Binding specificity and mutational analysis of the phosphotyrosine binding domain of the brain-specific adaptor protein ShcC.

Shc proteins (hereafter referred to as ShcA) represent major substrates of tyrosine phosphorylation by a wide variety of growth factors and cytokines. We have recently described a novel ShcA-like protein, ShcC, which like ShcA contains an NH-terminal phosphotyrosine binding domain (PTB), a central effector region (CH1) and a COOH-terminal Src homology 2 domain (SH2). Both the SH2 and PTB domains of ShcC bind a similar profile of proteins as the comparable regions of ShcA. In an effort to define the functional differences or similarities between ShcA and ShcC, we have further characterized the PTB domain of ShcC. Using a degenerate phosphopeptide library screen, we show that the PTB domain of ShcC preferentially binds the sequence His-hydrophobic-Asn/hydrophobic-Asn-Pro-Ser/Thr-Tyr(P). This sequence is similar to the binding site for the ShcA PTB domain, suggesting that these two proteins may have overlapping specificities. In addition, random mutagenesis of the ShcC PTB domain has identified several amino acids important for PTB function (Gly, Glu, Ala, Gly, and Asp). Mutation of these amino acids dramatically reduces the affinity of the ShcC PTB domain for the activated epidermal growth factor receptor in vitro.

Tyrosine phosphorylation represents a critical switch in the regulation of cell growth, differentiation, and development. Phosphorylation of cellular proteins on tyrosine residues creates high affinity binding sites for proteins containing Src homology 2 (SH2) 1 domains. SH2 domains recognize tyrosine and the 3-6 amino acids COOH-terminal to the phosphotyrosine. The selectivity of a particular SH2 domain is dictated by these COOH-terminal amino acids. Recently another phosphotyrosine binding domain (PTB) has been described (1)(2)(3). This domain, also known as PI (phosphotyrosine interaction domain) and SAIN (Shc and IRS-1 NPXY binding), recognizes phosphotyrosine in the context of amino acids NH 2 -terminal to the phosphotyrosine. Thus, PTB and SH2 domains represent distinct protein modules that recognize tyrosine-phosphoryl-ated proteins, but under entirely different contexts. PTB domains were first described in the adaptor protein ShcA (1,2). ShcA represents a major target of tyrosine phosphorylation following stimulation by a variety of growth factors and cytokines (4). Upon activation of receptor tyrosine kinases, ShcA becomes physically associated with the receptor and phosphorylated on tyrosine. This association was initially believed to occur through the SH2 domain of ShcA (5). Indeed, the ShcA SH2 binding site on the EGFR was mapped using a combination of in vitro binding and phosphopeptide competition assays (6,7). The peptide selectivity of the ShcA SH2 domain was determined using a degenerate phosphopeptide library screen (8). These results suggested that a number of receptors had putative ShcA binding sites. However, some confusion arose as to the true ShcA binding site due to the finding that ShcA association with the polyoma virus middle T antigen occurred through an Asn-Pro-Thr-Tyr sequence and not the consensus ShcA SH2 binding sequence (9,10). In addition, the association of ShcA with a 145-kDa phosphoprotein in plateletderived growth factor-stimulated cells was shown to occur not through the SH2 domain, but rather through the NH 2 terminus (2). The determination that ShcA contains two distinct phosphotyrosine binding motifs, a COOH-terminal SH2 and a NH 2terminal PTB, provided an explanation for these observations. PTB recognition sites are also present in the nerve growth factor receptor (TrkA), the insuln and insulin-related receptors, interleukin-2 receptor, and the EGFR.
We have recently described the identification of two shc-like genes which we called shcB and shcC (11). shcB is nearly identical in sequence to the partial human shc-like gene sck and most likely represents the mouse homolog of this gene (11). shcC, however, has not yet been found in other organisms. In contrast to the wide expression of shcA, shcC is restricted in expression to tissues of neural origin, suggesting a role for this adaptor protein in brain-specific tyrosine kinase signaling. Like ShcA, ShcC contains an NH 2 -terminal PTB domain, a central proline-rich region (CH1) and a COOH-terminal SH2 domain. In addition, ShcC binds to activated growth factor receptors through both its SH2 and PTB domains. In this report, we have further characterized the PTB domain of ShcC. We have determined the phosphopeptide selectivity of the ShcC PTB domain as well as describe a number of point mutations that dramatically reduce the affinity of the ShcC PTB domain for activated growth factor receptors. These mutations occur in conserved regions of the PTB domain, suggesting an important role for these amino acids in phosphotyrosine recognition and binding.

