Potent Low Molecular Weight Substrates for Protein-tyrosine Phosphatase*

The ability of protein-tyrosine phosphatases (PTP- ases) to catalyze the hydrolysis of simple aromatic phosphates has been recognized for some time. However, these compounds are significantly poorer substrates than their peptide-based counterparts containing phos- photyrosine. Consequently, the effort to create potent PTPase substrates has predominately focused on the use of peptidic carriers to deliver the phosphotyrosine moiety to the enzyme active site. We now report the synthesis and evaluation of several low molecular weight aromatic phosphates that serve as robust sub- strates for the rat PTPase, PTP1. We initially surveyed the ability of PTP1 to catalyze the hydrolysis of a variety of phenyl phosphate structural variants. Sterically demanding substituents positioned ortho and (to a lesser extent) meta to the phosphate group severely compro-mise the ability of these species to serve as phosphatase substrates. However, both benzylic and negatively charged substituents para to the hydrolyzable phos- phate dramatically promote hydrolytic efficiency, which appears to be augmented through a dramatic en- hancement in the affinity of the substrate for PTP1. The best substrate examined in this study exhibits a K m of 16 (cid:54) 3 (cid:109) M . In addition, it serves as an inhibitor of the PTP1-catalyzed hydrolysis of p -nitrophenyl phosphate with a K i of 4.9 (cid:54) 0.7 (cid:109) M . The extraordinary structural simplicity of this compound, as well as those of several others described herein, provides a promising starting point for the design of potent PTPase inhibitors.

The phosphorylation of tyrosine residues in proteins is one of several key molecular mechanisms by which cell growth and differentiation is regulated. Furthermore, the phosphorylation state of proteins is remarkably dynamic, which enables cellular behavior to respond rapidly to discrete changes in environmental conditions (1). This dynamic behavior is governed by the opposing actions of protein-tyrosine kinases and protein-tyrosine phosphatases (PTPases). 1 Protein-tyrosine kinases and PTPases, as well as their corresponding substrates, are integrated within an elaborate signal transducing network, an enzyme-based system which converts external environmental stimuli to internal cellular action. The defective or inappropriate operation of this network is at the root of a variety of disease states, such as cancer. Consequently, the characterization of the individual components and the delineation of the circuitry of this regulatory network has emerged as one of the most active fields in biomedical research. Ultimately, advances in this area may spawn enzyme-specific inhibitors, species that could serve as the basis for the creation of novel chemotherapeutic agents.
Clearly, a successful inhibitor of signal transducing enzymes must not only tenaciously bind to the specific target enzyme(s), but must do so without impeding the catalytic behavior of closely related species. Nature carefully regulates the activity of signaling enzymes by restricting their range of potential protein substrates to specific molecules containing the requisite structural recognition sites. In addition, substrate targeting is further enhanced by confining individual PTPases and protein-tyrosine kinases to specific cellular microenvironments. In much the same way, inhibitory agents must be engineered to recognize their intended target(s) in an exquisitely specific fashion. This has generally, although not exclusively, been approached by synthesizing peptides that encompass the site of phosphorylation (or dephosphorylation) in protein substrates. The inhibitor is simply obtained by replacing a key residue (e.g. a tyrosine phosphate) with an inert one (e.g. a tyrosine phosphonate) (2)(3)(4)(5). These peptide-based inhibitors have been particularly useful in elucidating the molecular and catalytic characteristics of individual enzymes under in vitro conditions. Furthermore, these enzyme-specific inhibitors can be microinjected into cells to provide a biochemical context for the role of specific enzymes. Unfortunately, peptide-based species are less attractive as lead compounds for the generation of medicinally useful drugs.
We have found that PTPases will also utilize non-naturally occurring residues as substrates (6 -8). Indeed, the "tyrosinespecific" PTPases from rat brain (PTP1) and Yersinia both catalyze the dephosphorylation of aromatic and aliphatic phosphates, even if these species are not contained within a peptidebased environment. However, in general, aromatic phosphates are significantly better substrates than their aliphatic counterparts (7). Unfortunately, even these aromatic phosphates are considerably poorer substrates than peptide-based systems (3). With these features in mind, we decided to explore, in greater detail, the ability of tyrosine-specific protein phosphatases to catalyze the hydrolysis of simple aromatic phosphates. Which aromatic substitution patterns are important for enzyme recognition? Is it possible to construct low molecular weight phosphates that are as efficiently hydrolyzed as phosphotyrosinecontaining peptides? We have addressed these questions and have found that the peptidic environment can be replaced with appropriately positioned functionality to produce remarkably efficient PTPase substrates.

