J. Med. Chem. 2007, 50, 5339-5356
`
`5339
`
`Discovery of 1-(4-Methoxyphenyl)-7-oxo-6-(4-(2-oxopiperidin-1-yl)phenyl)-4,5,6,7-tetrahydro-
`1H-pyrazolo[3,4-c]pyridine-3-carboxamide (Apixaban, BMS-562247), a Highly Potent, Selective,
`Efficacious, and Orally Bioavailable Inhibitor of Blood Coagulation Factor Xa
`
`Donald J. P. Pinto,* Michael J. Orwat, Stephanie Koch, Karen A. Rossi, Richard S. Alexander, Angela Smallwood,
`Pancras C. Wong, Alan R. Rendina, Joseph M. Luettgen, Robert M. Knabb, Kan He, Baomin Xin, Ruth R. Wexler, and
`Patrick Y. S. Lam
`DiscoVery Chemistry, Research and DeVelopment, Bristol-Myers Squibb Company, 31 Pennington-Rocky Hill Road,
`Pennington, New Jersey 08534
`
`ReceiVed May 1, 2007
`
`Efforts to identify a suitable follow-on compound to razaxaban (compound 4) focused on modification of
`the carboxamido linker to eliminate potential in vivo hydrolysis to a primary aniline. Cyclization of the
`carboxamido linker to the novel bicyclic tetrahydropyrazolopyridinone scaffold retained the potent fXa binding
`activity. Exceptional potency of the series prompted an investigation of the neutral P1 moieties that resulted
`in the identification of the p-methoxyphenyl P1, which retained factor Xa binding affinity and good oral
`bioavailability. Further optimization of the C-3 pyrazole position and replacement of the terminal P4 ring
`with a neutral heterocycle culminated in the discovery of 1-(4-methoxyphenyl)-7-oxo-6-(4-(2-oxopiperidin-
`1-yl)phenyl)-4,5,6,7-tetrahydro-1H-pyrazolo[3,4-c]pyridine-3-carboxamide (apixaban, compound 40). Com-
`pound 40 exhibits a high degree of fXa potency, selectivity, and efficacy and has an improved pharmacokinetic
`profile relative to 4.
`
`Introduction
`Thrombotic diseases remain the leading cause of death in
`developed countries despite the availability of anticoagulants
`such as warfarin,1a-c heparin and low molecular weight hep-
`arins,2,3 and antiplatelet agents such as aspirin and clopidogrel.
`The oral anticoagulant warfarin inhibits the post-translational
`maturation of coagulation factors VII, IX, and X and prothrom-
`bin and has proven effective in both venous and arterial
`thrombosis. However, warfarin’s usage is limited because of
`its narrow therapeutic index, slow onset of therapeutic effect,
`numerous dietary and drug interactions, and a need for monitor-
`ing and dose adjustment.4a,b This not withstanding, warfarin
`remains the standard orally administered anticoagulant available
`in the United States. Patients on warfarin therapy require regular
`monitoring in part because of its narrow therapeutic index and
`interactions with food and other drugs. Injectable agents that
`are also widely used include low molecular weight heparins
`and the synthetic pentasaccharide fondaparinux.5 Thus, discov-
`ering and developing safe and efficacious oral anticoagulants
`for the prevention and treatment of a wider range of thrombotic
`diseases has become increasingly important.
`A key strategy for the discovery and development of new
`anticoagulants has been the targeting of specific enzymes within
`the blood coagulation cascade. One approach is to inhibit
`thrombin generation by targeting the inhibition of coagulation
`factor Xa (fXa).5,6a-h Factor Xa, a trypsin-like serine protease,
`is crucial to the conversion of prothrombin to thrombin, the
`final enzyme in the coagulation cascade that is responsible for
`fibrin clot formation. Preclinical animal models have suggested
`that inhibiting fXa has the potential for providing excellent
`antithrombotic efficacy with minimal bleeding risk when
`compared to direct thrombin inhibitors.6a-g Recent disclosures
`from clinical studies with direct fXa inhibitors such as compound
`
`* To whom correspondence should be addressed. Phone:
`5295. Fax (609) 818-3460. E-mail: donald.pinto@bms.com.
`
`(609) 818-
`
`Figure 1. Schematic of important pyrazole fXa compounds.
`
`4,7a-c rivaroxaban (BAY 59-7939),8a,b 1H-indole-5-carboxylic
`acid {(R)-2-[4-(4-methylpiperazin-1-yl)-piperidin-1-yl]-2-oxo-
`1-phenylethyl}amide (LY-517717)9 and the indirect parenteral
`fXa inhibitor fondaparinux5 have confirmed the preclinical
`findings.10
`The discovery of the pyrazole scaffold, illustrated by SN429
`(compound 1, Figure 1, fXa Ki ) 13 pM, trypsin Ki ) 16 nM),11
`was a significant milestone in our search for molecules targeting
`coagulation fXa and proved to be crucial in the evolution of
`orally bioavailable fXa inhibitors such as DPC423 (compound
`2, fXa Ki ) 0.15 nM,
`trypsin Ki ) 60 nM),11 DPC602
`(compound 3, fXa Ki ) 0.87 nM, trypsin Ki ) 1500 nM),12a
`and razaxaban (compound 4, fXa Ki ) 0.15 nM, trypsin Ki >
`
`10.1021/jm070245n CCC: $37.00 © 2007 American Chemical Society
`Published on Web 10/03/2007
`
`MYLAN EXHIBIT 1005
`
`

