‘Srcasm: a Novel Src Activating and Signaling Molecule*

The Src family tyrosine kinase, Fyn, can facilitate regulation of cell proliferation and differentiation. Mice with mutations in the fyn gene have defects in the brain, immune system, and epidermal differentiation. To identify molecules that may interact with Fyn in the epidermis, we performed a yeast two-hybrid interaction screen of a murine keratinocyte library. A novel adaptor-like molecule was isolated and termed Srcasm for Src activating and signaling molecule. Murine Srcasm is a 52.7-kDa protein that contains a VHS membrane association domain and a number of tyrosine motifs suggesting that it may be a substrate for Src family kinases and serve as an adaptor protein. Northern blot analysis of murine tissues demonstrates that Srcasm expression is highest in brain and kidney. In situ hybridization analysis reveals that srcasm mRNA is expressed in regions of the epidermis and hair follicle where keratinocyte differentiation occurs. In the brain, srcasm mRNA distribution correlates with that of fyn , with both being highly expressed in the hippocampal and cerebellar Purkinje neurons. Fyn can phosphorylate Srcasm, and association of these molecules relies on cooperative binding between the SH2 and SH3 domains of Fyn and corresponding canonical binding sites in Srcasm. Srcasm is capable of interacting with Grb2 and the regulatory subunit of phosphoinositide 3-kinase, p85, in a phosphorylation-dependent polymerase probe. II murine srcasm cloned the Eco RV- HI sites of pKS (cid:1) ; digestion with Cla T7 polymerase digestion with Xba I and with T3 polymerase sense probe. The RNA probes using (cid:3) - 35 S]UTP. In situ hybridization conducted paraffin paraformaldehyde-fixed mu- rine Chromosomal C57BL/6 and Mus in M. to C57BL/6. The primers used to amplify the region around the polymorphism were -TACCT- TCCAGCCTTTCTG-3 and (cid:4) reaction were analyzed to resolve M. alleles. isolated into inserts hemagglutinin onto from HA-tagged then into pCDNA3.1 1 m M NaVO 4 ). The lysates were incubated on ice for 15 min and cleared by centrifugation at 14,000 (cid:2) g for 10 min at 4 °C. The supernatants were collected and assayed for protein content using the MicroBCA protein assay kit (Pierce Chemical Co.). Aliquots of lysate or washed immunoprecipitates were separated by SDS-PAGE and transferred to PolyScreen (PerkinElmer Life Sciences). Western blots were conducted in a standard manner with the indicated antibodies and developed using an enhanced chemiluminescence kit as de- scribed by the manufacturer (Lumilight Plus, Roche Molecular Biochemicals). Affinity Binding Assays— GST fusion proteins were induced with 0.2 m M isopropyl-1-thio- (cid:1) - D -galactopyranoside for 4 h; most proteins were produced in DH5 (cid:3) cells. The GST fusion proteins were purified from bacterial lysates using glutathione agarose (Amersham Biosciences, Inc.). The concentration of protein/bead volume was determined using a bromosulfalein protein assay (30). The purity and integrity of the fusion proteins were confirmed by SDS-PAGE followed by staining with Coo- massie Brilliant Blue. Cell lysates were incubated with a (cid:3) 1 (cid:2) M concentration of the indicated GST fusion protein for 14 h at 4 °C followed by incubation with glutathione-agarose. The bead-bound complexes were washed and subjected to SDS-PAGE and Western blotting with the appropriate antibody.

The Src family tyrosine kinase, Fyn, can facilitate regulation of cell proliferation and differentiation. Mice with mutations in the fyn gene have defects in the brain, immune system, and epidermal differentiation. To identify molecules that may interact with Fyn in the epidermis, we performed a yeast two-hybrid interaction screen of a murine keratinocyte library. A novel adaptor-like molecule was isolated and termed Srcasm for Src activating and signaling molecule. Murine Srcasm is a 52.7-kDa protein that contains a VHS membrane association domain and a number of tyrosine motifs suggesting that it may be a substrate for Src family kinases and serve as an adaptor protein. Northern blot analysis of murine tissues demonstrates that Srcasm expression is highest in brain and kidney. In situ hybridization analysis reveals that srcasm mRNA is expressed in regions of the epidermis and hair follicle where keratinocyte differentiation occurs. In the brain, srcasm mRNA distribution correlates with that of fyn, with both being highly expressed in the hippocampal and cerebellar Purkinje neurons. Fyn can phosphorylate Srcasm, and association of these molecules relies on cooperative binding between the SH2 and SH3 domains of Fyn and corresponding canonical binding sites in Srcasm. Srcasm is capable of interacting with Grb2 and the regulatory subunit of phosphoinositide 3-kinase, p85, in a phosphorylation-dependent manner. The evidence suggests that Srcasm may help promote Src family kinase signaling in cells.
The Src family kinases (SFKs) 1 comprise nine highly similar tyrosine kinases that regulate a variety of cellular responses including proliferation, migration, differentiation, and survival (1). The SFKs share a common domain structure differing primarily in the amino-terminal 60 -80 amino acids (1,2). There are several functional motifs common to all Src kinases: (i) the Src homology 2 (SH2) domain, which binds phosphotyrosine, preferentially within the context of acidic amino acids (3); (ii) the Src homology 3 (SH3) domain, which binds polyproline motifs (4 -6); (iii) the amino-terminal region Src homology 4 domain, which contains consensus sequences for myristoylation and/or palmitoylation (7); (iv) the carboxyl-terminal Src homology 1 domain, which contains the catalytic region and has a short carboxyl-terminal tail containing the major regulatory tyrosine (1,2). Phosphorylation of the carboxyl-terminal regulatory tyrosine leads to an intramolecular association between the phosphotyrosine residue and the SH2 domain; this interaction decreases substrate access to the active site, thereby inhibiting kinase activity (8 -11). When the association of the carboxyl-terminal tyrosine with the SH2 domain is disrupted, kinase activity is increased (1,2).
