Calmodulin Binds to and Inhibits the Activity of the Membrane Distal Catalytic Domain of Receptor Protein-tyrosine Phosphatase α*

cDNA expression library screening revealed binding between the membrane distal catalytic domain (D2) of protein-tyrosine phosphatase α (PTPα) and calmodulin. Characterization using surface plasmon resonance showed that calmodulin bound to PTPα-D2 in a Ca2+-dependent manner but did not bind to the membrane proximal catalytic domain (D1) of PTPα, to the two tandem catalytic domains (D1D2) of PTPα, nor to the closely related D2 domain of PTPε. Calmodulin bound to PTPα-D2 with high affinity, exhibiting a K D ∼3 nm. The calmodulin-binding site was localized to amino acids 520–538 in the N-terminal region of D2. Site-directed mutagenesis showed that Lys-521 and Asn-534 were required for optimum calmodulin binding and that restoration of these amino acids to the counterpart PTPε sequence could confer calmodulin binding. The overlap of the binding site with the predicted lip of the catalytic cleft of PTPα-D2, in conjunction with the observation that calmodulin acts as a competitive inhibitor of D2-catalyzed dephosphorylation (K i ∼340 nm), suggests that binding of calmodulin physically blocks or distorts the catalytic cleft of PTPα-D2 to prevent interaction with substrate. When expressed in cells, full-length PTPα and PTPα lacking only D1, but not full-length PTPε, bound to calmodulin beads in the presence of Ca2+. Also, PTPα was found in association with calmodulin immunoprecipitated from cell lysates. Thus calmodulin does associate with PTPα in vivo but not with PTPα-D1D2 in vitro, highlighting a potential conformational difference between these forms of the tandem catalytic domains. The above findings suggest that calmodulin is a possible specific modulator of PTPα-D2 and, via D2, of PTPα.

modulator of insulin receptor signaling (7). Structurally, PTP␣ has a short heavily glycosylated extracellular region, a transmembrane region, and an intracellular region containing two conserved catalytic domains. The membrane proximal catalytic domain (D1) is responsible for most if not all of the phosphatase activity measured in vitro and toward an in vivo substrate such as Fyn (8 -11). The membrane distal catalytic domain (D2), while possessing a distinct, yet weaker in vitro activity, does not dephosphorylate any cellular substrates identified to date.
A conserved D2 domain is present in many receptor-like PTPs (RPTPs), yet all have undetectable or extremely low activity. This indicates a non-redundant function of the tandem domains, with D2 perhaps playing a regulatory role. The RPTPs CD45, LAR, PTP, and PTP␣ all display altered D1 activity or in vitro substrate specificity in the absence of D2 (9,(12)(13)(14)(15)(16)(17). CD45-D2 is specifically required, independent of any catalytic activity, for the in vivo recognition by CD45-D1 of its substrate, the T-cell receptor -chain (18). The in vitro interaction of PTP␦-D2 with D1 of PTP inhibits PTP-D1 activity, suggesting that D2 may regulate the formation and activity of receptor heterodimers (19). Alternatively, some D2 domains may have a catalytic function but possess a highly restricted or unusual substrate specificity. RPTPs that this could apply to include, for example, PTP␣ and LAR, the D2 domains of which have minimal substitutions in the 42 residues identified as conserved among all D1 domains and thus likely to be essential for catalysis (20). Both D2 domains assume the folding of their respective D1 domains, as predicted by PTP␣-D2 modeling (11) and shown for LAR-D2 within the LAR crystal (21). The D2 domain of PTP␣ clearly has intrinsic activity, which with some in vitro substrates can be as much as 10% of the D1 activity (8 -10). Substitution of two amino acids in PTP-D2 or LAR-D2 with Tyr and Asp that are conserved in these positions in all D1 domains confers a D1-equivalent catalytic competence, but the mutated D2s retain a distinct non-D1-like substrate specificity (11,21,22). Crystal structure determination of PTP1B bound to substrate has shown that the aromatic ring of the conserved Tyr is involved in interactions with substrate phosphotyrosine and that the conserved general acid Asp is in close proximity to the bound phosphotyrosine (23). The equivalent residue in PTP␣-D2 and LAR-D2 is hydrophobic but not aromatic, whereas Asp is replaced by the acidic residue Glu which is more distant from the substrate phosphate. Thus, essential catalytic amino acids are intact, and the D2 domain is folded correctly, but positioning and protonation of the conventional phosphotyrosine in the active site pocket is not optimal.