`

`5340 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22
`
`Pinto et al.
`
`Scheme 1. Syntheses of C-3-carboxypyrazolo-pyridinone Analoguesa
`
`a (a) NaNO2, HCl, 0 (cid:176) C, NaOAc, EtOH, ethyl 2-chloroacetoacetate; (b) Et3N/toluene, reflux; (c) 3 N HCl or TFA, DCM; (d) 2-formylphenylboronic acid,
`(Ph3P)4Pd, toluene/EtOH or DME/water (4:1), Na2CO3 (2 N), reflux; (e) 3-(R)-OH-pyrrolidine (2 equiv), NaCNBH3, ZnCl2 (0.5 N, in THF), MeOH; (f)
`LiOH or NaOH (1 N), MeOH/water; (g) NH4OH, EtOH, 80 (cid:176) C; (h) amine, NaCNBH3, ZnCl2 (0.5 N, in THF), MeOH; (i) oxalyl chloride, DMF; (j) MeNH2
`or NHMe2, trimethylaluminum (1 N), DCM, 0 (cid:176) C to room temp; (k) ammonia/MeOH, 50 (cid:176) C; (l) DMAP, TFAA; (m) ether, 20% aq. HCl; (n)
`p-methoxyphenylhydrazine, MeOH reflux.
`
`5000 nM).7a Compounds 2 and 4 were advanced to clinical trials.
`Subsequently, compound 4 was further advanced to a phase II
`trial for the prevention of venous thromboembolism (VTE) after
`knee replacement surgery and was shown to be highly effica-
`cious when compared to enoxaparin.7c
`Consistent with our strategy of developing and advancing key
`follow-on candidates, our focus was directed toward the
`identification of novel entities that would be significantly
`differentiated from previous candidates in terms of improving
`on potential
`liabilities of earlier compounds. A common
`structural feature that is present with compound 4 and its
`predecessor candidates was the presence of the 5-carboxamido
`linker that connects the pyrazole scaffold to the P4 moiety. In
`the advancement of potential candidates for preclinical evalu-
`ations, it was necessary to determine the susceptibility of the
`amide linker to metabolic cleavage because this could potentially
`liberate a aniline fragment. Fortunately, for compound 4 and
`its predecessor clinical compound 2 the amide linker was stable
`to metabolic hydrolysis; however, this was not the case with
`our preclinical compound 3, which liberated the biarylamino
`group at a higher pH. In the bacterial reverse mutation (AMES)13
`assay, the aniline moiety of 3 tested positive, which was further
`
`confirmed in follow-up assays for mutagenicity.14 Therefore,
`as part of our optimization strategy, we sought to modify the
`carboxamido portion of the molecule to obviate the need for
`mutagenicity studies on potential aniline degradants. Toward
`this end, we recently disclosed several series of bicyclic pyrazole
`scaffolds15a-c,16a in which the carboxamido linker was cyclized
`into the pyrazole ring, some of which showed similar or better
`fXa potency compared with the previously disclosed monocyclic
`pyrazole analogues.7a,b,11 The optimization strategy with the
`bicyclic pyrazole scaffold led to the identification of BMS-
`740808 (compound 5, fXa Ki ) 0.03 nM, trypsin Ki > 5000
`nM, Figure 1),15a which was advanced to preclinical safety
`evaluation. Importantly, the discovery of the potent bicyclic
`scaffold set the stage for exploratory work employing additional
`P1 moieties,7c,16 many of which demonstrated subnanomolar fXa
`binding affinities and moderate to high clearance (Cl) and
`volume of distribution (Vdss) in dogs. However, the lack of
`adequate differentiation from compound 4 in terms of improve-
`ment in the overall pharmacokinetic profile made them less
`attractive for further development. In this paper, we report an
`optimization strategy that resulted in the identification of
`compound 40, a structurally novel and neutral bicyclic pyrazole
`
`

