Substrate Specificity of Lymphoid-specific Tyrosine Phosphatase (Lyp) and Identification of Src Kinase-associated Protein of 55 kDa Homolog (SKAP-HOM) as a Lyp Substrate*

A missense single-nucleotide polymorphism in the gene encoding the lymphoid-specific tyrosine phosphatase (Lyp) has been identified as a causal factor in a wide spectrum of autoimmune diseases. Interestingly, the autoimmune-predisposing variant of Lyp appears to represent a gain-of-function mutation, implicating Lyp as an attractive target for the development of effective strategies for the treatment of many autoimmune disorders. Unfortunately, the precise biological functions of Lyp in signaling cascades and cellular physiology are poorly understood. Identification and characterization of Lyp substrates will help define the chain of molecular events coupling Lyp dysfunction to diseases. In the current study, we identified consensus sequence motifs for Lyp substrate recognition using an “inverse alanine scanning” combinatorial library approach. The intrinsic sequence specificity data led to the discovery and characterization of SKAP-HOM, a cytosolic adaptor protein required for proper activation of the immune system, as a bona fide Lyp substrate. To determine the molecular basis for Lyp substrate recognition, we solved crystal structures of Lyp in complex with the consensus peptide as well as the phosphopeptide derived from SKAP-HOM. Together with the biochemical data, the structures define the molecular determinants for Lyp substrate specificity and provide a solid foundation upon which novel therapeutics targeting Lyp can be developed for multiple autoimmune diseases.

Protein-tyrosine phosphorylation-mediated signal transduction is essential for a wide range of eukaryotic functions, including cell proliferation, differentiation, migration, apoptosis, and the immune responses (1). This signaling mechanism is often dysregulated in human diseases, due to aberrant activity of the enzymes involved in the regulation of protein tyrosine phosphorylation status (1,2). Thus, a comprehensive understanding of the physiological roles of protein tyrosine phosphorylation and how this process is abrogated in the pathogenesis of human diseases must necessarily include characterization of proteintyrosine phosphatases (PTPs), 4 in addition to protein-tyrosine kinases.
The lymphoid-specific tyrosine phosphatase (Lyp) has garnered tremendous interest due to the observation that a missense C1858T single nucleotide polymorphism in the gene (PTPN22) encoding Lyp is associated with multiple autoimmune disorders, including type I diabetes (3), rheumatoid arthritis (4,5), Graves disease (6), and systemic lupus erythematosus (7). Lyp expression is restricted to human T cells, B cells, and macrophages (8). Lyp exerts an inhibitory effect on the proximal T cell receptor signaling pathways, probably through dephosphorylation of the Lck and ZAP-70 kinases (9 -11). The C1858T single nucleotide polymorphism changes Arg 620 into a Trp within the first Pro-rich region in the C terminus of Lyp, leading to loss of Lyp binding to the Src homology 3 domain of the Src C-terminal kinase (Csk) (3,4). Importantly, the autoimmune-predisposing variant of Lyp appears to represent a gain-of-function mutation, leading to increased inhibition of T-cell signaling relative to the wild-type enzyme (12). Interestingly, a loss-of-function variant of PTPN22 has been linked to reduced risk of systemic lupus erythematosus (13). Hence, Lyp is emerging as a potential target for therapeutic intervention of a broad spectrum of autoimmune disorders. Unfortunately, it remains unclear how an activating mutation in a negative regulator of T cell signaling gives rise to autoimmune diseases. Recent studies of PTPN22 C1858T carriers also point to a role for Lyp in B cell signaling, indicating that a combination of disregulation of T cell, B cell, and possibly macrophage function by the Lyp/R620W mutant may contribute to autoimmunity (14 -16).
