Article pubs.acs.org/jmc
Identification of 4‑(Aminomethyl)-6-(trifluoromethyl)-2(phenoxy)pyridine Derivatives as Potent, Selective, and Orally Efficacious Inhibitors of the Copper-Dependent Amine Oxidase, Lysyl Oxidase-Like 2 (LOXL2) Martin W. Rowbottom,* Gretchen Bain, Imelda Calderon, Taylor Lasof, David Lonergan, Andiliy Lai, Fei Huang, Janice Darlington, Patricia Prodanovich, Angelina M. Santini, Christopher D. King, Lance Goulet, Kristen E. Shannon, Gina L. Ma, Katherine Nguyen, Deidre A. MacKenna, Jilly F. Evans, and John H. Hutchinson PharmAkea Therapeutics, San Diego Science Center, 3030 Bunker Hill Street, Suite 300, San Diego, California 92109, United States S Supporting Information *
ABSTRACT: LOXL2 catalyzes the oxidative deamination of ε-amines of lysine and hydroxylysine residues within collagen and elastin, generating reactive aldehydes (allysine). Condensation with other allysines or lysines drives the formation of inter- and intramolecular cross-linkages, a process critical for the remodeling of the ECM. Dysregulation of this process can lead to fibrosis, and LOXL2 is known to be upregulated in fibrotic tissue. Small-molecules that directly inhibit LOXL2 catalytic activity represent a useful option for the treatment of fibrosis. Herein, we describe optimization of an initial hit 2, resulting in identification of racemic-trans-(3-((4-(aminomethyl)-6-(trifluoromethyl)pyridin-2-yl)oxy)phenyl)(3-fluoro-4-hydroxypyrrolidin-1-yl)methanone 28, a potent irreversible inhibitor of LOXL2 that is highly selective over LOX and other amine oxidases. Oral administration of 28 significantly reduced fibrosis in a 14-day mouse lung bleomycin model. The (R,R)-enantiomer 43 (PAT-1251) was selected as the clinical compound which has progressed into healthy volunteer Phase 1 trials, making it the “first-in-class” small-molecule LOXL2 inhibitor to enter clinical development.
■
INTRODUCTION The lysyl oxidase (LOX) family of enzymes are copperdependent amine oxidases consisting of five distinct isoforms, namely, lysyl oxidase (LOX) and lysyl oxidase-like (LOXL) 1 through 4.1,2 The LOX family is divided into two subgroups based on structural differences in the N-terminal domain. Both LOX and LOXL1 contain a highly basic propeptide that requires cleavage in order to generate active enzyme, whereas LOXL2, 3, and 4 contain four scavenger receptor cysteine-rich (SRCR) domains which are thought to mediate protein− protein interactions and cell adhesion.2 All LOX family members possess a conserved carboxy-terminal catalytic domain containing both a His-X-His-X-His copper binding motif and a post-translationally modified lysine tyrosylquinone (LTQ) cofactor. These enzymes localize to the extracellular matrix (ECM) where they catalyze the oxidative deamination of the terminal ε-amine of lysine and hydroxylysine residues in both collagen and elastin. This results in the generation of highly reactive aldehydes (allysine) that can further condense with other allysines or intact lysines to form inter- and intramolecular cross-linkages, a process critical for normal remodeling of the ECM.2,3 Dysregulation of this process can © 2017 American Chemical Society
lead to increased cross-linking, resulting in abnormal and uncontrolled ECM deposition in tissues. If left unchecked, this can lead to pathological fibrosis characterized by excessive scarring, impairment of organ function, and ultimately organ failure. Examples include fibrosis of the lung4 (such as idiopathic pulmonary fibrosis or IPF), liver,5 kidneys,6 and skin7 (systemic sclerosis). Unfortunately, current treatment options for patients diagnosed with fibrosis are extremely limited. Only for IPF have new antifibrotic medications been recently approved, these being nintedanib and pirfenidone.8 However, these agents offer only modest benefit to patients,9 so there is a need for additional therapies for both IPF and other types of fibrotic diseases. Given that aberrant ECM deposition is a driver in the progression of fibrotic disease and that LOX family enzymes are key mediators of this process, inhibition of one or more LOX proteins may delay or even halt disease progression. One family member in particular, LOXL2, has been extensively studied and shown to be upregulated in both fibrotic tissue and certain cancers.10 In addition to its role in the Received: March 7, 2017 Published: May 4, 2017 4403
DOI: 10.1021/acs.jmedchem.7b00345 J. Med. Chem. 2017, 60, 4403−4423
Journal of Medicinal Chemistry
Article
Figure 1. (a) Structures of the known pan-LOX(L) inhibitor BAPN 1, and the LOXL2 inhibitor 2. (b) Proposed mechanism for the oxidation of lysine amines by LTQ-containing lysyl oxidase enzymes. (c) Proposed complex between LOXL2 and either lysine residues or pyridine-derived inhibitors.