`

`Apixaban as Inhibitor
`
`Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22 5341
`
`fXa inhibitor (currently in phase III trials) with a superior
`pharmacokinetic profile (low clearance and volume of distribu-
`tion) compared to compound 4.
`
`Chemistry
`The synthesis of the C-3 trifluoromethylpyrazole analogue 6
`was accomplished via the cyclization methodology previously
`described.15,16a Scheme 1 illustrates the general synthetic
`methodology utilized to prepare diversified pyrazole C-3 ana-
`logues. Commercially available 4-methoxyaniline was diazotized
`(NaNO2, concentrated HCl, 0 (cid:176) C) and condensed in situ with
`either 1-chloro-1-(methylsulfonyl)propan-2-one or ethyl 2-chlo-
`roacetoacetate in the presence of sodium acetate17 to provide
`the requisite p-methoxyphenylchlorohydrazones 7a in 57% yield
`and 7b in 90% yield. Treatment of the chlorohydrazones 7a
`and 7b with compound 815 using excess triethylamine afforded
`the requisite [3 + 2] cycloadducts which, when treated with
`TFA in dichloromethane, led to compounds 9a (80% yield) and
`9b (71% yield) respectively. Suzuki coupling of 9a,b with
`2-formylbenzeneboronic acid as illustrated for compound 515
`afforded the biaryl o-carboxaldehyde intermediates 10a in 71%
`yield and 10b in 80% yield, respectively. Subsequent reductive
`amination with 3-(R)-hydroxypyrrolidine15,16a provided the
`bicyclic pyrazole compounds 11a (45% yield) and 11b (69%
`yield). Hydrolysis (LiOH in THF and water) of the ester group
`in 11b gave the desired C-3 carboxylic acid compound 12 in
`51% yield. Compounds 13a-h were prepared in a two-step
`sequence by the reductive amination of 10b followed by
`carboxamide formation as described above in yields that ranged
`between 80% and 90%. Alternatively, treatment of compound
`11b with ammonium hydroxide in ethanol at 80 (cid:176) C for 4 days
`provided the carboxamidopyrazole analogue 13f in 45% yield.
`Hydrolysis (NaOH (1 N) in THF/water) of 9b gave carboxylic
`acid intermediate 14a (90% yield). Treatment of the pyrazole
`ester 9b under the Weinreb amide conditions (methylamine or
`dimethylamine, trimethylaluminum (1 N) in DCM at 0 (cid:176) C to
`room temperature)18 provided 14b (92% yield) and 14c (88%
`yield). The compounds were subsequently converted to 15a,b
`in 55% and 46% yield, respectively, following the Suzuki and
`reductive amination procedures. To prepare the cyanopyrazole
`compound 18, compound 10b was first converted to the
`carboxamidobiarylcarboxaldehyde 16 in 66% yield by treatment
`with ammonia in methanol at 80 (cid:176) C. Dehydration (oxalyl
`chloride in DMF) to 17 (42% yield) followed by reductive
`amination gave the desired cyano compound 18 (27% yield).
`The aminopyrazole compounds 20-25 were accessed ac-
`cording to the methodologies outlined in Scheme 2. Curtius
`rearrangement19 of the pyrazolecarboxylic intermediate 14a
`provided the Boc protected aminopyrazole intermediate 19a in
`22% yield. Biarylcarboxaldehyde formation (20, 93% yield)
`followed by reductive amination with 3-(R)-hydroxypyrrolidine
`afforded compound 21 in 77% yield. Treatment of compound
`21 with TFA provided compound 22 in 10% yield. Alternatively,
`compound 21 was alkylated with sodium hydride and iodo-
`methane in anhydrous DMF to afford 23 in 42% yield. Treat-
`ment of 23 with TFA in dichloromethane afforded compound
`24 in 99% yield. To prepare compound 25, pyrazole derivative
`19a was deprotected with TFA and reductively aminated with
`formaldehyde (37%) and sodium cyanoborohydride in the
`presence of zinc chloride (0.5 M in THF) to afford the
`dimethylaminopyrazole compound 19b in 51% yield. Biaryl-
`carboxaldehyde formation followed by reductive amination with
`3-(R)-hydroxypyrrolidine afforded compound 25 in 34% yield.
`Tetrazolyl compounds 27 and 28 were prepared according
`to Scheme 3. Treatment of 14b with lutidine and triflic
`
`Scheme 2. Syntheses of 3-Aminopyrazole Analoguesa
`
`a (a) Oxalylchloride, DCM, catalyst DMF; (b) NaN3, water, acetone 0
`(cid:176) C; (c) toluene, 80 (cid:176) C, tBuOH; (d) TFA, DCM; (e) formaldehyde (37%,
`excess), ZnCl2 (0.5 M/THF), NaBH3CN, MeOH; (f) 2-formylphenylboronic
`acid, (Ph3P)4Pd, Na2CO3 (2 N), 4:1 toluene/EtOH, reflux; (g) 3-(R)-OH-
`pyrrolidine, NaCNBH3, ZnCl2 (0.5 N, in THF), MeOH; (h) NaH, DMF,
`MeI, room temp.
`
`anhydride generated the iminotriflate, which was directly treated
`with excess sodium azide to give the tetrazole derivative 26
`in 48% yield. Suzuki coupling with 2-formylboronic acid
`and reductive amination with 3-(R)-hydroxypyrrolidine led to
`27 in 47% yield. The tetrazole compound 28 was prepared
`in 52% yield by heating compound 18 with sodium azide in
`DMF.
`Heteroarylalkyl compounds 33a-e were synthesized accord-
`ing to procedures outlined in Scheme 4. Borane reduction of
`carboxylic acid20 intermediate 14a afforded the alcohol inter-
`mediate 29 in 89% yield, which was subsequently converted to
`the bromomethyl intermediate 30 by treatment with phosphorus
`tribromide (PBr3) in dichloromethane in 94% yield. Displace-
`ment of the crude bromide 30 with 1,2,3-triazole, 1,2,4-triazole,
`or 1H-tetrazole afforded mixtures of regioisomeric triazole-
`methyl or tetrazolylmethyl compounds 31a-e, which were
`subsequently converted to biarylcarboxaldehyde compounds
`32a-e and later to the desired compounds 33a-e.
`Variably substituted P4 anilino compounds 34, 35, and 36a-e
`were prepared according to the methods outlined in Scheme 5.
`Aryl amination of compound 9c according to the Buchwald
`amination methodology21 afforded compound 34 in 97% yield.
`Acetylation of 34 with acetic anhydride and triethylamine gave
`the acetyl derivative 35 in 97% yield. Alternatively, aniline 34
`was converted to the Boc protected derivative 36a by treatment
`with Boc anhydride (neat) at 80 (cid:176) C in 84% yield. Alkylation
`with iodomethane provided 36b in quantitative yield (100%).
`Removal of the Boc protecting group afforded 36c was
`acetylated to afford compound 36d. Alkylation of 36c with
`idodomethane and potassium carbonate provided compound 36e
`in 47% yield.
`
`

