Previous results from our and other groups suggest that neuregulin activates the receptor c-ErbB-2 to promote lobulo-alveolar development of the mammary gland(Yang et al., 1995; Jones et al., 1996; Niemann et al., 1998; Jones and Stern, 1999). However, little was known about the intracellular effectors that mediate this morphogenic effect. Here, a modified yeast two-hybrid screen was developed to search for new substrates of ErbB-2 that are involved in alveolar morphogenesis of the mammary gland. The guanine nucleotide exchange factor Vav2 was identified among other proteins as a novel interaction partner of ErbB-2. A full characterization of the ErbB-2 docking sites for these interacting proteins was performed in the yeast system, thus adding complexity to the previous knowledge on intracellular effectors of ErbB-2. Next, the potential function of Vav2 in ErbB2-induced morphogenesis was tested in EpH4 mammary gland epithelial cells that were cultured on a reconstituted basement membrane (Matrigel). These studies show two main points: first, Vav2 indeed promotes alveoli-like growth and reorganization of EpH4 cells, either when constitutively active or upon activation by ErbB-2; second, co-expression of Vav2 and ErbB-2 in mammary alveolar epithelium is observed in pregnant mice. These findings suggest a functional association between Vav2 and ErbB-2 during alveolar morphogenesis in vivo. Together, this work supports a model for neuregulin-induced mammary alveolar morphogenesis in vivo, in which Vav2 is a physiological target downstream of ErbB-2 that plays a crucial role in transduction of signals for lobulo-alveolar morphogenesis.
The commonly-used yeast two-hybrid baits to search for interaction partners of receptor tyrosine kinases are hybrid proteins, which consist of a LexA DNA-binding and dimerization sequence fused to the intracellular region of the receptors (O'Neill et al., 1994; Weidner et al., 1996; Grimm et al., 2001). Bait molecules dimerize via LexA when expressed in yeast cells, [page 55↓]thus enabling the activation of receptor tyrosine kinases without requirement of a ligand. This approach could not be used in case of the receptor ErbB-2, since the resulting hybrid proteins were rapidly degraded in the yeast. Therefore, modified yeast baits were generated: C-terminal sequences of ErbB-2 containing the autophosphorylation sites were fused to an oncogenic TprMet kinase in order to obtain constitutively active TprMet-ErbB2 baits. TprMet encodes the intracellular region of c-Met in frame with a leucin-zipper dimerization domain derived from a different locus (Park et al., 1986). The heterologous TprMet kinase efficiently phosphorylates tyrosine residues of ErbB-2, thus creating docking sites for signaling proteins with SH2 and PTB domains. By this means, known and novel interacting proteins for ErbB-2 were identified in a cDNA library screen in yeast.
This modified yeast two-hybrid system has proven to be useful for standard search of proteins that interact with peptides and proteins containing phosphorylated tyrosine residues; even unrelated protein sequences become tyrosine-phosphorylated when fused to the TprMet kinase and can be used as yeast baits. For example, coupling Gab1 to TprMet enabled the analysis of the interaction between phosphorylated Gab1 and the SH2 domains of PI-3-K, PLCγ, Shp2 and CRKL (Schaeper et al., 2000). Thus, the new method presented here offers an easy alternative for yeast screens as opposed to a tribrid approach, which includes an additional kinase (Licitra and Liu, 1996). The main advantage of the here described modified two-hybrid over a tribrid system is that only two selectable markers instead of three markers are necessary; therefore, the efficiency of yeast transformation is improved, and protein expression levels in the yeast are higher than those of the tribrid method.
