Acquisition of A Specific and Potent PTP1B Inhibitor from a Novel Combinatorial Library and Screening Procedure †

Protein tyrosine phosphatases (PTPases) form a large family of enzymes that serve as key regulatory components in signal transduction pathways. Defective or inappropriate regulation of PTPase activity leads to aberrant tyrosine phosphorylation, which contributes to the development of many human diseases including cancers and diabetes. For example, recent gene knockout studies in mice identify PTP1B as a promising target for anti-diabetes/obesity drug discovery. Thus, there is intense interest in obtaining specific and potent PTPase inhibitors for biological studies and pharmacological development. However, given the highly conserved nature of the PTPase active site, it is unclear whether selectivity in PTPase inhibition can be achieved. We describe a combinatorial approach that is designed to target both the active site and a unique peripheral site in PTP1B. Compounds that can simultaneously associate with both sites are expected to exhibit enhanced affinity and specificity. We also describe a novel affinity-based high-throughput assay procedure that can be used for PTPase inhibitor screening. The combinatorial library/high-throughput screen protocols furnished a small molecule PTP1B inhibitor that is both potent ( K i = 2.4 nM) and selective (little or no activity against a panel of phosphatases including Yersinia PTPase, SHP1, SHP2, LAR, HePTP, PTP α , CD45, VHR, MKP3, Cdc25A, Stp1, and PP2C). These results demonstrate that it is possible to acquire potent, yet highly selective inhibitors for individual members of the large PTPase family of enzymes. combinatorial library designed to target both the active site and an adjacent peripheral site in PTP1B. We also describe the development of an ELISA-based affinity selection procedure that was used to screen for potent PTP1B ligands. We have identified a highly potent PTP1B inhibitor (with a K i value of 2.4 nM) that exhibits several orders of magnitude selectivity in favor of PTP1B against a panel of PTPases. Our results demonstrate that it is feasible to achieve potency and selectivity for PTPase inhibition. rabbit anti-GST antibody and secondary horseradish peroxidase-conjugated mouse anti-rabbit IgG antibody.


4
The initiation, propagation, and termination of signaling events controlling many cellular processes are determined by the level of tyrosine phosphorylation. Phosphotyrosine level, in turn, is maintained in an exquisite balance by the reciprocal activities of protein-tyrosine kinases and protein-tyrosine phosphatases (PTPases). To date, a large number of PTPases has been identified. Because balanced protein tyrosine phosphorylation is critical for the maintenance of cellular homeostasis, it is not surprising that PTPase malfunction has been linked to many human diseases (1). Consequently, in those instances where PTPase activity is inappropriately high, PTPase inhibitors may provide a valuable new family of therapeutic agents. However, drug development targeted to PTPases was not seriously considered until recently. A major concern is that a PTPase may regulate multiple signaling pathways while at the same time, a single pathway may be controlled by several PTPases. Thus, PTPase inhibition may give rise to unwanted side effects. Significant progress has been made that is beginning to alleviate this concern.
For example, PTP1B has been suggested as a negative regulator of insulin signaling (2)(3)(4). In addition to a role in insulin signaling, PTP1B is overexpressed in association with the expression of p185 c-erbB-2 in human breast and ovarian cancers (5,6), and PTP1B is capable of suppressing transformation by Neu (7) and v-Src (8). Recently, PTP1B has been identified as the major PTPase that dephosphorylates and activates c-Src in several human breast cancer cell lines 7 in water for 2 hr. The resin was removed by filtration, and the remaining solution concentrated.
Dry diethyl ether was added and the precipitated peptides collected by centrifugation. The peptides were resuspended, washed twice with ether, dissolved in water, and purified by semipreparative reverse phase HPLC. All peptides were obtained in high purity (>95%) as analyzed by MALDI-TOF MS and analytical HPLC.