`

`5342 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22
`
`Scheme 3. Syntheses of C-3-Cyano, 3-Tetrazole Analoguesa
`
`Scheme 4. Syntheses of Substituted C-3 Heteroalkyl
`Analoguesa
`
`Pinto et al.
`
`a (a) Triflic anhydride, lutidine, NaN3, DMF; (b) 2-formylphenylboronic
`acid, (Ph3P)4Pd, Na2CO3 (2 N), 4:1 toluene/EtOH, Na2CO3 (2 N), reflux;
`(c) 3-(R)-OH-pyrrolidine (2 equiv), NaCNBH3, ZnCl2 (0.5 N, in THF),
`MeOH; (d) NaN3, DMF, heat.
`
`Analogues in which the P4 moiety is either the phenylpip-
`eridinyl or the corresponding phenyllactam groups were accessed
`according to the methods outlined in Scheme 6. Ullmann
`coupling22
`(K2CO3, CuI, 1,10-phenanthroline in DMSO,
`130 (cid:176) C) of compound 9c with excess piperidine in a sealed tube
`provided compound 37 in 5% yield. In a similar manner, the
`Ullmann coupling of 9c with (cid:228)-valerolactam or caprolactam led
`to the P4 phenyllactam analogues 38a,b in 20-25% yield.
`Likewise, treatment of pyrazole 9b with (cid:228)-valerolactam under
`similar Ullmann conditions provided compound 39 in 21% yield,
`which on aminolysis with ammonia in ethylene glycol at
`120 (cid:176) C provided compound 40 in 76% yield.
`The preparation of compound 47 is outlined in Scheme 7.
`Cycloaddition of chlorohydrazone compound 7b and morpholine
`derivative 42 (prepared in 65% yield from lactam 41) with
`triethylamine in toluene under reflux conditions followed by
`treatment with TFA afforded the bicyclic pyrazole 43 in 75%
`yield. Hydrogenation (palladium on carbon in methanol)
`provided aniline 44 in 96% yield. Boc protection of 44 (Boc2O,
`NaH in THF) followed by alkylation (NaH and iodomethane)
`and removal of
`the Boc group with TFA provided the
`N-methylaniline derivative 45 in 56% yield. Aminolysis of 45
`with ammonia in ethylene glycol at 120 (cid:176) C led to compound
`46, which was acetylated (acetyl chloride in the presence of
`sodium hydroxide (1 N) in DCM) to compound 47 in 30% yield.
`
`Results and Discussion
`in potency seen with the
`Because of the enhancement
`tetrahydropyrazolopyridone scaffold, efforts to extend the SAR
`to include neutral P1 groups such as the p-methoxyphenyl that
`previously showed reduced fXa binding in the monocyclic
`pyrazole series proved to be successful.16a-b Although the
`compounds with this P1 group showed potent fXa inhibition in
`
`a (a) BH3. THF, room temp; (b) PBr3, DCM, room temp; (c) NaH, 1,2,3-
`triazole or 1,2,4-triazole or 1H-tetrazole, DMF; (d) 2-formylphenylboronic
`acid, (Ph3P)4Pd, Na2CO3 (2 N), 4:1 toluene/EtOH, reflux; (e) 3-(R)-
`hydroxypyrrolidine, NaCNBH3, ZnCl2 (0.5 N in THF), MeOH.
`
`Scheme 5. Syntheses of Substituted P4 Amino Analoguesa
`
`a (a) Diphenylmethanimine, BINAP, NaOtBu, Pd2(dba)3, toluene, reflux;
`(b) hydroxylamine hydrochloride, NaOAc, MeOH; (c) Boc2O, neat, 80 (cid:176) C;
`(d) NaH, MeI, DMF; (e) TFA, DCM; (f) MeI, DMF, K2CO3, room temp;
`(g) Ac2O, TEA, DCM, room temp.
`
`the binding assay, the in vitro clotting activity as measured by
`the prothrombin time (PT) assay of these compounds was
`moderate to high. Further optimization of the p-methoxyphenyl
`
`

