Discovery of a Potent and Selective Covalent Inhibitor of Bruton’s Tyrosine Kinase with Oral Anti-Inflammatory Activity

Abstract

Bruton’s tyrosine kinase (BTK) is a cytoplasmic tyrosine kinase that plays a critical role in the activation of B cells, macrophages, and osteoclasts. Given the key role of these cell types in the pathology of autoimmune disorders, BTK inhibitors have the potential to improve treatment outcomes in multiple diseases. Herein, we report the discovery and characterization of a novel potent and selective covalent 4-oxo-4,5-dihydro-3H-1-thia-3,5,8-triazaacenaphthylene-2-carboxamide BTK inhibitor chemotype. Compound 27 irreversibly inhibits BTK by targeting a noncatalytic cysteine residue (Cys481) for covalent bond formation. Compound 27 is characterized by selectivity for BTK, potent in vivo BTK occupancy that is sustained after it is cleared from systemic circulation, and dose-dependent efficacy at reducing joint inflammation in a rat collagen-induced arthritis model.

Rheumatoid arthritis (RA) is a prototypical autoimmune disease with both genetic and environmental triggers characterized by the systemic dysregulation of the immune system resulting in an autoimmune response to self-antigens.¹ This break in self-tolerance ultimately leads to disease of the joints, including destruction of cartilage and bone.

Bruton’s tyrosine kinase (BTK) is a nonreceptor tyrosine kinase of the Tec family of kinases.² It is primarily expressed in B cells and myeloid cells such as monocytes, macrophages, neutrophils, and mast cells.³ BTK is a critical kinase involved in the signaling pathway of both B cell (BCR) and Fc receptors, thus playing a key role in B cell activation and inflammation triggered by immune complexes.⁴ By modulating both adaptive and innate immune responses, pharmacological inhibition of BTK may provide superior therapeutic benefit in autoimmune diseases such as RA compared to blockade of B cells alone.

BTK contains a pleckstrin homology (PH) domain, a Tec homology (TH) domain, Src homology (SH2 and SH3) domains, and a C-terminal catalytic domain. Upon antigen binding to the BCR, phosphorylation of the cytoplasmic tail occurs by SYK or SRC family kinases, which provides a docking site for BTK. Binding to phosphatidylinositol triphosphate (PIP3) via the PH domain recruits this cytoplasmic kinase to the membrane, where BTK autophosphorylates and then initiates signaling.⁵ Irreversible inhibitors have been described by binding the ATP-binding site of the C-terminal kinase domain and covalently trapping a noncatalytic cysteine residue (Cys481).⁶

There are multiple reported covalent and noncovalent inhibitors of the BTK kinase domain (Figure 1).^7,8 Ibrutinib (1) is a first-in-class irreversible BTK inhibitor that has been approved for the treatment of B-cell cancers such as mantle cell lymphoma, chronic lymphocytic leukemia, Waldenström’s macroglobulinemia, and marginal zone lymphoma.⁹ Two additional BTK inhibitors, acalabrutinib¹⁰ (2) and zanubrutinib¹¹ (3), have also been approved by the United States Food and Drug Administration for the treatment of mantle cell lymphoma. Although there has been marked clinical success with BTK inhibitors for the treatment of cancer, none have been approved for the treatment of autoimmune diseases such as RA. There are several reversible and irreversible BTK inhibitors in clinical evaluation for autoimmune diseases, including fenebrutinib (4) for RA;¹² branebrutinib¹³ (5) for RA, systemic lupus erythematosus, and Sjogren’s Syndrome (SjS); evobrutinib (6) for multiple sclerosis and RA;¹⁴ tirabrutinib¹⁵ (7) for SjS; and remibrutinib (8, Luo-064)¹⁶ for asthma and SjS (Figure 1). The results of a randomized, double-blind phase 2 trial in RA with fenebrutinib demonstrated similar efficacy to adalimumab (Humira) and a safety profile similar to current immunomodulatory therapies, which is promising for BTK inhibitors for the treatment of RA and other autoimmune diseases.¹²

Figure 1.