Despite its involvement in many autoimmune diseases, the precise biological functions of Lyp in signaling cascades and cellular physiology are poorly understood (17). Key issues that need to be addressed include the full repertoire of Lyp substrates and the signaling cascades modulated by Lyp activity. It is likely that in addition to ZAP-70 and Src family tyrosine kinases, Lyp may act on other as yet unidentified substrates. Identification and characterization of novel Lyp substrates will help define the chain of molecular events coupling Lyp dysfunction to diseases. In the current study, we sought to determine Lyp substrate specificity using an "inverse alanine scanning" peptide library approach (18). The obtained consensus peptide corresponds to a stretch of amino acid sequence in the integrinsignaling adaptor SKAP-HOM (19,20), which is a homolog of Src kinase-associated protein of 55 kDa (SKAP-55) (21). Biochemical and substrate-trapping studies support the notion that SKAP-HOM is a bona fide Lyp substrate. To determine the molecular basis for Lyp substrate recognition, we solved crystal structures of Lyp in complex with the consensus peptide as well as the phosphopeptide derived from SKAP-HOM. Together with the biochemical data, the structures define the molecular determinants for Lyp substrate specificity and provide a solid foundation upon which novel therapeutics targeting Lyp can be developed for multiple autoimmune disorders.

EXPERIMENTAL PROCEDURES
Materials-p-Nitrophenyl phosphate was purchased from Fluke. Glutathione-Sepharose 4B was from Amersham Biosciences. Rabbit Myc antibody was from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Crystallization reagents were from Hampton Research. FLAG tag affinity beads, FLAG antibody, and all other reagents were obtained from Sigma.
Protein Expression and Purification-Expression and purification of the His-tagged Lyp catalytic domain (residues 1-294) were performed as described previously (22). For substrate trapping, the Lyp catalytic domain was subcloned into the pGEX-2T vector, and the GST-tagged Lyp protein was expressed in BL21 Escherichia coli. In general, 3 liters of GSTtagged Lyp transformed E. coli were cultured, induced by 0.6 mM isopropyl 1-thio-␤-D-galactopyranoside, and pelleted by centrifugation at 5,000 rpm. The cell pellets were resuspended in 30 ml of a buffer containing 20 mM Tris, pH 7.5, 150 mM NaCl, 1 mM DTT, 2 mM EDTA, and 1 mM PMSF. The suspensions were twice frozen and thawed, and lysozymes were added at 1 mg/ml and incubated at room temperature for 30 min. Triton X-100 was subsequently added to a final concentration of 1% and incubated for another 20 min. The bacterial lysate was centrifuged at 12,000 rpm at 4°C for 1 h, and the supernatant was collected and incubated with 1 ml of glutathione-Sepharose 4B. The suspension was mixed by end-over-end rotation for 1 h at 4°C. The beads were pelleted at 1,000 rpm for 1 min, and the supernatant was discarded. The beads were washed four times for 10 min each at 4°C each time. The bound GST-Lyp protein was finally eluted with a buffer containing 50 mM Tris, pH 8.0, and 10 mM reduced glutathione. The protein was further concentrated and stored at Ϫ20°C. All Lyp mutants were generated by using the QuikChange site-directed mutagenesis kit from Stratagene.
Phosphatase Assay-Initial rate measurements for the Lypcatalyzed p-nitrophenyl phosphate hydrolysis were carried out as described (22,23). The Lyp-catalyzed hydrolysis of Tyr(P)containing peptides was continuously monitored at 305 nm for the increase in tyrosine fluorescence with excitation at 280 nm (18) Fluorometric determinations were performed on a PerkinElmer Life Sciences 50B fluorometer. All reactions were initiated by the addition of Lyp to a final 3 nM concentration. The data were analyzed using a nonlinear least-squares regression program (KaleidaGraph, Synergy Software) and Igor Software (Wave-Metrics, Lake Oswego, OR).
