Bruton's Tyrosine Kinase Inhibitors with Distinct Binding Modes Reveal Differential Functional Impact on B-Cell Receptor Signaling Open Access

This article is featured in Selected Articles from This Issue, p. 1

Bruton's tyrosine kinase (BTK) plays an important role in normal B-cell receptor (BCR) signaling during pre–B-cell development and in the malignant growth of B-cell lymphoma (1). Since its discovery in 1993 as the gene responsible for X-linked agammaglobulinemia (2, 3), BTK has become a prominent therapeutic target for hematologic cancers (4). In recent years, ATP-competitive inhibitors of BTK have become established as standard-of-care treatment for multiple cancers (1) and have shown encouraging clinical activity in several autoimmune diseases (1, 5).

More than 20 BTK inhibitors (BTKi) have been developed for clinical testing, and more than a dozen are approved or in active clinical trials. Ibrutinib was the first to gain approval and is currently used for the treatment of various B-cell malignancies, including chronic lymphocytic leukemia (CLL)/small lymphocytic lymphoma (SLL), mantle cell lymphoma (MCL), Waldenström macroglobulinemia (WM), marginal zone lymphoma (MZL), and chronic graft-versus-host disease (6). Survival benefits in patients with both previously untreated and relapsed/refractory CLL/SLL have been demonstrated with up to 8 years of long-term clinical data (7, 8). Subsequent to ibrutinib, acalabrutinib received approval by the FDA for MCL and CLL/SLL, and zanubrutinib received FDA approval for MCL, WM, and MZL (9, 10).

All four approved BTKis (ibrutinib, acalabrutinib, zanubrutinib, and tirabrutinib) form a covalent bond to cysteine 481 (C481) at the ATP-binding site of the BTK kinase domain (4). Reversible BTKis have demonstrated clinical activity in several patient populations, including in patients who have become resistant to covalent inhibitors (11). Whether covalent or noncovalent binders, all clinically evaluated BTKis can be categorized into one of three groups depending on their binding mode: back-pocket binders, represented by ibrutinib and the reversible BTKi pirtobrutinib; H3 pocket binders, represented by LOU064 (remibrutinib) and GDC-0853 (fenebrutinib); and hinge-only binders, represented by ABBV-105 (elsubrutinib). To date, all BTKis that are approved for oncology indications are covalent, back-pocket binders (5). Detailed structural studies using solution nuclear magnetic resonance (NMR) and hydrogen deuterium exchange mass spectrometry (HDX-MS) revealed that different binding modes of the various types of inhibitors lead to differential impacts on the BTK kinase domain and regulatory domain conformations (12).

Immediately downstream of BTK activation, phosphatidylinositol hydrolysis by phospholipase C gamma 2 (PLCγ2) and calcium mobilization are two major signaling events that occur upon BCR activation and are mediated by the concerted activity of BTK and its upstream kinases (13). Multiple tyrosine residues in PLCγ2 have been shown to undergo BTK-dependent phosphorylation and subsequently drive BCR-coupled calcium signaling (14). The BTK C481 mutations and PLCγ2-activating mutations were identified to be the two most prominent resistance mechanisms for patients with CLL treated with ibrutinib, further cementing the significant role that PLCγ2 plays downstream of BTK in promoting malignant B-cell proliferation (15). Studies in reconstituted COS-7 cells showed that ibrutinib-resistant PLCγ2 variants were activated by membrane-targeted and catalytically inactive BTK (E41K/K430R). Using the same cell system, it was reported that the Rac family small GTPase 2 (RAC2) was able to activate both the wild-type and ibrutinib-resistant variant of PLCγ2 in the presence of the catalytically inactive BTK (16). These results suggest that a noncatalytic function of BTK plays a role in transducing BCR signaling, likely through cooperation with RAC2 and PLCγ2.

In this study, we sought to understand whether treatment with different types of BTKis would result in differential functional impacts on BTK downstream signaling and the degree of efficacy in B-cell lymphoma models.

A docking model of BTK covalently bound to ABBV-105 was constructed on the basis of the crystal structure of wild-type BTK in complex with ibrutinib (PDB entry: 5P9J). First, the BTK-ibrutinib structure was imported into Maestro software from Schrödinger and prepared for modeling studies via the Protein Preparation Wizard. This computer-assisted drug design tool was used to assign bond orders, ionization and tautomeric states, and add missing hydrogen atoms to the residues and ligand in addition to optimizing the structure's H-bond network. The Protein Preparation Wizard's default restrained energy minimization protocol was run to relieve strain in the complex. Schrödinger's Covalent Docking program was then used to covalently dock ABBV-105 into BTK by identifying C418 as the reactive residue, choosing Michael addition as the reaction type (17, 18), using ibrutinib to define the receptor pocket, and defining the backbone carbonyl groups of E475 and M477 and the backbone amino group of M477 as functional groups with the potential to hydrogen bond to ligands.

