar to that of the Vpu-deficient variant (Fig. 2B, NL4-3

/Udel).
Together with the fact that CD4 did not perturb the maturation kinetics of Gag proteins ( Fig. 2A), these data suggest that CD4 directly or indirectly interferes with the activity of Vpu. Moreover, CD4 had a similar negative effect on particle release in HeLa cells whether transiently expressed by transfection (Fig.  2) or stably expressed (Fig. 1). This indicates that the effect observed in HeLa-T4 cells was due to CD4 rather than cellular factors specific to that cell line. Interestingly, no mature gp120 envelope protein was detected in the presence of CD4 (Fig. 2A,  compare NL4-3 and NL4-3 ϩ CD4). This confirms previous data showing that formation of intracellular complexes between CD4 and gp160 blocks the latter in the ER and therefore prevents access of gp160 to the Golgi compartment where it is normally cleaved into the mature gp120 and gp41 components (17,19). The fact that gp160 maturation was quantitatively impaired indicates that CD4 was expressed at a molar excess relative to gp160, and it is thus likely that, under these experimental conditions, a fraction of CD4 was transported to the cell surface.
The Inhibitory Effect of CD4 on HIV-1 Particle Release Correlates with Its Presence at the Cell Surface-To assess whether the inhibitory effect of CD4 on virus release correlates with its presence at the cell surface, HeLa cells were co-transfected with pNL4-3 and increasing amounts of pHIV-CD4⌬Bam, as indicated in Fig. 3. The total amount of CD4 as well as the fraction present at the cell surface were assayed on half of the transfected cells by Western blot (Fig. 3A). Cells were surface Env precursor, gp120 mature Env SU subunit, the p55 Gag precursor, the p24 mature Gag core protein, and the Vpu protein. SN, supernatant. B, HIV Gag proteins in A were quantified. Viral particle release was calculated as described for Fig. 1 and expressed as a function of chase time for each sample. biotinylated prior to lysis in Nonidet P-40-DOC buffer. A small fraction (10%) of each lysate was then removed for direct determination of total CD4. The remainder of each lysate was reacted with Neutravidin-agarose beads to isolate cell surfaceassociated biotinylated species. Both fractions were subsequently separated on acrylamide gels, transferred to nitrocellulose, and probed in Western blot with a mixture of the T4-4 and T4-Cy anti-CD4 polyclonal antibodies. Significant amounts of total CD4 were detected with as little as 2 g of transfected CD4 plasmid DNA. However, under the experimental conditions used here, i.e. in the presence of Env, Vpu, and Nef, at least 8 g of pHIV-CD4 were required to detect CD4 at the cell surface (Fig. 3A). Further increasing the amount of transfected plasmid DNA led to a proportional increase in CD4 cell surface expression (Fig. 3A).
The second half of transfected cells was used to determine the efficiency of NL4-3 particle release by pulse-chase. Cells were labeled for 30 min and chased for 4 h. At both time points, equal aliquots of the cells were lysed, and the virus secreted in the medium was recovered by centrifugation. Radiolabeled viral proteins were immunoprecipitated with TP serum and separated on polyacrylamide gels (data not shown). Bands corresponding to Gag proteins were quantified, and the particle release efficiency was calculated at the 4-h time point and plotted in Fig. 3B. The data show a linear correlation between the amounts of CD4 at the cell surface and the decrease in viral particle production (Fig. 3B). Interestingly, in the presence of 15 g of transfected pHIV-CD4 plasmid, the particle release efficiency of wild-type NL4-3 was similar to that observed for the Vpu-deficient variant (compare Fig. 1, NL-43/Udel, and   Fig. 3B). This indicates that at these levels, cell surface CD4 is able to fully inhibit the activity of Vpu on particle release. This phenomenon required the presence of CD4 at the cell surface because no inhibition of particle release was observed upon transfection of 4 g of pHIV-CD4, despite the presence of significant amounts of total CD4. Fig. 3 reveal an intriguing correlation between the levels of cell surface CD4 and the degree of inhibition of virus release. This suggests that the subcellular localization of CD4 rather than its steady-state level is relevant to this mechanism. To test this hypothesis, we examined the effect on virus release of a CD4 mutant (CD4/Q421) bearing an ER retention signal. The experiment was performed as described for Fig. 3, and the levels of total and cell surface CD4 were determined by Western blot. As shown in Fig. 4A, intracellular expression levels of CD4 and CD4/Q421 were similar when equivalent amounts of plasmid were transferred. In contrast, cell surface expression of the CD4/Q421 mutant was markedly impaired relative to wild-type CD4. Despite the high intracellular levels of CD4/Q421, the effect of the retention mutant on NL4-3 particle release was minimal, indicating that CD4 exerts its inhibitory action from a post-ER compartment (Fig. 4B). These data also indicate that the effect of CD4 is not due to nonspecific toxicity generated by overexpression of CD4 in cells. At the highest concentrations of CD4/Q421 a moderate 25% inhibition of particle release was observed (Fig. 4B), because of a small fraction of CD4/Q421 escaping the ER retention (Fig. 4A).