Peptide Synthesis and Characterization-The synthesis and characterization of the inverse alanine phosphopeptide library were described previously (18). The Lyp consensus peptides (Ac-YGEEpYDDLY-NH 2 and Ac-YGYEpYDDEY-NH 2 ) and the peptide 71 DGEEpYDDPF 79 derived from SKAP-HOM were prepared using standard solid phase Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry. The peptides were purified by HPLC using a Waters Breeze HPLC system equipped with a Waters Atlantis dC18 column (19 ϫ 50 mm). The eluted phosphopeptides were lyophilized and redissolved in water. The phosphopeptide concentrations were determined by complete dephosphorylation using PTP1B and determination of the released inorganic phosphate. The sequence and the purity of the peptides were verified using an Agilent 1200 LC-MS system equipped with an Agilent Eclipse XDB-C18 analytical column (4.6 ϫ 150 mm) and an Agilent 6130 Quadrupole MS detector.
GST Pull-down-Five 10-cm dishes of RAW264.7 or five 150-cm 2 flasks of Jurkat T cells at 90% confluence were treated with 200 mM pervanadate for 5 min and then lysed in a buffer containing 20 mM HEPES pH 7.5, 100 mM NaCl, 0.5% Nonidet P-40, 5 mM iodoacetic acid, and a protease inhibitor mixture and rotated end-over-end at 4°C for 1 h. After the addition of 10 mM DTT and 2 mM EDTA, the supernatant was recovered after centrifugation at 15,000 rpm for 15 min. The supernatant was left on ice for 10 min and then incubated with 5 g of purified GST, GST-tagged wild-type Lyp catalytic domain, or C227S or D195A mutant Lyp for 1 h with end-over-end rotation at 4°C. The GST-Lyp⅐substrate complex was then pulled down by incubating with 20 l of GST-agarose beads and precipitated by centrifugation. The precipitated agarose beads were washed with a binding buffer (20 mM HEPES, pH 7.5, 1 mM DTT, 1 mM EDTA, and 100 mM NaCl) three times, and then an equal volume of 2ϫ SDS loading buffer was added and subjected to Western blot analysis.
Co-immunoprecipitation and Western Blotting-HEK293 cells were transiently co-transfected with FLAG-tagged SKAP-HOM together with Myc-tagged Lyp. The cells were stimulated with 50 ng/ml EGF for 5 min and then collected in lysis buffer (50 mM HEPES, 0.5% Nonidet P-40, 250 mM NaCl, 10% glycerol, 2 mM EDTA, 1 mM DTT, and protease inhibitor). The cell lysates were centrifuged at 15,000 rpm for 15 min after a 1-h end-over-end rotation at 4°C. The complex of SKAP-HOM and Lyp was isolated by 20 l of FLAG tag affinity beads. The isolated complex was subjected to electrophoresis, and the amount of Lyp bound to SKAP-HOM was measured by Myc antibodies.
Crystallization, Data Collection, and Structure Determination-The N-terminal His 6 -tagged catalytically inactive Lyp/C227S (residues 1-294) mutant was crystallized with the phosphopeptides under conditions similar to those reported previously (22). The co-crystals of Lyp/C227S with the consensus or the SKAP-HOM peptide appeared after 3 days and grew to 0.1 ϫ 0.2 ϫ 0.2 mm after 5 days. X-ray diffraction data were collected up to 2.5 Å for Lyp/C227S in complex with the consensus peptide and 2.9 Å for Lyp/C227S in complex with the SKAP-HOM peptide at the APS 19BM beamline (Argonne, IL). The diffraction data were processed with HKL2000 (24). The crystals of Lyp/C227S with the consensus peptide belong to space group P1, whereas the crystals of Lyp/C227S with the SKAP-HOM peptide belong to space group P2 1 , with four Lyp molecules in each asymmetric unit.
Initial phases of the Lyp/C227S-peptide complexes were determined by molecular replacement with the software Phaser from the CCP4 software package. A single chain of the Lyp catalytic domain from Protein Data Bank entry 2QCT was used as the initial search model, with water and I-C11 deleted from the coordinates. After one round of refinement with rigid body by Refmac in the CCP4 package, the coordinates and reflection data were transformed to CNS format, and further refinement was carried out in CNS (25). Composite omitted maps were built, and density modification was carried out to reduce the phase biases during further refinement. The final refined model has R work /R free at 0.162/0.210 for Lyp/C227S in complex with the consensus peptide 1 and 0.164/0.223 for Lyp/C227S in complex with the SKAP-HOM Tyr(P) 75 peptide. All data collection and refinement statistics are listed in Table 2. The final structures were deposited in the Protein Data Bank.

