Potent Benzimidazole Sulfonamide Protein Tyrosine Phosphatase 1B

Jun 8, 2006 - Synthesis of substituted 2H-benzo[e]indazole-9-carboxylate as a potent antihyperglycemic agent that may act through IRS-1, Akt and GSK-3...
0 downloads 11 Views 3MB Size
3774

J. Med. Chem. 2006, 49, 3774-3789

Articles Potent Benzimidazole Sulfonamide Protein Tyrosine Phosphatase 1B Inhibitors Containing the Heterocyclic (S)-Isothiazolidinone Phosphotyrosine Mimetic Andrew P. Combs,*,† Wenyu Zhu,† Matthew L. Crawley,† Brian Glass,† Padmaja Polam,† Richard B. Sparks,† Dilip Modi,† Amy Takvorian,† Erin McLaughlin,† Eddy W. Yue,† Zelda Wasserman,† Michael Bower,† Min Wei,‡ Mark Rupar,‡ Paul J. Ala,‡ Brian M. Reid,‡ Dawn Ellis,‡ Lucie Gonneville,‡ Thomas Emm,§ Nancy Taylor,§ Swamy Yeleswaram,§ Yanlong Li,‡ Richard Wynn,‡ Timothy C. Burn,‡ Gregory Hollis,‡ Phillip C. C. Liu,‡ and Brian Metcalf† Incyte Corporation, DiscoVery Chemistry, Applied Technology, and Drug Metabolism, Experimental Station, Route 141 and Henry Clay Road, Wilmington, Delaware 19880 ReceiVed January 26, 2006

Potent nonpeptidic benzimidazole sulfonamide inhibitors of protein tyrosine phosphatase 1B (PTP1B) were derived from the optimization of a tripeptide containing the novel (S)-isothiazolidinone ((S)-IZD) phosphotyrosine (pTyr) mimetic. An X-ray cocrystal structure of inhibitor 46/PTP1B at 1.8 Å resolution demonstrated that the benzimidazole sulfonamides form a bidentate H bond to Asp48 as designed, although the aryl group of the sulfonamide unexpectedly interacts intramolecularly in a pi-stacking manner with the benzimidazole. The ortho substitution to the (S)-IZD on the aryl ring afforded low nanomolar enzyme inhibitors of PTP1B that also displayed low caco-2 permeability and cellular activity in an insulin receptor (IR) phosphorylation assay and an Akt phosphorylation assay. The design, synthesis, and SAR of this novel series of benzimidazole sulfonamide containing (S)-IZD inhibitors of PTP1B are presented herein. Introduction Protein tyrosine phosphatase 1B (PTP1Ba), the prototypical protein tyrosine phosphatase (PTP) from a family of ∼112 intracellular PTPs, has been strongly implicated by in vitro and in vivo experiments as a negative regulator of both insulin and leptin signal transduction pathways.1 The observation that PTP1B knockout mice exhibit increased insulin sensitivity and are highly resistant to weight gain upon high fat feeding strongly suggests that PTP1B is a viable drug target for the treatment of diabetes and obesity.2,3 Thus, a potent cell permeable inhibitor of PTP1B has been sought by nearly all pharmaceutical companies to test this hypothesis.4 Medicinal chemistry efforts to date have focused on the identification of inhibitors that contain strong binding, nonhydrolysable tyrosine phosphonate (pTyr) mimetics to affect potent binding to PTP1B. A variety of pTyr mimetics have been discovered and incorporated into potent small molecule inhibitors of PTP1B. Unfortunately, they lack cell permeability and oral bioavailability because of the strong negative charge carried by the pTyr mimetics as well as their high molecular weight and/or peptidic nature. We recently disclosed the structure-based design of novel isothiazolidinone (IZD) pTyr mimetics and their incorporation into the peptidic inhibitors of PTP1B.5 A thorough evaluation of the structure activity relationships of various pTyr heterocyclic mimetics and the peptidic portion of these inhibitors is * To whom correspondence should be addressed. Tel: 302-498-6832. Fax: 302-425-2750. E-mail: [email protected]. † Discovery Chemistry. ‡ Applied Technology. § Drug Metabolism. a Abbreviations: PTP1B, protein tyrosine phosphatase 1B; IR, insulin receptor; pTyr, phosphotyrosine; (S)-IZD, (S)-isothiazolidinone; TZD, thiadiazolidinone; DFMP, difluoromethylphosphonate.