oxidative-deamination of the trimethylated lysine 4 of histone H3 (H3K4me3).11 Although deamination of H3K4me3 to give the corresponding allysine requires catalytically competent LOXL2, the mechanism is not well understood but has been suggested to involve release of trimethylamine, followed by
ECM, an intracellular role has been proposed, and LOXL2 has been reported to be involved in the induction of epithelial to mesenchymal transition (EMT), a process central to tumor cell invasion and metastasis. LOXL2 can mediate EMT via either stabilization of the transcription factor Snail1 or catalytic 4404
DOI: 10.1021/acs.jmedchem.7b00345 J. Med. Chem. 2017, 60, 4403−4423
Journal of Medicinal Chemistry
Article
Scheme 1. Synthesis of 4-(Methylamino)-2-phenoxypyridine Derivatives 6, 9, 10, and 13−21a
Reagents and conditions: (a) phenol, Cs2CO3, DMA, 80 °C, 95%; (b) 10 wt % Pd/C, H2 gas (1 atm), EtOAc-MeOH, rt, 91%; (c) 2 M HCl in Et2O, DCM, rt, 100%; (d) methyl 4-hydroxybenzoate or methyl-3-hydroxybenzoate, Cs2CO3, DMA, 80 °C, 58−60%; (e) 10 wt % Pd/C, H2 gas (1 atm), EtOAc-MeOH, rt, 94−100%; (f) Boc2O, DIEA, THF, rt, 67−80%; (g) LiOH, THF-H2O, rt, 83−100%; (h) R1R2NH, HATU, DIEA, DCM, rt; (i) 2 M HCl in Et2O, DCM, rt, 9−84% (over 2-steps); (j) methyl 4-aminobenzoate, HATU, DIEA, DMF, rt, 68%; (k) LiOH, THF-H2O, rt, 59%; (l) 2 M HCl in Et2O, DCM, rt, 52%. a
oxidation to give the aldehyde (allysine).11b As such, an inhibitor of LOXL2 capable of cell and nuclear penetration could be effective in modulating EMT, and both extracellular and intracellular inhibition of LOXL2 may be more efficacious in the treatment of both fibrosis and metastatic cancers. Simtuzumab is a LOXL2-specific monoclonal antibody that only partially inhibits catalytic activity via binding to a distal allosteric site located in the fourth SRCR domain.10b Although the corresponding mouse antibody (AB0023) demonstrated efficacy in several preclinical models of fibrosis and cancer,10b,d subsequent Phase 2 trials with simtuzumab in patients with pancreatic cancer, IPF, and liver fibrosis failed to demonstrate efficacy, and development was recently discontinued.12 We envision that a small-molecule approach toward selective inhibition of LOXL2 will have significant advantages over an antibody, for instance, (i) complete inhibition of catalytic activity via direct binding to the catalytic active site, (ii) greater penetration into the fibrotic matrix, and (iii) access to both extra- and intracellular compartments, allowing for inhibition of LOXL2 activity in the ECM and within cells. There are currently no published structures of selective inhibitors of LOXL2, and the only published small-molecule inhibitor is the pan-LOX(L) inhibitor β-aminopropionitrile (BAPN) 1 (Figure 1a).13 However, LOX is a ubiquitous enzyme, and mice deficient in LOX show severe cardiovascular and pulmonary defects.14 Therefore, our goal was to identify inhibitors of LOXL2 that are selective over LOX. Given the paucity of selective small-molecule leads and the lack of NMR or X-ray structures for any of the LOX(L) enzymes, our approach started with the examination of the proposed mechanism by
