The HIV Nef protein alters Ca(2+) signaling in myelomonocytic cells through SH3-mediated protein-protein interactions.

Human immunodeficiency virus Nef plays an important role in AIDS pathogenesis. In addition to the well known down-regulation of cell surface receptors (CD4, MHCI), Nef is able to alter cellular signaling. Of particular interest for this study is the ability of Nef to bind with a very high affinity to SH3 domains of myelomonocyte-specific protein-tyrosine kinases of the Src family (Src-like PTK). We have therefore investigated Ca(2+) signaling in HL60 cells retrovirally transduced with wild type Nef or with a Nef mutant deficient in the SH3-interacting proline-rich motif (Nef((PXXP)4(-))). In differentiated HL60 cells, Nef markedly altered cellular Ca(2+) signaling; the amount of intracellularly stored Ca(2+) was increased, and as a consequence, store-operated Ca(2+)-influx was decreased. This effect was not observed in undifferentiated HL60 cells or in CEM T-lymphocytes and correlated with the differentiation-induced up-regulation of Src-like PTK. The Nef effect on Ca(2+) signaling depended entirely on the integrity of its PXXP motif. The Src-like PTK p56/59(hck) co-immunoprecipitated with both Nef and with the inositol 1,4,5-trisphosphate receptor, providing a possible mechanistic link between the viral protein and intracellular Ca(2+) stores of the host cell. Collectively, our results demonstrate that the human immunodeficiency virus 1 Nef protein manipulates intracellular Ca(2+) stores through SH3-mediated interactions in myelomonocytic cells.

Human immunodeficiency virus Nef plays an important role in AIDS pathogenesis. In addition to the well known down-regulation of cell surface receptors (CD4, MHCI), Nef is able to alter cellular signaling. Of particular interest for this study is the ability of Nef to bind with a very high affinity to SH3 domains of myelomonocyte-specific protein-tyrosine kinases of the Src family (Src-like PTK). We have therefore investigated Ca 2؉ signaling in HL60 cells retrovirally transduced with wild type Nef or with a Nef mutant deficient in the SH3interacting proline-rich motif (Nef (PXXP)4 ؊). In differentiated HL60 cells, Nef markedly altered cellular Ca 2؉ signaling; the amount of intracellularly stored Ca 2؉ was increased, and as a consequence, store-operated Ca 2؉influx was decreased. This effect was not observed in undifferentiated HL60 cells or in CEM T-lymphocytes and correlated with the differentiation-induced up-regulation of Src-like PTK. The Nef effect on Ca 2؉ signaling depended entirely on the integrity of its PXXP motif. The Src-like PTK p56/59 hck co-immunoprecipitated with both Nef and with the inositol 1,4,5-trisphosphate receptor, providing a possible mechanistic link between the viral protein and intracellular Ca 2؉ stores of the host cell. Collectively, our results demonstrate that the human immunodeficiency virus 1 Nef protein manipulates intracellular Ca 2؉ stores through SH3-mediated interactions in myelomonocytic cells.
Nef is a highly conserved 27-34-kDa myristoylated HIV 1 protein that belongs to a set of accessory proteins characteristic of primate lentiviruses. Nef is expressed early in the viral replication cycle (1,2). Results of studies in rhesus macaques suggest that it is a key factor in the pathogenesis of HIV/simian immunodeficiency virus infection and that it plays a particularly important role for the persistence of infection (3,4). An important role for Nef in HIV pathogenesis was also suggested by analysis of the HIV genome in long term surviving patients (5,6).