RESULTS AND DISCUSSION
Determination of Lyp Substrate Specificity-Although Lyp is associated with a wide range of autoimmune diseases, its physiological substrates and in vivo function remain poorly defined. Knowledge about the substrate specificity of Lyp will facilitate the identification of its physiological substrates and may ultimately assist in the design of Lyp-specific inhibitors. Previous kinetic and structural studies using synthetic Tyr(P)-containing peptides reveal that PTPs display a range of k cat /K m (substrate specificity constant) values (10 2 to 10 7 M Ϫ1 s Ϫ1 ) for relatively short peptide substrates (26). In addition, the k cat /K m values for the peptides are orders of magnitude higher than that of Tyr(P) alone, suggesting that amino acids flanking the Tyr(P) also contribute to catalytic efficiency (27,28). In fact, substrate recognition by PTPs requires the presence of amino acids on both the N-and C-terminal sides of Tyr(P) (29 -31). However, because of the differences in sequence and size of the individual peptides examined, it is difficult to draw definitive conclusions regarding the structural requirements for PTP substrate recognition. A more systematic and thorough approach, such as the use of peptide libraries, is needed for the determination of substrate specificity for individual PTPs. Indeed, combinatorial peptide libraries have been useful in the determination of optimal amino acid sequence for PTP recognition (18,(32)(33)(34)(35)(36)(37)(38).
To identify novel Lyp substrates in an unbiased manner, we employed an "inverse alanine scanning" strategy (18) to profile the sequence specificity of Lyp. Briefly, each Ala residue in the parent peptide, Ac-AAAApYAAAA-NH 2 , is separately and sequentially replaced by the 19 non-Ala amino acids to generate a library of 153 well defined phosphopeptides (Fig. 1). This method allows the acquisition of explicit kinetic data for all library members and enables the assessment of the contribution of individual amino acids to PTP substrate recognition based on the actual enzymatic activity of the PTP against its putative peptide substrates in solution. The Lyp-catalyzed dephosphorylation of each individual Tyr(P) peptide within the library was monitored by the increase in tyrosine fluorescence (18, 39) at pH 7.0 and 25°C. The k cat /K m value, a measure of substrate specificity, was directly calculated from the reaction progress curve for each peptide (see "Experimental Procedures").
The results of the inverse alanine scanning of the entire peptide library for Lyp are shown in Fig. 1. The k cat /K m values for all 153 peptides ranged from 1.0 ϫ 10 2 to 1.6 ϫ 10 5 M Ϫ1 s Ϫ1 , covering over 3 orders of magnitude and varied significantly for different substitutions within a single position. The data indicate that incorporation of a Glu residue at position Ϫ1 furnishes the single most dramatic enhancement in substrate efficacy. To further substantiate this observation, we also resynthesized the parent peptide Ac-AAAApYAAAA-NH 2 , Ac-AAAEpYAAAA-NH 2 , and Ac-AAAQpYAAAA-NH 2 and assayed them against Lyp under the same conditions. In full agreement with the library screening results, replacement of the Ala at the Ϫ1 position by Glu increased the k cat /K m value by 32-fold over the all-Ala parent peptide (5.0 ϫ 10 3 M Ϫ1 s Ϫ1 ), whereas the Gln substitution was only 6-fold better than the parent peptide (Table 1). A closer examination of the scanning data revealed a number of factors that control Lyp substrate specificity. Notable features of Lyp substrate specificity include the strong preference of acidic residues (Asp and Glu) at the Ϫ1, Ϫ2, ϩ1, and ϩ2 sites, although hydrophobic and aromatic residues could also be accommodated by Lyp at the Ϫ1 (Ile, Leu, Phe, and Tyr) and Ϫ2 (Trp and Tyr) positions. The ϩ4 position is also an important determinant, which favors Trp and Tyr and, to a lesser extent, Glu and Gln as well. Lyp also shows a striking selectivity for Gly at the Ϫ3 position. A modest preference for Trp and Tyr is observed for position Ϫ4, and several structurally diverse residues, including Glu, Leu, Gln, Thr, and Val, can be tolerated at the ϩ3-position. Finally, Lyp dislikes basic residues (Lys and Arg) at all positions.