which LOX(L) proteins enzymatically convert lysines to allysines. A simplified proposed mechanism is shown in Figure 1b.10a Briefly, the ε-amine (of lysine residues) reversibly binds one of the carbonyl groups of the LTQ cofactor (of I) to yield the substrate Schiff base intermediate (II). Subsequent irreversible conversion of the substrate Schiff base to the product Schiff base intermediate (III) occurs via proton abstraction (postulated to be mediated by a nearby active site basic residue) and concomitant reduction/aromatization of the quinonoid ring (of II). Hydrolysis of III yields the aldehyde (allysine) and the intermediate aminophenol (IV), which is then recycled, via a number of steps involving O2 and Cu, to regenerate the catalytically active species (I) with release of both hydrogen peroxide and ammonia.15 Upon the basis of the proposed mechanism, we hypothesized that a substrate-like aminomethyl-containing small molecule should be able to compete with lysine in binding directly to the LTQ carbonyl, thus forming a small molecule-containing substrate Schiff base (II) which is presumably converted to the product Schiff base (III). Furthermore, whereas the lysine-containing Schiff base (III) readily undergoes hydrolysis releasing allysine, the small molecule-containing Schiff base may be able to form a tightly bound Schiff base complex which could be more resistant to hydrolytic cleavage.16 In addition, LOXL2 is glycosylated at the active site, while LOX is not, and this structural difference may provide an opportunity to design inhibitors selective for LOXL2 versus LOX.2,10a With this in mind, we executed a focused screening campaign and rapidly identified a series of 2substituted pyridine-4-ylmethanamines as moderately selective LOXL2 inhibitors. The best compound, 4-(aminomethyl)-24405
DOI: 10.1021/acs.jmedchem.7b00345 J. Med. Chem. 2017, 60, 4403−4423
Journal of Medicinal Chemistry
Article
Scheme 2. Synthesis of 4-(Methylamino)-6-(trifluoromethyl)pyridine Derivatives 25−33, 36, 37, and 39a
Reagents and conditions: (a) methyl 3-hydroxybenzoate, K2CO3, THF-DMF, 60 °C, 91%; (b) CoCl2, NaBH4, THF-MeOH, 0 °C, 92%; (c) Boc2O, DIEA, THF, rt, 78%; (d) LiOH, THF-H2O, rt, 87%; (e) R1R2NH, HATU, DIEA, DCM-DMF, rt; (f) 2 M HCl in Et2O, DCM, rt, 53−89% (over 2steps); (g) (R)- or (S)-1-N-Boc-3-hydroxypiperidine, NaH, THF, 0 °C-rt, 47−53%; (h) 2 M HCl in Et2O, DCM, rt, 96−100%; (i) PhNCO, DIEA, THF, rt, 62−80%; (j) 10 wt % Pd/C, H2 gas (1 atm), EtOAc-MeOH, rt, 50−73%; (k) 3-phenoxyphenol, K2CO3, THF, 90 °C, 45%; (l) 10 wt % Pd/ C, H2 gas (1 atm), EtOAc, rt, 83%. a
chloropyridine 2 (Figure 1a) was shown to have ∼30-fold selectivity over LOX and was also selective against a panel of amine oxidases.16 Herein we describe the SAR and lead optimization campaign around 2 which led to the identification of racemic-trans-(3-((4-(aminomethyl)-6-(trifluoromethyl)pyridin-2-yl)oxy)phenyl)(3-fluoro-4-hydroxypyrrolidin-1-yl)methanone 28, a highly potent LOXL2 inhibitor with significantly improved selectivity over LOX. Compound 28 is an orally bioavailable, irreversible inhibitor of LOXL2 that exhibits efficacy in a mouse bleomycin lung fibrosis model with once-daily oral dosing. Subsequent profiling of the individual enantiomers of 28 ultimately led to the selection of the (R,R)enantiomer 43 (PAT-1251)17 as the lead compound.
