HIV-1 Vpu Sequesters β-Transducin Repeat-containing Protein (βTrCP) in the Cytoplasm and Provokes the Accumulation of β-Catenin and Other SCFβTrCP Substrates*

The human immunodeficiency virus type 1 Vpu protein acts as an adaptor for the proteasomal degradation of CD4 by recruiting CD4 and β-transducin repeat-containing protein (βTrCP), the receptor component of the multisubunit SCF-βTrCP E3 ubiquitin ligase complex. We showed that the expression of a Vpu-green fluorescent fusion protein prevented the proteosomal degradation of βTrCP substrates such as β-catenin, IκBα, and ATF4, which are normally directly targeted to the proteasome for degradation. β-Catenin was translocated into the nucleus, whereas the tumor necrosis factor-induced nuclear translocation of NFκB was impaired. β-Catenin was also up-regulated in cells producing Vpu+ human immunodeficiency virus type 1 but not in cells producing Vpu-deficient viruses. The overexpression of ATF4 also provoked accumulation of β-catenin, but to a lower level than that resulting from the expression of Vpu. Finally, the expression of Vpu induces the exclusion of βTrCP from the nucleus. These data suggest that Vpu is a strong competitive inhibitor of βTrCP that impairs the degradation of SCFβTrCP substrates as long as Vpu has an intact phosphorylation motif and can bind to βTrCP.

Vpu is unique to HIV-1, 1 the major causative agent of the acquired immune deficiency syndrome, and is not found in other lentiviruses such as HIV-2 or in most simian immunodeficiency viruses. Vpu is an integral membrane protein of 81 amino acids (aa). Its two principal biological functions are associated with different structural domains of the protein.
Vpu increases the release of virus particles from the plasma membrane (1) through its N-terminal transmembrane domain (aa , whereas it mediates the degradation of the CD4 receptor in the endoplasmic reticulum (ER) (2) via its cytoplasmic domain (aa 28 -81). It was shown recently that Vpu exerts a positive effect on HIV-1 infectivity by down-modulating CD4 receptor molecules at the surface of HIV-1-producing cells (3). CD4 degradation requires the phosphorylation of the serine residues at positions 52 and 56 of the Vpu cytoplasmic domain by casein kinase II (4) and the binding of Vpu to CD4 trapped in the ER via the formation of a complex with the HIV-1 envelope precursor gp160 (5,6). We showed previously that Vpu binds to the F box protein ␤-transducin repeat-containing protein (␤TrCP), the receptor component of the multisubunit SCF-␤TrCP E3 ubiquitin ligase complex, and connects CD4 to the ubiquitin-proteasome machinery. ␤TrCP is linked to the SCF complex by binding to Skp1 through its N-terminal F box motif and interacts with Vpu through its C-terminal WD repeat region (7). We and others have shown that ␤TrCP is also involved in the ubiquitination and proteasome targeting of ␤-catenin (8 -13), IB␣ (12, 14 -18), and ATF4 (19). ␤TrCP also plays a role in the processing and degradation of p105 and p100, the precursors of the active p50 and p52 NFB subunits (20 -22). The signal for the recognition of all these cellular ligands by ␤TrCP is the phosphorylation of one or two serine residues present in a conserved motif, DSGXXS for Vpu, ␤-catenin, and IB␣ and DSGXXXS for ATF4, p105, and p100.
As ␤TrCP controls essential cellular signaling pathways, such as the Wnt-␤-catenin and the NFB pathways, by degrading the above-mentioned substrates via the ubiquitin-proteasome system, we wondered whether Vpu exerts a generalized inhibitory effect on multiple substrates of ␤TrCP and impairs their degradation. This could have major consequences on signaling in HIV-infected cells in vivo. Recently, Bour et al. (23) demonstrated that Vpu blocks IB␣ degradation mediated by TNF␣ signaling. Subsequently, Akari et al. (24) showed that Vpu induces proapoptotic activity through this effect. Here we demonstrate that expression of Vpu, either as a full-length HIV-1 proviral DNA or fused to green fluorescent protein, impairs ␤-catenin degradation and up-regulates its transcriptional activity. Moreover, the degradation of ATF4 and IB␣ was inhibited, as was TNF␣-induced NFB nuclear translocation. Finally, we demonstrated that Vpu induces the cytoplasmic sequestration of ␤TrCP. All these effects required an intact DSGXXS phosphorylation motif.