The chemical structures of ibrutinib (compound CID: 24821094, synthetized in-house; ref. 19), acalabrutinib (compound CID: 71226662, synthetized in-house; ref. 20), and zanubrutinib (compound CID: 135565884, synthetized in-house; ref. 21) are publicly available at PubChem (https://pubchem.ncbi.nlm.nih.gov/; ref. 22) and PubChem BioAssay (https://pubchemdocs.ncbi.nlm.nih.gov/bioassays). The structures of GDC-0853 (Compound CID: 86567195, Selleckchem, catalog no. S8421; ref. 23), ABBV-105 (compound CID: 117773770, synthetized in-house; ref. 24), and poseltinib (compound CID: 56644522, Selleckchem, catalog no. E2843; ref. 25) are available in corresponding publications or patents. Detailed procedures for compounds 1,2,3 synthesis and structures are included in the Supplementary Methods.

Simple Western was performed for all total and phospho-protein level analyses in this study, except where noted, which used traditional Western blotting.

Protein lysates were prepared from cell lines and mouse tumor tissues using RIPA lysis buffer (Thermo Fisher Scientific, catalog no. 89900), supplemented with protease and phosphatase inhibitors (Thermo Fisher Scientific, catalog no. A32961). Total protein from the lysates was quantified using the Rapid Gold BCA Protein Assay Kit (Thermo Fisher Scientific, catalog no. A55860).

For Simple Western, 0.6 μg to 1 μg of total protein was mixed with the sample loading buffer and a fluorescent standard, boiled at 95°C for 5 minutes, and loaded into the capillary on the micro-plate (ProteinSimple, catalog no. SM-W001 and catalog no. PS-ST01EZ-8). Sample separation, antibody addition, incubation, wash, and chemiluminescence detection were sequentially carried out using the automated Simple Western system. Primary antibodies were used at 1:100 dilution and secondary antibodies were prediluted and provided with the standard detection modules (ProteinSimple, catalog no. DM-001 and catalog no. DM-002). Capillaries were imaged with the Compass software (ProteinSimple, RRID: SCR_022930). The amount of a specific phospho- or total protein in each capillary lane was quantified using the peak height value at its’ corresponding molecular weight. Results were normalized to β-actin loading control.

For traditional Western blotting, 10 μg of total protein was loaded into each lane after boiling with the sample loading buffer (Thermo Fisher Scientific, catalog no. NP0007) at 95°C for 5 minutes. Samples were separated by electrophoresis on a 4%−12% Bis-Tris mini gel (Thermo Fisher Scientific, catalog no. NP0322BOX) and transferred to a polyvinylidene difluoride membrane (Thermo Fisher Scientific, catalog no. 88585) using the XCell II blot module (Thermo Fisher Scientific, catalog no. EI9051) under full wet transfer conditions. The membranes were incubated with Odyssey Blocking Buffer (LI-COR Biosciences, catalog no. 927–40000) for 1 hour and then at 4°C overnight with primary antibody at a 1:1,000 dilution. After washing with 0.1% Tween-TBS, the membranes were incubated with IRDye800CW- or IRDye680RD–conjugated secondary antibodies (LI-COR Biosciences, catalog no. 926–35010 and 926–35011) for 1 hour in the dark and detected using the Odyssey Imaging System (LI-COR Biosciences, RRID: SCR_023765).

The following primary antibodies were used for Western or Simple Western applications in this study: phospho-BTK Y223, clone D9T6H rabbit mAb (Cell Signaling Technology, catalog no. 87141, RRID:AB_2800099); phospho-BTK Y551 (BD Biosciences, catalog no. 558034, RRID: AB_2067823); total BTK, clone D3H5 (Cell Signaling Technology, catalog no. 8547, RRID:AB_10950506); pY1217 PLCγ2 (Cell Signaling Technology, catalog no. 3871, RRID:AB_2299548); and total PLCγ2 (Cell Signaling Technology, catalog no. 3872, RRID:AB_2299586); RAC2 pAb (Abcam, catalog no. ab191527); RAC2 mAb (ProteinTech, catalog no. 60077–1, RRID:AB_2253445); RAC1 mAb, clone 23AB (Millipore, catalog no. 05–389, RRID:AB_309712); pS473 AKT mAb (Cell Signaling Technology, catalog no. 4060, RRID:AB_2811246); total AKT mAb (Cell Signaling Technology, catalog no. 4685, RRID: AB_2225340); pY410 HCK pAb (Abcam, catalog no. ab61055, RRID: AB_942255); pY522 HCK pAb (Abcam, catalog no. ab192578, RRID: RRID:AB_2934010); and total HCK mAb (Cell Signaling Technology, catalog no. 14643, RRID: AB_2687496) β-actin mAb (Cell Signaling Technology, catalog no. 4970, RRID:AB_2223172).

