Cell surface CD4 interferes with the infectivity of HIV-1 particles released from T cells.

The CD4 protein is required for the entry of human immunodeficiency virus (HIV) into target cells. Upon expression of the viral genome, three HIV-1 gene products participate in the removal of the primary viral receptor from the cell surface. To investigate the role of surface-CD4 in HIV replication, we have created a set of Jurkat cell lines which constitutively express surface levels of CD4 comparable to those found in peripheral blood lymphocytes and monocytes. Expression of low levels of CD4 on the surface of producer cells exerted an inhibitory effect on the infectivity of HIV-1 particles, whereas no differences in the amount of cell-free p24 antigen were observed. Higher levels of cell surface CD4 exerted a stronger inhibitory effect on infectivity, and also affected the release of free virus in experiments where the viral genomes were delivered by electrotransfection. The CD4-mediated inhibition of HIV-1 infectivity was not observed in experiments where the vesicular stomatitis virus G protein was used to pseudotype viruses, suggesting that an interaction between CD4 and gp120 is required for interference. In contrast, inhibition of particle release by high levels of cell-surface CD4 was not overcome by pseudotyping HIV-1 with foreign envelope proteins. Protein analysis of viral particles released from HIV-infected Jurkat-T cells revealed a CD4-dependent reduction in the incorporation of gp120. These results demonstrate that physiological levels of cell-surface CD4 interfere with HIV-1 replication in T cells by a mechanism that inhibits envelope incorporation into viral membranes, and therefore provide an explanation for the need to down-modulate the viral receptor in infected cells. Our findings have important implications for the spread of HIV in vivo and suggest that the CD4 down-modulation function may be an alternative target for therapeutic intervention.

The CD4 protein is required for the entry of human immunodeficiency virus (HIV) into target cells. Upon expression of the viral genome, three HIV-1 gene products participate in the removal of the primary viral receptor from the cell surface. To investigate the role of surface-CD4 in HIV replication, we have created a set of Jurkat cell lines which constitutively express surface levels of CD4 comparable to those found in peripheral blood lymphocytes and monocytes. Expression of low levels of CD4 on the surface of producer cells exerted an inhibitory effect on the infectivity of HIV-1 particles, whereas no differences in the amount of cell-free p24 antigen were observed. Higher levels of cell surface CD4 exerted a stronger inhibitory effect on infectivity, and also affected the release of free virus in experiments where the viral genomes were delivered by electrotransfection. The CD4-mediated inhibition of HIV-1 infectivity was not observed in experiments where the vesicular stomatitis virus G protein was used to pseudotype viruses, suggesting that an interaction between CD4 and gp120 is required for interference. In contrast, inhibition of particle release by high levels of cell-surface CD4 was not overcome by pseudotyping HIV-1 with foreign envelope proteins. Protein analysis of viral particles released from HIV-infected Jurkat-T cells revealed a CD4dependent reduction in the incorporation of gp120. These results demonstrate that physiological levels of cell-surface CD4 interfere with HIV-1 replication in T cells by a mechanism that inhibits envelope incorporation into viral membranes, and therefore provide an explanation for the need to down-modulate the viral receptor in infected cells. Our findings have important implications for the spread of HIV in vivo and suggest that the CD4 down-modulation function may be an alternative target for therapeutic intervention.
HIV 1 down-modulates its own receptor, the CD4 protein.
Removal of CD4 from the surface of infected cells is achieved by the Nef, Vpu, and Env products (1)(2)(3)(4)(5)(6)(7)(8). Nef acts early after infection by accelerating the internalization of CD4 and targeting it to late endosomes for degradation (6, 9 -13). Env and Vpu cooperate inside the cell to block the transport of CD4 to the cell surface and redirect this protein to the ubiquitin-proteosome machinery for degradation (1, 14 -16). The Env protein sequesters CD4 in the endoplasmic reticulum by an interaction mediated by the extracellular domains of CD4 and Env (8,17), whereas Vpu induces the degradation of the HIV receptor by a mechanism which requires an interaction between the cytoplasmic tails of CD4 and Vpu (1,7,18,19). The fact that the combined action of three HIV-1 gene products is required for the nearly complete elimination of CD4 from the cell surface encouraged investigators to examine why HIV has to reduce the cell surface expression of its own receptor. By analogy with other retroviruses, it was first speculated that removal of CD4 from the cell surface would impede superinfection of cells, an event that might augment cytotoxic effects and lead to cell death before productive infection and release of viral particles occurs (20,21). Although down-modulation of the viral receptor as a way to impede superinfection has been demonstrated in some retroviruses (21,22), the short half-life of HIV-infected PBLs would suggest that impeding superinfection confers no physiological advantage to the virus in the periphery. Two recent reports have proposed other explanations for the need to down-modulate CD4 expression. Overexpression of CD4 in HIV-transfected 293T cells leads to the reduction in infectivity (23) and release (24) of HIV-1 particles. Although the precise mechanism of action remains to be elucidated, both groups showed evidence that specific interactions between the CD4 and the viral envelope proteins are required for the inhibitory action of CD4. Pseudotyping of HIV-1 particles with foreign envelope proteins (VSVg or MLV) eliminates the negative effects of CD4. Furthermore, mutant CD4 proteins unable to bind to the envelope protein fail to interfere with HIV replication (23,24). These experiments were performed in human embryonic kidney cells (293T) which do not resemble the cellular environment found in infection of CD4-positive primary T cells and macrophages. Furthermore, transfected 293T cells may not reflect the physiological levels of CD4 found in the natural targets of infection. For instance, Ross et al. (24) used 293T cells expressing surface CD4 levels 12-fold higher than those found on SupT1 cells, a human T cell line with high levels of surface CD4. Therefore, several issues were raised as to whether physiological levels of surface CD4 would exert a similar effect in T cells (25). To investigate the relevance of the CD4-mediated inhibitory effect, we have extended our studies to transformed T cells constitutively expressing specific levels of surface CD4. Our findings demonstrate that physiological levels of surface CD4 interfere in an envelope-dependent manner with the infectivity of HIV-1 particles released from T cells.