Insights into the functional roles of SFKs can be gleaned from the limited yet distinct phenotype(s) of mice deficient for a specific kinase. For example, mice lacking Src exhibit defects in bone remodeling leading to osteopetrosis, and src Ϫ/Ϫ endothelial cells demonstrate decreased vascular endothelial growth factor-induced vascular permeability (12,13). Mice deficient for Fyn demonstrate phenotypes distinct from src mutants; these include abnormalities in lymphocyte development, brain structure and function, as well as keratinocyte differentiation (14 -19).
Stratified squamous epithelia are self-renewing tissues that maintain a stable integument throughout the life of the organism by carefully regulating keratinocyte growth and terminal differentiation. The molecular mechanisms governing the switch from a proliferative to a differentiated state in epithelia, although poorly characterized, are essential to an understanding of epithelial physiology and pathophysiology. There is good evidence to implicate Fyn in the regulation of keratinocyte differentiation. It is the only SFK that has increased enzymatic activity during keratinocyte differentiation (17,20). Keratinocytes from Fyn-deficient mice exhibit impaired squame formation and decreased expression of differentiation markers such as filaggrin and transglutaminase, suggesting that these cells display an aberrant differentiation program. This may be the result, in part, of impaired adhesion caused by decreased tyrosine phosphorylation of E-cadherin-associated catenins (17,21). Moreover, elevated Fyn expression in primary murine keratinocytes induces a number of markers characteristic of early differentiation events. This includes decreased cell proliferation, increased transglutaminase expression, and inhibition of EGF signaling (20). However, Sik and Rak, which comprise a tyrosine kinase subfamily that shares some structural features with SFKs, are also activated during calcium-induced keratino-cyte differentiation, and increased Sik activity appears to correlate with increased filaggrin expression (22). Thus, completion of the keratinocyte differentiation program may rely on the concerted action of a variety of tyrosine kinases.
To determine whether Fyn may exert some of its effects by interacting with and phosphorylating other molecules, we performed a yeast two-hybrid interaction screen with a murine keratinocyte library using Fyn as the bait. In this report, we describe the initial cloning and characterization of a molecule capable of interacting with Fyn and other Src family members. We have termed this protein Srcasm: Src activating and signaling molecule. Srcasm is a 52.7-kDa molecule that contains motifs consistent with a substrate for SFKs (23). Srcasm contains an amino-terminal VHS membrane binding domain (24), and motifs predicted to interact with the SH2 and SH3 domains of SFKs, and the SH2 domains of Grb2, and p85 PI3K (3,5). In addition, Srcasm can serve as a Fyn substrate and activate Fyn kinase activity. In situ analysis for mRNA reveals that srcasm and fyn show coincident yet distinct expression patterns in the skin and brain. The characteristics of Srcasm make it a good candidate molecule for transducing some signals from SFKs in the brain and during keratinocyte differentiation.

EXPERIMENTAL PROCEDURES
Isolation of Murine Srcasm-The bait vector was constructed by cloning murine FynT in-frame and distal to the GAL4 DNA binding domain (G4DB) using the pPC97 vector (25,26). A cDNA library derived from cultured primary murine keratinocytes was cloned into the pPC86 vector using SalI-MluI linkers to generate a fusion with the GAL4 activation domain (25,26). The bait vector was transformed into the yeast strain MAV103 using the lithium acetate method (27) and selection on synthetic medium lacking leucine. Expression of G4DB-FynT fusion protein did not stimulate ␤-galactosidase production as measured in a filter assay with 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside as substrate. The yeast transfected with the bait vector was subsequently transfected with the cDNA library; transformants were selected on synthetic medium containing 5 mM 3-amino-1,2,4-triazole in the absence of leucine, tryptophan, and histidine. ␤-Galactosidase activity in ϳ7 ϫ 10 4 transformants was measured by a filter assay, yielding 15 positive clones. DNA containing the positive clones was isolated and sequenced.
Sequence Analysis and Data Base Searches-The original Srcasm nucleic acid sequence was subjected to BLAST analysis in the GenBank data bases at NCBI. Nucleotide and protein sequence analysis was performed using PALIGN and CLUSTAL from Intelligenetics or AlignX (InforMax, Inc.).
Northern Blot Analysis-Total RNA was isolated from murine tissues using RNA STAT 60 (Tel Test "B", Inc.) according to the manufacturer's instructions. The RNA (10 g) was subjected to electrophoresis in formaldehyde buffer and blotted to Hybond-N filter (Amersham Biosciences, Inc.). A 2.1-kilobase SalI-NotI DNA fragment containing the complete Srcasm clone served as template for synthesizing a random primed probe (Megaprime DNA labeling system, Amersham Biosciences, Inc.) with [␣-32 P]dCTP (PerkinElmer Life Sciences). After washing, autoradiographic images of the hybridized filter were obtained.
In Situ Hybridization-A polymerase chain reaction-generated fragment containing nucleotides 238 -474 (unique region) of the murine fyn coding region was cloned into pKSϪ (Stratagene); digestion with ClaI and transcription with T7 RNA polymerase were used to generate the antisense probe, and digestion with XbaI and transcription with T3 RNA polymerase were used to generate the sense probe. A HincII-BglII fragment covering the 5Ј 520 bp of the coding region of murine srcasm was cloned into the EcoRV-BamHI sites of pKSϪ; digestion with ClaI and transcription with T7 polymerase were used to generate the antisense probe, and digestion with XbaI and transcription with T3 polymerase were used to generate the sense probe. The RNA probes were prepared using [␣-35 S]UTP. In situ hybridization was conducted as described (28) using paraffin sections of paraformaldehyde-fixed murine tissue.