EXPERIMENTAL PROCEDURES
Peptide Library Screen-The peptide library used for these studies has the sequence Met-Ala-X-X-X-Asn-X-X-Tyr(P)-X-Ala-Lys-Lys-Lys, where X corresponds to any amino acid except for Trp and Cys. This library was synthesized as described previously (8). The theoretical * This work was supported by Grants CA42978, CA55008, and CA63071 from the National Institutes of Health (to C. J. D.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Supported by a National Institutes of Health National Research Service Award.
Random Mutagenesis and Bacterial Expression-The PTB domain of ShcC (amino acids 28 -212) was amplified by polymerase chain reaction (PCR) as described previously (11) and then either directly subcloned into the pGEX bacterial expression plasmids or randomly mutagenized with hydroxylamine treatment (13,14), subcloned into pCRII (Invitrogen) for sequence analysis, and then subcloned into pGEX. For hydroxylamine mutagenesis, 30 l of a PCR fragment encoding the PTB domain was mixed with 150 l of ethylene glycol, then heated to 70°C for 5 min. To this mixture was added 16 l of hydroxylamine solution (0.5 M hydroxylamine, 0.2 M sodium pyrophosphate, pH 6.0) and then heated to 70°C for 20 min. After this incubation 80 l of stop solution (0.6 M Tris, pH 8.0, 1.0 M NaCl, 20% acetone) was added and the mutagenized DNA purified over a G-50 Sephadex column equilibrated with TE (10 mM Tris, pH 7.5, 1 mM EDTA). The resulting DNA was subcloned into the pCRII vector using the TA Cloning Kit (Invitrogen). Point mutations were identified by dideoxy sequence analysis. All the PTB fragments were subcloned into pGEX vectors as BamHI-EcoRI fragments. Mutants 4a and 4b were constructed by digesting the pGEX-PTB constructs for the wild-type and mutant 4 PTB domains with BstEII, which produces two fragments of approximately 2 kbp (encoding amino acids 28 -59 of the PTB) and 3.4 kbp (encoding the amino acids 60 -213 of the PTB domain). Mutant 4a was constructed by ligating the 3.4-kbp wild-type fragment with the 2-kbp fragment from mutant 4. Mutant 4b was constructed by ligating the 3.4-kbp fragment from mutant 4 with the 2-kbp wild-type fragment. Plasmid constructs were sequenced to confirm the presence of each mutation. Mutant proteins were expressed in Escherichia coli strain DH5␣ at 37°C and purified on glutathione-agarose beads after induction with IPTG for 3 h. Beads containing GST-PTB were washed with phosphate-buffered saline containing 20% glycerol, 0.5% Tween 20, 1 mM dithiothreitol, 10 g/ml aprotinin, and 10 g/ml leupeptin, resuspended in a 50% slurry, and then stored at Ϫ70°C. Aliquots of each PTB domain were fractionated on SDS-polyacrylamide gel electrophoresis and the gel stained with Coomassie to assess concentrations.
In Vitro Binding Assays-A431 cells (a human epidermoid carcinoma) were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, penicillin, and streptomycin in the presence of 5% CO 2 . Epidermal growth factor (EGF) was a kind gift of Dr. H. Shelton Earp and used at a concentration of 100 ng/ml. Briefly, lysates were prepared from A431 cells stimulated with EGF for 2 min at room temperature as described previously (11). Approximately equal amounts of each GST-PTB were incubated with equivalent amounts of lysate from EGF stimulated A431 cells at 4°C for Ͼ90 min. These mixes were then centrifuged and pellets washed three times with phosphate-buffered saline containing 20% glycerol, 0.5% Tween 20, 1 mM dithiothreitol, 1 mM sodium vanadate, 10 g/ml aprotinin, and 10 g/ml leupeptin. The pellets were resuspended in 50 l in 2 ϫ SDS sample buffer, boiled for 5 min, fractionated on SDS-polyacrylamide gel electrophoresis, and transferred to Immobilon filters. Filters were probed with either anti-phosphotyrosine antibodies (PY20, Transduction Laboratories) to detect the phosphorylated EGFR or anti-GST antibodies (GST(Z-5)HRP, Santa Cruz Biotechnology) to assure equal amounts of fusion protein were used in the binding experiments. Blots were developed using ECL reagents (Amersham Corp.). Signals were quantitated using a Bio-Rad phosphorimager. Relative binding affinities were determined by dividing the anti-Tyr(P) signal by the anti-GST signal. The resulting ratio for each sample was divided by the ratio for the wild-type PTB domain to normalize the samples relative to the wild-type PTB.

