The Role of the C-terminal Domain of Protein Tyrosine Phosphatase-1B in Phosphatase Activity and Substrate Binding*

Protein tyrosine phosphatase 1B (PTP-1B) has been implicated in the regulation of the insulin receptor. Dephosphorylation of the insulin receptor results in decreased insulin signaling and thus decreased glucose uptake. PTP-1B–/– mice have increased insulin sensitivity and are resistant to weight gain when fed a high fat diet, validating PTP-1B as a potential target for the treatment of type 2 diabetes. Many groups throughout the world have been searching for selective inhibitors for PTP-1B, and most of them target inhibitors to PTP-1B-(1–298), the N-terminal catalytic domain of the enzyme. However, the C-terminal domain is quite large and could influence the activity of the enzyme. Using two constructs of PTP-1B and a phosphopeptide as substrate, steady state assays showed that the presence of the C-terminal domain decreased both the Km and the kcat 2-fold. Pre-steady state kinetic experiments showed that the presence of the C-terminal domain improved the affinity of the enzyme for a phosphopeptide 2-fold, primarily because the off-rate was slower. This suggests that the C-terminal domain of PTP-1B may contact the phosphopeptide in some manner, allowing it to remain at the active site longer. This could be useful when screening libraries of compounds for inhibitors of PTP-1B. A compound that is able to make contacts with the C-terminal domain of PTP-1B would not only have a modest improvement in affinity but may also provide for specificity over other phosphatases.

Protein tyrosine phosphatases (PTPs) 2 are essential components of signal transduction cascades. They work in conjunction with tyrosine kinases to control the phosphorylation of proteins, thereby regulating processes such as cell growth and metabolism (1)(2)(3). The family of PTPs is rather extensive, as there have been more than 100 human genes identified as protein tyrosine phosphatases via sequence analysis of the human genome (2). There is mounting interest in this family of enzymes, and different PTPs have been implicated in human diseases, including diabetes, autoimmunity, and cancer (4,5).
PTP-1B is a member of the intracellular PTP family that comprises monomeric enzymes containing a single catalytic domain and various additional domains believed to be essential for cellular localization or enzymatic regulation (21)(22)(23). The full-length PTP-1B enzyme is made up of 435 amino acids, including an N-terminal 298-amino acid catalytic phosphatase domain and a C-terminal domain (amino acids 299 -435) involved in protein-protein interactions and cellular localization (24,25). For practical reasons, most studies on PTP-1B have employed a 298-or 321-amino acid enzyme fragment, devoid of the C terminus (26 -31). However, no systematic study has verified that the enzymatic activity of this fragment is the same as the full-length enzyme. A recent study showed that an allosteric inhibitor of PTP-1B was 2-fold more potent on a more full-length construct relative to the catalytic domain alone (32).
The current study investigated whether the C-terminal domain of PTP-1B contributes to the catalytic function of the enzyme. PTP-1B-(1-411) was used to mimic the full-length enzyme, and PTP-1B-(1-298) contained only the catalytic domain. Steady state phosphatase assays and pre-steady state substrate binding assays revealed a difference between these two enzymes in their affinity for a phosphorylated peptide substrate. Specifically, the phosphorylated peptide was retained longer by PTP-1B-(1-411) compared with PTP-1B-(1-298), suggesting that the C-terminal domain of PTP-1B helps to stabilize the enzyme-substrate complex.