All cell lines were cultured according to ATCC-recommended culture conditions. Passage numbers were limited to <20 passages from the original culture obtained from vendors. Ramos Burkitt's lymphoma cell line (ATCC, catalog no. CRL-1596, RRID: CVCL_0597) and REC-1 mantle cell lymphoma (ATCC, catalog no. CRL-3004, RRID:CVCL_1884) were obtained from ATCC. TMD8 DLBCL cell line (RRID: CVCL_A442) was licensed and procured from Tokyo Medical and Dental University in Japan and used for in vitro assays. All cell lines used in this study were authenticated using short tandem repeat DNA profiling (CellCheck, IDEXX BioAnalytics). Cell line identity and Mycoplasma testing using real-time PCR (STAT-Myco, IDEXX BioAnalytics) were last conducted on the Ramos Burkitt's lymphoma cell line on October 9, 2020, and on the REC-1 mantle cell and TMD8 DLBCL cell lines on October 19, 2020. The TMD8 and the REC-1 MCL xenograft models were used to assess in vivo activities of compounds. BTK, RAC1, and RAC2 knockouts or knockdowns were generated in Ramos and TMD8 cells using CRISPR technology (Supplementary Methods).

Freshly drawn human (purchased from the Stanford Blood Center for research purposes only) or mouse whole blood was incubated with compounds for 2 hours at 37°C. The whole blood was lysed in lysis buffer containing phosphatase inhibitors, and the lysates were frozen at –80°C overnight. BTK pY223 was measured using a multiarray MSD assay (Meso Scale Discovery). The capturing antibody was rabbit anti–human-BTK, clone D3H5 (Cell Signaling Technology, catalog no. 8547, RRID:AB_10950506) and the detection antibody was a biotinylated rabbit anti–pY223-BTK, clone D9T6H (Cell Signaling Technology, catalog no. 16135BC, RRID:AB_2800099).

For total BTK and occupancy quantitation, MSD plates were precoated with BTK capturing antibody (Cell Signaling Technology, catalog no. 8547, RRID:AB_10950506) for 6 hours at room temperature. A BTK-selective biotinylated probe was then incubated with either spleen or tumor lysates on ice for 6 hours and added to the plate for overnight incubation at 4°C. Unoccupied BTK was detected the next day using the Streptavidin SULFO-TAG. Total BTK was detected using anti-BTK antibody (BD Biosciences, catalog no. 611117, RRID:AB_398428) followed by sulfo-secondary antibody (MSD, catalog no. R32AC-1). Spleen and tumor lysates were subjected to Simple Western analyses for pY223 and total BTK quantification.

Cells were grown in culture for 6 days before being plated and used for calcium flux fluorescent imaging plate reader (FLIPR) assay. The cells were split once in the 6-day time frame. Cell culture medium was RPMI1640, with 10% heat-inactivated FBS and penicillin/streptomycin. At the time of cell plating for the FLIPR assay, cell culture medium was replaced with the one containing 5% FBS. The day prior to the assay, plated cells were incubated at 37°C/5% CO₂ for approximately 18 hours. Compounds were preincubated with cells for 1, 2, and 4 hours. Cells were stimulated with 10 μg/mL goat F(ab’)2 anti-human IgM (Southern Biotech, catalog no. 2022–01) and immediately read on the FLIPR for calcium flux.

Compounds were added in a 384-well low-volume plate using D300e digital dispenser (TECAN), and 50,000 Ramos cells (wild-type and BTK knockout) were incubated with compounds in cell growth medium for 4 hours before 10 μg/mL goat F(ab’)2 antihuman IgM (Southern Biotech, catalog no. 2022–01) was added in stimulation buffer (IP-One kit, Cisbio, catalog no. 40451). The cells were incubated overnight in a 37°C cell incubator. The next day, inositol phosphate 1 (IP1)-D2 and anti-IP1 cryptate (IP-One Kit, Cisbio, catalog no. 40451) were added and incubated for 1 hour at room temperature according to the manufacturer's instructions. The assay was read on an EnVision Multilabel Plate Reader (Perkin Elmer) using the HTRF setting.