RESULTS AND DISCUSSION
A number of groups have defined the sequence requirements of a phosphopeptide for binding the ShcA PTB domain (12,(15)(16)(17)(18) (Table I). The ShcA and ShcC PTB domains share a high degree of amino acid sequence homology (78% identity; Fig. 1). Given our interest in further defining functional differences or similarities between ShcA and ShcC, we have examined the peptide specificity of bacterially expressed ShcC PTB using a degenerate phosphopeptide library screen (see "Experimental Procedures"). Using this strategy, the ShcA PTB was shown to select phosphopeptides containing the sequence Asn-Pro-X-Tyr(P)-Phe-X-Arg with the strongest selectivity at posi-  (20). The alignment was created using the Pileup program of the Genetics Computer Group software analysis package then imported into the Maligned multiple sequence alignment program and modified based on the reported NMR structure (21). Shown above the alignment is the predicted secondary structure of the PTB domains based on the NMR structure of ShcA. We have not included the ␣1 helix in this alignment, since our PTB constructs did not contain this region. Asterisks indicate the positions of amino acid changes. The arrow indicates the position of the Arg 175 mutation of ShcA (18,21). ShcA and ShcC represent the predicted peptide sequences encoded by the respective mouse genes (GenBank™ accession numbers U15784 and U46854, respectively). The ShcC sequence begins at amino acid 29. The ShcA sequence begins at amino acid 46. The accession numbers for the remaining sequences are listed in Ref. 20.

FIG. 2. Phosphopeptide selectivity of the PTB domain of ShcC.
A degenerate phosphopeptide library was used to examine the specificity of the ShcC PTB domain (see "Experimental Procedures"). The peptides that bound to immobilized GST-ShcC PTB were purified and sequenced, and the sequence of the purified peptides was compared with that of the starting peptide library. The data were normalized such that a value of 1 or less indicates no selectivity for a given amino acid (33). A, B, C, D, E and F show, respectively, the selectivity ot the Tyr(P) Ϫ6 , Tyr(P) Ϫ5 , Tyr(P) Ϫ4 , Tyr(P) Ϫ2 , Tyr(P) Ϫ1 , and Tyr(P) ϩ1 positions. Amino acids are given in their one-letter codes.
tions Ϫ3 and Ϫ2 relative to the Tyr(P) (12) (Table I). Indeed, this motif (Asn-Pro-X-Tyr(P)) is present in a number of receptor tyrosine kinases, including TrkA, EGFR, and the insulin receptor as well as polyoma virus middle T-antigen.
Using a modification of the above phosphopeptide library screen, we determined the phosphopeptide selectivity of the ShcC PTB domain (Table I). Previous experiments with ShcA indicated that Asn at the Ϫ3 position was absolutely required for efficient PTB binding (12). In addition, most of the selectivity of PTB domains appears to be dictated by residues NH 2 -terminal to Tyr(P). Therefore, to increase the sensitivity of our experiments and to examine the importance of residues at Ϫ6 to Ϫ4, the Ϫ3 position of the phosphopeptide library was fixed as Asn. The library consisted of peptides containing the sequence Met-Ala-X-X-X-Asn-X-X-Tyr(P)-X-Ala-Lys-Lys-Lys, where X corresponds to any amino acid except for Trp and Cys. The ShcC PTB domain selects phosphopeptides containing the sequence His-hydrophobic-Asn/hydrophobic-Asn-Pro-Ser/Thr-Tyr(P) (Fig. 2 and Table I). There also appears to be some selectivity at the ϩ1 position for small chain amino acids (Table I). These data suggest that the ShcC PTB may bind to similar phosphoproteins as the ShcA PTB. In agreement with these data, we have shown that the ShcA and ShcC PTB domains bind in vitro to the activated NGFR and EGFR in growth factor-stimulated cells with relatively equal affinities and this binding can be competed away with a phosphopeptide modeled on the Tyr 490 juxtamembrane autophosphorylation site of TrkA (11). In addition, both ShcC and ShcA PTB domains bind a 170-kDa phosphoprotein in EGFstimulated A431 cell lysates (Fig. 3A). The identity of the protein is currently unknown. Thus, the ShcC PTB domain shares common recognition specificities with the corresponding PTB domain of ShcA.