The down-regulation of immunologically relevant cell surface proteins (CD4 and MHCI) is the best documented biological activity of Nef (7). Several studies indicate that Nef physically connects CD4 with components of the trafficking machinery, leading to the formation of endocytic structures that trigger the accelerated internalization and lysosomal sorting of CD4 (8 -13). In addition, Nef exerts genetically and biochemically distinguishable effects on the biology of the cell and on viral growth (14,15). Nef enhances virus replication by facilitating the early post-entry steps of infection, indirectly promoting the efficiency of reverse transcription (16). Nef also perturbs cellular activation pathways both in lymphoid and nonlymphocytic cells (17)(18)(19)(20). The mechanisms and significance of these latter effects are poorly understood. The identification of cellular proteins interacting with Nef both in vitro and in vivo have strengthened the concept that Nef acts as a modulator of intracellular activation pathways. Indeed, tyrosine and serine/threonine protein kinases have been reported to interact with Nef (21). In particular, studies in vitro with recombinant proteins and crystal structure determinations have shown that the core region of Nef, via a proline-rich motif, binds with a high affinity to the SH3 domains of Src-like protein-tyrosine kinases (Src-like PTKs) (14,22). Four members of the Src kinases family, p56 lck , p56/59 hck , p59 fyn , and p53/56 lyn have been shown to be able to interact physically with Nef (14,(22)(23)(24)(25). Interestingly, the affinity of Nef-p56/59 hck interaction is the highest known for an SH3-mediated proteinprotein binding (23).
The Src family of nonreceptor tyrosine kinases consists of nine members (Src, Lck, Hck, Fyn, Fgr, Yes, Blk, Lyn, and Yrk) that share common structures and regulation (26). These kinases are widely expressed in cells of hematopoietic origin, with most cells expressing more than one family member (27). Src-family kinases have been implicated in multiple signaling events including cell growth (28), T cell antigen receptor signaling (29), Fc receptor signaling (30,31), integrin signaling (32), glycosyl phosphatidylinositol-anchored protein signaling (33), phagocytosis (34), apoptosis (35), and tumor necrosis factor release (36). There is also growing evidence that Src-like tyrosine kinases are implicated in the control of cytosolic free Ca 2ϩ concentration (37)(38)(39)(40)(41)(42). In these studies, we used HL60 cells as a cellular model to investigate a possible effect of Nef on Ca 2ϩ -associated signal transduction pathways in myelomonocytic cells.

EXPERIMENTAL PROCEDURES
Reagents-Thapsigargin (Tg), ionomycin, 1,25(OH) 2 D 3 , phorbol 12myristate 13-acetate (PMA), and Me 2 SO were purchased from Sigma, fetal calf serum and geneticin disulfate (G418) were from Life Technologies, Inc., fura-2/AM was from Molecular Probes (Eugene, OR), Com-* This work was supported by Swiss National Science Foundation (FNRS) Grants 3139 055226.98 (to D. P. L.) and 31.50560.37 (to D. T.), by National Institutes of Health Grant R37 AI34306 (to D. T.), and by a professorship from the Giorgi-Cavalieri Foundation (to D. T.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Cell Culture and HL60 Cell Differentiation-The promyelocytic HL60 cell line stably expressing either Nef, Nef (PXXP)4 Ϫ, or the empty vector was cultured in RPMI 1640 medium supplemented with 10% fetal calf serum and selected in the presence of 1 mg/ml G418. The CEM cell line expressing Nef was described elsewhere (9). CRE and CRIP cells (43) were maintained in Dulbecco's modified Eagle's medium supplemented with 10% newborn calf serum. HL60 differentiation to the granulocyte phenotype was triggered by adding Me 2 SO (final concentration 1.3% v/v for 3 days and then 0.65% for 2 days) to the culture medium. Differentiation toward the monocyte and macrophage phenotype was obtained by adding, respectively, 100 nM of 1,25(OH) 2 D 3 for 5 days and 20 nM of PMA for 2 days to the culture medium (44 -49). Nef expression did not affect the general morphology of differentiated cells, and cytofluorimetry analysis of specific cell surface markers such as CD14, CD11b, CD11c, CD18, HLA DR, HLA ABC, the fMet-Leu-Phe (fMLP) receptor, and CD71 confirmed that differentiation to each phenotype was efficient under these conditions (data not shown).