Protein precipitation method was used for the determination of plasma compound concentration. Testing compounds and an internal standard were extracted from mouse plasma with 10 volume equivalents of acetonitrile in a 96-well deep well plate, vortex-mixed, and then centrifuged to pellet the proteins. One-hundred microlitres of supernatant were transferred into another 96-deep well plate and mixed with 100 μL of 0.2% formic acid in 10% methanol in deionized water. Analytes were separated using reverse-phase chromatography on a Waters XBridge C18 Column (3.5-μm particle size, 50 mm × 2.1 mm), prior to analysis on a SCIEX 6500 mass spectrometer with a turbo-ion spray source. The mobile phases were (A) 10 mmol/L ammonium acetate in high-performance liquid chromatography (HPLC) grade water, pH ∼5 adjusted with glacial acetic acid and (B) acetonitrile. Peak areas were determined using SCIEX Analyst v1.6.2 software. Actual concentrations were calculated by power fit without weighting regression analysis of the peak area ratio (parent/internal standard) of the spiked standards versus concentrations. The lowest limit of quantification was 0.5 ng/mL with linearity demonstrable to 1,000 ng/mL.

Female Fox Chase SCID mice (C.B.17 SCID), 6–8 weeks old, 16–20 g were implanted with human tumor cells subcutaneously at the flank region and mice were randomized into treatment groups of 10 mice per group based on tumor volumes, as calculated by formula: (length × width)²/2, using matched distribution algorithm. Protocols for all in vivo studies were approved by AbbVie's Institutional Animal Care and Use Committee, in compliance with the NIH Guide for Care and Use of Laboratory Animals guidelines.

All results are shown as mean ± SEM. The data were subjected to one-way ANOVA with Tukey multiple comparisons test. P values of less than 0.05 were accepted as statistically significant (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).

The data generated in this study are available upon request from the corresponding author.

Although ibrutinib, ABBV-105 and GDC-0853 are known to bind to the ATP-binding site of BTK, their binding modes have been shown or predicted to be distinct. The cocrystal structure of the BTK:ibrutinib complex (26) revealed that the inhibitor forms a covalent bond with C481 through the acryloyl group, while the phenoxyphenyl moiety projects into the back-pocket of BTK, where the terminal phenyl group of the inhibitor is surrounded by hydrophobic residues of BTK (including M449, I472, and L542) and forms an edge-to-face π–π interaction with F450 of the DFG motif (Fig. 1A). In contrast, the cocrystal structure of the BTK:GDC-0853 complex (27) revealed that this inhibitor does not directly interact with C481 on BTK and that the 7,7-dimethyl-3,4,7,8-tetrahydro-2H-cyclopenta[4,5]pyrrolo[1,2-a]pyrazin-1(6H)-one group instead occupies the H3 pocket, where it sequesters Y551 and interacts with residues Q412, F413, L542, and S543 (Fig. 1A). Given that a crystal structure of BTK bound to ABBV-105 was not available, a docking model was constructed using the X-ray structure of the BTK-ibrutinib complex, as detailed in Materials and Methods. On the basis of this model, ABBV-105 is predicted to form a covalent bond with C481 through the acryloyl group, similarly to ibrutinib. However, unlike ibrutinib and GDC-0853, ABBV-105 is predicted to form three H-bonds at the hinge region of BTK. These include interactions between the 7-carboxamide group of ABBV-105 and the backbone carbonyl and amino groups of E475 and M477, respectively, and a third interaction between the inhibitor's indole -NH group and the backbone carbonyl group of M477 (Fig. 1A). Another distinguishing feature of ABBV-105’s predicted binding mode compared with those of ibrutinib and GDC-0853 was its exclusive occupation of the hinge and ribose-binding regions of the BTK ATP-binding site. On the basis of these observations, ABBV-105 is referred to as a hinge-only binder throughout this article.

In addition to ABBV-105, three hinge-only binders (compounds 1, 2, and 3) were synthetized and examined (Supplementary Methods). To assess proximal and distal pharmacodynamic BCR signaling inhibition with the different types of BTKis, a series of functional assays were performed (Fig. 1B).

We evaluated the impact of BTKi treatment in vitro on autophosphorylation of the Y223 residue in the BTK SH3 domain as well as phosphorylation of PLCγ2 as its direct substrate upon anti-IgM stimulation. All tested BTKis dose dependently inhibited the autophosphorylation of Y223 of BTK in the Ramos Burkitt's lymphoma cell line (Fig. 1C) and in the TMD8 diffuse large B-cell lymphoma (DLBCL) cell line (Fig. 1D), suggesting that the enzyme's kinase activity was fully inhibited. In contrast, only the H3 pocket binders were able to potently inhibit the phosphorylation of the Y551 residue, reflecting their unique ability to sequester and inhibit phosphorylation of this residue by upstream kinases (Fig. 1C and D).

In addition to these BTK tyrosine phosphorylation events, various tyrosine residues on PLCγ2 have been shown to be phosphorylated by BTK and are required for calcium signaling in response to BCR stimulation (14). In Ramos cells, stimulation with anti-IgM resulted in phosphorylation of Y759 and Y1217 on PLCγ2, of which pY1217 showed a clear dose response to BTKi treatment (Supplementary Fig. S1). Therefore, we used transphosphorylation of Y1217 on PLCγ2 as the pharmacodynamic marker of BTK kinase activity on its direct substrate. No significant differences on Y1217 phosphorylation were detected between back-pocket, H3 pocket, and hinge-only binders (Fig. 1C and D).