DNA Constructs-A SalI-NotI fragment isolated from the pPC86 murine Srcasm clone was cloned into HA-pKS, a plasmid that inserts the hemagglutinin epitope in-frame onto cDNAs isolated from pPC86 libraries. The HA-tagged Srcasm fragment was excised with KpnI and NotI, then cloned into the corresponding sites of pCDNA3.1ϩ (Invitrogen). The 3.1HA-Srcasm d388 (deletion of amino acids 389 -474) construct was created by excising an XhoI fragment from pCDNA3.1 HA-Srcasm vector followed by religation. Both murine FynB and a kinasedefective variant were cloned into pCDNA3.1.
Cell Culture and Transient Transfections-COS-7 cells were maintained in Dulbecco's modified Eagle's medium (Cellgro, Mediatech) containing 10% fetal bovine serum (Hyclone Laboratories), 2 mM glutamine, and antibiotic (200 units/ml ampicillin and 200 mg/ml streptomycin). COS-7 cells at ϳ60 -70% confluence were transfected with the indicated expression vectors using LipofectAMINE (Invitrogen). In general, a 10-cm plate of cells was transfected with 7 g of Mock (pCDNA3.1) or HA-Srcasm DNA Ϯ 1-2 g of the appropriate Fyn expression vector. Cells were transfected for 5 h in Dulbecco's modified Eagle's medium using a ratio of 1 g of DNA/6 l of LipofectAMINE according to the manufacturer's instructions. Cells were allowed to recover in medium with serum for 18 h before lysis.
Immunoblotting and Immunoprecipitation-Cell lysates were prepared using Nonidet P-40 lysis buffer (150 mM NaCl, 20 mM Tris-HCl, pH 7.5, 1% Nonidet P-40, 0.4 mM EDTA, 10 mM NaF, 10 mM sodium pyrophosphate, 1 g/ml aprotinin, 0.1 mM leupeptin, 1 mM phenylmethylsulfonyl fluoride, 1 mM NaVO 4 ). The lysates were incubated on ice for 15 min and cleared by centrifugation at 14,000 ϫ g for 10 min at 4°C. The supernatants were collected and assayed for protein content using the MicroBCA protein assay kit (Pierce Chemical Co.). Aliquots of lysate or washed immunoprecipitates were separated by SDS-PAGE and transferred to PolyScreen (PerkinElmer Life Sciences). Western blots were conducted in a standard manner with the indicated antibodies and developed using an enhanced chemiluminescence kit as described by the manufacturer (Lumilight Plus, Roche Molecular Biochemicals).
Affinity Binding Assays-GST fusion proteins were induced with 0.2 mM isopropyl-1-thio-␤-D-galactopyranoside for 4 h; most proteins were produced in DH5␣ cells. The GST fusion proteins were purified from bacterial lysates using glutathione agarose (Amersham Biosciences, Inc.). The concentration of protein/bead volume was determined using a bromosulfalein protein assay (30). The purity and integrity of the fusion proteins were confirmed by SDS-PAGE followed by staining with Coomassie Brilliant Blue. Cell lysates were incubated with a ϳ1 M concentration of the indicated GST fusion protein for 14 h at 4°C followed by incubation with glutathione-agarose. The bead-bound complexes were washed and subjected to SDS-PAGE and Western blotting with the appropriate antibody.
In Vitro Kinase Assays-Partially purified Fyn (Upstate Biotechnology) was incubated with purified GST fusion proteins, and kinase assays were performed according to the manufacturer's specifications. Srcasm fusion proteins containing amino acid 389 -474 were used because this region contained the Fyn phosphorylation sites. Typical reactions utilized 5 units of Fyn and 0.25 M indicated fusion proteins: GST, Wt (GST-Srcasm; amino acids 389 -474); -SH2L (GST-Srcasm Y457F); -SH3L (GST-Srcasm 421 PPLP to 421 AALA). Phosphorylation was monitored by isotopic labeling with 5 Ci of [␥-32 P]ATP and quantitated using a PhosphorImager. Fold stimulation is reported relative to background values from the Fyn ϩ GST lane.

RESULTS
Cloning and Sequence Analysis of Srcasm-A murine keratinocyte cDNA library was screened using a yeast two-hybrid assay with Fyn as the bait. The screen yielded 12 positive clones; these clones were sequenced, and one was selected for further analysis (Fig. 1A). Data base sequence analysis of the murine cDNA showed significant homology with a human gene called TOM1-like (GenBank AJ010071). The designation TOM1-like is based on limited amino acid homology to the amino-terminal third of the TOM1 molecules (GenBank hTOM1, AJ006973; mTOM1, AJ006972; cTOM1, Y08741); to our knowledge, no experimental characterization of the human TOM1-like molecule has been published. We suggest that the cDNA that was isolated in the screen probably represents the mouse ortholog of human TOM1-like based on the following data. The open reading frame of murine clone encodes a 52.7-kDa molecule containing 474 amino acids. The predicted amino acid sequence and alignment of murine and human cDNA demonstrate 81% homology at the amino acid level, and these molecules share a number of conserved motifs (Fig. 1B). The amino-terminal region amino acids 1-155 comprises a VHS domain (Fig. 1, A and C); VHS domains are found on the amino termini of a number of proteins that are known to associate with endosomal membranes (24,31) and are thought to be involved with receptor tyrosine kinase signaling (32). The VHS domain may help to localize TOM1-like molecules to the cytoplasmic face of endosomal membranes where the acylated SFKs are found (33,34).