HL60 Retroviral Transduction-HL60 cell lines stably producing the various murine leukemia virus-based retroviral vectors were generated by infecting CRIP cells with the supernatant of transfected CRE cells in the presence of 8 g/ml polybrene (43). Transduced CRIP clones were selected in the presence of 0.4 mg/ml G418 (Life Technologies Inc.). HL60 promyelocytes were infected with the various retroviral expression vectors by cocultivation with CRIP producer cell lines and selected in the presence of 1 mg/ml G418 (9).
Flow Cytometry Analysis-Flow cytometry was performed on a Becton Dickinson FACScalibur. Labeling was performed by incubating cells with fluorochrome-conjugated antibodies for 2 h in phosphatebuffered saline with 0.5% bovine serum albumin at 4°C. Next cells were washed twice in cold phosphate-buffered saline and immediately analyzed on the flow cytometer. For fluorochrome-unconjugated antibodies, a second incubation with fluorescein-5-isothiocyanate-conjugated secondary antibodies was performed 1 h at 4°C in phosphate-buffered saline, bovine serum albumin 0.5%. Time course of antibody binding showed that equilibrium was reached for all antibodies after 90 min at 4°C, and binding was saturating at the concentration used (data not shown). Specificity of the binding was assessed by incubating the cells with a 10-fold excess of nonspecific immunoglobulins before the specific antibody (data not shown).

Measurements of Intracellular Free Ca 2ϩ Concentrations-[Ca 2ϩ
] c was measured with the fluorescent Ca 2ϩ indicator fura-2. Cells (2 ϫ 10 7 /ml) suspended in Ca 2ϩ medium containing 0.1% bovine serum albumin were loaded for 45 min at 37°C with 2 M fura-2/AM, then diluted to 10 7 /ml and kept on ice. Just before use, a sample of loaded cells (5ϫ10 6 /ml) was centrifuged and resuspended in the desired medium at 37°C. Fluorescence measurements were performed on a Per-kin-Elmer fluorometer (LS3, Perkin-Elmer) thermostated at 37°C. Excitation and emission wavelengths were 340 and 505 nm, respectively. Calibration was performed for each cuvette by the sequential addition of 4 -5 mM Ca 2ϩ (to reach a 5 mM final Ca 2ϩ concentration), 2 M ionomycin to measure Ca 2ϩ -saturated fura-2 (F max ), followed by 24 mM EGTA, 60 mM Tris, pH 9.3, and 0.1% Triton X-100 (TX-100) to measure Ca 2ϩ -free fura-2 (F min ). Results are expressed as concentration (nm) of intracellular free Ca 2ϩ .
Protein Analysis-For quantitative protein detection, cells were lysed in TX-100 1% buffer (25 mM Hepes, pH 7.4, 145 mM NaCl, 5 mM MgCl 2 , 1% sucrose, 1% TX-100 and a mixture of protease inhibitors) or alternatively in radioimmune precipitation buffer (9.1 mM dibasic sodium phosphate, 1.7 mM monobasic sodium phosphate, pH 7.4, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, and a mixture of protease inhibitors) for 20 min on ice. Cell lysates were next centrifuged 20 min at 4°C 1200 ϫ g to discard aggregates, and protein concentrations were determined using the BCA assay from Pierce. The proteins were then separated by SDS-polyacrylamide gel electrophoresis and transferred on nitrocellulose membranes. Western blot analysis was performed using the ECL kit from Amersham Pharmacia Biotech.