EXPERIMENTAL PROCEDURES
Enzymes-Human PTP-1B (amino acids 1-411 or 1-298) was cloned by PCR from a human hippocampal cDNA library (Clontech) and inserted into a pET 19-b vector (Novagen) at the NcoI restriction site. The sequence of the forward primer used to clone both PTP-1B-(1-411) and PTP-1B-(1-298) was GCC CGC CAT GGA GAT GGA AAA GGA G. The sequence of the * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1  reverse primer for PTP-1B-(1-411) was GCA AGC CAC CCA TGG CTA GAC CAG AA and for PTP-1B-(1-298) was CTC GGG TGG CCA TGG CTA GTC CTC GTG GGA. Escherichia coli strain BL21 (DE3) was transformed with this clone and stored in 20% glycerol at Ϫ80°C. For enzyme production, a stock culture was grown at 37°C in Luria Bertani medium containing ampicillin (50 g/ml). Expression of PTP-1B was initiated by induction with 1 mM isopropyl ␤-D-thiogalactopyranoside after the culture had reached an A 600 of 0.6. After 4 h, the bacterial pellet was collected by centrifugation and the cells were resuspended in lysis buffer (50 mM Tris-Cl, 100 mM NaCl, 5 mM dithiothreitol, 0.1% Triton X-100, 100 M phenylmethylsulfonyl fluoride, 100 g/ml DNase I, 2 mM MgCl 2 , 100 g/ml lysozyme, 1 tablet/50 ml complete protease inhibitors (Roche Diagnostics), pH 7.6). The cells were lysed by sonication, and the lysate was centrifuged at 100,000 ϫ g for 60 min. The supernatant was collected, the buffer was exchanged with a cation exchange buffer (50 mM MES, 75 mM NaCl, 5 mM dithiothreitol, pH 6.5), loaded onto a POROS 20SP column (PerSeptive Biosystems, Inc.), and the protein was eluted using a linear NaCl gradient (0 -500 mM). The fractions containing PTP-1B were pooled, and the buffer was exchanged with an anion exchange buffer (50 mM Tris-Cl, 50 mM NaCl, 5 mM dithiothreitol, pH 8.0). The PTP-1B protein was purified on a Source 30 Q column (Amersham Biosciences) using a linear NaCl gradient (0 -300 mM). PTP-1B was pooled and frozen at Ϫ80°C. As judged by SDS-polyacrylamide gel electrophoresis, the preparations were Ͼ95% pure. The protein concentration was determined by absorbance measurements at 280 nm in 8 M urea (extinction coefficient is 60,000 M Ϫ1 cm Ϫ1 for PTP-1B-(1-411) and 45,900 M Ϫ1 cm Ϫ1 for PTP-1B-(1-298)).
Phosphopeptide-The sequence of the phosphopeptide used in the experiments mimicked a sequence of insulin receptor substrate 1 (IRS-1) that is a substrate of PTP-1B (GNGD-pYMPMSPKS) (33). The peptide was synthesized using a RINK resin (Midwest Bio-Tech, Fishers, IN) using standard FMOC (N-(9-fluorenyl)methoxycarbonyl) solid phase peptide chemistry and purified using reversed-phase high pressure liquid chromatography. Amino acid analysis was performed (Commonwealth Biotechnologies, Inc.) for accurate determination of the peptide concentration.
pNPP Phosphatase Assay-The para-nitrophenyl phosphate (pNPP) phosphatase assays were carried out at room temperature in a microtiter plate using a Molecular Devices Spectra-MAX Plus spectrophotometer. PTP-1B-(1-411) or PTP-1B-(1-298) (50 nM) was mixed with pNPP (Sigma) (0 -60 mM) in Buffer A (50 mM Tris-Cl, pH 7.5, 50 mM NaCl, 3 mM dithiothreitol). The absorbance of the product, para-nitrophenol (pNP), was measured every 12 s for 300 s at 405 nm and was converted into molar product using an extinction coefficient of 12,852 M Ϫ1 cm Ϫ1 . The molar pNP was plotted as a function of time, and the slope of the plot provided the reaction rate. The experiment was repeated at multiple concentrations of pNPP, and the rate of each reaction was plotted as a function of the pNPP concentration. The data were fit to a hyperbola to obtain the K m and k cat values (Sigma Plot 2000; SPSS Inc.).
Phosphorylated Peptide Phosphatase Assay-The phosphorylated peptide phosphatase assays were carried out at room temperature in a cuvette (650 l) using a LS50B PerkinElmer fluorometer. PTP-1B-(1-411) or PTP-1B-(1-298) (4 nM) was mixed with the phosphorylated peptide (0 -150 M) in Buffer A. After excitation at 280 nm, the fluorescence of the phosphorylated tyrosine was measured at 305 nm every 10 s for 300 s (34). The data were plotted and analyzed as described above. To ensure that less than 10% of the substrate was converted to product, several reactions were allowed to proceed to saturation. This allowed for conversion from fluorescence units to micromolar product. Experiments using the same peptide, but phosphorylated with 33 P, showed that PTP-1B converted 100% of the substrate to product (data not shown).
Stopped Flow Peptide Binding-The pre-steady state peptide binding studies were done at 22°C using a stopped flow instrument ( To accurately measure the observed rate at which peptide bound to PTP-1B, 500 data points were collected over six halflives of the first phase of increasing fluorescence. After the first phase of changing fluorescence was saturated, there were additional phases of changing fluorescence. However, the rates for these additional phases were slower than the steady state rate of PTP-1B activity, indicating that they were not occurring during the first turnover. Therefore, only the first phase of changing fluorescence was monitored. At each concentration of peptide tested, the first phase of changing fluorescence was fit to an equation describing a single exponential to obtain the observed rate for peptide binding, F ϭ a*e (Ϫkobs*t) ϩ C, where F is the fluorescence as a function of time (t), a is the amplitude of the fluorescence change, and k obs is the observed rate constant of the fluorescence change. C is a constant.