described elsewhere.6 In summary, we determined that the (S)IZD heterocycle is the most potent pTyr mimetic known to date, being ∼10-fold more potent than the difluoromethylphosphonate (DFMP) when incorporated into dipeptide inhibitors and 5-fold more potent than the analogous thiadiazolidinone (TDZ) heterocyclic pTyr mimetic. However, despite the significant improvement in the enzyme potency of these peptidic (S)-IZD inhibitors, they also lacked cellular activity because of poor permeability. We rationalized that the diffusely monoanionic (S)-IZD pTyr mimetic was not solely responsible for the lack of permeability and that the reduction in the peptidic nature of the inhibitors might cause a net improvment in permeability. The C-terminal primary amide was replaced with a benzimidazole with a 20fold improvement in potency. The N-terminal truncation of the peptide followed by parallel synthesis and purification of N-acylated libraries identified the sulfonamide as a potent replacement for the two terminal peptides. The introduction of ortho substituents on the aromatic ring to the (S)-IZD provided relatively low molecular weight (MW ∼450-650) PTP1B inhibitors with low nanomolar enzyme potency. The orthomethyl benzimidazole sulfonamide derivative 79a has exceptional enzyme potency and was shown to be cell active in an IR phosphorylation assay as well as in an Akt phosphorylation assay. Details of the synthesis and SAR are presented herein. Chemistry. The synthesis of peptides and benzimidazole sulfonamide derivatives incorporating the (S)-IZD pTyr mimetic followed our route previously reported to derive peptide derivative 1 starting with commercially available phenylalanine boronic acid derivative 3 (Scheme 1).5 Acid 3 was coupled with chloroisothiazolinone 4 in a Suzuki7 cross-coupling to generate oxidized core 5 in moderate yield. The reduction-deprotection of 5 using palladium on carbon under 50 psi of hydrogen gas gave 6 in high yield. The coupling of acid 6 with benzenedi-

10.1021/jm0600904 CCC: $33.50 © 2006 American Chemical Society Published on Web 06/08/2006

Protein Tyrosine Phosphatase 1B Inhibitors

Journal of Medicinal Chemistry, 2006, Vol. 49, No. 13 3775

Scheme 1. Synthesis of Benzimidazole-Containing (S)-IZD Analoguesa

a (a) 16 mol % Pd(dppf)Cl , 3 equiv Et N, DCM, 90 οC, 16 h, 68%; (b) 10 wt % Pd/C, EtOH, 93%; (c) 7, HATU, DIEA, rt, 16 h, 48%; For compds 9, 2 3 10, 12, 14, and 16, chiral separation of (R/S)-IZD diastereomers was performed; (d) AcOH, 40 οC, 2 h; (e) TFA, 30 min, rt, (83% for 2 steps); (f) Boc-Phe, HATU, DIEA, rt, 2 h; (g) TFA, 30 min, rt, (h) Ac2O, DIEA, DCM, 2 h, rt (80% for 2 steps); (i) MsOH, ACN, rt, 60% (j) R3Cl, Et3N, DCM, rt, 5 h; (k) TFA, µW, 1 min (20-80% for 2 steps).

Scheme 2. Synthesis of ortho-Fluoro Substituted (S)-IZD-Containing Analoguesa

a (a) 10 mol % Pd(dppf)Cl , 4, 3 equiv Et N, toluene, 90 οC, 16 h, 61%; (b) 10% Pd/C, EtOH, rt, 16 h, 82%; (c) NBS, benzoyl peroxide, CCl , 80 οC, 2 3 4 4 h, 67%; (d) AgNO3, EtOH, H2O, reflux, 1 h, 85%; (e) DBU, 61, CH2Cl2, 74%; (f) 0.5 mol % R,R-(-)-1,2-bis(o-methoxyphenyl)(phenyl)phosphinoethan(1,5cyclooctadiene rhodium; (I), EtOH, 50 psi H2, 12 h, 93%, 97% ee, 1:1 dr; (g) Chiralcel AD column, 47%, 97% ee, single diastereomer; (h) TFA, 130 οC, µW, 6 min, 74%; (i) Boc2O, Et3N, CH2Cl2, 97%; (j) LiOH, H2O, MeOH, 99%; (k) BOPCl, 7, iPr2NEt, DMF, rt, 1 h; (l) AcOH, 40 οC, 1.5 h; (m) TFA/ CH2Cl2 ) 1:5; (n) Et3N, 66, CH2Cl2.

amine 7 followed by cyclization under acidic conditions at room temperature for 24 h afforded the desired benzimidazole. Careful

control of the reaction temperature was critical because ring closure at higher temperatures proceeded more rapidly but

3776

Journal of Medicinal Chemistry, 2006, Vol. 49, No. 13

Combs et al.