Subsequent HATU-mediated coupling of 8 with the appropriate amine followed by N-Boc deprotection gave final compounds 9 and 10 as HCl salts. Final compounds 13−20 (Scheme 1) were obtained in analogous fashion via reaction of 2-chloro-4-pyridinecarbonitrile 4 with methyl 3-hydroxybenzoate to afford the meta-substituted 2-phenoxypyridine derivative 11. Subsequent nitrile reduction, N-Boc protection, and then ester hydrolysis gave acid-derivative 12, which was readily converted to compounds 13−20. Coupling of acidderivative 12 with methyl 4-aminobenzoate, followed by ester hydrolysis and N-Boc deprotection, gave final compound 21 (Scheme 1). 4-(Methylamino)-6-(trifluoromethyl)pyridine derivatives 25−33 were prepared as described in Scheme 2. Treatment of 2-chloro-6-(trifluoromethyl)isonicotinonitrile 22 with methyl 3-hydroxybenzoate in the presence of K2CO3 gave 2-(phenoxy)-6-(trifluoromethyl)pyridine derivative 23. Reduction of the nitrile to the corresponding methylamine via treatment with NaBH4 and CoCl2, followed by N-Boc protection then ester hydrolysis, afforded acid-containing derivative 24. Desired compounds 25−33 were obtained via previously described amide-coupling procedures followed by NBoc deprotection. Piperidine phenylurea-containing derivatives 36 and 37 were prepared as shown in Scheme 2. Treatment of 2-chloro-6-(trifluoromethyl)isonicotinonitrile 22 with either (R)- or (S)-1-N-Boc-3-hydroxypiperidine and NaH, followed by N-Boc deprotection, afforded NH-piperidine derivatives 34
■
CHEMISTRY All compounds described herein were prepared as illustrated in Schemes 1−4. Compounds 6, 9, and 10 were obtained in the following manner (Scheme 1). Reaction of 2-chloro-4pyridinecarbonitrile 4 with phenol in the presence of Cs2CO3 gave 2-phenoxypyridine derivative 5. Subsequent nitrile reduction followed by treatment with HCl afforded compound 6. Treatment of 4 with methyl 4-hydroxybenzoate in the presence of Cs2CO3 gave para-substituted 2-phenoxypyridine derivative 7. Reduction of the nitrile followed by N-Boc protection of the resulting amine, then base-mediated hydrolysis of the ester, afforded carboxylic acid derivative 8. 4406
DOI: 10.1021/acs.jmedchem.7b00345 J. Med. Chem. 2017, 60, 4403−4423
Journal of Medicinal Chemistry
Article
Scheme 3. Synthesis of 4-(Methylamino)-6-(trifluoromethyl)pyridine Derivatives 43 and 44 via Preparative Normal-Phase Chiral HPLC Separationa
a
Reagents and conditions: (a) racemic-trans-4-fluoro-3-hydroxypyrrolidine·HCl, HATU, DIEA, DCM-DMF, rt, 87%; (b) separation via preparative normal-phase chiral HPLC, 59−68%; (c) 2 M HCl in Et2O, DCM, rt, 100%.
and 35, respectively. Subsequent reaction with phenyl isocyanate followed by reduction of the nitrile, afforded final compounds 36 and 37. In addition, the 3-phenoxyphenylcontaining derivative 39 was prepared from 2-chloro-6(trifluoromethyl)isonicotinonitrile 22 and 3-phenoxyphenol via a simple two-step SNAr/reduction procedure (Scheme 2). Optically pure trans-4-fluoro-3-pyrrolidinol-containing derivatives 43 and 44 were prepared as shown in Scheme 3. The aforementioned acid derivative 24 was coupled with racemic trans-4-fluoro-3-pyrrolidinol to give the racemate 40. At the time this work was performed, only racemic trans-4-fluoro-3pyrrolidinol was commercially available. Thus, the individual enantiomers (of 40) were obtained via preparative normalphase chiral HPLC separation, wherein the (R,R)-enantiomer 41 and (S,S)-enantiomer 42 were the first and second to elute, respectively. Subsequent N-Boc deprotection of 41 and 42 afforded optically pure final compounds 43 and 44, respectively. The absolute stereochemistry of 41 was confirmed via its preparation from 24 and optically pure (R,R)-trans-4fluoro-3-pyrrolidinol 2,3-dibenzoyl-D-tartrate 47 (under identical HATU mediated conditions used previously). Optically pure 47 was readily obtained in 94.4% ee from racemate 46 via diastereomeric salt crystallization with 2,3-dibenzoyl-D-tartaric acid (Scheme 4), and the (R,R)-configuration of the 4-fluoro-3pyrrolidinol 47 was confirmed via X-ray crystallography (see Figure 2).
Scheme 4. Preparation of Optically Pure (R,R)-trans-4Fluoro-3-pyrrolidinol 47 via Diastereomeric Salt Crystallizationa
a
Reagents and conditions: (a) NaOH, EtOH, rt, 94%; (b) 2,3dibenzoyl-D-tartaric acid, THF, reflux to rt. Obtained precipitate was repeatedly recrystallized from MeOH; 6% overall yield, 94.4% ee.