MATERIALS AND METHODS
All phenol derivatives and common reagents were obtained from commerical suppliers and used without further purification. Solvents were distilled and dried as required. All phosphorylated phenols were prepared by one of two general methods, except for compounds 1 and 19 * 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.
¶ To whom correspondence should be addressed: Dept. of Molecular Pharmacology and Biochemistry, Albert Einstein College of Medicine, 1300 Morris Park Ave., Bronx, NY 10461. 1 The abbreviations used are: PTPase, protein-tyrosine phosphatase; THF, tetrahydrofuran.
(which were purchased), and 24 (which was prepared from the hydrolysis of commercially available 1,2-phenylene phosophorochloridite). All compounds, with the exception of 18 and 20, were purified by recrystalization as the cyclohexylammonium salt and were characterized as such. The structure of new compounds were confirmed by 1 H NMR (400 Mhz), 13 C NMR (22.5 Mhz), 31 P NMR (161.9 Mhz), and fast atom bombardment mass spectral analysis (negative ion). Enzyme assay solutions were prepared with deionized/distilled water. The catalytic domain of rat PTP1 was expressed and purified as described previously (9).

Preparation of Aryl Phosphates
Note: The carboxylic acid functionalities in compounds 10 -15 were protected prior to the phosphorylation step. A solution of tert-butyldimethylsilyl chloride (2.1 mmol) and N-methylmorpholine (2.1 mmol) in dry THF (5 ml) was added to the carboxyphenol (2.0 mmol) in dry THF (20 mL) and the resultant mixture was stirred at room temperature for 15 min (N 2 ). The solvent was removed in vacuo, and the resultant oily residue subsequently phosphorylated as described below.
Procedure A (8)-The following procedure was employed for the preparation of compounds 2-15 and 17: POCl 3 (2.1 mmol) in dry pyridine (20 ml) was added dropwise (30 min) to a solution of aromatic alcohol (2.0 mmol) in dry pyridine (30 ml) (N 2 ). The reaction was stirred for 1 h at 0°C and then for 1 h at room temperature. The mixture was subsequently poured over ice (approximately 3 g) and then stirred for 15 min. Cyclohexylamine was added to precipitate out the desired aromatic phosphate. The solid was collected via filtration and then twice recrystalized from ethanol.
Procedure B (10)-The following procedure was employed for the preparation of compounds 16, 18, and 20. 1-H-tetrazole (6.0 mmol) in dry THF (5 mL) was added to a solution of aromatic alcohol (2.0 mmol) in dry THF (20 ml) and stirred for 5 min at room temperature (N 2 ). Di-tert-butyldiethylphosphoramidite (2.1 mmol) was subsequently added and the reaction mixture was then stirred for an additional 2 h at room temperature. The mixture was cooled to 0°C and m-chloroperbenzoic acid (4.0 mmol) in CH 2 Cl 2 (5 mL) was added. Finally, 5 ml of NaHSO 3 (10%) was introduced and the mixture extracted with ethyl ether. The combined ethereal solvent was dried, filtered, and removed in vacuo to furnish material which was subsequently purified by silica gel column chromatography (the solvent varied from 20:1 CHCl 3 : CH 3 OH to 10:1 CHCl 3 :CH 3 OH depending upon the aromatic phosphate).
Hydrolysis of 1,2-Phenylene Phosphorochloridite (Preparation of 24)-Water (0.25 ml) was added to a solution of 1,2-phenylene phosphorochloridite (2.0 mmol) in THF (10 ml) maintained at Ϫ20°C. The reaction was stirred and allowed to warm to room temperature over a period of 30 min. The solvent was subsequently removed in vacuo. Cyclohexylamine (6 ml) was added to the residue and the remaining solid was isolated via filtration and then twice recrystalized from ethanol.

PTPase Assay
All enzyme assays were performed at 30°C in 50 mM succinate, 1 mM EDTA, pH 6.0 buffer with a constant ionic strength of 0.15 M (adjusted with NaCl). Initial rates for the enzyme-catalyzed hydrolysis of phosphate monoesters were measured by the production of inorganic phosphate using a colorimetric method described previously (11,12). Michaelis-Menten kinetic parameters were determined from a direct fit of the velocity versus substrate concentration data to the Michaelis-Menten equation using the nonlinear regression program GraFit (Erithacus Software).