Structures of BTK inhibitors approved for cancer treatment (compounds 1–3) and BTK inhibitors in clinical trials for autoimmune diseases that inhibit BTK by reversible (compound 4) and irreversible (compounds 5–8) mechanisms.

Covalent inhibition of kinases by targeting noncatalytic cysteine residues is a validated strategy for achieving sustained target engagement without requiring high systemic drug exposure.^17,18 The BTK Cys481 is proximal to the ATP binding site, and an analogous cysteine residue is present in only ten other human kinases.^19,20 The relatively low prevalence of the corresponding Cys residue in the human kinome makes covalent inhibition of BTK an attractive strategy for achieving high selectivity. Furthermore, the estimated rate of BTK resynthesis following irreversible inhibition in chronic lymphocytic leukemia patients is relatively slow, ranging from 3.6 to 31.4% per day,²¹ and provides the opportunity to achieve a prolonged pharmacodynamic effect in vivo following covalent inhibition.²² Covalent BTK inhibitors acalabrutinib, branebrutinib, and tirabrutinib engage the Cys481 with a butynamide electrophile while ibrutinib, zanubrutinib, evobrutinib, and remibrutinib utilize an electrophilic acrylamide.

Here, we report the design and biological characterization of a new series of covalent irreversible BTK inhibitors. For structure–activity relationships, the potency was measured as the second order rate constant k_inact/K_I ratios. Unlike IC₅₀ values, k_inact/K_I ratios are independent of preincubation times and substrate concentrations and are considered the best measure of potency for irreversible inhibitors.²³ We focused on optimizing the noncovalent binding potency (K_I) by engaging lipophilic residues in the BTK back pocket and adjusting the linker to the electrophile to increase potency without increasing the reactivity of the electrophile, which could compromise proteome-wide selectivity. This work led to the identification of compound 27, a potent covalent BTK inhibitor with high kinome selectivity and efficacy in vivo, attenuating paw inflammation when dosed therapeutically after disease onset in a rat collagen-induced arthritis model of RA.

Establishing a Covalent Inhibitor Chemotype

Compound 9 (Table 1) was identified in the Janssen compound collection as an ATP-competitive inhibitor of BTK displaying micromolar affinity in a Lanthascreen binding assay. A docking model of 9 bound to BTK shows a single point hinge interaction in which the Met477 backbone forms a hydrogen-bond with the aromatic nitrogen, and the phenyl group is directed toward the gatekeeper residue Thr474 (Figure 2). The piperidine amine in 9 projects toward the Cys481. It was envisioned appending an acrylamide via the introduction of a linker that positions an electrophile in proximity to Cys481 would convert 9 into a potent and selective irreversible inhibitor of BTK. Molecular modeling estimated a distance of 5 Å from the piperidine nitrogen to the sulfur atom of Cys481, along a vector that could be spanned by an acrylamide electrophile.

Table 1. Identification of a Potential Covalent BTK Inhibitor Chemical Series.

^an = 1.

Figure 2.

Docking model of compound 9 bound to BTK defines the vector and distance to Cys481 (for clarity, P-loop not shown).

Incorporation of an acrylamide electrophile (10) increased potency approximately 30-fold in the BTK binding assay. In contrast, acetylating the piperidine nitrogen as a propionamide (11), a noncovalent acrylamide isostere, indicated the amide alone did not substantially impact BTK binding affinity. Additional studies were initiated to characterize the potential for 10 to inhibit BTK as an irreversible covalent inhibitor.