In independent cDNA library screens with tyrosine phosphorylated ErbB-2 baits, novel ErbB2-interacting proteins like Nck, Grb-10 and Vav2 were identified, along with the known partners Src and PLCγ1 (Muthuswamy and Muller, 1995b; Fazioli et al., 1991). These proteins belong to the group of multiadaptor signaling molecules; the most conspicuous [page 56↓]feature of such proteins is the presence of several specialized regions that allow protein-protein interactions (reviewed in Pawson and Scott, 1997). The ErbB-2 partners identified in the present workt contain SH2 domains; indeed, these SH2 domains were always present in the interacting clones and therefore mediate the association with phosphotyrosine residues of ErbB-2. The direct binding sites of ErbB-2 for these proteins were mapped by mutational analysis of the baits in the yeast system. So far, tyrosines Y2 and Y4 of the ErbB-2 multidocking site were known to directly bind Grb-2 and Shc, respectively (Ricci et al., 1995; Dankort et al., 1997). Here, direct binding of Vav2 to Y1 and Y3, Grb-10 to Y4, Nck to Y3 and Src to complexed Y3-Y4 were characterized. The overlap of Vav2, Nck and Src for binding to Y3 suggests that these proteins may represent alternative effectors of ErbB-2. Alternatively, these proteins may assemble to form a hierarchy of different protein multicomplexes; any of such supramolecular complexes may lead to a unique outcoming signal, regardless of the protein-protein interaction pattern within the complexes. The formation of functionally redundant protein alignments between Src, Nck and Vav2 (possibly together with some other proteins) is indeed conceivable. Preliminary data indicate that Vav2 can directly bind Src or Nck; using the modified yeast two-hybrid method, a TprMet-Vav2 fusion bait was recently tested with several SH2 domains for interaction in the yeast. Indeed, tyrosine-phosphorylated Vav2 interacted with the SH2 domains of Src and Nck but not with that of Grb2 (data not shown). A mechanism whereby Vav2 is ultimately recruited and activated via different protein-protein associations may account for the direct and indirect binding of Vav2 to ErbB-2 (see also below).
Direct association of the SH2 domain of Vav2 with ErbB-2 involves recognition of the motif pYLVP, present at Y1 (Y1028) and Y3 (Y1201). Using an entirely different selection procedure, the sequence pYXEP (where X is L, M or E) has been identified to bind the SH2 domain of Vav (Songyang et al., 1994). The SH2 domains of Vav and Vav2 share 55% identity(Schuebel et al., 1996), and it is therefore possible that they have distinct binding [page 57↓]preferences. From all the ErbB-2 partners that were found in the yeast screens, only Vav2 showed a particular affinity for binding to the various ErbB receptors. Vav2 also interacts with the PDGFβ receptor the yeast system (Table 3). In fact, recent biochemical and mass spectrometry analyses have shown that all three Vav proteins can form complexes with EGF or PDGFβ receptors and become phosphorylated on tyrosine residues following stimulation with EGF or PDGF (Bustelo et al., 1992; Margolis et al., 1992; Lopez-Lago et al., 2000; Moores et al., 2000; Pandey et al., 2000). However, it was not clear from these studies whether Vav proteins couple directly to the receptors or indirectly via binding to adaptor proteins. The results from yeast two-hybrid analyses that are here presented do not rule out the possibility of indirect interaction (see below); however, they provide the first line of evidence that indeed Vav2 can directly bind to receptor tyrosine kinases via its SH2 domain, a mechanism that may be shared by the other Vav proteins.
The above identified pYLVP consensus motif for binding to Vav2 is conserved in the C-terminal region of ErB-4 (Y1022) but not in the EGF receptor, ErbB-3 or the PDGFβ receptor; however, the rather similar sequences pYLIP in the EGF receptor (at Y1012), pYLMP in ErbB-3 (at Y1159) and pYIIP in PDGFβ receptor (at Y1021) may be predicted as Vav2 binding sites. This suggests that positions 1+ and 2+ of the consensus motif for Vav2 are flexible but, in contrast to the pYXEP binding motif of Vav, they may selectively be occupied by amino acids with a non-polar side chain.
The examination of ErbB-2/Vav2 interactions in mammalian cells rendered an unexpected result: in 293T cells, a mutant ErbB-2 receptor that lacks both Y1 and Y3, the direct docking sites for Vav2 identified in the yeast system, still binds Vav2. This indicates that, in addition to direct binding, Vav2 may also interact indirectly with ErbB-2 via a further adaptor protein. Similarly, direct and indirect binding has been reported for recruitment of the adaptor Gab1 to the receptor tyrosine kinase c-Met (Fixman et al., 1995; Holgado-Madruga et al., 1996; Weidner et al., 1996; Bardelli et al., 1997; Nguyen et al., 1997; Schaeper et al., 2000). ErbB-2 [page 58↓]binds to Grb2, Shc, the Csk-homologous kinase, PLCγ1, Nck, Grb10 and Src (Fazioli et al., 1991; Ricci et al., 1995; Muthuswamy and Muller, 1995; Dankort et al., 1997; Zrihan-Licht et al., 1998; this work), which may indirectly recruit Vav2 to the receptor. Sequence analysis of Vav2 reveals the presence of putative consensus motifs for binding the SH2 domains of Src, Lck (a Src-related tyrosine kinase) and Shc (Songyang et al., 1994). Ongoing studies suggest that Vav2 may directly interact with the SH2 domains of Src and Nck.