■
RESULTS AND DISCUSSION All compounds (Tables 1−3) were profiled for inhibitory activity against human LOXL2 and counter-screened against human LOX. The LOXL2 assay was carried out using serumfree concentrated conditioned media (CCM) from CHO cells stably expressing human LOXL2, and amine oxidase activity was determined via measurement of hydrogen peroxide released during the oxidative deamination of the artificial substrate 1,5-diaminopentane (DAP).18 The pan-LOX(L) inhibitor 1 exhibits time-dependent inhibition of LOXL2 with IC50 values shifting from 0.24 to 0.066 μM when the preincubation period is increased from 15 min to 2 h. We observed no further significant change in potency upon extending the preincubation out to 4 h.16 In order to allow for any time-dependent inhibition with our compounds, amine 4407
DOI: 10.1021/acs.jmedchem.7b00345 J. Med. Chem. 2017, 60, 4403−4423
Journal of Medicinal Chemistry
Article
enzyme, so we explored substitution of the phenyl ring of 6. Numerous derivatives were initially prepared incorporating a variety of substituents on the phenyl ring (not shown). From these, it was found that substitution with a functionalized amide moiety proved particularly productive and that the position of the substituent amide is important for potency. Derivatives incorporating the amide substituent meta to the pyridine Olinker exhibited improved potency for LOXL2 compared to the corresponding para-substituted analogues (compare 9 and 10 with 13 and 14, respectively). This was particularly apparent in the blood assay. For example, the meta-benzamide derivative 14 (LOXL2 CCM IC50 = 0.056 μM; hWB IC50 = 1.3 μM) was 30fold more potent in human whole blood than the corresponding para-benzamide derivative 10 (LOXL2 CCM IC50 = 0.27 μM; hWB IC50 = 39 μM). Additional meta-amide containing analogues demonstrated that substituents of varying sizes and lipophilicities may be incorporated while maintaining good potency for LOXL2. For example, benzyl 15 (LOXL2 CCM IC50 = 0.062 μM) and 2-(methylene)benzothiophene 17 (LOXL2 CCM IC50 = 0.079 μM) derivatives were equipotent to benzamide derivative 14. Furthermore, substitution of the benzamide ring (of 14) with groups of varying electronic character proved well tolerated as, for example, incorporation of either an electron-donating OMe (18) or an electronwithdrawing fluoro (19) had little effect on overall LOXL2 potency or selectivity. Even the incorporation of ionizable groups such as amidine (20) or carboxylic acid (21) proved well tolerated by the enzyme. Overall, these data suggest that the amide is important for potency; however, substituents on the amide nitrogen are not directly involved in critical binding interactions with the enzyme. Rather, this area of the molecule probably sits in a relatively open (solvent exposed) region next to the LTQ-containing active site, which is presumably the same region normally occupied by the protein backbone immediately adjacent to the active lysine or hydroxylysine residues of both collagen and elastin substrates (illustrated in Figure 1c).20 This is the case for the related topaquinone (TPQ) containing copper amine oxidase, Pichia pastoris LOX, which has a broad funnel leading to the active site, thus allowing access to (and subsequent oxidation of) lysine residues in polypeptides.21 The identification of a region within our small molecule scaffold amenable to significant structural variation proved an important finding and ultimately allowed for the efficient modulation and optimization of key physicochemical properties of this series without affecting potency at LOXL2 (vide infra). Despite having identified a series of potent pyridine-derived LOXL2 inhibitors (in both serum-free and human whole blood assays), we felt that the selectivity over LOX (∼30-fold) could be further improved, particularly given the physiological role of LOX in normal healthy tissue.14 Key structural differences between the catalytic domains of LOX and LOXL2 are known. In particular, the catalytic domain of LOXL2 contains a predicted N-glycosylated asparagine residue (N644) close to the LTQ-Cu active site, which is absent in LOX.2,10a This may result in structural differences between the active site binding pockets of LOX and LOXL2, leading to differential binding of small-molecule inhibitors. Using our hypothesis that the pyridine-4-ylmethylamino group (or warhead) of our inhibitors covalently binds to the quinone carbonyl of the active site LTQ (Figure 1b), we reasoned that targeted modification around the warhead may further reduce LOX activity while maintaining potency for LOXL2. Substitution at either the warhead amine nitrogen
Figure 2. X-ray crystal structure of (R,R)-trans-4-fluoro-3-pyrrolidinol 2,3-dibenzoyl-D-tartrate 47.
