|Home | About | Journals | Submit | Contact Us | Français|
The dbl oncogene encodes a prototype member of the Rho GTPase guanine nucleotide exchange factor (GEF) family. Oncogenic activation of proto-Dbl occurs through truncation of the N-terminal 497 residues. The C-terminal half of proto-Dbl includes residues 498 to 680 and 710 to 815, which fold into the Dbl homology (DH) domain and the pleckstrin homology (PH) domain, respectively, both of which are essential for cell transformation via the Rho GEF activity or cytoskeletal targeting function. Here we have investigated the mechanism of the apparent negative regulation of proto-Dbl imposed by the N-terminal sequences. Deletion of the N-terminal 285 or C-terminal 100 residues of proto-Dbl did not significantly affect either its transforming activity or GEF activity, while removal of the N-terminal 348 amino acids resulted in a significant increase in both transformation and GEF potential. Proto-Dbl displayed a mostly perinuclear distribution pattern, similar to a polypeptide derived from its N-terminal sequences, whereas onco-Dbl colocalized with actin stress fibers, like the PH domain. Coexpression of the N-terminal 482 residues with onco-Dbl resulted in disruption of its cytoskeletal localization and led to inhibition of onco-Dbl transforming activity. The apparent interference with the DH and PH functions by the N-terminal sequences can be rationalized by the observation that the N-terminal 482 residues or a fragment containing residues 286 to 482 binds specifically to the PH domain, limiting the access of Rho GTPases to the catalytic DH domain and masking the intracellular targeting function of the PH domain. Taken together, our findings unveiled an autoinhibitory mode of regulation of proto-Dbl that is mediated by the intramolecular interaction between its N-terminal sequences and PH domain, directly impacting both the GEF function and intracellular distribution.
The proto-Dbl protein is the prototype member of a large family of guanine nucleotide exchange factors (GEFs) for Rho GTPases (8, 50). Oncogenic activation of proto-Dbl occurs by truncation of the amino-terminal 497 residues (41), resulting in constitutively active carboxyl-terminal sequences that include a Dbl homology (DH) domain in tandem with a pleckstrin homology (PH) domain, the conserved motifs of the Dbl family. Many members of this family, including Vav, Ect2, Tim, Ost, Dbs, Lbc, Lfc, Lsc, and Net, possess transformation or invasion ability, similar to onco-Dbl upon activation. In many cases, the DH-PH module represents the minimum structural unit that is required for cell transformation (8, 50).
A large body of evidence has helped establish that the biological functions of Dbl family members are intimately dependent upon their ability to interact with and activate Rho GTPases and that the cellular effects of Dbl-like proteins, including actin cytoskeletal reorganization, cell growth stimulation, and transformation, are likely the consequences of coordinated action of their immediate downstream substrates, the Rho family GTPases (8, 47, 50). The evidence includes the findings that Dbl family oncoprotein-induced foci are morphologically similar to those transformed by constitutively activated Rho GTPases but distinct from that seen when cells are transformed by Ras, Raf, or Src (23); coexpression of Dbl family members with dominant negative mutants of Rho family GTPases blocks their transforming activity (20, 23, 32, 52); mutants of the GEFs that are no longer able to interact or activate Rho protein substrates behave dominant-negatively in cells (46, 54); and many cellular activities induced by Dbl family proteins, such as actin cytoskeleton reorganization, c-Jun kinase (JNK) activation, SRF transcriptional activation, and NF-κB activation, are associated with the activation of signaling pathways known to be mediated by the Rho GTPase effector targets (24, 30, 36, 48). Therefore, the ability to interact and activate Rho proteins is essential for Dbl family functions.
Current biochemical and structural data have pointed to the conserved structural motif of the Dbl family, the DH domain, as the primary interactive site with Rho GTPases (2, 20, 31, 44, 54). The DH domain does not have significant sequence homology with other subtypes of small GTPase activators such as the Cdc25 domain and Sec7 domain, which are specific to Ras and ARF, respectively (6, 14), indicating that the DH-Rho protein interaction employs a distinct mechanism (9). Deletions or mutations within the DH domain have been reported to result in loss of GEF activity and cellular functions by the GEFs (20, 40, 43, 54, 55), suggesting that an intact DH domain, likely its Rho GTPase-interactive ability, is critical for the cellular effects of Dbl family members.
The invariable location of a PH domain immediately C-terminal to the DH domain of the Dbl family GEFs suggests a functional interdependence between the two domains. Indeed, a regulatory role of the PH domain in the function of Dbl family members has been recognized. Derivatives of the Dbl family members onco-Dbl, Lbc, Lfc, and Dbs that are truncated within the PH domain are impaired in their transforming activity (38, 48, 49, 53). In these cases, the PH domain was found to promote the translocation of the Dbl family proteins to the plasma membrane or cytoskeleton, where the Rho GTPase substrates reside. It is therefore likely that the PH domain of the Dbl proteins, acting similarly to the SH2/SH3 domains in the Ras pathway (10, 39), serves to bring the catalytic DH domain to specific intracellular locations to effectively activate the Rho GTPases.
