Characterization of the phosphotyrosine-binding domain of the Drosophila Shc protein.

The phosphotyrosine-binding (PTB) domain of Drosophila Shc (dShc) binds in vitro to phosphopeptides containing the sequence motif NPXpY, and physically associates with the activated Drosophila epidermal growth factor receptor homologue (DER) in vivo. The structural elements, specificity and binding kinetics of the dShc PTB domain have now been characterized. The dShc PTB domain appeared similar to the insulin-like receptor substrate-1 PTB domain in secondary structure as suggested by Fourier transform infrared spectroscopy. Surface plasmon resonance measurements indicated that the dShc PTB domain bound with high affinity to phosphopeptides (Der) derived from the Tyr1228 site of the DER receptor. The kinetics of the dShc PTB domain-Der phosphopeptide interaction differed from those of a typical SH2 domain-ligand interaction, in that the PTB domain displayed slower on/off rates. Competition binding assays using truncated versions of the Der peptides revealed that high affinity binding to the dShc PTB domain requires, in addition to the NPXpY motif, the presence of hydrophobic residues at both positions −5 and −7 relative to phosphotyrosine. The dShc PTB domain showed a similar binding specificity to the human Shc (hShc) PTB domain, but subtle differences were noted; such that the hShc PTB domain bound preferentially to a phosphopeptide from the mammalian nerve growth factor receptor, whereas the dShc PTB domain bound preferentially to phosphopeptides from the Drosophila DER receptor. The invertebrate dShc PTB domain therefore possesses a binding specificity for tyrosine-phosphorylated peptides that is optimally suited for recognition of the activated DER receptor.

Interactions between signaling proteins are often mediated by protein modules such as Src homology (SH) 1 2, SH3, and pleckstrin homology domains (1,2). These domains represent common structural elements found in a diverse array of enzymes involved in signaling transduction, including protein kinases, protein phosphatases, lipid kinases, lipid phosphatases, phospholipases, cytoskeletal proteins, and transcription factors (2)(3)(4). A distinct group of molecules, termed adaptor proteins, possess such modules but lack catalytic domains, and therefore have a more specialized function in promoting intermolecular interactions that control the activation of signaling pathways (5). Shc is a member of the adaptor protein family that apparently participates in multiple signal transduction pathways (6). The mammalian shc gene encodes three widely expressed isoforms with molecular masses of 46, 52, and 66 kDa (7). These Shc proteins become phosphorylated on both tyrosine and serine residues following exposure of cells to a wide spectrum of extracellular stimuli, including growth factors, antigens, and cytokines (8 -12). Shc is also phosphorylated on tyrosine in cells transformed by oncogenic tyrosine kinases such as v-Src, v-Fps, and Bcr-Abl (13,14). Moreover, it has been recently demonstrated that activation of G-protein coupled receptors can result in Shc phosphorylation (15)(16)(17)(18). Phosphorylated Shc proteins may couple to the Ras signaling pathway through the formation of a Shc-Grb2-Sos complex (19 -23).
The diverse functions of mammalian Shc can be rationalized, in part, from its unique structure. The 52 kDa Shc isoform contains three distinct domains: a C-terminal SH2 domain that preferentially recognizes pY(I/E/Y/L)X(I/L/M) motifs (24); a recently identified N-terminal phosphotyrosine-binding (PTB) domain (also termed the PI domain) (25)(26)(27), and a central collagen homology (CH1) region (7). In contrast to SH2 domains, which select residues C-terminal to phosphotyrosine (24), the PTB domain of Shc recognizes tyrosine phosphorylated sites in the consensus sequence XNPXpY (where signifies hydrophobic residues). PTB domain binding specificity is therefore determined by residues N-terminal to the phosphotyrosine (28 -30). In vivo, the Shc PTB domain interacts with a range of activated receptor tyrosine kinases (8 -10), the p145 inositol phosphatase (31), and polyoma virus middle T antigen (32). The coexistence of SH2 and PTB domains in a single molecule potentially allows Shc to interact with a wide range of tyrosine-phosphorylated proteins. The Shc CH1 region con-* This work was supported in part by grants from the Human Frontier Science Program, Bristol-Myers Squibb, the National Cancer Institute of Canada, and the Medical Research Council of Canada, and by a Howard Hughes International Research Scholar award (to T. P.). 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  tains a principle phosphorylation site at Tyr 317 , within the sequence YVNV, which when phosphorylated binds the SH2 domain of Grb2. Multiple potential SH3 binding sites are also found in the CH1 region, together with a motif that mediates interactions with ␣and ␤-adaptins, and might therefore regulate endocytosis (7,33).
