Orelabrutinib

Discovery of Evobrutinib: An Oral, Potent, and Highly Selective, Covalent Bruton’s Tyrosine Kinase (BTK) Inhibitor for the Treatment of Immunological Diseases

Richard D. Caldwell, Hui Qiu, Ben C. Askew, Andrew T. Bender, Nadia Brugger, Montserrat Camps, Mohanraj Dhanabal, Vikram Dutt, Thomas Eichhorn, Anna S. Gardberg, Andreas Goutopoulos, Roland Grenningloh, Jared Head, Brian Healey,⊥ Brian L. Hodous,∥ Bayard R. Huck, Theresa L. Johnson, Christopher Jones, Reinaldo C. Jones, Igor Mochalkin, Federica Morandi, Ngan Nguyen, Michael Meyring, Justin R. Potnick, Dusica Cvetinovic Santos, Ralf Schmidt, Brian Sherer, Adam Shutes, Klaus Urbahns, Ariele Viacava Follis, Ansgar A. Wegener, Simone C. Zimmerli, and Lesley Liu-Bujalski
EMD Serono Research & Development Institute, Inc. (a Business of Merck KGaA, Darmstadt, Germany), 45 A Middlesex Turnpike, Billerica, Massachusetts 01821, United States

■ INTRODUCTION
Bruton’s tyrosine kinase (BTK) is a member of the Tec family
of tyrosine kinases and is expressed in B cells, macrophages, and monocytes but not in T cells.1 BTK plays a crucial role in signaling through the B cell receptor (BCR) and the Fcγ receptor (FcγR) in B cells and myeloid cells, respectively.2 Indeed, mutations in the human Btk gene can result in the B cell specific immunodeficiency, X-linked agammaglobulinemia, which is characterized by a block in pre-B cell differentiation; this leads to a lack of B cells and plasma cells and markedly reduced levels of serum immunoglobulins.3 Likewise, Btk mutations in mice with X-linked immunodeficiency and targeted deletion of Btk in knockout mice result in defects in B cell development and proliferation.3,4 However, pharmaco-diseases involving B cell and/or macrophage activation such as B cell malignancies, asthma, rheumatoid arthritis, systemic lupus erythematosus, and multiple sclerosis.6,7
The small molecule irreversible BTK inhibitor, ibrutinib, which covalently targets Cys481 within the ATP-binding pocket, is approved for the treatment of mantle cell lymphoma (MCL), chronic lymphocytic leukemia (CLL)/small lympho- cytic lymphoma, Waldenström’s macroglobulinemia, and marginal zone lymphoma, as well as graft versus host disease.8 Ibrutinib demonstrated clinical efficacy in previously difficult- to-treat lymphomas such as treatment-resistant/refractory CLL.8,9 However, this first in class BTK inhibitor has also shown side effects, including bleeding, rash, diarrhea, and atrial fibrillation, which have been attributed, in part, to off-target logical inhibition of BTK has not resulted in depletion of B cells in preclinical models.5 Consequently, BTK is considered a promising therapeutic target for the treatment of various kinase-inhibitory related effects on the epidermal growth factor receptor (EGFR) and other Tec family kinases.10−13
With the aim of improving the overall tolerability of BTK inhibitors while maintaining the efficacy of ibrutinib, new BTK inhibitors are being developed that have more refined pharmacologic profiles.13 Additionally, these agents are designed to circumvent the inhibitory effect of ibrutinib on antibody-dependent cell-mediated cytotoxicity, which may be important for the activity of other oncology therapies administered concomitantly with a BTK inhibitor.13 In 2017, acalabrutinib became the initial second-generation, covalent BTK inhibitor to be approved by the FDA for the treatment of MCL.9,14 However, for chronic autoimmune indications, clinical development progress has lagged behind that of oncology indications. Recently, there has been increased interest in the development of BTK inhibitors in the autoimmune/inflammatory arena.