Many members of the Dbl family appear to exist in an inactive, basal state prior to full activation. The incoming upstream signals, such as the heterotrimeric G-protein Gα and Gβγ subunits, protein tyrosine or serine/threonine kinases, adaptor or scaffolding proteins, and phosphoinositol lipids, may contribute in varying degrees to GEF activation processes (12, 13, 18, 19, 26, 34, 46). The best-understood example of self-regulation among the family members is the proto-Vav protein. The N-terminal autoinhibitory extension of proto-Vav forms an α-helix that binds in the DH domain active site through direct contact with the Rho GTPase binding pocket, blocking access to GTPases (3). Phosphorylation of Tyr174, which is an integral part of the autoinhibition interface, by Syk or Src-like kinases causes the N-terminal peptide to become unstructured and released from the DH domain, resulting in proto-Vav activation (3). The yeast Dbl family member Cdc24, which is a Cdc42-specific GEF, forms a protein complex with the scaffolding molecule Far1 and the Gβγ subunits to mediate the mating response of Saccharomyces cerevisiae (34). Mammalian p115RhoGEF becomes activated as a Rho GEF upon Gα13 binding to its N-terminal RGS domain, suggesting that the coupling between a Gα and p115Rho GEF may relieve the intrinsic constraint of the DH domain (19). Moreover, phosphorylation of the Rac1-specific GEF Tiam1 by Ca2+/calmodulin-dependent protein kinase II has been shown to lead to its translocation to the plasma membrane and activation (13), possibly by interference of the PH domain function of Tiam1, which has previously been demonstrated to determine its subcellular location (45). These cases suggest that the Dbl family GEFs employ a diverse range of self-regulatory mechanisms to maintain themselves in the basal state.
Proto-Dbl activation occurs through truncation of N-terminal 497 amino acids (42), suggesting that the N-terminal half of the molecule contains a negative regulatory element(s) for the C-terminal DH-PH functional module. A previous database search found limited similarities between the N terminus of proto-Dbl and the intermediate filament protein vimentin, spanning a 300-amino-acid region which was predicted to consist of an extended α-helical coiled-coil structure (41). However, where the inhibitory function resides upstream of the DH domain (residues 498 to 690) and how the N terminus exerts the inhibitory function remain unclear. In the present article, we report the finding that proto-Dbl protein involves an intramolecular interaction between the N terminus and the PH domain to maintain an autoinhibited, inactive state. The N- and C-terminal domain interaction effectively limits the access of the Rho GTPase substrates RhoA and Cdc42 to the catalytic site of the DH domain and masks the intracellular targeting function of the PH domain, resulting in suppression of its GEF function and a unique perinuclear localization pattern in cells. Such an autoinhibition state prevents proto-Dbl from transforming cells, and presents a basal mode that could be subject to modulation by a variety of upstream signals.
Constructs pZipneo-onco-Dbl, pZipneoGST-DH-PH, pKH3-DH-PH, pKH3-Cdc42, and KH3-RhoA were generated as described before (54). Various proto-Dbl truncation mutants, including T1-T7, N1-N4, and the DH and PH domains (Fig. (Fig.1A),1A), were generated by PCR cloning using the high-fidelity Pfu DNA polymerase (Stratagene) as described (28). The resulting constructs in the pZipneo vector were subsequently sequence proofed by automated fluorescence sequencing. The cDNAs encoding T1-T7, DH-PH, DH, or PH were subcloned into the pKH3 vector for expression as the trihemagglutinin (HA3)-tagged proteins in Cos-7 cells. The Myc-tagged N2 and Flag-tagged N1 constructs were generated by subcloning the corresponding cDNA sequences to the pCMV6 and pCMV2B vectors, respectively. The BamHI fragments encoding T1, N1, and the DH-PH module were also subcloned into the BglII and BamHI sites of pVL1392 vector together with the cDNAs encoding the glutathione S-transferase (GST) or His6 sequences for insect cell expression (51). The N2, N3, and N4 cDNAs were subcloned into the BamHI and EcoRI sites of the pGEX-2T vector for expression in Escherichia coli as GST fusions. The pSR-lbc and pSR-v-ras plasmids used for the transformation assays were described before (52).
Expression and purification of GST fusion small GTP-binding proteins (GST-Cdc42, GST-RhoA, GST-N17Cdc42, and GST-N19RhoA) from pGEX vector-transformed E. coli were carried out as described previously (20). Production of GST-N2, -N3, and -N4 and the GST-PH domain of Dbl in E. coli was carried out similarly. Production and purification of the Sf9 insect cell expressed His6-tagged T1, DH-PH module, or N1 polypeptide were performed as described (51). The concentration and integrity of purified proteins were estimated by Coomassie blue staining after sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE).
Cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% calf serum (NIH 3T3) or 10% fetal bovine serum (Cos-7). Transfections were carried out using the Lipofectamine reagent (Gibco Life Sciences, Inc.). To generate stable cell lines, NIH 3T3 cells were transfected with pZipneoGST constructs or a combination of the pKH3 construct with pCMV2B vector (Stratagene), which contains a neomycin resistance selection marker and were selected in DMEM supplemented with 5% calf serum and G418 (350 μg/ml). The drug-resistant colonies were cloned and subcultured in the same medium after 18 days.
