hSos1 Contains a New Amino-terminal Regulatory Motif with Specific Binding Affinity for Its Pleckstrin Homology Domain*

The protein hSos1 is a Ras guanine nucleotide exchange factor. In the present study, we investigated the function of the amino-terminal region of the hSos1 protein, corresponding to the first 600 residues, which includes the Dbl and pleckstrin homology (DH and PH) domains. We demonstrated, using a series of truncated mutants, that this region is absolutely necessary for hSos1 activity. Our results suggest that the first 200 residues (upstream of DH domain), which we called the HF motif on the basis of their homology with histone H2A, may exert negative control over the functional activity of the whole hSos1 protein. In vitro binding analysis showed that the HF motif is able to interact specifically with the PH domain of hSos1. The amino-terminal region of hSos1 may be associatedin vivo with an expressed HF motif. These findings document the existence of the HF motif located upstream of the DH domain in the hSos1 protein. This motif may be responsible for the negative control of hSos1, probably by intramolecular binding with the PH domain.

The protein hSos1 is a Ras guanine nucleotide exchange factor. In the present study, we investigated the function of the amino-terminal region of the hSos1 protein, corresponding to the first 600 residues, which includes the Dbl and pleckstrin homology (DH and PH) domains. We demonstrated, using a series of truncated mutants, that this region is absolutely necessary for hSos1 activity. Our results suggest that the first 200 residues (upstream of DH domain), which we called the HF motif on the basis of their homology with histone H2A, may exert negative control over the functional activity of the whole hSos1 protein. In vitro binding analysis showed that the HF motif is able to interact specifically with the PH domain of hSos1. The amino-terminal region of hSos1 may be associated in vivo with an expressed HF motif. These findings document the existence of the HF motif located upstream of the DH domain in the hSos1 protein. This motif may be responsible for the negative control of hSos1, probably by intramolecular binding with the PH domain.
Sos guanine nucleotide exchange proteins mediate Ras activation induced by various tyrosine kinase receptors (1). Sos consists of several defined domains, each of which has a distinct function. For example, nucleotide exchange activity on Ras is mediated by a central domain of Sos that, among the various Ras guanine nucleotide exchange factors, is very well conserved (1) and for which the structure has been determined in complex with Ras (2). The carboxyl-terminal region of Sos is proline-rich and contains specific sequences (P⌿⌿PPR) that bind the SH3 (Src homology 3) domains of Grb2 (3,4). Finally, the amino-terminal region of Sos is ϳ600 amino acids long and contains regions of homology to Dbl (DH) 1 and pleckstrin (PH) domains involved in Rac1 activation (5,6) and phospholipid binding (7), respectively, for which the structures have also been determined (8,9).
Stimulation of cells with growth factor leads to the association of Sos-Grb2 complexes with activated receptors, leading to the stimulation of Ras through the juxtaposition of Sos and Ras at the membrane (1). Sos-Ras binding involves the switch 1 and switch 2 regions of Ras (2). Whereas the interaction with switch 2 mediates the anchoring of Ras to Sos, the interaction with switch 1 leads to the disruption of the nucleotide-binding site and GDP dissociation (10). In this model, both the cytosolic and membrane-bound Sos forms are thought to exhibit similar nucleotide exchange activity; no variation of this activity is thought to occur as a consequence of relocation inside the cell. In support of this idea, constitutive or conditional membrane targeting of these exchange factors has been shown to strengthen Ras activation in transfected cells (11,12). However, some reports suggest that regardless of sub-cellular location, the intrinsic Ras guanine nucleotide exchange activity of Sos (Ras-GEF activity) may differ before and after stimulation of surface tyrosine kinase receptors (13,14).
Several reports suggest that the carboxyl-terminal portion of Sos exerts negative regulation over the activity of Sos1 (15)(16)(17)(18)(19). However, the role of the amino-terminal region of Sos is not fully understood. Thus, some reports suggest that this region is responsible for positive regulation of Sos1 activity (17,18), basically through the cytoplasmic membrane localization of Sos1 protein. Nevertheless, stable membrane association of Sos by addition of a myristoylation signal to this protein, still lacking the amino-terminal region, is not sufficient for Sos to be biologically active (20). Others reports suggest that the aminoterminal portion of Sos1 is involved in the negative regulation of its catalytic activity and exerts negative allosteric control on the interaction of the Sos catalytic domain with Ras (15). Most studies of the amino-terminal region of Sos1 focus on the DH and PH domains and disregard other possibilities such as the peptide region upstream of DH.