Scheme 3. Synthesis of ortho-Methyl Substituted Derivativesa

a (a) LiBH , THF, rt, 2 days, 82%; (b) DMSO, oxalyl chloride, CH Cl , -78 οC, 96%; (c) DBU, 70, CH Cl , 85%; (d) 0.5 mol % R,R-(-)-1,2-bis(o4 2 2 2 2 methoxyphenyl)(phenyl)phosphinoethan(1,5-cyclooctadiene rhodium (I), EtOH, 50 psi H2, 12 h, 83%, 98% ee; (e) Pd(OAc)2, Et3N, 4,4,5,5-tetramethyl1,3,2-dioxaborolane, 91% (f) NaIO4, NH4OAc, THF, H2O, rt, 18 h, 80%; (g) 15 mol % Pd(dppf)Cl2, 4, 3 equiv Et3N, toluene, 80 οC, 24 h, 62%; (h) L-selectride, THF, -78 οC, 88%; (i) TFA, 130 οC, µW, 6 min, 74%; (j) BnOC(O)Cl, Et3N, CH2Cl2; (k) LiOH, H2O, MeOH; (l) BOPCl, iPr2NEt, 7, DMF, rt, 1 h, 82%; (m) AcOH, 40 οC, 1.5 h; (n) 2.2 equiv SEMCl, iPr2NEt, CH2Cl2; (o) Chiralcel AD column, 98% ee, single diastereomer; (p) 10 wt % Pd/C, MeOH, H2; (q) Et3N, 66, CH2Cl2; (r) TFA, 130 οC, µW, 6 min;

returned mixtures of diastereomers at the alpha center of the amino acid. At this point in the synthesis, the (R/S)-IZD diastereomers were easily separated by chiral HPLC to afford two discrete isomers, (S)-IZD and diastereomer 8 is shown in Scheme 1. Both diastereomers were further elaborated to the final compounds, providing an active and inactive pair. The active isomer was assigned to the (S)-IZD on the basis that all X-ray cocrystal structures solved to date have the (S)-IZD isomer bound, such as 46/PTP1B. All other analogues were derived from diastereomer 8. The removal of the Boc group from 8 with TFA furnished the free amine, which could be acylated with a variety of reagents under mild conditions to give amides, ureas, sulfonamides, and carbamates such as 10-56. The synthesis of each desired analogue ortho substituted on the aryl ring to the (S)-IZD necessitated the optimization of a unique sequence of chemical reactions. The ortho-fluoro and ortho-methyl derivatives were synthesized via a similar Suzuki cross-coupling reaction of chloroisothiazolinone 4 and a properly substituted arylboronic acid. The construction of the orthobromo and ortho-chloro derivatives relied on a chemoselective and regioselective Heck coupling of an ortho substituted arylidodide with isothiazolidinone 82. The separation of (R/S)IZD isomers was also necessary in these routes, and the (S)IZD isomer was again assigned by synthesis of the two (R/S)-

IZD diastereomers and the determination of the active compound. The details of each sequence of reactions are described below. The ortho-fluoro derivatives were synthesized via a similar Suzuki cross-coupling approach, though accessing the key amino acid-containing (S)-IZD intermediate 64 was more challenging. Starting with commercially available fluoroboronic acid 57 and chloro-isothiazolinone heterocycle 4, palladium catalyzed crosscoupling employing dppf as the ligand gave a 61% yield of core 58 (Scheme 2). The reduction of heterocyclic olefin followed by the formation of the dibromo acetal equivalent 59 proceeded in 51% yield for two steps. Aldehyde 60 was unmasked using silver nitrite in refluxing aqueous ethanol in 85% yield. A Horner-Wadsworth-Emmons type olefination with phosphonate 61 afforded unsaturated amino acid 62. The reduction of 62 utilizing R,R-(-)-1,2-bis(o-methoxyphenyl)(phenyl)phosphinoethan(1,5-cyclooctadiene) rhodium8 proceeded with >97% ee to give a 1:1 mixture of diastereomers at the heterocycle, which were separated on a Chiral AD HPLC column. Single isomer 63 was then globally deprotected by microwave irradiation in TFA to afford the free amine. Classical thermal conditions gave significantly lower yields in this deprotection sequence for most derivatives. Reprotection of the amine as the Boc carbamate and saponification of the ester generated the key amino acid containing (S)-IZD intermediate

