J Biol Inorg Chem DOI 10.1007/s00775-017-1500-1

ORIGINAL PAPER

Kinetic characterization of the inhibition of protein tyrosine phosphatase‑1B by Vanadyl ­(VO2+) chelates Jason Hon1 · Michelle S. Hwang1 · Meara A. Charnetzki1 · Issra J. Rashed1 · Patrick B. Brady1 · Sarah Quillin1 · Marvin W. Makinen1 

Received: 21 July 2017 / Accepted: 12 October 2017 © SBIC 2017

Abstract  Protein tyrosine phosphatases (PTPases) are a prominent focus of drug design studies because of their roles in homeostasis and disorders of metabolism. These studies have met with little success because (1) virtually all inhibitors hitherto exhibit only competitive behavior and (2) a consensus sequence H/V-C-X5-R-S/T characterizes the active sites of PTPases, leading to low specificity of active site directed inhibitors. With protein tyrosine phosphatase1B (PTP1B) identifed as the target enzyme of the vanadyl ­(VO2+) chelate bis(acetylacetonato)oxidovanadium(IV) [VO(acac)2] in 3T3-L1 adipocytes [Ou et al. J Biol Inorg Chem 10: 874–886, 2005], we compared the inhibition of PTP1B by VO(acac)2 with other V ­ O2+-chelates, namely, bis(2-ethyl-maltolato)oxidovanadium(IV) [VO(Etmalto)2] and bis(3-hydroxy-2-methyl-4(1H)pyridinonato) oxidovanadium(IV) [VO(mpp)2] under steady-state conditions, using the soluble portion of the recombinant human enzyme (residues 1–321). Our results differed from those of previous investigations because we compared inhibition in the presence of the nonspecific substrate p-nitrophenylphosphate and the phosphotyrosine-containing undecapeptide DADEpYLIPQQG mimicking residues 988–998 of the epidermal growth factor receptor, a relevant, natural substrate. While VO(Et-malto)2 acts only as a noncompetitive inhibitor in the presence of either subtrate, VO(acac)2 Electronic supplementary material  The online version of this article (doi:10.1007/s00775-017-1500-1) contains supplementary material, which is available to authorized users. * Marvin W. Makinen [email protected] 1



Department of Biochemistry and Molecular Biology Center for Integrative Science, The University of Chicago, 929 East 57th Street, Chicago, IL 60637, USA

exhibits classical uncompetitive inhibition in the presence of DADEpYLIPQQG but only apparent competitive inhibition with p-nitrophenylphosphate as substrate. Because uncompetitive inhibitors are more potent pharmacologically than competitive inhibitors, structural characterization of the site of uncompetitive binding of VO(acac)2 may provide a new direction for design of inhibitors for therapeutic purposes. Our results suggest also that the true behavior of other inhibitors may have been masked when assayed with only p-nitrophenylphosphate as substrate. Keywords  Protein tyrosine phosphatase-1B · Steadystate kinetics · Uncompetitive inhibition · Vanadyl ­(VO2+) chelates · Bis(acetylacetonato)oxidovanadium(IV) · VO(acac)2 Abbreviations Ac Acetate CM Carboxymethyl DMSO Dimethyl sulfoxide EDTA  N,N,N′,N′-ethylene diamine-tetraacetic acid EGFR Epidermal growth factor receptor EGFR988−998 The phosphotyrosine containing undecapeptide DADEpYLIPQQG simulating residues 988–998 of phosphorylated EGFR ENDOR Electron nuclear double resonance EPR Electron paramagnetic resonance IPTG Isopropyl-β-D-1-thiogalactopyranoside IR Insulin receptor IRS-1 Insulin receptor substrate-1 MES 2-(N-morpholino)-ethanesulfonic acid pNPP  para-nitrophenylphosphate PTPases Protein tyrosine phosphatases

13

Vol.:(0123456789)



J Biol Inorg Chem

PTP1B Protein tyrosine phosphatase-1B pY Phosphotyrosine UV Ultraviolet VO(acac)2  Bis(acetylacetonato)oxidovanadium(IV) VO(Et-malto)2  Bis(2-ethyl-maltolato) oxidovanadium(IV) VO(Me-malto)2  Bis(2-methyl-maltolato) oxidovanadium(IV) VO(mpp)2  Bis(3-hydroxy-2-methyl-4(1H)pyridinonato)oxidovanadium(IV) TCEP  Tris(2-carboxyethyl)phosphine Tris  Tris(hydroxymethyl)aminomethane

Introduction The protein tyrosine phosphatases ­(PTPases§) constitute a superfamily of enzymes that together with protein tyrosine kinases regulate a vast number of cellular processes through post-translational phosphorylation and dephosphorylation of other enzymes and proteins. Because of their critical roles not only in metabolic homeostasis but also in disorders of metabolism, they have commanded the attention of a large number of studies directed towards drug design for therapeutic intervention. In contrast to protein tyrosine kinases, however, these efforts have met with little success because (1) identification of physiological substrates of the PTPases has been limited; (2) the active site is characterized by a consensus sequence H/V-C-X5-R-S/T, in which X is any amino acid and C is the catalytically required, nucleophilic cysteine residue, a condition predisposed to low specificity of active site directed inhibitors; (3) PTPases are themselves subject not only to both tyrosine and serine phosphorylation and dephosphorylation but also sumoylation, post-translationally modulating their activity in the cell; and (4) the catalytic action of PTPases, while straightforward with respect to chemistry, has been often difficult to link to its biological consequences. Protein tyrosine phosphatase-1B (PTP1B) is the first member of the PTPases to be purified to homogeneity [1, 2]. Its role in dephosphorylation of receptor tyrosine kinases, endocytosis, the stress response of the endoplasmic reticulum, energy balance, cell–cell communication, and vesicle

O O

O V

a

O O

O

O O O V O O

b

trafficking and how these processes are affected in metabolic disease [3] underscore the breadth of metabolic pathways in which PTP1B takes part. The chemical reaction catalyzed by PTP1B and other PTPases, the hydrolysis of a peptide containing a phosphorylated tyrosine residue, results in the free phosphate anion and the dephosphorylated peptide. Because identification of physiologically relevant substrates of the PTPases has been limited, one of the most widely used substrates for assays of catalytic phosphatase activity is the minimal substrate para-nitrophenylphosphate (pNPP) because of its low cost, chromophoric properties, and commercial availability. While the pNPP assay is robust and economically amenable for screening applications, its use in the search for effective therapeutic inhibitors of PTP1B is limited because of the breadth of metabolic roles of PTP1B. It is well established that active site residues, as well as residues surrounding the active site of PTP1B experience a much larger number of hydrogen-bonding and electrostatic interactions with a phosphotyrosine-containing peptide substrate than with the phosphotyrosine moiety alone [4], implying that structural changes induced in the protein are not identical for both types of substrates. We note that of the large number of studies carried out to identify compounds potentially useful therapeutically and based on pNPP assays [5–9], only one inhibitor has been identified [9] that is not a competitive inhibitor. We believe, therefore, that to develop inhibitors of therapeutic efficacy requires use of phosphotyrosine-containing peptides that mimic the molecular determinants of recognition of physiologically relevant substrates. In previous studies from this laboratory to identify the target enzyme of organic vanadyl (­ VO2+) chelates as insulin mimetics, we demonstrated on the basis of phosphotyrosine immunoblots of cultured 3T3-L1 adipocytes that the target enzyme of bis(acetylacetonato)oxidovanadium(IV) [VO(acac)2] regulated the phosphotyrosine level of the insulin receptor (IR) and of the insulin receptor substrate-1 (IRS-1) protein and that the ­VO2+-chelate did not influence other downstream components of the insulin signaling cascade [10–12]. Because PTP1B is well recognized as a major negative regulator of insulin signaling [13], our observations led us to focus on possible inhibition of this enzyme by VO(acac)2 and other related organic V ­ O2+-chelates illustrated in Fig. 1. In preliminary investigations [14], our