The aim of the present study is to explore the function of the amino-terminal region of hSos1, specifically the small region (200 residues) upstream of DH domain, as well as to examine further its involvement in the regulation of hSos1 activity. Toward these ends, we characterized the effect of this region on biological and biochemical signaling events related to Ras activation. In addition, we analyzed its in vitro and in vivo spe-cific association with the PH domain of hSos1 and its function in the physiological activity of hSos1.

EXPERIMENTAL PROCEDURES
Cell Lines, Transfections, and Antibodies-NIH3T3 cells were maintained in Dulbecco's modified Eagle's medium (DMEM; Invitrogen) supplemented with 10% calf serum (Invitrogen). COS1 and the human 293T (kidney keratinocyte) cell lines were maintained in DMEM supplemented with 10% fetal calf serum (fetal calf serum, Invitrogen). Transient transfections in COS1 and 293T cells were performed with the LipofectAMINE reagent (Invitrogen). COS1 cells for serum starvation received DMEM containing 0.5% fetal bovine serum 24 h after transfection and were then incubated for another 24 h. All assays were done 48 h after transfection. NIH3T3 cells were transfected (transient or stable) by the calcium phosphate precipitation technique. Morphologically transformed foci were scored after 2-3 weeks in culture (21). Transfected cells were also selected in medium supplemented, as appropriate, with Geneticin 750 g/ml (Invitrogen). Monoclonal antibody to phospho-MAPK protein was purchased from New England Biolabs. Rabbit polyclonal antiserum to MAPK (ERK1/ERK2) and rabbit anti-GST polyclonal antibody came from Santa Cruz Biotechnology Inc., monoclonal anti-His, from Sigma, and anti-HA and anti-AU5 monoclonal antibodies, from the Berkeley Antibody Co.
Bacterial Expression of Fusion Proteins-S. J. Taylor (Cornell University) kindly provided the plasmid pGEX-RBD containing the Raf-Ras-binding domain (amino acids 1-149) fused to glutathione S-transferase (GST). The GST-RBD protein was purified (from Escherichia coli BL21(DE3) harboring that plasmid) following the method described previously (19). Similarly, the BL21(DE3) strain of E. coli was transformed with the vector pGEX-4T-1 encoding the fusion protein GST-NDP, GST-HF, GST-DH, GST-PH, GST-DH-PH (all containing domains of hSos1), or GST-PH2 (containing the PH domain 2 of Ras-GRF1) kindly provided by J. S. Gutkind. This bacterial strain was also transformed with the vector pGEX-4T-3 encoding the fusion protein GST-PH (containing the PH domain of PKD) kindly provided by T. Iglesias (22). Protein purification was performed according to the method described previously (21). The M15 strain of E. coli harboring plasmid pREP4 (Qiagen) was transformed with the vector pQE30 encoding the HF motif of hSos1 containing six consecutive histidine residues (the 6xHis tag) at their amino terminus. The 6xHis-HF peptide was purified as described previously (23).
HF in Vitro Binding Assays-Fifty pmol of 6xHis-HF protein, purified and eluted, were incubated with 50 pmol of either GST-NDP, GST-HF, GST-DH, GST-PH, GST-DH-PH of hSos1, GST-PH2 of Ras-GRF1, GST-PH of PKD, or GST proteins coupled to glutathione-Sepha-rose beads and incubated for 60 min at 4°C following the methodology described previously (21).
Reporter Gene Analysis-NIH3T3 cells were transfected with 0.6 g of constructs encoding for either hSos1 or Ras, together with 16 ng of pCDNAIII-Gal4-Elk-1, 0.1 g of pRL-TK (a plasmid expressing the enzyme Renilla luciferase), and 0.3 g of the reporter plasmid (pGal4-Luc). Cells were kept for 24 h in DMEM supplemented with 0.5% calf serum and 18 h later were stimulated for 8 h with 30% fetal calf serum. The assays were performed as described previously (19).