To examine if Shc function is evolutionarily conserved, we have previously cloned a Drosophila shc gene homologue, dshc (34). This gene encodes a 45-kDa protein that is closely related in organization to mammalian p52 Shc and contains a C-terminal SH2 domain, an N-terminal PTB domain, and a central CH1 region. Sequence alignment revealed 51% identity between the PTB domains of p52 Shc and dShc, and the dShc PTB domain was found to interact with phosphopeptides containing an NPXpY motif, including a phosphopeptide modeled after the Tyr 1228 site in the DER receptor tyrosine kinase. Autophosphorylation of Tyr 1228 in DER may therefore create a physiological binding site for the dShc PTB domain, based on the observation that the dShc protein associates physically with activated DER in vivo (34).
Functional PTB domains have been identified in Shc and its relatives ShcB and ShcC (35), as well as in the IRS-1 and IRS-2 proteins that serve as prominent substrates for the insulin receptor and the cytokine-activated tyrosine kinases (36). Binding studies and structural analysis have shown that both Shc and IRS-1 PTB domains recognize a ␤-turn structure adopted by the sequence NPXpY (28 -30, 37-42). However, they differ from each other in that the Shc PTB domain prefers a hydrophobic residue at the Ϫ5 position N-terminal to Tyr(P), while the IRS-1 PTB domain favors a patch of hydrophobic residues located at positions Ϫ6/Ϫ7/Ϫ8 (37)(38)(39). Here, we have investigated the specificity, binding kinetics, and structural composition of the dShc PTB domain. Our results suggest that the dShc PTB domain has many features in common with its mammalian counterparts, but shows a unique specificity, which is likely to be of physiological relevance.

Expression and Purification of the PTB Domains of dShc and hShc-
The N-terminal region of dShc (residues 1-203) containing the PTB domain was cloned into the pET-4b expression vector (Novagen). PCR was used to amplify the corresponding cDNA and to generate restriction endonuclease sites. Specifically, BamHI (5Ј) and EcoRI (3Ј) sites were added to the ends of the DNA fragment to facilitate its cloning. The resulting pET-4b-dShc PTB plasmid was transformed into Escherichia coli BL21(DE3) cells, and the transformed bacteria were grown in LB culture at 30°C. pGEX GST-hShc fusion constructs were generated by PCR subcloning of human p52 Shc cDNA fragments encoding residues 1-225. The PCR products were sequenced to confirm fidelity. Expression of the recombinant proteins were induced by adding 1 mM isopropyl-1-thio-␤-D-galactopyranoside into the bacteria culture when its optical density at 600 nm reached at 0.6 -0.8. Bacteria cells were harvested 3 h after induction by centrifugation. The cell pellets were resuspended in a lysis buffer containing 50 mM sodium phosphate, pH 7.5, 150 mM sodium chloride, 0.5 mM EDTA, 5 mM dithiothreitol, and 0.5 mM benzamidine, and sonicated for five bursts of 15 s at 0°C. The lysate was cleared by spinning at 10,000 rpm for 15 min at 4°C.
For purification of dShc PTB, the clear lysate was diluted with two equal portions of 20 mM Tris buffer containing 5 mM dithiothreitol, 0.5 mM EDTA, and 0.5 mM benzamidine, pH 7.5. The diluted lysate was then loaded onto a Tyr(P) affinity column equilibrated with the same buffer. The Tyr(P)-agarose beads were prepared using the protocol provided by the manufacturer (Pharmacia Biotech Inc.). Proteins bound to the beads were eluted using a linear gradient of 0 -0.6 N NaCl. Fractions containing dShc PTB were confirmed by SDS-PAGE and pooled. The PTB protein purified using the Tyr(P)-agarose affinity column was essentially pure as indicated by SDS-PAGE (Fig. 1). For structural studies, the protein was further purified by passing through a Superdex-200 gel filtration FPLC column (Pharmacia). Pure dShc PTB fractions (Fig. 1) were collected and concentrated. It should be noted that all purification procedures were carried out at 4°C under reducing conditions to prevent denaturation and oxidization of the protein. GST-fused hShc PTB protein was purified according to published procedures (27,28).