To assay transforming activity, NIH 3T3 cells were transfected with the pZipneo-proto-Dbl constructs onco-Dbl and lbc or v-ras cDNA by the calcium phosphate method as described (53). For inhibition assays, different doses of the cDNAs encoding the N-terminal polypeptide N1 or N2 in the pCEFL vector or the pCEFL vector alone were cotransfected with onco-Dbl, lbc, or v-ras cDNAs into the cells. The transfected NIH 3T3 cells were fed every 2 days with fresh DMEM supplemented with 10% calf serum. At 12 to 14 days posttransfection, the cell culture dishes were either visualized directly under the microscope for focus formation or stained with a 2% solution of Giemsa for focus scoring (53).
The time courses for [3H]GDP/GTP exchange of Rho family GTPases in the presence and absence of purified His6-tagged or HA3-tagged proto-Dbl mutants were determined as previously described using the nitrocellulose filtration method (51). The GEF reaction buffer contains [3H]GDP-loaded Cdc42 with 20 mM Tris-HCl (pH 7.6), 100 mM NaCl, 10 mM MgCl2, 0.5 mM GTP, and 1 mM dithiothreitol (DTT) supplemented with various proto-Dbl mutants.
Cos-7 cells were transfected with various proto-Dbl constructs or the N-terminal polypeptide N1 by the Lipofectamine method (54). At 48 h posttransfection, complex formation between the HA3-tagged proto-Dbl mutants and GST-fused dominant negative Cdc42 (N17Cdc42) were carried out by incubation of the mutant proto-Dbl-expressing cell lysates with the immobilized GST fusion proteins (54). Complex formation between GST-N1, GST-N2, GST-N3, or GST-N4 and the DH-PH, DH, or PH protein or between the GST-PH domain and the HA3-N1 polypeptide were carried out similarly. The coprecipitation complexes were probed with anti-HA monoclonal antibody and visualized with chemiluminiscence reagents (Amersham Pharmacia). To detect coimmunoprecipitation between the N2 polypeptide and various Dbl mutants, the Myc-N2-encoding cDNAs in vector pCMV6 were cotransfected with the DH-PH, DH, or PH construct in the pKH3 vector into Cos-7 cells, and the cell lysates were subjected to anti-HA immunoprecipitation with an anti-HA monoclonal antibody immobilized on agarose beads (Roche Biochemicals). The coprecipitates were washed three times before being probed with anti-HA or anti-Myc in Western blots.
The glutathione-agarose-immobilized GST-PAK1, which contains the p21-binding domain (PBD) of human PAK1 (residues 51 to 135), and GST-PKN, which contains the site required for RhoA-GTP recognition of protein kinase N (residues 1 to 128), were expressed and purified in E. coli by using the pGEX-KG vector as previously described (29). The active, GTP-bound form of Cdc42 or RhoA in fresh Cos-7 cell lysates coexpressing the small GTPase and various proto-Dbl constructs was captured by incubation with the GST-fused effector domains for 40 min at 4°C (54).
Log-phase growing fibroblasts were seeded at a density of 3 × 104 cells per 12-mm round coverslip (Fisher Scientific) overnight before fixation in phosphate-buffered saline containing 4% paraformaldehyde for 10 min at room temperature. The cells were permeabilized in Tris-buffered saline containing 0.2% Triton X-100 for 5 min and double stained for HA-tagged protein and F-actin using anti-HA monoclonal antibody and rhodamine-phalloidin (Molecular Probes). Coverslips were mounted onto slides in 50% glycerol–Tris-buffered saline. Stained cells were analyzed with a conventional fluorescence microscope and a Zeiss confocal microscope (54).
To delineate a possible structural motif embedded within the N-terminal sequences that confers an inhibitory function, we generated a series of deletion mutants of proto-Dbl in which the N-terminal 101, 171, 285, 348, 407, 442, or 497 residues and/or the C-terminal 100 amino acids (T1 to T7 and DH-PH) were removed while leaving the DH-PH module intact (Fig. (Fig.1A).1A). To evaluate the transforming potential of these proto-Dbl constructs, the respective cDNAs were cloned into the mammalian pZipneo vector and transfected into NIH 3T3 cells. As positive and negative controls, pZipneo-proto-Dbl, pZipneo-onco-Dbl, and the pZipneo vector alone were tested in parallel.