EXPERIMENTAL PROCEDURES
Cells and Transfections-HeLa cells and 293T human embryo kidney cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and antibiotics (Invitrogen). HeLa cells were transiently transfected by electroporation when using plasmids encoding HA-ATF4, Vpu-GFP, Vpuc-GFP, or ␤TrCP-FLAG or by the calcium method when using pNL4-3 or its variants. 293T cells were transiently transfected by the calcium phosphate method.
Plasmids-Proteins bearing the Vpu cytoplasmic domains (Vpuc) or wild-type Vpu fused to the GFP were constructed by PCR from plasmids expressing the full-length Vpu molecular clone of HIV-1 LAI or a Vpuphosphorylated mutant with an HA epitope at the C terminus (6) by using appropriate primers. Vpu2.6 carries serine to asparagine mutations at positions 52 and 56. Amplified fragments were digested with HindIII and ApaI and then inserted into the corresponding sites of a pcDNA3.1/Myc-His plasmid (Invitrogen) in which the Myc-His sequence had been replaced by the green fluorescent protein (GFP) deleted of its ATG (19). These constructs (Vpuc-HA-GFP, Vpuc2.6-HA-GFP, Vpu-HA-GFP, and Vpu2.6-HA-GFP) were named Vpuc-GFP, Vpuc2.6 GFP, Vpu-GFP, and Vpu2.6-GFP, respectively, for simplicity.
The pcDNA3-⌬F␤TrCP construct was described previously (7). We used a pcDNA3-⌬F␤TrCP-tagged Myc/Hist, in which the ⌬F␤TrCP coding sequence was amplified by PCR and inserted in fusion with the Myc/histidine double tag in the pcDNA3.1 Myc/HisA vector (15). The hemagglutinin (HA)-tagged human ATF4 plasmid was described previously (19). ␤TrCP-FLAG was constructed by inserting the human ␤TrCP cDNA into the XhoI and BamHI sites of pSGFLAG (a gift from P. Jalinot, Ecole Normale Superieure de Lyon). pNL4-3, the HIV1 clone, and its Vpu-deficient mutant were gifts from Dr. K. Strebel.
Pulse-Chase Analysis-To perform pulse-chase analysis of endogenous ␤-catenin, 8 ϫ 10 6 HeLa cells were transiently transfected with 30 g of Vpuc-GFP or Vpuc2.6-GFP. Twenty four hours later, cells were divided into four fractions, incubated for 30 min in methionine-and cysteine-free medium, and pulse-labeled for 30 min with 0.130 mCi of [ 35 S]methionine and -cysteine. They were then chased with fresh medium for the indicated times (see Fig. 3). Cells were lysed in 1% Nonidet P-40 lysis buffer (0.05 M Tris buffer (pH 7.4), 0.1 M NaCl, 1% Nonidet P-40, 0.001 M EDTA, 10% glycerol) supplemented with protease inhibitors. After centrifugation of the lysates, ␤-catenin was immunoprecipitated by a mouse anti-␤-catenin monoclonal antibody (dilution 1:100) for 90 min and then incubated with protein G-agarose beads (Sigma) for 1 h. After three washes with lysis buffer supplemented with NaCl (0.150 M final concentration) and one wash in lysis buffer with 0.3 M NaCl (final concentration), immune complexes were eluted with Laemmli buffer and subjected to 10% SDS-PAGE. Gels were fixed in 10% acetic acid, 30% methanol and dried. They were then used to expose film and analyzed with an ImageQuant PhosphorImager (Amersham Biosciences).
Luciferase Assays-293T cells were plated in 6-well flat-bottom plates on the day prior to transfection at a density of 5 ϫ 10 4 cells/35mm-diameter well. Transfections were performed using the calcium phosphate co-precipitation method with the mammalian transfection kit (Stratagene). We used the luciferase reporter Top-TK-Luciferase plasmid containing multimerized synthetic Lef/Tcf-binding sites fused to the luciferase reporter gene under the herpes simplex virus-thymidine kinase promoter (TK) (8). Cells were co-transfected with 0.2 g of luciferase reporter plasmid and 0.1 g to 1 g of the constructs. One ng of pRL-TK-Renilla luciferase (PRL-TK from Promega) was included as an internal transfection efficiency control. The total amount of transfected plasmids was held constant at 5 g by a DNA carrier. After 24 h, the cells were lysed, and the ratios between both luciferase activities were determined with enzyme assay kits (Dual Luciferase Reporter Assay System, Promega). Luminescence was quantified by use of a Lumat LB9507 luminometer (EG & G Instruments) from duplicate plates. The results represent data from four independent experiments. The data were normalized by function of transfection efficiency.