Peptide Synthesis and Purification-Peptides were synthesized on an Applied Biosystems 431A peptide synthesizer using standard 9-fluorenyl methoxycarbonyl (Fmoc) solid phase chemistry. Phosphotyrosine was directly incorporated as its N ␣ -fluorenylmethoxycarbonyl-O-phosphate-L-tyrosine derivative. Amino acids were activated with 2-(1Hbenzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate, and the peptide chain was elongated on a p-hydroxymethylphenoxymethyl polystyrene type resin (Applied Biosystems). Peptides were cleaved from the resin using a mixture of trifluoroacetic acid/H 2 O/1,2ethanedithiol/thioanisol (90/5/2.5/2.5, v/v). Crude peptides were precipitated with tert-butyl methyl ether cooled on dry ice. After lyophilization, the crude peptide was separated directly on a Primesphere 10 C18-MC HPLC column (250 ϫ 10 mm, Phenomenex). Pure peptides were obtained after eluting with a linear gradient of acetonitrile (0.1% trifluoroacetic acid was added to the solvent). The identities of the peptides were confirmed by mass spectroscopy and amino acid analysis. The concentrations of peptide stock solutions were determined by triple amino acid analysis.
Surface Plasmon Resonance (SPR) Analysis of dShc PTB Binding to Phosphopeptides-Surface plasmon resonance measurements were conducted on a Bia-core apparatus (Pharmacia Biosensor). Immobilization of the Der phosphopeptide on a Biosensor chip was accomplished following essentially the same protocol described earlier (28). Specifically, the peptide IGVPVSVDNPEpYLLNAQK was injected across the surface of the chip at 1 mM in 50 mM HEPES, 1 M NaCl, pH 7.5. The immobilized peptide typically gave resonance signals in the range of 4000 -5000 resonance units when saturated with the dShc PTB protein.
For competition assays, either 0.5 M dShc PTB protein or 1 M hShc PTB-GST fusion protein was mixed with the peptides of interest in aqueous buffer (containing 20 mM MES, pH 6.5, 1 mM dithiothreitol, 0.5 mM benzamidine, and 0.5 mM EDTA) before injecting across the surface of the sensor chip immobilized with the Der peptide. Signals obtained using different concentrations of the competing peptide were monitored and used for calculating the IC 50 values. Kinetic analysis of the binding data were carried out using the BIAevaluation software (Pharmacia, Version 2.1).
In Vitro Peptide Competition Assay-A Drosophila strain carrying the transgene encoding the torso-DER fusion protein (with fusion of the extracellular domain of torso 4021 to the intracellular kinase domain of DER) under the control of the hsp70 promoter in the pW8 transformation vector was stimulated at 37°C for 45 min and then allowed to recover at room temperature. Drosophila lysates prepared after heat shock were probed by GST-dShc PTB fusion protein with or without the presence of the Der phosphopeptides. The protein complexes were washed three times with HNTG (20 mM HEPES, pH 7.5, 150 mM sodium chloride, 10% glycerol, 0.1% Triton X-100, and 1 mM sodium orthovanadate) boiled for 5 min in SDS sample buffer, resolved by SDS-PAGE, and transferred to nitrocellulose. Filters were blocked in 5% bovine serum albumin, 1% ovalbumin in TBS-T and probed with affinity-purified anti-Tyr(P) antibodies (1 g/ml). Immunoblots were developed using an ECL kit (Amersham).
Spectroscopy-FTIR measurements were performed on a Bruker IFS48 spectrometer. The spectrometer was continuously purged with nitrogen to eliminate the background absorbance of water vapor. The spectrum of the protein was recorded for 100 interferograms in D 2 O at a resolution of 2 cm Ϫ1 . The amide I region (1600 -1700 cm Ϫ1 ) was analyzed using the Spectra Calc version of Savitzky-Golay convolution method (43) available from Galactic Industries Co. (Salem, NH).

Kinetics of dShc PTB Binding to a Phosphopeptide
Derived from the DER Receptor-A polypeptide containing residues 1-203 of dShc, and therefore encompassing the PTB domain, was expressed in E. coli. This protein, which did not contain any fused sequences from another source, was purified to homogeneity ( Fig. 1) as described under "Materials and Methods." To test the ability of the purified PTB domain to recognize the Tyr 1228 site of the DER receptor in vitro, an 18-mer phosphopeptide (Ile-Gly-Val-Pro-Val-Ser-Val-Asp-Asn-Pro-Glu-Tyr(P)-Leu-Leu-Asn-Ala-Gln-Lys, designated Der1) was synthesized corresponding to the amino acids flanking Tyr 1228 . This peptide contains an 11-residue extension N-terminal, and 5 residues C-terminal to the phosphotyrosine. The C terminus was extended to include a Lys residue to facilitate immobilization of the peptide to a Biosensor chip.