EXPERIMENTAL PROCEDURES
Cell Lines-The parental Jurkat-T cell line used in this study is a CD4-negative derivative of the Jurkat lymphoblastoid cell line (26). Jurkat-T cells express the large T antigen protein from SV40, whose gene is maintained with G418 selection. These cells allow high expression of genes driven from plasmids containing cytomegalovirus promoters. Jurkat-T cells were kindly provided by Dr. N. Coudronniere (La Jolla Institute for Allergy and Immunology, San Diego). To construct CD4-positive cells, Jurkat-T cells were transfected with a full-length CD4 expression plasmid and selected with zeocin (0.1 mg/ml) in RPMI medium also containing penicillin, streptomycin, glutamine, Hepes, and 10% fetal calf serum. Two weeks after selection, cells were stained with a phycoerythrin-conjugated CD4 mAb and sorted with a Becton-Dickinson flow cytometry apparatus for high and low surface expression of CD4. After sorting, cells were selected for another week in zeocin (0.1 mg/ml) and G418 (0.2 mg/ml), and frozen in liquid nitrogen until usage. Before use, cells were thawed and grown in the presence of G418 (0.2 mg/ml) for 3-4 days. High-CD4 cells were further purified with CD4 microbeads as specified by the manufacturer (Miltenyi Biotec Inc.). CD4-negative cells were also incubated with CD4 microbeads and purified in the same way, but this time the unbound fraction was collected. Cells were maintained in 0.2 mg/ml G418 for 2-3 additional days before electroporation. Levels of surface CD4 were always monitored before each experiment. CD4 surface levels were maintained constant for at least 10 days after enrichment with microbeads.
Transfections, Virus Preparations, and Infections-Stocks of viruses were produced by chemical transfection of the human kidney fibroblastic cell line 293T, or by electroporation of Jurkat-T cells. Jurkat-T cells were electroporated in 2-mm gap cuvettes with a BTX electroporation system set to deliver 20-ms pulses at 150 volts. Twenty-four to thirtysix hours after electroporation supernatant fluids were collected, filtered through 0.45-m units, and used immediately for infectivity assays. The amount of p24 antigen in supernatant fluids was used to normalize samples containing virus. The levels of p24 antigen were estimated with an enzyme-linked immunoassay (PerkinElmer Life Sciences). Infectivity assays were performed in either CD4-positive Jurkat-T cells, CEM-GFP, or HeLa P4 reporter cell lines. CEM-GFP is a T-lymphoid cell line with a stably integrated green fluorescent protein gene (GFP) whose expression is driven by the HIV-1 long terminal repeat promoter (27), whereas P4 are HeLa-derived CD4-positive cells expressing the ␤-galactosidase gene under the control of the HIV promoter (28). Unless specified, infection of Jurkat-T and CEM-GFP cells were performed following a previously described centrifugation method (29,30). Briefly, 0.5 ϫ 10 6 cells were incubated with viral supernatants (0.5-1 g of p24 protein) in 24-well plates in the presence of 4 g/ml Polybrene and 10 mM Hepes, and centrifuged at room temperature for 90 min (2,500 rpm) in a table-top centrifuge (Sorvall RT6000B). After centrifugation cells were washed and incubated at 37°C in RPMI supplemented with 10% fetal calf serum. Infection of HeLa P4 cells was performed in the presence of 20 g/ml DEAE dextran. Routinely, after infection of reporter cells 5 M azidothymidine was added to block new cycles of replication.
Plasmids-To collect HIV particles, Jurkat-T cells were electroporated with the proviral construct NL4.3-GFP (kindly provided by Dr. Cohen, Harvard University), or infected with R7 HXB2 viruses (viral stocks produced from transfected 293T cells). NL4.3-GFP is a replication-competent vector which contains all the HIV proteins including Env and Nef, together with a GFP reporter gene used to monitor infected cells. Nef-deleted versions (NL4.3⌬Nef-GFP or R7 HXB2 ⌬Nef) were used to study the effect of CD4 down-regulation. In some experiments, additional pseudotyping proteins were provided by co-transfection with the vectors pMDG (31) (encoding VSVg) or pMLV (encoding the amphotropic MLV envelope protein) (32). Expression of CD4 in 293T cells was achieved with the pCMX-CD4 plasmid (23).