oxidase activity for both LOXL2 and LOX was measured after a 2 h preincubation with test compounds. To assess the selectivity of compounds versus human LOX, a slightly modified assay was used which required the addition of BSA to the serum-free media from HEK cells stably expressing human LOX in order to obtain measurable activity. The corresponding LOXL2 assay (+ BSA in serum-free conditioned media) was also run in order to generate an accurate head-tohead comparison of selectivity. In addition, LOXL2 activity was assessed in a more clinically relevant human whole blood assay wherein recombinant human LOXL2 and test compounds were incubated for 2 h in whole blood prior to the addition of DAP and subsequent measurement of hydrogen peroxide release. Structure−Activity Relationships of LOXL2 Inhibitors. Compound 2 (Figure 1a and Table 1)16 is an attractive lead for several reasons including good in vitro potency for LOXL2 (IC50 = 0.13 μM), selectivity for LOXL2 over LOX (∼30-fold vs 6-fold for 1), and a modest 10-fold shift in potency in a human whole blood assay (IC50 = 1.5 μM). In addition, the molecular properties of 2 fall well within the “rule of five” (RO5) criteria,19 thus an attractive starting point for allowing efficient access to compounds with excellent physicochemical and drug-like properties. The goals for our SAR campaign included improving LOXL2 potency and selectivity, generating a compound with acceptable bioavailability to demonstrate in vivo efficacy in a mouse model of fibrosis, and, ultimately, selecting a clinical development candidate. Initial studies focused on the 2-position of the pyridine, as summarized in Table 1. We first investigated the replacement of the 2-Cl moiety (of 2) with other small groups of varying electronic and lipophilic character such as CF3 (48), OMe (49), or Ocyclohexyl (50). This led to a 6 to 9-fold drop in potency for LOXL2, although selectivity over LOX was improved for the more electronegative CF3-containing derivative 48 (40-fold selective). Potency in human whole blood was also significantly reduced, particularly for both OMe (49, hWB IC50 = 35 μM) and O-cyclohexyl (50, hWB IC50 = 29 μM) derivatives. However, replacement of O-cyclohexyl with an aromatic Ophenyl (6) led to improved potency at LOXL2 (versus 49 and 50), particularly in blood where 6 (hWB IC50 = 6.7 μM) was ∼4-fold more potent than 50. One hypothesis is that the phenyl group is picking up a favorable interaction with the 4408
DOI: 10.1021/acs.jmedchem.7b00345 J. Med. Chem. 2017, 60, 4403−4423
Journal of Medicinal Chemistry
Article
Table 1. SAR of 4-(Methylamino)pyridine-Derived LOXL2 Inhibitors
a
Assay measuring inhibition of H2O2 produced from the oxidative deamination of the substrate, DAP, after a 2 h preincubation with test compound. Reported IC50 values are the geometric mean values ×/÷ geometric standard deviation from at least 3 separate experiments. bCarried out using concentrated conditioned media (CCM) from CHO cells stably expressing hLOXL2. cAssay performed using CCM containing 0.1% BSA prior to concentration. dCarried out using CCM containing 0.1% BSA prior to concentration and collected from HEK cells stably expressing hLOX. eCarried out using fresh human blood spiked with purified recombinant hLOXL2. 4409
DOI: 10.1021/acs.jmedchem.7b00345 J. Med. Chem. 2017, 60, 4403−4423
Journal of Medicinal Chemistry
Article
Table 2. SAR of 4-(Methylamino)-6-(trifluoromethyl)-pyridine-Derived LOXL2 Inhibitors
a
Assay measuring inhibition of H2O2 produced from the oxidative deamination of the substrate, DAP, after a 2 h preincubation with test compound. Reported IC50 values are the geometric mean values ×/÷ geometric standard deviation from at least 3 separate experiments. bCarried out using concentrated conditioned media (CCM) from CHO cells stably expressing hLOXL2. cAssay performed using CCM containing 0.1% BSA prior to concentration. dCarried out using CCM containing 0.1% BSA prior to concentration and collected from HEK cells stably expressing hLOX. eCarried out using fresh human blood spiked with purified recombinant hLOXL2. fC57BL/6 mice (n = 3 per time point) dosed orally (po) with 30 mg/kg of test compound (formulated in 0.5% methylcellulose). Reported values are the means from 3 animals. gCalculated using CDD Vault software from Collaborative Drug Discovery, Burlingame, CA. hND indicates data not determined. 4410
DOI: 10.1021/acs.jmedchem.7b00345 J. Med. Chem. 2017, 60, 4403−4423
Journal of Medicinal Chemistry
Article
Table 3. Selectivity versus Other Amine Oxidases for LOXL2 Inhibitors 28, 43, and 44
a