Despite similar inhibitory potency of BTK catalytic activity exhibited by all three classes of BTKis, the hinge-only binders (ABBV-105; compounds 1, 2, and 3) demonstrated a deficiency in inhibiting calcium flux in a FLIPR calcium assay (Fig. 2; Supplementary Table S1) in both the Ramos and TMD8 cell lines. The percent maximal (max%) inhibition of calcium was <50% when the cells were preincubated with ABBV-105 or compounds 1, 2, and 3 (Fig. 2B), whereas the back-pocket binders ibrutinib, acalabrutinib, and zanubrutinib or the H3 pocket binder GDC-0853, all achieved maximal inhibition of more than 50% in TMD8 cells and up to 90% in Ramos cells (Fig. 2C; Supplementary Table S1). The covalent BTKis ibrutinib, acalabrutinib, and zanubrutinib also exhibited a time-dependent decrease in their IC₅₀ values reflecting the two-step kinetic mechanism of the covalent inhibition (Fig. 2C).

The differential effects on calcium flux by the various BTKis were not explained by the similar levels of pY1217 inhibition observed among the BTKis. Therefore, we hypothesized that phosphorylation-independent activation of PLCγ2 may play a role in calcium flux induction in the BTKi-treated cells. To test this hypothesis, we next used an IP1 HTRF assay to measure PLCγ2-mediated inositol phosphate production as a surrogate readout for assessing its enzymatic activity. Compared with the Ramos wild-type cell line, a significant amount of IP1 production was still observed in the BTK knockout cell line upon anti-IgM stimulation (Fig. 2D). Therefore, when calculating the max% inhibition with BTKi treatment in wild-type cells, the amount of IP1 produced in BTK knockout cells upon anti-IgM stimulation was subtracted from that produced in wild-type cells to adjust for BTK-dependent IP1 production. This normalization method did not affect the compound IC₅₀ calculation. Hinge-only binders showed incomplete inhibition of IP1 production in Ramos cells upon anti-IgM treatment, while back-pocket and H3 pocket binders were capable of inhibiting IP1 production (Fig. 2E). The level of inhibition in IP1 production observed from treatment with different BTKis correlated with their potency in inhibiting calcium flux in Ramos cells (Supplementary Table S1).

We next sought to understand the mechanism for PLCγ2 and calcium flux activation in the absence of BTK kinase activity. RAC2 is an integral part of BCR signaling and contributes significantly to calcium flux and growth of B-cell lymphoma (28). This small GTPase has been shown to directly activate PLCγ2 in vitro and X-ray structural data support that RAC2 can activate PLCγ2 by inducing conformational changes (29). RAC2 was also identified in a CRISPR knockout screen as a dependency factor in ibrutinib-resistant DLBCL cell lines (30). We therefore generated RAC2 knockout clonal cell lines on the TMD8 background to further investigate the role of RAC2 in mediating BCR-coupled calcium signaling. A one base pair insertion in the exon 3 of the RAC2 gene led to the complete elimination of the RAC2 protein in these cells (Supplementary Methods; Supplementary Fig. S2A).

Stimulation of RAC2 knockout TMD8 cells with anti-IgM to activate the BCR showed a 20% reduction in calcium flux compared with the TMD8 wild-type cells, confirming the contribution of RAC2 to BCR-mediated calcium mobilization in wild-type TMD8 cells (Supplementary Fig. S2B). When the RAC2 knockout cells were treated with different types of BTKis, all compounds exhibited increased potency and maximal inhibition of calcium flux relative to the RAC2 wild-type cells, suggesting that elimination of the RAC2 bypass mechanism enabled BTKis to more completely block calcium mobilization (Table 1). Among the hinge-only binders tested, ABBV-105 and another clinical stage BTKi, poseltinib, showed the most significant increase in potency in the RAC2 knockout cells, with a maximal inhibition of calcium flux both above 60% and IC₅₀ under 100 nmol/L (Table 1). In contrast, both compounds were not able to inhibit calcium flux to above 50% in TMD8 wild-type cells. Back-pocket binders showed superior potency in preclinical non-Hodgkin lymphoma (NHL) models compared with hinge-only binders.