Assay of Substrates 2, 13, and 17 as PTPase Inhibitors
The inhibition constants for 2, 13, and 17 were determined for the PTP1-catalyzed hydrolysis of p-nitrophenyl phosphate. At various fixed concentrations of inhibitors, the initial rates at various p-nitrophenyl phosphate concentrations were measured as described (6). The data were fit to equation 1 using KINETASYST (IntelliKinetics, State College, PA) to obtain the inhibition constant

RESULTS AND DISCUSSION
We have recently shown that PTPases will utilize a variety of non-naturally occurring aromatic and aliphatic phosphates as substrates (6, 7). For example, p-nitrophenyl phosphate, which is hydrolyzed by a variety of phosphatases, generally exhibits a K m in the mM range (13). Analogous aromatic and aliphatic phosphates display K m values that vary from 1 to 90 mM (7). Although we have found that the substrate efficacy of low molecular weight species can be dramatically improved by attaching them to active site-directed peptides (14), these "fusion" compounds still possess the principle disadvantage that inhibitor analogs are unlikely to serve as chemotherapeutic agents. With these features in mind, we investigated the active site substrate specificity of PTP1 with a structurally diverse array of low molecular weight aromatic phosphates. Several substrates examined in this study exhibit K m values in the low micromolar range and, as such, compare favorably with the very best peptide-based substrates ever reported for this enzyme (3).
p-Nitrophenyl phosphate 1 serves as a substrate for several PTPases. For example, it exhibits a K m of 1.7 mM with the Yersinia PTPase (3). With PTP1, we obtain a K m of 0.62 mM for the hydrolysis of 1 (Table I). For comparison, PTP1 hydrolyzes DADEpYLIPQQG with a K m of 2.6 M (3). Although the K m values of these two substrates are dramatically different, the k cat values are nearly identical (63.7 s Ϫ1 for 1 versus 75.7 s Ϫ1 for DADEpYLIPQQG). In general, the enhancement in catalytic efficiency for the peptide-based (versus p-nitrophenyl phosphate) PTP1 substrates is primarily a consequence of differences in K m . In spite of the fact that the relationship between K m and K d is often complex, it is tempting to ascribe the lower K m associated with peptide substrates to enhanced binding as a consequence of additional noncovalent interactions (outside of the active site) between enzyme and peptide.
Our initial survey of the active site substrate specificity of PTP1 focused on the relationship between aromatic substitution patterns and substrate efficacy. We prepared the ortho-, meta-, and para-ethyl-substituted phenyl phosphates 2, 3, and 4 ( Table I). The ortho derivative is a dramatically poorer substrate than its meta and para counterparts. This is largely due to a precipitous drop in the k cat term, which is somewhat reminiscent of differences that we previously observed in the PTP1-catalyzed hydrolysis of aromatic versus aliphatic phosphates (7). We found that the former exhibits k cat values that are several orders of magnitude greater than those displayed by the latter. This appears to be a reflection of different ratedetermining steps in the PTPase-catalyzed hydrolysis of these two different groups of substrates. In short, whereas the rate- limiting process for aromatic phosphates is breakdown (k 3 ) of the phosphoenzyme intermediate, the slow step for aliphatic phosphates appears to be formation (k 2 ) of this intermediate (Scheme 1). In the case of compound 2 the ortho-substituent may sterically impede the ready formation of the phosphoenzyme and thereby render the k 2 step rate-limiting as well. This could account for the dramatically reduced k cat value exhibited by this compound relative to those values displayed by compounds 3 and 4, where the ethyl moieties are positioned at a more distant site from the phosphate moiety. We also investigated the activity of the isopropyl derivatives 5 and 6. The k cat associated with the meta-substituted derivative is an order of magnitude less than that obtained with its para-substituted counterpart. Furthermore, a similar pattern is evident with the corresponding phenyl substituted derivatives 7 and 8. One likely explanation for these observations is that relatively large substituents, even at the meta position, can interfere with phosphoryl transfer to the enzyme. Such a sensitivity to steric effect at the meta position has also been noted for the nitro-and chloro-substitutions (6). However, the results with compound 7 are notable for an additional reason. The alkyl and aryl substituted derivatives 2-6, and 8 all exhibit K m values above 1 mM. In contrast, the K m associated with 7 is 320 M. Furthermore, the benzyl-containing derivative 9 exhibits a comparatively low K m as well. However, unlike 7, which is hydrolyzed with a low k cat , compound 9 enjoys a turnover rate that not only compares favorably with 1, but also with previously described peptide-based substrates (3). Due to the promising behavior of the para-benzyl derivative, we utilized this structural motif in several of the compounds described below.