Flow Cytometry, Antibodies, and Recombinant Proteins-Labeled cells were routinely fixed with 1% paraformaldehyde after staining and analyzed in a FACScalibur system (Becton-Dickinson) running with the CellQuest software program. CD4 surface expression was analyzed by staining with the OKT4 mAb, followed by goat anti-mouse Cy5-conjugated antibodies (CalTag Laboratories). The epitope recognized by the OKT4 mAb does not overlap with the gp120-binding domain (33). OKT4 was routinely purified in our laboratory from the precipitates of ascites fluids saturated with ammonium sulfate. Intracellular staining of p24 was performed with the IntraStain kit following the recommendations of the manufacturer (DAKO) using a p24-specific mAb (KAL-1, from DAKO) followed by fluorescein isothiocyanate-conjugated goat antimouse IgG (DAKO).
Protein Analyses-To examine virion-associated proteins, particles were pelleted through a 20% sucrose cushion at 26,000 rpm in a SW28 rotor (Beckman) for 1.5 h at 4°C. The liquid was aspirated and the walls of the tube wiped before resuspending the pellet. Pellets were resuspended in 1 ml of phosphate-buffered saline and remaining sucrose was washed by centrifugation at 14,000 rpm (4°C for 90 min) in a table-top Eppendorf Microfuge. These pellets were resuspended in 100 -200 l of phosphate-buffered saline. Equal amounts of p24 protein were separated in 10% SDS-polyacrylamide gels, transferred to polyvinylidene difluoride membranes, probed with specific antibodies, and then revealed with horseradish peroxidase-conjugated secondary antibodies (Amersham Bioscience, Inc.). HIV-1-specific antibodies were obtained from the NIH AIDS Research and Reference Program (NIH-ARRP). Envelope incorporation was estimated with a gp120-specific sheep antiserum, whereas p24 levels were analyzed with the 183-H12-5C mAb. Western blot analysis of CD4 was performed with the T4-4 antiserum (NIH-ARRP).

Construction of T Cell Lines Expressing Physiological Levels
of Surface CD4 -The CD4 protein is essential for entry of CD4-dependent HIV strains into target cells (34). However, the role of CD4 after viral entry remains controversial. To investigate the role of CD4 during HIV replication we engineered a set of Jurkat cell lines expressing different levels of surface CD4. By utilizing Jurkat-T cell lines that differed solely in the levels of expression of CD4 we minimized the contribution of unknown factors, which might vary among different T cell lines. CD4-positive cells were derived from a Jurkat cell line constitutively expressing the SV40 large T antigen. This modification was needed to achieve levels of CD4 expression comparable with those found in CD4-positive lymphocytes. Jurkat-T cells were then transfected with a CD4 expression plasmid (pCMX-CD4), selected with 0.1 mg/ml zeocin, and subjected to magnetic separation based on their levels of surface CD4. After magnetic separation, Jurkat-CD4 cells were grown for 2-3 days and stained with a CD4-specific mAb (OKT4) to estimate surface expression (Fig. 1). The mean values of their fluorescence intensities (m.f.) were estimated and compared with the levels found in CD4-positive peripheral blood lymphocytes (PBLs) and monocytes stained in parallel. Jurkat-T high-CD4 cells expressed levels of surface CD4 10-fold higher than low-CD4 cells (m.f. 296 versus 28). CD4 surface levels on Jurkat-T low-CD4 cells were comparable with the levels found in monocytes (m.f. 39), whereas expression in high-CD4 cells was 106% higher than in CD4-positive PBLs and 28% higher than in SupT1 cells. These analyses show that surface levels of CD4 in these cell lines fairly reflect the levels of expression observed in the natural targets of HIV infection. Two days after sorting Jurkat-T cells were subjected to electroporation with proviral constructs (Fig. 2). We used a proviral vector (NL4.3-GFP) encoding a GFP reporter gene that allowed us to readily monitor HIV producer cells. NL4.3-GFP is a replication competent vector that encodes all the HIV-1 proteins, including Env, Vpu, and Nef. The nef gene is expressed from a bicistronic RNA in which translation is driven by an internal ribosomal entry site (2). NL4.3-GFP down-modulates MHC class I protein to an extent similar to the parental wild-type virus (35). Twenty hours after electroporation, cells were stained with the CD4-specific OKT4 mAb and the level of surface CD4 in GFP-positive cells was analyzed by flow cytometry (Fig. 2). Interestingly, expression of the complete set of HIV genes was needed to achieve maximal down-modulation of CD4. In fact, even the entire set of CD4 down-modulating genes present in wild-type viruses (env, vpu, and nef) did not achieve complete removal of CD4 from the surface of producer cells. This fact was observed even in Jurkat-T cells expressing low levels of surface CD4, but was more evident in high-CD4 cells, where the CD4-specific fluorescence signal was significantly above that observed in CD4-negative cells or in CD4positive cells stained with isotype-matched control antibodies (Fig. 2B). The requirement of the nef gene for efficient CD4 down-modulation was confirmed in low-and high-CD4 cells. A NL4.3⌬Nef-GFP vector lacking the nef gene showed a significant defect in the ability to down-modulate CD4. This defect was more evident in high-CD4 cells, where more than 50% of the HIV producer cells did not show significant reductions in surface CD4. These results strongly support the idea that down-modulation of surface CD4 requires a complete set of HIV down-modulator genes, and that this function may be saturated at physiological levels of CD4.