Binding of the dShc PTB domain to peptide Der1 was studied by flowing various concentrations of the purified PTB protein over a Biosensor chip to which the phosphopeptide was coupled. The PTB protein was found to interact with immobilized Der1 in a concentration-dependent manner as shown in the corresponding SPR sensorgrams ( Fig. 2A). The interaction was specific, since the binding signals were completely eliminated by addition of excess amount of free Der1 peptide. A Scatchard plot of the SPR response (cRU) at equilibrium versus response/ PTB concentration (cRU/[PTB]) yielded a linear line (Fig. 2B), suggesting that binding of the dShc PTB domain to the Der1 peptide followed a bimolecular process. The corresponding dissociation constant, K d , was determined at ϳ0.43 M from the Scatchard analysis. The same data could also be used for kinetic analysis (see "Materials and Methods"). Assuming a simple A ϩ B 3 AB model of association and dissociation, the kinetic analysis yielded an association rate constant, k on , of 3.48 Ϯ 0.38 ϫ 10 4 M Ϫ1 s Ϫ1 and a dissociation rate constant, k off , of 1.74 Ϯ 0.03 ϫ 10 Ϫ2 s Ϫ1 . The K d value could then be calculated as K d ϭ k off /k on ϭ 0.50 Ϯ 0.06 M. This value agrees essentially with that determined using the Scatchard analysis.
It is interesting to compare the kinetics of phophopeptide binding of the PTB domain and the SH2 domain. High affinity phosphopeptides bind to SH2 domain very rapidly with k on rate constants in the range of ϳ1 ϫ 10 5 M Ϫ1 s Ϫ1 to 2 ϫ 10 6 M Ϫ1 s Ϫ1 . Dissociation of the peptide ligand is also relatively fast with typical k off values of ϳ0.1 s Ϫ1 (44). In comparison, the k on and k off values for the interaction of Der1 with the PTB domain were approximately 1 order of magnitude slower than those of a typical peptide-SH2 interaction.
High Affinity Binding of dShc PTB to the Der Phosphopeptides Requires Hydrophobic Residues at Both Ϫ7 and Ϫ5 Positions-Since hydrophobic residues N-terminal to the Tyr(P) have been implicated in the interactions of the Shc and IRS-1 PTB domains with their respective phosphopeptide ligands, we investigated the roles of the hydrophobic residues present at positions Ϫ5, Ϫ7, Ϫ9, and Ϫ11 of the Der1 peptide in binding to the dShc PTB domain. For this purpose, we synthesized a series of truncated peptides based on the sequence of peptide Der1 (Table I). We then tested these peptides for their ability to compete for dShc PTB binding to full-length Der1 using surface plasmon resonance techniques.
As shown in Fig. 3A, solubilized Der1 peptide inhibited binding of dShc PTB to immobilized Der1 at nanomolar concentrations. Complete inhibition was achieved at a peptide concentration of ϳ150 nM. The peptide Der2 lacks the C-terminal 4 residues of Der1. This peptide inhibited PTB-binding with a similar efficiency to that of the free, full-length Der1 peptide, indicating that these 4 C-terminal residues do not contribute significantly to PTB-binding. Deletion of the 4 N-terminal amino acids, 2 residues at a time, yielded peptides Der3 and Der4 containing 9 and 7 residues N-terminal to the phosphotyrosine, respectively. These two peptides also displayed full binding activity, suggesting that the hydrophobic residues Ile Ϫ11 and Val Ϫ9 are not required for efficient PTB binding. However, over 80% of the binding affinity was lost when a single additional residue, Val Ϫ7 , was deleted in peptide Der5, suggesting that the hydrophobic residue at position Ϫ7 relative to phosphotyrosine plays a significant role in dShc PTB-binding. Deletion of Ser at position Ϫ6 in peptide Der6 had no further effect. As shown in Fig. 3B, a residual signal was still detectable when 715 nM of Der6 was used to compete for binding of dShc PTB to the full-length Der1 peptide. However, deletion of Val Ϫ5 rendered the resulting Der7 peptide completely inactive (Table I). To probe the role of Leu-1, we replaced this hydrophobic residue by the hydrophilic amino acid Ser. The binding affinity of the resulting peptide, Der8, was approximately 60% lower than that of Der1. In contrast, replacement of the negatively charged Glu residue within the NPEpY motif by the neutral amino acid Thr had a negligible effect on PTB binding. The pivotal role played by Tyr(P) is demonstrated by peptides Der10 and Der11, in which the presence of non-phosphorylated Tyr resulted in a total loss of binding affinity.