For immunoprecipitations, cells were solubilized 20 min on ice with either solubilization buffer A (50 mM Tris-HCl, 50 mM NaCl, 10 mM MgCl 2 , 1 mM EGTA, pH 7.4, 1% TX-100, 10 mM orthovanadate, and a mixture of protease inhibitors) or alternatively with radioimmune precipitation buffer (9.1 mM dibasic sodium phosphate, 1.7 mM monobasic sodium phosphate, pH 7.4, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 10 mM orthovanadate and a mixture of protease inhibitors). After a brief sonication, 400 -500 g of protein were adjusted to the same volume (800 l) and precleared for 1 h with 3 mg/ml protein A-Sepharose. The beads were discarded by centrifugation, and 1.5 g of anti-p56/59 hck antibodies were added to each sample overnight at 4°C with gentle mixing. The immune complexes were then precipitated by adding 3 mg/ml protein A-Sepharose. The beads were washed three times with the corresponding solubilization buffer and boiled 5 min. The immunoprecipitates were finally separated by SDSpolyacrylamide gel electrophoresis and transferred on nitrocellulose membranes. Western blot analysis was performed using the ECL kit from Amersham Pharmacia Biotech.

RESULTS
Myelomonocytic cells (especially macrophages) are important targets of HIV. It therefore seems very relevant that the HIV Nef protein interacts with a high affinity with Src-like PTKs of myelomonocytic cells. However, the effects of Nef in these cells have so far not been examined. We therefore decided to use HL60 cells as a myelomonocytic cell model to investigate this question. Indeed, HL60 cells cannot only be used in their undifferentiated form as promyelocytes but can also be induced to differentiate major subtypes of myelomonocytic cells (44,45). Differentiation with Me 2 SO, 1,25(OH) 2 D 3 , and PMA will lead to a granulocyte (46), monocyte (47,48), and macrophage phenotype (49), respectively. Retroviral transduction was used to generate HL60 cell lines stably expressing the wild type form of HIV1 Nef (Nef) or a mutated Nef protein (Nef (PXXP)4 Ϫ), where prolines 69, 72, 75, and 78 were replaced by alanines. Previous in vitro studies have shown that, although Nef interacts with high affinity with four members of the Src-like PTK family (Lck, Fyn, Hck, and Lyn), no such interactions are observed for the Nef (PXXP)4 Ϫ) mutant (14,(22)(23)(24)(25).
Nef Expression in HL60 Cell Lines-As assessed by Western blot analysis, Nef and (Nef (PXXP)4 Ϫ) were expressed to significant levels both in undifferentiated promyelocytes and differentiated cells (Fig. 1A). To verify whether the expressed Nef wild type and PXXP mutant were functional, we investigated the down-regulation of CD4 cell surface expression by cytofluorimetry. Indeed, it has been shown previously that the PXXP motif of Nef is not crucial for CD4 down-regulation (14,15). As shown in Fig. 1B, CD4 surface expression was efficiently downregulated by Nef in undifferentiated and differentiated HL60 cells showing that Nef was functional in this cellular background. The Nef (PXXP)4 Ϫ) mutant was also able to down-regulate CD4, but, as described previously in other cells (14,50), somewhat less efficiently than the wild type form. Thus, the expressed Nef and Nef (PXXP)4 ) are functional with respect to CD4 down-regulation.