Surface CD4 Interferes with the Infectivity of HIV-1 Produced in Jurkat-T Cells-Twenty-four hours after electroporation, viruses were collected from the supernatant and their infectivity assessed in Jurkat-T low-CD4 cells. Target cells were infected with viruses produced either in the absence of CD4, or with low-or high-CD4 surface levels. The number of GFP-positive cells was estimated by flow cytometry 24 h after infection. First, we wanted to assure that the relationship between the number of GFP-positive cells and the input virus was linear. Target cells were exposed to different amounts of viral supernatant and the infectivity assessed. As shown in Fig. 3, the relationship between the number of GFP-positive cells and the added volume was linear. The infectivity titer obtained in the linear range of the curve was 1,260 GFP-positive cells/ng of p24 capsid protein. Therefore, this parameter can be used to estimate the infectivity titer of viral supernatants. Infections described below were analyzed within the linear range observed in Fig. 3. In Fig. 4A, the infectivity of HIV particles released from producer cells expressing different levels of surface CD4 was analyzed. Expression of CD4 in Jurkat-T cells reduced the infectivity of HIV, as compared with particles produced in CD4negative cell cultures. Interestingly, the infectivity of wild-type particles released from low-CD4 cells was inhibited by 75%. These results suggest that residual CD4 levels found in the surface of low-CD4 cells transfected with wild-type particles ( Fig.   FIG. 1. Construction of T cell lines expressing CD4. Jurkat-T cells were transfected with a CD4 expression plasmid. Utilizing a combination of flow cytometry and antibody-conjugated microbeads, cells were sorted into low-and high-CD4 populations expressing different levels of surface-CD4 (see "Experimental Procedures"). Twenty-four hours after sorting with microbeads, surface expression of CD4 was analyzed by flow cytometry after staining with a CD4-Cy5 conjugated antibody. As a comparison, surface expression of CD4 in PBLs, monocytes, and in the SupT1 cell line, are also shown. Flow cytometry settings were adjusted to correct for size differences among the different cell types. All cell types showed similar background levels of fluorescence when stained with control isotype-matched antibodies. All measurements are from a single experiment and were run in parallel. The fluorescence mean values were as follows: Jurkat-T No-CD4 (4), Jurkat-T Low-CD4 (28), Jurkat-T high-CD4 (296), monocytes (39), PBLs (143), and SupT1 (232). CD4-negative Jurkat-T cells (dotted line) produced the same signal that CD4-positive cells stained with isotype-matched antibody. Jurkat-T low-CD4 and high-CD4 cells are shown with light gray and dark gray histograms, respectively.

FIG. 2. CD4 surface levels in Jurkat-T cells expressing HIV proteins. Jurkat-T cells expressing either low (panel A) or high (panel B)
surface CD4 were electroporated with 20 g of a proviral construct (NL4.3-GFP, indicated as wild-type) or its Nef-defective version (NL4.3⌬Nef-GFP, shown as ⌬Nef). Twenty hours after electroporation cells were stained with a CD4-specific mAb (OKT4) and the levels of surface CD4 were analyzed by flow cytometry. The histograms show the fluorescence in GFP-positive cells transfected with either wild-type virus (solid line) or the Nef-defective version (filled histogram). Levels of surface CD4 in untransfected cells (GFP-negative cells from the same electroporation) are shown as thin line histograms. Staining with isotype-matched antibody is also shown (dotted lines).
2A) are sufficient to diminish HIV infectivity. The infectivity of wild-type particles released from high-CD4 cells was decreased by 90%. The inhibitory effect of CD4 was also revealed in Nefdeleted viruses. In these viruses infectivity was reduced by 95% when released from high CD4-expressor cells (Fig. 4A). It is important to note that expression of surface CD4 in high-CD4 cells was not homogenous. About 10 -13% of the culture showed levels of surface CD4 comparable with those in low-CD4 cells (see Fig. 1). Therefore, it cannot be excluded that the low number of infectious particles released from the high-CD4 cultures are indeed produced from these low expressor cells. Attempts to remove these low-CD4 contaminant cells by further rounds of purification were not successful. To determine the specificity of the results, cells were electroporated with a mixture of NL4.3-GFP proviral vector and a plasmid encoding the VSVg envelope glycoprotein (pMDG). Co-expression of VSVg allowed mixed pseudotyping of HIV particles and abrogated the CD4 inhibitory effect even in cells expressing high surface levels of the HIV receptor (Fig. 4A). Therefore, expression of CD4 does not compromise the ability of the cells to produce fully infectious HIV particles if pseudotyped with an heterologous envelope. The CD4inhibitory effects were revealed only when HIV enters into cells via its own envelope. To a lesser extent, co-expression of the amphotropic MLV envelope also overcame the CD4-mediated inhibitory effect. The partial recovery observed in high-CD4 cells is likely due to the lower efficiency of MLV-pseudotyping, as compared with VSVg pseudotypes, and the fact that a fraction of the mixed pseudotyped particles may still enter into cells using CD4 as receptor. Under the experimental conditions used above, VSVg-pseudotyped particles were 10-fold more infectious than either HIV-or MLV-pseudotyped particles (for absolute values see Fig. 5).
To investigate whether overexpression of CD4 inhibits virion release we used an immunocapture ELISA to estimate the amount of p24 antigen released from producer cells (Fig. 4B). Expression of low levels of surface CD4 did not reduce virion release in Jurkat-T cells. However, higher levels of CD4 expression inhibited the release of HIV particles by 75-85%, as compared with CD4-negative cells. Neither VSVg nor MLV envelope coexpression abrograted the CD4-mediated inhibition of particle release, suggesting that high levels of CD4 expression may also inhibit virus release in an envelope-independent manner. Our findings confirm a recent report by Bour et al. (15) which showed that the CD4-mediated inhibition of particle release from CD4-positive HeLa cells is not specific for HIV Env-pseudotyped viruses (15).