O

HN

O O O V O O

NH

c

Fig. 1  Comparison of chemical bonding structures of a bis(acetylacetonato)oxidovanadium(IV), VO(acac)2; b bis(2-ethyl-maltolato) oxidovanadium(IV), VO(Et-malto)2; and c bis(3-hydroxy-2-methyl-4(1H)pyridinonato)oxidovanadium(IV), VO(mpp)2

13

J Biol Inorg Chem

results suggested that the mode of inhibition of VO(acac)2 in hydrolysis of pNPP catalyzed by PTP1B differed from that in the presence of the phosphotyrosine-containing undecapeptide DADEpYLIPQQG mimicking residues 988–998 of the epidermal growth factor receptor (EGFR) [15]. Reasoning that structural details of the interaction of PTP1B with the phosphotyrosine-containing peptide substrate could directly affect the mode of inhibition, we compared the inhibition of dephosphorylation of the synthetic substrate pNPP and of DADEpYLIPQQG by the three ­VO2+-chelates illustrated in Fig. 1. The results of our studies, confirming our earlier observations [14], demonstrate that VO(acac)2 acts as an uncompetitive inhibitor of PTP1B with DADEpYLIPQQG as the substrate. Not only is VO(acac)2 the only uncompetitive inhibitor of any PTPase identified hitherto, but the results also show that this V ­ O2+-chelate exhibits only apparent competitive inhibition of pNPP hydrolysis when catalyzed by PTP1B, differing from that observed in the hydrolysis of the phosphotyrosine-containing undecapeptide. With respect to development of inhibitors of PTP1B and possibly other PTPases for therapeutic intervention of disease, the results are important because uncompetitive inhibitors are widely recognized to be more potent pharmacologically in open-cell systems than their competitive counterparts [16, 17].

Experimental procedures Materials Crystalline VO(acac)2, pNPP, TCEP, IPTG, EDTA, Tris, MES, spectrophotometric grade DMSO, and CM Sephadex G-50 were purchased from Sigma-Aldrich (St. Louis, MO 63103) and used without further purification. The phosphotyrosine containing undecapeptide DADEpYLIPQQG simulating residues 988–998 of the EGFR [15] was obtained from C. S. Bio, Co. (Menlo Park, CA 94025). VO(Et-malto)2 [18] and VO(mpp)2 [19, 20] were synthesized according to published methods. Analytical data for elemental composition of each chelate showed good agreement with theoretical expectation. Sodium chloride, sodium orthovanadate, and sodium acetate were of analytical grade and were obtained from Fisher Scientific (Pittsburgh, PA 15275). All other reagents were of the highest quality available; deionized, distilled water was used throughout. Expression and purification of PTP1B Competent E. coli BL21 (DE3) cells were obtained from Invitrogen (Carlsbad, CA 92008). The plasmid DNA corresponding to the soluble portion of human PTP1B (residues 1–321) was kindly provided by Professor ZY Zhang

(Department of Biochemistry and Molecular Biology, Indiana University School of Medicine, Indianapolis, Indiana 462020) and cloned into E. coli BL21 (DE3) according to the method described by Zhang and coworkers [21, 22]. Transformed cells were stored at −80 °C until further use. Single cell colonies were selected from agar plates containing 100 μg/mL ampicillin for overnight precultures in Luria–Bertani broth. Inoculation of precultures into Luria–Bertani medium initiated E. coli growth at 34 °C under constant aeration. Protein expression was initiated by addition of 0.005 M IPTG to bacterial cultures when growth had reached an optical density of 0.6–0.8 at 600 nm. Cells were harvested at 4 °C by centrifugation at 4500×g for 30 min; the cell pellets were resuspended in a minimal volume of 0.01 M NaCl buffered to pH 6.1 with 0.01 M MES containing 0.002 M TCEP, and cell lysis was achieved by application of a French press. The cell lysate was centrifuged at 15,000×g for 30 min at 4 °C, and the supernatant was added to 4–6 g CM Sephadex G-50 that had been pre-equilibrated with the same buffer. The Sephadex–lysate mixture was poured into a 2 × 50 cm column and washed with ~ 250 mL of the same low ionic strength buffer. PTP1B was eluted with a NaCl gradient of 0.01–0.4 M buffered to pH 6.1 with 0.01 M MES. Because sulfonate-based buffers are competitive inhibitors of the enzyme, we have also used sodium 3,3′-dimethylglutarate at pH 6.1 in place of MES to buffer the NaCl gradient. This buffer system has been used by Zhang and coworkers in kinetic studies [23]. While the yield of the recombinant enzyme was generally higher than with use of MES, a sizeable fraction frequently remained irreversibly adsorbed to CM-Sephadex during chromatography. As a competitive inhibitor, MES likely helps to stabilize the tertiary structure of the enzyme during purification. Protein concentration was determined on the basis of 1% the absorptivity coefficient A280 nm of 1.24 [24, 25]. Specific activity of chromatographic fractions was determined by following the hydrolysis of pNPP at 349 nm in 0.1 M NaCl buffered to pH 5.0 with 0.01 M sodium acetate. Fractions showing absence of contaminant protein on the basis of SDS-PAGE with the highest specific activity were pooled and concentrated to a final concentration of ~ 12 mg/mL with use of Amicon Ultra centrifugal filtration tubes (15,000 molecular mass cut-off; Millipore, Billerica, MA 01821). Enzyme was stored for further use in 30% (v/v) glycerol at − 80 °C. Kinetic studies For the chromophoric substrate pNPP, real-time data acquisition was carried out with use of a Cary-15 recording spectrophotometer modified by On-Line Instrument Systems, Inc. (Bogart, GA 30622). Initial velocity data were collected at 22 °C under conditions corresponding to the

13



steady-state approximation by following the hydrolysis of pNPP at 349 nm catalyzed by PTP1B in 0.1 M NaCl buffered with either 0.01 M sodium acetate or sodium 3,3′-dimethylglutarate to pH 5.0. pNPP concentrations ranged from 0.5–1.3 × KM. Higher concentrations were generally associated with indications of substrate inhibition. In the absence of metal-chelate inhibitors, the reaction was initiated by addition of enzyme; in the presence of metal-chelates to characterize their inhibitory properties, pNPP and enzyme were added separately but simultaneously to avoid potential inhibitory complexing of the chelate with the substrate or the enzyme. Experiments designed to investigate the potential influence of complex formation between the enzyme and vanadyl ­(VO2+) chelates are described in “Results”. The value of the (εsubstrate − εproduct) difference extinction coefficient of 1.911 × 10−3 M−1 cm−1 used to quantify turnover with pNPP as substrate is a maximum at 349 nm and pH 5, and was applied as in preliminary studies [14]. Buffer solutions were thoroughly purged with nitrogen prior to data collection to avoid oxidation of ­VO2+-chelates or of enzyme; stock solutions of V ­ O2+-chelates were prepared just prior to use with nitrogen-purged DMSO and kept in airtight Hamilton syringes (Reno, NV 89502) equipped with an automatic dispenser. The fractional volume of DMSO in the reaction mixture was maintained constant at 0.33% (v/v) for all measurements. Initial velocity data were similarly collected following the hydrolysis of the undecapeptide substrate DADEpYLIPQQG by fluorescence emission with use of a Horiba Jobin–Yvon FluoroMax3 fluorescence spectrometer (Edison, NJ 08820). The reaction was monitored at 20 °C in 0.1 M NaCl buffered to pH 6.0 with 0.01 M Tris-Ac, corresponding to the maximum in the pH profile of kcat for this substrate [26]. Substrate concentrations ranged from 0.8–3.6 × KM. Excitation and emission wavelengths were set at 272 and 300 nm, respectively, with a 3-mm excitation slit width and a 10-mm emission slit width. The fluorescence emission intensity was monitored at 0.5 s intervals with a 0.5-s integration time. Buffers were similarly purged with nitrogen prior to use, and stock solutions of ­VO2+-chelates were prepared just prior to use, as described above. Prior to collecting initial velocity data with ­VO2+-chelates as inhibitors, we tested whether they facilitated substrate hydrolysis under the conditions used for steady-state velocity measurements but in the absence of enzyme. None was detected. Data analysis Initial velocity data were analyzed first with use of Origin 5.0 (OriginLab Corp., Northampton, MA 01060). Modeling of inhibitor effects was evaluated with Origin 2017. With pNPP as substrate, the initial velocity was estimated from