Protein Tyrosine Phosphatase 1B Inhibitors

Journal of Medicinal Chemistry, 2006, Vol. 49, No. 13 3777

Scheme 4. Synthesis of ortho-Chloro Substituted (S)-IZD-Containing Analoguesa

a (a) Zn0, NH Cl, MeOH, H O; (b) NCS, DMF; (c) NaNO , 1 N aq HCl, KI, 51%) (3 steps); (d) Pd(OAc) , Bu NCl, 81, Et N, DMF, 100 οC, 64%; (e) 4 2 2 2 4 3 LiBH4, MeOH, 0 οC, 72%; (f) Chiralcel AD column, 48%, 97% ee, single diastereomer; (g) TFA, 130 οC, µW, 0.25 h, 88%; (h) Boc2O, Et3N, CH2Cl2; (i) LiOH, H2O, MeOH, 81% (2 steps); (j) BOPCl, 7, iPr2NEt, DMF, rt, 1 h (k) AcOH, 40 οC, 1.5 h, 72% (2 steps); (l) TFA/CH2Cl2 ) 1:5; (m) Et3N, 66, CH2Cl2, 76% (2 steps).

Scheme 5. Synthesis of ortho-Bromo Substituted (S)-IZD-Containing Analoguesa

a (a) Zn0, NH Cl, MeOH, H O; (b) NBS, DMF; (c) NaNO , 1 N aq. HCl, KI 64% (3 steps); (d) Pd(OAc) , Bu NCl, 82, Et N, DMF, 100 οC, 52%; (e) 4 2 2 2 4 3 LiBH4, MeOH, 0 οC, 82%; (f) TFA, 130 οC, µW, 0.25 h, 99%; (g) CbzCl, Et3N, CH2Cl2; (h) LiOH, H2O, MeOH, 81% (2 steps); (i) BOPCl, iPr2NEt, DMF, ο rt, 1 h (j) AcOH, 40 C, 1.5 h, 72% (2 steps); (k)TMSI, CH3CN; (l) Et3N, 66, CH2Cl2, 68% (2 steps).

64. Peptide coupling with diamine 7 followed by in situ cyclization in acetic acid afforded benzimidazole 65. The N-terminal amine of the benzimidazole derivatives were deprotected with TFA and reacted with various arylsulfonyl chlorides 66 to furnish benzimidazole sulfonamide derivatives 67 (a-e). Synthesis of the ortho-methyl scaffold also employed a Suzuki cross-coupling to attach the IZD to the scaffold (Scheme

3). However, it was again necessary to reconstruct the overall synthetic route compared to those of the unsubstituted and orthofluoro syntheses. Commercial bromide 68 was converted in 80% yield over two steps to aldehyde 69 via a reduction-oxidation sequence. Olefination of the aldehyde with phosphonate ester 70 afforded unsaturated amino acid 71 in 85% yield. The reduction of 71 utilizing R,R-(-)-1,2-bis(o-methoxyphenyl)-

3778

Journal of Medicinal Chemistry, 2006, Vol. 49, No. 13

Combs et al.

Figure 2. Peptidomimetic approach to optimizing peptide-(S)-IZD inhibitors. Table 1. SAR of the N-terminal Truncation of Peptide Benzimidazoles

Figure 1. X-ray crystal structure of compound 1 bound to PTP1B. The molecular surface of the protein is shown with the (S)-IZD heterocycle binding in the phosphotyrosine binding pocket. The bidentate hydrogen bond of the inhibitor’s peptide-like backbone with Asp48, also shown, is indicated with dotted lines.