To understand whether the differential impact of the BTKis on calcium signaling directly affects the potency of the BTKis in inhibiting tumor cell growth in vitro, we treated the TMD8 cells with various inhibitors and assessed cell viability using the CellTiter-Glo (CTG) assay (Fig. 3A). Cell killing half-maximal effective concentrations (EC₅₀) of the hinge-only binders were more than 50-fold higher compared with EC₅₀ of back-pocket and H3 binders in a 4-day cell viability assay, despite similar potency in inhibiting BTK kinase activity (Fig. 3A and B). The lack of TMD8 tumor cell growth inhibition demonstrated by hinge-only binders was consistent with their inability to inhibit calcium flux in these cells (Table 1). Furthermore, compound 1 was defective in inhibiting the phosphorylation of Akt at residue S473 (Fig. 3B), a key event of the PI3K signaling downstream of BCR activation that is partially mediated by BTK (31). In addition, none of the BTKis had an impact on HCK phosphorylation level at residues Y410 and Y522 (Supplementary Fig. S3), which was shown to be regulated by BTK kinase–dead mutants in B lymphoma cells (32).

Next, we sought to understand whether kinase-independent function of BTK can contribute to acquired resistance in cells treated with various BTKis. To induce acquired resistance, TMD8 cells were treated with increasing concentrations of ABBV-105 over 6 weeks and the resulting population of ABBV-105–resistant cells was tested for sensitivity to different BTKis in a 4-day CTG assay. All BTKis showed a reduction or loss of potency in killing of the ABBV-105–resistant cells, compared with parental cells (Fig. 3C). Ibrutinib retained nanomolar cytotoxic potency in ABBV-105–resistant cells, whereas acalabrutinib, GDC-0853, and all three hinge binders tested were not cytotoxic at submicromolar concentrations. In addition, the hinge binders showed paradoxical stimulation of cell proliferation compared with control DMSO treatment in the ABBV-105–resistant cells (Fig. 3C). This observation suggests hyperactivation of the BCR signaling by the hinge binders in the ABBV-105–resistant cells, which is dependent on BTK scaffolding function and likely involves PLCγ2 and calcium flux activation via RAC2.

To corroborate the in vitro data, in vivo efficacy of the hinge-only and back-pocket binders was tested in TMD8 and REC-1 xenograft models (Fig. 4). For the TMD8 model, the hinge-only binder ABBV-105 was compared with the back-pocket binder acalabrutinib. In vitro, both molecules inhibited BTK pY223 and PLCγ2 pY1217 dose dependently with similar potencies, yet a 50-fold difference was observed in EC₅₀ in the TMD8 CTG assay, suggesting their differential impact on the BTK noncatalytic functions (Fig. 3A). For the in vivo efficacy study, TMD8 tumor–bearing mice were treated with multiple doses of ABBV-105 and acalabrutinib on a twice-daily schedule. Plasma exposure of the compounds were determined at highest concentration (C_max) and lowest concentration (C_trough), and apparent total exposure (area under the curve) was calculated using these parameters (Fig. 4A and B). BTK occupancy was determined for both compounds in spleen and tumor lysates at 12 hours after the final dosing, where both compounds showed ≥90% tumor BTK occupancy levels at 30 mg/kg twice daily at C_trough. Despite similar plasma exposure and degree of BTK occupancy in tumor tissues (∼90% BTK occupancy at C_trough), ABBV-105 was not able to achieve >50% tumor growth inhibition (TGI) in this model, whereas acalabrutinib achieved tumor regression (TGI = 91%, 92% BTK occupancy at C_trough). Similar to acalabrutinib, treatment with back-pocket binder ibrutinib led to tumor regression in the TMD8 model with twice-daily dosing (Supplementary Fig. S4).

Insufficient inhibition of tumor growth by the hinge-only binders was also observed in the REC-1 MCL model (Fig. 5). All inhibitors used in this study showed dose-dependent inhibition of BTK pY223 levels in both Simple Western analysis of the tumor tissues and multiarray MSD analysis of the whole blood at 12 hours after the last dose. Compound 2 showed the strongest inhibition of the BTK pY223 level and significantly higher BTK occupancy levels compared with ibrutinib (Fig. 5A–C). However, ibrutinib achieved better TGI (85%) than both hinge-only binders tested (TGI = 28% for compound 1, 67% for compound 2; Fig. 5D). In summary, the in vitro and in vivo tumor cell growth data from multiple B lymphoma cells reveal the insufficient inhibition of the aberrant BCR signaling by hinge-only binders.

ATP-competitive BTKis have shown clinical activity, including molecules that bind in reversible or irreversible fashion. These different classes can be further categorized by the manner in which they bind to the protein. In this study, we investigated ATP-competitive BTKis with three different binding modes to the ATP-binding site, all of which potently inhibited BTK-Y223 autophosphorylation. The inhibitors also indistinguishably inhibited the substrate phosphorylation of the PLCγ2 Y1217 residue of BTK. However, the hinge-only binders exhibited deficiency in inhibiting downstream anti-IgM–stimulated calcium flux in B-cell lymphoma. Incomplete inhibition of calcium flux by the hinge-only binders led to their lower efficacy in inhibiting tumor cell growth in vitro and in vivo, compared with the back-pocket and H3 pocket binders.