Nef Decreases Agonist-induced Ca 2ϩ Influx through an Increase in Stored Ca 2ϩ -The cytosolic free Ca 2ϩ concentration, [Ca 2ϩ ] c , is a crucial intracellular signaling mediator in myelomonocytic cells (51,52). In response to specific agonists, [Ca 2ϩ ] c elevations occur in two different steps. First, Ca 2ϩ is released from intracellular stores in response to the generation of inositol phosphate (ϭCa 2ϩ release). Second, the emptying of intracellular stores activates, by a yet unknown mechanism, Ca 2ϩ channels in the plasma membrane (ϭCa 2ϩ influx). To investigate a potential effect of Nef on agonist-induced [Ca 2ϩ ] c elevations, fura-2-loaded HL60 cells were stimulated with the bacterial peptide fMLP. To separate the initial Ca 2ϩ release phase from the subsequent Ca 2ϩ influx phase, the following protocol was used. Cells were kept in a Ca 2ϩ -free medium. Under these conditions, the [Ca 2ϩ ] c increase in response to fMLP stimulation was used as a measure for Ca 2ϩ release. Subsequently, Ca 2ϩ was added to the extracellular medium, and the resultant [Ca 2ϩ ] c elevation was used as a measure of Ca 2ϩ influx (Fig. 2). In control cells, basal [Ca 2ϩ ] c was around 70 -80 nM. Ca 2ϩ elevations due to Ca 2ϩ release in response to 100 nM fMLP were 250 -350 nM. Ca 2ϩ influx, induced by Ca 2ϩ readdition, raised [Ca 2ϩ ] c to 300 -400 nM (Fig. 2B). Wild type Nef expression affected neither basal [Ca 2ϩ ] c nor Ca 2ϩ release from intracellular stores. By contrast, Ca 2ϩ influx was de-creased by approximately 50% in the presence of the viral protein (Fig. 2B). The Nef inhibition of Ca 2ϩ influx was also present when the lag time between fMLP addition and Ca 2ϩ readdition was reduced to 1 min, suggesting that activation of Ca 2ϩ channels is decreased by Nef already in the early phase of fMLP stimulation (not shown).
Since in myelomonocytic cells agonist-induced Ca 2ϩ influx is regulated by the filling state of intracellular Ca 2ϩ stores (53,54), the decrease in agonist-induced Ca 2ϩ influx in Nef-transfected cells might be due to an increase in residual Ca 2ϩ store content after stimulation. To test this hypothesis, we performed experiments essentially similar to the Ca 2ϩ influx protocol described above. However, instead of adding back Ca 2ϩ in the medium to measure Ca 2ϩ influx, the Ca 2ϩ ionophore ionomycin was added to assess the amount of residual Ca 2ϩ store content. Ionomycin-releasable Ca 2ϩ from internal stores was approximately double in wild type nef-expressing cells as compared with control cells (Fig. 2, C and D). This would be compatible with the concept that an increased Ca 2ϩ store content accounts for the decreased Ca 2ϩ influx in Nef-transfected cells.
Thapsigargin, an inhibitor of the Ca 2ϩ pumps of agonistsensitive Ca 2ϩ stores, is able to induce a depletion of intracellular Ca 2ϩ stores and a subsequent store-operated Ca 2ϩ influx independently from receptor activation and Ins(1,4,5)P 3 production. We therefore studied thapsigargin-induced [Ca 2ϩ ] c changes using protocols similar to those described above for fMLP. In contrast to fMLP, thapsigargin induced an equivalent Ca 2ϩ influx in nef-expressing and in control cells (Fig. 3, A and  B). The absence of inhibition of thapsigargin-induced Ca 2ϩ influx by Nef might have two possible explanations. (i) Either the presence of a receptor agonist (i.e. fMLP) is necessary to unravel the inhibitory effect of Nef on Ca 2ϩ influx, or (ii) after efficient depletion of intracellular Ca 2ϩ stores by thapsigargin, a normal activation of store-operated Ca 2ϩ influx occurs. When fMLP was coapplied with thapsigargin, no inhibition was observed (Fig. 3, C and D), excluding the former possibility. Instead, thapsigargin efficiently depleted intracellular Ca 2ϩ stores both in nef-expressing and in control cells (Fig. 3, E and  F), demonstrating that after efficient inhibition of the Ca 2ϩ -ATPase, (i) the Ca 2ϩ store contents are down to similarly low levels in both nef-expressing and control cells, and (ii) normal size Ca 2ϩ influx occurs.