The carboxyl-terminal region of the predicted protein contains four motifs thought to be important for association with other signaling molecules (Fig. 1, A and C). A polyproline motif at 420 LPPLP is predicted to interact with the SH3 domain of SFKs; the tyrosine motif 457 YEEI is predicted to be phosphorylated by SFKs and to interact with the SH2 domain (3,5,23,35). In addition, the carboxyl terminus contains the motifs 392 YDNF and 441 YEVM, which, when phosphorylated, are predicted to interact with the SH2 domains of Grb2 and PI3K p85, respectively (Fig. 1, A and C) (3). All four motifs are conserved between the mouse and human molecules (Fig. 1B). The predicted primary structure of the TOM1-like molecule is suggestive of an adaptor molecule; based on the experimental data presented, we propose to rename it Srcasm.
Chromosomal mapping using the Jackson Laboratory mouse BSS interspecific species back-cross panel (29) demonstrated that srcasm localized to the distal segment of mouse chromosome 11. The srcasm gene cosegregates with  (24,31); bold designates the motif predicted to interact with the SH2 domain of Grb2 (3,44); a bold underline designates the motifs predicted to interact with the SH3 and SH2 domains of SFKs, respectively (3,5). Bold italics show the motif predicted to interact with the SH2 domains of PI3K p85 (3). B, sequence homology of murine and human Srcasm molecules. The amino acid sequence of the murine (mSrcasm) and human (hSrcasm) Srcasm molecules were aligned using the PALIGN program from Intelligenetics. The conserved motifs predicted to interact with the SH2 domain of Grb2 and p85 are designated as in A (3,44). The motifs predicted to interact with the SH3 and SH2 domains of Fyn also are conserved (bold underlined) (3,5). C, schematic diagram of Srcasm domain structure. BD, binding domain.
Chad and D11Mit122, resulting in a gene order of D11Mit179 where R is the percent recombination between adjacent loci, and S.E. is the standard error for each R (Fig. 2). Comparison with the human genome confirms that the syntenic organization is conserved between species and maps to the region around human chromosome 17q21-22. Taken together, these data suggest that murine Srcasm and human TOM1-like are orthologs and may function as an adaptor molecule linking signals from SFKs to downstream events.
Expression Pattern of Murine Srcasm-Northern blot analysis of murine tissues identified a single transcript of ϳ2.1 kilobases (Fig. 3). Srcasm mRNA is prevalent in brain and kidney (Fig. 3). An intermediate level of srcasm expression is observed in skin and heart (Fig. 3). This pattern of srcasm expression correlates with the tissues that exhibit phenotypes in fyn mutants, namely abnormal brain development/function and defective keratinocyte differentiation (15,17,36). Low levels of srcasm mRNA are observed in thymus (Fig. 2). No significant srcasm mRNA is seen in liver and spleen.
To determine whether fyn and srcasm were expressed in the same cells, in situ hybridization for fyn and srcasm mRNA was performed on sections of skin and brain, tissues containing intermediate and high levels of this transcript, respectively.
Within the epidermis, srcasm mRNA is expressed most prominently within the lower layers (Fig. 4A, arrows); hybridization for fyn mRNA yielded a weak but consistent signal within the suprabasilar region (data not shown). In the hair follicle, srcasm and fyn mRNA are expressed within the precortical zone of hair follicles (superior to the dermal papillae), where follicular keratinocyte proliferation decreases, and differentiation begins (Fig. 4B, arrows). Srcasm mRNA is more widely expressed throughout the precortical region, whereas fyn mRNA expression is restricted to a more central cell population (Fig. 4B). Srcasm is also expressed in prostate and oral epithelia (data not shown); the data suggest that Srcasm may be expressed by a wide variety of epithelia.
In the brain, srcasm and fyn mRNA demonstrate corresponding yet distinct expression patterns. Both srcasm and fyn mRNA are highly expressed in the hippocampal neurons, whereas a prominent signal was detected in the choroid plexus and ependymal epithelium only with the Srcasm probe (Fig.  5A). In the cerebellum, both srcasm and fyn mRNA are expressed within the Purkinje neurons that line the cerebellar foliations (Fig. 5B). The results of the in situ hybridization studies demonstrate that fyn and srcasm mRNA are probably coexpressed within some of the same cells, as might be expected for a Fyn substrate.

FIG. 1-continued
Srcasm Is Phosphorylated on Tyrosine in the Presence of Fyn-To facilitate the characterization of Srcasm, we inserted a HA epitope onto its amino terminus. When HA-Srcasm was transfected into COS cells, it was not significantly tyrosine-phosphorylated. However, if Srcasm was cotransfected with Fyn, prominent tyrosine phosphorylation of Srcasm was observed (Fig. 6A). Based on sequence analysis, the most likely Fyn phosphorylation sites were carboxylterminal to amino acid 389; to test this hypothesis a deletion mutant lacking amino acids 389 -474 (d388) was cotransfected with Fyn and compared with the native molecule. The d388 molecule was not phosphorylated in the presence of Fyn (Fig. 6B). These data suggest that Fyn phosphorylation sites in Srcasm lie distal to amino acid 388; however, these results do not exclude additional or alternative amino-terminal tyrosines as potential Fyn phosphorylation sites. Both HA-Srcasm and d388 yielded a doublet on Western blotting; the frequency of the Srcasm doublet was variable but was usually observed in immunoprecipitations, but less frequently seen in pulldown experiments and blots of whole cell lysates. The cause of the Srcasm doublets in the full-length protein is unknown, but it does not appear to represent an aminoterminal or carboxyl-terminal proteolytic product given that the HA epitope and the Tyr-457 phosphorylation sites are preserved in both molecular species. Second, the d388 molecule gives a similar doublet. One hypothesis is that conformational changes may cause the doublets to form.