Accumulation of ␤-Catenin in HeLa Cells in the Presence of
Vpu-To determine whether the expression of Vpu results in the accumulation of ␤-catenin, we constructed expression vectors in which either wild-type Vpu (Vpu-GFP) or Vpu mutated at the two serine residues (S52N-S56N) of the DSGXXS motif required for binding to ␤TrCP (Vpu2.6-GFP) was fused with GFP. HeLa cells were transiently transfected with these plasmids, and the level of ␤-catenin expression was assessed by Western blotting using monoclonal anti-␤-catenin antibodies. The expression of increasing amounts of Vpu-GFP resulted in the accumulation of ␤-catenin (Fig. 1, compare lanes 1-5 with lane 13). In contrast, similar amounts of the mutated form of Vpu had no effect on the amount of ␤-catenin present (Fig. 1, compare lanes 11 and 12 to lane 13). Similar amounts of ␤-catenin accumulated when only the cytoplasmic domain (residues 21-81) of Vpu (Vpuc-GFP) was fused to GFP (Fig. 1, compare lanes 6 -10 to lane 13). By comparing different dilutions of the cell lysates, we estimated that the accumulation of ␤-catenin was more than 10-fold higher in the presence of Vpuc-GFP than in untransfected cells, whereas the mutated form (Vpuc2.6-GFP) had no effect (data not shown). Thus, the cytoplasmic domain of Vpu with an intact DSGXXS phosphorylation motif is necessary and sufficient to allow the accumulation of ␤-catenin.
Expression of Vpu Results in the Accumulation of ␤-Catenin Both in the Cytoplasm and in the Nucleus-We carried out immunofluorescence studies to determine whether the accumulated ␤-catenin was translocated into the nuclei of the Vpu-transfected HeLa cells. The intracellular localizations of Vpu-GFP, Vpuc-GFP, Vpu2.6-GFP, and Vpuc2.6-GFP were evaluated by direct fluorescence after transfection (Fig. 2, left panels). Vpu-GFP and Vpu2.6-GFP were found in the cytoplasm and in the perinuclear region of transfected cells (Fig. 2, a and c), confirming previous observations (25,26). In contrast, Vpuc-GFP and Vpuc2.6-GFP, both of which lack the Vpu transmembrane anchor domain, were found both in the nucleus and in the cytoplasm (Fig. 2, e and g). The intracellular level of ␤-catenin was considerably higher in Vpu-GFP-or Vpuc-GFPtransfected cells than in non-transfected cells (Fig. 2, b and f). ␤-Catenin accumulated both in the cytoplasm and in the nucleus. However, no ␤-catenin accumulated in the nonphosphorylated mutants Vpu2.6-GFP-and Vpuc2.6-GFP (Fig.  2, d and h)-transfected cells.
Thus, the expression of Vpu-GFP and Vpuc-GFP not only leads to elevated levels of ␤-catenin but also to the translocation of the accumulated ␤-catenin from the cytoplasm to the nucleus.
␤-Catenin Degradation Is Impaired by Expression of Vpu-To characterize further the mechanism by which expression of Vpu results in the accumulation of ␤-catenin, we measured ␤-catenin protein stability by pulse-chase analysis after 35 (Fig. 3B). Thus, ␤-catenin accumulates following the expression of Vpu because its constitutive degradation by the ubiquitin-proteasome pathway is inhibited. Once again, the DSGXXS-binding motif is required for this effect.