Assay measuring the inhibition of H2O2 produced from the oxidative deamination of the substrate, DAP, after a 2 h preincubation with test compound. Reported IC50 values are the geometric mean values ×/÷ geometric standard deviation from at least 3 separate experiments. bCarried out using fresh human blood spiked with purified recombinant hLOXL2. cCarried out using concentrated conditioned media (CCM) from CHO cells stably expressing hLOXL2. dAssay performed using CCM containing 0.1% BSA prior to concentration. eCarried out using CCM containing 0.1% BSA prior to concentration and collected from HEK cells stably expressing hLOX. fCarried out using purified recombinant hLOXL3. gAssay carried out using human plasma containing endogenous SSAO with benzylamine as substrate. Percent inhibition was determined after a 2 h preincubation with test compound and calculated relative to both vehicle control and control containing 1 μM PXS-4681A32 (a known SSAO inhibitor). hAssay carried out using purified recombinant hDAO with DAP as substrate. Percent inhibition was determined after a 2 h preincubation with test compound and calculated relative to both vehicle and enzyme-free controls. iAssay carried out using either purified recombinant hMAO-A or hMAO-B with tyramine and benzylamine as substrate, respectively. Percent inhibition was determined after a 2 h preincubation with test compounds and calculated relative to both vehicle and enzyme-free controls.
amide substituents, and the results are shown in Table 2 where compounds are listed in order of decreasing lipophilicity (cLogP of 5.1 to 0.8). Most compounds proved potent for LOXL2 under serum-free conditions with IC50 values between 0.034 and 0.11 μM. An outlier was compound 36 (CCM IC50 = 0.57 μM) which contains the (3R)-piperidine ring in place of phenyl. However, potency was restored with the corresponding (3S)-piperidine enantiomer 37 (CCM IC50 = 0.13 μM), suggesting that relative spatial orientation of both pyridine and piperidine rings in the enzyme binding pocket is important. As predicted, the decrease in lipophilicity was accompanied by a significant improvement in whole blood potency for LOXL2. For instance, for cLogP values ∼5 (compounds 26 and 39) the blood shift is ∼300-fold. When cLogP values are reduced by just 1 log unit to 4.3 (25), the blood shift is reduced to ∼70fold, and for compounds with cLogP values 90% inhibition of LOXL2 activity (solid bars, Figure 3a). Washing, followed by a 2 h equilibration, in most cases, led to minimal (10 to 20%) recovery of LOXL2 activity (open bars, Figure 3a) when compared with the corresponding nonwashed controls, thus indicating that these are irreversible inhibitors of LOXL2. The exception was the initial lead 4-(methylamino)pyridine 2, which was found to be largely reversible at LOXL2, exhibiting ∼65% recovery of enzyme activity after a 2 h postwash equilibration.16 We speculate that the observed difference in reversibility between 2 and t he 2-(phenoxy)-6(trifluoromethyl)pyridine containing inhibitors is due to the presence of additional bulky substituents, which upon formation of the initial substrate Schiff base complex can participate in additional noncovalent binding interactions further facilitating the formation and stabilization of a more tightly bound product Schiff base complex.16 Given that increased levels of LOXL2 have been observed in lung tissue
intermediate (see III, Figure 1b), which can rearrange to a more stable enamine product (V, Figure 1b) potentially allowing for an additional covalent interaction between the nitrile and a nearby nucleophilic residue in the LOX active site.13b Our inhibitors are likely to form a Schiff base complex with LOXL2; however, these inhibitors are unlikely to form additional covalent bonds with the enzyme as proposed in V (Figure 1b). Instead, we believe multiple noncovalent binding interactions help drive the formation of a tightly bound Schiff base complex and irreversibly lock down the enzyme in an inactive form. A major advantage of irreversible inhibition is the potential for an extended pharmacodynamic (PD) response that outlasts systemic drug levels (PK). This assumes that the target protein is not rapidly regenerated in vivo at the site of action. Such minimization of overall exposure in vivo helps limit any off-target activities and can lead to an improved safety profile and higher therapeutic index.25 Select compounds, including 1, were assessed in a human LOXL2 reversibility assay, and the results are shown in Figure 3a. Briefly, purified recombinant human LOXL2 bound to the bottom of 96-well plates was incubated with the inhibitor compound (1000 × IC50 concentration) for 2 h at 37 °C. After repeated washing with PBS to remove any unbound inhibitor, incubation was 4412