Overexpression or Deletion of the Amino-terminal Half of hSos1 (NDP) Impairs the Transforming Phenotype Induced by Normal Ras; the HF Motif of hSos1 Upstream DH Domain Is Critical for
This Biological Effect-To ascertain the function of the aminoterminal half of hSos1, we cloned this region (NDP, residues 1-600; NDP ϭ Nter ϩ DH ϩ PH) in a mammalian expression vector and also generated a truncated mutant of this region (⌬NDP-hSos1). The NDP region includes the DH and PH domains and a small amino-terminal (Nter) region (residues 1-200) upstream of DH. This amino-terminal region displays very close sequence similarity to histone H2A (24) (the core of homology are residues 90 -173), denoted by us as HF (histone fold) motif ( Fig.  1). A classical effect of the Ras pathway activation is to elicit formation of transformed foci in NIH3T3 cells. We cotransfected NIH3T3 cells with the hSos1 constructs and H-Ras wt as described above and assessed the transformed foci (Fig. 1). The difference in transforming activity between hSos1 wt and the NDP truncated mutant was shown in experiments in which the hSos1 constructs were cotransfected with normal ras genes. The over-expression of H-Ras wt alone produced weak but reproducible transforming activity, which was enhanced severalfold when hSos1 wt was included in the cotransfection experiments ( Fig. 1). However, over-expression of ⌬NDP-hSos1 ( Fig. 1A) or NDP region ( Fig. 1B) together with H-Ras wt consistently resulted in a significant reduction in the number of transformed foci produced by normal H-Ras alone. These results suggest that hSos1 needs its amino-terminal half (NDP region) for the synergistic effect with H-Ras wt. This NDP region was involved in the positive control of hSos1 function, in agreement with previously published observations on the regulatory effect of the amino-terminal region of Sos (17,18,20).
To determine whether the ability to reduce the number of transformed foci induced by normal H-Ras can be assigned to any particular region of NDP, we over-expressed the different domains (HF, DH-PH, DH, and PH) together with H-Ras wt. Thus, ectopic expression of HF, DH-PH, and DH domains led to the same effect as the complete NDP region, inhibiting the transforming activity of H-Ras wt (Fig. 1B). In sharp contrast, overexpression of the PH domain had no effect on the number of transformed foci (Fig. 1B). Further, cotransfection of the truncated mutant ⌬HF-hSos1 (hSos1 without HF motif) together with H-Ras wt more actively induced focus formation than hSos1 wt (Fig. 1A). The ectopic expression of the different hSos1 constructs did not affect the expression levels of endogenous Sos1 (data not shown). Taken together, these results suggest that the HF motif of hSos1 protein is involved in the negative control of hSos1 activity.
Overexpression or Deletion of the NDP Region (or Their Domains) of hSos1 Impairs MAP Kinase Activation Induced by Mitogenic Stimulation-We confirmed the negative effect of  Fig. 1. The transfected cells were serum-starved for 18 h and then treated without (Ϫ lanes) or with epidermal growth factor (EGF; 100 ng/ml, 10 min; ϩ lanes). Cell lysates were prepared, and equal amounts of protein were resolved on 10% SDS-PAGE and immunoblotted for activated MAPK using an antibody that recognizes phosphorylated activated MAPK. Immunoreactive bands were viewed by ECL. Activated MAPK (p-ERK1 and p-ERK2) are indicated in the upper autoradiograms (IB: ␣-p-ERK). Equal protein loading was confirmed by immunoblotting for MAPK protein levels (IB: ␣-ERK). The expression levels of the transfected HA-hSos1 constructs and AU5-H-RasV12 were detected by immunoblotting the cell extracts with the corresponding anti-tag monoclonal antibody the HF motif on Sos activity by evaluating its function in the signaling pathways downstream of Sos. Specifically, we investigated whether over-expression of the NDP region or its different domains (HF, DH-PH, DH, and PH) affected the Raf-MEK-MAPK pathway. Therefore, 293T cells were transiently transfected with full-length hSos1 (hSos1 wt), the truncated mutants ⌬HF-hSos1 and ⌬NDP-hSos1, the NDP region, or their different domains. After serum starvation, cells were stimulated with epidermal growth factor (EGF, Fig. 2A). As a positive control we used the oncogenic version of H-Ras (H-RasV12). The results show that over-expression of full-length hSos1 induces, under starved conditions, a level of ERK activation comparable with that seen for naive cells stimulated by epidermal growth factor, whereas ⌬HF-hSos1 appears more potent than full-length hSos1 in eliciting this response. However, the over-expression of ⌬NDP-hSos1, NDP region, HF, DH-PH and DH domains, but not the PH domain, led to a significant reduction of activated ERK1/ERK2 ( Fig. 2A). This inhibitory effect on the MAP kinase pathway produced by overexpression of NDP occurs upstream of Ras because the cotransfection of H-RasV12 and NDP induced similar levels of activated ERK1/ERK2 as H-RasV12 alone ( Fig. 2A). The same results were obtained with ⌬NDP-hSos1 mutant and by the HF, DH-PH, or DH domains (data not shown).