Herein, we describe work which led to the identification of the potent and irreversibly covalent BTK15 inhibitor, evobrutinib (A18, M2951), for the treatment of autoimmune diseases. Evobrutinib has excellent kinome selectivity, partic- ularly against EGFR, an acceptable preclinical pharmacoki- netic/pharmacodynamic (PK/PD) profile, and is efficacious in in vivo efficacy models of rheumatoid arthritis. Evobrutinib is being evaluated in clinical development for treatment of various immunological diseases.16

RESULTS AND DISCUSSION
Medicinal Chemistry Strategy. The target profile for a development candidate for chronic administration in auto- immune diseases included potency, high BTK kinase- selectivity, particularly against EGFR, and moderate clearance. It was hypothesized this profile would result in a low daily dose which should minimize the risk of liver damage; as covalent binding, combined with a high daily dose, has been reported to be linked to drug-induced liver injury.17
To achieve BTK potency and kinase selectivity simulta- neously, we employed multiple strategies. First, it was desired to have an obligate type covalent inhibitor in which the covalent binding is essential for the potency of the molecule. Additionally, targeting Cys481 with a warhead to form a covalent bond enhances selectivity because only 10 other kinases, BLK, BMX, EGFR, ERBB2, ERBB4, ITK, JAK3, TEC, MKK7, and TXK, contain cysteine in the same region of the receptor. Finally, interaction with the gatekeeper Thr474 further increases selectivity as only 20% of all human kinases have threonine as a gatekeeper residue.18,19
Identification of a Novel Hinge Binder. Work to identify suitable chemical starting points began with analysis of the X-ray crystal structure of B43 (Figure 1, PDB 3gen).18,20 B43 is a moderately potent inhibitor of BTK and occupies the ATP binding pocket of the BTK kinase domain. The amino- pyrimidine group, acting as a hinge binder, forms interactions with the gatekeeper residue Thr474 as well as backbone carbonyl and amide N interactions with Glu475 and Met477, respectively. The O-linked phenoxy phenyl group occupies the long and narrow selectivity pocket adjacent to gatekeeper Thr474, and the lipophilic cyclopentyl group sits in the ribose pocket, which is near Cys481, providing an indication of the proper vectors and angles for the desired covalent connection. Structure−activity relationship (SAR) exploration focused on replacing the pyrrolopyrimidine of B43 with other similarly substituted hinge binding moieties bearing aryl linkers appended with an acrylamide warhead. This work revealed that several hinge binding moieties were tolerated for potent BTK inhibitory activity (Table 1). The pyrimidine analogue A3 showed excellent inhibition of BTK in a biochemical assay,21 while the removal of the 2-amino moiety led to nearly 10-fold loss in potency in compound A4. Pyridine analogue A5 also showed excellent BTK inhibitory activity, and both A3 and A5 showed 25−30-fold more potent inhibition of BTK versus EGFR. However, A5 had unacceptable activity in the hERG (human ether-a-go-go-related gene) channel assay (Supporting Information). In contrast to A3 and A5, pyrimidine analogues such as A7, with opposite orientation of the hinge binder, did not provide an acceptable ratio of inhibition of BTK over EGFR and were subsequently deprioritized. In addition, analysis of the crystal structures of A5 and A7 demonstrated that these hinge binders both achieve interactions in the identified hinge region while maintaining the critical vectors and angles required to effectively engage the back pocket and Cys481 as covalent BTK inhibitors (Figure 2). Finally, attaching the linker to the hinge binder through an N atom as in compound A8 also provided good inhibition of BTK but was equipotent at inhibiting EGFR. On the basis of these data, the amino-pyrimidine hinge binder with the amino moiety oriented adjacent to gatekeeper residue Thr474, as in A3 and A5, was selected for further exploration.