X-ray crystallography, solution NMR, and HDX-MS methods have been applied to explore the effects of inhibitor binding on the dynamic conformations of full-length BTK (12). The hinge-only binder examined in the Joseph and colleagues study (12) induced limited conformational change to the N-lobe of the kinase domain. The H3 pocket binder GDC-0853 altered the conformation of both the N-lobe and C-lobe of the BTK kinase domain without affecting the conformation of its regulatory domain. In contrast, the back-pocket binder ibrutinib markedly stabilized the αC -helix of the kinase domain in an inactive conformation compared with other BTKis and partially released the SH2 and SH3 domains from the distal side of the kinase domain. These structural features could impact the scaffolding function of BTK in addition to its catalytic activity (33). Consistent with its most limited interaction with the BTK ATP-binding site, and minimal impact on structural conformations outside the BTK kinase domain, the hinge-only binders in our study had the lowest efficacy in inhibiting calcium mobilization and tumor cell growth in vitro and in vivo, among the three types of BTKis examined. These results suggest that a wider perturbance of the BTK kinase domain and potentially regulatory domain conformations is required for disrupting the noncatalytic function of BTK and fully inhibiting its downstream signaling.

Furthermore, a multicomponent BCR/toll-like receptor (TLR) signaling complex was described by Phelan and colleagues to form and signal at the endosomal membrane in ibrutinib-sensitive ABC DLBCL cell lines (34). Ibrutinib was shown to disrupt PLCγ2 interaction with MyD88 in this MyD88–TLR9–BCR super complex, suggesting that a back-pocket binder could alter the interaction between BTK and its signaling partners in multiprotein complexes, in addition to inhibiting its kinase activity. The efficacy of back-pocket binders in inhibiting aberrant B-cell proliferation in humans is clinically validated by the approval of ibrutinib, acalabrutinib, zanubrutinib, and tirabrutinib for multiple oncology indications (4).

A unique feature of an H3 pocket binder is its ability to markedly inhibit the phosphorylation of Y551 by upstream kinases. This phenomenon results from a side-chain rotamer change of the Y551 residue upon inhibitor binding, where the side chain becomes sequestered in the H3 pocket and is thus rendered inaccessible to phosphorylation (26). However, it is unclear what functional consequences result from phosphorylation of the Y551 residue in malignant B cells, although it has been shown to mediate Fc receptor signaling (35).

Despite the complete inhibition of the BTK kinase activity, our data indicate the incomplete inhibition of PLCγ2 activation upon BCR stimulation as a major underlying mechanism of the functional deficiency of the hinge-only binders. All BTKis inhibited PLCγ2 phosphorylation to the same extent yet showed differential inhibition of PLCγ2 activation, as demonstrated by varying degrees of reduction in inositol phosphate production observed upon BTKi treatment. We postulated that, in the condition of BTK kinase inhibition, RAC2-mediated activation of PLCγ2 could serve as a potential bypass mechanism for intracellular calcium mobilization in these cells. To this end, RAC2 knockout in TMD8 cells improved the potency of all types of BTKis in inhibiting calcium flux and rescued the deficiency of some hinge-only binders, thereby confirming the role of RAC2 in BCR-coupled calcium signaling. PLCγ2-activating mutations are commonly found in ibrutinib-resistant patients with CLL (35, 36), and RAC2 was identified to be an epigenetic mediator of ibrutinib resistance in ABC DLBCL cell lines (30). In the study by Shaffer and colleagues (30), the ibrutinib-resistant cell pools were shown to be more reliant on RAC2 than their parental cells. In addition, RAC2 was shown to interact with BTK in quantitative mass spectrometry. CRISPR knockout of BTK was toxic to the ibrutinib-resistant cells, indicating the resistant cells still relied on noncatalytic BTK function for survival. These data further support that both PLCγ2 and RAC2 could be potential drug targets to overcome BTKi resistance in B-cell malignancies.

Building on prior published findings (29, 37), our study provides further evidence that catalytically inhibited BTK can coordinate with RAC2 to activate PLCγ2, establishing RAC2/PLCγ2 mediated BTK scaffolding function as a bypass mechanism for BCR signaling in the absence of BTK kinase activity. RAC2 knockout improved the potency and max% inhibition of calcium flux by all BTKis; for hinge-only binders, however, RAC2 knockout alone was not able to fully rescue the deficient inhibition of calcium flux. These data suggest the presence of additional bypass mechanisms capable of activating PLCγ2, while BTK kinase activity is inhibited. We did observe deficiency in inhibiting Akt phosphorylation at S473 by compound 1, suggesting that elevated Akt-mediated signaling could further contribute to the diminished potency of the hinge-only binders in inhibiting TMD8 proliferation. A recent study demonstrated that kinase-dead BTK mutants can recruit hematopoietic cell kinase (HCK) to phosphorylate and activate PLCγ2, thus transducing the BCR signaling in the absence of BTK kinase activity (32). However, we did not observe changes in HCK phosphorylation upon treatment with the three subtypes of BTKis, suggesting that the HCK-mediated BCR signaling could be specifically associated with gain-of-function BTK C481F or C481Y kinase-dead mutants in tumor cells.