To confirm the results presented above, the NDP region and its domains were studied for their ability to affect the MAPK pathway. We used a reporter assay in NIH3T3 cells cotransfected with hSos1 or H-RasV12 constructs, together with a chimerical Gal4-Elk1 transcription factor and the reporter plasmid TATA-Gal4-Luc. Fig. 2B shows the results obtained in a set of experiments in which we measured the induction of luciferase activity under starved and serum-stimulated conditions. In accordance with the above p44/p42 MAP kinase results, we also detected inhibition of serum-induced Gal4-Elk1 activation by the NDP region of hSos1 or its domains, excepting the PH domain. In the case of NDP, the inhibitory effect was also detected with a version constitutively targeted to the cell membrane (NDP-myristoylated) (Fig. 2B). Finally, according to the phospho-ERK detection assays, over-expression of the NDP truncated mutant of hSos1 (⌬NDP-hSos1) strongly inhibited serum-induced Gal4-Elk1 activation (Fig. 2C), whereas the truncated mutant ⌬HF-hSos1 was more potent than hSos1 wt in inducing Gal4-Elk1 activation (Fig. 2C).
The HF Motif Has a Negative Regulatory Effect on the Activation of Ras Induced by hSos1-Because the inhibitory effect on the MAP kinase activation produced by the over-expression of NDP occurs upstream of Ras, we investigated whether the overexpression of the NDP region of hSos1 (or its domains) affects Ras activation induced by mitogenic stimulation. As a read-out, we measured the level of Ras-GTP in transient cotransfectants overexpressing hSos1 or Ras wt. After transient transfection, COS1 cells were serum-starved for 18 h and then stimulated with fetal calf serum (Fig. 3). The Ras-GTP levels were detected using a nonradioactive Ras-GTP detection assay (19). Fig. 3A shows representative results wherein we compared the levels of Ras-GTP in transient cotransfected COS1 cells expressing AU5-K-Ras wt and HA-tagged constructs of either full-length hSos1 or the NDP region (or their domains), under both basal and mitogenic conditions. As controls, COS1 cells harboring AU5-K-Ras wt alone or AU5-K-RasV12 were analyzed in parallel. As expected, analysis of GTP-bound Ras in cell lysates expressing full-length hSos1 showed Ras activation under basal conditions (Fig. 3A). However, the over-expression of the NDP region of hSos1 or any of its domains, with the exception of the PH domain, inhibited AU5-K-Ras wt activation induced by serum stimulation (Fig. 3A). Again, this inhibition was also observed with a version of NDP constitutively targeted to the cell membrane (NDP-myristoylated) (Fig.  3, A and B). This assay was performed with the three types of mammalian Ras (K-, H-, and N-Ras). Fig. 3B summarizes the results obtained when we analyzed the AU5-Ras-GTP/AU5-Ras levels under basal and stimulated conditions. In line with the focus formation and MAP kinase results, whereas Ras activation was completely blocked by the over-expression of NDP and HF peptides, the levels of Ras-GTP upon serum stimulation were unaffected by ectopic expression of PH domain (Fig. 3B). Nevertheless, the inhibitory effect on Ras activation due to the over-expression of DH-PH and DH domains was more clearly detected with K-Ras than with H-Ras or N-Ras (Fig. 3B).