It has been previously reported that overexpression of CD4 in 293T cells interferes with the incorporation of HIV Env into viral membranes (23). Attempts to analyze the protein content of these HIV particles were unsuccessful. Consistently, supernatants collected 24 h after electroporation resulted in yields of 1-2 ng of p24 protein/ml of culture. These low amounts of HIV particle production precluded analysis of viral proteins. To investigate the role of viral gp120 in CD4-mediated inhibition we decided to assess the ability of HIV to infect target cells with different levels of surface CD4. The level of expression of CD4 modulates the efficiency of HIV entry into target cells. This is particularly noticeable when the concentration of chemokine receptor is low, suggesting that the requirement for CD4 increases when the other HIV co-receptor is expressed in limiting amounts (36). HIV particles carrying a GFP reporter gene were produced from Jurkat-T cells expressing either no surface CD4 or high levels of the receptor, and the infectivity of these particles was analyzed as shown in Fig. 4. This time the assays were performed using target cells expressing either low or high levels of CD4 (Fig. 5). As expected, viral particles using the HIV envelope glycoprotein showed higher levels of infectivity when tested in high-CD4 target cells. Thus, HIV particles collected from the supernatant of CD4-negative cells showed a 140% increase in infectivity when assayed in high-CD4, as compared with low-CD4 cells (Fig. 5A). A similar pattern was observed with HIV particles lacking the nef gene and produced in CD4negative Jurkat-T cells (Fig. 5B). HIV particles pseudotyped with HIV Env and VSV glycoprotein did not show significant changes in infectivity when tested in cells with low or high levels of surface CD4 (Fig. 5, C and D), as would be expected if these particles predominantly use a CD4-independent route for entry. A similar pattern was found in HIV particles produced from high-CD4 expressor cells. However, the increase in infectivity observed in high-CD4 versus low-CD4 target cells was significantly higher in viruses produced in the presence of CD4 than in its absence (305% increase versus 140%, Fig. 5, E and  A). This fact was more evident in HIV particles produced in the absence of a functional nef gene. These viruses showed a 775% increase in infectivity when analyzed in high-CD4, as compared with low-CD4 target cells (compared with a 145% increase in viruses produced in CD4-negative cells). VSVgpseudotyped particles produced in high-CD4 cells did not show any increase on infectivity when analyzed in high-CD4 target cells (Fig. 5, G and H). These results indicate that the defect in HIV particles produced from high-CD4 cells can be partially overcome by increasing the concentration of CD4 in target cells.
The experiments described above show an inverse correlation between surface CD4 levels and infectivity of released HIV particles in an HIV envelope-dependent manner. However, the method of choice to deliver HIV genomes into these cells (electroporation) may impose conditions that are different from virus production in infected cells. To investigate the role of surface CD4 in HIV-infected cells we created a new set of CD4-positive Jurkat-T cells. High-CD4 cells were sorted with the same CD4 surface settings used in the experiments described in Figs. 1-4. However, low-CD4 cells were selected with even lower levels of surface CD4 than those previously studied in electrotransformation experiments. These cells were infected with HIV viruses and analyzed in the experiments shown in Figs. 6-9. In the experiment shown in Fig. 6, Jurkat-T cells were infected by spinoculation with wild-type R7 HXB2 , a T-cell tropic virus that expresses all the CD4 down-modulator genes from their natural regulatory elements, or its Nef-deleted version (R7⌬Nef). Surface CD4 was analyzed 36 h after infection with a CD4-specific mAb, whereas infected cells were scored by intracellular staining with a p24 mAb. The requirement for Nef to efficiently down-modulate CD4 was again confirmed. In general, the ability of the virus to down-modulate CD4 was similar to that observed with NL4.3-GFP-electropo- rated Jurkat-T cells (Fig. 2). However, down-modulation of the viral receptor in this set of low-CD4 cells was complete, as estimated by the CD4 mean fluorescence in cells infected with wild-type virus and stained with either anti-CD4 (m.f. 5.9) or isotype-matched control antibody (m.f. 5.5) (Fig. 6B). Nef, Vpu, and Env contributed to decrease 7-8-fold the surface levels of CD4 in high-CD4 cells (Fig. 6E) although, as previously observed in electroporation experiments, the HIV receptor was not completely eliminated from the cell surface. Confirming previous results (Fig. 5), high-CD4 cells were 3 to 4 times more susceptible to infection with HIV than low-CD4 cells, as estimated by the percentage of p24-positive cells (compare Fig. 6, B and E, or C and F). It could be argued that signals transduced through CD4 may affect expression of the HIV genome, leading to misinterpretation of infectivity assays in which expression of a viral protein is analyzed. However, we did not find differences in the amount of viral proteins expressed in either low-CD4 or high-CD4 HIV-infected cells (compare intracellular p24 mean fluorescence levels, Fig. 6, B and E), suggesting that in our experimental conditions signaling through CD4 does not modulate HIV expression. Viruses released from infected cells were collected, filtered, and assayed immediately for infectivity in either CEM-GFP (Fig. 7A) or HeLa P4 reporter cells (Fig. 7B). Eight hours after infection, 5 M azidothymidine was added to the cultures to block new rounds of infection. Viruses produced in high-CD4 cells showed a 4 -8-fold decrease in infectivity, as compared with particles produced in low-CD4 cells. However, the release of viral particles, as estimated by p24 ELISA, was not reduced in high-CD4 cells. In fact, these cells show a 4 -5fold increase in the release of p24 (Fig. 7C), as expected if these cells are infected with higher efficiencies than low-CD4 cells.