To define further which tyrosines within the carboxyl-terminal region of Srcasm are important for Fyn phosphorylation, we constructed a series of point mutations. Mutation of the putative Src SH2 binding site (Y457F) abolishes nearly all Srcasm tyrosine phosphorylation, confirming this tyrosine as a critical Fyn phosphorylation site (Fig. 6C). Conversion of the potential Src SH3 ligand from 421 PPLP to 421 AALA did not diminish Srcasm phosphorylation, suggesting that this motif is not essential for promoting phosphorylation by Fyn (Fig. 6C). Mutation of both the putative Src SH2 and SH3 binding sites (Y457F/-SH3L) decreased Srcasm tyrosine phosphorylation to levels similar to that of the Y457F mutant (Fig. 6C). Mutagenesis of the putative Grb2 SH2 ligand (Y392F) or the PI3K p85 SH2 binding site (Y441F) resulted in a mild decrease in Srcasm phosphorylation (Fig. 6C). Together, these experiments indicate that Tyr-457, located in the presumed Src SH2 binding site, is the predominant tyrosine residue that is phosphorylated by Fyn.
Characterization of the Fyn-Srcasm Association-Because Srcasm contains the sequences 421 PPLP and 457 YEEI, we hypothesized that these two motifs may interact with the Fyn SH3 and SH2 domains in a cooperative manner. Lysates from cells transfected with Srcasm and Fyn were incubated with GST fusion proteins containing various domains of Fyn (Fig.  7A). The Fyn SH2 domain was capable of interacting with Srcasm, whereas the Fyn SH3 fusion protein bound little or no HA-Srcasm. However, the Fyn SH2ϩSH3 fusion protein bound significantly more than the Fyn SH2 domain alone (Fig. 7A), and a fusion protein containing the amino-terminal half of Fyn bound amounts similar to those of the Fyn SH2ϩSH3 protein.
Quantitative Western blotting using 125 I-protein A demonstrated that approximately three times more Srcasm associated with the Fyn SH2ϩSH3 fusion protein than with the SH2 domain, consistent with cooperative interactions between the SH2 and SH3 domains of Fyn and corresponding ligands in Srcasm. Similar evidence of cooperative binding was observed using Hck fusion proteins; Hck bound more Srcasm than the Hck SH2 domain alone (Fig. 7B). In addition, the Src SH2 domain was also capable of interacting with Srcasm in a phosphorylation-dependent manner. Srcasm association with Fyn also required kinase activity because HA-Srcasm transfected in the absence of Fyn (Fig. 7A) or coexpressed with kinase-defective Fyn (data not shown) did not bind to GST-Fyn in affinity binding assays. Together, these results indicate that Srcasm can interact with multiple members of the Src family in a phosphorylation-dependent manner.
Because 457 pYEEI, a canonical Src family SH2 ligand, is a major Srcasm phosphorylation site, we tested the ability of the Y457F mutant molecule to interact with the Fyn SH2 domain. The Y457F mutant did not significantly associate with the Fyn SH2 domain, demonstrating that Tyr-457 is critical not only for Srcasm phosphorylation but also for its association with Fyn (Fig. 7C).
Srcasm Interacts with the Grb2 Adaptor Molecule and the p85 Subunit of PI3K in a Phosphorylation-dependent Manner-Srcasm contains motifs at tyrosines 392 and 440/441 that are possible binding sites for the SH2 domains of Grb2 and p85, respectively (3). Within Srcasm, the motif 392 YDNF constitutes a good consensus sequence for binding to the Grb2 SH2 domain because it contains a conserved asparagine in the Yϩ2 position (3). The Srcasm motif 441 YEVM constitutes a good consensus sequence for binding to the PI3K p85 SH2 domains, given a conserved methionine in the Yϩ3 position (3). When HA-Srcasm and Fyn were cotransfected, HA-Srcasm bound to p85 and Grb2 GST fusion proteins; however, no binding was observed when Srcasm was transfected alone (Fig. 8, A and B). The blots were stripped and reprobed with ␣-pTyr antibody, confirming that Srcasm bound to p85 and Grb2 was tyrosine-phosphorylated (data not shown).
To confirm binding specificity, Srcasm molecules with mutations of tyrosines 392 and 440/441 to phenylalanine were tested. The Y392F mutant eliminated binding of Srcasm to Grb2 but did not affect its ability to bind to Fyn (Fig. 8A). Similarly, a Y440F/Y441F mutant was unable to bind PI3K p85 yet still retained the ability to interact with GST-Fyn 2ϩ3 (Fig.  8B). In addition, tyrosine 392 is critical for Grb2 binding, whereas tyrosines 440/441 are necessary for p85 binding. These results suggest that Srcasm can associate with Grb2 and p85 and require tyrosine phosphorylation.