Given that the truncated mutant ⌬HF-hSos1 was more potent than the full-length hSos1 inducting transforming foci ( Fig. 1) and MAP kinase activation (Fig. 2), we decided to determine whether this mutant was also more efficient in inducing Ras activation. To this end, COS1 cells were cotransfected with AU5-H-Ras wt and either full-length HA-hSos1 or HA-⌬HF-hSos1, and Ras-GTP levels were detected as described above. As controls, COS1 cells transfected with AU5-H-Ras wt alone or AU5-H-RasV12 were analyzed in parallel. Analysis of GTP-bound Ras in cell lysates expressing either full-length hSos1 or ⌬HF-hSos1 showed Ras activation under basal conditions (Fig. 4A). The amounts of AU5-H-Ras-GTP under basal and stimulated conditions, standardized to AU5-H-Ras levels (Fig. 4B), in the cells containing HA-hSos1 wt were higher than in cells transfected with AU5-H-Ras wt alone but lower than in cells containing AU5-H-RasV12. Interestingly, under the same basal and stimulated conditions, HA-⌬HF-hSos1 induced Ras-GTP levels to a greater extent than full-length hSos1 (Fig. 4B), which is in agreement with the results observed in focus formation and MAP kinase activation assays.
The HF Motif Binds Specifically to the PH Domain of hSos1 in Vitro as Well as in Vivo-Taken together, the above results suggest that the HF motif of hSos1 exerts a negative regulatory effect on hSos1 activity. Because the NDP region (containing HF, DH, and PH domains) is necessary for hSos1 function, a possible mechanism to explain this negative effect could be through molecular interactions of the HF motif with the domains of the NDP region. To test this hypothesis, we analyzed the in vitro interaction between HF and NDP as well as with each one of its corresponding domains. The NDP region or its domains (HF, DH, DH-PH, PH) were expressed as GST fusion proteins, and the HF motif was also expressed as a 6xHis fusion peptide. The purified 6xHis-HF peptide was incubated with similar amounts of purified GST, GST-HF, GST-DH, GST-DH-PH, GST-PH, and GST-NDP proteins coupled to glutathi-(two lower autoradiograms). Results are from a representative experiment with similar results obtained in three additional, separate experiments. B and C, NIH3T3 cells were cotransfected with the plasmid pcDNAIII-Gal4-Elk-1 encoding the Gal4-Elk1 fusion protein (containing the Gal4 DNA-binding domain and the activation domain of Elk1), the plasmid pGal4-Luc containing the Photinus luciferase gene controlled by six copies of a Gal4-responsive element, and the plasmid pRL-TK containing the Renilla luciferase gene under control of the HSV-TK promoter region, together with expression plasmids pCEFL-KZ-HA containing the indicated hSos1 constructs described. The transfected cells were serum-starved for 18 h and then treated without (Ϫ lanes) or with calf serum (30%, 8 h; ϩ lanes) and assayed for luciferase activity. The data represent Photinus luciferase activity standardized for the Renilla luciferase activity present in each cellular lysate, expressed as fold induction compared with control cells. The data give the average and standard deviation of four separate assays performed in triplicate. one-Sepharose beads, and the proteins bound to the beads were analyzed by immunoblotting with antibodies to 6xHis (Fig. 5A). Whereas purified GST alone, GST-HF, and GST-DH (Fig. 5A) did not bind any HF, high amounts of 6xHis-HF bound to GST-PH beads (Fig. 5A). Consistent with these results, GST-DH-PH and GST-NDP proteins (both containing PH domain) also bound HF (Fig. 5A). The yeast two-hybrid approach (data not shown) gave the same results.