Reduced Levels of gp120 Incorporation in Viruses Produced from High-CD4 Jurkat-T Cells-To analyze the effect of CD4 in protein composition of viral particles, we purified virus from the culture medium of low-or high-CD4 cells. Cell-free viral particles were separated by low-speed centrifugation and concentrated through a sucrose cushion. Equal amounts of p24 protein was loaded onto SDS-polyacrylamide gels, transferred to polyvinylidene difluoride membranes, and probed with antibodies specific to gp120, CD4, and p24 proteins (Fig. 8A). HIV envelope incorporation was slightly reduced in Nef-defective viruses produced in low-CD4 cells, as compared with wild-type viruses. This reduction was more significant in viruses produced from high-CD4 cells, Nef-defective viruses showed a 80% reduction in the amount of gp120 incorporated, as compared with wild-type virions. Analysis with a CD4 antiserum revealed incorporation of this protein into the membranes of viral particles produced in high-, but not low-CD4 cells. Viral particles produced in 293T cells transfected with a Nef-defective proviral construct, in the absence or presence of CD4, were included for comparison (Fig. 8B). These particles showed a similar pattern of decreased gp120 levels and incorporation of CD4 into viral membranes. These findings suggest that expression of surface CD4 interferes with incorporation of envelope into HIV particles produced in infected Jurkat-T cells, and also explain the results presented in Fig. 5. Reductions of gp120 protein in viral membranes may be overcome by elevated levels of the HIV receptor in the surface of target cells.
To address the role of CD4 in multiple rounds of infection, low-and high-CD4 cells were infected with R7 HXB2 or its Nefdeleted version, and the replication of HIV was monitored at intervals by measuring the amount of p24 protein released into the medium (Fig. 9A). Despite the fact that high-CD4 cells were previously found to be more permissive to infection (Fig. 6), both wild-type and Nef-defective viruses replicated at lower rates in high-CD4 than in low-CD4 cells. Wild-type particles started to appear in cultures of low-CD4 cells around day 11 after infection, whereas the same viruses were not detected in high-CD4 cultures until 19 days post-infection. Similar results were found with Nef-defective viruses although with lower levels of replication. These findings demonstrate that surface CD4 is detrimental for the replication of HIV in T cell lines. Interestingly, once the p24 signal was detected in cultures of high-CD4 cells, the rate of replication (as estimated by the slope of the curves) was similar in both low-or high-CD4 cells.
To explain these findings we measured the profile of surface CD4 expression in mock-infected cultures at days 1 and 19 after infection. As shown in Fig. 9B, fresh cultures of high-CD4 cells were composed mainly of cells with high levels of CD4 (89%), whereas a small fraction (11%) showed values comparable with those found in low-CD4 cells. However, this latter fraction increased throughout the experiment and, at day 19, already constituted 50% of the population (Fig. 9C). Therefore, the onset of replication observed in high-CD4 cultures after 19 days of infection could be due to viral replication in the contaminant population of low-CD4 cells. Supporting this hypothesis, Nef-deleted viruses collected from "late" cultures of high-CD4 cells showed low levels of CD4 incorporation and high gp120 content (data not shown), characteristics that marked the source of these virions as produced in cells expressing low levels of surface CD4. Therefore, the detrimental effects of CD4 expression in multiple rounds of infection are likely to be more drastic than those observed under the experimental conditions used in Fig. 9. Our results are consistent with those by Marshall et al. (37), who reported that high levels of surface CD4 in the tumor cell line HSB prevents the establishment of persistent HIV infections. DISCUSSION The CD4 protein is essential for infection of target cells with most HIV-1 strains (34). Paradoxically, despite the requirement for this protein to enter target cells, three HIV-1 proteins are involved in the removal of CD4 from the surface of productively infected cells (2,38). Experiments with epithelial adherent cells have shown that CD4 expression inhibits HIV replication at a late stage of the viral cycle (15, 23, 24). These  2) or high-CD4 (lanes 3-8) Jurkat-T cells were infected with wildtype or Nef-defective R7 HXB2 viruses. Thirty-six hours after infection viral particles were collected from the medium, filtered, and concentrated through a sucrose cushion. Four micrograms of p24 protein (as estimated by ELISA) were separated onto a 10% SDS-polyacrylamide gel, transferred to polyvinylidene difluoride membranes, and then probed with anti-gp120, anti-CD4, or p24-specific antibodies. As a control the supernatant of high-CD4 Jurkat-T cells was processed in parallel (lane 5). Serial dilutions of wild-type viruses produced in high-CD4 cells were run in lanes 6 -8. Panel B shows the levels of gp120, CD4, and p24 proteins found in Nef-defective viral particles produced by transfection of 293T cells. Twenty-five g of a Nef-defective proviral construct were transfected alone (lane 9) or in the presence of 10 g of a CD4-expresion plasmid (lane 10). As a control supernatants from cells transfected with CD4 alone are shown in lane 11.