Vpu Promotes ␤-Catenin Activation in 293T Cells-To test whether the ␤-catenin accumulated upon expression of Vpu was able to activate the transcription of target genes, we used the TOP luciferase reporter system (27). This reporter gene contains multiple ␤-catenin-dependent Lef/Tcf sites that drive the luciferase gene. 293T cells were co-transfected with this reporter plasmid plus either Vpuc-GFP, Vpuc2.6-GFP, or ⌬F␤TrCP (a positive control for the accumulation of ␤-catenin) (8). We observed about a 3-fold increase of the luciferase activity in cells transfected by Vpuc-GFP compared with cells transfected with Vpuc2.6-GFP (Fig. 4). Such differences in luciferase activity were significant (p Ͻ 0.05) as determined by a t test (Statview F software) between cells transfected by Vpuc-GFP or Vpuc2.6-GFP (the luciferase activity is 21.17 Ϯ 2.22 in the presence of 0.5 g of Vpuc-GFP versus 6.25 Ϯ 0.585 in the presence of 0.5 g of Vpuc2.6-GFP; p ϭ 0.0006). Transfection with the Vpuc2.6-GFP had no effect compared with the negative control, and transfection with the negative transdominant form of ␤TrCP, ⌬F␤TrCP, had a slightly greater effect than Vpu (about 4.5-fold activation) (Fig. 4). Thus, the expression of Vpu results not only in the physical accumulation of the ␤-catenin protein but, more importantly, in the constitutive transcriptional activation of the ␤-catenin-Lef/Tcf-dependent genes.
␤-Catenin Also Accumulates in Cells Expressing Full-length HIV-1 Proviral DNA Containing Vpu-To determine whether Vpu in the context of a full-length HIV-1 genome also induces the accumulation of ␤-catenin, HeLa cells were transfected with pNL4-3 plasmid expressing Vpu or with a Vpu-deficient pNL4-3 plasmid (28). Production of HIV-1 proteins and of ␤-catenin was checked by immunofluorescence double labeling using total anti-HIV1 human antiserum from HIV-infected patients and mouse monoclonal anti-␤-catenin antibodies, respectively. ␤-Catenin accumulated in cells transfected with pNL4-3 expressing Vpu (Fig. 5, d or f) but not in cells trans-fected with a Vpu-deleted proviral NL4 -3 DNA (c or e). Furthermore, the accumulated ␤-catenin is translocated to the nucleus (h). Hence, HIV-1-producing cells produce enough Vpu to induce the accumulation of ␤-catenin. This suggests that the up-regulation of ␤-catenin detected in HIV-1-producing cells in vitro should also take place in HIV-1-infected cells in vivo.
Delocalization of ␤TrCP from the Nucleus to the Cytoplasm of HeLa Cells Expressing Vpu-Next, we investigated whether ␤TrCP, which is essentially found in the nucleus (19,29,30), was sequestered in the cytoplasm following the production of Vpu, which is itself produced in the cytoplasm and more specifically in the ER and the Golgi. The ␤TrCP fused to the FLAG epitope (␤TrCP-FLAG) was mainly detected in the nucleus (Fig. 6A, a). In cells co-transfected with ␤TrCP-FLAG and Vpu-GFP, ␤TrCP was clearly redistributed from the nucleus to

FIG. 5. ␤-Catenin accumulates in cells producing HIV-1 NL4 -3 Vpu؉ but not in cells producing HIV-1 NL4 -3 Vpu؊.
HeLa cells were transfected with pNL4-3-expressing Vpu (right panels) or with a Vpu-deficient pNL4-3 (left panels) and cultured on glass coverslips. Twenty-four hours later, HeLa cells were washed, fixed, and permeabilized. HIV-1-producing cells were detected by use of a human anti-HIV-1 serum (a and b) and a secondary Cya3-conjugated anti-human antibody. ␤-Catenin was detected with a mouse anti-␤-catenin antibody and a secondary Cya5-conjugated anti-mouse antibody (c and d). Confocal microscopy was performed with a Bio-Rad MRC1000 microscope. e and f show the same images as c and d, respectively, but more exposed in white and black to improve the visualization of the untransfected cells. g and h show the overlay. the cytoplasm (Fig. 6A, d-f). In contrast, in cells co-transfected with the phosphorylation mutant Vpu2.6-GFP, ␤TrCP was only detected in the nucleus, as in the control (see Fig. 6A, g-i). Although the antibodies against ␤TrCP were not excellent, rabbit anti-␤TrCP antibodies showed that endogenous ␤TrCP was excluded from the nucleus to the cytoplasm following the production of Vpu-GFP (Fig. 6B). No ␤TrCP was detected in the nucleus of Vpu-transfected cells, whereas ␤TrCP remained in the nucleus of untransfected cells. These results suggest that the binding affinity of Vpu for ␤TrCP is sufficiently high to sequester most of the endogenous ␤TrCP in the cytoplasm of the transfected cells.