13

J Biol Inorg Chem

a least-squares fit to the linear region within the first 20 s of progress curves and within the first 40 s with DADEpYLIPQQG as substrate. For each data set, KM and Vmax were estimated from hyperbolic fits of the initial velocity data in the absence of chelate inhibitors. Competitive, uncompetitive, and noncompetitive models of inhibition were then evaluated on the basis of hyperbolic fits to the full data set with KM held constant, and KI and Vmax allowed to vary. Double reciprocal plots of the data were then constructed with use of the parameters generated from the hyperbolic fits. Final model interpretation was determined qualitatively by the fit of hyperbolic and Lineweaver–Burk plots, and quantitatively on the basis of the χ2 value. In each case the initial velocity data were evaluated by Dixon [27] or Cornish-Bowden [28] plots to confirm the type of inhibition and the value of KI.

Results Spectral characterization of ­VO2+‑chelates Because VO(acac)2 and PTP1B exhibit spectral contributions in the near UV region overlapping with that of pNPP and its hydrolysis products, our first objective was to identify conditions under which changes in the spectra of the enzyme or chelate could impact data collection for substrate hydrolysis. Dissolution of VO(acac)2 into aqueous mixtures results in formation of up to four different, pH-dependent species that are detectable by EPR [10, 29]. These qualitative changes in EPR spectra as a function of time and pH have been ascribed to alterations in ligand composition and coordination geometry by Crans and coworkers [29–31]. The suggested changes include (1) transformation of the equatorial trans conformation of the V ­ O2+-chelate to a cis conformation whereby one of the carbonyl oxygen donor atoms of the acetylacetonate ligand becomes axially coordinated and an H ­ 2O molecule is incorporated into an equatorial coordination site or (2) dissociation of one of the organic ligands with acquisition of two equatorially coordinated ­H2O molecules. Crans and coworkers have not corroborated the purported changes in ligand composition and structure on the basis of any measurable spectroscopic parameter. Such changes in ligand composition and structure can be identified in principle via the superhyperfine or ligand hyperfine structure in EPR spectra. For VO(acac)2 the hyperfine broadening of protons in the nearby vicinity of the ­V4+ ion is detectable by EPR only from equatorial ligands, is small (approximately 1.25 G for a vanadium coordinated –OH group), and adds to the underlying broadening of EPR absorption features due to the I = 7/2 51V nucleus [32–34]. We analyzed the ligand hyperfine interactions of VO(acac)2 as a function of pH by application of electron-nuclear double

J Biol Inorg Chem

resonance (ENDOR) spectroscopy [10, 35, 36]. Analysis of ENDOR spectra of VO(acac)2 at neutral and low pH showed no dissociation of organic ligands, no incorporation of an equatorially coordinated ­H2O molecule, and no transformation of the ­VO2+-chelate from a trans to a cis conformation [10] although such changes would have been detectable, had they occurred [36]. These observations, thus, directly contradict the suggestion of changes in speciation due to pH [29–31] and indicate that the structure of VO(acac)2 is by itself stable in solution. In earlier studies we have shown that the action of VO(acac)2 as an insulin mimetic in cultured 3T3-L1 adipocytes is restricted to the enzyme that controls the phosphotyrosine levels of the insulin receptor and the insulin receptor substrate-1 protein [11]. PTP1B is likely the most important phosphatase with this role in cells [13]. However, to investigate the properties of V ­ O2+-chelates as inhibitors of PTP1B, our initial objective was to determine conditions in which binding of V ­ O2+-chelates to PTP1B remained compatible with the rapid equilibrium assumption of steady-state kinetics. We also needed to identify conditions in which alterations in the absorption spectrum of the protein or the ­VO2+-chelate might impact data collection for the time-dependent hydrolysis of substrates catalyzed by PTP1B. Figure 2 compares changes in the UV absorption spectrum of VO(acac)2 in sodium acetate buffered 0.1 M sodium chloride to changes in the spectrum of VO(acac)2 in the presence of PTP1B. The same buffer was used for the collection of kinetic data. Because oxidation of VO(acac)2 is accompanied by an increase in absorption at 270 nm relative to the decrease at 312 nm [37], we conclude that the spectral changes in Fig. 2a reflect only alterations in solvation and possibly ligand geometry, as earlier characterized by EPR and ENDOR methods [10, 36]. In aqueous solutions, VO(acac)2 acquires both an axially coordinated water molecule in its inner coordination shell and a water molecule hydrogen bonded to the oxygen atom of the axial V=O moiety [36], both of which are absent in the DMSO solution. We, therefore, attribute the spectral changes in Fig. 2a to equilibration of VO(acac)2 to its aqueous environment. Despite relatively small changes in absorption at 349 nm seen in Fig. 2a, the wavelength at which hydrolysis of pNPP is best monitored at pH ~ 5 [14], addition of the chelate to buffer had a measurable negative rate at this wavelength that would have impacted collection of initial velocity data for hydrolysis of pNPP by as much as 20%. A 5-min equilibration time, therefore, was allowed for the spectral changes to go to completion prior to addition of substrate and enzyme. This precaution ensured that following the reaction spectrophotometrically reflected only the rate of hydrolysis of pNPP catalyzed by the enzyme in the presence of the metal-chelate without interference from spectral changes of the type in Fig. 2a. To minimize the potential effects of complexing of

Fig. 2  Comparison of UV absorption spectra of VO(acac)2 in the absence (Panel a) and presence (Panel b) of PTP1B. In Panel a an aliquot of VO(acac)2 dissolved in neat DMSO was added to a final concentration of 6  ×  10−5 M and mixed in N ­ 2-purged 0.1  M NaCl buffered to pH 5 with 0.01 M sodium acetate at 22 °C. Spectra were collected, in direction of the arrow, at 0, 1.5, 3, 4.5, 6, and 9 min. In Panel b an aliquot of VO(acac)2 was added and allowed to come to equilibrium as in Panel a. Small aliquots of a stock PTP1B solution were then added to final concentration of 0, 0.089, 0.18, 0.23, 0.35, 0.43, 0.51, and 0.59 × 10−6 M, as indicated by the arrow, and mixed. The spectra were collected only after equilibrium had been reached for addition of each enzyme aliquot. The addition of enzyme aliquots contributed to less than a 2% increase in the total volume of the solution in the cuvette

the chelate to either the substrate or the enzyme on the rate of the reaction when followed either by absorption or by fluorescence, substrate and enzyme were added simultaneously but separately to the solution containing VO(acac)2 to initiate the reaction. Figure 2b shows the spectral changes of VO(acac)2 in the presence of increasing concentrations of PTP1B. Equilibrium of the interaction of the chelate with the enzyme was reached within 3 min of addition of each aliquot of the enzyme solution. Plotting the maximum change in absorption at 312 nm against the enzyme concentration yielded an approximate equilibrium dissociation constant of 2.6 × 10−7