CD4-mediated Inhibition of HIV Infectivity
findings addressed an important topic in HIV biology as to why the viral receptor needs to be down-modulated in producer cells (25). However, these results also raised important issues that we have investigated here. First, previous reports made use of transformed cells that overexpressed the HIV receptor to very high levels (exceeding 10-fold the levels found in the surface of SupT1 cells) (24), which did not reflect the conditions found in HIV permissive T cells. Second, the experiments were performed in adherent fibroblast cells which in normal conditions do not express CD4, and lack the molecular machinery that transduces signals through this protein and controls its internalization rate (39). To better reproduce the in vivo conditions, we have created a set of transformed T cell lines expressing physiological levels of surface-CD4. Although comparison of unrelated T cell lines with different levels of surface-CD4 could have been an alternative approach, we decided to manipulate a parental Jurkat-T cell line to minimize variation in other cell factors that could affect HIV replication. In a first set of experiments HIV particles were produced by electroporation rather than direct infection, since this approach allowed us to test the specificity of the findings by pseudotyping viruses with foreign envelopes. Expression of CD4 on the surface of these cells correlated with a reduction in the infectivity of HIV particles. These inhibitory effects were overcome by pseudotyping HIV viruses with either VSVg or MLV envelope, indicating that the observed interference of HIV infectivity is a gp120-dependent effect.
In a second set of experiments we analyzed the infectivity of HIV viruses produced from infected rather than transfected cells. Similar results were found in this experimental setting, which allowed us to concentrate and analyze sufficient amounts of HIV particles that had undergone only a few rounds of replication. These experiments showed that Nef-deleted HIV particles produced in high-CD4 cells present low levels of gp120 protein in their membranes. This reduction likely accounts for the low infectivity values observed in these viruses, and explains why this defect is partially overcome by using high-CD4 target cells, where an excess of CD4 may overcome a reduction of gp120 in viral membranes. Interestingly, decreased infectivity values were also observed in low-CD4 cells transfected with wild-type viruses. These viruses were unable to achieve complete down-modulation of CD4 from the surface of infected cells, suggesting that even small amounts of CD4 in the surface of producer cells may interfere with HIV infectivity. Unfortunately, due to the small amounts of virus produced upon electrotransformation, we were not able to determine whether in this experimental setting interference was mediated by a block in Env incorporation.
Unlike previous reports, production of HIV particles was analyzed in a T cell line, and infectivity assays were performed in the same cell line and confirmed in at least two more CD4positive reporter cells, CEM-GFP and HeLa P4. It is unlikely that reductions in infectivity can be attributed to changes other than CD4 expression, since these cells maintained the ability to produce infectious viruses if a foreign envelope protein that does not bind to CD4 was used to pseudotype HIV particles. It is important to emphasize that in electrotransfection experiments significant reductions in infectivity were seen even in wild-type HIV particles produced in cells with surface CD4 levels comparable with those found in monocytes. Our results should be validated in HIV-infected PBLs and macrophages, natural targets of HIV infection. The amount of CD4 on the surface of T cells and macrophages could be manipulated by transducing these cells with CD4-expressing viral vectors. However, obtaining homogeneously transduced cultures of primary cells remains challenging. Alternatively, the role of surface CD4 could be studied with specific inhibitors of CD4 endocytosis. In this regard, ikarugamycin, a novel antibiotic that blocks PMA-and Nef-mediated down-modulation of CD4 (40), could represent a useful tool to perform these studies in primary cells.
Besides the decrease in infectivity, a defect in the release of HIV particles was also observed at higher levels of surface CD4 in cells electrotransfected with NL4.3-GFP proviral constructs. However, we did not observe reductions in HIV released from high-CD4 cells infected with R7 HXB2 viruses. Differences in the level of expression of viral genes delivered by either infection or electrotransfection may account for these variations. These results suggest that an inhibition of particle release may not significantly contribute to the CD4 inhibitory effect in in vitro infections. However, we cannot rule out the possibility that, in vivo, both effects might play a role in HIV infection. Our results contrast with those presented by Ross et al. (24) who observed a reduction in HIV particle release in a gp120-specific manner. In our electroporation experiments HIV particles pseudotyped with either MLV or VSVg envelope proteins were also poorly released from high-CD4 producer cells, suggesting that the observed interference does not require a gp120-CD4 interaction. However, it cannot be ruled out that, since these viruses are mixed pseudotypes, the HIV envelope protein still present in these particles is enough to diminish virus budding at high levels of CD4. Our findings are in agreement with a more recent report by Bour et al. (15), which showed that CD4 inhibits HIV particle release in an Env-independent manner. These authors suggested that CD4 interferes with the ability of Vpu to enhance particle release. Unfortunately, this study did not analyze the infectivity of the released particles. Interestingly, inhibition of particle release required the cytoplasmic domain of CD4 (15), whereas the block in HIV infectivity that we observe does not require this domain (23), suggesting that the CD4-dependent interference of infectivity and particle release are mechanistically independent events.