We have previously shown that negatively charged residues, particularly on the NH 2 -terminal side of the phosphotyrosine moiety, substantially enhance the efficacy of PTP1 peptidebased substrates (3). Can appropriately positioned charged functionality on low molecular weight species likewise enhance substrate efficacy? Table II lists several derivatives of phenyl phosphate that contain a charged carboxylate moiety. The meta-substituted derivative 10 is an exceedingly poor substrate, as it exhibits both an elevated K m and a depressed k cat relative to p-nitrophenyl phosphate 1. In contrast, when the carboxylate functionality is situated at the para position (11), the k cat is virtually identical to that displayed by 1. The k cat /K m associated with the hydrolysis of these carboxyl-substituted derivatives improves somewhat when the -CO 2 Ϫ moiety (10 and 11) is replaced with a CH 2 CO 2 Ϫ substituent (12 and 13). Indeed, although the turnover rate associated with the hydrolysis of 13 is relatively modest, the K m (80 M) exhibited by this compound is comparatively impressive. Interestingly, when an additional methylene unit is inserted between the benzene and the carboxylate in 13 (to generate 14), compound 14 exhibits a 18.3and 3.5-fold increase in k cat and K m respectively, in comparison with 13. The PTPase-catalyzed hydrolysis of compound 15 is also characterized by a noteworthy K m (90 M). The relative orientation between the carboxylate moiety and the hydrolyzable phosphate group in phosphopodocarpic acid is not only fixed (16, see structure in Scheme 2) but also differs considerably from that found in 15. In addition, the former contains considerable steric bulk at the meta position as well, a feature which has a deleterious impact on the k cat terms associated with the hydrolysis of 5 and 7 as well. Not unexpectedly, phosphopodocarpic acid is a poorer substrate than 15, primarily due to a drop in the rate of turnover.
The relatively low k cat values exhibited by compounds 13 and 16 may provide an unforeseen opportunity in the design of novel phosphatase inhibitors. For example, it may be possible to create phosphate-bearing compounds that bind well to specific phosphatases, yet are hydrolyzed slowly or not at all. Indeed, work by Greengard and his colleagues offers some enticing evidence that at least one naturally occurring inhibitor may utilize this mechanism as a basis for its inhibitory activity. These investigators have demonstrated that the active form of the protein phosphatase-1 inhibitor, DARPP-32 (the dopamineand cAMP-regulated phosphoprotein) (15), is phosphorylated at an essential threonine residue. Synthetic peptides containing this phosphothreonine moiety are potent inhibitors of protein phosphatase-1 as well (16,17). In contrast, the corresponding nonphosphorylated peptides are poor inhibitors. In short, although the precise fashion by which the phosphate moiety on DARPP-32 interacts with the target phosphatase remains to be established, it is tempting to ascribe the inhibitory potency of this species to the ability of a key phosphothreonine residue to occupy the active site in a fashion that promotes binding but precludes ready hydrolysis.
We have also examined the hydrolytic efficacy of several diphosphate-bearing compounds. The Michaelis constant (16 M) associated with the hydrolysis of the diphenylmethane derivative 17 approaches those exhibited by the very best PTP1 peptide-based substrates (Table III). Indeed, 17 is the most potent low molecular weight substrate for PTPase that we have identified thus far. Nevertheless, the relatively low k cat is surprising, particularly given the fact that the closely related SCHEME 1. analog 9 exhibits a robust turnover rate. One possible explanation for these observations is that the additional phosphate moiety on 17 may enhance enzyme affinity, but at the expense of the ideal active site alignment required for rapid PTP1catalyzed hydrolysis. Alternatively, 17 (as well as 13, 15, and others) may be engaged in a combination of productive and nonproductive binding modes. In the latter case, 17 may bind to the active site, but in a fashion that precludes phosphate hydrolysis. Such nonproductive binding would lower both the k cat and the K m (18). The k cat would be lowered since only a fraction of the substrate would be productively bound when the enzyme is saturated. The K m would be lowered because the existence of additional binding modes must lead to apparently tighter binding. Such a mechanism (nonproductive binding) has also been proposed to account for the decreased k cat and K m values by a series of substituted anilides substrates for chymotrypsin (19). Finally, the low k cat values associated with 13, 15, and 17 may be due to the presence of negatively charged substituents that serve to render the release of the phenol product rate-limiting. Clearly, additional work is required to assess if and how specific structural features influence the interaction of substrates with the enzyme active site as well as the nature of the rate-determining step.