Taken together, these results demonstrate that, after agonist stimulation, an abnormally high residual Ca 2ϩ store content prevents the normal activation of store-operated Ca 2ϩ influx in nef-expressing cells. The normal activation of Ca 2ϩ influx by thapsigargin excludes the possibility that Nef directly blocks store-operated Ca 2ϩ channels. The residual Ca 2ϩ after agonist stimulation is unlikely to be due to an inhibition of agonist-induced signaling to Ca 2ϩ stores, as the amplitude of fMLP-induced Ca 2ϩ release was identical in nef-expressing and in control cells. Thus, we considered the possibility that the Ca 2ϩ content of intracellular stores is constitutively increased in nef-expressing cells. To test this hypothesis, we directly added the Ca 2ϩ ionophore ionomycin to cells kept in a Ca 2ϩfree medium without any prestimulation. As shown in Fig. 4A, nef-expressing cells had indeed a markedly increased ionomycin-releasable Ca 2ϩ pool. Thus, we conclude that Nef is able to constitutively increase cellular Ca 2ϩ store content in differentiated HL60 cells.
The Nef-induced Increase in Cellular Ca 2ϩ Store Content Requires Factors Expressed upon Differentiation of Myelomonocytic Cells-To investigate the mechanisms underlying the Nefinduced increase in cellular Ca 2ϩ content, we analyzed the Nef effect in undifferentiated HL60 cells, in three different types of differentiated HL60 cells, and in CEM T-lymphocytes. As shown in Fig. 4B, an increase of Ca 2ϩ content ranging from 30 to 60% was observed in all types of differentiated HL60 cells, i.e. HL60 granulocytes (Me 2 SO), HL60 monocytes (1,25(OH) 2 D 3 ), and HL60 macrophages (PMA). In contrast, in undifferentiated HL60 promyelocytes and in CEM lymphocytes, no increase in Ca 2ϩ store content was observed.
Thus Nef appears to require the presence of differentiationinduced cofactors not present in T-lymphocytes to exert its action on intracellular Ca 2ϩ stores. Importantly, the two myelomonocytic cell types, which are thought to be relevant for HIV infection, namely monocytes and macrophages, clearly are sensitive to the Nef effect on Ca 2ϩ store content.
SH3-mediated Protein-Protein Interactions via the PXXP Motif in the Nef Core Is Required for the Nef Effect on Ca 2ϩ Stores-Myelomonocyte-specific Src-like PTKs would be good candidates as differentiation-induced cofactors that mediate the Nef effect on Ca 2ϩ signaling. Nef binds to four members of this PTK family (p56 lck , p56/59 hck , p59 fyn , and p53/56 lyn (14,(22)(23)(24)(25) with a high affinity via interactions between a prolinerich motif in the Nef core and SH3 domains of Src-like PTKs. The mutation of the Nef proline motifs prevents the binding of Nef to Src-like PTKs and their subsequent activation but does not abolish other Nef effects such as cell surface CD4 downregulation (Fig. 1B). We therefore studied Ca 2ϩ signaling in HL60 cells expressing a proline mutant of Nef (Nef (PXXP)4 ). As shown above, this mutant was well expressed and functional with respect to CD4 down-regulation (Fig. 1). As shown in Fig.  5, cells expressing the Nef (PXXP)4 mutant behaved like control cells. (i) fMLP-induced Ca 2ϩ influx was not diminished (Fig.  5A), and ii) the amount of ionomycin-releasable Ca 2ϩ was not increased either in unstimulated cells or after fMLP stimulation (Fig. 5B).