The mechanism of CD4-mediated interference of incorporation of envelope into viral membranes is yet unknown. Experimental evidence suggest that assembly and budding of HIV particles occur in regions of the plasma membrane that have defined lipid and protein compositions (41,42), and several HIV proteins have been found to associate with these regions (41,43,44). These regions have been named as glycolipidenriched membrane lipid rafts and can be selectively enriched by taking advantage of their resistance to treatment with Triton X-100 (45). CD4 has been found to distribute in both raft and non-raft regions on the surface of T cells (46). However, the determinants that modulate the compartmentalization of CD4 into membrane microdomains are unknown. It is interesting to note that CD4 blocks Env incorporation without affecting its transport to the cell surface (23), suggesting that the HIV receptor may alter either the distribution or the structure of gp120 molecules in a way that makes them incompetent for incorporation into membranes. Elevated levels of surface CD4 may saturate the budding sites where the envelope protein is selectively recruited through interactions between the matrix (MA) protein and the cytoplasmic domain of gp41 (47,48). In this manner CD4 might also block the recruitment in lipid rafts of other viral proteins such as Gag. This fact would explain the HIV Env-independent inhibition of virus budding observed in some instances with high levels of surface CD4. Alternatively, CD4, by interacting with the extracellular domain of gp120, may redistribute the envelope away from budding sites. Experiments aimed at analyzing the distribution of CD4 and gp120 in budding sites on the surface of infected cells, and characterizing the factors that govern the localization of these proteins are currently being examined in our laboratory. The recruitment of CD4 into viral particles also suggests that its inhibitory effect may be mediated not only by impeding Env incorporation, but also by saturating CD4-binding sites on the surface of the viral membranes. Supporting the existence of alternative mechanisms, both wild-type and Nef-defective particles showed decreased infectivity values when produced in high-CD4 cells and incorporated CD4 in their viral membranes, however, only Nef-defective viruses showed significant reductions in the amount of Env incorporation (Fig. 8). Therefore, CD4 might be interfering with Env function at two levels, by preventing its incorporation into viral membranes, and by diminishing the ability of incorporated Env proteins to bind CD4 molecules in the surface of target cells. Experiments to analyze the contribution of both effects are currently underway.
It is also important to note that, as shown in our experiments, high surface CD4 levels in target cells may mask the CD4-mediated inhibitory effect, possibly by reducing the threshold of gp120 in viral particles required to promote entry. This fact should be taken into consideration when analyzing the phenotypic properties of Nef-defective HIV viruses, which therefore may be specific for the producer and the target cells. Infectivity assays that utilized different target cells might have confounded the results in previous studies.
In addition to the CD4-mediated interference in envelope incorporation and particle release other plausible reasons to explain why the HIV receptor needs to be down-modulated have been presented. It has been suggested that removal of CD4 would prevent superinfection of cells (21). However, this fact appears to be of little relevance in the periphery, where low concentrations of viral particles make reinfection events unlikely. It has also been proposed that signals transduced through CD4 may negatively regulate expression of HIV by a mechanism depending on CD4 oligomerization (49). However, controversy exists as to whether these signals are indeed of negative or positive nature (49,50). The Jurkat-T cell lines used in our experiments expressed full-length CD4 molecules, yet we did not observe increased amounts of HIV proteins in infected cells expressing low levels of surface CD4, as compared with high CD4 expressers. Whether signals transduced through the CD4 protein play a role in HIV expression is an issue that needs to be re-examined. The inhibition of HIV infectivity described here may account by itself for the need to down-regulate CD4 in infected cells.
Nef functions appear to be modulated during HIV infection in humans. Nef alleles obtained during late stages of infection do not efficiently down-modulate class I major histocompatibility complex (MHC) but are highly active in down-regulation of CD4. Early during infection, however, the CD4 down-modulatory function is diminished (51). These results suggest that at early steps of infection the inability to efficiently down-modulate CD4 from the surface of infected cells might make HIV more sensitive to the CD4 inhibitory effect. To overcome this block, it would be expected that viruses replicating in low-CD4 target cells, such as macrophages, be selected during the early stages of infection. On the other hand, the scenario would be reversed at late steps of infection where the CD4 down-modulating activity is stronger, and therefore viruses replicating in CD4-positive lymphocytes (high-CD4 expressor cells) could propagate efficiently. Thus, specific selection of Nef alleles may CD4-mediated Inhibition of HIV Infectivity be the driving force modulating the switch from CCR5-to CXCR4-dependent strains that occurs at late stages of infection. Nef's function in class I MHC down-modulation allows infected cells to escape recognition by cytotoxic T lymphocytes (29,35). It has been suggested that immune escape may help maintain class I MHC down-modulation activity in Nef alleles early during infection (51). However, the selection forces promoting the switch from weak to potent CD4-down-modulation strains are speculative. One could reason that mutations in Nef conferring potent down-modulation of class I MHC may diminish the CD4 regulatory activity, that would not be enhanced until immune evasion is no longer needed. In this model Nef would act as a molecular clock controlling chemokine receptor usage and HIV pathogenesis.
From our findings it is also implicit that strategies targeting CD4 expression in infected cells could be used as potential avenues to interfere with HIV replication. Delivery into infected cells of CD4 recombinant proteins insensitive to HIVmediated down-modulation would exert an effect similar to protease inhibitors, leading to the production of non-infectious particles. Provided with high levels of CD4 expression, particle release might be blocked as well. This approach could help reduce viral load and slow progression to disease. Furthermore, it is predicted that like HIV, other viruses containing lipid membranes may use similar mechanisms to avoid the interference associated with the simultaneous expression of envelope and receptor proteins on the surface of infected cells.