The specificity of the in vitro interaction between the HF motif and PH domain of hSos1 was analyzed by comparing the in vitro binding of HF to the PH domains of Ras-GRF1 (PH2 domain) (1) and PKD (22). To this end, the purified 6xHis-HF peptide was incubated with similar amounts of purified GST-PH (hSos1), GST-PH2 (Ras-GRF1), or GST-PH (PKD) proteins coupled to glutathione-Sepharose beads; and the bound proteins were detected as described above (Fig. 5B). The 6xHis-HF peptide was found to bind only to GST-PH (hSos1) but not to GST-PH2 (Ras-GRF1) or GST-PH (PKD) proteins.
To extrapolate these results to an in vivo situation, we carried out transient cotransfections of COS1 cells with the plasmids pCEFL-KZ-HA-NDP (coding for the epitope-tagged (HA) NDP region of hSos1), together with either pCEFL-KZ-AU5-HF (coding for the epitope-tagged (AU5) HF motif of hSos1), or pCEFL-KZ-AU5 (as negative control). Cellular lysates and anti-AU5 immunoprecipitates obtained under starving or stimulated cellular conditions were further analyzed by immunoblotting with anti-HA antibodies. We consistently detected HA-NDP coimmunoprecipitated with AU5-HF (Fig. 6). The immunoblot analyses with anti-HA demonstrated that HA-NDP is associated with AU5-HF, mainly under stimulated conditions (Fig. 6). Similar results were observed in 293T cells (data not shown). All of these results suggested that the NDP region of hSos1 may establish in vivo stable complexes with the HF motif, depending on mitogenic conditions.  Fig. 1. The transfected cells were serum-starved for 18 h and then treated without (Ϫ lanes) or with fetal calf serum (30%, 10 min; ϩ lanes). As described in the legend for Fig. 3, Ras-GTP was recovered from cell lysates by binding to immobilized GST containing the Ras-GTP binding domain of Raf and detected by immunoblotting with anti-AU5 monoclonal antibody (top autoradiogram). As control, the filter was stripped and reblotted against polyclonal GST antibody (bottom autoradiogram). The expression levels of the transfected HA-hSos1 constructs and AU5-H-Ras were detected by immunoblotting of the cell extracts with the corresponding anti-tag monoclonal antibody (middle autoradiograms). Results are from a representative experiment with similar results obtained in three additional, separate experiments. B, quantitation of AU5-H-Ras-GTP standardized (by GelWoks Analyses) to AU5-Ras levels for the experiments indicated in A. The histogram represents the average and standard deviation of four separate assays.

FIG. 5. In vitro binding of the 6xHis-HF peptide to GST-PH (hSos1).
A, 50 pmol of purified 6xHis-HF peptide were incubated with 50 pmol of purified GST, GST-HF, GST-DH, GST-PH, GST-DH-PH, and GST-NDP proteins coupled to glutathione-Sepharose beads. After washing, proteins bound to the beads were run in SDS and immunoblotted against monoclonal 6xHis antibody. As control, the filter was stripped and reblotted against polyclonal GST antibody. The autoradiogram shown is from a representative experiment that was repeated four more times with similar results. B, 50 pmol of purified 6xHis-HF peptide were incubated with 50 pmol of purified GST, GST-PH (hSos1), GST-PH 2 (Ras-GRF1), or GST-PH (PKD) proteins coupled to glutathione-Sepharose beads. After washing, proteins bound to the beads were analyzed as described in A. The autoradiogram is from a representative experiment repeated four times with similar results.