Up-regulation of Src-like PTKs during of Myelomonocytic Differentiation-Among the four Src-like PTKs able to interact with Nef, p56/59 hck , p53/56 lyn , and p56 lck are restricted in their expression to cells of the hematopoietic lineage; p56/59 hck expression is restricted to myelomonocytic cells (55), whereas p53/56 lyn is found either in myelomonocytic or in B-lymphocytic cells (56), and p56 lck is specifically expressed in lymphocyte. By contrast p59 fyn is more ubiquitously expressed. We thus investigated the expression of these four Src-like PTKs under our conditions of HL60 cell differentiation by Western blots analysis of total cell extracts. We did not detect p56 lck and p59 fyn in undifferentiated or differentiated HL60 cells (data not shown). By contrast p56/59 hck and p53/56 lyn were detected to varying amounts in HL60 cells depending upon the differentiation type (Fig. 6). p56/59 hck and p53/56 lyn were not expressed to detectable levels in CEM T cells (data not shown). The p56/59 hck -and p53/56 lyn -associated signals were further quantitated by densitometry in wild type and nef-or nef (PXXP)4 Ϫ -expressing HL60 cells (Fig. 6, B and C). In undifferentiated HL60 cells, only a very weak p56/59 hck -associated signal was detected. However, upon differentiation, the p56/59 hck signal strongly increased, reaching the highest level in Me 2 SO-differentiated cells. By contrast, the p53/56 lyn -associated signal was already relatively high in undifferentiated cells and showed only relatively small increases with differentiation as previously reported by others (57).
Note that, although the expression of Src-like PTKs was clearly differentiation-dependent, it was not influenced by the expression of Nef or the Nef mutant. Thus, although Nef might possibly modulate activity or cellular localization of Src-like PTKs, it does not influence protein levels of these kinases.
P56/59 hck Interacts in Vivo with Nef and the Ins (1,4,5) Receptor-Interactions of HIV1 Nef and Src-like PTKs have been clearly demonstrated in vitro (14). In vivo Nef-p56/59 hck interaction has been observed only in reconstituted cell systems overexpressing both Nef and p56/59 hck (58). The results presented so far would be compatible with a mediation of the Nef effect on Ca 2ϩ signaling by p56/59 hck and p53/56 lyn . To study whether Nef is indeed able to interact with these Src-like PTKs in myelomonocytic cells, we immunoprecipitated p56/59 hck and p53/56 lyn and then examined the immunocomplexes by blotting with Nef-specific antibodies. We first performed immunoprecipitations with lysates from differentiated HL60 cells using anti p56/59 hck antibodies. Coimmunoprecipitation of Nef was observed with cells expressing wild type Nef but not the Nef-(PXXP)4 mutant (Fig. 7). Using anti-p53/56 lyn antibodies, we were unable to detect any Nef-associated signal in the immu-noprecipitates (data not shown). Thus, Nef clearly interacts with p56/59 hck through its proline-rich motifs. The negative results concerning p53/56 lyn should be taken with caution. They might suggest that Nef does not interact in vivo with p53/56 lyn . Alternatively however, the interaction between Nef and p53/56 lyn might be of too low affinity to be detected by immunoprecipitation, or the antibody used could have disrupted the interaction.
Previous studies in lymphocytes have implicated nonreceptor tyrosine kinases in the control of Ca 2ϩ store regulation. More specifically, the Src-like PTK p59 fyn regulates the activity of the Ins(1,4,5)P 3 receptor in lymphocytes through direct interaction with this Ca 2ϩ release channel (42). We could not detect p59 fyn in HL60 cells (not shown). It is, however, conceivable that myelomonocyte-specific Src-like PTKs play a role in the regulation of Ca 2ϩ store function. We therefore examined whether p56/59 hck is able to interact with the Ins(1,4,5)P 3 receptor in HL60 cells. As shown in Fig. 7, the Ins(1,4,5)P 3 receptor coimmunoprecipitated with p56/59 hck . This observation provides a possible link between the known regulation of Src-like PTKs activity by Nef and the SH3-dependent regulation of myeloid cell Ca 2ϩ signaling demonstrated in our study. Note, however, that the amounts of Ins(1,4,5)P 3 receptor, which coimmunoprecipitated with p56/59 hck , were comparable in control and nef-transfected cells. This would be compatible with a concept where Nef regulates in a qualitative, rather than a quantitative manner, the association between p56/59 hck and the Ins(1,4,5)P 3 receptor. DISCUSSION In this study we show a novel mechanism whereby the HIV Nef protein manipulates cellular signaling; Nef expression in myelomonocytic cells leads to an increased Ca 2ϩ content of intracellular stores. The Nef PXXP motif was crucial for the observed effect. This suggests that proteins bearing SH3 domains are able to interact with Nef, such as Src-like PTKs, might act downstream to Nef to alter the regulation of intracellular Ca 2ϩ signaling pathways.