As shown in Fig. Fig.1B,1B, under assay conditions in which proto-Dbl displayed 1 to 2% of the transforming activity of onco-Dbl, the vector alone consistently yielded null foci. Deletion of the C-terminal 100 residues from proto-Dbl (T1) yielded a similar number of foci as proto-Dbl itself, indicating that the C-terminal sequences after the PH domain do not contribute directly to proto-Dbl regulation. Sequential removal of the N-terminal sequences, however, apparently unleashed the transforming activity in a two-step manner: the T2, T3, and T4 constructs, which lack the N-terminal 101, 171, and 285 residues, respectively, displayed a minor increase in transforming activity, with 7 to 9% of the transforming activity of onco-Dbl, whereas further truncation to residue 348, 407, or 442 (T5, T6, and T7, respectively) resulted in significant activation of transforming activity indistinguishable from that of onco-Dbl. Since the deletion mutants were expressed equally well in NIH 3T3 cells and Cos-7 cells, giving rise to polypeptides of the expected molecular weights (data not shown; see Western blots described below), the differences between the mutants in transformation are likely to reflect the true biological activities in cells rather than their differences in stability. Consistent with a previous observation (20), the DH-PH module (residues 498 to 825) behaved like onco-Dbl (Fig. (Fig.1B),1B), implying that the DH and PH domains together constitute the structural module sufficient for maximum transforming activity. These results suggest that the extreme N terminus (residues 1 to 101) of proto-Dbl contains a minor negative regulatory element and that sequences between residues 286 and 348 contain an additional element(s) that is involved in imposing a major constraining effect on the oncogenic activity of the subsequent DH-PH module.
The transforming activity of onco-Dbl was found to correlate closely with its Rho GEF activity (54). We pondered whether proto-Dbl, under the constraint of the N-terminal sequences, is defective in functioning as a GEF for Cdc42 and RhoA, resulting in the apparent suppression of transforming activity. To this end, the T1 mutant, which bears the C-terminal 100-amino-acid truncation and functions like proto-Dbl in transformation assays, and the DH-PH module, which mimics onco-Dbl in both GEF catalysis and transformation ability (20), were expressed in Sf9 insect cells as His6-tagged fusion proteins and purified by Ni2+-agarose affinity chromatography. When equal molar amounts of His6-T1 and His6-DH-PH were assayed for the ability to stimulate [3H]GDP/GTP exchange of Cdc42, we observed that while DH-PH was very efficient in accelerating the GEF reaction, such that over 90% of bound [3H]GDP was dissociated from Cdc42 within 5 min under its stimulation, T1 was only marginally active in stimulating the GEF reaction, such that only ~10% of Cdc42-bound [3H]GDP was replaced by GTP within the same time period compared with that in the absence of T1 (Fig. (Fig.2A).2A). Thus, the N-terminal sequences of proto-Dbl negatively regulate the GEF activity of the DH-PH module.
Next, we examined the catalytic GEF activity of a panel of proto-Dbl mutants on Cdc42 in vitro. The T1, T4, T5, T6, and DH-PH cDNAs in the pKH3 vector, which provides an HA3 tag at the N terminus, were transiently expressed in Cos-7 cells, and the proteins were purified by immunoprecipitation from the cell lysates by using anti-HA agarose beads. Upon elution by HA peptides, the purified deletion mutants were analyzed by anti-HA Western blot, and the amount of each protein was visualized by chemiluminiscence imaging of the blot (Fig. (Fig.2B).2B). While the T5 and T6 mutants, which lacked the N-terminal 348 and 407 residues, respectively, showed activities in stimulating [3H]GDP dissociation from Cdc42 similar to that of the DH-PH module, T1 and T4, which contain an intact N terminus or residues 286 to 825, respectively, were comparable in displaying a significantly weaker GEF activity at equal molar quantities (Fig. (Fig.2C).2C). These results indicate that the N-terminal sequences directly impose an inhibitory effect on the GEF activity of the DH-PH module. To test if the N terminus interacts with the catalytic DH domain, resulting in inhibition, the N1 polypeptide encoding residues 1 to 482 was generated in Sf9 insect cells as a GST fusion and included in the GEF activity assays with the DH-PH module. As shown in Fig. Fig.2D,2D, no detectable effect was observed when a fourfold molar excess of GST-N1 was present together with the DH-PH module in stimulating [3H]GDP dissociation from Cdc42 compared with DH-PH alone. We conclude that the N terminus of proto-Dbl interferes with the GEF function through a mechanism other than direct blockage of substrate binding to the DH domain as is the case with Vav (3).
To determine the Rho GTP ase exchange potential of the mutants in cells, the HA-tagged proto-Dbl mutants were transiently cotransfected with HA-tagged, wild-type Cdc42 or RhoA in Cos-7 cells. The expression level of the mutants and Cdc42 or RhoA could be directly compared (Fig. (Fig.3).3). The relative amounts of activated GTPases in the cell lysates were measured by GST-PAK1 or GST-PKN pull-down, which specifically recognizes and stabilizes the GTP-bound form of Cdc42 or RhoA (29). As shown in Fig. Fig.3,3, both the T1 and T4 mutants demonstrated significantly lower Cdc42 activating potential, while T5 and T6 were similar to the DH-PH module in their ability to generate Cdc42-GTP. Similar observations were also made when RhoA was examined as an in vivo substrate (data not shown). These results indicate that the cellular Rho GTPase-activating potential of the proto-Dbl mutants correlates with their in vitro GEF activity. This pattern of GEF activity and the Rho protein-activating potential of the mutants are reminiscent of the above-described transforming abilities of the mutants (Fig. (Fig.1B).1B). We deduce from these results that the sequences between residues 286 and 348 contain the critical structural element(s) that appears to hinder the GEF function of the DH-PH module and that the lack of transforming activity of proto-Dbl reflects the suppressive effect on the catalytic GEF activity by the N-terminal negative regulatory constraint(s).