Based on the results from inverse alanine scanning of the 153-peptide library (Fig. 1), we synthesized two consensus peptides by combining the most preferred residue at each position. The consensus peptide 1 (Ac-YGEEpYDDLY-NH 2 ) and the consensus peptide 2 (Ac-YGYEpYDDEY-NH 2 ) represent the best peptide substrates ever reported for Lyp, exhibiting k cat /K m values of 4.6 ϫ 10 6 M Ϫ1 s Ϫ1 and 2.1 ϫ 10 6 M Ϫ1 s Ϫ1 , respectively, at pH 7.0 and 25°C. The ϳ2-fold lower activity observed for consensus peptide 2 is consistent with the finding that Lyp has a higher preference for Glu than Tyr at the Ϫ2 position (Fig. 1). Previous studies showed that Lyp preferentially dephosphorylates Lck at Tyr(P) 397 in the activation loop over the autoinhibitory Tyr(P) 505 at its C terminus (11,22). Consistent with this finding, the k cat /K m value for the Lck Tyr 394 peptide (Ac-ENDEpYTARE-NH 2 ) is 10.3-fold higher than that of the Tyr 505 peptide (Ac-TEPQpYQPGE-NH 2 ) (Table 1). Notably, the k cat /K m value for the Lck Tyr 394 peptide is still 7.4-fold lower than that for the consensus peptide 1 ( Table 1). The differences in the catalytic efficiency of Lyp toward these peptide substrates can be explained by the specificity data gained from inverse alanine scanning. Although the N-terminal amino acid sequence in the Lck Tyr 394 peptide matches the specificity profile for Lyp, the Thr, Ala, and Arg at the C-terminal side of the Tyr(P) do not correspond to the most favored residues at the ϩ1, ϩ2, and ϩ3 sites. The low activity of Lck Tyr 505 peptide could be due to the fact that, with the exception of Gln Ϫ1 and Glu ϩ4 , none of the other residues outperform Ala at the remaining positions (Table 1). Collectively, the kinetic analyses further validate the inverse alanine scanning approach for studying PTP substrate specificity (18), and the resulting specificity profile leads to the identification of consensus peptides that prove to be the best Lyp substrates known to date.
Sequence Specificity-based Substrate Identification for Lyp-Although PTPs share a common core catalytic domain with a conserved catalytic mechanism, in cellular environments, they carry out highly distinct, substrate-specific dephosphorylation events, which are essential for maintaining normal cellular processes. Knowledge about the substrate specificity of a given PTP can be used to facilitate the identification of its physiological substrates. Our current study establishes that Lyp exhibits   (19,20), which is a homolog of Src kinaseassociated protein of 55 kDa (SKAP-55) (21). Both SKAP-55 and SKAP-HOM are involved in the regulation of integrin activation and the related biological functions in immune cells (19). Whereas SKAP-55 is expressed solely in T lymphocytes (40), SKAP-HOM is more widely expressed in lymphohematopoietic cells (41)(42)(43). SKAP-55 and SKAP-HOM have the same molecular architecture, with an N-terminal dimerization domain, a central pleckstrin homology domain, and a C-terminal Src homology 3 domain. Interestingly, the tyrosine motif 71 DGEEYDDPF 79 located in the linker between the dimerization domain and the pleckstrin homology domain in SKAP-HOM does not exist in SKAP-55.