We examined BTK inhibition by compound 10 in greater detail by measuring BTK kinase activity using a kinetic assay in which ADP production was detected continuously by coupling to the NADH oxidation using pyruvate kinase/lactic dehydrogenase (Supporting Information). As shown in Figure 3A, the BTK reaction was linear over a 2-h period in the absence of 10. However, in the presence of increasing levels of 10, the progress curves exhibited curvature and time-dependent inhibition as typically expected from an irreversible mechanism of inhibition. The data were fit to eq 1 (Supporting Information) to determine the pseudo-first-order rate constant k_obs, the rate of inactivation at each inhibitor concentration. Plotting these k_obs values as a function of the concentration of 10, as described by eq 2 (Supporting Information), revealed saturable inactivation kinetics, indicating a two-step mechanism of BTK inactivation in which the first step involves reversible binding of 10 to BTK, followed by the second chemical step, which results in covalent bond formation (Figure 3B). Furthermore, the y-intercept of a plot of k_obs vs [I] is zero, as expected from an irreversible covalent inhibitor. Based on this model, the k_inact and K_I values for 10 were determined to be 0.000256 ± 0.000008 s^–1 and 4.22 ± 1.32 μM, respectively (Supporting Information). The second-order rate constant and the overall inactivation potency, k_inact/K_I ratio, was 60.7 ± 18.0 M^–1 s^–1 (Figure 3B).

Figure 3.

Inhibition of BTK by compound 10. Data are averages, and the error bars represent the SD from two separate experiments. (A) Progress curves for BTK inhibition by 10 varied 0.0041–10 μM. The k_obs values were obtained (Supporting Information). (B) The k_obs values were fit into eq 2 (Supporting Information) to obtain the individual k_inact and K_I values as described in the Supporting Information, which were used to calculate the overall potency k_inact/K_I ratio. V₀ is the initial rate and t is time to obtain the first order rate constant for enzyme inactivation (k_obs) at each inhibitor concentration. k_inact is the maximal rate of inactivation and K_I is the inhibitor concentration that yields half the rate of maximal inactivation.

Based on the covalent irreversible mechanism of inhibition, further optimization of 10 and SAR studies were conducted using the second-order rate constant k_inact/K_I ratios. The initial potency optimization surveyed the role of substituents on the N-aryl ring that projects into the back pocket region of BTK (Table 2). The unsubstituted phenyl (12) was approximately twofold more potent than the parent compound 10 with para-isopropoxy substitution. The potency (k_inact/K_I ratio) increased approximately threefold by the addition of an ortho-methyl substituent (13) compared to the unsubstituted phenyl (12) and further increased by threefold when paired with the para-isopropoxy substitution (14). By contrast, shifting the isopropoxy substituent to the meta position (15) was detrimental. Compound 14 possessed a 14-fold higher k_inact/K_I ratio than that of 10 and displayed potent on-target cellular activity, inhibiting anti-IgM-induced B cell activation in mouse splenocytes (IC₅₀ = 42 nM) and anti-IgD-induced activation of B cells in rat whole blood (IC₅₀ = 460 nM), consistent with blockade of B-cell signaling through the BTK mechanism.

Table 2. Impact of Back Pocket Substituents on Potency.

^aEach k_inact and K_I value was obtained from an average of two independent determinations (±SD). The method is described in the Supporting Information.

^bCD69 expression on mouse splenocytes was measured after stimulation with anti-IgM. nd = not determined.

^cInhibition of anti-IgD-induced activation of B cells (B220+CD86+).

^dn = 1.

Introduction of the ortho-methyl substituent produced a mixture of atropisomers that did not readily interconvert at ambient temperature. The atropisomers were chromatographically resolved and characterized as separate phenoxyphenyl ether analogs 16 and 17, having unassigned absolute stereochemistry, indicated as *R and *S. Analogue 17 displayed an approximately 30-fold higher potency and 20-fold improvement in cellular activity relative to the rotational isomer 16, indicating the importance of the configuration in the N-aryl bond. Introduction of halogen atoms -F and -Cl in the ortho-position caused a modest decline in K_I (18, 19), but an ortho-CF₃ substituent (20) was not tolerated. It is noteworthy that the improvement in biochemical potency observed for compound 17 over 10 is primarily due to an improvement in initial reversible binding affinity (K_I values) with little change in k_inact values. For example, the overall biochemical potency improvement of 1500-fold from 10 to 17 corresponded to a 870-fold improvement in K_I values with a change of only 1.8-fold in the k_inact value. These improvements in K_I values are also reflected in the increased lipophilic ligand efficiency of 17 (LLE = 3.3, cLogP 5.0) over 10 (LLE = 2.2, cLogP 3.2). Despite its high potency, the utility of compound 17 was limited by high metabolic clearance (human microsomal half-life <4 min, Supporting Information).