Recent work has shown that normal development of the mammary gland during pregnancy to prepare lactogenesis can be mimicked in vitro. The reorganization of mammary epithelial cells to form lobulo-alveolar structures is markedly influenced by the matrix on which these cells are cultured. Barcellos-Hoff et al. (1989) demonstrated that primary mammary epithelial cells form functional, albeit rudimentary, alveoli-like structures when cultured on a reconstituted three-dimensional matrix, in the presence of lactogenic hormones and absence of serum. Within the first days of culture, cells remodel the exogenous basement membrane and form matrix-ensheated aggregates, which subsequently cavitate; by day 6 of culture, cells are reorganized into hollow spheres composed of morphologically polarized cells facing a lumen. These cells are functionally differentiated and secrete milk proteins vectorially into the luminal compartment. A reconstituted basement membrane alone, however, does not account for the complex epithelial-mesenchymal interactions that are linked to mammary development. The spontaneously immortalized, non tumorigenic mouse mammary cell line IM-2 consists of both epithelial and fibroblastic cell populations (Reichmann et al., 1989); it has been shown that the fibroblastic cells render the epithelial cells competent to undergo cytoskeletal rearrangements and to functionally differentiate on Matrigel. The physiological signals from the mammary mesenchyme can alternatively be supplied by addition of growth factors to mammary epithelial cells in organotypic culture. Previous studies pointed to neuregulin as a mesenchymal growth factor that stimulates alveolar morphogenesis of [page 59↓]mammary glands in organ culture (Yang et al., 1995). In view of these results, Niemann et al. (1998) used EpH4 mammary cells (an epithelial subclone derived from the abovementioned IM-2 cell line) to test the effect of neuregulin in a Matrigel system; EpH4 cells indeed form large alveoli-like structures when cultured on Matrigel in the presence of neuregulin. Additionally, biochemical studies revealed tyrosine phosphorylation of endogenous ErbB-2 in EpH4 cell lysates following neuregulin treatment (Fig. 8). Formation of similar alveolar structures is observed when EpH4 transfectants that stably overexpress a TrkErbB2 chimeric receptor are cultured on Matrigel in the presence of nerve growth factor (Niemann et al., 1998). These results indicate that activation of the overexpressed ErbB-2 receptor is sufficient to elicit alveolar morphogenesis in organotypic cultures of mammary epithelial cells. Together with the mesenchymal localization of neuregulin transcripts and the epithelial expression of ErbB-2 in mammary tissue at mid-pregnancy, these findings strongly suggest that stromal neuregulin activates epithelial ErbB-2 to induce alveolar morphogenesis of the mammary gland in vivo. Moreover, this model of neuregulin signaling supports the increasing evidence of the role of epithelial–mesenchymal interactions in postnatal growth and differentiation of the mammary gland (reviewed in Cunha and Hom, 1996; Robinson et al., 1999; Silberstein, 2001).