(phenyl)phosphinoethan(1,5-cyclooctadiene) rhodium proceeded with >97% ee to give 72. Bromide 72 was converted to boronic acid 73 via coupling to the borate ester followed by oxidative cleavage. Boronic acid 73 was cross-coupled with chloroisothiazolinone heterocycle 4 employing the same conditions utilized for the ortho-fluoro derivative to give IZD scaffold 74. The reduction of the heterocycle with sodium borohydride and subsequent global deprotection furnished amine salt 75 in 66% yield. Reprotection of the amine as the Cbz carbamate followed by saponification afforded acid 76 in quantitative yield. The coupling of acid 76 with diamine 7 and then in situ cyclization with acetic acid provide the desired benzimidazoles 77. At this point, the 1:1 diastereomeric mixture at the heterocycle of 77 was separated utilizing a Chiralcel AD HPLC column. The single diastereomer was then bis-SEM protected to supply intermediate 78. The removal of the Cbz group using hydrogenation followed by coupling with arylsulfonyl chlorides 66 and subsequent deprotection with TFA afforded final products 79. The synthesis of the ortho halogenated (S)-IZD derivatives (chloro and bromo) could not be affected by a Suzuki crosscoupling because of chemoselectivity issues. A novel synthetic strategy was devised to couple the IZD to ortho-halo iodophenyl derivatives 81 and 88 through chemoselective and regioselective Heck reactions (Schemes 4 and 5). The synthesis of ortho-chloro derivative 87 is outlined below. The synthesis of ortho-bromo derivative 93 was performed as shown in Scheme 5 with only a slight modification in the protecting group strategy. Key chloro-iodo-derivative 81 was readily obtained from commercial nitro-phenylalanine 80. The nitro group of 80 was reduced to an amino group using zinc and ammonium chloride in methanol, followed by ortho chlorination to give an intermediate 2-chloroaniline. The aniline was diazatized using sodium nitrite under acidic conditions and the resulting intermediate trapped with potassium iodide to afford 81 in 51% yield over three steps. This intermediate was used in the key Heck9 reaction with heterocycle 82. The halogenated core 81 was coupled with 82 in 64% yield with classical ligandless conditions utilizing palladium acetate, tetra-n-butylammonium chloride, and triethylamine in DMF.10 The reduction of 83 using lithium borohydride followed by the separation of the diastereomers utilizing chiral HPLC afforded single enantiomer 84. Global deprotection using trifluoroacetic acid in the microwave at 130 °C, followed by Boc reprotection of the amine with di-tertbutyl dicarbonate and then saponification of the ester supplied

a

The pNPP assay. b The mixture of (R/S)-IZD diastereomers.

acid 85 in 71% yield over three steps. The coupling of benzenediamine 7 to scaffold 85 and cyclization under acidic conditions gave benzimidazole 86 in 72% yield. The removal of the carbamate with TFA followed by sulfonylation afforded the desired ortho-chloro benzimidazole sulfonamide 87 in 76% yield. Results and Discussion In our effort to improve the cell permeability and the overall pharmacokinetic properties of our peptidic (S)-IZD-containing inhibitors of PTP1B, we endeavored to eliminate the peptidic nature of lead compound 1. We initially focused on replacing the C-terminal amides with less polar, more hydrophobic groups. An inspection of the X-ray structures of several inhibitors, including 1, revealed a conserved bidentate hydrogen bond between the C-terminal amide NH and the adjacent amide NH with Asp48 of PTP1B (Figure 1). Previous analogs from our lab had also shown that N-terminal tertiary amides were completely inactive against PTP1B, corroborating that this is an important interaction. Several heterocycles 2 were thus

Protein Tyrosine Phosphatase 1B Inhibitors

Journal of Medicinal Chemistry, 2006, Vol. 49, No. 13 3779

Table 2. SAR of Arylsulfonamide Substituents

Table 3. SAR of Benzimidazole Substituents

PTP1B pNPP

R2 compdb 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 a

2-

3-

4-

5-

6-

F F F Cl Cl Cl Br Br Br Me Me Me CF3 CF3 CF3 CN CN OMe OMe Ph Cl Cl Cl Cl

Cl Cl Cl Cl Cl Cl

F F F

Cl Cl F F F

F F

F F

IC50 (nM)a 100 (32)c 120 66 (43)c 80 160 60 110 170 77 110 300 85 80 130 67 51 45 63 48 94 80 240 280 160 460 79 51 110 140 160 110 72

b

A pNPP assay. The compounds were synthesized in parallel as a mixture of 4 diastereomers (two major: (R/S)-IZD and two minor: alphacenter of amino acid) and purified by preparative LCMS. c The pNPP data for the separated single diastereomer.