Specificity of the dShc PTB Domain-These assays using the truncated series of Der peptides showed that high affinity binding of the Der peptides to the dShc PTB domain requires the presence of hydrophobic residues at both Ϫ5 and Ϫ7 positions relative to Tyr(P). In addition, a hydrophobic residue at the ϩ1 position appeared to be favored. The binding specificity of the dShc PTB domain was further tested using phosphopeptides derived from the PTB binding sites in various mammalian phosphoproteins, including the insulin receptor (IR), the nerve growth factor receptor (TrkA), the IL4 receptor (mIL4R), and the polyomavirus middle T antigen (mT) ( Table II). The IC 50 values of these peptides in inhibiting dShc PTB binding to immobilized Der1 peptide were determined and compared to that of peptide Der4 to yield their relative affinities. Peptide Der4 was chosen as the reference peptide, since it contains the minimum sequence required for high affinity binding to dShc PTB.
As shown in Table II, the TrkA peptide that binds strongly to the mammalian Shc PTB (45) did not interact with dShc PTB as favorably as peptide Der4. The TrkA peptide contains hydrophobic Ile residues at both Ϫ5 and Ϫ6 positions but lacks a hydrophobic amino acid at the Ϫ7 position, consistent with the view that the Ϫ7 position is important for high affinity binding to dShc PTB. Substitution studies on the NPXpY motif of the TrkA peptide reinforced the conclusion from previous studies (27,28,(37)(38)(39) that Asn Ϫ3 is critical, and Pro Ϫ2 plays an auxiliary role in high affinity PTB binding. Interestingly, the mT phosphopeptide, which was shown to associate tightly with the human Shc PTB domain (28,39), competed poorly for binding to the dShc PTB domain (Fig. 3C). The lack of a hydrophobic residue at the Ϫ7 position in conjunction with the presence of a hydrophilic Ser residue at the ϩ1 position may account for the inability of the mT phosphopeptide to bind strongly to the dShc PTB domain.
In contrast to its rather low affinity for the human Shc PTB domain (28,39), the insulin receptor peptide IR(pY960) was found to effectively inhibit dShc PTB binding to Der1 with an IC 50 of 0.62 M. The IR(pY960) peptide contains a relatively hydrophobic residue at the Ϫ7 position relative to Tyr(P), which may explain its favorable interaction with the dShc PTB domain. As the peptide also contains a Ser residue at the Ϫ5 position, which might offset its binding affinity, it was expected that replacing this residue by a bulky, hydrophobic residue would generate a high affinity ligand for dShc PTB. This proved to be the case as the resulting peptide IR (Ser 3 Ala at Ϫ5) competed effectively against Der1 for dShc PTB binding at nanomolar concentrations (Table II). The relatively low affinity exhibited by the phosphopeptide derived from the IL4 receptor for binding to the dShc PTB domain can be accounted for by the fact that it lacks bulky, hydrophobic residues at both Ϫ5 and ϩ1 positions. Finally, a phosphopeptide representing a high affinity binding sequence for the SH2 domain of Grb2 was found to be completely inactive in PTB binding, consistent with the notion that the binding specificities of the PTB domain and SH2 domain do not overlap outside their common recognition of phosphotyrosine.