To understand the molecular mechanisms of neuregulin/ErbB signaling in mammary alveolar morphogenesis, this work aimed at the identification of intracellular effectors of ErbB-2 that mediate these morphogenic effects. In yeast two-hybrid screens with ErbB-2 baits, Vav2 was identified as a novel partner of ErbB-2. Vav2 belongs to a family of guanine nucleotide exchange factors for small GTPases of the Rho superfamily, like Rho, Rac and Cdc42 (reviewed in Bustelo, 2000). Vav proteins contain a common array of domains, which include an N-terminal regulatory acidic region, a catalytic Dbl-homology domain and several C-terminal SH2 and SH3 domains (see Fig. 2). Unlike Vav, Vav2 expression is not restricted to the hematopoietic system, but is also found in several epithelia, for example in the mammary gland epithelium (Fig. 9). Though recent reports involve Vav2 in immune responses of certain [page 60↓]hematopoietic lineages (Billadeau et al., 2000; Doody et al., 2000), a function of Vav2 in epithelial tissues is also possible. Upon phosphorylation of a single regulatory tyrosine residue in the N-terminus, Vav proteins gain catalytic activity towards Rho GTPases, thus leading to changes in the actin cytoskeleton and gene transcription (reviewed in Bustelo, 2000). Vav2 associates not only with ErbB-2 in the yeast system, but also with all other ErbB receptors (Fig. 4), which are known to be involved at different stages of mammary development (Xie et al., 1997; Jones and Stern, 1999; Jones et al., 1999). The capability of Vav2 to induce cytoskeletal reorganization, together with its high affinity for ErbB receptors, favored the choice of Vav2 as candidate to mediate morphogenic signals of ErbB-2.
The morphogenic potential of Vav2 was here tested in a Matrigel assay. It was known that truncation of the N-terminal region of Vav proteins eliminates an autoinhibitory loop and exposes the catalytic site to small GTPases, thus enhancing GDP/GTP exchange (Aghazadeh et al., 2000). Therefore, EpH4 mammary epithelial cells were stably transfected with a cDNA encoding an N-truncated, constitutively active Vav2 protein (Schuebel et al., 1996), and transfectants were cultured on Matrigel under serum-free conditions and in the absence of growth factors. In this organotypic system, catalytically active Vav2 was sufficient to induce formation of large alveolar structures that resembled those that are observed following activation of ErbB-2 by neuregulin (see Niemann et al., 1998). Alveoli consisted of a monolayer of polarized epithelial cells, which enclose a luminal compartment by means of apical tight junctions. These results represent the first evidence of a biological response elicited by overexpression of active Vav2 in epithelial cells. Constitutively active but not wild-type Vav2 elicited these morphogenic events in EpH4 cells, indicating that GDP/GTP exchanger activity, and therefore cytoskeletal reorganization, may be essential in these processes. In line with these findings, it has been reported that truncated forms of all Vav family members have a deregulated activity as guanine nucleotide exchange factors in vitro, are highly transforming in focus formation assays, and elicit changes in cell morphology upon transient transfection into NIH 3T3 fibroblasts (Bustelo, 2000 and references therein).
Activation of an overexpressed TrkErbB2-Vav2 fusion protein by nerve growth factor also induced alveolar morphogenesis of EpH4 transfectants. This finding suggests that the activated ErbB-2 kinase of the fusion protein efficiently phosphorylated wild-type Vav2. Phosphorylation may occur on the putative regulatory tyrosine Y172 of Vav2; thus, the autoinhibitory loop of wild-type Vav2 is disrupted and therefore, phosphorylated full-size Vav2 can now behave as its oncogenic truncated counterpart to elicit morphogenesis.
Biochemical studies revealed association between Vav2 and ErbB-2 in lysates from EpH4 cells following neuregulin treatment (Fig. 8B, left panel). Activation of ErbB-2 was observed after stimulation with neuregulin as an increase in phosphotyrosine content, despite high basal phosphorylation levels (Fig. 8B, right panel). Importantly, endogenous complexes between Vav2 and ErbB2 were found in lysates of mammary glands from pregnant mice (Fig. 8C). In addition, in situ hybridization studies showed spatial co-localization of Vav2 and ErbB-2 in mammary alveolar epithelium (Fig. 9), which allows the physical interaction between the receptor and its putative effector in the mammary gland during pregnancy.
Taken together, Vav2 can induce alveolar morphogenesis of EpH4 cells, both in its constitutively active state or when activated by ErbB-2. This places Vav2 as a downstream effector of ErbB-2 for this biological response. In both cases, alveolar structures were similar to those that are formed upon neuregulin treatment or activation of ectopic ErbB-2 (Niemann et al., 1998); along with the functional link between neuregulin and ErbB-2, and with the physical association of ErbB-2 and Vav2 in mammary tissue, these findings support a relevant role of Vav2 as a downstream effector of neuregulin/ErbB-2 signals for alveolar morphogenesis of the mammary gland in vivo. Furthermore, they provide clear evidence of the as yet suggested function of Vav2 in signaling of receptor tyrosine kinases that lead to a concrete biological response.