13



M. The value of this dissociation constant differs significantly from the inhibitor binding constant of 6.5 × 10 −6 M determined under steady-state conditions (see below). In addition, the equilibration time of the spectral changes observed in Fig. 2b is significantly greater than the nearly instantaneous reversible association exhibited in kinetic studies, suggesting that the spectral changes in Fig. 2b represent additional types of binding to the enzyme that are not compatible with rapid equilibrium binding. Addition of pNPP after addition of saturating amounts of VO(acac)2 to PTP1B showed complete loss of catalytic activity. Because of the loss of absorption in the long wavelength end of the spectrum, we believe that the changes observed in Fig. 2b indicate a change in the immediate coordination environment of the ­VO2+ moiety through binding to the protein with possible displacement of one or both of the acetylacetonate ligands. Scatchard plot analysis of the absorption changes at 312 nm in Fig. 2b indicated a site of 1:1 binding stoichiometry of tight binding with multiple sites of lower binding affinity. Because the inhibitor constants for reversible binding measured under steady-state conditions, as shown below, differed greatly from that estimated under conditions described in Fig. 2b, we conclude that the binding sites of VO(acac)2 to PTP1B in the absence of substrate do not overlap with that for reversible inhibition under steady-state conditions. Also, the time dependence of the changes in Fig. 2b indicate that they cannot be compatible with rapid equilibrium binding under steady-state conditions. Figure 3 compares changes in the absorption spectrum of VO(Et-maltol)2 in sodium acetate buffered 0.1 M sodium chloride to changes in the spectrum of the ­VO2+-chelate in the presence of PTP1B. This chelate and its methyl analog have received much attention through in vitro and in vivo studies as a potential antidiabetic reagent [38–42]. While VO(Et-maltol)2 has not been subjected to spectroscopic studies comparable to those cited above for VO(acac)2, its methyl analog VO(Me-maltol)2 has been widely studied because of its identical inner-shell coordination structure and similar ligand environment. As for VO(acac)2, dissolution of VO(Me-maltol)2 into aqueous medium also is associated with changes in ionization detectable by magnetic resonance methods [43, 44]. However, NMR analysis of VO(Me-maltol)2 over the pH 3–8 range shows no change in ligand stoichiometry or speciation of this ­VO2+-chelate [44]. In contrast to the time-dependent spectral changes observed for VO(acac)2 in Fig. 2a, the changes for VO(Et-maltol)2 in Fig. 3a were very small and occurred within the mixing time. In contrast to the changes observed for the binding of VO(acac)2 to PTP1B, binding of VO(Et-maltol)2 to PTP1B was accompanied by an increase in absorption near 278 nm and a decrease in absorption near 325 nm, as seen in Fig. 3b.

13

J Biol Inorg Chem

Fig. 3  Comparison of UV absorption spectra of VO(Et-maltol)2 in the absence (Panel a) and presence (Panel b) of PTP1B. In Panel a an aliquot of VO(Et-maltol)2 dissolved in neat DMSO was added to a final concentration of 4.5  ×  10−5 M to a solution and mixed in ­N2-purged 0.1 M NaCl buffered to pH 5 with 0.01 M sodium acetate at 22 °C. In Panel b an aliquot of VO(Et-maltol)2 was added to a concentration of 3 × 10−5 M in acetate buffered 0.1 M NaCl as in Fig. 2b. Aliquots of a stock PTP1B solution were added, in the order indicated by the arrow, to final concentrations of 1.0, 3.0, 3.9, and 4.9 × 10−7 M, respectively, and mixed. The spectra were collected only after equilibrium had been reached for addition of each enzyme aliquot. The two different concentrations of the ­VO2+-chelate in panels a and b were chosen to maintain approximately equivalent signal-to-noise in both panels for comparison. Other conditions as in Fig. 2

Because these optical changes could suggest oxidation of the ­VO2+-chelate [37], we have tested the effects of VO(acac)2 and VO(Et-maltol)2 in aerated solutions to induce oxidation of the ­VO2+ moiety. While the results are described later (vide infra), oxidation of VO(Et-maltol)2 did not account for either the spectral changes in Fig. 3b or the reversible inhibition properties under steady-state conditions. Moreover, spectroscopic studies of the analogous chelate VO(Memaltol)2 under comparable conditions demonstrate suitable stability of the chelate against oxidation by dissolved, molecular ­O2 [43, 44]. In contrast to the interaction of VO(acac) 2 with PTP1B, plots of the change in absorption at 278  nm,

J Biol Inorg Chem

as illustrated in Fig. 3b, against enzyme concentration showed no indication of an approach to saturation of a binding site. In kinetic assays the same procedure was employed, as described for VO(acac)2, with the modification that VO(Et-maltol)2 did not require an equilibration time longer than the mixing time for addition of substrate and enzyme to the cuvette to initiate the reaction. While spectral changes are observed through interaction of both VO-(acac)2 and VO(Et-malto)2 with PTP1B, as described above, no chelate elicited spectral changes in the presence of the substrates employed in this investigation within the range of substrate concentrations used in kinetic assays. Also, there was no evidence that any of the chelates alone catalyzed hydrolysis of pNPP. We similarly examined the interaction of VO(mpp)2 with the enzyme and observed tight binding characterized by a dissociation constant of 1.6 × 10−7 M (data not shown). The time dependence was similarly not consistent with reversible inhibition under steady-state conditions. We, therefore, concluded that steady-state experiments using VO(mpp)2 could be carried out in an identical manner to that for VO(acac)2 and VO(Et-malto)2. Steady-state kinetic studies of substrate hydrolysis catalyzed by PTP1B. The nonspecific phosphotyrosine substrate analog para-nitrophenylphosphate. The synthetic organic substrate pNPP has been widely used to characterize the catalytic activity of PTP1B and other PTPases because it is readily available commercially and its chromophoric properties allow a convenient means to monitor the catalytic reaction spectrophotometrically. Nonetheless, it simulates at best only the phosphorylated tyrosine residue of a physiologically relevant substrate. In preliminary kinetic studies of this enzyme, we observed that the inhibition of PTP1B by VO(acac)2 in the presence of pNPP as substrate differs from that observed in the presence of the phosphotyrosinecontaining undecapeptide substrate DADEpYLIPQQG [14], which we refer to as ­EGFR988−998. Because hydrolysis of ­EGFR988−998 in that study was analyzed by quantifying the amount of a phosphate-malachite green precipitate, precluding steady-state conditions, we have carried out a detailed comparison of the influence of V ­ O2+-chelates on the catalytic action of PTP1B with both pNPP and E ­ GFR988−998 on

Table 1  Steady-state kinetic parameters governing the hydrolysis of substrates catalyzed by PTP1B

the basis of initial velocity measurements under the steadystate approximation. Kinetic parameters for the hydrolysis of both pNPP and the phosphotyrosine containing undecapeptide in the absence of inhibitors determined in this investigation and summarized in Table  1 show excellent agreement with respective literature values. Because we observed different inhibitory behavior of VO(acac)2 dependent on the substrate, we subjected initial velocity data collected for substrate hydrolysis in the presence of VO(acac)2 to a two-stage, leastsquares fitting method [45] to arrive at the best interpretation of its mode of inhibition. Because of the success of this approach, we extended the analysis also to VO(Et-malto)2 and VO(mpp)2. To this end Fig. 4 illustrates a double reciprocal plot of initial velocity data for hydrolysis of pNPP in the presence of VO(acac)2 analyzed according to two classical models of reversible inhibition: competitive in Fig. 4a and noncompetitive in Fig. 4b. Each panel shows an expanded view of the region near the origin of the plot to better illustrate the intersection point on the ordinate axis in Panel A and on the abscissa in Panel B. Mixed inhibition and uncompetitive inhibition were also tested but were not supported by the results, as discussed below. We note that in preliminary kinetic studies, mixed noncompetitive inhibition was observed [14]. In that study the equilibration time of VO(acac)2, when added to an aqueous solution, as described in Fig. 2, was not taken into account. Also, the enzyme was purified with the use of dithiothreitol instead of TCEP, as in the present investigation. Because dithiothreitol is known to form covalent adducts with free sulfhydryl groups, this difference may account in part for the different mode of inhibition. In addition to C215, there are five other cysteinyl residues within the soluble portion of the enzyme (residues 1–321) employed in this investigation [46]. Selection of the model mathematically best constrained by the data was carried out by first calculating the Vmax and K M values for the dataset, then constraining them during total least-squares fitting to the four kinetic models, competitive, noncompetitive, and uncompetitive, under comparison [45]. We also tested mixed inhibition but were able to rule it out in every case, as described in