DOI: 10.1021/acs.jmedchem.7b00345 J. Med. Chem. 2017, 60, 4403−4423
Journal of Medicinal Chemistry
Article
Prophylactic treatment with 28 dose-dependently reduced lung fibrosis in mice as evidenced by a reduction in mean Ashcroft score (Figure 4). Dosing at 3 mg/kg QD had little effect on
and serum samples from patients with IPF and that LOXL2 has been implicated as a pro-fibrotic mediator of fibrosis in murine lung fibrosis models,10b,26 the mouse oropharyngeal bleomycin lung fibrosis model was chosen as our primary in vivo efficacy screen. A key consideration for advancing candidates into in vivo studies is ensuring that appropriate drug concentrations are achieved in tissues of interest. Normal C57BL/6 mice were administered the compound as a single oral dose (30 mg/kg), and both lung and plasma concentrations were assessed over time (Cmax data are shown in Table 2). Compounds 28 and 37 showed the greatest concentrations in lung tissue with Cmax values of 21 and 30 μM, respectively, with compounds 25, 27, and 29 also exhibiting reasonable lung concentrations (Cmax ∼10 μM). However, compounds 30 through 33 exhibited significantly lower lung concentrations; in particular, 33 resulted in an average lung concentration of only 0.54 μM. In general, the lung to plasma ratios proved greater for compounds of a more lipophilic nature. Compounds with cLogP values ≥2 (25, 27, 28, and 37) displayed between 7- to 16-fold higher maximal concentrations in lung tissue versus plasma, whereas compounds with cLogP < 2 (29 through 33) had lung concentrations only 0.5 to 1.8 times that seen in plasma. It is known that the lungs are a site for both uptake and accumulation of certain drugs. In particular, lipophilic amines with pKa values ≥8 often exhibit significant pulmonary uptake (for example, fentanyl and propranolol).27 The pKa of the aminomethyl group of these 6-CF3-pyridine derivatives is ∼8,28 which is in line with known lung-penetrable compounds. The above data indicate that for this series of basic amine-containing inhibitors, a cLogP of at least 1.8 is required to ensure good distribution to mouse lung. Furthermore, the combined data set presented in Table 2 shows that by “tuning” overall lipophilicity, one can readily access compounds with both optimal lung exposure and high potency in human whole blood. This suggests that, for this series, the lipophilic sweet spot sits in a narrow range between cLogP of ∼1.8 to 3, and compound 28 (cLogP = 2.0) emerged as the lead compound suitable for advancement in to the mouse lung bleomycin model. Additionally, we measured the potency of 28 in mouse blood to ensure that the concentrations observed in mouse lung tissue would be sufficient to fully inhibit lung LOXL2. The mouse blood LOXL2 IC50 and IC90 values were 0.86 μM and 5.6 μM, respectively, which is very similar to the human blood IC50 and IC90 values of 0.87 μM and 5.4 μM, respectively. Therefore, we reasoned that a single 30 mg/kg dose should be sufficient to cover the LOXL2 IC90 in mouse lung for at least 2 h, thus allowing adequate time for irreversible binding to the enzyme. Evaluation of Compound 28 in the Mouse Bleomycin Lung Fibrosis Model. Lung fibrosis was induced in 6 to 7week old male C57BL/6 mice via the oropharyngeal administration of a single dose of bleomycin (1.5 U/kg). Compound 28 was administered orally (PO) once daily (QD) at 3, 10, 30, or 60 mg/kg for a total of 15 days, beginning 1 day prior to the instillation of bleomycin (prophylactic dosing regimen). At day 14 post-bleomycin administration, lungs from each animal were removed for histological examination of lung slices stained with H&E and Masson’s trichrome and an Ashcroft score of fibrosis and inflammatory damage determined.29 All doses were well tolerated with no mortality or significant body weight loss (400-fold). For the more closely related LOXL isoforms, activity was only evaluated against LOXL3, where both compounds exhibit IC50 values of ∼1.2 μM. Attempts to generate stable cell lines for LOXL1 and LOXL4 were unsuccessful, so potencies against these two enzymes were not determined. Compounds 43 and 44 also proved highly selective for LOXL2 over other key members of the amine oxidase family, including the copperdependent amine oxidases semicarbazide-sensitive amine oxidase (SSAO) and diamine oxidase (DAO), as well as the flavin-dependent monoamine oxidases A (MAO-A) and B (MAO-B). They all showed