DISCUSSION
The amino-terminal region of hSos1 (NDP region) contains DH and PH domains involved in Rac1 activation (5) and phospholipid binding, respectively. In addition, it also contains a small region (residues 1-200) upstream of DH displaying high sequence similarity with histone H2A (24), which we denoted here as the HF motif. Ectopic over-expression of the NDP region and its domains (HF, DH-PH, and DH) in mammalian cells blocks Ras, MAPK activation, and cellular transformation induced by normal Ras. In addition, using a NPD deletion mutant of hSos1, we have shown here that this region is absolutely necessary for MAPK activation and cellular transformation. This suggests, on the basis of the published observations (17,18,20), that this region may be responsible for the positive regulation of Sos1 activity. However, ectopic overexpression of the PH domain does not have any inhibitory effect on the Ras pathway, in agreement with the observed effects of isolated PH domain peptide on the insulin-dependent induction of germinal vesicle breakdown in Xenopus oocytes (23). In apparent contrast with these results, Karlovich et al. (18) showed that the PH domain of Drosophila melanogaster interfered with Sos-dependent eye development in the fly. However, the same group reported that deletion of the PH domain from wild-type Drosophila Sos did not inhibit its ability to activate Ras in COS1 or 293 cells (16). Although, at least in our hands, ectopic overexpression of the isolated PH domain did not inhibit Ras activation, this does not mean that this domain is unnecessary for Sos1 activity. The PH domain isolated from hSos1 has been shown to associate with the plasma membrane in a serum-dependent manner (7) and is localized to the leading edges of motile cells rather than being randomly distributed in the membrane. As we explain further on, the PH domain interacts with the HF motif, which makes our results not incompatible with a positive role of the PH domain in the physiological function of Sos (18,20). Previously, several reports have been published using mutational analysis (deletions or point mutations) of the DH and PH domains in the context of full-length Sos (16,18,20). All known DH domains have associated carboxyl-terminal PH domains involved in targeting and regulatory functions. Furthermore, crystallographic data suggest that the tandem DH-PH is closely interconnected (8,9). In the case of hSos1, the entirely ␣-helical DH domain is unrelated in architecture to other nucleotide exchange factors (9). The active site of the DH domain, identified on the basis of sequence conservation and structural features, lies near the interface of the DH and PH domains (9), and mutations in DH or PH could affect function in the other domain. For this reason, the inhibition of Sos activity observed in DH or PH mutants (16,18,20) could be due to a conformational distortion of its PH or DH partner, respectively. Because of this possibility, we used ectopic over-expression of isolated DH, PH or DH-PH domains. We found to our surprise that the inhibitory effect on Ras activation upon ectopic over-expression of DH-PH or DH domains was more evident with K-Ras than with H-Ras and N-Ras. Although we still have not explained these differences, the explanation could lie in the fact that the inhibitory effects were related to differences in how Ras proteins are routed to the plasma membrane and where they localize once they get there (25).
To date, the HF motif is a region that is absolutely unknown in the context of Sos1 activity. We found that deletion of the HF motif increases the hSos1 functionality measured by MAPK activation, Ras-GTP levels, or assessing its cooperative effects on the transforming activity of Ras wt. Taken together, these results and those presented above indicate inhibitory behavior of the isolated HF motif, suggesting that this domain is involved in negative regulation of Sos1 activity. No crystallographic data on the HF motif of Sos1 exist yet, but this small region seems to be independent of the DH-PH domain (9). The negative effect of the HF motif could explain the results of Corbalan-Garcia et al. (15), who found that the amino-terminal portion of Sos1 was involved in the negative regulation of its catalytic activity (15). Our results indicate that the HF motif exhibits significant and specific binding affinity for the PH A, three-dimensional model of the core of the HF motif with homology to histone H2A. This model was derived from the sequence alignment between the hSos1 amino-terminal end and histone H2A (24). The structure of H2A was taken from the nucleosome structure (32). Shown in green are those residues that, in histone H2A, bury more than 10% of their solvent-accessible surface area in protein-protein interactions. The remaining residues are shown in magenta. B, accessibility plot of the hSos1 HF residues aligned with H2A histone, as derived from the known alignment (24). These accessibility changes are associated with the existence of protein-protein interactions. For purposes of clarity, we labeled only those residues that buried more than 50% of their solventaccessible surface area.