Synthesis of PTP1B Ligand Library. The library was synthesized on a cystaminemodified Tentagel S NH 2 resin 1 using Fmoc chemistry (25). pTyr was attached to the amino terminus of the resin-linked cystamine (8 g). After Fmoc removal by two 5 min treatments with 30% piperidine in DMF, the resin was washed with DMF, CH 2 Cl 2 , isopropanol, and ether, and then the residual solvent removed in vacuo. The resin was distributed in 220 mg quantities into 20 mL polypropylene filtration tubes (Supelco) for coupling of the next component. The linking diversity elements 4 -25 (Fig. 2) were incorporated (except for the absence of a diversity element 26) into the library in the Fmoc-protected form, which were either commercially available or prepared by treatment of commercially available amino acids with Fmoc-Osu in 1:1 THF/10% Na 2 CO 3 . Coupling was accomplished by one 2 hr and one 15 hr treatments with 6 eq. of the amino acid, 6 eq. of PyBOP, 6 eq. of HOBt, and 12 eq. of NMM in 4 mL DMF. The phosphate group of pTyr used in the library synthesis was mono-benzyl ester protected, and the acid side chains of Asp and Glu t-butyl ester protected. The N-terminal Fmoc group was deprotected by two 5 min treatments with 30% piperidine in DMF. The resin was then washed with DMF, CH 2 Cl 2 , isopropanol, and ether, and the residual solvent removed in vacuo. The coupling and deprotection steps were monitored by examination of free amine substitution level or Fmoc release during the course of the library synthesis until the coupling of the terminal diversity elements. The resin from each filtration tube was then distributed in 5.0 mg quantities 8 treatment with phosphoryl chloride in pyridine (26) followed by basic hydrolysis. 2,2′bipyridine-4,4′-diacid was prepared from 4,4′-dimethyl-2,2′-bipyridine (GFS Chemicals) by treatment with KMnO 4 in 25% H 2 SO 4 (27). Upon completion of the solid-phase assembly, side chain deprotection was accomplished by two 1 hr treatments with 90% TFA and 5% phenol in water. The resulting resin 3 was then washed extensively with CH 2 Cl 2 , DMF, MeOH, and H 2 O before treatment with 10 mM DTT in 500 µL 50 mM Tris buffer (pH 8.0) for 3 hr. Finally the solution phase was filtered into the 96-well receiving plate to afford the spatially separated library members 3 at a concentration of 0.1 mM (assuming complete conversion for each member). Several library members were resynthesized on larger scale using the same procedure in high yield and purity (about 90%) as assessed by HPLC and MOLDI-TOF MS analysis. These Resynthesis of Selected High-Affinity PTP1B Ligands. Several high-affinity members of the library were selected based on the initial ELISA screening results, and their analogs without a thiol tail were synthesized on Rink resin according to the above peptide synthesis procedure.
These compounds were again subjected to the ELISA evaluation and the highest-affinity compound 21B was synthesized on large scale. 1 (29). Silver tetrafluoroborate (28) (2.3 g, 11.8 mmol) was dissolved in dry DMSO (10 mL) and a solution of benzyl 4-(bromomethyl)phenylacetate 28 (3.0 g, 9.4 mmol) in dry DMSO (10 mL) was slowly added. The mixture was stirred at room temperature for 12 hr and then triethylamine (2 mL) was added. The mixture was kept for additional 15 min and then subjected to CH 2 Cl 2 /water extraction. The organic phase was concentrated via rotary evaporation and purified by flash column chromatography to afford a white solid 29 (1.96 g, 82%). 1  in dry CH 2 Cl 2 (5 mL), 260 mg activated MnO 2 (85%, 2.5 mmol) was added in one portion. The mixture was stirred for 24 hr and then filtered through acid-washed silica gel. The filtrate was rotary evaporated and dried in vacuo to afford the ketophosphonate intermediate as a colorless oil. Without further purification, the oil was chilled to 0 °C and 1 mL DAST (7.5 mmol) added dropwise. The solution was stirred at room temperature for 6 hr and then diluted by 10 mL CH 2 Cl 2 . The resulting solution was added slowly to 15 mL saturated Na 2 CO 3 soluion at 0 °C.     (38). A solution of the amino acid 37 (535 mg, 1.5 mmol) and NaHCO 3 (128 mg, 1.5 mmol) in water (5 mL) and
The coding sequence of the catalytic domain (amino acid residues 1-288) of the human T cell PTPase (TCPTP) was a generous gift from Dr. Harry Charbonneau and TCPTP was expressed and purified as described (38). Recombinant HePTP and the catalytic domains of SHP1 and SHP2 were expressed and purified as (His) 6

RESULTS AND DISCUSSION
As noted in the Introduction, biochemical and genetic studies suggest that PTP1B is a major modulator of insulin sensitivity and fuel metabolism. Thus PTP1B represents a potential therapeutic target for the treatment of Type II diabetes and obesity. Consequently, small molecules designed to inhibit PTP1B not only hold promise as pharmaceutical agents but also could function as probes for elucidating the roles of PTP1B in specific intracellular pathways involved in normal cellular processes. However, given the highly conserved nature of the PTPase active site (i.e., pTyr binding site), it has been assumed that it would be difficult to obtain specific inhibitors targeted to the active site of individual PTPases.
Our previous kinetic studies with pTyr-containing peptides and small molecule aryl phosphates showed that pTyr alone is not sufficient for high affinity binding by PTPases and residues surrounding the pTyr contribute to efficient substrate recognition (24,26,(40)(41)(42). In inhibitor that is able to simultaneously occupy both the active site and a unique second site on PTP1B.
Library Design and Construction. Our first-generation library was designed to contain two linked motifs, one targeted to the pTyr-binding catalytic site, and the other targeted to a unique adjacent noncatalytic site in PTP1B. Due to the demanding synthetic requirements associated with the preparation of nonhydrolyzable phosphonate analogs (vide infra), we felt it prudent to prepare a library of synthetically accessible phosphate-based derivatives. Once a high affinity lead from the latter is identified, it can then be converted into an inhibitor by replacing the phosphate moiety with a difluorophosphonate group. Since pTyr is the canonical ligand for PTPase active site, we decided to structurally bias the library with pTyr in order to direct library members to the active site. A small array of structurally disparate aryl acids (A -H) (Fig. 1) were chosen and linked to pTyr in order to access binding interactions removed from the active site.