CD4 Molecules Trapped in the ER Have No Effect on Viral Particle Release-Results from
CD4 Inhibition of Particle Release Is Independent of the HIV-1 Env-The inhibition of gp160 cleavage concomitant with decreased virus production observed in Fig. 2 suggests that intracellular interactions between CD4 and gp160 might play a  Fig. 2. HIV Gag proteins were quantified, and viral particle release calculated as described in the legend to Fig. 1 was plotted on the left ordinate. Cell surface CD4 detected in A was quantified and plotted on the right ordinate as the total pixel value of the band. One-tenth of the lysates was directly loaded on the gel (Total CD4), and the rest was immunoprecipitated with Neutravidin-agarose (Cell-surface CD4). CD4 was detected by Western blotting using anti-CD4 polyclonal antibodies. B, cells were pulse-labeled for 30 min in the presence of [ 35 S]methionine and chased for 4 h. Viral particle release and levels of cell surface CD4 were calculated as described in the legend to Fig. 1. role in the negative effect of CD4 on particle release. We therefore addressed the role of the HIV-1 envelope protein in this process. The effect of CD4 on particle release by an Env-deficient variant of NL4-3, NL43-K1, was assessed in a CD4 doseresponse experiment similar to that detailed in Fig. 3. Cell surface CD4 detected by surface biotinylation appeared at significantly lower concentrations of pHIV-CD4⌬Bam plasmid as was observed in the case of NL4-3 (compare Figs. 3A and 5A). This was expected because the envelope protein is a major component of CD4 down-regulation and is absent in NL43-K1. Accordingly, inhibition of viral particle release was observed at lower concentrations of CD4 with 24% inhibition obtained with as little as 2 g of transfected pHIV-CD4 DNA. In the presence of 12 g of CD4 plasmid, the inhibition reached a maximal value of 69% of the NL4-3 control, a level similar to that of a Vpu(Ϫ) virus (compare Fig. 1, NL4-3/Udel, and Fig. 5). Further increasing the amount of CD4 present at the cell surface had little additional effect on particle release, indicating that once Vpu activity is fully inhibited, no other mechanism is in place to further reduce particle release (Fig. 5, 15 g of pHIV-CD4). These data indicate that the HIV-1 envelope glycoprotein is not required for the inhibitory effect CD4 exerts on virus production by HIV-1. On the contrary, the removal of Env allows more CD4 to reach the cell surface, leading to more pronounced inhibitory effect.
The CD4 Cytoplasmic Tail Is Required for Inhibition of Particle Release-To further demonstrate that the ability of CD4 to inhibit virion production is related to direct targeting of Vpu, chimeric proteins bearing parts of the wild-type CD4 were tested for their ability to interfere with virus release. Previous reports have shown that the cytoplasmic tail and possibly the transmembrane domain of CD4 are required for interaction with Vpu (22, 38, 39). The particle release efficiency of NL43-K1 was examined in the presence of either wild-type CD4, a Vpu-interacting chimera where the CD4 ectodomain was replaced by that of CD8 (CD8/4) or two chimeras bearing the CD4 ectodomain fused to the transmembrane and cytoplasmic tail of CD8 or CD3 (CD4/8 and CD4/3, respectively). All chimeras were efficiently expressed at the cell surface (data not shown). As shown in Fig. 6, neither the CD4/8 nor the CD4/3 chimera, both of which are unable to interact with Vpu (data not shown), affected the particle release of NL43-K1. In contrast, wild-type CD4, and to a similar extent CD8/4, reduced particle release efficiency to the level observed in the case of a Vpu(Ϫ) virus. Taken together, these data indicate that the inhibitory effect of CD4 on particle release is specific and that the transmembrane and cytoplasmic tail of CD4 are necessary and sufficient for inhibition of Vpu-mediated enhancement of viral particle release.