Protein-protein interactions are key determinants of cell regulation and signal transduction events. Several such interactions have been described for PTP␣ and proposed to modulate its function. The extracellular region of PTP␣ associates in cis with the glycosylphosphatidylinositol-linked cell surface molecule contactin, possibly forming a receptor complex in which PTP␣ transduces a signal to activate the tyrosine kinase Fyn (24). A proposed dimerization of D1 involves the interaction of the N-terminal "wedge" region of one D1 domain with the active site of the partnering D1 domain, with consequent inhibition of catalytic activity (25). This wedge region can also interact in trans with the D2 domains of PTP␣ and other RPTPs, perhaps affecting PTP␣ dimer formation through the wedge-to-D1 interaction described above (26). A phosphotyrosine located in the C-terminal tail region binds to Grb2-SH2 and Src-SH2. The former binding facilitates the subsequent binding of the Grb2-SH3 domain to a region in D1, which is proposed to inhibit D1 catalysis (27)(28)(29)(30). Binding of the SH2 domain of Src to the same phosphotyrosine in the PTP␣ tail that can bind to Grb2 is hypothesized to facilitate PTP␣-catalyzed dephosphorylation of the regulatory Tyr-527 residue in Src (31).
Other than PTP␣ homodimer formation, no protein interactions with PTP␣-D2 have been reported. To search for such interacting proteins, the identification of which might illuminate D2 function, we have used radiolabeled PTP␣-D2 to screen a cDNA expression library. The Ca 2ϩ -binding protein calmodulin was discovered to associate with PTP␣-D2, and the binding site on D2 and the effect of calmodulin binding to D2 have been characterized. In addition, an in vivo association of calmodulin with the wild-type PTP␣ protein was observed.
Purification and Radiolabeling of GST-PTP Fusion Proteins-Bacterially expressed GST-PTP fusion proteins were bound to glutathione-Sepharose beads (Amersham Pharmacia Biotech) and purified either by elution with reduced glutathione (Sigma) or by cleavage with thrombin (Sigma). Purified PTPs were quantitated by Bradford or BCA analysis (Pierce) and visualized by SDS-PAGE. For radioactive labeling, pGEX-2TK-PTP␣-D2 was expressed in Escherichia coli DH5␣FЈ and bound to glutathione-Sepharose beads (Amersham Pharmacia Biotech). Bovine heart muscle kinase (Sigma) and [␥-32 P]ATP (3000 mCi/mmol, NEN Life Science Products) were added and incubated with the beads at 4°C for 30 min as described (34). To remove unreacted [␥-32 P]ATP, the beads were washed extensively with phosphate-buffered saline until no signal could be detected in the wash. The 32 P-labeled PTP␣-D2 polypeptide was then isolated following thrombin cleavage and the specific activity determined. A small portion of the sample was applied to SDS-PAGE and visualized by autoradiography as a single band of the expected size.
Far Western Screen of cDNA Expression Library-2 ϫ 10 6 clones from a human cerebellum cDNA expression library constructed in ZAP (a gift from T. Leung) were plated at 42°C for 4 h and then overlaid with 10 mM isopropyl-␤-D-thiogalactopyranoside-saturated nitrocellulose membranes (NEN Life Science Products) at 37°C for another 8 h. Cells grown on the membranes were processed through a series of steps of denaturation and renaturation as described (34). 32 P-Labeled PTP␣-D2 (10 6 cpm/ml) was added and incubated with the membranes in Hyb buffer (20 mM Hepes-KOH, pH 7.7, 75 mM KCl, 0.1 mM EDTA, 2.5 mM MgCl 2 , 1 mM dithiothreitol, 0.05% Nonidet P-40, 50 -100 M CaCl 2 , 1% non-fat dry milk) for 12 h. All procedures after the cell lysis step were performed at 4°C. The membranes were washed three times in Hyb buffer, dried briefly, and exposed for autoradiography. Sixteen clones showing a positive signal were picked up and plated for the secondary screen. Fifteen of these had over 20% positive clones in 100-mm Petri dishes and were kept as putative PTP␣-D2-binding molecules for further analysis.