Substrate

kcat ­(s−1)

KM (M × 106)

kcat/KM ­(M−1s−1 × 10−6)

References

pNPP

62.9 ± 12.0 69.2 ± 1.1 44.4 ± 4.3 44.6 ± 1.8

460 ± 120 339 ± 16 4.1 ± 0.6 3.9 ± 0.9

0.14 0.20 10.8 11.0

This study [64]a This study [4]

EGFR988−998 a

 Rat PTP1 (residues 1–323) with 97% chemical identity to human PTP1B (residues 1–321) was studied; kinetic parameters are compared for 0.1 M ionic strength and pH 5.5, corresponding closely to conditions employed in this study

13



J Biol Inorg Chem

Fig. 4  Lineweaver–Burk plots of initial velocity data of PTP1B catalyzed hydrolysis of pNPP in the presence of VO(acac)2. Lines are plotted according to parameters obtained from hyperbolic fits to the data as described in the text for (Panel a) competitive inhibition and (Panel b) noncompetitive inhibition. The reaction mixture consisted of N ­ 2-purged 0.1 M NaCl buffered to pH 5 with 0.01 M sodium

acetate. Enzyme concentration was 3.2  ×  10−8 M. Inhibitor concentrations were (×  106 M): closed circle (0), closed up triangle (6.0), closed down triangle (10), closed diamond (14). The area near the origin is expanded in the inset to emphasize that the intersection of each straight-line with the ordinate was drawn according to parameters based on hyperbolic fits to initial velocity data

Supplementary Material. On this basis, each model, competitive, noncompetitive, and uncompetitive, had only one free parameter and could be compared to the others on the basis of its associated χ2 value [47]. By this analysis, the χ2 differs from the Bayesian Information Criterion merely by a constant [48]. Results of this statistical analysis for hydrolysis of pNPP catalyzed by PTP1B are compared for ­VO2+-chelates in Table 2. While the best interpretation of the results in Fig. 3 based on the χ2 values associated with each KI determination assigns the inhibition as competitive, the χ2 value for the noncompetitive fit to the data is visually difficult to exclude as a possibility. However, although the value of χ2 for the noncompetitive fit is low, model selection is based on the lowest value, and that belongs to the competitive fit. The uncompetitive model is readily discarded because of its associated χ2 value of 9.44. We note that Kuzmic and coworkers used a similar statistical method based on information theory to rule out kinetic patterns of enzyme–inhibitor binding in the case of the anthrax lethal factor protease [49].

Figure 5 illustrates the double reciprocal plot of initial velocity data for hydrolysis of pNPP in the presence of VO(Et-malto)2. The analysis showed more straightforwardly that VO(Et-malto)2 exhibited noncompetitive inhibition for hydrolysis of pNPP. The KI value of the VO(mpp)2 chelate in the presence of pNPP as substrate is also included in Table 2 and proved to be compatible only with competitive inhibition on the basis of its χ2 value and Dixon plot. As described in the Methods section, we have collected initial velocity data to characterize the inhibitory properties of these vanadyl chelates, taking every precaution to prevent or at least minimize oxidation of the ­VO2+ moiety by solubilized molecular ­O2. In order to test the effects of oxidation of chelates, we aerated solutions of the chelates prior to addition of enzyme and substrate and determined their inhibition properties under identical conditions compared with those that were used for kinetic assays of the ­VO2+-chelates. All of the chelates exhibited unambiguously competitive inhibition with smaller and distinguishably different KI values (data not shown). The resultant change in

Table 2  Dissociation inhibitor constants of ­VO2+-chelates

Substrate

Pnpp

EGFR988−998 a

Inhibitor

VO(acac)2 VO(Et-malto)2 VO(mpp)2 VO(acac)2 VO(Et-malto)2

KI (M × 106)

6.5 ± 0.4 14.8 ± 1.4 38.2 ± 0.5 0.16 ± 0.01 0.14 ± 0.01

χ2 Compa

Noncompa

Uncompa

0.45 24.72 5.17 18.02 10.49

1.77 14.69 9.70 7.92 1.72

9.44 17.02 34.2 2.30 7.02

 comp competitive, noncomp noncompetitive, uncomp uncompetitive

13

J Biol Inorg Chem

Fig. 5  Lineweaver–Burk plots of initial velocity data of PTP1B catalyzed hydrolysis of pNPP in the presence of VO(Et-malto)2. Solution conditions and determination of kinetic parameters from hyperbolic plots as in Fig. 3. Enzyme concentration was 2.9 × 10−8 M. Inhibitor concentrations were (× 106 M): closed square (0); closed circle (2.0), closed up triangle (3.0), closed down triangle (4.0), closed diamond (6.0)

kinetic behavior upon introduction of atmospheric ­O2 into the medium affirms that the oxidation state and coordination environment of the vanadium atom in the presence of dissolved ­O2 differed from that associated with the central V ­ 4+ 2+ cation of ­VO -chelates. On this basis we conclude that the inhibitory behavior of the three V ­ O2+-chelates employed in this investigation, as shown in Figs. 4 and 5 and summarized in Table 2, is attributable to each chelate with a central ­VO2+ moiety. The phosphotyrosine-containing undecapeptide substrate analog of the EGFR. In preliminary kinetic studies of PTP1B, we observed that inhibition of hydrolysis of the phosphotyrosine-containing undecapeptide substrate DADEpYLIPQQG catalyzed by PTP1B in the presence of VO(acac)2 strongly suggested uncompetitive behavior, differing from that observed for the hydrolysis of pNPP [14]. Because hydrolysis of E ­ GFR988−998 in that study was analyzed by quantifying the amount of a phosphate-malachite green precipitate generated over a 30-min period, precluding steady-state conditions, we have carried out a detailed comparison of the influence of V ­ O2+-chelates on the catalytic action of PTP1B under the steady-state approximation. Initial velocity data for the PTP1B catalyzed hydrolysis of ­EGFR988−998 were collected by fluorescence. In the absence of ­VO2+-chelates the values of the kinetic parameters for PTP1B catalyzed hydrolysis of this substrate, summarized in Table 1, agreed well with those published by others [4, 26]. Figure 6 compares the results of steady-state kinetic analysis of PTP1B activity as a function of the concentration