Mutational analysis revealed that a functional Dbl-homology domain is necessary and sufficient for Vav2 to elicit alveolar morphogenesis. The Dbl-homology domain enables Vav2 to function as guanine nucleotide exchange factor towards small GTPases of the Rho family with an as yet unclear specificity (Schuebel et al., 1998; Abe et al., 2000). Disruption of this domain in a mutant Vav2 protein completely abolished its ability to promote alveolar morphogenesis of EpH4 mammary epithelial cells. Moreover, the isolated Dbl-homology domain of Vav2 still retained morphogenic properties, suggesting that the GDP/GTP exchange activity is critical in this process, whereas the SH2/SH3 adaptor domains are dispensable. Similarly, integrity of the Dbl-homology domain in the absence of SH2/SH3 modules is required by Vav3 to promote formation of lamellipodia and membrane ruffling in transfected NIH3T3 fibroblasts (Movilla et al., 1999). In view of these results, it is evident that the Dbl-homology domain of Vav proteins is sufficient to reorganize the actin cytoskeleton, a process that may be required for Vav2-mediated alveolar morphogenesis.
Rho proteins constitute a subgroup of the Ras superfamily of small GTPases that are key regulators of the actin cytoskeleton and of several cellular processes like gene transcription, membrane trafficking, growth, movement and morphogenesis (reviewed in Van Aelst and D’Souza-Chorey, 1997; Hall, 1998; Mackay and Hall, 1998). Rho small GTPases switch from an inactive GDP-bound state to an active GTP-bound state; this process is accelerated by guanine nucleotide exchange factors containing a Dbl-homology domain. Activated small GTPases trigger actin polymerization to form stress fibers, focal adhesion complexes, lamellipodia, membranes ruffles and filopodia. An increasing amount of genetic studies in invertebrates provide ample evidence that GEFs and Rho proteins play a central role in several morphogenic events during the development of a multicellular organism. It has been shown that the putative GDP/GTP exchanger DRhoGEF2 and the small GTPase DRho1 are part of a signaling pathway that determines cell shape changes during gastrulation of Drosophila [page 63↓](Barrett et al., 1997; Häcker et al., 1998). The recently described Drosophila Trio GEF or its C. elegans homolog UNC-73 have been shown to regulate axon guidance and cell migration in the central nervous system (Steven et al., 1998; Awasaki et al., 2000; Liebl et al., 2000; Newsome et al., 2000; Bateman et al., 2000). Still life, another DrosophilaRhoGEF, has been identified in a mutational screen for motor activity defects, and its absence determines reduced locomotion as well as male infertility (Sone et al., 1997). Furthermore, Rho proteins have also been shown to control dorsal closure at late stages of Drosophila embryogenesis (Magie et al., 1999), directed cell movement in the Drosophila ovary (Murphy et al., 1999) and planar polarity in eyes and wing hair (Eaton et al., 1995; Eaton et al., 1996; Strutt et al., 1997). In mammals, the complexity of the small GTPase superfamily, and therefore the potential redundancy of its members, has made the study of their role during embryogenesis difficult. In humans, mutations of the FGD1 (faciogenital dysplasia) gene, which encodes the Vav-related guanine nucleotide exchanger FGD1, cause an X-linked autosomal developmental disorder (known as Aarskog-Scott syndrome or faciogenital dysplasia) involving skeletal and genito-urinary anomalies (Pasteris et al., 1994; Olson et al., 1996; Zheng et al., 1996). In view of the accumulating evidence of a role for GEFs and Rho proteins in epithelial development, it is conceivable that Rho small GTPases are downstream effectors of Vav2 in the mammary gland. Matrigel assays with EpH4 cells overexpressing constitutively active or dominant negative forms of Rho, Rac and Cdc42 small GTPases may elucidate their contribution to alveolar morphogenesis.