`

`Apixaban as Inhibitor
`
`Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22 5343
`
`Scheme 6. Syntheses of P4 Lactam Analogues and Compound
`37a
`
`Table 1. In Vitro Activity for Substituted C-3 Pyrazolopyridinonesa
`
`a (a) 1.5 equiv of piperidine, K2CO3, catalyst CuI, DMSO, sealed tube,
`130 (cid:176) C, 24 h; (b) (cid:228)-valerolactam or azepan-2-one, K2CO3, catalyst CuI,
`catalyst 1,10-phenanthroline, DMSO, 130 (cid:176) C 24 h; (c) ammonia in ethylene
`glycol, 120 (cid:176) C, 4 h.
`
`Scheme 7. Synthesis of Compound 47a
`
`a (a) 3 equiv of PCl5, CHCl3, reflux; (b) morpholine reflux; (c) TEA,
`toluene, reflux; (d) TFA, DCM; (e) H2, Pd/C (10%), MeOH; (f) Boc2O,
`NaH, THF; (g) NaH, THF, MeI; (h) TFA, DCM; (i) ammonia, MeOH/
`ethylene glycol, 120 (cid:176) C, 4 h; (j) acetyl chloride, NaOH (1 N), DMC.
`
`bicyclic pyrazole series required careful adjustment for potency
`and polarity at the C-3 pyrazole position for possible alternatives
`to the lipophilic trifluoromethyl substituent (Table 1). In the
`course of our investigation of the C-3 pyrazole position, we
`
`a Ki values were obtained from purified human enzymes and are averaged
`from two experiments (n ) 2).28,29 PT values are measured according to
`refs 7a and 11. Human trypsin Ki values for all compounds above are >3000
`nM. NT indicates “not tested”.
`
`were gratified to see the breadth of substitutions that were readily
`accommodated in this region of the fXa active site. For example,
`in addition to the trifluoromethyl analogue 6,16a subnanomolar
`inhibitory activity was seen for the methylsulfonyl compound
`11a (fXa Ki ) 0.25 nM), the carboxamido compound 13f
`
`