FIG. 6. Analysis of HF-NDP complexes in vivo. COS1 cells were transiently cotransfected with pCEFL-KZ-HA-NDP (coding the NDP region of hSos1) together with either pCEFL-KZ-AU5-HF (coding the HF motif of hSos1) or pCEFL-KZ-AU5 (as negative control). The transfected cells were serum-starved for 18 h and then treated without (Ϫ lanes) or with fetal calf serum (30%, 10 min; ϩ lanes). Cell extracts were incubated with anti-AU5 monoclonal antibodies. The anti-AU5 immunoprecipitates or whole cell extracts were then analyzed by immunoblotting using anti-HA monoclonal and anti-AU5 antibodies as described under "Experimental Procedures." The autoradiogram shown is from a representative experiment repeated three times with similar results. domain of hSos1. However, the HF motif does not bind to the DH domain, and neither does it bind to itself. The homology shown by the HF motif with histone H2A (24) (involving residues 90 -173 of hSos1) suggests that HF and histone H2A have a similar structure. Histone H2A is an ␣ protein with a relatively simple structure (Fig. 7A). In this structure a long helix occupies an axial position, with one short helix at each end. As a result of this organization, a central cleft, which can be used for interactions with other molecules, remains. In fact, visual inspection of different protein complexes involving the H2A fold shows that the central cleft is probably a good binding site. This is also confirmed in the case of the nucleosome by accessibility computations (Fig. 7, A and B), which is an indicative that central helix residues are involved mainly in proteinprotein contacts. In particular, helices from other proteins tend to lie against the long H2A helix through this cleft. This suggests that the observed interaction between the hSos1 HF and PH domains could be caused by binding of the PH to the cleft in the HF motif. In vivo assays showed that the isolated HF motif originated a stable complex with NDP upon mitogenic stimulation. Interestingly, this HF-NDP complex was not detected under starvation conditions. We propose a hypothetical model, based on our data, that may help to envisage the role of the HF motif in hSos1 activity regulation. In a basal state the HF motif could inhibit Sos activity by intramolecular binding to the PH domain. This would prevent the association of the DH-PH region with an unknown molecule(s) required for the efficient activation of Sos (18,20). Upon stimulation of the cells by engagement of mitogenic receptors, Sos1 function is up-regulated by events that weaken the contacts between the HF motif and the PH domain. This in turn would allow the interaction of the DH-PH region with specific targets necessary for Ras activation. Oligomerization of Ras-GRFs through their DH domains was also reported (26); although, until now, oligomerization of Sos1 has not been described, this possibility cannot be ruled out.
Similar autoinhibitory regulation has been described in others GEF proteins. Indeed, the autoinhibitory mode of regulation of proto-Dbl involves an intramolecular interaction between its amino-terminal sequences and the PH domain (27). The amino-terminal region of proto-Vav forms an ␣-helix that binds directly to the DH domain, blocking the access to Rho GTPases (28). Furthermore, mammalian p115RhoGEF becomes activated upon G␣ 13 binding to its amino-terminal domain, suggesting that this interaction may relieve the intrinsic constraint of the DH domain (29).
There are several possibilities as to how the HF motif-PH domain interaction could be affected. Phosphorylation events that modify either the HF motif or the PH domain may relieve the structural constraint of the HF motif-PH domain interaction. However, serine/threonine rather than tyrosine phosphorylation would be more likely to play a role in this effect, because no tyrosine phosphorylation was detected in the NDP region of hSos1. 2 Likewise, another potential mechanism may involve the interaction of either the HF motif or the PH domain with some docking/scaffold complex proteins upon mitogenic cell stimulation. This occurs with Cdc24 (GEF in budding yeast), which needs to be recruited to the targeting site via interaction with Far1, a scaffolding protein that recognizes a conserved motif found in the amino-terminal region of Cdc24 (30). Finally, the phosphoinositol lipids may affect the HF motif-PH domain interaction. The Sos1 PH domain shows an ϳ5-fold higher affinity for phosphatidylinositol 3,4,5-trisphos-phate than for phosphatidylinositol 4,5-diphosphate (31), and the specific phosphoinositide composition in a local environment could trigger conformational changes of the PH domain that may eventually reduce the HF-PH binding affinity. Some reports suggest that, irrespective of the subcellular location, the intrinsic Ras-GEF activity of Sos may be different before and after stimulation of surface tyrosine kinase receptors (13,14). Our proposed model for HF-PH interactions may help to reconcile these two apparently contradictory views. Whether this model is applicable to other Ras exchange factors is currently being studied, as is the identification of the putative proteins that interact with the HF motif upon mitogenic stimulation.