The ability of GST-Grb2 and p85 to bind Srcasm directly was also evaluated by Far Western analysis. In contrast to the GST pulldown assays, both molecules bound Srcasm in a phosphorylation-independent manner (data not shown). One possibility is that upon denaturation, the 420 LPPLP motif in Srcasm becomes accessible and is capable of interacting with the SH3 domains present in Grb2 and p85. The results obtained from the Far Westerns as well as the studies in Fig. 8 suggest that Srcasm probably binds Grb2 and p85 directly rather than using an intermediary protein, such as Fyn, as a bridging molecule. Lysates were subjected to immunoprecipitation with ␣-HA followed by Western blotting with ␣-HA. The blot was stripped and probed with ␣-pY. Y457F, mutant that eliminates putative Fyn phosphorylation site and canonical Fyn SH2 ligand; -SH3, 420 LPPLP to 420 LAALA mutant that eliminates putative Fyn SH3 binding site; Y392F, mutant that eliminates putative Grb2 SH2 ligand; Y441F, mutant that eliminates putative p85 PI3K SH2 ligand.

FIG. 4. In situ hybridization for murine srcasm and fyn mRNA in skin.
Formalin-fixed sections of murine skin were examined for srcasm and fyn mRNA expression. A, sense or antisense 35 S-labeled RNA srcasm probes hybridized to sections of footpad and back skin. B, antisense 35 S-labeled RNA probes corresponding to srcasm and fyn were hybridized to back skin, and expression in the hair follicle was determined. DP, dermal papilla. Arrows indicate zones of expression in the suprabasilar epidermis (A) and precortical zone of the hair follicles (B). Light microscopy, magnification ϫ788.
To test for selectivity of binding, we performed similar experiments using Nck, an adaptor molecule containing one SH2 domain and three SH3 domains. Based on sequence analysis, the carboxyl-terminal tyrosine motifs in Srcasm do not appear to be good ligands for the Nck SH2 binding domain (3). Consistent with this sequence analysis, no association between Nck and phosphorylated Srcasm was detected (data not shown).
Srcasm Is a Fyn Substrate That Activates the Kinase in Vitro-One mechanism of activating SFKs involves disrupting the inhibitory intramolecular interactions of the SH2 and SH3 domains with the carboxyl-terminal regulatory tyrosine and the polyproline type II helix linker region, respectively (1,8,11). In fact, a putative activator of these kinases would contain ligands for the SH2 and SH3 domains and could disrupt these inhibitory interactions (1,8,11). Given the structural features of Srcasm, we tested the ability of purified Srcasm fusion proteins to serve as Fyn substrates and stimulate kinase activity as measured by autophosphorylation.
Wild-type GST-Srcasm stimulated Fyn autophosphorylation 3.2-fold relative to GST alone. In contrast, the GST-Y457F/-SH3L double mutant was the weakest activator tested and stimulated autophosphorylation only 10% above background (1.1-fold activation) (Fig. 9, A and B). To determine the relative contribution of the Src SH2 and SH3 ligands in stimulating Fyn kinase activity, the Y457F and -SH3L single mutants were tested. The Y457F mutation, which eliminates the canonical Fyn SH2 binding site but preserves the putative SH3 binding site, stimulated Fyn autophosphorylation 1.6-fold (Fig. 9, A and  B). The -SH3L mutant, which eliminates the putative SH3 binding site but preserves the putative SH2 ligand, stimulated Fyn autophosphorylation 2.8-fold (Fig. 9, A and B).
Regarding Srcasm substrate phosphorylation, both GST-Srcasm and GST-Srcasm-SH3L exhibited ϳ25-fold increase in phosphorylation relative to background, whereas the double mutant molecule was phosphorylated 1.9-fold (Fig. 9, A and C). Phosphorylation of the GST-Srcasm Y457F molecule was increased 4-fold over background (Fig. 9, A and C).
These results confirm that the Srcasm 457 pYEEI motif, the ligand for the Fyn SH2 domain, is important for stimulating Fyn autophosphorylation and for Srcasm substrate phosphorylation. The putative SH3 binding motif, 420 LPPLP, appears to play a role in promoting Fyn autophosphorylation, possibly by disrupting intramolecular SH3-dependent interactions. The data suggest that Srcasm contains two motifs, a probable Lysates were subjected to pulldown assays using 1 M GST-Src SH2, GST-Hck SH2, or GST-Hck fusion protein. The GST protein complexes were analyzed by Western blotting with ␣-HA to detect Srcasm. Data are representative of two experiments. C, tyrosine 457 is the major site of interaction between Srcasm and the Fyn SH2 domain. COS cells were transfected with Fyn and HA-Srcasm or the Y457F mutant. Lysates were split and subjected to ␣-HA immunoprecipitation (IP) or a pulldown assay using GST-FynSH2. The immunoprecipitates and GST protein complexes were analyzed by Western blotting with ␣-HA to detect Srcasm.
FIG. 8. Tyrosine-specific association of Srcasm with PI3K p85 and Grb2. A, Srcasm binds to Grb2 in a phosphorylation-dependent manner, and the Y392F mutation inhibits this association. GST-Grb2, GST-Fyn 2ϩ3, or ␣-HA was incubated with COS cell lysates transfected with HA-Srcasm, HA-SrcasmϩFyn, or HA-Srcasm Y392FϩFyn. The protein complexes were subjected to Western blot analysis with ␣-HA to detect Srcasm. B, Srcasm binds to PI3K p85 in a phosphorylation-dependent manner, and Y440F/Y441F mutations inhibit this interaction. GST-p85, GST-Fyn 2ϩ3, or ␣-HA was incubated with COS cell lysates transfected with HA-Srcasm, HA-SrcasmϩFyn, or HA-Srcasm Y440F/ Y441FϩFyn. The protein complexes were subjected to Western blot analysis with ␣-HA to detect Srcasm. PD, GST pulldown. SH2 and SH3 ligand, that work in concert to promote maximal Fyn autophosphorylation and Srcasm substrate phosphorylation.