To further investigate the functional properties of the various deletion mutants of proto-Dbl, we next compared their abilities to directly bind to the Rho GTPase substrate Cdc42. Cos-7 cell lysates expressing similar amounts of HA-tagged T1, T4, T5, T6, and the DH-PH module (Fig. (Fig.4)4) were incubated with the glutathione-agarose-immobilized, GST-fused, dominant negative form of Cdc42, N17Cdc42, that is known to bind to the DH domain of onco-Dbl with high affinity (20). After extensive washing the coprecipitates of GST-N17Cdc42 were subjected to anti-HA Western blotting. Among the five HA-tagged polypeptides, T5, T6, and the DH-PH module displayed a similarly strong binding pattern to N17Cdc42. T4 bound significantly more weakly, and T1 binding was barely detectable (Fig. (Fig.4).4). These results are consistent with the relative effectiveness of the mutants in activating the guanine nucleotide exchange of Cdc42 in vitro and in vivo (Fig. (Fig.22 and and3)3) and coincide with the mutants' transformation abilities (Fig. (Fig.1B).1B). They suggest that the N-terminal residues, particularly residues 285 to 348, are involved in negative allosteric control of proto-Dbl activity by limiting the access of Rho GTPase to the catalytic site on the DH domain.
Onco-Dbl and the PH domain of Dbl were colocalized with Triton X-100-insoluble particulates in previous cell fractionation studies (53). Whether the N-terminal constraining sequences of proto-Dbl would interfere with the PH domain intracellular targeting function, thereby causing an alteration in the intracellular distribution pattern of proto-Dbl from that of onco-Dbl, has not been assessed. To examine the effect of the N-terminal sequences on proto-Dbl cellular localization and to determine the contribution of individual structural motifs to the proto-Dbl localization pattern, we have generated stable transfectants of NIH 3T3 cells expressing the HA-tagged proto-Dbl (T1), DH-PH module, DH domain, PH domain, or the N1 polypeptide (residues 1 to 482), as well as a cell clone coexpressing Flag-tagged N1 together with HA-DH-PH (DH-PH+N1). Western blot analysis of the anti-HA immunoprecipitates from the respective cell lysates confirmed that HA-tagged polypeptides of the expected molecular sizes were expressed in the cell lines (Fig. (Fig.5A).5A). In addition, an anti-Flag Western blot further confirmed that Flag-N1 was coexpressed with HA-DH-PH in the DH-PH+N1 cells (data not shown). After fixation, the cells were double stained with anti-HA monoclonal antibody plus fluorescein-labeled anti-mouse immunoglobulin antibody and rhodamine-labeled phalloidin to visualize HA-tagged polypeptide distribution and cellular actin structures. Under a confocal microscope, we found that proto-Dbl displayed a mostly perinuclear distribution pattern similar to that of the N1 polypeptide (Fig. (Fig.5B).5B). The DH-PH module, on the other hand, was found to colocalize with actin stress fibers, like the PH domain alone, whereas the DH domain was mostly diffused throughout cells, with some degree of concentration around the nucleus (Fig. (Fig.5B).5B). These results are consistent with the previous cell fractionation data showing a significant portion of the DH-PH module and the PH domain in the Triton X-100-insoluble fraction and the DH domain mostly in the cytosol (53) and suggest that the N-terminal sequences in proto-Dbl affect the cellular distribution pattern of proto-Dbl. In the N1 polypeptide-coexpressing cells, the actin-stress fiber colocalization pattern of DH-PH was disrupted and changed to the mostly perinuclear localization, similar to that of the N1 polypeptide alone (Fig. (Fig.5B).5B). Thus, the N-terminal sequences of proto-Dbl contain the structural element(s) that dictates the intracellular distribution of proto-Dbl. This is likely due to the interference of the PH domain targeting function that determines the localization pattern of onco-Dbl by the N-terminal sequences.
The negative regulatory function of the N-terminal sequences raised the possibility that the isolated N terminus of proto-Dbl might interfere with the biological activity of oncogenic Dbl. To test this hypothesis, the cDNAs encoding the N1 (residues 1 to 482) and N2 (residues 286 to 482) polypeptides were cloned into the mammalian expression vector pCEFL and cotransfected with onco-Dbl into NIH 3T3 cells. Compared with the empty vector, N1 reduced onco-Dbl transforming activity by ~90% at a dose of 2 μg/100-mm dish, while N2 consistently caused ~25% inhibition under similar conditions (Fig. (Fig.6).6). At a lower dose (0.2 μg/100-mm dish), however, only N1 showed significant inhibition of onco-Dbl activity. Neither N1 nor N2 had a detectable effect on proto-Dbl transforming activity when the foci were induced by proto-Dbl overexpression (2 μg of cDNA/100-mm dish; data not shown). To confirm that the N-terminal sequence-caused inhibition was specific for onco-Dbl, we examined the ability of N1 and N2 to affect the transforming functions of a dbl-related oncogene, lbc, and oncogenic v-ras. Distinct from their effects on onco-Dbl, neither peptide showed any sign of inhibiting oncogene-induced focus formation at 2 μg of cDNA/dish (Fig. (Fig.6).6). These results indicate that the N terminus of proto-Dbl can specifically act on onco-Dbl or the onco-Dbl pathway and negatively influence its biological activity.