To examine whether the phosphorylated tyrosine motif 71 DGEEYDDPF 79 from SKAP-HOM could serve as a Lyp substrate, we synthesized the SKAP-HOM Tyr 75 peptide (Ac-DG-EEpYDDPF-NH 2 ) and assayed it against Lyp. Consistent with the prediction, the SKAP-HOM Tyr 75 peptide is an excellent substrate for Lyp, with a k cat /K m value of 4.0 ϫ 10 6 M Ϫ1 s Ϫ1 , comparable with that of the consensus peptide 1 (Table 1). Thus, the kinetic analysis suggests that SKAP-HOM Tyr 75 is a potential cellular target of Lyp. To further establish that SKAP-HOM is a bona fide Lyp substrate, we decided to conduct substrate-trapping experiments in cells.
Two types of "substrate-trapping" mutants have been developed to identify PTP substrates. In the first, the active site Cys residue is replaced by a Ser (e.g. Lyp/C227S) (44), whereas in the second, the general acid Asp residue is substituted by an Ala (e.g. Lyp/D195A) (45). These mutant PTPs retain the ability to bind substrates but are either unable (the Cys to Ser mutant) or severely impaired (the Asp to Ala mutant) to carry out substrate dephosphorylation, allowing the capture of the PTP-substrate complex. In most cases, the Asp to Ala mutant displays a higher affinity toward substrates than the Cys to Ser mutant (45,46). To investigate whether SKAP-HOM can serve as a substrate of Lyp in a cellular context, we first performed a GST pull-down experiment with pervanadate-treated (to increase and preserve tyrosine phosphorylation) RAW264.7 cell lysates with either GST alone, wild-type Lyp fused to GST (GST-Lyp), or the substrate-trapping fusion proteins GST-Lyp/C227S and GST-Lyp/ D195A. Whereas no protein retained by GST or GST-Lyp was detectable, the GST-Lyp/C227S and GST-Lyp/D195A trapping mutants specifically bound SKAP-HOM ( Fig. 2A). The failure to detect association of SKAP-HOM with wild-type Lyp indicates that the interaction between the Lyp trapping mutants and SKAP-HOM requires tyrosine phosphorylation, which is supported by the observation that the trapping mutant (especially GST-Lyp/D195A)-bound SKAP-HOM is tyrosine-phosphorylated ( Fig. 2A). Because SKAP-55 lacks the SKAP-HOM Tyr(P) 75 motif (the corresponding sequence in SKAP-55 is GQDSSDDNH), SKAP-55 is not expected to serve as a Lyp substrate. Indeed, Lyp trapping mutants GST-Lyp/C227S and GST-Lyp/D195A failed to bind SKAP-55 from Jurkat T cell lysates under similar pull-down conditions used for SKAP-HOM. 5 Finally, we analyzed whether Lyp forms a complex with SKAP-HOM inside the cell. Substrate trapping was performed in HEK293 cells expressing the FLAG-tagged SKAP-HOM together with the vector control pCDNA4, Myc-tagged Lyp, and Myc-tagged Lyp/D195A trapping mutant. Protein complexes were isolated from crude cell extracts using anti-FLAG antibodies to capture the FLAG-tagged SKAP-HOM, and the immunoprecipitates were examined for the presence of Myctagged Lyp by immunoblot analysis. As shown in Fig. 2B, SKAP-HOM readily coimmunoprecipitated with the substrate-trapping mutant Lyp/D195A. However, no complex formation was detected with wild-type Lyp or control transfectants lacking Lyp (Fig. 2B). Collectively, the results demonstrate specific interaction between the Lyp substrate-trapping mutant and SKAP-HOM, further confirming SKAP-HOM as a bona fide Lyp substrate.
SKAP-HOM is a substrate for the Src family protein-tyrosine kinase Fyn and is involved in regulating leukocyte adhesion (19, 47, 48). In SKAP-HOM-deficient mice, B cell receptor-mediated proliferation is strongly attenuated, and adhesion of activated B cells to fibronectin as well as to ICAM-1 is strongly reduced (48). In addition, the loss of SKAP-HOM also results in a less severe clinical course of experimental autoimmune encephalomyelitis following immunization of mice with the encephalitogenic peptide of myelin oligodendrocyte glycoprotein. Thus, it appears that SKAP-HOM is required for proper activation of the immune system, probably by regulating the cross-talk between immunoreceptors and integrins in B cells. Future studies will be required to delineate the role of Lypmediated SKAP-HOM dephosphorylation in B cell signaling and behaviors and to furnish new insight into how deregulation of Lyp activity contributes to autoimmune diseases.