`

`5344 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22
`
`Pinto et al.
`
`Table 2. Comparative Permeability and Dog Pharmacokinetic
`Parametersa
`
`assay. Amongst the tetrazole analogs, the tetrazole compound
`27 showed moderate clotting activity, though less potent in the
`binding assay, and the reverse was true for tetrazole 28.
`Triazolylmethyl and tetrazolylmethyl analogues 33a-d exhib-
`ited good potencies in both assays. Overall, the compounds
`shown in Table 1 were highly selective (>1000-fold) for fXa
`relative to other serine proteases such as thrombin and trypsin.
`Table 2 lists the pharmacokinetic profile in dogs of a repre-
`sentative set of the most optimized C-3 substituted compounds.
`The carboxamide 13f demonstrated an excellent pharmacokinetic
`profile, with low clearance (Cl ) 0.32 L kg-1 h-1), moderate
`volume of distribution (Vdss ) 1.6 L kg-1), and a half-life (T1/2)
`of 5.6 h. The high oral bioavailability (F ) 100%) exhibited
`by 13f was consistent with the high apparent permeability (Papp
`) 2.3 (cid:2) 10-6 cm s-1) of this compound in the Caco-2 assay.23
`In contrast, the pharmacokinetic profiles of the triazole analogue
`33a and the 1,2,3,4-tetrazole analogue 33d were poor with high
`clearance, moderate volume of distribution, and poor oral
`bioavailability.
`Given the excellent fXa activity exhibited by compound 13f
`and its high oral bioavailability, we shifted the focus on further
`P4 optimization (Table 3). In general, the compounds retained
`subnanomolar fXa binding affinity and potent clotting activity,
`good selectivity (trypsin/thrombin, >100-fold), and showed
`good permeability (Papp) in the Caco-2 assay. The unsubstituted
`amino compound 13a (fXa Ki ) 0.97 nM) was the least potent,
`whereas the substituted amino compounds 13b-d exhibited
`Table 3. C-3 Carboxamido Pyrazoles: In Vitro and in Vivo Profile of the P4 Biarylmethylamino Moietiesa
`
`a Compounds were dosed (po/iv) as TFA salts in a cassette dosing N-in-
`one format7a,34a-c at 0.4 mg/kg iv and at 0.2 mg/kg po (n ) 2).10 Caco-2
`and dog PK parameters were measured according to refs 7a and 11.
`(fXa Ki ) 0.07 nM), the nitrile compound 18 (fXa Ki )
`0.33 nM), and the dimethylamino compound 25 (fXa Ki )
`0.31 nM). The binding affinity and clotting activity of the
`carboxamide analog 13f showed significant improvement when
`compared to the corresponding trifluoro-methyl compound 6.
`Compared to the parent carboxamide 13f,
`the substituted
`carboxamides 15a and 15b, though significantly less potent in
`the binding assay, were only about 2- to 3-fold less potent in
`the clotting (PT) assay. The ester analogue 11b and its
`corresponding carboxylic acid 12 were less potent in both assays.
`Among the C-3 amino analogues investigated, the order of fXa
`potency was NMe2 > NHMe g N(Me)Boc g NH2, NHBoc.
`The unsubstituted amino analogue 22 and the N-methylamino
`compound 24 demonstrated acceptable activity in the clotting
`
`a Ki values were obtained from purified human enzymes and are averaged from two experiments (n ) 2).28,29 Prothrombin time (PT) values are measured
`according to refs 7a and 11. Human trypsin Ki values for all compounds listed in Table 2 are >3000 nM. Caco-2 and dog PK parameters were measured
`according to refs 7a and 11. b Compounds were dosed (po/iv) as TFA salts in a cassette dosing N-in-one format at 0.4 mg/kg iv and 0.2 mg/kg po (n )
`2).34a-c NT indicates “not tested.”
`
`