EpH4 mammary epithelial cells express endogenous c-Met and ErbB receptors and are competent to elicit biological responses following stimulation by HGF/SF or neuregulin, the ligands for such receptors (Niemann et al., 1998). EpH4 cells exhibit ductal growth when cultured on Matrigel in the presence of HGF/SF, while they form alveoli-like structures upon neuregulin treatment. Therefore, they provide a versatile system to study different developmental events of the mammary gland in vitro. A Dbl-defective mutant of Vav2 did not [page 64↓]affect cell proliferation but completely blocked neuregulin-induced morphogenesis in a dominant-negative manner (Fig. 11). In contrast, this catalytically-inactive Vav2 mutant did not interfere with the morphogenic stimulus of HGF/SF for ductal growth (Fig. 12). These results suggest that Vav2 is specifically involved in the morphogenic but not the mitogenic signaling cascades of neuregulin/ErbB-2, whereas other intracellular effectors may mediate the stimulatory effects of HGF/SF and c-Met. It is known that one such effector is the multiadaptor protein Gab1 (Grb2-associated binder; Holgado-Madruga et al., 1996). Gab1 specifically couples to activated c-Met (Weidner et al., 96; Schaeper et al., 2000), thus transmitting unique HGF/SF signals for essential developmental processes during embryonic life; genetic analysis revealed that both Gab1- and c-Met-deficient mice exhibit similar phenotypes (Bladt et al., 1996; Sachs et al., 2000). In the last years, several multiadaptor proteins have been described as specific substrates of receptor tyrosine kinases and key transducers of their biological function. DOS, a Gab1-related multiadaptor protein of Drosophila, mediates retinal development downstream of the receptor tyrosine kinase Sevenless (Herbst et al., 1996; Raabe et al., 1996). IRS-1 and IRS-2 are major substrates of the insulin and insulin-like growth factor receptors (Sun et al., 1991; Sun et al., 1995). Mice lacking IRS-1 are mildly insulin-resistant due to compensation by IRS-2, but show a dramatic impairment of glucose uptake in skeletal muscle (Araki et al., 1994; Tamemoto et al., 1994; Yamauchi et al., 1996; Bruning et al., 1997). Disruption of the IRS-2 gene results in diabetes (Withers et al., 1998), thus clearly showing distinct and critical roles for IRS-1 and IRS-2 in different tissues that are targeted by insulin. Recently, novel Dok proteins have been characterized as interaction partners of c-Ret and, like the receptor, induce neurite outgrowth in PC12 cells (Grimm et al., 2001). The present work supports a putative role of Vav2 in mediating neuregulin signals for mammary alveolar morphogenesis. Moreover, all these studies provide substantial evidence that specific biological responses may result from specific signaling pathways involving particular adaptors and hence, they put forward an alternative mechanism for the generation of the unique developmental responses of receptor tyrosine kinases.
Recent studies show that activation of Vav2 and RhoA also inhibits HGF/SF-induced cell scattering upon overexpression in Madin-Darby canine kidney cells (MDCK; Kodama et al., 2000). This observation suggests a negative role of Vav2 in branching morphogenesis and raises the possibility of a dual role in vivo. Vav2 may therefore mediate neuregulin-induced alveolar morphogenesis while simultaneously repressing HGF/SF-induced ductal growth. Other groups, in contrast, report that activation of RhoG is essential for differentiation and neurite outgrowth in PC12 cells following NGF treatment (Katoh et al., 2000). In the present work, catalytically active Vav2 as GDP/GTP exchanger is strictly required for neuregulin-induced alveolar morphogenesis. Future experiments in more physiological systems are required to clarify the regulatory role of GEF factors and their targets in different biological responses.