To better understand the importance of particular amino acid residues in the PTB domain for recognition of target phosphoproteins, we set out to mutagenize the ShcC PTB domain (Fig. 1). In the absence of structural information regarding the importance of particular regions of the PTB domain, we employed a random mutagenesis approach with the aim of identifying amino acids important for PTB func-tion. Using hydroxylamine mutagenesis and PCR, we isolated a number of PTB mutants, several of which are impaired in their ability to bind the phosphorylated EGFR (Table II). Mutant 2, which contains a single point mutation (E63G), has dramatically reduced binding (Ͼ10-fold) to the EGFR as compared with the wild-type PTB (Fig. 3). This Glu residue at position 63 of ShcC (amino acid 80 of ShcA) is conserved in 14 of 22 PTB domains described thus far ( Fig. 1 and Ref. 19). Four of the remaining eight PTB domains possess a conserved Asp. These findings suggest that a negatively charged side chain amino acid at this position plays a critical role in the recognition of tyrosine-phosphorylated substrates of the ShcC PTB domain.
In addition to the E63G single mutation, PTB mutant 7, which possesses two tandem amino acid substitutions (G139E/ D140N), also has dramatically reduced binding (Ͼ10-fold) to phosphotyrosine substrates (Fig. 3). These two amino acids occur in a region of the PTB domain which is highly conserved in Shc family members as well as other PTB containing proteins (Fig. 1). In particular, Asp is present in 15 of the 22 PTB domains (19,20), suggesting an important role in PTB function. PTB mutant 4 also has dramatically reduced affinity (Ͼ9-fold) for the activated EGFR. This mutant contains three amino acid substitutions (G32R, A136T, C166R; Table II). The Gly is conserved in 11 of 19 PTB domains, and the Ala is present in 6 of 22 PTB domains, suggesting a conserved role for these amino acids in PTB function. The Cys, however, is only present in 3 of 21 PTB domains, which suggests that this amino acid may not be critical for PTB function (19,20). We have separated this triple mutant into a single mutant containing the G32R mutation and a double mutant containing the A132T and the C166R mutations. The relative binding affinities of these two mutants compared with the triple mutant and wildtype PTB domains were determined. Although the triple mutant had approximately a 90% decrease in binding the activated EGFR, neither the single nor double mutant were severely impaired in binding. The G32R mutation resulted in approximately a 20% decrease in binding, and the A136T/ C166R double mutation resulted in approximately a 50% reduction in binding. We have not assessed the individual con-  tributions of the A136T or C166R mutations. We believe the severity of the triple mutant is due to a synergistic effect of the three mutations on PTB binding function. We do not believe that the triple mutant is defective in binding due to a decrease in the stability of the protein, since equivalent amounts of fusion protein are obtained for mutants 4, 4a, and 4b (data not shown).
During the course of this study, the NMR solution structure of the PTB domain of ShcA complexed to a TrkA phosphopeptide was described (21). The tertiary structure of the PTB domain is composed of two antiparallel ␤-sheets formed by a series of seven ␤ strands and three ␣ helices. The overall topology of the PTB domain bears a striking resemblance to that of another modular domain, the pleckstrin homology (PH) domain, although these two domains lack any sequence homology. The ShcA and ShcC PTB domains share a high degree of sequence identity. Overall these two domains are 78% identical particularly in the regions that form specific contacts with the phosphopeptide as determined by NMR (21). For example, the ␤5 strand is 100% identical in ShcA and ShcC and forms four contacts with the phosphopeptide ligand (21). These findings suggest that the solution structure of the ShcC PTB domain may be very similar to that of ShcA. Therefore, we have analyzed our mutations using the ShcA PTB structure as a framework for comparison. The E63G mutation occurs in the middle of ␣2 helix, which connects the ␤1 and ␤2 strands. These ␤ strands comprise part of a ␤-sheet that forms a hydrophobic pocket into which the phosphopeptide binds. Thus, the E63G mutation likely disrupts important ionic interactions with the ␣2 helix, thereby abrogating phosphopeptide recognition. The G139E/D140N mutations occur in a loop between the ␤5 and ␤6 strands. The ␤5 strand forms several contacts with the phosphopeptide backbone (21). Gly 139 appears important for forming a proper turn between these two ␤ strands, which allows for their antiparallel arrangement. Thus, the G139E/D140N double mutant likely disrupts these contacts by restricting the ability of the loop to form a turn, thereby disrupting the alignment of ␤5 and ␤6 and diminishing the affinity of the PTB for phosphopeptide.