Interestingly, resistance induced by treating TMD8 cells with increasing concentration of a hinge-only binder (ABBV-105) also rendered partial resistance of this tumor cell line to back-pocket and H3 pocket binders. These results suggest that BTKis with different binding modes are susceptible to common bypass mechanisms that can sustain cancer cell proliferation when BTK kinase activity is fully inhibited, in the context of acquired resistance to BTKi treatment (30). It is unlikely that such resistance can be overcome by switching between different catalytic inhibitors.

In summary, we demonstrated that while back-pocket and H3 binders inhibited both the catalytic and scaffolding BTK activity leading to decreased in vitro and in vivo cancer cell proliferation, hinge-only binders did not affect the BTK scaffolding function and failed to suppress tumor growth. These data collectively indicate that BTKis have differential downstream impact upon BTK signaling and efficacy in BTK-mediated disease models, and support the consideration of binding mode when designing and evaluating BTKi for treatment of human diseases.

W. Li reports employment with AbbVie at the time the study was conducted; current employment with Neomorph; and stock/stock options with AbbVie and Neomorph. M. Yang reports employment with AbbVie at the time the study was conducted; current employment with Merck; and stock/stock options with AbbVie. G. Lundgaard reports employment with AbbVie. M. Apatira reports employment and stock/stock options with AbbVie and employment and stock/stock options of immediate family member with The Permanente Medical Group. R. Sano reports employment and stock/stock options with AbbVie. T. Gururaja reports employment, stock/stock options, and compensation for a leadership role with AbbVie. J. Liu reports employment and stock/stock options with AbbVie. Y. Zhai reports employment with AbbVie. C. Pan reports employment and stock/stock options with AbbVie. W.H. Yoon reports employment and stock/stock options with AbbVie. A.J. Souers reports employment and stock/stock options with AbbVie and patents, royalties, and other intellectual property interests with AbbVie. C. Tse reports employment and stock/stock options with AbbVie. L. Wang reports employment and stock/stock options with AbbVie. F. DeAnda reports employment and stock/stock options with AbbVie and patents, royalties, and other intellectual property interests with Pharmacyclics LLC, an AbbVie Company. C.-H. Lee reports employment and stock/stock options with AbbVie Inc.

W. Li: Conceptualization, data curation, formal analysis, investigation, writing–original draft, writing–review and editing. R. Sano: Data curation, formal analysis, investigation, writing–original draft, writing–review and editing. M. Apatira: Data curation, formal analysis, investigation, writing–original draft, writing–review and editing. F. DeAnda: Data curation, formal analysis, investigation, writing–original draft, writing–review and editing. T. Gururaja: Data curation, formal analysis, investigation, writing–original draft, writing–review and editing. M. Yang: Data curation, formal analysis, investigation, writing–original draft, writing–review and editing. G. Lundgaard: Data curation, formal analysis, investigation, writing–original draft. C. Pan: Data curation, formal analysis, investigation, writing–original draft, writing–review and editing. J. Liu: Data curation, formal analysis, investigation, writing–original draft. Y. Zhai: Data curation, formal analysis, investigation, writing–original draft. W.H. Yoon: Data curation, formal analysis, investigation, writing–original draft. L. Wang: Data curation, formal analysis, investigation, writing–original draft, writing–review and editing. C. Tse: Resources, writing–original draft, writing–review and editing. A.J. Souers: Resources, writing–original draft, writing–review and editing. C.-H. Lee: Conceptualization, data curation, formal analysis, investigation, writing–original draft, writing–review and editing.

All sources of support, including funding, equipment, and compounds, were provided by AbbVie Inc. We thank Mee Rhan Kim, Sharanya Ford, Kojo Osei-Bonsu, and James P. Dean (all employed with AbbVie Inc.) for their critical reviews and discussions about the manuscript; Guowei Fang for providing insightful suggestions on experimental design; Jim Yan (employed with AbbVie Inc.), Joe Cefalu, and Xuan Ma (employed with AbbVie Inc.) for dosing and pharmacokinetic/pharmacodynamic sample collection for in vivo studies. Editorial support was provided by Jaya Kolipaka, MS, and Agnieszka Looney, PhD, and was funded by Pharmacyclics LLC, an AbbVie Company.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.