This work provides evidence that Vav2 and ErbB-2 may be functionally associated in vivo to promote alveolar morphogenesis of mammary epithelium. However, it does not resolve the issue whether Vav2 is a direct or indirect substrate of the ErbB-2 kinase. A TrkErbB2-Vav2 fusion protein becomes morphogenic upon NGF stimulation (Fig. 7); nevertheless, it is possible that an additional kinase is recruited to this hybrid protein via Vav2. Indeed, activation of all Vav proteins by phosphorylation through Src tyrosine kinases has been reported (Crespo et al., 1997; Schuebel et al., 1998; Movilla et al., 1999). In addition, recent studies show that the PDGF receptor stimulates Vav2 through tyrosine phosphorylation by Src (Chiariello et al., 2001). Recently, in vitro kinase assays were here performed with immunoprecipitated TrkErbB-2 receptor from lysates of stimulated cells, together with recombinant Vav2 protein as substrate. In line with the aforementioned report, these data suggests that Vav2 is not directly phosphorylated by ErbB-2 in vitro (data not shown). It is thus possible that the recruitment and activation of Src by ErbB-2 represents an intermediate step resulting in the engagement of Vav2 for morphogenesis. However, activation of Src has [page 66↓]largely been implicated in the etiology of ErbB2-overexpressing breast tumors but not in normal mammary development (Muthuswamy and Muller, 1994; Muthuswamy and Muller, 1995a); in fact, targeted overexpression of activated c-Src in the mammary epithelium results in epithelial hyperplasia and impaired lobulo-alveolar development followed by severe lactational failure (Webster et al., 1995). Latest experiments show that phosphorylated Vav2 interacts with the SH2 domain of Src (data not shown). This interaction is indeed likely, since Vav2 contains a Src-binding consensus motif involving the regulatory tyrosine Y172; however, it requires previous phosphorylation of Vav2. It is therefore possible that ErbB-2 phosphorylates Vav2 first on some of its various, non-regulatory tyrosine residues, thus creating docking sites for Src other than the predicted consensus motif, and then Src futher phosphorylates and activates Vav2. In support of this hypothesis, none of the known autophosphorylation sites of ErbB-2 corresponds to the optimal Src consensus binding motif (Songyang et al., 1993), though direct interaction between Src and ErbB-2 was here observed in the yeast system and has also been reported by others (Muthuswamy and Muller, 1995). Similarly, the direct binding site for Src on the PDGFβ receptor (DGHEYpYIpYVDP; Mori et al., 1993) does not match the predicted motif; instead, it resembles the amino acid sequence of ErbB-2 encompassing tyrosine Y3 (pYLVP), which was mapped in the yeast system as part of the Src binding site (Table 2). However, such sequence is not present in Vav2, thus raising the possibility of a broader range of phosphotyrosine-containing sequences that may be recognized by the SH2 domain of Src. Alternatively, it has been proposed that Src phosphorylates its own binding sites on the EGF receptor (Olayioye et al., 1999); such a model would account for simultaneous binding of Src and activation of Vav2. Taken together, these latest experiments favor that the indirect recruitment of Vav2 to ErbB-2 is of physiological relevance. Nevertheless, further research that addresses the role of direct and indirect binding of Vav2 to ErbB-2 may contribute to understand the mechanisms of Vav2 recruitment and activation in neuregulin signaling.
Previous studies suggest that activation of the MAPK/ERK (mitogen-activated protein kinase/extracellular signal regulated protein kinase) pathway is a necessary step in neuregulin-induced alveolar morphogenesis (Niemann et al., 1998). Indeed, MAP kinases are stimulated by all ligand-activated combinations of ErbB receptors (Ben-Levy et al., 1992; Graus-Porta et al., 1995; Karunagaran et al., 1996; Pinkas-Kramarski et al., 1996; Dankort et al., 1997; Pinkas-Kramarski et al., 1998). The MAPK pathway is triggered via recruitment of Grb2 or Shc/Grb2 complexes by the activated receptors (reviewed in Schlessinger and Bar-Sagi, 1994 and references therein). It has been shown that Grb2 constitutively binds Sos (Son of Sevenless) via its SH3 domains. Association of the SH2 domain of Grb2 with activated receptor tyrosine kinases leads to recruitment of the GDP/GTP exhanger Sos to the membrane, where it can switch the membrane-anchored Ras GTPase to the active GTP-bound form. Activation of Ras triggers a signaling cascade of kinases involving Raf (also termed MAP/ERK kinase kinase MEKK, or MAPKKK), MAPK/ERK kinase (MEK, also termed MAPKK) and finally the MAPK, which ultimately regulates transcription (reviewed in Schaeffer and Weber, 1999). Tyrosines Y2 and Y4 of the ErbB-2 multidocking site directly bind Grb-2 and Shc, respectively (Ricci et al., 1995; Dankort et al., 1997), and Grb2/Sos/ErbB-2 complexes have been detected in breast cancer cells (Janes et al., 1994). Moreover, different patterns of MAPK activation by α- and β-isoforms of NRG-1 and NRG-2 were reported (Pinkas-Kramarski et al., 1998); it has been suggested that duration of coupling to MAPK pathway contributes to the signaling specificity by several ErbB heterodimers, and the presence of ErbB-2 in receptor combinations contributes to prolong MAPK activation (Karunagaran et al., 1996). Overexpression of constitutively active Vav proteins does not activate MAPK; instead, Vav proteins strongly activate the c-Jun N-terminal kinase/stress-activated protein kinase (JNK/SAPK; Crespo et al., 1996; Abe et al., 2000; data not shown). Activation of JNK/SAPK by Vav proteins depends on the integrity of the Dbl-homology domain, as it requires activation of the small GTPase Rac1. Nevertheless, a connection between Vav2 and Ras has been suggested by synergism for cellular transformation (Schuebel et al., 1998). It has been shown that Vav indirectly enhances Ras signaling; Vav-mediated [page 68↓]activation of small GTPases leads to subsequent activation of p21-activated protein kinases (PAKs), which in turn activate Raf and MEK (Bustelo, 2000 and references therein). It is still unclear whether Vav2 engages similar phosphorylation cascades. The possible requirement of JNK/SAPK activation in neuregulin-induced alveolar morphogenesis therefore requires attention.