The above-characterized negative regulatory function of the N terminus of proto-Dbl could be rationalized by an intramolecular interaction involving the N-terminal regulatory sequences and the C-terminal functional module, thereby affecting the GEF activity of the DH domain and the intracellular targeting function of the PH domain. To test this hypothesis, we first employed a glutathione-agarose pull-down assay using the insect cell-expressed GST-N1 peptide as a probe to detect possible interaction with the C-terminal functional motifs, i.e., the DH-PH module, DH domain, or PH domain. The DH-PH module, DH domain, and PH domain were transiently expressed in Cos-7 cells as HA-tagged proteins, and the cell lysates were incubated with immobilized GST or GST-N1. As shown in Fig. Fig.7A,7A, GST-N1 was able to stably associate with the DH-PH module and the PH domain but not with the DH domain, whereas the GST control did not bind to any of the three proteins. When the N1 polypeptide was expressed in Cos-7 cells and subjected to the pull-down assay with the immobilized GST-PH domain, we found that GST-PH was able to tightly bind to HA-N1 in the glutathione-agarose coprecipitates under conditions in which GST alone did not interact with N1 (Fig. (Fig.7B).7B). Therefore, the N-terminal 482 residues of proto-Dbl can form a physical complex with the C-terminal DH-PH functional module via the PH domain.
To further delineate the region of amino acids in the N terminus that may contribute to direct interaction with the PH domain, the sequentially deleted N2, N3, and N4 polypeptides, encoding residues 286 to 482, 349 to 482, and 408 to 482, respectively, were employed as GST-tagged probes to detect possible binding to the PH domain. Figure Figure7C7C shows that while none of the three N-terminal peptides bound to the DH domain at a detectable level, N2, but not N3 or N4, was capable of binding directly to the PH domain. These results imply that the region between amino acids 286 and 348 contains an important element(s) that is involved in interaction with the PH domain.
To test whether stable association between the N terminus and the PH domain could occur in cells, a Myc-tagged N2 peptide was coexpressed with the HA-tagged DH-PH module, DH domain, or PH domain in Cos-7 cells, and the coimmunoprecipitation pattern of Myc-N2 with the HA-tagged proteins was determined. In contrast to the lack of detectable association by HA-DH, both HA-DH-PH and HA-PH formed a stable complex with Myc-N2, as revealed by the anti-Myc Western blot of the anti-HA immunoprecipitates (Fig. (Fig.7D).7D). Thus, the N-terminal sequences of proto-Dbl can bind tightly to the C-terminal PH domain in cells and are likely to do so intramolecularly. Such an interaction may have a direct impact on both the GEF activity and intracellular localization of proto-Dbl, which are essential for its transforming activity, causing the observed autoinhibitory behaviors.
Although most Dbl family members contain diverse multifunctional motifs, they all have the structural array of a central DH domain in tandem at the carboxyl terminus with a PH domain. Previous studies of onco-Dbl have established that the DH domain is primarily responsible for Rho GTPase binding and GEF activity, while the PH domain is involved in intracellular targeting and is necessary for the transforming activity of onco-Dbl (8, 50). Truncation of the N-terminal 497 residues of proto-Dbl results in oncogenic activation (42), suggesting that the N-terminal sequences contain negative regulatory elements imposing a constraint on the C-terminal DH-PH module. To date, the mechanism which the N terminus employs in the negative regulation of proto-Dbl activity and the regions of the molecule contributing to the regulation remain unclear. In this study, we provide direct evidence that proto-Dbl adopts an autoinhibitory mechanism for controlling its biochemical and biological activities. We identified the region between amino acids 275 and 349 as a critical inhibitory motif that impairs GEF catalytic activity and the intracellular targeting activity of the DH-PH module. The autoinhibitory effects are likely to arise through direct intramolecular interaction of this region with the PH domain, limiting the access of Rho GTPases to the DH domain and masking the intracellular targeting function of the PH domain. Such an autoinhibitory mode may be optimal for proto-Dbl regulation, since incoming upstream signals could exert their effects by modulation of either the N-terminal motif or the C-terminal PH domain.