The observed rate constant for peptide binding was then plotted as a function of the peptide concentration, and the data were fit to a straight line to obtain estimates for the intrinsic on and off rates for peptide binding to PTP-1B, k obs ϭ k on *(peptide) ϩ k off .

RESULTS
Two PTP-1B enzymes were characterized, PTP-1B-(1-298) and PTP-1B-(1-411), to investigate whether the C terminus of PTP-1B affects the enzymatic activity of the N-terminal domain. PTP-1B-(1-411) lacked the C-terminal 24 residues that localize the enzyme in the endoplasmic reticulum (Fig. 1A) and was used to mimic the full-length enzyme, because it was possible to obtain large quantities by overexpression in bacteria. PTP1B-(1-298) consisted of the catalytic domain alone. The enzymes were routinely purified to Ͼ95% purity, as indicated in Fig. 1.
Steady State Hydrolysis of pNPP-pNPP is commonly used as a substrate of tyrosine phosphatases. It changes its absorbance when it is converted to product (pNP) and therefore can be monitored relatively easily. Steady state hydrolysis of pNPP by PTP-1B-(1-411) and PTP-1B-(1-298) was monitored over time by measuring the absorbance of pNP at 405 nm. As the concentration of pNPP was increased, the rate of the reaction increased in a hyperbolic manner. The rate of the reaction was plotted versus the concentration of substrate and was fit to a hyperbola to obtain Michaelis-Menten constants. The k cat and the K m for pNPP were similar for both forms of the enzyme, ϳ3 s Ϫ1 and 2.2 mM, respectively ( Fig. 2A and Table 1).
Steady State Hydrolysis of a Phosphopeptide-The K m of a peptide containing a phosphotyrosine residue is smaller than the K m of pNPP or phosphotyrosine alone, indicating that there are additional residues besides those found directly in the active site of PTP-1B that contribute to recognition of the peptide substrate (28,35). Additional steady state phosphatase assays were done using a phosphorylated peptide (GNGDpYMPM-SPKS) as substrate, measuring the fluorescence change of the tyrosine residue upon dephosphorylation (excitation wavelength 280 nm, emission wavelength 305 nm) (34). In contrast to what was observed using pNPP as a substrate, PTP-1B-(1-411) and PTP-1B-(1-298) showed differences for both the k cat and the K m (Fig. 2B and Table 1). Although PTP-1B-(1-298) hydrolyzed the phosphorylated peptide with a k cat that was 2-fold faster when compared with PTP-1B-(1-411), the K m for PTP-1B-(1-298) was 2-fold lower.
Pre-steady State Peptide Binding-Pre-steady state stopped flow experiments were designed as shown in Fig. 3A. Briefly, PTP-1B was rapidly mixed in a stopped flow instrument with phosphorylated peptide, and a fluorescence change was meas-

Role of C-terminal Domain of PTP-1B
FEBRUARY 2, 2007 • VOLUME 282 • NUMBER 5 ured in real time during peptide binding. It was not clear whether the fluorescence signal was a result of changes within PTP-1B or the phosphorylated peptide. However, the fluorescence signal increased upon peptide binding. The experiment was repeated at multiple peptide concentrations to elucidate the intrinsic rate constants (Fig. 3B). At each peptide concentration tested, the data were fit to an equation describing a single exponential function to obtain the observed rate for peptide binding. Fig. 3B shows data collected with PTP-1B-(1-411), but the same type of fluorescence changes were observed with PTP-1B-(1-298). The observed rate constants for peptide binding to PTP-1B increased linearly with peptide concentration (Fig. 4, A and B), indicating that the fluorescence changes were due to a bimolecular interaction between PTP-1B and the peptide. The data were fit to a straight line, and the slope and y-intercept provided the k on and k off , respectively ( Table 2). The on-rate of the peptide to either form of the enzyme was the same. However, the off-rate of the peptide from PTP-1B-(1-411) was 2-fold slower than from PTP-1B-(1-298), 59 versus 120 s Ϫ1 , respectively.