Fig. 6  Lineweaver–Burk plot of inhibition of PTP1B catalyzed hydrolysis of E ­ GFR988−998 in the presence of VO(acac)2. The double reciprocal plot was constructed on the basis of the parameters generated from hyperbolic fits of initial velocity data. For each inhibitor concentration, KM and Vmax values determined in the absence of the inhibitor were fixed. Competitive, noncompetitive, and uncompetitive models of inhibition were then evaluated on the basis of hyperbolic fits to the full data set by varying KI and, for noncompetitive and uncompetitive models, Vmax. The final interpretation of the mode of inhibition for the chelate was selected on the basis of the best fit, evaluated visually on the basis of double-reciprocal plots and quantitatively by the χ2 value. The reaction mixture consisted of ­N2-purged 0.1 M NaCl buffered to pH 6 with 0.01 M Tris-Ac. The concentration of PTP1B was 3.75 × 10−9 M. Initial velocity data in the absence of inhibitor are indicated by squares (closed square). VO(acac)2 concentrations were (× ­109 M): closed circle (50), closed up triangle (100), closed down triangle (150)

of VO(acac)2 with E ­ GFR988−998 as the substrate. To arrive at the best interpretation for both ­VO2+-chelates, we tested the data against competitive, mixed, noncompetitive, and uncompetitive models comparing the rectangular hyperbolic and Lineweaver–Burk plots and the associated χ2 values, as described above. The best fit to the initial velocity data illustrated in Fig. 6 and summarized in Table 2 showed that VO(acac)2 is an uncompetitive inhibitor in the presence of the phosphotyrosine-containing undecapeptide substrate (also, cf., Supplementary Material). This is the only uncompetitive inhibitor of PTP1B reported heretofore. The results in Fig. 6 and Table 2 confirm our earlier assessment of VO(acac)2 as an uncompetitive inhibitor in the presence of ­EGFR988−998 as substrate [14]. Figure 7 compares the results of steady-state kinetic analysis of PTP1B catalyzed hydrolysis of ­EGFR988−998 in the presence of VO(Et-malto)2. The initial velocity data were similarly tested against all four models on the basis of the rectangular hyperbolic and Lineweaver–Burk plots and associated χ2 values. The KI value and its associated χ2 values are given in Table 2. For VO-(Et-malto)2 the noncompetitive

13



J Biol Inorg Chem

Peters and coworkers state that to generate protein bound VO(Me-malto)2 in crystals of PTP1B, the chelate, dissolved in water, was added to the polyethylene glycol based mother liquor containing crystals followed by diffusion of the chelate into the crystals for 2 h at room temperature and open to air prior to X-ray data collection [50]. These conditions were obviously sufficient for the organic ligands to be stripped from the ­VO2+ moiety. Generation of the orthovanadate anion is likely due to subsequent oxidation by dissolved O ­ 2 in the solvent channels of the crystal and X-ray induced oxidation of the ­V4+ to a ­V5+ species.

Discussion Fig. 7  Lineweaver–Burk plot of initial velocity data of PTP1B catalyzed hydrolysis of E ­ GFR988−998 in the presence of VO(Et-malto)2. Conditions as in Fig. 3. Initial velocity data in the absence of inhibitor are indicated by squares (closed square). VO(Et-malto)2 concentrations were (× 109 M): closed down triangle (30), closed circle (60), closed up triangle (90)

model is clearly favored statistically. Also, the data strongly argue against a competitive model. The inhibition patterns observed for VO(Et-malto)2 for both pNPP and E ­ GFR988−998, as substrates, contradict the interpretation of Peters and coworkers [50], who suggest that inhibition of PTP1B by the structurally analogous chelate VO(Me-malto)2, containing a methyl group instead of an ethyl group at the 2-position of the pyrone ring, is due to oxidation of the V ­ O2+ moiety stripped of its organic ligands and bound in the active site as the orthovanadate anion. Although Peters and coworkers do not provide an estimate of the value of KI for competitive inhibition of PTP1B by VO(Me-malto)2, orthovanadate has been characterized in detail as a competitive inhibitor and has a KI value of ~ 3 × 10−7 M [51]. The orthovanadate anion cannot account for the noncompetitive inhibition pattern of VO(Et-malto)2 in Figs. 5 and 7. To identify the origins of the discrepancy between our results and those of Peters and coworkers [50], we note that the Lineweaver–Burk plot for VO(Me-malto)2 inhibition of PTP1B catalyzed hydrolysis of 6,8-difluoro-4-methylumbelliferyl phosphate, similarly a nonspecific substrate, (Fig. 2d, p 325, of ref. [50]) does not show a monotonically increasing slope dependent on VO(Me-malto)2 concentration. This pattern is inconsistent with linear competitive inhibition. Furthermore, Peters and coworkers state that for kinetic studies they incubated the chelate with the enzyme for 10 min prior to initiating the reaction by addition of substrate. On the basis of the spectral changes for the analogous chelate in Fig. 3, this treatment cannot result in binding of the chelate to the enzyme consistent with reversible, rapid equilibrium binding. Furthermore,

13

A variety of vanadium containing chelates of oxidation states + 3, + 4, and + 5, have been investigated with respect to their capacity to enhance cellular uptake of glucose [12, 42, 52]. While the main objective of most of these investigations has been to develop a pharmacological reagent for potential treatment of diabetes, the target enzyme or mechanism underlying the capacity of vanadium compounds to enhance glucose uptake has remained undefined in most cases. In this respect the demonstration through Figs. 4, 5, 6, 7 that the ­VO2+-chelates investigated in this study inhibit PTP1B indicates that inhibition of this critical enzyme underlies at least part of their capacity to enhance glucose uptake. PTP1B is known as an important enzyme regulating in vivo the phosphotyrosine status of the insulin receptor [13]. We have shown that increased phosphotyrosine forms of the IR and of IRS-1 detected on the basis of Western immunoblots of 3T3-L1 adipocytes in the presence of VO(acac)2 are due to inhibition of PTP1B in a manner that is synergistic with insulin [11]. Observation of synergism indicates that VO(acac)2 and insulin influence the same pathway [10, 11]. The noncompetitive inhibition of hydrolysis of pNPP and of the phosphorylated undecapeptide substrate E ­ GFR988−998 in the presence of VO(Et-malto)2 suggests that this chelate binds to the same site on the enzyme in the presence of both substrates and indicates that it is not overlapping with the active site. Because noncompetitive inhibition of hydrolysis of pNPP catalyzed by PTP1B is observed for an aromatic derivative of benzbromarone designated as Compound 2 [9], the only other noncompetitive inhibitor of this enzyme reported hitherto, we examined by molecular modeling whether the noncompetitive binding exhibited by VO(Etmalto)2 is sterically compatible with the binding site of the organic benzbromarone inhibitor. We were led to examine this hypothesis because of the butterfly-like structure of Compound 2, [3-(3,5-dibromo-4-hydroxy-benzoyl)-2-ethylbenzofuran-6-sulfonic acid (4-sulfamoyl-phenyl)amide] [9], and of the cis conformation of VO(Et-malto)2 modeled according to the structure of the 2-methyl-maltolato analog