Our previous structural mapping studies indicate that the DH-PH module of Dbl represents the minimum constitutively active structural unit that confers Rho GTPase exchange activity and cell-transforming potential (20). We show here that the presence or absence of the C-terminal 100 amino acids just outside the PH domain did not cause any change in transforming activity for either proto-Dbl or onco-Dbl, indicating that these sequences are not involved in regulation of the DH-PH module. A series of deletion mutations into the N terminus resulted in an apparent two-step activation of the proto-Dbl transforming activity: removal of the N-terminal 100 or 274 residues caused minor but significant enhancement, while further truncation to residue 348 led to full-blown activation of the focus-forming activity that is similar to the capacity of the DH-PH module (Fig. (Fig.1).1). These results prompted us to speculate that the N-terminal sequences impose a two-layer regulatory mechanism on the DH-PH module. The N-terminal 100 residues and sequences between residues 275 and 349 either act independently in negatively controlling the DH-PH activity or act coordinately so that the N-terminal 100 residues may further reinforce the negative constraint imposed by the following sequences. This is analogous to the situation of the vav proto-oncogene product, in which removal of the N-terminal 66 or 127 residues led to only partial activation of the transforming activity, while full activation was achieved by truncation of the N-terminal 186 residues (1).
The cellular transforming activity of Dbl is intimately dependent upon its catalytic GEF activity on Rho GTPases (54). The inhibitory effects of deletions of proto-Dbl on transformation indeed reflect their relative GEF activities on Cdc42 and RhoA when the purified mutants were tested in vitro (Fig. (Fig.2).2). The N terminus appears to potently inhibit the GEF ability of the DH-PH module, implying that an intramolecular interaction is at work in the regulatory mechanism. Residues 275 to 349, in particular, seem to contain the critical inhibitory element(s) for GEF activity, reminiscent of the major inhibitory region involved in the regulation of transforming activity (Fig. (Fig.1B).1B). We show that the significantly reduced GEF activity of proto-Dbl and the deletion mutants that were generated at the N terminus of residue 274 was due to their reduced ability to interact with Rho GTPase, further establishing that the N-terminal sequences limit the access of the DH catalytic site for Rho proteins. The fact that the in vitro GEF activities of the proto-Dbl mutants closely correlate with their Cdc42- and RhoA-activating potential in vivo and with their cellular transforming activity strongly suggests that proto-Dbl maintains a low basal transforming capability, at least in part by downregulation of its ability to activate Rho GTPases through the N-terminal sequences.
By comparing the distinct intracellular distribution patterns of proto-Dbl and the DH-PH module, we come to the conclusion that the N-terminal sequences also dictate the intracellular location of proto-Dbl. Consistent with previous subcellullar fractionation results (53), our confocal microscopy data clearly demonstrate that the PH domain of Dbl colocalizes with the actin structure of cells, and this property of the PH domain is responsible for bringing the DH-PH module to a similar location. However, the presence of the N-terminal sequences in proto-Dbl appeared to overrule the PH function, leading the molecule to a perinuclear location. In cells in which the polypeptide encoding the N-terminal sequences was overexpressed together with the DH-PH module, the DH-PH protein was found to translocate from the actin-associated locations to the perinucleus, further confirming that the N-terminal sequences are involved in regulation of the cellular localization pattern of proto-Dbl. These observations raise the possibility that the N-terminal sequences may block ligand binding to the PH domain, resulting in loss of the targeting function of PH.
Although the mechanism of intramolecular interaction was considered in the regulation of many Dbl family members, including Vav, Tiam1, Ect2, Ost, Net1, and proto-Dbl (1, 4, 21, 33, 42), and in the case of the Ras-specific GEF Sos1 (11, 40), there has been little biochemical evidence available to directly support such a mode of regulation for the GEFs until recently. A nuclear magnetic resonance spectroscopic study on an extended DH domain of proto-Vav most recently has revealed that the immediate N-terminal sequences, including the critical Tyr174 residue, could interact directly with the catalytic core region of the DH domain, achieving an autoinhibition conformation (3). Similar to but distinct from the proto-Vav protein, proto-Dbl was found in the current study to maintain a basal inactive state by specific binding of the N-terminal sequences to the PH domain, providing a biochemical rationale for another autoinhibitory mechanism in the negative regulation of a Dbl family member.
The structural arrangement of the N-terminal sequences of proto-Dbl is not known but appears to be different from those of proto-Vav protein and other Dbl family GEFs. Limited sequences homologies between a >300-residue span of the N terminus and the intermediate filament protein vimentin suggest that the N-terminal region may contain an extended α-helical coiled-coil structure (41). In agreement with the functional analysis, i.e., transformation capacity, GEF activity, and Rho GTPase binding activity, our sequential N-terminal deletion mutants point to the region between amino acids 275 and 349 as critical in interaction with the PH domain. The observed direct binding interaction between the N terminus and the PH domain allows us to present a model to rationalize the previous (20, 42, 53) and above-described functional data: by interacting with the PH domain, the N terminus of proto-Dbl maintains the molecule at an autoinhibited basal state. This intramolecular interaction would mask the PH-ligand binding site, resulting in an inactive PH domain and allowing the N terminus to dictate the cellular location of the molecule. Although the DH domain may not be directly inhibited by the N terminus, as proto-Vav is (3), since no binding interaction was detected between them, the allosteric hindrance brought by the N-terminal interaction with PH domain could effectively limit the access of the Rho GTPase substrates to the DH catalytic sites, indirectly affecting the Rho protein-activating potential. This model could also explain the observations that proto-Dbl (T1) remains weakly active as a GEF and can bind to Cdc42 with reduced affinity (Fig. (Fig.22 and and44).