In addition to 17 and 18, we have found that the diphosphate of phenolphthalein (19) is a modest substrate (K m ϭ 360 M). There is little doubt that the second phosphate in 19 is a key contributor to the low K m since the corresponding phenolphthalein monophosphate (20) displays a Michaelis constant of 2.2 mM! Interestingly, 3,6-fluorescein diphosphate has also been shown to display K m values in the low M range toward a number of PTPases (20).
Finally, we have examined electron rich aromatic phosphates as PTP1 substrates. In general, the simple derivatives that we investigated in this study do not significantly differ, in terms of their substrate efficacy, from their more electron deficient counterparts. For example, the ortho-substituted derivatives 2 and 21 are both inefficiently hydrolyzed relative to the corresponding meta (3 and 22) and para (4 and 23) species (Table IV). In both instances, this is due to a dramatic drop in the rate of turnover associated with the ortho-modified compounds. However, the relatively small hydroxyl moiety (24) can be accommodated at the ortho site without dire consequences for the k cat term. Nevertheless, compounds 21-24 are modest substrates and all display K m values in the low millimolar range. Only the benzoylated derivative 25 exhibits a submillimolar Michaelis constant, behavior which is reminiscent of the benzylated phenylphosphate 9.
We, as well as other investigators, have previously demonstrated that protein phosphatases will catalyze the hydrolysis of low molecular weight, nonpeptidic, aromatic and aliphatic phosphates (6, 7, 21-24). However, these species are poor sub-strates relative to their peptidic counterparts, typically displaying K m values that are in excess of 1 mM. We have now established that low molecular weight aromatic phosphates can be hydrolyzed by PTP1 nearly as efficiently as full length peptide-based substrates. Indeed, compound 17 exhibits a very respectable K m of 16 M. Is this Michaelis constant an accurate assessment of the affinity of 17 for PTP1? Fortunately, several of the substrates described in this study exhibit relatively low turnover rates, which provides us with an opportunity to estimate their PTP1 affinity. In particular, we investigated the ability of 13 and 17 to serve as inhibitors for the PTPasecatalyzed hydrolysis of p-nitrophenyl phosphate. We found that both 13 and 17 act as purely competitive inhibitors for PTP1. The K m values associated with these two substrates appear to slightly underestimate their affinity for the enzyme. Both exhibit K i values that are somewhat less than their corresponding K m values (13, K m (80 Ϯ 18 M) and K i (5.8 Ϯ 0.1 M); 17, K m (16 Ϯ 3 M) and K i (4.9 Ϯ 0.7)). As a control, we assessed the inhibitory activity of compound 2, which displays a K m of 4.2 mM. As expected, species 2 is a comparatively poor inhibitor (K i ϭ 2.9 Ϯ 0.2 mM). In short, the K m values for 2, 13, and 17, appear to provide a reasonable indication of how well these compounds bind to PTP1. Most importantly, the extraordinary inhibitory efficacy of 13 and 17 augurs well for the design of nonpeptidic protein phosphatase inhibitors.
In summary, we have examined the active site substrate specificity of rat brain PTP1 with a diverse structural array of aromatic phosphates. The active site of this enzyme is sensitive to steric bulk positioned ortho to the phosphate moiety. Furthermore, the enzyme exhibits some sensitivity to certain sterically demanding residues at the meta site as well. In contrast, hydrophobic substituents, in conjunction with appropriately appended negatively charged residues, markedly enhance substrate efficacy. Although we have now established that low molecular weight compounds can serve as potent PTPase substrates, it still remains to be seen if all PTPases catalyze the hydrolysis of these simple aromatic phosphates with equal efficiency.