DISCUSSION
In an effort to delineate further SFK signaling pathways, we have isolated a novel murine gene that we call Srcasm. Murine Srcasm probably represents the mouse ortholog of the human gene, TOM1-like. Inspection of GenBank sequences indicates that Srcasm and human TOM1-like (GenBank AJ010071) enjoy significant homology, ϳ81%, at the amino acid level (Fig.  1B). Residues that we have demonstrated to be potentially important for Srcasm function are conserved between the two molecules. In addition, murine Srcasm and human TOM1-like map to syntenic regions of mouse chromosome 11 and human chromosome 17, respectively. Our experimental evidence suggests that the TOM1-like molecules are substrates for SFKs, stimulate Fyn autophosphorylation, and have structural features of adaptor molecules. Therefore, we propose the designation Srcasm: Src activating and signaling molecule.
Although virtually nothing was known previously about the function of the human TOM1-like protein, its designation was based on limited amino acid homology within the amino-terminal third of the molecule to a family of proteins termed TOM1 (37). The amino-terminal 155 amino acids of TOM1-like/Srcasm comprises a VHS domain (Fig. 1, B and C) that is thought to promote endosomal membrane association (24, 31) and may be important for receptor tyrosine kinase signaling (32). Significantly, EGF receptor signaling can occur while the receptor is localized within endosomal compartments, and this event is associated with recruitment of Grb2, Shc, and SOS to endosomal membranes (38 -40). Previous work performed in fibroblasts has demonstrated that Src can be found in early endosomes as well (33,34). Moreover, SFKs play a major role in growth factor-induced mitogenesis of fibroblasts as increased expression of Src potentiates mitogenesis, whereas intracellular microinjection of antibodies recognizing these kinases inhibits this process (41,42). Srcasm contains several motifs that would allow it to promote signaling while at endosomal membranes. The protein contains 392 YDNF and 441 YEVM motifs, which, when phosphorylated, are predicted to interact with the SH2 domains of Grb2 and PI3K p85; indeed, Srcasm can bind both Grb2 and p85 in a phosphorylation-dependent manner (Fig. 8). Srcasm also contains a 104 LLNPRY motif and a 143 YLDL motif, which, when phosphorylated, are predicted to interact with the PTB domain and SH2 domain of Shc, respectively (43)(44)(45); a possible phosphorylation-dependent interaction between Srcasm and Shc is being evaluated. Thus, a molecule such as Srcasm may serve as a platform to nucleate functional signaling complexes on the endosome.
The effect of Srcasm on cell signaling remains to be determined, but Srcasm may influence PI3K and Grb2 signaling pathways. Association of Srcasm with p85, the regulatory subunit of PI3K, could target the p110 catalytic subunit to membranous surfaces, such as endosomes (24,31,32,46), promoting its lipid kinase activity. Importantly, endosomes are enriched in a number of polyphosphate inositide lipids, creating a local- FIG. 9. Srcasm is a Fyn substrate that enhances kinase activity. A, partially purified Fyn (Upstate Biotechnology) was incubated with 0.25 M purified GST fusion proteins and subjected to an in vitro kinase assay. The Y457F mutant deletes the proposed ligand motif recognized by Src SH2 domains. The -SH3L mutant ( 420 LPPLP to 420 LAALA) disrupts binding to Src SH3 domains. Phosphorylation was detected by isotopic labeling with [␥-32 P]ATP followed by phosphorimaging analysis. B and C, bar graphs demonstrating relative levels of Fyn autophosphorylation and Srcasm phosphorylation; x axis, GST-Srcasm molecules as indicated above; y axis, fold stimulation relative to background. Data are representative of three experiments.
ized environment to promote signaling (47,48). Increased intracellular phosphoinositide production would promote cell survival by stimulation of p70 S6 kinase and Akt activity (49 -52). It is of interest to note that the PI3K p110 catalytic subunit exhibits high homology with Vps34p, a protein associated with yeast vacuoles, and is probably the yeast ortholog of PI3K (53). The combination of p85 binding to Srcasm and the potential for the catalytic subunit to associate with endosomallike structures may serve as dual anchors to secure binding of PI3K to endosomal membranes and ensure adequate production of phosphoinositol lipids.
The ability of Srcasm to bind Grb2 may represent a novel mechanism to nucleate activation of the Ras/mitogen-activated protein kinase pathway. It is known that Grb2, Shc, and SOS can localize to endosomal membranes during EGF receptor signaling (39,54). In fact, Grb2 appears to play a key role in EGF receptor endocytosis and signaling (55). Although endocytosis is generally thought to be a mechanism for down-modulating receptor signaling, more recent studies suggest that the endosomal location may represent active sites of signaling (38). Within 5 min of insulin addition, the insulin receptor is found in early endosomes and colocalizes with activated forms of the signaling intermediates from the Ras/mitogen-activated protein kinase cascade (56). Moreover, SFKs are activated by EGF and increase the rate of translocation of EGF receptor signaling complexes to endosomes (38,57). The ability of Srcasm to stimulate Src kinases and bind Grb2 may allow it to clear receptors from the cell surface and assemble signaling complexes on the endosome.
Expression of fyn and srcasm is found in cells that are phenotypically affected in fyn mutants. Fyn-deficient mice exhibit structural and functional defects in the hippocampus (15). By in situ hybridization, srcasm and fyn mRNA are coexpressed in hippocampal and Purkinje neurons (Fig. 5); based on these data and the molecular features of Srcasm, Srcasm may mediate some Fyn-dependent signals in these neuronal cell populations.