Molecular Basis of Lyp Substrate Recognition-To determine the molecular basis of Lyp substrate recognition, we solved the crystal structures of the catalytically inactive mutant Lyp/ C227S complexed with consensus peptide 1 and the SKAP-HOM Tyr(P) 75 peptide at 2.5 and 2.9 Å resolution, respectively (Figs. 3 and 4). Data collection and structure refinement statistics are summarized in Table 2. The final models for both structures include Lyp residues 1-294. Although the two structures belong to different space groups, both of them contain four monomers in an asymmetric unit with a 1:1 binding (Lyp/ C227S to phosphopeptide) stoichiometry within each monomer. Initial F o Ϫ F c omit maps displayed good density for Tyr(P) and its four neighboring residues (Ϫ2, Ϫ1, ϩ1, and ϩ2) at the Lyp catalytic active site (Figs. 3A and 4A). After iterative model building and refinement, the rest of the peptide was modeled into the well defined 2F o Ϫ F c electron density (Figs. 3B and 4B). The peptide backbones of the bound Tyr(P) peptides adopt an "M"-shaped overall conformation, with the pentapeptide core EEpYDD assuming the central "V" shape, which is stabilized by two hydrogen bonds formed by the side chain of Asp 62 with the main chain amides of Tyr(P) and Asp ϩ1 . From the N to C terminus, the peptides interact with several surface loops in Lyp (Figs. 3B and 4B), including the Tyr(P) recognition loop (residues 58 -63), the Lyp-specific insert (residues [35][36][37][38][39][40][41][42], the phosphate-binding loop (P-loop, residues 226 -235), and the Q-loop (residues 269 -278). In particular, the Lyp substratebinding site contains a cluster of polar residues from the P-loop and additional basic residues (Lys 32 , Lys 39 , Arg 59 , Lys 61 , Lys 136 , Lys 138 , and Arg 266 ) that provides a complementary positive surface to the highly electronegative pentapeptide core EEpYDD (Figs. 3C and 4C). This electrostatic complementarity dictates the overall preference of Lyp for acidic residues at the Ϫ2, Ϫ1, ϩ1, and ϩ2 positions.
In the structure of Lyp/C227S-consensus peptide 1 complex, a rich network of interactions is responsible for the precise positioning of the peptide substrate (Fig. 3D). The central Tyr(P) is primarily anchored by six hydrogen bonds with backbone amides of Ser 228 , Ala 229 , Cys 231 , Gly 232 , and Arg 233 , as well as the ⑀-N of Arg 233 in the P-loop. Together with Van der Waals contacts with the side chains of Tyr 60 , Ile 63 , Ala 229 , and aliphatic carbons of Gln 274 , the Tyr(P) is tightly fixed at the active site pocket. As mentioned above, two hydrogen bonds, formed by the side chain of Asp 62 with the backbone amides of Tyr(P) and Asp ϩ1 , define the "V"-shaped binding conformation of the EEpYDD motif, allowing the N-terminal portion of the peptide to spread along the Lyp-specific insert and Tyr(P) recognition loop, whereas the C-terminal portion extends around the P-loop, Lyp-specific insert, and Q-loop. Specifically, Glu Ϫ1 engages in electrostatic interaction with Lys 61 and also makes Van der Waals contacts with Lys 61 and Asp 62 . Glu Ϫ2 interacts electrostatically with Lys 61 , Lys 136 , and Lys 138 , and its aliphatic carbons are involved in hydrophobic interactions with Tyr 60 . These structural observations are consistent with the preference of Lyp for acidic residues and its plasticity for hydrophobic residues at Ϫ1 and Ϫ2 positions. The backbone amide of Gly Ϫ3 forms a hydrogen bond with Lys 61 . Tyr Ϫ4 forms a hydrogen bond with Arg 59 and makes hydrophobic interactions with Leu 106 , Tyr 60 , and the aliphatic carbons of Arg 59 and Lys 138 , which agrees well with the observed preference for Tyr and Trp at the Ϫ4 position. At the C-terminal side of Tyr(P), electrostatic interactions occur between Asp ϩ1 and Lys 32 , Gln 274 , and Arg 266 , and Asp ϩ2 only makes long range polar interactions with Lys 32 and Lys 39 . These electrostatic interactions probably explain the overwhelming preference for acid residues at both the ϩ1 and ϩ2 positions. The following Leu ϩ3 has no obvious interaction with Lyp, which is in accord with the fact that sev- eral structurally diverse residues can be tolerated at the ϩ3 position. Tyr ϩ4 mainly makes hydrophobic interactions with Met 134 and aliphatic carbons of Gln 274 and Thr 275 , in agreement with observed preference for large hydrophobic residues (Trp and Tyr) at the ϩ4 position.