J Biol Inorg Chem

in aqueous medium established by ENDOR spectroscopy [36]. The atomic numbering scheme of VO(Et-malto)2 and of the benzbromarone inhibitor known as Compound 2 [9] are shown in Fig. S1. The fit of the cis conformation of VO(Et-malto)2 in the binding site of Compound 2 in PTP1B is illustrated in Fig. S2. As seen in Fig. S2, the van der Waals surface of the amino acid residues surrounding Compound 2 [9] accommodates VO(Et-malto)2 without violation of steric interactions. Because VO(Et-malto) 2 is associated with noncompetitive inhibition, as is Compound 2, we suggest that the binding site of VO(Et-malto)2 may overlap with that defined for Compound 2 [9]. Compound 2 and VO(Etmalto)2 are the only inhibitors of PTP1B identified, thus far, as noncompetitive. In the presence of pNPP as substrate, Li and coworkers observed mixed inhibition by VO(Me-malto)2 of an engineered, fused protein construct consisting of glutathioneS-transferase and the full length PTP1B polypeptide chain (435 residues) [53]. While the VO(Me-malto)2 chelate is essentially structurally identical to VO(Et-malto)2 employed in this investigation, it is difficult to relate the results of Li and coworkers to those presented here because of their protein construct. No comparison was made to compare the inhibitory properties of VO(Me-malto)2 in the presence of phosphotyrosine-containing peptide mimics as substrates. The important challenge in this investigation is to explain the apparently different modes of inhibition exhibited by VO(acac)2 in the presence of pNPP and E ­ GFR988−998 as substrates. Given the care in our experiments to exclude potential artifacts in data collection, e.g., partial oxidation of the ­VO2+ moiety, demonstration that the site of reversible binding of VO(acac)2 under steady-state conditions is not identical to that observed in the absence of substrate (cf., Fig. 2) and accounting for small absorption changes in the spectrum of the chelate during its equilibration with aqueous solvent require that the site of reversible inhibition of VO(acac)2 with the enzyme is the same for both pNPP and E ­ GFR888−899 substrates under steady-state conditions. In contrast to the cis conformation of VO(Me-malto)2, VO(acac)2 retains a trans, planar coordination geometry free in aqueous solution [36], bound to serum albumin [10], and upon uptake into xenograft tumor cells [54]. Based on these results, we expect that VO(acac)2 similarly retains only a trans, planar geometry bound to PTP1B. This assertion is further strengthened by the observation that Zn(acac)2 and Mg(acac)2, as closedshell analogs of VO(acac)2 that do not possess an axial oxygen atom, exhibit values of KI > 600 × 10−6 M (Makinen MW and Zhou KI, unpublished observations), suggesting that the ­VO2+ moiety is necessary for inhibition and likely interacts directly with the protein. In addition, because we have tested and found by molecular modeling that the planar structure of VO(acac)2 could not be accommodated sterically into the region occupied by the fused ring system of the

benzbromarone inhibitor [9], we conclude that the binding site of VO(acac)2 differs from that suggested in Fig. S2 for VO(Et-malto)2. Because changes in protein conformation due to interaction of the enzyme with its physiological substrate analog E ­ GFR988−998 are uncharacterized and because an uncompetitive inhibitor binds to the substrate:enzyme complex, as discussed below, a similar modeling analysis could not be carried out for VO(acac)2. The action of a classical competitive inhibitor that competes with the substrate in the active site is independent of substrate structure. On this basis we conclude that, in the presence of pNPP as substrate, VO(acac)2 does not bind directly in the active site but binds in a site, inducing structural alterations that thwart binding of sterically small substrates such as pNPP. This form of apparent competitive inhibition has been described [55]. While the site of VO(acac)2 binding may be near the active site, it cannot be overlapping with that of the much larger DADEpYLIPQQG undecapeptide substrate. Furthermore, because active site residues of the enzyme experience a much larger number of hydrogen-bonding and electrostatic interactions with the undecapeptide substrate than with pNPP [4, 25], the resultant structural alterations induced in the protein favor tighter binding of VO(acac)2 than with pNPP, as reflected by their respective KI values in Table 2. Because of its central role as a negative regulator of insulin and leptin signaling and its emergence as a positive factor in tumorigenesis [56–59], PTP1B has been a target for therapeutic intervention through drug design. Nonetheless, despite the concerted efforts of both academia and the pharmaceutical industry, no PTP1B-specific drug has yet been introduced into pre-clinical studies. A significant obstacle in these efforts has been that the active site of the enzyme, towards which most drug design studies have been directed, attracts highly negatively charged substrates that bind to multiple, positively charged residues. Design of highly anionic substrate mimetics results in compounds with limited bioavailability and poor penetration into cells. Furthermore, drugs targeted for the active site of PTP1B exhibit low specificity because of the H/V-C-X5-R-S/T consensus sequence of the catalytic motif that PTP1B shares with other tyrosine phosphatases. In this respect, VO(acac)2 differs sharply from other inhibitors of PTP1B not only as an uncompetitive inhibitor but also as an electrostatically neutral compound that readily enters cells. Scheme 1 below illustrates uncompetitive inhibition, where E represents the enzyme, S the substrate, and I the uncompetitive inhibitor. In uncompetitive inhibition the inhibitor binds to the enzyme–substrate or Michaelis complex. Since the enzyme is known to undergo a conformational change from open to closed upon substrate binding, involving the WPD loop, i e., residues ­Trp 179 –Pro 180 –Asp 181 [60, 61], after which substrate

13



J Biol Inorg Chem

References

Scheme  1  Diagrammatic illustration of an enzyme catalyzed reaction in the presence of an uncompetitive inhibitor

hydrolysis occurs, VO(acac)2 must bind to the open conformation, preventing the conformational change and substrate hydrolysis. Identification of the site of uncompetitive inhibition of hydrolysis of the ­EGFR888−899 substrate by VO(acac)2 may lead to development of inhibitors with enhanced specificity because, in comparison to pNPP, phosphorylated EGFR is a natural, physiologically relevant substrate of PTP1B. Furthermore, identification of the site of uncompetitive inhibition would be particularly important with respect to drug design because uncompetitive inhibitors in open, cellular systems are pharmacologically more potent than competitive inhibitors [16, 17]. Most studies to identify inhibitors of PTP1B have been carried out on the basis of detecting inhibition of hydrolysis of pNPP. Except for the characterization of the analog of benzbromarone as a noncompetitive inhibitor [9], all other inhibitors have been described as competitive. Our observations of the inhibition of hydrolysis of pNPP in the presence of VO(acac) 2 as of apparent competitive character implies that some inhibitors described only on the basis of inhibition of hydrolysis of pNPP may act similarly to VO(acac)2. For this reason we conclude that development of pharmacologically effective inhibitors of PTP1B is likely to be more successfully pursued through use of phosphotyrosine containing peptides that mimic physiologically relevant substrates. Because of the diversity of protein substrates of protein tyrosine phosphatases [59, 62, 63], this approach may help to identify inhibitors of greater metabolic specificity Acknowledgements  We thank Professor Z. Y. Zhang for providing the DNA for overexpression of human PTP1B. IJR was supported by an NIH Roadmap Training Program (T90 DK070076). PBB was supported by a training Grant of the National Institutes of Health at the Interface of Chemistry and Biology (T32GM008720). This research was supported by Grants from the National Institutes of Health (P50 CA125183 and P30 DK020595) and by the Department of Biochemistry and Molecular Biology at The University of Chicago. Author contributions  PBB and SQ synthesized VO(mpp)2 and VO(Et-malto)2, respectively; MAC, JHH, MSH, and IJR overexpressed and purified the enzyme and collected and analyzed preliminary steadystate kinetic data; JHH and MWM conducted the final analysis of kinetic data and wrote the manuscript.