In Vitro Binding Measurements Using a Surface Plasmon Resonance Biosensor-A BIAcore 2000 instrument (Amersham Pharmacia Biotech) was used. Bovine brain calmodulin (a gift from M. Z. Zhang) was covalently linked to a sensor chip CM5 (BIAcore), and unreactive groups on the chip were inactivated by ethanolamine according to the manufacturer's instructions. Recombinant protein in running buffer (20 mM Hepes, pH 7.7, 150 mM NaCl, 0.005% Tween 20, 1 mM CaCl 2 , or EGTA) was injected over the surface using the KINJECT command at a flow rate of 15 l/min and a total volume of 240 l. A 20-min dissociation period was provided, and the surface was regenerated by injection of 10 l of 0.05% SDS. Cycles of injection and regeneration followed by running buffer flow were computer programmed for the binding comparison of different samples. BIA evaluation 3.0 software (BIAcore) was used to analyze data. The K D value was calculated according to a 1:1 interaction model by direct fitting of ligand binding at multiple concentrations around the K D .
Cell Culture and Transient Transfections-COS-1 cells were obtained from American Type Culture Collection (Manassas, VA). NIH3T3 cells were a gift from P. Lobie. Cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and penicillin/streptomycin in an atmosphere of 5% CO 2 at 37°C. Transient transfections were performed at 50 -70% cell confluency (100-mm dishes) using 5 g of plasmid DNA and liposome-mediated transfection with LipofectAMINE reagent (Life Technologies, Inc.) as described by the manufacturer. Cells were harvested 24 -36 h after transfection. Transfection with the empty expression plasmid pJX41neo was used as a control.
Immunoprecipitation and Western Blots-Monoclonal antibodies toward VSVG (Sigma), FLAG (Sigma), and calmodulin (Upstate Biotechnology) were used for immunoprecipitation and Western blotting. Transfected NIH3T3 cells were lysed in buffer A (20 mM Tris, pH 7.5, 150 mM NaCl, 1% Brij-96, 20 g/ml aprotinin, 2 mM phenylmethylsulfonyl fluoride), and the lysates were clarified by centrifugation. For immunoprecipitation, cell lysates were incubated with antibodies at 4°C for 12 h. Protein G plus/protein A-agarose suspension (Calbiochem) was then added for another 3 h at 4°C. The immunoprecipitates were washed five times in buffer B (20 mM Tris, pH 7.5, 150 mM NaCl, 0.2 mM CaCl 2 ), resolved by SDS-PAGE, transferred to a polyvinylidene difluoride membrane, and immunoblotted using the ECL system (Amersham Pharmacia Biotech).
In Vitro Binding Assay Using Calmodulin-Sepharose Beads-Lysates of COS-1 cells expressing VSVG-PTP␣, VSVG-PTP␣-D2⌬D1, or FLAG-PTP⑀ were incubated with 50 l of calmodulin-Sepharose beads (Amersham Pharmacia Biotech) in the presence of 1 mM of CaCl 2 or EGTA at 4°C for 3 h. The beads were washed three times in buffer C (20 mM Tris, pH 7.5, 150 mM NaCl) containing 1 mM CaCl 2 or EGTA and eluted with 50 l of buffer D (20 mM Tris, pH 7.5, 150 mM NaCl, 5 mM EGTA). The eluates were resolved by SDS-PAGE and immunoblotted with anti-VSVG or anti-FLAG antibody.
Phosphatase Assays-Phosphatase assays toward pNPP were per-formed at 30°C in 450-l reactions containing 50 mM sodium acetate, pH 5.5, 0.5 mg/ml bovine serum albumin, 5 mM dithiothreitol, and with 1 mM CaCl 2 or EGTA added as indicated in the figure legends. Some reactions contained bovine brain or recombinant calmodulin, and similar results were obtained with either form. The N2 peptide used in the competition experiments was synthesized commercially with an automated peptide synthesizer (BTC, NUS, Singapore) to Ͼ95% purity and added at the indicated concentrations. The apparent K m and V max of PTP␣-D2 in the absence or presence of calmodulin was manually extrapolated from Lineweaver-Burk inverse plots of PTP␣-D2 activity toward pNPP concentrations ranging from 1.25 to 10 mM. For K i determination, the linear regression of the K m(app) /V max values (the slopes in the inverse plot) were plotted against inhibitor concentration. The intersection of the linear regression with the x axis gives the negative K i value. All reactions were carried out at 30°C and terminated during the linear portion of the reaction. Released p-nitrophenol was quantitated as described previously (8).

Identification of Calmodulin as a PTP␣-D2-binding
Protein-Recombinant 32 P-labeled PTP␣-D2 was used to screen a human cerebellum cDNA expression library. Fifteen positive clones were identified after the secondary screen, and two of these encoded calmodulin.