DISCUSSION
Despite the extensive mechanistic and structural data collected for tyrosine phosphatases, there are many questions that remain elusive. It was shown that PTP-1B resides at the cytoplasmic face of the endoplasmic reticulum, yet there have been additional reports localizing PTP-1B in microsomes or even cytoplasm (36,37). The insulin receptor is functional at the plasma membrane; therefore, it is unclear where PTP-1B acts upon the insulin receptor (38,39). In addition, there are many biological functions attributed to PTP-1B. It has been implicated in a variety of biological pathways, including insulin signaling (13,16,17,40,41), leptin signaling (42)(43)(44), platelet activation (45,46), cytokine signaling (47), integrin-mediated signaling (46, 48 -51), p210 bcr-abl-induced transformation (52), and cell cycle regulation (53,54). In light of these multiple functions of PTP-1B, it is not clear what the consequences may be of inhibiting such an enzyme. Finally, there have been reports of various forms of PTP-1B derived from alternative splicing and post-translational proteolytic cleavage (36,55), raising the question as to the biologically relevant form. It is quite likely that each of these questions has a different answer depending upon the cell type studied, resulting in a tremendous amount of information yet to be learned about the physiological function of PTP-1B and its mechanism of regulation.
Because of the strong evidence that PTP-1B counter-regulates the insulin signal, there has been a concerted effort to find inhibitors of PTP-1B as therapeutic agents for type 2 diabetes (2, 20, 29 -31). Curiously, many groups have relied upon recombinant PTP-1B forms lacking the C-terminal domain. The C terminus of PTP-1B has several striking features. First, it has a proline-rich region (amino acid residues 301-315) shown to interact with Src homology (SH3)-containing proteins such as p130 Cas (56). It also has a hydrophobic tail (amino acid residues 400 -435) that localizes PTP-1B to the endoplasmic reticulum (25,37). In addition, two serine residues (amino acid residues 352 and 386) become phosphorylated during mitosis and in response to stress (53,57). Taken together, these data suggest that the C terminus of PTP-1B is critical for enzyme regulation as well as cellular localization. An extensive search of proteins interacting with PTP-1B indicates that there are more than 50 documented interactions, including erythropoietin receptor (58), focal adhesion kinase (59), caveolin-1 (60), and Janus kinase 2 (44). In light of this information, and contrary to the view from numerous crystal structures of the enzyme catalytic core, it is impossible to envision PTP-1B inside a cell without including a host of binding interactions and protein complexes, some of which occur through the C terminus of the enzyme. But does the C terminus contribute to the phosphatase activity of the N-terminal domain? Can the N-terminal phosphatase domain be studied independently, or does the C terminus contribute to its catalytic activity? The answer to this question is particularly important for interpreting studies using recombinant PTP-1B in relation to in vivo predictions for potential drug candidates.
In this study, the steady state activity of two PTP-1B enzymes was compared. One of the enzymes consisted of only the catalytic domain (1-298) and the other enzyme was used to mimic the full-length protein . Interestingly, the steady state experiments using a phosphopeptide showed that the K m was 2-fold higher in the absence of the C terminus, yet the k cat was 2-fold faster ( Table 1). The K m is often thought to be equivalent to K D , but it is a macroscopic constant and in this case it is unlikely to be equivalent to the K D . Because there are so many reaction intermediates that occur as PTP-1B dephosphorylates a peptide, the K m is likely lower than the K D (62). In addition, although the ratio of the k cat /K m was the same for both constructs, this apparent on-rate is not likely to be equivalent to the intrinsic on-rate. The only time the k cat /K m , an apparent second-order rate constant, can be considered equal to the true microscopic on-rate is when substrate binding is the rate-limiting step of the catalytic reaction (62). This is likely not the case for PTP-1B. There are many conformational changes that must occur after binding, including closure of the WPD loop (amino acids 179 -181), formation and hydrolysis of a covalent phosphopeptide intermediate, release of the dephosphorylated peptide, and release of the phosphate (23). Therefore, the k cat /K m in this case is likely the lower limit for the true on-rate.
Pre-steady state stopped flow studies were initiated to directly measure the intrinsic rate constants for peptide binding to the two forms of PTP-1B. Interestingly, the measured onrate was not so different from the apparent second-order rate constant (1.5 versus 0.8 M Ϫ1 s Ϫ1 for PTP-1B-(1-411) and 1.4 versus 1.0 M Ϫ1 s Ϫ1 for PTP-1B-(1-298)). However, the affinity of a phosphopeptide for the construct containing the C-terminal domain was 2-fold tighter ( Table 2). The phosphorylated peptide substrate was retained on PTP-1B-(1-411) twice as long as it was on PTP-1B-(1-298), indicating that the C terminus lends conformational stability to the enzyme-peptide complex. Either there are amino acids in the C terminus that directly contribute to peptide binding or the presence of the C terminus influences the structure of the phosphatase domain, allowing the phosphatase domain to retain the peptide longer.