Optimization of the Linker. Further characterization of the hit molecule A3 showed that, although it demonstrated good potency in the human peripheral blood mononuclear cell (PBMC) assay21 (IC50 = 9.9 nM), this molecule had poor solubility (<2.4 μM) (Table 2) and poor metabolic stability (high intrinsic clearance, Clint = 277 μL/min/mg) in human liver microsomes. Subsequent efforts focused on modification of A3 to improve solubility and metabolic stability. Preliminary SAR and X-ray cocrystal information suggested that the modification of the linker moiety might be amenable for addressing low solubility of the series. Replacing the flat aromatic ring in compound A3 with a saturated moiety, such as the diamino cyclohexane ring in analogue A9, showed good improvement in solubility versus A3 (134 vs 2.4 μM) and metabolic stability (Clint = 27 vs 277 μL/min/mg), without significant loss in enzyme potency. Various saturated linkers were evaluated with the goal of improving potency in the enzymatic assay (<10 nM). Modifications such as moving the amino group around the cyclohexyl ring, changing the size of the ring or adding a bridge unit into the linker, maintained the high solubility, however, did not lead to any meaningful boost in potency for A10−A12. Continued exploration of the linker moiety showed that attachment of the linker to the hinge binding core via a secondary amine, such as in compound A13, was not tolerated. It was hypothesized that this substitution caused steric collision between the linker and the protons at the proximal phenyl ring, which altered the vector of the linker toward the Cys481. In contrast, analogue A14, with the reversed linker, was 220-fold more potent in enzymatic assay but suffered from a poor human microsomal stability. Additional exploration revealed that the pyrrolidine linker in A15 was less well tolerated. Introduction of a spiral linker boosted the potency in molecule A16. Unfortunately, A16 showed poor microsomal stability and an unacceptable hERG profile. Shifting the N atom in A14 as depicted in A17 led to a more than 200-fold reduction in enzymatic potency (IC50 1510 nM for A17 vs 6.5 nM for A14). Insertion of one methylene group into A17 led to the discovery of A18 (evobrutinib). Evobrutinib A18 showed excellent potency in the biochemical and PBMC assays, favorable permeability in the Caco-2 assay, good solubility and microsomal stability, and acceptable hERG inhibition (see Supporting Information for assay methods). Finally, a hydroxyl group was added in molecule A19 to reduce lipophilicity. Back Pocket Optimization. Additional modifications directed to the back pocket to further improve microsomal stability of A18 (evobrutinib) proved to be unsuccessful (Table 3). Substitutions at the proximal phenyl ring (A20,A21) resulted in a dramatic loss in potency on BTK inhibitory activity. No improvement in metabolic stability was observed from modifications of the distal phenyl ring (A22− A25). Only analogues A23 and A24 displayed similar BTK inhibition compared to A18, both with no obvious improve- ment in clearance. Subsequently, efforts to optimize the back pocket were not pursued further. Warhead Optimization. Identification of an optimal warhead to engage the Cys481 residue was driven by the goal of designing an obligate covalent inhibitor of BTK. A diverse array of warheads was evaluated to optimize potency, reactivity, and plasma stability and to identify a suitable covalent inhibitor for further in vivo studies (Table 4). Analogues bearing an acrylamide warhead such as that of ibrutinib19 displayed acceptable biochemical and cellular potency, as exemplified in analogue A18. Replacement of the acrylamide of A18 by the acetamide moiety in compound A27 resulted in an analogue with no significant BTK inhibitory activity, demonstrating the requirement of a warhead for efficient inhibition of BTK. Compounds such as A28−A30, bearing warheads that are theoretically highly reactive,22,23 did show excellent enzyme inhibitory activity but suffered from unacceptable plasma stability. These results were attributed to the high reactivity of the warheads. It was speculated that these compounds were binding nonspecifically to cellular components before they were able to reach the target. Substitution with Michael acceptors that are theoretically less reactive than acrylamide22,23 led to a decrease in BTK inhibitory activity compared to compound A18, as seen with analogue