`

`Apixaban as Inhibitor
`potent fXa inhibitory activity (fXa Ki < 0.3 nM) and good
`clotting activity (PT EC2(cid:2) < 1.5 (cid:237)M). As was observed with
`the 3-(R)-hydroxypyrrolidine compound 13f, the 4-hydroxypi-
`peridinyl analogue 13g also demonstrated low clearance, moder-
`ate volume of distribution, and half-life in the same range as
`observed with 13f. Again, the high dog oral bioavailability seen
`for 13g (F ) 55%) correlated well with the observed Caco-2
`(Papp) permeability value. In the rabbit arteriovenous shunt
`(AVShunt) thrombosis model,6g compounds 13b, 13c, 13f, and
`13g inhibited thrombus formation in a dose-dependent manner
`with IC50 values of 445, 175, 120, and 180 nM, respectively,
`and with the exception of 13b were slightly more potent than
`compound 4 (AVShunt IC50 ) 340 nM)7a and were in the same
`range when compared to compound 5 (AVShunt IC50 ) 140
`nM).15a Of the compounds in this series, 13b had the longest
`half-life in dogs, albeit with relatively high clearance and high
`volume of distribution. Interestingly, these data did not correlate
`well with the observed 13b half-life in the human liver
`microsome (HLM) assay24 (T1/2 > 100 min). In the same assay,
`the HLM half-life for compounds 4 and 5 was 38 and 42 min,
`respectively. Taken together therefore, the carboxamide pyrazole
`analogue 13f emerged as a potent alternative to compound 4,
`with excellent potency both in vitro and in vivo and a good
`pharmacokinetic profile in dogs.
`In a parallel effort, the compounds containing P4 nitrogen
`atom (as the point of attachment) were also explored (Table
`4).12c This strategy proved to be highly successful in that a potent
`compound 36d bearing a N-methylacetyl group was quickly
`identified. This discovery was significant in that it differed in
`structure from all our previous pendent P4 groups we had
`explored. The compound, though more potent than the aniline
`derivatives 34, 36c, and 36e, was weak in the clotting assay,
`suggesting high protein binding. The high level of potency
`exhibited by compound 36d (fXa Ki ) 0.50 nM) suggested that
`the orientation of the P4 N-methylacetyl substituent in the S4
`region of the fXa active site is very important. This type of
`observation was unique in the fXa literature at the time it was
`discovered. To explain this finding, a closer look at the model
`of 36d in the active site of fXa clearly showed the N-methyl P4
`group forming a lipophilic (cid:240) interaction with the bottom S4
`Trp215 residue,12d and thus positioning the acetyl carbonyl
`functionality perpendicular to the inner P4 phenyl ring, thereby
`forming a hydrophobic interaction with the other residues in
`this region. The importance of the orientation of the N-methyl
`group was confirmed by the loss in fXa affinity with the
`acetamide analogue 35 (fXa Ki ) 180 nM) where the planarity
`of this group positioned it in an unfavorable orientiation in the
`S4 region of the enzyme. The dimethylamino analogue 36e (fXa
`Ki ) 6.0 nM) and piperidinyl 37 (fXa Ki ) 2.1 nM) were also
`less potent, suggesting a planar orientation with these moieties
`in the S4 region as well. Cyclization of the P4 N-methyl acetyl
`group in 36d to form lactam analogues 38a and 38b retains the
`subnanomolar fXa binding affinity. Unfortunately,
`lactam
`analogues 38a (PT ) 23 (cid:237)M) and 38b (PT ) 26 (cid:237)M) exhibited
`poor anticoagulant activity. This could be explained by the high
`lipophilicity (cLogP > 7) and high human serum protein binding
`(>99%)25 exhibited by these compounds.
`In order to modulate the lipophilicity of 36d and 38a, we
`reintroduced the polar C-3 carboxamido moiety (Table 5) that
`was shown to be important in compounds 13a-h. The car-
`boxamidopyrazole analogue 40 (fXa Ki ) 0.08 nM, PT ) 3.8
`(cid:237)M) and 47 (fXa Ki ) 0.61 nM, PT ) 3.1 (cid:237)M) not only
`maintained subnanomolar fXa binding affinity but also dem-
`onstrated much improved potency in the clotting (PT) assay,
`
`Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22 5345
`
`Table 4. In Vitro Profile of P4 Substituents with Imbedded Nitrogena
`
`a Ki values were obtained from purified human enzymes and are averaged
`from two experiments (n ) 2).28,29 Prothrombin time (PT) values are
`measured according to refs 7a and 11. Human trypsin Ki values for all
`compounds listed in Table 2 are >3000 nM. NT indicates “not tested”.
`
`especially when compared to the related trifluoromethylpyrazole
`analogues 38a and 36d.
`Selectivity and Liability Profiling. Compound 40 shows a
`high degree of selectivity versus other proteases (see supple-
`mental section), even compared to compounds 47a and 5.15a
`Additionally, the compound shows weak activity against various
`P450 isozymes (IC50 > 25(cid:237)M) and weak activity against the
`hERG potassium channel (IC50 > 25 (cid:237)M, patch clamp assay).26a-e
`The solubility of compound 40 was shown to be approximately
`40-50 (cid:237)g/mL.27 In the human liver microsome assay, com-
`pound 40 was very stable with a T1/2 of >100 min (the HLM
`T1/2 of 47 was not measured). The Caco-2 permeability values
`for compounds 40 (Papp ) 0.9 (cid:2) 10-6 cm s-1) and 47 (Papp )
`2.5 (cid:2) 10-6 cm s-1) were moderate to high.
`Dog Pharmacokinetics and Rabbit Antithrombotic Ef-
`ficacy. As a result of the excellent
`in vitro potency and
`selectivity of compounds 40 and 47, the pharmacokinetic profiles
`of both compounds were studied in dogs using a cassette dosing
`paradigm (“N-in-one” study, Table 6).7a,34a,b The acetylated
`N-methyl analogue 47 was orally bioavailable; but showed high
`clearance (Cl ) 2.8 L kg-1 h-1), moderate volume of distribu-
`tion (Vdss ) 1.7 L kg-1), and unacceptable half life. The dog
`pharmacokinetics for compound 40 was outstanding with very
`low clearance (Cl ) 0.02 L kg-1 h-1), and low volume of
`distribution (Vdss ) 0.2 L kg-1). These values were significantly
`
`