Although srcasm mRNA appears to be present in basilar epidermal keratinocytes, it is also present in keratinocyte populations undergoing differentiation within the suprabasilar epidermis and precortical zone of the hair follicle. This expression profile partially overlaps with the expression pattern of Fyn (Fig. 4). Significantly, increased expression of Fyn or elevated Fyn kinase activity can induce many aspects of early keratinocyte differentiation (17,20). Furthermore, Fyn-deficient mice also demonstrate abnormal keratinocyte differentiation (17). We hypothesize that under normal physiological conditions, Srcasm may activate Fyn to promote keratinocyte differentiation.
Srcasm is phosphorylated by Fyn in transfected cells and in vitro. Based on sequence analysis, the most likely Fyn phosphorylation sites in Srcasm are distal to amino acid 388; in fact, deletion of this region ablates tyrosine phosphorylation by Fyn (Fig. 6B). The predominant site is probably tyrosine 457 because a Y457F mutation dramatically decreases phosphorylation (Figs. 6C and 9). Secondary phosphorylation sites probably occur at Tyr-392 and Tyr-441 because they are present in motifs that are thought to be recognized by Src kinases (23), and mutations of these residues mildly decrease the level of Srcasm tyrosine phosphorylation (Fig. 6C).
Our experimental data demonstrate that Srcasm is a Fyn substrate that can stimulate kinase autophosphorylation (Fig. 9). Based on structural studies, a potential activator of SFKs has been postulated to contain ligands for the SH3 and/or SH2 domain (11). Phosphorylated Srcasm contains the 457 pYEEI motif, a canonical SFK SH2 domain ligand. It has a stronger affinity for the Fyn SH2 domain than the endogenous 528 pYQPG motif present in the carboxyl-terminal regulatory tail (58). Phosphotyrosine peptide binding studies involving the Src SH2 domain demonstrate that the K d for a pYEEI-containing peptide is ϳ0.2 M but is 29 M for a pYQPG-containing peptide (59). This ϳ150-fold difference in binding affinity suggests that Srcasm phosphorylated at Tyr-457 would be expected to displace the carboxyl-terminal regulatory tyrosine, thereby stabilizing the active conformation of the kinase (Fig. 10).
We have demonstrated that phosphorylated Srcasm associates in a cooperative fashion with the SH2 and SH3 domains of Fyn (Fig. 7A), and it also associates the SH2 domains of Src and Hck (Fig. 7B). Because the 457 YEEI and 420 LPPLP motifs both contribute to Fyn activation as measured by autophosphorylation (Fig. 9), we propose the following model for activation of Src kinases by Srcasm. The initial site of interaction occurs between the SH3 domain and the 420 LPPLP motif in Srcasm (Fig. 10B). Although this interaction is relatively weak (60), it is capable of partially stimulating kinase activity. Once the SFK is activated, Tyr-457, in close proximity to the active site, would be phosphorylated and then engage the SH2 domain to stabilize Srcasm-SFK interactions and maintain the kinase in the fully active configuration (Fig. 10C).
Other molecules such as Nef, FAK, Sin, and CD19 are thought to activate SFKs by associating with the SH2 and/or SH3 domain of the kinase (61)(62)(63)(64). The HIV-1 Nef protein can activate Hck by disrupting the inhibitory intramolecular interaction between the Hck SH3 domain with the polyproline type II helix present in the region linking the SH2 and catalytic domains (65). It has also been demonstrated that a SH3 ligand in the Sin molecule can stimulate c-Src activity (61). Maximal activation of SFKs by FAK requires engagement of both the SH3 and SH2 domains. FAK peptides containing only one binding motif did not stimulate as well as peptides containing both the SH2 and SH3 ligands (64). In contrast to FAK, Srcasm encodes canonical SFK SH3 and SH2 ligands (3,35); therefore, Srcasm may have a stronger binding affinity for SFKs than FAK and lead to a more robust activation of Src kinases. CD19 activates Src kinases by a slightly different mechanism. Multiple phosphotyrosine residues in the cytoplasmic region of CD19 engage the SH2 domains of Lyn, forming a cluster of activated kinases (63,66); this activation mechanism of SFKs does not appear to involve SH3-dependent interactions.
The ability of Srcasm to activate Src kinases and bind multiple signaling molecules is reminiscent of the proposed mechanism for how polyomavirus middle T antigen functions. The tumorigenic capacity of polyomavirus middle T antigen is attributed to its ability to bind PI3K, Shc, and activate c-Src. Mutations that prevent binding of these molecules greatly attenuate its growth-promoting properties (67,68). Given the binding similarities between Srcasm and polyomavirus middle T antigen, it seems plausible that both act as molecular scaffolds to assemble signaling complexes and raise the possibility that Srcasm may exhibit oncogenic potential.
The chromosomal location of Srcasm is particularly intriguing. It is found on mouse chromosome 11 and human chromosome 17q21-22. The gene organization is syntenically conserved in mice and humans and in both cases, places the gene very close to the BRCA1 locus. This region of the chromosome contains a number of genes involved in breast cancer, such as erbB2/Neu and BRCA1 (69). In addition, genes encoding Grb2 and the EGF receptor map nearby and are amplified in breast carcinomas (70,71). Thus, it is tempting to speculate that elevated Srcasm expression could enhance intracellular signaling sufficiently to create a permissive environment that would predispose cells to developing a neoplastic phenotype.
Srcasm is a novel substrate and activator of SFK which stably associates with SFKs in a phosphorylation-dependent manner. In addition, it can associate with p85 PI3K and Grb2. These characteristics suggest that Srcasm lies at an interesting nexus of Src-dependent signaling. Further studies will determine the role of Srcasm in mediating signals from SFKs.