CD4 and Vpu Interact at the Cell Surface-The requirement for the transmembrane and cytoplasmic tail of CD4 in inhibiting Vpu-mediated particle release activity from the cell surface suggests that direct interactions between the two proteins might be involved. To directly demonstrate the presence of CD4-Vpu complexes at the cell surface, co-immunoprecipitation experiments were performed. HeLa cells were transfected with pHIV plasmids encoding wild-type CD4, the ER retention mutant CD4/Q421, or the cytoplasmic tail truncation mutant CD4/Nar. Wild-type or nonphosphorylated Vpu was provided by co-transfection of pNL-A1 or the pNL-A1/U26 plasmid, respectively. The pNL-A1/Udel plasmid was used as a Vpu(Ϫ) control. Cells were surface biotinylated and lysed in buffer containing the mild detergent digitonin to preserve the integrity of CD4-Vpu complexes. The lysates were split in three fractions. To assess CD4-Vpu complexes in the whole cell lysate, 25% of each lysate was subjected to direct immunoprecipitation with the anti-CD4 T4-4 antibody. The immunoprecipitates were separated on a polyacrylamide gel and probed in Western blot with the anti-Vpu antibody to detect co-immunoprecipitated Vpu (Fig. 7, Anti-CD4). As previously reported (22) and confirmed in this experiment, CD4 forms intracellular complexes with both wild-type and nonphosphorylated Vpu. Also in agreement with previous studies (22), the CD4/Q421 but not the CD4/Nar mutant retained the ability to interact with Vpu (Fig. 7, Anti-CD4). Another 25% of each cell lysate was subjected to immunoprecipitation with anti-Vpu followed by Western blot analysis with the same antibody. As can be seen in Fig. 7 (Anti-Vpu), Vpu was present at comparable levels in all extracts except Vpu/del. To demonstrate cell surface complexes of Vpu and CD4, the third fraction (50%) of the cell extracts was first subjected to Neutravidin-agarose immuno-  Fig. 3. B, pulse-chase was performed as indicated in Fig. 2. Viral particle release was plotted on the left ordinate and expressed as the percentage of that of NL4-3/K1 in the absence of CD4. Cell surface CD4 quantified from A is shown on the right ordinate. HIV Gag proteins were quantified, and the particle release efficiency of each sample was calculated as described in the legend to Fig. 1. precipitation for the isolation of cell surface biotinylated molecules, including CD4. The immunoprecipitates were then probed in Western blot with the Vpu-specific antibody (Fig. 7, Neutravidin). Because Vpu lacks an extracellular domain, we did not expect Vpu to be accessible to cell surface biotinylation. Therefore, the detection of Vpu on the blot was indicative of co-immunoprecipitation with a biotinylated molecule. Interestingly, both Vpu and Vpu/26 were co-immunoprecipitated by cell surface CD4, indicating that stable interactions exist between CD4 and both forms of Vpu at the cell surface. The specificity of the assay for cell surface interactions was confirmed by the fact that the CD4/Q421, which is capable of interacting with Vpu in the ER, failed to co-immunoprecipitate Vpu following neutravidin immunoprecipitation. In addition, the CD4/Nar protein, which is expressed at the cell surface in large amounts (data not shown) but is incapable of interacting with Vpu, also failed to show cell surface co-immunoprecipitation of Vpu. The later control also confirms that Vpu itself, because of the lack of an ectodomain, was not biotinylated. Taken together, these data establish a clear correlation between the ability of CD4 to interact with Vpu at the cell surface and its inhibition of viral particle release. DISCUSSION Receptor interference, the process by which a cellular receptor is rendered inaccessible for superinfection is a hallmark of retroviral infections (1). For most retroviruses, it is the direct result of high affinity interactions, in the endoplasmic reticulum or at the cell surface, between the cellular receptor and the viral envelope glycoprotein (40). Although the prevention of superinfection has been linked to maintenance of latent infections and reduced viral cytotoxicity (2)(3)(4)(41)(42)(43), the question remains as to whether this process is a key step in the viral life cycle or simply the unavoidable consequence of the high affinity between the cellular receptor and the viral Env protein. In the case of HIV-1, however, two novel activities provided by the Vpu and Nef proteins appear to have been specifically selected for the purpose of down-regulation of the CD4 receptor. We therefore asked whether interference of CD4 expression during HIV-1 infection served a more essential function than prevention of superinfection. We now report that expression of CD4 severely impairs viral production and that CD4 down-regulation by Env, Vpu, and Nef serves to alleviate this inhibitory effect. Accordingly, we found that the negative activity of CD4 on viral particle release correlated with its presence at the cell surface and that an ER retention mutant of CD4 had no such effect. Moreover, the transmembrane and cytoplasmic domains of CD4 played a key role in the mechanism of inhibition because chimeric proteins bearing the corresponding domains of CD8 or CD3 had no effect on particle release. These data, together with the observation that Vpu-deficient viruses are immune to the negative effect of CD4, indicate that CD4 acts directly on Vpu by inhibiting its particle release-promoting activity. Vpu has been shown both in vitro and in virus-producing cells to form homo-oligomeric complexes consisting of 4 -6 subunits (35). There is evidence to suggest that formation of such oligomeric structures is required for Vpu-mediated enhancement of viral particle release. For instance, we and others have previously shown that Vpu has the propensity to form ion conductive membrane pores (44 -47). In addition, there is a clear correlation between this property of Vpu and its ability to enhance virus release (47). This is consistent with the notion that the N-terminal ␣-helix of Vpu, which constitutes the membrane anchor, contains the major determinants for activity on particle release (27,29,48). It is thus likely that, by virtue of its affinity for Vpu, CD4 could insert itself into the Vpu oligomers leading to either dissociation of their structure or disruption of their ion channel activity. Our ability to demonstrate direct interactions between CD4 and Vpu at the cell surface strongly supports such a proposed mechanism. Furthermore, data indicating that CD4 and Vpu interact through both their transmembrane and cytoplasmic domains reinforce the notion that CD4 could successfully compete with Vpu-Vpu interactions (38,39).