In Vitro Binding of Calmodulin to PTP␣-D2-The recombinant and mammalian expressed proteins used throughout this study are defined in Table I. The interaction of PTP␣-D2 and calmodulin was confirmed using surface plasmon resonance. PTP␣-D2 bound in a Ca 2ϩ -dependent manner to bovine brain calmodulin coated on the sensor chip, with no binding observed in the presence of EGTA (Fig. 1A). In comparison, a greatly reduced but Ca 2ϩ -dependent binding occurred between calmodulin and PTP␣-D1 (Fig. 1B) or PTP␣-D1D2 (Fig. 1C). Increasing concentrations of PTP␣-D2 resulted in increased binding (RU) to the calmodulin-coated sensor chip. The kinetics of interaction were fitted to a single interaction site model, and yielded a dissociation constant (K D ) of 2.86 ϫ 10 Ϫ9 M Ϯ 0.31, suggestive of a specific and high affinity interaction.
The Calmodulin-binding Site Is in the N-terminal Region of PTP␣-D2-To define the calmodulin-binding site within PTP␣-D2, two constructs were generated that expressed an N-terminal portion of D2 (D2-N, aa 486 -542) or the remaining Cterminal portion of D2 with the tail region (D2-C, aa 543-774) (Fig. 2A). These regions of PTP␣ were expressed in bacteria as GST fusion proteins, affinity purified on glutathione-Sepharose beads, and either eluted from the beads (GST-D2-N) or cleaved from the GST with thrombin (D2-C) to give soluble purified polypeptides. Their interaction with a calmodulin-coated sensor chip was analyzed by surface plasmon resonance. The D2-C polypeptide showed little or no interaction with calmodulin, as was found with a control preparation of purified GST protein (Fig. 2, B and C). However, the GST-D2-N polypeptide bound to the immobilized calmodulin with very similar kinetics to the complete D2 polypeptide (Fig. 2B). The irregularities in the D2-N binding curve may be due to the presence of degraded forms of the fusion protein that were detected by SDS-PAGE. The purified GST-D2-N fusion protein was also difficult to produce in any significant quantity. Nevertheless, these results indicate that the calmodulin-binding site resides within the N-terminal amino acids 486 -542 of PTP␣-D2.
Analysis of the amino acid sequence in D2-N for candidate calmodulin-binding sites identified two overlapping subregions, N1 (aa 502-521) and N2 (aa 520 -538) ( Fig. 2A), with   positive charge distributions and the potential abilities to form an amphipathic helix, a favored structure for calmodulin binding (35). Engineered forms of the D2-C polypeptide with either N1 or N2 at the amino terminus were produced and tested for calmodulin binding. The N1 region conferred weak calmodulin binding upon the polypeptide. The presence of the N2 region conferred much stronger calmodulin binding ability to the polypeptide, and this was equivalent to the interaction with calmodulin observed with the complete D2 protein (Fig. 2D). Thus the 19-amino acid sequence of N2 contains the calmodulin-binding site of PTP␣-D2.
Identification of Amino Acid Residues in PTP␣-D2 That Are Essential for Calmodulin Binding-PTP␣ shares considerable amino acid sequence homology with its closest relative, PTP⑀ (32). The D2 domains of these PTPs have 72% amino acid identity, yet PTP⑀-D2 exhibited a relatively weak interaction with calmodulin (Fig. 3, A and B). Following the identification of the N2 subregion of PTP␣-D2 as the calmodulin-binding site, the corresponding N2 sequence in PTP⑀ (aa 422-440) was examined. The 19-amino acid N2 regions were identical with the exceptions of two conservative substitutions (Asp-520 and Leu-537 in PTP␣ to Glu-422 and Ile-439, respectively, in PTP⑀) and two non-conservative replacements (Lys-521 and Asn-534 in PTP␣ to Asn-423 and Ala-436, respectively, in PTP⑀). This suggested that these amino acid differences, in particular the non-conserved residues, could be the basis for the differential interactions of PTP␣-D2 and PTP⑀-D2 with calmodulin. Indeed, mutations of the positively charged Lys-521 of PTP␣ to either a negatively charged glutamic acid (K521E) or to the neutral Asn (K521N) present in this position in PTP⑀ resulted in a greatly reduced binding to calmodulin (Fig. 3C). Mutation of neutral Asn-534 of PTP␣ to a positively charged lysine residue (N534K) or to the hydrophobic alanine residue (N534A) found in this position in PTP⑀ also significantly reduced the strength of the interaction with calmodulin, although to a lesser extent than did the mutation of Lys-521 (Fig. 3C). This indicates that both Lys-521 and Asn-534 of PTP␣ play a role in the interaction of PTP␣-D2 with calmodulin and that substitution of either is sufficient to disrupt binding. Consistent with these results, a single point mutation in position 423 (N423K) of the PTP⑀ N2 sequence (which restored the lysine residue found in this position in PTP␣ N2) was unable to restore calmodulin binding to a (PTP⑀-N2)-(PTP␣-D2C) fusion protein. However, two simultaneous mutations of the non-conserved residues 423 and 436 of PTP⑀ (N423K/A436N) that restored the amino acids correspondingly found in PTP␣ increased calmodulin binding by about 2-fold (Fig. 3D). Since the double mutation in PTP⑀ N2 did not restore binding to a level comparable to that observed with PTP␣ N2, this indicates that both the conservative and the non-conservative amino acid differences between the PTP␣ and PTP⑀ N2 subregions contribute to the distinct calmodulin binding properties of PTP␣-D2 and PTP⑀-D2.