A32. On the basis of these data, compound A18, bearing an acrylamide warhead, possessed the optimal combination of potency, reactivity, and plasma stability to be selected for additional profiling. The obligatory nature of covalent inhibition with A18 was further confirmed by comparing its activity with that of A27 against a mutant BTK kinase domain with a cysteine to serine substitution, which resulted in a lack of covalent binding by either molecule. Both compounds inhibited the BTK mutant with similarly low potency (IC50 8.6−8.7 μM), reflecting their reversible affinities for the BTK kinase domain without any covalent binding. Thus, while A27 showed minimal inhibitory activity regardless of Cys481 binding, the approximately 1000-fold increase in potency of A18 against the wild-type BTK kinase domain demonstrated that its potency is derived from covalent engagement of Cys481. Both A18 and A19 were further characterized in PK/PD studies in mice. PK/PD Studies. To confirm whether the in vitro effects with A18 and A19 translated into B cell inhibition in vivo, these compounds were tested head-to-head in a B cell stimulation assay in normal C57BL/6 mice.24 Both com- pounds were dosed at 1 mg/kg orally. Compound A19 inhibited B cell activation by 42% and 15% at 1 and 24 h after dosing, respectively (Figure 3). Compound A18 (evobrutinib) achieved a greater B cell inhibition of 71% and 25% at 1 and 24 h. This was despite a lower maximum plasma concentration (Cmax) of A18 compared to A19 (52 versus 257 nM). These data showed that A18 was a more potent B cell inhibitor than A19 in vivo. Synthesis of Evobrutinib (A18). The synthesis of evobrutinib (A18) began with regioselective nucleophilic substitution of the 6-chloride of 5,6-dichloropyrimidin-4- amine by the primary amine of tert-butyl 4-(aminomethyl)- piperidine-1-carboxylate, which led to generation of tert-butyl 4-{[(6-amino-5-chloropyrimidin-4-yl)amino]methyl}- piperidine-1-carboxylate (Figure 4). Subsequent Suzuki coupling of this chloropyrimidine intermediate with phenox- yphenylboronic acid yielded the molecule tert-butyl 4-({[6- amino-5-(4-phenoxyphenyl)pyrimidin-4-yl]amino}methyl)- piperidine-1-carboxylate. Deprotection under acidic conditions and subsequent neutralization afforded the piperidine deriva- tive 5-(4-phenoxy-phenyl)-N-piperidin-4-ylmethyl-pyrimidine- 4,6-diamine. Finally, the acrylamide was introduced to give the target compound N-[(1-acryloylpiperidin-4-yl)methyl]-5-(4- phenoxyphenyl)pyrimidine-4,6-diamine (A18; evobrutinib; step d). Characterization of Evobrutinib: Biophysical Con- firmation of Covalent, Irreversible Binding. Covalent binding to Cys481 is confirmed by analysis of the X-ray crystal structure of evobrutinib bound to the BTK kinase domain, showing continuous electron density between Cys481 and the evobrutinib warhead (Figure 5). Mass Spectrometry. Reversed-phase (RP)−LC−MS analysis of intact BTK and BTK plus evobrutinib further confirmed the covalent nature of evobrutinib binding and indicated 1:1 stoichiometry (Figure S1 in the Supporting Information). Thus, evobrutinib is bound to only one cysteine residue on the BTK kinase domain. Orthogonal confirmation of evobrutinib−BTK covalent binding with 1:1 stoichiometry was also achieved by isothermal calorimetry (see Figure S2 in the Supporting Information). Jump Dilution Assay. BTK IC50 was measured routinely for each compound to drive medicinal chemistry optimization. In addition, a jump dilution assay was used to qualitatively check the irreversible nature of key compounds (see Supporting Information). Lack of recovery of enzymatic activity was consistent with irreversible inhibition of BTK by evobrutinib (Supporting Information, Figure S3). In general, time-dependent BTK IC50 correlated well with the Kinact/Ki ratio (where Ki is the affinity of initial noncovalent binding, and Kinact is the maximum potential rate of enzyme inactivation), which represents the second-order rate constant for the reaction of evobrutinib with BTK (Figure S4 and Table S1 in the Supporting Information).25 Evobrutinib Kinase Selectivity. The high kinase selectivity of evobrutinib is derived from multiple factors. Its obligate covalent binding to a rare cysteine residue and interaction with a unique threonine gatekeeper-mediated selectivity pocket in structurally related kinases accounts for much of the agent’s high kinase selectivity. For kinases with a similarly situated cysteine residue, selectivity