Drug candidates that exist as atropisomers can be challenging to develop based on the energy barrier to rotation and corresponding half-life for atropisomer interconversion.^24,25 Atropisomers with long interconversion half-lives of years or decades at room temperature are preferred because they can be separated and treated as stable stereoisomers. Compounds 21 and 23 were selected to estimate the rotational interconversion barriers of the sterically hindered N-aryl bonds bearing ortho-methyl and ortho-fluoro substituents with the reactive acrylamides removed to avoid competing solvolysis. The resolved atropisomers were subjected to thermal interconversion in ethanol, and the appearance of the alternate atropisomers 22 and 24 was detected and quantified by chiral SFC (254.4 nm absorbance, Figure 4) to determine the rotational energy barriers ΔG^⧧_rot (Supporting Information). Compound 21 displayed a ΔG^⧧_rot = 30.3 kcal/mol, corresponding to an interconversion half-life of 63 years at 25 °C. The rotational energy barrier for the less sterically hindered ortho-fluoro derivative 23 was reduced (ΔG^⧧_rot = 24.2 kcal/mol), leading to a shorter interconversion half-life of 46 h at 25 °C. Retaining the ortho-methyl was preferred in phenoxyphenyl derivatives based on the higher barrier to atropisomer interconversion.

Figure 4.

Rotational barriers for ortho-substituted phenyl analogues.

Having increased the noncovalent affinity through back pocket interactions, we next turned to evaluating aliphatic diamine linkers (Table 3) to optimally position the electrophile to engage Cys481. Structurally diverse linkers achieved potent BTK inhibition, including the flexible acyclic linker in 25 and rigidified ring-constrained linkers 26–30. The shortest 3-aminoazetidine (26) did not appropriately position the acrylamide to efficiently engage Cys481 and resulted in a reduced k_inact/K_I ratio. Aminopyrrolidine derivative 27 imparted improved mouse primary B cell potency relative to the initially evaluated amino-piperidine derivative 17 and maintained potent rat whole blood activity. Upon further characterization, 27 also inhibited anti-IgM-induced activation of B cells in human whole blood with an IC₅₀ = 0.084 μM. The biochemical potency for BTK inactivation of 27 (k_inact/K_I = 12 000 M^–1 s^–1) is similar to the reported potency of the clinical BTK inhibitors acalabrutinib (k_inact/K_I = 30 000 M^–1 s^–1) and tirabrutinib (k_inact/K_I = 24 000 M^–1 s^–1).¹⁵ Based on the high potency in covalent BTK inhibition, in the CD69 cellular assay and in whole blood, 27 was progressed into selectivity profiling and in vivo studies.

Table 3. Lead Optimization through Modification of the Acrylamide Linker.

^aEach k_inact and K_I value was obtained from an average of two independent determinations (±SD). The method is described in the Supporting Information.

^bCD69 expression on mouse splenocytes was measured after stimulation with anti-IgM. nd = not determined.

^cInhibition of anti-IgD-induced activation of B cells (B220+CD86+). Potency values are reported for the more potent atropisomer in the mouse CD69 assay.

^dn = 1.

To verify the covalent mechanism, compound 27 was incubated with BTK protein, and the measured mass of the adduct under both denaturing and nondenaturing conditions confirmed 27 as an irreversible covalent inhibitor of BTK (Supporting Information).

Selectivity of Compound 27

The kinase selectivity profile of 27 was established using radiometric kinase activity assays (Eurofins KinaseProfiler and IC₅₀Profiler; Supporting Information) to define an off-target profile as summarized below. Compound 27 exhibited high kinome human selectivity, resulting in inhibition of >50% at 1 μM for only 10 kinases out of 330 in the panel (3.3%). The IC₅₀ values for these kinases are shown in Table 4. Among the off-target kinases detected in the panel, only BMX, ErbB4, TEC, and TXK overlap with the 10 other kinases in the human genome that share a cysteine residue in the same position as the BTK Cys481 and would therefore have the potential to form a covalent bond with 27.¹⁹ Thus, 27 could exhibit time-dependent inhibition of BMX, ErbB4, TEC, and TXK, and measurement of the k_inact/K_I ratios would be needed for a more accurate selectivity assessment. The potential of 27 to modulate T-cell function through off-target inhibition of LCK was evaluated in a T-cell activation assay. The positive control ITK inhibitor caused potent suppression of anti-CD3/anti-CD28 induced IL-2 secretion, but 27 was inactive at inhibiting IL-2 secretion up to 10 μM (Supporting Information), demonstrating that 27 does not modulate T-cell activation.