The three mutations present in PTB mutant 4 occur in different regions of the PTB domain. Of particular interest is the fact that the A136T mutation occurs in the ␤5 strand, which forms part of the cleft into which the phosphotyrosine binds (21). Several amino acids in this ␤ strand, including Ala 136 , are in close proximity with the phosphopeptide. Mutation of Ala 136 likely disrupts these contacts, thereby abrogating binding to the activated EGFR. Thus, the A136T mutation likely accounts for the majority of the reduction in EGFR binding by mutant 4. The A136T/C166R double mutant does not appear to be as impaired in binding as the triple mutant (Fig. 3C). Although we have not assessed the individual effects of these two substitutions, we believe that the C166R mutation may have some compensatory effect in the context of the double mutant. This compensation in binding may not occur in the context of the triple mutant.
In addition to mutations which affect phosphotyrosine binding, a number of PTB mutants are unaffected in their interaction with the activated EGFR (Fig. 3). Many of these mutations represent conservative substitutions that likely do not have a profound affect on the structure or interactions with other amino acids within the PTB domain itself or the phosphopeptide. Many of the nonconservative substitutions occur in loop regions that tend to be more resistant to mutational effects due to the ability of these regions to move freely in space.
We have identified several mutations in the PTB domain of a novel adaptor protein, ShcC, which dramatically reduce the affinity of its PTB domain for the activated EGFR. Based on the predicted structure of the ShcA PTB domain, these mutations likely disrupt regions important in phosphopeptide recognition and binding. Several groups have identified additional mutants in the ShcA PTB domain that affect PTB binding to phospho-  The relative binding affinities for mutants 4a and 4b were determined in independent binding experiments with the wild-type and mutant 4 PTB domains only. In these experiments, the relative binding affinities for the wild-type and mutant 4 PTB domains were 1 and 0.055 Ϯ 0.019, respectively. Quantitation of the results from these experiments is shown in Fig. 3B. tyrosine containing proteins (18,20,21). Interestingly, Yajnik et. al. (20) independently isolated an Ala to Thr mutation at amino acid 153 in ShcA, which is identical to the A136T mutation present in mutant 4 of ShcC. This mutation resulted in a 74% reduction in binding of the ShcA PTB to the activated EGFR, further supporting the notion that the A136T mutation of mutant 4 is indeed the critical amino acid mutation affecting binding.
Mutation of Arg 175 of ShcA to either Glu, Met, or Lys completely abolishes phosphotyrosine binding, suggesting a critical role for this Arg in substrate binding (18,21). Based on the recently described structure of the ShcA PTB domain, Arg 175 directly participates in binding the phosphotyrosine residue of the phosphopeptide ligand. Interestingly, this Arg is not absolutely conserved in all PTB domains (19). This finding is in contrast to SH2 domains, which contain an absolutely conserved Arg, mutation of which blocks SH2 binding to tyrosinephosphorylated proteins (22,23). In addition to Arg 175 , mutation of Phe 198 drastically reduces binding (Ͻ1% of wild type) (20). Phe 198 , in contrast to Arg 175 , is conserved in the majority of PTB domains described thus far (19,20). Interestingly, the IRS-1 PTB domain binds to a similar sequence as the ShcA and ShcC PTB domains, yet shares no apparent sequence homology. These observations suggest that although different PTB domains may lack primary sequence homology, they may adopt similar three-dimensional structures. Indeed, the PTB domain of ShcA and the PH domain of pleckstrin share a similar three-dimensional structure in the absence of sequence homology (21). Alternatively, PTB domains lacking a comparable Arg 175 as found in the PTB domains of Shc family members may adopt a different structure and, thus, employ a different mechanism for phosphopeptide recognition and binding. Determining the structures of other PTB domains will address these possibilities.
We have examined the peptide selectivity of the ShcC PTB domain. This PTB domain has a similar selectivity as compared with the PTB domain of ShcA. Similar results were obtained with the SH2 domains of Shc family members (11). The similarity in the sequence and the peptide selectivities of both ShcA and ShcC PTB and SH2 domains suggests that these two adaptor proteins may share overlapping functions, but in different cell types. Indeed, both proteins interact with similar receptors and tyrosine-phosphorylated proteins in vitro (11). The identification of mutant PTB domains presents the possibility of designing ShcC dominant interfering mutants that may block signaling from tyrosine kinases as well as other proteins that signal through Shc family members.
In addition to binding tyrosine-phosphorylated proteins, the PTB domain of ShcA has also been shown to bind phospholipids (21). Furthermore, the SH2 domains of phosphatidylinositol 3Ј-kinase, as well as Src and Abl, have been shown to bind phospholipids (24). This interaction with lipids provides a possible explanation for how ShcA may translocate to the membrane to activate Ras in the absence of direct binding to activated growth factor receptors (25)(26)(27)(28). In addition, the interaction of ShcA with phospholipids may play a role in the regulation of phospholipid metabolism.