targeted, which could maintain the bidentate hydrogen bonding interaction with Asp48 (Figure 2). The benzimidazole moiety proved to be a very effective replacement for the amide. Benzimidazole peptide 10 was ∼10-fold more potent than the primary amide 1 with an IC50 ) 35 nM (Table 1). The mode of inhibition experiments unequivocally demonstrated that 10 was a competitive and reversible inhibitor of PTP1B. The truncation of the N-terminus of peptide inhibitor 10 demonstrated that the peptide portion of the inhibitor significantly contributed to the binding of this molecule to PTP1B (Table 1). The elimination of the terminal acetamide in compound 11 gave a surprising >20-fold loss in activity. It is clear that the hydrogen bond between the amide carbonyl and Arg54 provides a significant amount of the binding energy observed with these peptidic ligands. Truncating further to the simple acylbenzimidazole derivative 12 resulted in an additional 2-fold loss in activity, suggesting that the hydrophobic phenyl substituent of the peptide plays only a minor role in binding to PTP1B. The desolvation of the hydrophobic side chain in this solvent exposed region is likely a contributing factor. A diverse compound library of different acyl groups (amides,

compdb

R1

PTP1B pNPP IC50 (nM)a

34 47 48 49 50 51 52 53 54 55 56c

H 4-Me 5-Me 5-F 5-CN 5-Cl 5-CO2Me 5-CF3 5-OMe 4-OH 4-tert-butyl

80 190 140 120 53 140 150 160 130 240 460

a The pNPP assay. b The compounds were synthesized in parallel as a mixture of 4 diastereomers (two major: (R/S)-IZD and two minor: alphacenter of amino acid) and purified by preparative LCMS. c The compound was submitted as a mixture of 2 diastereomers ((R/S)-IZD).

ureas, carbamates, sulfonamides) was synthesized using scaffold 8, as a mixture of diastereomers, in an effort to identify replacements for the N-terminal peptide portion of the inhibitor. Gratifyingly, arylsulfonamides were rapidly identified as potent PTP1B inhibitors. The new benzimidazole sulfonamide lead 14 is nonpeptidic and lower in molecular weight than parent 10 yet has comparable potency (IC50 ) 32 nM). The aryl functionality was important for potency since the simple methylsulfonamide 13 was 5-fold less active. A focused compound library of substituted arylsulfonamides was synthesized in parallel and purified by automated LCMS11,12 to optimize this functionality. A partial, yet representative set of compounds is shown Table 2. The SAR from this study revealed that a variety of meta or para substituted derivatives were more potent than the ortho substituted compounds, although they are only 2-3-fold better than the unsubstituted parent 14. A small library of substituted benzimidazoles was synthesized and showed that a wide variety of functional groups could also be appended to the benzimidazole (Table 3). All 3-substituted derivatives were approximately equipotent against PTP1B, whereas the large tert-butyl substituent 56 at the 2-position was less active. Modeling predicted that the benzimidazole would bind in a solvent exposed region on the surface of PTP1B, which we have termed the E-site. The expected orientation of the benzimidazole effectively explains the lack of SAR in the 5-position because these groups would project into the solvent and would not interact with the protein. The loss of activity with large substituents 56 at the 4-position could also be rationalized on the basis of a steric clash with Asp48 or Phe182. The relatively flat SAR for the sulfonamide was more difficult to rationalize on the basis of our design and modeling, which predicted that the arylsulfonamide NH and benzimidazole NH bidentate H-bond interaction with Asp48 would necessitate that the aryl group of the sulfonamide bind in an extended conformation against the protein in the so-called C-site. An X-ray cocrystal structure of benzimidazole sulfonamide 46/ PTP1B solved to 1.8 Å resolution confirmed the presence of the bidentate H-bonding pattern and the position of the

3780

Journal of Medicinal Chemistry, 2006, Vol. 49, No. 13

Combs et al.