Table 4. Kinase Selectivity of Compound 27.

kinase	IC50 (μM)
BTK	0.142
BMX	0.129
LCK	0.130
ErbB4	0.377
TEC	0.409
TXK	1.77
FGR	4.76
B-RAF	7.84
SRC	3.49
HCK	>30

The IV and PO pharmacokinetics of compound 27 were investigated in nonfasted rats (1 and 5 mg/kg IV and PO) and fasted dogs (0.5 and 2.5 mg/kg IV and PO) from 20% hydroxypropyl-β-cyclodextrin solutions. The PK parameters are shown in Table 5. IV pharmacokinetics were characterized by moderate clearance in rat and low clearance in dog, a moderate volume of distribution, and a short plasma half-life across both species. Absorption was rapid with T_max achieved within 1 h. The oral bioavailability was 30 and 68% in rat and dog, respectively. These were considered acceptable PK attributes for the development of a covalent inhibitor, where a short drug half-life may achieve prolonged pharmacodynamic effect based on the irreversible inactivation of BTK and slow rate of enzyme resynthesis.

Table 5. Rat and Dog Pharmacokinetic Data Following Oral and IV Administration of 27.

species	Cl (mL/min/kg)	Vss (L/kg)	IV T1/2 (h)	F (%)	oral Cmax (μM)	Tmax (h)
rat	39.6 ± 6.4	1.0 ± 0.1	0.3 ± 0.0	29.7 ± 4.5	0.6 ± 0.1	1.0 ± 0.0
dog	8 ± 3	1.1 ± 0.2	1.9 ± 0.4	68 ± 21	1.5 ± 0.3	0.8 ± 0.3

In Vivo Target Occupancy of 27

The activity of compound 27 was assessed in a rat BTK occupancy model to validate the potential of covalent inhibition to achieve sustained target engagement in vivo (Figure 5). When orally administered to rats at 3 mg/kg, compound 27 exhibited a rapid decrease in exposure between 1 and 4 h. However, BTK occupancy was decoupled from compound exposure levels, maintaining approximately 70% target occupancy at 24 h and 50% occupancy at 48 h, at which point 27 was not detectable in the plasma. The extended pharmacokinetic effect of 27 is consistent with a covalent mechanism requiring BTK protein synthesis for recovery of catalytic activity.

Figure 5.

Comparison of plasma concentration and BTK occupancy after a single oral dose of compound 27 (3 mg/kg).

Preclinical Efficacy in CIA Model

A rat collagen-induced arthritis (CIA) model was used to define the effectiveness of compound 27 at reducing joint inflammation in rats (Figure 6). Arthritic inflammation was induced in female Wistar rats by injecting collagen on day 0 and day 7, and treatments were initiated therapeutically on day 9 and continued until day 16. Once daily oral administration of compound 27 attenuated hind paw inflammation in a dose-dependent manner. A dose of 3 mg/kg/day of 27 achieved an effect on inflammation that was equivalent to the anti-TNFα comparator treatment dosed at 10 mg/kg QOD. Efficacy approaching resolution of disease was achieved at the higher dose groups while remaining well-tolerated. These results are consistent with reported exposure-efficacy relationships requiring >12 h coverage of the BTK IC₇₀ to achieve efficacy in preclinical models of rheumatoid arthritis.²⁷

Figure 6.

Efficacy of 27 in a rat model of collagen-induced arthritis.