These aryl acids include three phenylphosphate-containing species (A -C), three phenolcontaining species (D -F), and two additional aromatic species (G -H). Members of the aryl acid array were separately linked to pTyr either directly (26) or via twenty-two different amino acids (4 -25) (Fig. 2), which include nine linear aliphatic species (4 -10, 15, 23), eleven ringcontaining species (11 -14, 16 -20, 24 -25), and two natural acidic amino acids (21 -22).   (43) and that the PTP1B/C215S mutant exhibits similar affinity for substrates as the wild-type enzyme (44). We have also shown that the hexameric pTyr-containing peptide DADEpYL-amide is a high affinity PTP1B substrate (41,44). We prepared the biotinyl-caproic acid-DADEpYL-NH 2 peptide and found that it displayed kinetic parameters similar to those reported for the DADEpYL-NH 2 peptide with the wild-type PTP1B (data not shown). Thus, in this assay the binding affinity of the library members was assessed by their ability to compete with the biotinylated phosphopeptide for binding to PTP1B/C215S. There are several key points to be noted concerning the ELISA-based assay. First, since the reference ligand (biotinylated DADEpYL-NH 2 ) is known to bind to the PTP1B/C215S active site (18), compounds that displace the reference ligand from PTP1B/C215S most likely bind to the active site as well. Second, since the catalytically inactive PTP1B/C215S binds ligands with equal potency as the wild-type enzyme, this assay furnishes a true assessment of the PTP1B binding ability of the library members. Third, it is known that the invariant active site Cys residue is essential for PTPase catalytic activity (19). Consequently, PTPases are prone to inactivation by oxidizing reagents, heavy metals, and alkylating compounds. This has presented a serious problem for the PTPase activity-based inhibitor screening projects in which hits are identified based on the ability of the compounds to reduce the PTPase activity. The substitution of the active site Cys by a Ser (e.g., PTP1B/C215S) renders the mutant PTPase less sensitive to oxidation and alkylation and thus will likely eliminate "false" positives due to interactions with the active site Cys that destroy the phosphatase activity. Finally, since the assay is ELISA-based, it can be easily implemented for high-throughput PTPase inhibitor discovery. Finally, PTP1B is clearly quite sensitive to the structural nature of the N-terminal element given the fact that closely related elements (A and C) which differ by a single methylene group are less effective than the lead B.

Identification of High
In order to obtain a more accurate assessment of the affinity of these compounds for PTP1B/C215S, we measured the IC 50 values (compound concentrations that block 50% of the ELISA readout at 450 nm) of the lead compounds (21B and 24B) using 39 as a reference (Fig.   4). For comparison, we also measured the IC 50 values of compounds 4A and 4B, which were less effective than 21B and 24B in displacing biotinylated DADEpYL-NH 2 from PTP1B/C215S (Fig.   3). To avoid potential problems associated with the possible oxidation of the thiol tail in the library compounds, we resynthesized compounds 4A, 21B, and 24B without the thiol tail. Table   1 lists the ratio of the IC 50 values of the test compounds relative to that of the reference compound 39. Since 39 is an established competitive inhibitor for PTP1B with a K i value of 1 µM (39), this IC 50 ratio should reflect the true affinity of the test compounds for PTP1B (in units of µM). As can be seen from Table 1, the presence of the thiol tail in the compounds does not affect the affinity of these compounds for PTP1B/C215S. It can be concluded that compounds 21B and 24B display binding affinities significantly higher than that of 39. In addition, compounds 21B and 24B also exhibit higher affinity for PTP1B than that of 4A and 4B, consistent with the ELISA results obtained at a single compound concentration (250 nM) (Fig.   3). Finally, although PTP1B can accommodate both Tyr (24) and Asp (21) at the P-1 position (20,24), it appears that in the context of the terminal element B, the linker Asp (21) is slightly favored over Tyr (24).  (44,45,47). This has been attributed to a direct interaction between the fluorine atoms and PTP1B active site residues (47). Thus we decided to replace the ester oxygens in 21B with the difluoromethylene group.
The corresponding nonhydrolyzable analog (40, Fig. 4 (39,(48)(49)(50)(51)(52), achieving selectivity, particularly between PTP1B and TCPTP, has been a considerable challenge. As shown in Table   2, compound 40 is highly selective for PTP1B, exhibiting a greater than three orders of magnitude preference for PTP1B versus nearly all phosphatases examined. More importantly, compound 40 also displays >10-fold selectivity in favor of PTP1B over TCPTP, which is the closest structural homologue of PTP1B (the catalytic domain of PTP1B (residues 1 -279) is 69% identical and 85% homologous to that of TCPTP). The high selectivity that is observed for compound 40 without any further optimization is quite impressive, considering the general lack of selectivity that has been observed for inhibitors of the PTPase family members. These results demonstrate that it is possible to achieve both potency and selectivity in PTPase inhibitor development.     Scheme III .
Synthesis of the difluorophosphonate-containing unnatural amino acid 38.    Scheme III