Calmodulin Inhibits PTP␣-D2 Activity-The phosphatase activity of PTP␣-D2 was assayed toward two defined D2 substrates, pNPP, and the RR-Src peptide, with and without calmodulin. Although calmodulin alone or in the presence of added EGTA had no effect on PTP␣-D2 activity, calmodulin in the presence of added Ca 2ϩ inhibited D2 activity toward pNPP (Fig. 4A) and RR-Src (data not shown). However, neither PTP␣-D1 nor PTP␣-D1D2 activity was affected by calmodulin in the presence of Ca 2ϩ or EGTA (Fig. 4, B and C), consistent with the inability of these recombinant forms of PTP␣ to bind to calmodulin. In the presence of Ca 2ϩ , calmodulin inhibited PTP␣-D2 in a concentration-dependent manner (data not shown). The inhibition exerted by calmodulin on PTP-D2 activity could be progressively blocked by the addition to the reaction of increasing concentrations of a synthetic peptide corresponding to the N2 sequence (Fig. 4D). Further investigation of the action of calmodulin revealed that it acts as a competitive inhibitor of PTP␣-D2, with an apparent K i of 337 nM (Fig. 5). Thus calmodulin binding reduces the affinity of PTP␣-D2 for substrate.
In Vivo Association of PTP␣ and Calmodulin-VSVG-tagged full-length PTP␣ and VSVG-tagged PTP␣ lacking only D1 (PTP␣-D2⌬D1) were transiently expressed in cells, and the cell lysates were incubated with calmodulin-Sepharose beads in the presence of added EGTA or Ca 2ϩ . After washing, proteins bound to the beads were eluted with EGTA. PTP␣ and PTP␣-D2⌬D1 were both eluted from the beads that had been incubated with the lysates in the presence of Ca 2ϩ but were not detected in eluates from beads incubated with lysates and EGTA (Fig. 6A). Thus cellular PTP␣ possessing the extracellular, transmembrane, and juxtamembrane regions can undergo Ca 2ϩ -dependent binding to calmodulin. The two bands that are immunoreactive with the anti-VSVG antibody (Fig. 6, A and B) represent differentially glycosylated forms of PTP␣, and it is apparent that there is preferential binding to calmodulin of the lesser glycosylated, faster migrating form. In a similar experiment using lysates of cells expressing full-length PTP⑀, no binding of PTP⑀ to the calmodulin-Sepharose beads was observed (Fig. 6B), in accord with the lack of binding of PTP⑀-D2 to calmodulin. To determine whether PTP␣ could bind to calmodulin in the cell, anti-calmodulin immunoprecipitates of cell lysates were probed for the presence of PTP␣. PTP␣ was found in association with calmodulin (Fig. 6C). DISCUSSION We have shown that calmodulin binds to and modulates the phosphatase activity of PTP␣-D2. Calmodulin binds in a Ca 2ϩdependent manner and with high affinity (K d ϳ3 nM) to a 19-amino acid sequence located at the N-terminal region of D2 and comprising residues 520 -538. This sequence contains 5 basic residues with a cluster near the C-terminal end, 1 acidic residue, and several hydrophobic residues. This is in general agreement with the nature of other calmodulin-binding peptides, which, although having no consensus sequence, tend to have a net positive charge with several basic residues interspersed with hydrophobic amino acids which often, but not always, appear at defined intervals (36). The homologous region of LAR-D2 (21) and the calmodulin-binding site of a model of PTP␣-D2 (based on the LAR-D2 structure template, data not shown) adopts a bent shape due to the presence of a proline residue in the middle (Pro-528 in PTP␣-D2 and Pro-1636 in LAR-D2). The N-terminal side of the bend contains a hydrogenbonded turn, whereas the C-terminal side contains a short 3 10 helix. This is unexpected as many known calmodulin-binding peptides consist of a continuous amphipathic helical structure (35)(36)(37). Generally, a 3 10 helix is known to be slightly less stable than an ␣-helix because of steric hindrance and the awkward geometry of its hydrogen bonds. The presence of a predicted turn and a 3 10 helix would provide flexibility, and upon binding to calmodulin, it may be induced to fold into a continuous helix which presents residues in a manner for optimum and continuous binding. This is indeed the case for high affinity calmodulin binding to melittin, a bee venom peptide with an unusual bent conformation and 5% helical content that exhibits an increase in helical content to 72% upon complex formation (38,39).