Chemistry

Compound 27 was synthesized in 8 steps as shown in Scheme 1.²⁸ The phenoxy-phenyl ether 32 was assembled using an S_NAr reaction between phenol and p-fluoronitro compound 31. The nitro group was reduced to an aniline by palladium-catalyzed hydrogenation and elaborated in a subsequent S_NAr reaction with 2-chloro-4-iodonicotinitrile to give 34, with substituents poised to build the tricyclic core. Nucleophilic addition of 2-mercaptoacetic acid methyl ester followed by intramolecular cyclization yielded 35, and the final ring of the tricyclic core was completed by treatment with CDI. The resultant ester was saponified with lithium hydroxide, and the atropisomers of the resulting carboxylic acid (37) were resolved by chiral SFC with pure but unknown configurations designated *R and *S. The final product 27 was prepared by introduction of the acrylamide-substituted diamine with a HATU coupling.

Scheme 1. Synthesis of Compound 27.

Reagents and conditions: (a) phenol, K₂CO₃, DMF, 90 °C for 16 h, 92%; (b) H₂, Pd/C, MeOH, 50 PSI for 12 h, 90%; (c) 2-chloro-4-iodonicotinitrile, Pd(OAc)₂, K₃PO₄, DPE(phos), 1,4-dioxane, 100 °C for 16 h, 63%; (d) 2-mercaptoacetic acid methyl ester, NaOMe, MeOH, reflux 18 h, 75%; (e) CDI, 1,4-dioxane, reflux 18 h, 86%; (f) LiOH, MeOH/THF/H₂O, 60 °C for 16 h, 91%; (g) Chiral SFC column (stationary phase: Whelk O1 (S,S), 5 μm, 250 × 21.1 mm column. The mobile phase was: 40% CO₂, 60% MeOH (0.2% formic acid); (h) HATU, (R)-1-(3-aminopyrrolidin-1-yl)prop-2-en-1-one, TEA, DMF, 1 h, rt, 32%.

Toward the goal of developing BTK inhibitors with high potency and kinome selectivity for immunology indications, a weakly binding noncovalent BTK chemotype was converted to irreversible covalent inhibitor 27, characterized by excellent kinome selectivity, potent cellular activity, and favorable properties for oral bioavailability. BTK inhibitors were profiled by determining the second order rate constant k_inact/K_I, demonstrating the improvement in potency was achieved by optimizing the initial reversible binding affinity (K_I) and not to an increase in electrophile reactivity. Compound 27 demonstrates a benefit of covalent inhibitors by achieving sustained pharmacodynamic activity, resulting in efficacy in a collagen-induced arthritis disease model despite having a short drug half-life. The optimization of additional compounds from this chemotype will be reported in due course.

Glossary

Abbreviations

BCR

B cell receptor

CIA

collagen-induced arthritis

HP-β-CD

2-hydroxypropyl-β-cyclodextrin

LCK

lymphocyte-specific protein tyrosine kinase

PK

pharmacokinetics

PO

oral dosing

QD

once daily

QOD

once every other day

*R and *S

resolved stereoisomer with unknown absolute stereochemistry

RA

rheumatoid arthritis

SAR

structure–activity relationship

SjS

Sjogren’s syndrome

S_NAr

nucleophilic aromatic substitution

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.1c00044.

• Determination of atropisomer interconversion barriers, characterization of a BTK covalent adduct, kinase selectivity panel, synthesis methods, and biological protocols (PDF)

Author Present Address

^□ J.J.M.W.: Lundbeck La Jolla Research Center, Inc., 10835 Road to the Cure Suite #250, San Diego, California 92121, United States

Author Present Address

^○ K.D.K.: Empress Therapeutics, 790 Memorial Drive, Cambridge, Massachusetts 02139, United States

Author Present Address

^△ M.H.: Vividion Therapeutics, 5820 Nancy Ridge Drive, San Diego, California 92121, United States

Author Present Address

^▽ J.J.W.: 11383 Pacific Shores Way, San Diego, California 92130, United States

Author Present Address

^■ K.A.: Kojin Therapeutics, One Kendall Square, Building 200 Suite 001, Cambridge, Massachusetts 02139, United States

Author Present Address

^● S.D.B.: Denovicon Therapeutics, 1130 Wall Street #558, La Jolla, California 92037, United States

The authors declare no competing financial interest.