TNF␣-induced Nuclear Translocation of NFKB Is Inhibited by Expression of Vpu-To check the effect of Vpu expression on signal-induced IB␣ degradation, HeLa cells were transfected with Vpu-GFP or Vpu2.6-GFP, treated or untreated with human TNF␣ (10 ng/ml), and analyzed either for degradation of IB␣ by Western blot using anti-IB␣ antibodies or for nuclear translocation of NFB by immunofluorescence using anti-p65 antibodies (Fig. 7, A and B, respectively). IB␣ was completely degraded within 5 min of treatment with TNF␣. However, IB␣ was not degraded in these conditions when Vpu-GFP was present (Fig. 7A). Similarly, the TNF␣-mediated nuclear translocation of NFB was completely prevented by Vpu-GFP (compare transfected cells with untransfected cells in d-f, in Fig. 7B).
The expression of mutated Vpu2.6-GFP had no effect on TNF␣mediated nuclear translocation of NFB (compare transfected cells with untransfected cells in Fig. 7B, j-l). These results are in agreement with data reported previously (23) on the inhibition of TNF␣-mediated NFB activation by Vpu. In conclusion, by inhibiting the TNF␣-induced degradation of IB␣, Vpu prevents the nuclear translocation of NFB and consequently impairs the activation of the NF〉 pathway.
Expression of Vpu Results in Accumulation of ATF4 -To examine whether ATF4, a newly identified substrate of ␤TrCP, accumulated in the presence of Vpu like ␤-catenin and IB␣, we analyzed the effect of Vpuc-GFP on cells co-transfected with HA-ATF4 expression vector. The amount of HA-ATF4 was always higher in Vpuc-GFP-expressing cells (Fig. 8A, 4th lane) than in cells expressing HA-ATF4 alone or in cells expressing the mutant Vpuc2.6-GFP (Fig. 8A, 1st or 3rd lanes). The amount of ␤-catenin was also higher in cells expressing exogenous HA-ATF4 (Fig. 8A, 1st lane) than in untransfected cells (Fig. 8A, 2nd lane). The most likely explanation for this result is that the overproduction of one of the ligands of ␤TrCP (either Vpuc-GFP or HA-ATF4) results in the competitive inhibition of the degradation of the other ␤TrCP ligands like ␤-catenin. Next, we wanted to compare the magnitude of the effect of Vpuc-GFP with that of HA-ATF4 on the level of ␤-catenin. Although Vpuc-GFP was produced at a lower level than HA-ATF4, it had a much greater effect on the accumulation of ␤-catenin (Fig. 8B, compare 1st and 3rd lanes).
In conclusion, these data suggest that Vpu is a more potent competitive inhibitor of ␤TrCP for the inhibition of ␤-catenin degradation than the overproduced ATF4.

FIG. 6. ␤TrCP is excluded out of the nucleus and sequestered in the cytoplasm upon Vpu-GFP expression in HeLa cells.
A, delocalization of ␤TrCP-FLAG; HeLa cells were either transfected with ␤TrCP-FLAG alone (a-c) or co-transfected either with Vpu-GFP (d-f) or with Vpu2.6-GFP (g-i). Cells were examined by direct fluorescence for the detection of Vpu-GFP constructs or stained with a mouse anti-FLAG antibody and a secondary anti-Cya3 mouse antibody for the detection of ␤TrCP-FLAG. Cells were analyzed by confocal microscopy. B, delocalization of endogenous ␤TrCP; HeLa cells were transfected with Vpu-GFP. Cells were examined by direct fluorescence for the detection of Vpu-GFP constructs (b) or stained with a rabbit anti-␤TrCP antibody and a secondary anti-Cya3 rabbit antibody for the detection of endogenous ␤TrCP (a). Transfected cells are indicated by arrows. Cells were analyzed by confocal microscopy. NT indicates untransfected cells.

FIG. 7. Nuclear translocation of NFB induced by TNF␣ treatment of HeLa cells is prevented by Vpu-GFP expression.