13

1. Tonks NK, Diltz CD, Fischer EH (1988) J Biol Chem 263:6731–67372 2. Tonks NK, Diltz CD, Fischer EH (1988) J Biol Chem 263:6722–6730 3. Bakke J, Haj FG (2015) Semin Cell Dev Biol 37:58–65 4. Sarmiento M, Zhao Y, Gordon SJ, Zhang ZY (1998) J Biol Chem 273:26368–26370 5. Zhang S, Zhang ZY (2007) Drug Discov Today 12:373–381 6. Zhang ZY (1998) Crit Rev Biochem Mol Biol 33:1–52 7. Zhang ZY (2002) Annu Rev Pharmacol Toxicol 42:209–234 8. He R, Yu ZH, Zhang RY, Wu L, Gunawan AM, Lane BS, Shim JS, Zeng LF, He Y, Chen L, Wells CD, Liu JO, Zhang ZY (2015) ACS Med Chem Lett 6:782–786 9. Wiesmann C, Barr KJ, Kung J, Zhu J, Erlanson DA, Shen W, Fahr BJ, Zhong M, Taylor L, Randal M, McDowell RS, Hansen SK (2004) Nat Struc Mol Biol 11:730–737 10. Makinen MW, Brady MJ (2002) J Biol Chem 277:12215–12220 11. Ou HS, Yan LM, Mustafi D, Makinen MW, Brady MJ (2005) J Inorg Biol Chem 10:874–886 12. Makinen MW, Salehitazangi M (2014) Coord Chem Rev 279:1–22 13. Elchebly M, Payette P, Michaliszyn E, Cromlish W, Collins S, Loy AL, Normandin D, Cheng A, Himms-Hagen J, Chan CC, Ramachandran C, Gresser MJ, Tremblay ML, Kennedy BP (1999) Science 283:1544–1548 14. Makinen MW, Rivera SE, Zhou KI, Brady MJ (2007) In: Kustin K, Pessoa JC, Crans DC (eds) Vanadium: the versatile metal, Am Chem Soc, Washington, DC, pp 82–92 15. Milarsk KL, Zhang ZY, Dixon JE, Saltiel AR (1993) J Biol Chem 268:23634–23639 16. Cornish-Bowden A (1986) FEBS Lett 203:3–6 17. Westley AM, Westley J (1996) J Biol Chem 271:5347–5352 18. Thompson KH, Liboiron BD, Bellman YSKDD, Setyawati IA, Patrick BO, Karunaratne V, Rawji G, Wheeler J, Sutton K, Bhanot S, Cassidy C, McNeill JH, Yuen VG, Orvig C (2003) J Biol Inorg Chem 8:66–74 19. Rangel M, Tamura A, Fukushima C, Sakura H (2001) J Biol Inorg Chem 6:128–132 20. Harris R (1976) Aust J Chem 29:1329–1334 21. Zhang ZY, Thiemesefler AM, Maclean D, McNamara DJ, Dobrusin EM, Sawyer TK, Dixon JE (1993) Substrate-specificity of the protein–tyrosine phosphatases. Proc. Natl. Acad. Sci. (USA) 90:4446–4450 22. Puius YA, Zhao Y, Sullivan M, Lawrence DS, Almo SC, Zhang ZY (1997) Proc Natl Acad Sci (USA) 94:13420–13425 23. Xie LP, Zhang YL, Zhang ZY (2002) Biochemistry 41:4032–4039 24. Wang SS, Tabernero L, Zhang M, Harms E, Van Etten RL, Stauffacher CV (2000) Biochemistry 39:1903–1914 25. Sarmiento M, Puius YA, Vetter SW, Keng YF, Wu L, Zhao Y, Lawrence DS, Almo SC, Zhang ZY (2000) Biochemistry 39:8171–8179 26. Zhang ZY, Thiemesefler AM, MacLean D, McNamara DJ, Dobrusin EM, Sawyer TK, Dixon JE (1993) Proc Natl Acad Sci (USA) 90:4446–4450 27. Dixon M (1953) Biochem J 55:170–171 28. Cornish-Bowden A (1974) Biochem J 137:143–144 29. Amin SS, Cryer K, Zhang B, Dutta SK, Eaton SS, Anderson OP, Miller SM, Reul BA, Brichard SM, Crans DC (2009) Inorg Chem 39:406–416 30. Crans DC (1998) In: Tracey AS, Crans DC (eds) Vanadium compounds: chemistry, biochemistry, and therapeutic applications. Am Chem Soc, Washington, pp 82–103 31. McLauchlan CC, Peters BJ, Wilsky GR, Crans DC (2015) Coord Chem Rev 301–302:163–199

J Biol Inorg Chem 3 2. Chasteen ND, Francavilla J (1976) J Phys Chem 80:867–871 33. Albanese NF, Chasteen ND (1978) J Phys Chem 82:910–913 34. Chasteen ND (1981) In: Berliner LJ, Reuben J (eds) Biol Magn Reson, vol 3. Plenum Press, NY, pp 53–119 35. Makinen MW, Mustafi D (1995) In: Sigel H, Sigel A (eds) Metal ions in biological systems, vol 31. Marcel Dekker, NY, pp 89–127 36. Mustafi D, Makinen MW (2005) Inorg Chem 44:5580–5590 37. Grybos R, Samotus A, Popova N, Bogolitsyn K (1997) Trans Metal Chem 22:61–64 38. Yuen VG, Orvig C, McNeill JH (2003) Can J Physiol Pharmacol 81:1049–1055 39. Thompson KH, Orvig C (2003) FASEB J 17:A1132 40. Setyawati IA, Thompson KH, Yuen VG, Sun Y, Battell M, Lyster DM, Vo C, Ruth TJ, Zeisler S, McNeill JH, Orvig C (1998) J Appl Physiol 84:569–575 41. Mehdi MZ, Srivastava AK (2005) Arch Biochem Biophys 440:158–164 42. Thompson KH, Barta CA, Orvig C (2006) Chem Soc Rev 35:545–556 43. Hanson GR, Sun C, Orvig C (1996) Inorg Chem 35:6507–6512 44. Buglyo P, Kiss E, Fabian I, Kiss T, Sanna D, Garribba E, Micera G (2000) Inorg Chim Acta 306:174–183 45. Jukic D, Sabo K, Scitovski R (2007) J Comp Appl Math 201:230–246 46. Romsicki Y, Kennedy BP, Asante-Appiah E (2003) Arch Biochem Biophys 414:40–50 47. Rice JA (1988) Mathematical statistics and data analysis. Wadsworth Inc, Belmont, p 595 48. Schwarz G (1978) Ann Stat 6:461–464 49. Kuzmic P, Cregar L, Millis SZ, Goldman M (2006) FEBS J 273:3054–3062

50. Peters KG, Davis MG, Howard BW, Pokross M, Rastogi V, Diven C, Greis KD, Eby-Wilkens E, Maier M, Evdokimov A, Soper S, Genbauffe F (2003) J Inorg Biochem 96:321–330 51. Huyer G, Liu S, Kelly J, Moffat J, Payette P, Kennedy B, Tsaprailis G, Gresser MJ, Ramachandran C (1997) J Biol Chem 272:843–851 52. Buglyo P, Crans DC, Nagy EM, Lindo RL, Yang LQ, Smee JJ, Jin WZ, Chi LH, Godzala ME, Willsky GR (2005) Inorg Chem 44:5416–5427 53. Li M, Ding W, Baruah B, Crans DC, Wang R (2008) J Inorg Biochem 102:1846–1853 54. Mustafi D, Peng B, Foxley S, Makinen MW, Karczmar GS, Zamora M, Ejnik J, Martin H (2009) J Biol Inorg Chem 14:1187–1197 55. Segal IH (1993) Enzyme kinetics. John Wiley & Sons, Inc., New York, p 957 56. Dadke S, Chernoff J (2003) Curr Drug Targets Immune Endocr Metabol Disord 3:299–304 57. Zhang ZY (2005) Biochim Biophys Acta-Proteins Proteom 1754:100–107 58. Lee S, Wang Q (2007) Med Res Rev 27:553–573 59. Yip SC, Saha S, Chernoff J (2010) Trends Biochem Sci 35:442–449 60. Jia ZC, Barford D, Flint AJ, Tonks NK (1995) Science 268:1754–1758 61. Choy MS, Li Y, Machado LESF, Kunze MBA, Connors CR, Wei XY, Lidorff-Larsen K, Page R, Peti W (2017) Mol Cell 65:644–658 62. Tiganis T, Bennett AM (2007) Biochem J 402:1–15 63. Li X, Wilmanns M, Thornton J, Kohn M (2013) Sci Signal 6:rs10 64. Zhang ZY (1995) J Biol Chem 270:11199–11204

13