What Are the Downstream Mediators of the Nef Effect on Ca 2ϩ Signaling in Myelomonocytic Cells?-Our results allowed assembly of the following information concerning elements downstream of Nef involved in the increase in Ca 2ϩ store content.
The absence of the Nef effect in lymphocytes and undifferentiated HL60 cells suggests that factors induced by myelomonocytic differentiation are involved in the mediation of this effect.
The absolute requirement for the proline-rich regions of Nef suggests the involvement of SH3 domain-containing proteins, in particular Src-like PTKs.
The coimmunoprecipitation of Nef and the Ins(1,4,5)P 3 receptor with p56/59 hck makes the latter Src-like PTK a particularly attractive candidate.
Thus, our results are compatible with the following scheme. Nef interacts with Src-like PTKs (in particular p56/59 hck ), which in turn interact with elements of intracellular Ca 2ϩ stores (in particular the Ins(1,4,5)P 3 receptor). Further experiments will however be necessary to definitively prove this hypothesis.
Possible Molecular Mechanisms of the Nef-induced Increase in Ca 2ϩ Storage-The amount of Ca 2ϩ stored within intracellular Ca 2ϩ stores is essentially a function of two parameters, the activity of Ca 2ϩ uptake and the permeability of the Ca 2ϩ store membrane. Thus, either an increased Ca 2ϩ -ATPase activity or a decreased Ins(1,4,5)P 3 receptor activity are the most likely mechanisms accounting for the increase Ca 2ϩ storage observed in nef-expressing cells. The observation that p56/59 hck is not only coimmunoprecipitating with Nef but also with the Ins(1,4,5)P 3 receptor together with the known regulation of the Ins(1,4,5)P 3 receptor by Src-like PTKs in lymphocytes suggest the latter possibility. However, a modulation of the Ca 2ϩ -ATPase activity and thereby of the Ca 2ϩ uptake mechanisms of intracellular stores is an attractive alternative or additional hypothesis to explain the Nef effect. Note that Nef has also been shown to interact with a cellular H ϩ -ATPase (59).
Potential Physiological Role of Increased Ca 2ϩ Storage-What could be the potential physiological role of the observed increase in intracellular Ca 2ϩ storage? A most obvious explanation might be to allow cells to release increased amounts of Ca 2ϩ during stimulation, although we could not observe such an effect in nef-expressing cells. Indeed, after fMLP stimulation, an increased residual Ca 2ϩ store content rather than an increased Ca 2ϩ release, was observed. Thus, the relevant effect of the Nef-induced Ca 2ϩ increase should reside within other cellular events, which are controlled by the content of Ca 2ϩ stores. Indeed, there is a decrease in store-operated Ca 2ϩ influx in response to receptor activation in nefexpressing cells. As Ca 2ϩ influx is involved in a variety of crucial cellular functions, a decreased Ca 2ϩ influx might be a pathophysiologically relevant Nef effect. In addition, Nef could alter other cellular functions independent of an extracellular Ca 2ϩ influx but which may be regulated by the Ca 2ϩ content of the endoplasmic reticulum. Indeed, the Ca 2ϩ content of the endoplasmic reticulum plays an important role in protein synthesis (60,61), regulation of apoptosis (62,63), gene expression (64), and regulation of the nuclear pore activity (65). Future studies will have to address the question of how the Nefinduced increase in intracellular Ca 2ϩ storage and the subsequent potential effects relate to the specific properties of HIV infection in myelomonocytic cells.