HeLa cells were transfected with 25 g of Vpu-GFP for 24 h. A, expression of Vpu-GFP impairs TNF␣-mediated IB␣ degradation. Vpu-GFP-transfected cells or untransfected cells (NT) were treated or not with TNF␣ (10 ng/ml for 5 min). IB␣, Vpu-GFP, ␤-catenin, and actin were detected by immunoblotting with corresponding specific antibodies. B, immunofluorescence assay showing that TNF␣-mediated nuclear translocation of NFB is inhibited by Vpu-GFP expression. The same transfection experiment as described in A was studied by immunofluorescence for nuclear translocation of endogenous NFB using anti-NFB (p65) antibody. Cells treated for 30 min or untreated with TNF␣ were fixed and stained with an anti-NFB (p65) antibody to detect endogenous NFB in the left panel (a, d, g, and j). GFP immunofluorescence was detected in the middle panels (b, e, h, and k) and overlaid in the right panels (c, f, i, and l). d, some untransfected cells are indicated by stars, and some cells transfected with GFP constructs are indicated by arrows. Cells were analyzed under a confocal microscope.
Overexpression of ␤TrCP Inhibits the Effects of Vpu on Accumulation of ␤-Catenin-If we assume that by interacting with endogenous ␤TrCP, Vpu provokes ␤-catenin accumulation, it is expected that overexpression of ␤TrCP should abrogate this effect. This is indeed what we observed as shown in Fig. 9 when the level of ␤-catenin in the presence of Vpu (lane 1) is compared with that in the presence of Vpu and overexpressed ␤TrCP-FLAG (lanes 2 and 3). This result confirms that Vpu acts as a competitive inhibitor of the degradation of the SCF ␤TrCP substrates. DISCUSSION The SCF ␤TrCP complex is one of the best characterized E3 ubiquitin ligases, whose multiple substrates are targeted to ubiquitination and proteasomal degradation. In this study, we found that HIV-1 Vpu, the first ligand of ␤TrCP to be identified (7), is a general inhibitor of the degradation of the ␤TrCP substrates. In fact, ␤-catenin, IB␣, and ATF4 accumulate, and the transcriptional activity of ␤-catenin is up-regulated in the presence of Vpu. Importantly, ␤-catenin also accumulates in cells expressing full-length HIV-1 proviral DNA containing Vpu but not in cells expressing an HIV-1 proviral DNA deleted of Vpu. We also found that ␤TrCP is redistributed from the nucleus to the cytoplasm in cells expressing Vpu. The production of Vpu completely prevented the TNF␣-induced nuclear translocation of NFB. Bour et al. (23) have shown that NFB activity is inhibited by Vpu. However, their NFBluciferase reporter study detected that the inhibition was only partial. In the presence of TNF␣, 100% of cells transfected with Vpu-GFP sequestered NFB in the cytoplasm, whereas 100% of cells not transfected or transfected with Vpu2.6-GFP translocated NFB normally to the nucleus. Our results suggest that Vpu is a powerful inhibitor of IKB␣ degradation and of NFB activity. All these effects require the presence of an intact phosphorylation motif DSGXXS in the cytoplasmic domain of Vpu. This ␤TrCP-binding site is well conserved in all ␤TrCP ligands identified so far. All these results support a model in which each ligand competes with all other ligands to interact with the same WD repeat domain of ␤TrCP, through an analogous phosphorylation motif. Besides serving as an adaptor for CD4 degradation, Vpu is a strong competitor that can inhibit the degradation of the ␤TrCP substrates, leading them to accumulate in HIV-1producing cells expressing Vpu.
As a very homologous protein, ␤TrCP2 also leads to degradation of ␤-catenin and IKB␣ (31). We showed by double-hybrid experiments or by coimmunoprecipitation studies that Vpu interacts also with ␤TrCP2 (data not shown). These results suggest that inhibition of the degradation of ␤-catenin that we observed is because of the interaction of Vpu both on endogenous ␤TrCP and ␤TrCP2.
This generalized inhibition of the SCF-␤TrCP complexes and the accumulation of undegraded substrates within cells infected by HIV-1 producing Vpu should have major effects on the essential cellular signaling pathways controlled by ␤TrCP.
The impairment of the nuclear translocation of NFB following TNF␣ treatment confirms that Vpu can strongly inhibit activation of NFB by preventing IB␣ degradation. This would in turn have pleiotropic consequences on the multiple signaling pathways controlled by this master transcription factor such as the TNF␣, interleukin-1, and Toll-like receptors pathways. All these pathways are involved in innate immunity, and Vpu may provide HIV-1 with means of counteracting these innate immunity responses. In most cells, NFB activation protects from apoptosis by controlling programmed cell death genes (32). Vpu can induce apoptosis effectively, and this effect is amplified by TNF␣ treatment (24,33).