It is clear that activation of the transcription factor Stat5a (signal transducer and activator of transcription 5a) is critical in lobulo-alveolar morphogenesis. High levels of activated Stat5a are found in the mammary gland at late pregnancy and during lactation. Inactivation of the stat5a gene is accompanied by failure in terminal mammary differentiation but normal production of milk proteins during pregnancy (Liu et al., 1997). Moreover, activation and function of Stat5a during alveolar morphogenesis seems to be located downstream of ErbB-4 and prolactin signaling (Jones et al., 1999; Ihle and Kerr, 1995). Phosphorylation and activation of Stat5 following neuregulin treatment has been observed in NIH3T3 fibroblasts overexpressing both ErbB-2 and ErbB-4 (Olayioye et al., 1999); this observation indicates that either heterodimers of ErbB2/erbB4, homodimers of ErbB-4 or both are responsible for Stat5 activation. Since transgenic mice overexpressing dominant-negative ErbB-2 and ErbB-4 receptors in the mammary gland also show lactational failure, it is likely that heterodimers of ErbB-2 and ErbB-4 control the activity of Stat5a in vivo. The activity of Stat5a in EpH4 cells that overexpress Vav2 proteins has not been evaluated.
The results presented in this work suggest that Vav2 is a specific interacting partner of the ErbB subgroup of receptor tyrosine kinases. This particular affinity of Vav2 for ErbB receptors has a functional outcome: Vav2 can mediate specific signals from these receptors to elicit alveolar morphogenesis of mammary epithelial cells. The enzymatic activity of Vav proteins as GDP/GTP exchangers may provide a direct link to cytoskeletal rearrangements, an essential step in morphogenic events. In vivo, Vav2 and ErbB-2 co-localize and are associated [page 69↓]in mammary alveolar epithelium during pregnancy, while neuregulin is simultaneously synthesized in the mammary stroma. Therefore, this work supports a model whereby neuregulin activates ErbB-2 in mammary epithelium, which then recruits Vav2 to trigger unique signaling cascades that lead to lobulo-alveolar morphogenesis of the gland during pregnancy.
As discussed above, there are some mechanistic points of ErbB-2/Vav2 signaling that still need to be clarified. In addition, data on the function of Vav2 in mammary development in vivo are required. Genetic ablation of the vav2 gene in mice revealed defective immune response to thymus-independent antigens, and the additional loss of vav led to a severe defect in B cell maturation (Doody et al., 2001; Tedford et al., 2001); however, vav2-deficient mice apparently lack an overt epithelial phenotype, and it is not clear whether the mammary glands from vav2 null mice undergo normal alveolar morphogenesis during pregnancy. An explanation for this lack of an evident phenotype in epithelia is that Vav3, the other epithelial Vav protein, may compensate for Vav2 function in tissues where both Vav2 and Vav3 are co-expressed, as is the case of the mammary epithelium. Thus, generation of vav2/vav3 double knockout mice may shed light on the physiological function of these guanine nucleotide exchange factors in mammalian epithelial morphogenesis.
|© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.|
|DiML DTD Version 3.0||Zertifizierter Dokumentenserver|
der Humboldt-Universität zu Berlin