Calmodulin can interact with cellular PTP␣ and with recombinant PTP␣-D2 but not with the PTP⑀ equivalents. Mutagenesis of the calmodulin-binding sequence in PTP␣-D2 and of the counterpart non-binding sequence in PTP⑀-D2 showed that two amino acids near either end of the PTP␣-D2 sequence, Lys-520 and Asn-534, are essential for optimal calmodulin binding and are not conserved in PTP⑀. All other RPTP D2 domains (except PTP-OST) possess the same Asn residue within a conserved KNR sequence as PTP␣-D2, but none have a Lys residue in the equivalent position to that in PTP␣-D2. Although PTP␦-D2, PTP-D2, and LAR-D2 all have a positively charged arginine residue in this latter position, it may not be sufficient to confer calmodulin binding in the absence of other PTP␣-D2 features within the N2 sequence, for example a second positively charged residue two positions C-terminal to the Lys/Arg. It will be of interest to determine whether calmodulin can interact with these D2 domains. Nothing resembling the N2 calmodulin-binding sequence is found in a comparable position in any of the RPTP D1 domains, including PTP␣. This suggests that the interaction with calmodulin may be a specific property of PTP-D2, and perhaps a unique means of PTP␣-D2, and through this, PTP␣ regulation.
An intriguing difference is apparent in the abilities of recombinant PTP␣-D1D2 and cellular PTP␣ to bind calmodulin. While recombinant D2 can bind calmodulin, the additional presence of D1 in the recombinant protein appears to preclude binding. Nevertheless, cellular PTP␣ containing D1 can associate with calmodulin. This suggests that the two catalytic domains may not assume the same relative conformation in the recombinant protein as they do in cellular PTP␣, with the calmodulin-binding site being inaccessible in the former conformation. Even so, the D2 catalytic cleft is accessible in both forms, at least to a small molecule substrate such as pNPP, as D2 can contribute to the same extent to the total activity of recombinant PTP␣-D1D2 or immunoprecipitated cellular PTP␣ in in vitro assays (9). Cellular conditions that could modify conformation and thus calmodulin-binding site accessibility include phosphorylation (cellular PTP␣ is a phosphoprotein, although not all phosphorylation sites are defined), membrane insertion of the mature protein, and intra-or intermolecular interactions, including, for example, those involving the juxtamembrane region of PTP␣ that is lacking in the recombinant protein.
Calmodulin-binding activates many enzymes. However the phosphatase activity of PTP␣-D2 is inhibited upon interaction with calmodulin. Kinetic studies show that calmodulin acts as a competitive inhibitor, likely reflecting the close proximity of the calmodulin-binding site in PTP␣-D2 to the substrate-binding site. The C-terminal end of the N2 calmodulin-binding sequence comprises the lip of the D2 catalytic cleft as determined from PTP crystal structures and predicted from PTP␣-D2 modeling (11,21,23,25). Thus calmodulin binding could prevent substrate binding, either by direct occlusion of the catalytic cleft, and/or possibly through a distortion of the catalytic cleft resulting from calmodulin pulling the binding site region out from the PTP structure and concomitant rearrangement of the binding site secondary structure. If PTP␣-D2 indeed has a catalytic function, this would be an unusual example of calmodulin directly blocking enzymatic activity. If PTP␣-D2 acts in a non-enzymatic capacity through binding phosphotyrosyl proteins (20), this function could also be abrogated by association with calmodulin.