The accumulation and constitutive up-regulation of ␤-catenin in HIV-1-producing cells expressing Vpu can also have serious consequences. ␤-Catenin is a cytoplasmic protein that participates in cell adhesion (34). Its abnormal accumulation is linked with various types of cancer and results in the transcriptional activation of several target genes with oncogenic effects (reviewed in Refs. 35 and 36). Accumulated ␤-catenin translocates into the nucleus where it interacts with a member of the lymphoid enhancer factor/T cell factor (Lef/Tcf) family of transcription factors and additional nuclear factors, forming a complex that induces the transcription of a set of oncogenic genes critical for cell transformation (37)(38)(39). Given that Vpu can up-regulate the transcriptional activity of ␤-catenin, it may be one of the viral factors that affects the occurrence of transformed lymphoid cells during chronic HIV-1 infection. Consequently, Vpu may contribute to the development of tumors in AIDS patients by activating the expression of TCF target genes. Recently, accumulations of ␤-catenin were found in Kaposi's sarcoma, the most common AIDS-related cancer (40).
Besides its well documented role in cell transformation, ␤-catenin also plays an important role in development, in particular in thymocyte cell differentiation. ␤-Catenin is critical for promoting the maturation and survival of thymocytes (41)(42)(43). As ␤-catenin accumulates in HIV-1-producing cells expressing Vpu, and given that the thymus is one of the major targets of the HIV-1 infection in AIDS patients, Vpu may be one of the viral factors responsible for the renewal of CD4 ϩ T lymphocytes following their depletion, which is a major consequence of HIV-1 infection.
One of the functions of ATF4, a member of the b-Zip family of transcription factors, is to control the cellular uptake of amino acids. The uptake of amino acids is important for the metabolism of T lymphocytes. The accumulation of ATF4 as a result of Vpu production might disrupt the metabolism of T cells infected by HIV-1.
Interestingly, overexpression of ATF4 results in accumulation of ␤-catenin. This finding suggests a novel mechanism underlying regulation of SCF ␤TrCP ubiquitin ligase activities toward specific cellular substrates in the absence of viral infection. Competition between cellular ␤TrCP substrates for access to the SCF ␤TrCP ubiquitin ligase may regulate their signaling activity and have potential physiologic or pathologic outcomes. Such a competition is probably influenced by the level of expression of each substrate and its affinity for ␤TrCP upon phosphorylation. Further studies will be necessary to explore whether upon activation of the NFB signaling pathway by TNF␣ the binding of IKB␣ to ␤TrCP might be correlated with modifications in the level of ␤-catenin, for example.
The massive delocalization of ␤TrCP from the nucleus to the cytoplasm provoked by Vpu production, as observed here, indicates that most of the endogenous ␤TrCP is redistributed. Such a ␤TrCP sequestration in the cytoplasm probably plays an essential role in the degradation of CD4 mediated by Vpu. Instead of using one of the E3 ubiquitin ligases located in the ER, which are responsible for the ER-associated degradation process (44), HIV-1 has evolved in such a way that, using Vpu, it subverts a mostly nuclear E3 ubiquitin ligase like ␤TrCP to ensure degradation of the CD4 receptor at the ER.
In conclusion, by interfering with cellular proteins like ␤TrCP, HIV-1 Vpu probably has major consequences on various functions and signaling pathways in HIV-1-infected cells. These consequences are still not fully understood and should be carefully studied in vivo during the course of HIV-1 infection. Such in vivo studies are not so easy to do, but the availability of simian human immunodeficiency virus chimeric viruses producing or not producing Vpu could be of great help. It has already been shown that a simian human immunodeficiency virus producing Vpu is far more pathogenic than a simian human immunodeficiency virus without Vpu (45). Thus, although Vpu is an accessory protein that is not essential for the infection of human CD4ϩ cells in vitro, it is probably a major pathogenic determinant for HIV infection in vivo. The serious pathogenic consequences of HIV-1 infection are probably partly related to the presence of six non-essential, so-called accessory genes in the HIV genome. Like Vpu, the viral proteins encoded by these genes act as connectors with various specialized ma-chineries in the infected host cells. Detailed knowledge of the intimate relationships between these viral proteins and their cellular partners and of the mechanisms involved could help scientists to design novel therapeutic approaches that may complement the currently used combination therapies.