Binding of hShc PTB Domain to Der Phosphopeptides-Direct Bia-core comparison of the binding specificity of the human Shc (hShc) PTB domain with that of the dShc PTB domain was conducted using the same set of Der and Trk phosphopeptides (Table III). The hShc PTB domain bound to immobilized peptide Der1 with a typical dissociation constant (K d ) of ϳ6.5  M (data not shown), which is approximately 1 order of magnitude greater than that of the dShc PTB-Der1 interaction. Of interest, the same group of Der phosphopeptides were shown to inhibit the hShc PTB-Der1 interaction in a significantly different fashion from that observed for the dShc PTB domain. In particular, peptide Der4, which extends only to the Ϫ7 residue, inhibited the binding of hShc PTB domain to immobilized Der1 at nanomolar concentrations, whereas addition of residues to the N terminus of peptide Der4 to include Val Ϫ9 and Ile Ϫ11 , as in peptides Der1, Der2, and Der3, greatly decreased the affinities of these peptides for the hShc PTB domain. Similarly, deletion of Val Ϫ7 in peptide Der4 also greatly reduced the affinity of the resulting peptides Der5 and Der6. Peptide Der7, which lacks Val Ϫ5 , showed no detectable binding to the hShc PTB domain. It thus appears that optimal binding of the Der site to the human Shc PTB domain is achieved only when the Der peptide contains residues from the Ϫ7 to ϩ2 positions. Interestingly, the TrkA(pY490) peptide was found to be more effective than peptide Der4 in inhibiting hShc PTB binding to the Der1 chip. This is in distinction to the dShc PTB domain, which binds preferentially to Der4 compared to TrkA(pY490) ( Tables I and II). Thus, while the binding properties of the human and Drosophila Shc PTB domains are similar, a subtle difference exists between these two PTB domains in binding to the Ϫ6/Ϫ7 region of the peptide. It is likely that the hShc PTB domain favors a hydrophobic residue at the Ϫ6 position (such as the Ile Ϫ6 in peptide TrkA(pY490)), whereas the dShc PTB domain biases more strongly to a hydrophobic amino acid at the Ϫ7 position (such as the Val Ϫ7 in peptide Der4). Moreover, hydrophobic residues upstream of the Ϫ7 position may play an important but different role in modulating phosphopeptide binding to the hShc PTB domain versus the dShc PTB domains. As a consequence, the dShc PTB domain binds preferentially to a phosphopeptide derived from the Drosophila epidermal growth factor receptor (DER), while the human Shc PTB domain prefers a phosphopeptide from TrkA, the mammalian nerve growth factor receptor.

I G V P V S V D N P E pY L L N A Q K
Phosphopeptides Compete for Binding of Activated DER to the dShc PTB Domain-The relative affinities of the Der phosphopeptides were also examined in a competition study using a GST-fused dShc PTB domain, and an activated DER receptor variant, in which the extracellular domain of a torso 4021 mutant receptor is fused to the DER cytoplasmic region (34). As shown in Fig. 4A, the GST-PTB protein bound to the activated DER receptor in a Drosophila lysate while GST alone did not, suggesting that the binding was mediated by the dShc PTB domain. Addition of free peptide Der1 to the lysate was found to inhibit the binding of the DER receptor to dShc PTB, with complete inhibition observed at a peptide concentration of ϳ10 M (Fig. 4A). In contrast, the non-phosphorylated peptide Der 11 was not able to compete for binding at the same concentration. At a 5 M peptide concentration, peptides Der1-Der4 competed effectively for dShc PTB binding to the activated DER receptor (Fig. 4B). The efficiency of competition decreased significantly for peptides Der5 and Der6, which lack the hydrophobic residue at the Ϫ7 position. Peptide Der7, which lacks hydrophobic residues at positions Ϫ5 and Ϫ7, was found to be essentially inactive in competition for dShc PTB-binding to the DER receptor. These results are in agreement with those obtained from the Bia-core studies.
Secondary Structure of the PTB Domain of dShc-The structure of the PTB domain was investigated by FTIR. Deconvolution of the corresponding IR spectrum in the amide I region produced five discrete bands centered at 1620, 1633, 1652, 1673, and 1682 cm Ϫ1 , respectively (Fig. 5). The major band at 1652 cm Ϫ1 is assigned to ␣-helix, while absorbances at 1620 cm Ϫ1 , 1633 cm Ϫ1 , and 1682 cm Ϫ1 are attributed to ␤-sheets. The band at 1673 cm Ϫ1 may have originated from turn structure elements in the protein (46). Quantitative analysis of the IR bands provided an estimation of the secondary structure of the dShc PTB domain. The secondary structure contents of the dShc PTB domain based on its IR spectra were then compared ND a IC 50 values were determined using the same procedures as described in the legends to Table I. b Der4 was chosen as the reference peptide due to its similar size to most of the other peptides listed in the table. As well, it contained the minimal elements for high affinity dShc PTB binding. ND, not determined.   Table I except that 1 M hShc PTB (GST fused) was used. b Relative affinity of the phosphopeptides to hShc PTB calculated according to their respective IC 50 values. The relative affinity of peptide Der4 was set at 100%. ND, not determined.