Although similar in its concept, the data presented in this paper contrast with a recent study proposing that cell surface CD4 decreases viral production by interacting with the HIV-1 envelope protein (49). Accordingly, cell surface interactions between CD4 and Env were proposed to be responsible for blocking the budding of newly formed virions. However, our data obtained with the Env-deficient NL43-K1 variant of HIV-1 suggest that Env is largely dispensable for CD4-mediated inhibition of particle release (Fig. 4). In fact, rather than contributing to the negative effect of CD4, the presence of Env in our system reduces the effect of CD4 on viral release, presumably by trapping CD4 in the ER. The mechanism proposed by Ross et al. (49) requires Env to be present at the cell surface together with CD4 to inhibit particle release. However, previous reports showed that CD4-gp160 complexes formed in the ER are blocked in their transport to the cell surface. In instances where the steady state levels of Env exceed that of CD4, CD4 is quantitatively trapped in the ER (15)(16)(17). Under such conditions, only Env is transported to the cell surface. Conversely, if CD4 levels are higher than Env, gp160 is blocked in the ER and CD4 is expressed at the cell surface (19,50). This phenomenon is exemplified in Fig. 2 where expression of excess CD4 led to complete inhibition of gp160 cleavage. Our data rather support a model in which CD4 decreases viral particle release by interfering with the activity of Vpu at the cell surface in an Env-independent fashion. These results provide the first evolutionary link between the two functionally distinct activities of Vpu. Indeed, by promoting CD4 degradation, Vpu would contribute to alleviate the inhibitory effect of CD4 on its particle release activity. Although mechanistically distinct, both biological activities of Vpu thus appear to contribute to the enhancement of viral particle production.
Our model system focused on demonstrating the importance of receptor down-regulation beyond the simple prevention of superinfection. By successfully defeating the mechanisms put in place by HIV-1 to down-regulate CD4, we showed that receptor down-regulation is essential for one of the most important steps of the viral life cycle: production of progeny virus. HIV-1 isolates that express the full complement of CD4 downregulation activities, i.e. Env, Vpu, and Nef, are likely to efficiently remove cell surface CD4 and therefore allow for full particle release activity of Vpu (51). However, this delicate FIG. 7. HeLa cells were transfected with 10 g of pHIV-CD4⌬Bam (CD4) or pHIV-CD4/Q421 (CD4/Q421) or pHIV-CD4/ Nar (CD4/Nar). Wild-type Vpu or the Vpu mutants indicated on the figure were expressed by co-transfection of 20 g of pNL-K1 (Vpu), pNL-K1/Udel (Vpu/del), or pNL-K1/U26 (Vpu/26). Following cell surface biotinylation, digitonin lysates were separated into three aliquots and subjected to immunoprecipitation as indicated (Immunoprecipitation). Immunoprecipitates were separated on acrylamide gel, transferred to nitrocellulose, and probed in Western blot with the anti-Vpu U2-3 antibody. balance could be perturbed by events such as mutations that decrease the affinity between CD4 and Env or inactivate the nef gene. Although not fully explained, the presence of Nefdeficient viruses has been correlated with lower virus loads in long term survivors of HIV-1 infection (52). Whether this phenomenon is related to the inhibition of viral release by increased amounts of cell surface CD4 is presently not known. However, there is experimental evidence that high levels of CD4 can effectively lead to abortive HIV-1 infection (53). This in turn suggests possible avenues for therapeutic intervention of HIV-1 infection. Indeed, stimulation of CD4 expression levels, by treatment with T-cell activating cytokines such as interleukin-2 for instance, could contribute to a reduction in progeny virus production.