Such a mode of autoinhibition is interesting from another angle. It provides a physiologically relevant peptide ligand, i.e., the coiled-coiled N-terminal motif, for a PH domain. Although the significance of various phosphoinositol phosphates as PH domain binding partners is well established (10, 39), physiologically important protein ligands for the PH domain remain rare. Consistent with our previous cell fractionation studies, which implicated the PH domain of Dbl in association with the Triton X-100-insoluble component of the cytoskeleton (53), we demonstrate in this study by immunofluorescence that Dbl PH extensively colocalizes with actin stress fibers in cells, suggesting that the PH domain is involved in interaction with a protein factor. Thus, it is possible that the N terminus binding to the PH domain may function as a switch in proto-Dbl regulation so that the basal, autoinhibited state can be relieved of the N-terminal constraint when a cytoskeletal protein factor binds to the PH domain by competing with the N-terminal sequences. It will be of great interest to obtain the molecular details of the N terminus-PH domain complex by structural biology means.
Understanding the autoinhibitory mechanism of proto-Dbl is an important step toward elucidating the mechanism of its activation. Recent progress in studies of the mechanism of Rho GTPase activation, in particular, the Dbl-like GEF activation, has provided quite a few possibilities on how the autoinhibited state of proto-Dbl could become activated. First, a phosphorylation event that modifies either the N-terminal constraining sequences or the PH domain may result in the desired relief of the structural constraint of the N terminus-PH domain interaction. However, serine/threonine rather than tyrosine phosphorylation is more likely to play a role in proto-Dbl activation, since proto-Dbl was found to be phosphosphorylated mainly on serine and threonine residues (15). Second, interaction with heterotrimeric G-protein α or βγ subunits may lead to effective translocation and activation of proto-Dbl. The analogy here includes p115RhoGEF, which utilizes its N terminus to couple directly to Gα12/Gα13, resulting in enhanced GEF catalytic activity and membrane translocation (5, 19). Moreover, a recent study also found that Gβγ can bind directly to the N-terminal 100 amino acids of proto-Dbl (35). Third, various phosphoinositol lipids may be involved in proto-Dbl activation, given our finding that the PH domain of Dbl can bind to PIP2 (4, 5) and PIP3 (3, 4, 5) selectively. The conformational change of the PH domain induced by such an interaction could therefore lead to reduced binding to the N terminus of proto-Dbl, resulting in an open, active conformation. Finally, similar to the Cdc42-specific exchange factor in budding yeast Cdc24, proto-Dbl may need to be recruited to the targeting site via interaction with a Far1-like scaffolding protein, which recognizes a conserved motif found in the N termini of both Cdc24 and proto-Dbl (7). If Cdc24 is a good analogue for proto-Dbl, it is likely that a combination of the above events will be required for the production of a fully activated proto-Dbl molecule. These possible proto-Dbl-activating events are currently under investigation.
The autoinhibitory mode of regulation utilized by proto-Dbl falls into an emerging theme that appears to govern many aspects of small-G-protein signaling. Sos1, one of the best-understood Ras GEFs, is known to involve both N-terminal and C-terminal sequences to achieve autoinhibitory control of its Ras GEF activity (11). In addition to proto-Dbl, p115RhoGEF, Ost, Tiam1, Ect2, Net1, and Asef of the Dbl family all seem to employ some degree of autoinhibition to maintain themselves in a basal state (16, 19, 21, 22, 33, 41), although the molecular basis for their negative regulation remains unclear. In the context of the present results on proto-Dbl regulation, the recently illustrated tertiary structure of the autoinhibitory mode of Vav regulation, which involves an intramolecular interaction between the N-terminal Tyr174 containing an α-helix and the Rho GTPase binding site of the DH domain (3), suggests that the biochemical role of the N-terminal sequences in the Dbl family GEFs may be diverse, so that direct modulation of either the DH or the PH domain could be involved in controlling the activity of the DH-PH module. The activation scheme for the GEFs, by the same token, would reflect a similar divergence which could involve an array of distinct mechanisms in absorbing different incoming signals to provide signal transduction divergence.
The recent characterization of the autoinhibitory mechanisms of the Rho GTPase effectors PAK1 and WASP also provides interesting parallels to the Dbl family Rho GEFs (25, 27). In both cases, the N-terminal CRIB motif of the molecules interacts specifically with the C-terminal functional module, the kinase domain for PAK1, or the VCA domain for WASP to achieve an autoinhibition state by masking the access of the immediate downstream target, the PAK1 substrates, or the Arp2/3 complex (25, 27). Activation of such an autoinhibition state occurs when activated Cdc42/Rac recognizes the CRIB motif, leading to allosteric relief of the target-interactive sites. Therefore, from Rho GEFs to effectors, the autoinhibition mechanism seems to be a common feature that the small-G-protein cascades utilize to mediate signal flows in a highly regulated manner.
This work was supported by National Institutes of Health grant GM 53943 and U.S. Army grant BC990290 to Y.Z.F.B. is an American Heart Association Southern Consortium postdoctoral fellow.
We acknowledge the technical assistance of Catherine Ottaviano.