with those of the hShc PTB and of the IRS-1 PTB derived from their NMR and x-ray structures, respectively (Table III) (41,42). Although it is difficult to make a direct comparison between the dShc PTB domain and either the human Shc PTB or the IRS-1 PTB domain, as the dShc PTB construct used in the present study (encompassing residues 1-203 of dShc) is significantly longer than the PTB proteins used in the NMR and the x-ray crystallographic studies (Table IV), it is nonetheless interesting to note that the dShc PTB domain contains a large proportion of ␤-sheet, similar to that found in the IRS-1 PTB domain (Table IV). In contrast, the secondary structure elements of the PTB domains of human Shc and dShc differ significantly from each other. As sequence alignment has indicated that residues important for the tertiary structure of the protein as well as residues critical for phosphopeptide binding are well conserved between these two domains (41), it is likely that these differences arise from local structural re-adjustments. The dShc PTB domain contains a deletion of several residues that correspond to the loop connecting the second ␣-helix (␣2) and the second ␤-strand (␤2) in human Shc PTB (41). The same loop region and the entire ␣2-helix are also absent in the IRS-1 PTB domain (42). It remains to be seen whether the observed difference in the binding specificity of the dShc PTB domain from that of the human Shc PTB domain stem from subtle structural alterations in these loop regions. DISCUSSION It is now established that the PTB domain is a protein module that can mediate the formation of protein complex through its recognition of specific phosphotyrosine-containing motifs (27, 28, 36 -42). Apart from their common recognition of phosphotyrosine, the PTB domain and SH2 domain differ in both their structures and binding affinities (40 -42, 45). Our Bia-core analysis of the interaction of the dShc PTB domain with the 18-mer phosphopeptide Der1, derived from the Tyr 1228 site in the DER receptor tyrosine kinase, suggests that the kinetics of PTB domain binding may also differ from those of the SH2 domain. The PTB domain of dShc has a significantly slower association rate constant (k on ϭ 3.48 Ϯ 0.38 ϫ 10 4 M Ϫ1 s Ϫ1 ) than the SH2 domain of the 85-kDa subunit of the phosphotidylinositol 3Ј-kinase (k on ϭ 1.6 -3.3 ϫ 10 6 M Ϫ1 s Ϫ1 ) (44). Moreover, the observed dissociation rate constant for the dShc PTB-Der1 peptide interaction (k off ϭ 1.74 Ϯ 0.03 ϫ 10 Ϫ2 s Ϫ1 ) is approximately 1 order of magnitude slower than a typical SH2ligand interaction (ϳ0.1 s Ϫ1 ) (3,44). In a similar study, Laminet et al. (47) measured the kinetics of human Shc PTB binding to a tyrosine phosphorylated peptide derived from the c-ErbB2 receptor tyrosine kinase. This study yielded a similar association rate constant (4.7 Ϯ 0.56 ϫ 10 4 M Ϫ1 s Ϫ1 ) and an even slower dissociation rate constant (2.5 Ϯ 0.21 ϫ 10 Ϫ3 s Ϫ1 ). It is therefore likely that an SH2 domain binds to its ligand much more rapidly than does a PTB domain. However, once the complex between the PTB domain and its ligand is formed, its dissociation may be much slower than for a typical SH2-ligand complex. The differences in binding kinetics between the SH2 and PTB domains may provide another level of regulation in signal transduction, aside from the different phosphopeptide motifs recognized by these two domains. While the rapid association and dissociation rates allow an SH2 domain to sample a series of candidate phosphotyrosine-containing sites for opti-  a Secondary structure content of the dShc PTB domain was estimated from its IR spectra using Spectra Calc (43). The NMR structure of the hShc PTB domain (41) and the crystal structure of the IRS-1 PTB domain (42) were used to derive the secondary structure compositons of these two domains. b ␣, ␣-helix; ␤, ␤-sheet; others, secondary structural elements other than the ␣-helix and ␤-sheets, including random coil, loops, and turn structures. mal binding, a PTB domain may be more specific and long-lived in its interaction with its ligands. Regulation at the level of differential binding kinetics may be of general importance in signal transduction. It is of interest in this regard that PTB domains are found in proteins such as Shc and IRS-1, which apparently function as kinase substrates and subsequently as docking proteins for polypeptides with SH2 domains. In contrast, SH2-containing proteins are frequently enzymes that regulate biochemical signaling pathways and may therefore benefit from rapid detachment from their binding sites.