can be explained by examination of the primary structure. The tyrosine kinase BMX, whose sequence identity is the most similar to that of BTK, presented an insurmountable challenge in achieving selectivity (Table S2 in the Supporting Information). The Tec kinase ITK also shares a high degree of homology to BTK but contains a bulky gatekeeper phenylalanine residue, which explains the higher observed selectivity of evobrutinib for BTK over ITK. Conversely, other cysteine containing kinases that are not inhibited by evobrutinib diverge more structurally from BTK, BMX, and ITK (Figure S5 in the Supporting Information). Evobrutinib’s selectivity for BTK over EGFR is explained by analysis of the global differences between the structures of the two cysteine-containing kinases. Upon binding to BTK, evobrutinib and ibrutinib resulted in a relatively small displacement of N- and C-lobes in relation to one another. On the other hand, EGFR showed a dramatic rearrangement in the N-lobe when targeted by ibrutinib (Figure S5 in the Supporting Information). It is hypothesized that the 1,3- disubstituted pattern on the piperidine ring in ibrutinib allows the warhead to rotate to adapt to kinases that move their N lobes.24 In contrast, the evobrutinib warhead is more constrained, having a 1,4-substitution on the piperidine linker and thus cannot accommodate kinases that upon binding rearrange their N lobe, such as EGFR. This is demonstrated in Table 5, which shows that selectivity for BTK over EGFR was lost in A14 which, similar to ibrutinib, had a 1,3 attachment to the piperidine ring. The higher kinase selectivity of evobrutinib compared with ibrutinib is illustrated in Table 6. Ibrutinib inhibited seven Cys481-containing kinases compared with only two for evobrutinib. Unlike evobrutinib, ibrutinib does not require covalent binding for potency, as it showed only an approximately 10-fold reduction in potency against the Cys481S mutant compared to wild-type BTK. The ibrutinib scaffold, lacking the acrylamide warhead, showed similar potency toward BTK and comparable selectivity against 236 kinases tested. Preclinical Pharmacokinetic Studies. Evaluation of evobrutinib PK in mice, rats, and dogs indicated that evobrutinib is rapidly absorbed after oral administration, with Cmax reached between 0.25 and 1 h (Table S3 in the Supporting Information). Evobrutinib demonstrated moderate clearance in mice and high clearance in rats and dogs. The plasma half-life was short (∼1 h) in all animal species. The oral bioavailability was low (<5% in rats) to moderate (up to 25% in mice) restricted by hepatic and possibly intestinal first-pass metabolism. Activity of Evobrutinib in Collagen-Induced Arthritis. Evobrutinib was tested in a rat collagen-induced arthritis (CIA) model. Female Lewis rats with semiestablished type II collagen arthritis26,27 were treated once daily for 11 days (days 6−16) by oral gavage with vehicle (20% hydroxy-propyl-β- cyclodextrin in H2O), evobrutinib (0.3, 1, 3, 10, or 30 mg/kg), or the reference compound methotrexate (MTX, 0.1 mg/kg). Animals were terminated on study day 17. Efficacy evaluation was based on daily ankle caliper measurements and histopathologic evaluation of ankles and knees. Disease-induced ankle-swelling was significantly reduced (P ≤ 0.05) for rats treated with evobrutinib 3 mg/kg (days 11− 17), 10 mg/kg (days 11−17), 30 mg/kg (days 10−17), or MTX (days 11−17), compared with vehicle (Figure 6A). Ankle histopathology scores were also reduced with evobrutinib in a dose-dependent manner compared to vehicle (Figure 6B). The IC50 for evobrutinib was 5.180 mg/kg. ▪ CONCLUSIONS In summary, we have described the discovery and preclinical characterization of Orelabrutinib, a potent, obligate covalent BTK inhibitor with high kinase selectivity. Evobru- tinib was derived from design and subsequent optimization of analogues based on the analysis of the crystal structure of the BTK inhibitor, B43. Evobrutinib displayed suitable preclinical pharmacokinetic and promising pharmacodynamic character- istics. Evobrutinib also demonstrated efficacy in a rat model of rheumatoid arthritis.16 Moreover, the high selectivity of evobrutinib for BTK over EGFR and other Tec family kinases indicates that it may have a low potential for off-target related adverse effects. The results of ongoing clinical trials in various autoimmune diseases will enhance our understanding of the efficacy and safety of evobrutinib.