Considering the results and conclusions from the pre-steady state data, it is interesting to re-examine the steady state data. The k cat /K m for the phosphorylated peptide was ϳ600-fold faster when compared with pNPP, regardless of which form of PTP-1B was used. This is primarily because of the significant difference in the K m values for the two substrates, confirming that the interactions between PTP-1B and the amino acids surrounding the tyrosine in the phosphorylated peptide contribute significantly to substrate recognition (28). Interestingly, the k cat was the same for both forms of the enzyme when pNPP was used as a substrate, but not when the phosphorylated peptide was used as a substrate (Fig. 2, Table 1). The small molecule pNPP binds directly in the active site of PTP-1B without making extensive interactions with the surface of the enzyme, whereas the phosphorylated peptide lays on the surface of PTP-1B, extending the phosphorylated tyrosine residue into the active site (26,28,30). This suggests that the C terminus does not profoundly influence the amino acids making contact with the substrate in the active site. Rather, the C terminus has a more global change in the conformation of the catalytic domain of PTP-1B, which in turn contributes to substrate recognition.
The current understanding of the structure and function of PTP-1B is derived from a number of different sources and tech- FEBRUARY 2, 2007 • VOLUME 282 • NUMBER 5 niques, including a wealth of crystal structures of the catalytic domain, protein-protein interaction data, post-translational modifications, biochemical methods, and computer-aided predictions (30,31,(63)(64)(65). The composite information from the abundant crystal structures defines the catalytic core of the enzyme (1-298) and the interaction of various substrates and inhibitors. These methods have defined binding sites and interactions in great detail, including the role of the PTP loop (216 -221), the WPD loop (181-182), residues involved in substrate recognition (Lys-36, Val-49, Lys-120), and a secondary substrate site including Tyr-20, Arg-24, His-25, Phe-52, and Arg-254. The details of substrate interactions were elucidated by an iterative affinity selection method using non-hydrolyzing phosphor-tyrosine mimics of a substrate peptide derived from the epidermal growth factor receptor (61). Using this method, Pellegrini et al. found an interesting preference within peptide substrates for an aromatic residue immediately prior to the tyrosine, a site known to be in proximity to Arg-47 and Asp-48.

Role of C-terminal Domain of PTP-1B
Recently, using a yeast-based selection system, Montalibet et al. (32) identified mutant enzymes that showed a significant resistance to difluoro-napthalenely methylphosphonic acid inhibitors. Mutant PTP-1B enzymes allowing for selection with this system included numerous point mutations at residues Glu-297, Ser-295, Trp-291, Ala-189, Ser-187, and Glu-186, a large preference for the far C-terminal region of the catalytic domain. The Ser-295 mutation was extensively characterized, and a theory of how this effect is carried to the active site that is 21 Å away was presented. The data indicate that there are extensive hydrogen bonds between two helices (helices 7 and 3), and disruption of these interactions likely results in structural changes to Phe-182, leading directly to the WPD loop.
Src homology 3 domain proteins that interact at the C-terminal proline-rich area would bind to PTP-1B close to helix 7. As such, these interactions may influence enzyme activity by the hydrogen bond network through helix 3, as depicted by Montalibet et al. (32). One can easily envision a protein-protein interaction occurring immediately after the catalytic domain playing an important role in catalysis at the active site.
The presented data suggest that the C-terminal domain of PTP-1B is involved in more than subcellular localization and protein-protein interactions. It also participates in substrate recognition of the phosphatase domain. It is possible that there are direct contacts between the C-terminal domain of PTP-1B and phosphorylated protein substrates. Alternatively, the C-terminal domain may cause a global conformational change influencing the N-terminal domain, such that there are additional interactions between the N-terminal domain and phosphorylated substrate. Either way, these data could influence the way in which inhibitors of PTP-1B are screened. The catalytic domains of protein tyrosine phosphatases are rather conserved, and finding a selective inhibitor has been difficult. Using an enzyme that contains both the N-and C-terminal domains of PTP-1B may allow for modest improvements in affinity but a significant improvement in specificity.