By immobilizing the Der1 peptide onto a Biosensor chip and studying its interaction with the purified dShc PTB protein in the presence of competing peptides, we were able to determine the binding specificity of the dShc PTB domain by using systematically truncated peptides based on the amino acid sequence of the Der1 peptide. The minimal sequence for high affinity dShc PTB binding was found to be Val-Ser-Val-Asp-Asn-Pro-Glu-Tyr(P)-Leu-Leu (peptide Der4). In addition to residues such as Asn and Tyr(P) in the Asn-Pro-Glu-Tyr(P) motif, which were shown to be indispensable for dShc PTB-binding, several hydrophobic residues in the peptide, including Val Ϫ7 , Val Ϫ5 , and Leu ϩ1 , were found to be crucial for high affinity binding to the dShc PTB domain. Specifically, deletion of Val Ϫ7 resulted in a significant loss of binding affinity, while deletion of Val Ϫ5 rendered the peptide completely inactive. The hydrophobic Leu residue at the ϩ1 position plays an significant but less important role than the hydrophobic residues at positions Ϫ5 and Ϫ7. Nevertheless, substitution of this residue by Ser in peptide Der8 decreased its affinity for the dShc PTB domain by ϳ60%. The relative binding affinities of the Der series of peptides were found to correlate well with their ability to compete for dShc PTB binding to the activated DER receptor tyrosine kinase in vitro.
Our observation that hydrophobic residues at both the Ϫ5 and Ϫ7 positions are needed for high affinity binding of the Der peptide to the dShc PTB domain indicates that the specificity of the Shc PTB domain is conserved from invetebrates to man. The mammalian Shc PTB domain recognizes a bulky, hydrophobic residue at the Ϫ5 position. In addition, a hydrophobic residue at the Ϫ6 or Ϫ7 position relative to phosphotyrosine appears to be favored (27,28,39). Most of the identified Shc PTB binding sites, including those from TrkA, TrkB, TrkC, the epidermal growth factor receptor, middle T antigen, and c-erbB-2 and c-erbB-3, share a consensus sequence of XN-PXpY (27,48). In contrast, both in vitro (39) and in vivo (38) binding studies have shown that the PTB domain of IRS-1 favors a patch of hydrophobic residues at positions Ϫ6/Ϫ7/Ϫ8 N-terminal of the NPXpY motif. The insulin receptor, the IL4 receptor, and the insulin-like growth factor receptor all contain high affinity IRS-1 PTB-binding sites that fit the consensus sequence XXNPXpY. The structural basis for such distinctive recognition of hydrophobic residues N-terminal of Tyr(P) by the IRS-1 and Shc PTB domain has recently been unraveled with the determination of the three-dimensional structures of these two domains in complex with their phosphopeptide ligands, by NMR and x-ray crystallography (40 -42). In each case, the NPXpY motif forms a type I ␤-turn, and the Nterminal residues in the peptide forms an additional ␤-strand on the surface of a ␤-sheet of the PTB protein. However, the Shc PTB domain has a deep hydrophobic pocket for the Ϫ5 residue, while the IRS-1 PTB domain lacks this pocket but has more extended hydrophobic contacts with the Ϫ6 to Ϫ8 residues of the corresponding peptide ligand. Such extensive hydrophobic interactions are not observed for the Shc PTB domain-peptide complex, in which the side chains of residues at Tyr(P)-6 and Tyr(P)-7 positions of the peptide are found to be more solvent-exposed.
Such structural distinctions between the hShc and IRS-1 PTB domains may be used to explain our finding that peptides Der1, Der2, and Der3, which contain hydrophobic residues at the Ϫ11 and Ϫ9 positions, displayed significantly reduced affinities for the hShc PTB domain than their shorter counterpart, peptide Der4. It is possible that the Ϫ9/Ϫ11 hydrophobic residues in the Der peptides cannot be accommodated efficiently by the relatively limited hydrophobic surface in the binding site of hShc PTB domain, and consequently interfere with peptide binding. It is interesting to note that the same residues are tolerated by the dShc PTB domain as peptides Der1, Der2, and Der3 displayed similar affinities to the dShc PTB domain as peptide Der4. The subtle difference between these two closely related PTB domains is also reflected in the observation that peptide Der4, which contains a hydrophobic residue at the Ϫ7 position, is superior in binding to the dShc PTB domain as compared with peptide TrkA(pY490), which lacks a Ϫ7 hydrophobic amino acid but contains an Ile at the Ϫ6 position. In contrast, the latter peptide displayed greater affinity to the hShc PTB domain than the former. These results suggest that the Drosophila and human Shc PTB domains have developed to bind optimally to receptor sites of physiological relevance.