Figure 4. Modeling of ortho-methyl (S)-IZD. (A) Molecular surface of 46 (orange mesh) shown with the molecular surface of the phosphotyrosine binding site of PTP1B (transparent gray), illustrating the complementarity of the ligand in the enzyme active site. (B) Molecular surface of a model of ortho-methyl derivative 79a with the unchanged enzyme surface showing that ortho-methyl occupies the entire available volume of the D-site, a small preformed pocket adjacent to the catalytic site of PTP1B. Figure 3. The X-ray cocrystal structure of compound 46/PTP1B at 1.8 Å resolution. (A) Structure of compound 46; (B) bidentate hydrogen bond with Asp48 shown between the benzimidazole and sulfonamide NHs; (C) top view showing the intramolecular pi-stacking of the ligands’ arylsulfonamide and benzimidazole groups.

benzimidazole in the E-site, but revealed an unexpected conformation for the arylsulfonamide (Figure 3). One observes a 180° rotation of the sulfonamide apparently driven by the intramolecular pi-stacking of the aryl ring of the sulfonamide and benzimidazole while maintaining the H bond to Asp48. The X-ray structure indicates that the aryl of the sulfonamide does not bind with the protein in any site but extends into the solvent with only intramolecular interactions. The failure to significantly improve the binding of these ligands by substitution on the arylsulfonamide is consistent with the observed binding mode in the solid state because the aryl ring does not interact with the protein, although the observed high affinity of these benzimidazole sulfonamide derivatives attests to the strength of this intramolecular interaction. This unique conformation of the ligand has been observed in all X-ray cocrystal structures of benzimidazole sulfonamide inhibitors solved to date. A detailed analysis of these cocrystal structures will be reported elsewhere. Further optimization of the benzimidazolesulfonamide lead focused on substitutions ortho to the (S)-IZD heterocyclic pTyr mimetic on the aryl ring. Substituting ortho to pTyr mimetics has been shown to be an effective strategy for significantly

improving the binding (5-20-fold) of nearly all PTP1B inhibitor chemotypes by binding into the relatively small D-site imbedded deep into the protein adjacent to the catalytic site. Thus, a series of ortho substituents to the (S)-IZD, including halo and alkyl groups, was targeted. The synthesis of each different ortho substituted scaffold required the development of a unique synthesis of 14-18 linear steps. The significant synthetic effort was rewarded with the identification of the most potent enzyme inhibitors of this series, and the first compounds to show cellular activites. The ortho-F derivatives 67 were the most potent, giving 2-4-fold increases in enzyme potency compared to that of the unsubstituted parent, whereas the ortho-chloro 87 and ortho-methyl 79 congeners were approximately equipotent (Table 4). The sterically larger ortho-Br derivative 93 was much less active. Although it is known that the D-site accommodates halogens and small hydrophobic substituents, it was surprising that the ortho-bromo derivative was not active because orthobromo difluoromethylphosphonate (DFMP) derivatives have been reported to increase potency as much as 20-fold over the unsubstituted DFMP derivatives.13 One possible explanation is that the steric bulk of the bromide adjacent to the (S)-IZD causes the optimal binding conformation of the heterocycle to be of much higher energy, thus resulting in a net loss in binding affinity. For the ortho-chloro and ortho-methyl derivatives, we propose that this negative interaction with the heterocycle is less severe for these smaller substituents and that they are nearly optimal for filling the available space in the D-site (Figure 4).

Protein Tyrosine Phosphatase 1B Inhibitors

Journal of Medicinal Chemistry, 2006, Vol. 49, No. 13 3781

Table 4. SAR of ortho-Substituted (S)-IZD Benzimidazole Sulfonamides

compd

X

R2

R1

PTP1B pNPP IC50 (nM)a

14 67a 67b 67c 87 93 79a 79bd 79c

H F F F Cl Br Me Me Me

H H H 5-Cl H H H H H

32 18 15 10 50 340c 35 >10000 16

79d 79e

Me Me

H H 3-F 3-CF3 H H H H 3-CF3, 4-Br 2-CN 3-Cl

H H

18 76

IR phos assay fold shift at 80 µMb

caco-2 Pm (×10-6 cm/s)

1.0 ( 0.3c 1.6 ( 0.6 1.1 2.3 1.0 0.7c 2.8 ( 0.8 1.0 2.0

0.4 0.2