The structure of Lyp/C227S⅐SKAP-HOM Tyr(P) 75 peptide reveals that the EEpYDD motif takes up a conformation similar to that observed in the Lyp/C227S-consensus peptide 1 structure (Fig. 4). Consequently, the interactions between Lyp and substrate residues from Ϫ2 to ϩ2 are conserved for both consensus peptide 1 and the SKAP-HOM Tyr(P) 75 peptide. Different from consensus peptide 1, the SKAP-HOM Tyr(P) 75 peptide assumes a more extended conformation at both termini. The Ϫ4 position is occupied by an Asp in the SKAP-HOM Tyr(P) 75 peptide, which is displaced from the position taken by Tyr Ϫ4 in consensus peptide 1 and is surrounded by Arg 59 , Lys 136 , Lys 137 , and Lys 138 . The flexibility of a Gly at the Ϫ3 position may facilitate this displacement to optimize Lyp-peptide interactions based on the hydrophobic or acidic properties of the Ϫ4 residue. At the C terminus, no obvious interaction exist between Pro ϩ3 and Lyp. However, the special backbone torsion angles of Pro ϩ3 orient Phe ϩ4 closer to the Q-loop in comparison with Tyr ϩ4 in consensus peptide 1, probably resulting in stronger hydrophobic interactions between Phe ϩ4 and Met 134 as well as side chains of Gln 274 , Thr 275 , and Glu 277 in the Q-loop. Collectively, the observed interactions between Lyp and the SKAP-HOM Tyr(P) 75 peptide agree with the excellent activity of Lyp toward the SKAP-HOM peptide and support the hypothesis that SKAP-HOM is a cellular substrate of Lyp.
In summary, consensus sequence motifs for Lyp substrate recognition were identified using an "inverse alanine scanning" combinatorial peptide library approach. The intrinsic sequence specificity data led to the discovery of SKAP-HOM as a potential cellular target of Lyp. Further biochemical and substratetrapping experiments confirmed SKAP-HOM as a bona fide Lyp substrate. SKAP-HOM is a cytosolic adaptor protein required for proper activation of the immune system, probably by regulating the cross-talk between immunoreceptors and integrins in B cells. Further investigation of the functional consequence of Lyp-mediated SKAP-HOM dephosphorylation will shed new light on the molecular events coupling Lyp dysfunction to autoimmune diseases. The successful identification of SKAP-HOM as a Lyp substrate from peptide specificity profile suggests that the intrinsic sequence specificity of Lyp is a major determinant of the enzyme's in vivo substrate selectivity. Structural analysis of Lyp/C227S bound with consensus peptide 1 and SKAP-HOM Tyr(P) 75 peptide revealed unique molecular determinants supporting the observed Lyp substrate specificity. The defined structural features of Lyp interactions with highly efficient peptide substrates could be exploited for the design of novel inhibitors of this enzyme for both mechanistic studies of Lyp signaling and therapeutic development.