`

`5346 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22
`
`Table 5. Optimization of the Amido and Lactam P4 Moietiesa
`
`Pinto et al.
`
`a Ki values were obtained from purified human enzymes and are averaged from two experiments (n ) 2).28,29 PT and APTT values, Caco-2, and solubilities
`were measured according to refs 7a and 11. Human trypsin Ki values for all compounds listed in Table 2 are >3000 nM. NT indicates “not tested.”
`
`Table 6. Comparative in Vitro and in Vivo Profiles of Compounds 40, 47, 4, and 5a
`
`a “h” and “r” refer to human and rabbit species, respectively. Ki values were obtained from purified human enzymes and are averaged from two experiments
`(n ) 2). Compounds were dosed in a cassette dosing (po/iv) N-in-one format7a,34a-c a refers to the cassette dosing (po/iv) dog pharmacokinetic parameters.
`P.B. refers to serum protein binding. NT indicates “not tested”.
`
`lower than those observed with compounds 4 and 5. FXa being
`a vascular target, the pharmacokinetic profile for compound 40
`was viewed as highly desirable and less likely to have nontarget-
`related adverse effects. Importantly, compound 40 had a
`moderate half-life (T1/2 ) 5.8 h) and good oral bioavailability
`(F ) 58%). The human serum protein binding as measured by
`equilibrium dialysis for 40 was 87%.25 In the rabbit AVShunt
`thrombosis model (Figure 4), compound 40 inhibited thrombus
`formation in a dose-dependent manner with an IC50 value of
`329 nM.15d,e This is comparable to the antithrombotic potency
`obtained for compound 4 (IC50 ) 340 nM) in the same model.
`X-ray Crystallography of Compound 40. The X-ray
`structure for compound 40 bound to fXa (Figure 2, 2.3 Å
`resolution with an R value of 0.229 and an Rfree of 0.277) shows
`a tight inhibitor-enzyme complex30-33 and shows a similar
`binding mode compared to compound 211 and 4.7a The p-
`methoxy group in the S1 specificity pocket does not appear to
`interact with any specific residue in this region of the enzyme,
`and is oriented in a planar manner relative to the phenyl P1
`moiety. Other interactions that were found to be similar to those
`observed for compounds 47a and 515a included the pyrazole N-2
`nitrogen atom interaction with the backbone of Gln192 (3.2 Å)
`and the carbonyl oxygen (scaffold carboxamide) interaction with
`the NH of Gly216 (2.9 Å). The pyrazole C-3 carboxamido
`moiety shows the NH within bonding distance to the Glu146
`carbonyl oxygen (3.1 Å) with the carbonyl oxygen solvent
`exposed. The orientation of the pendant P4 phenyllactam of 40
`in the S4 region shows an edge to face interaction with Trp215
`and is appropriately positioned between the Tyr99 and Phe174
`
`Figure 2. X-ray structure of fXa bound to compound 40. Atomic
`coloring is used with cyan for fXa cartoon and carbon atoms and gray
`for compound 40 carbon atoms. Water molecules are shown as red
`spheres. Initial electron density (2Fo - Fc contoured at 1(cid:243)) is shown
`in magenta. Hydrogen bonds between protein and ligand are shown as
`dashed black lines. The figure was created using PyMol.35
`
`residues. The X-ray structure does not appear to show the
`carbonyl oxygen group of the pendent lactam moiety directly
`interacting with any residues in the S4 pocket but appears to
`show it interacting with a water molecule. Importantly, the P4
`lactam carbonyl brings about a conformational bias toward
`
`

`

`Apixaban as Inhibitor
`
`Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22 5347
`
`Table 7. Diffraction Data for Compound 40 (BMS-562247)
`
`(A) Data Collection Statistics
`P212121
`
`space group
`unit cell parameters
`a, Å
`b, Å
`c, Å
`
`diffraction limits, Å
`no. of unique reflections
`data completeness
`
`56.4
`72.8
`79.4
`overall
`19.4-2.3
`14544
`96.8
`
`highest shell
`2.4-2.3
`1645
`94.9
`
`2238
`34
`122
`0.008
`1.5
`0.229
`0.277
`5.9
`
`(B) Refinement Statistics
`no. of non-H atoms in protein
`no. of non-H atoms in inhibitor
`no. of refined water molecules
`rmsd bond distances from ideal, Å
`rmsd angles from ideal, deg
`crystallographic residual R value
`crystallographic residual Rfree
`test set, % of all reflections
`
`Figure 3. Superposition X-ray structures of fXa bound to compounds
`40 and 4. Atomic coloring is used with cyan for fXa cartoon and carbon
`atoms, gray for compound 40 carbon atoms, magenta for compound 4
`carbon atoms, and orange for compound 4 fluorine atoms. The
`orientation is the same as in Figure 2. The figure was created using
`PyMol.35
`
`Figure 4. Rabbit arteriovenous shunt (AVShunt) profile of compound
`40 (BMS-562247).
`
`orthogonality to optimally position itself between the S4 enzyme
`residues Trp215, Tyr99, and Phe174. Overall, compound 40 fits
`into the fXa enzyme active site in a highly complementary
`manner. Figure 3 shows an overlay