Structural and biophysical insights into the mode of covalent binding of rationally designed potent BMX inhibitors

The bone marrow tyrosine kinase in chromosome X (BMX) is pursued as a drug target because of its role in various pathophysiological processes. We designed BMX covalent inhibitors with single-digit nanomolar potency with unexploited topological pharmacophore patterns. Importantly, we reveal the first X-ray crystal structure of covalently inhibited BMX at Cys496, which displays key interactions with Lys445, responsible for hampering ATP catalysis and the DFG-out-like motif, typical of an inactive conformation. Molecular dynamic simulations also showed this interaction for two ligand/BMX complexes. Kinome selectivity profiling showed that the most potent compound is the strongest binder, displays intracellular target engagement in BMX-transfected cells with two-digit nanomolar inhibitory potency, and leads to BMX degradation PC3 in cells. The new inhibitors displayed anti-proliferative effects in androgen-receptor positive prostate cancer cells that where further increased when combined with known inhibitors of related signaling pathways, such as PI3K, AKT and Androgen Receptor. We expect these findings to guide development of new selective BMX therapeutic approaches.

This journal is © The Royal Society of Chemistry 2020

Figure S1
Structure of the analogues prepared S3

Figure S9
Synergistic anti-proliferative effects of JS24-26 with AKT inhibitor AKT1/2, androgen receptor inhibitor Flutamide and PI3K inhibitor LY294002 on the viability of LNCaP prostate cancer cells. S10 Figure S10 Kinase selectivity of compound JS25 over 36 BMX-related kinases in the Eurofins DiscoverX's KINOMEscan platform S11 Figure S11 BMX degradation after treatment with JS25 and BMX-IN-1 in PC3 cells S12 Scheme S1 Synthetic route for the preparation of compounds JS9a-e, JS10-13 and JS14-23 S13 Scheme S2 Synthetic route for the preparation of compounds JS24-27 S14 Scheme S3 Synthetic route for the preparation of compound JS28 S14 Scheme S4 Synthetic route for the preparation of compound JS29 S14

Table S1
In silico cLogP and LogS calculation and in vitro artificial membrane permeability (PAMPA) and colloidal aggregation (DLS) determination S15

Table S2
Melting temperature (Tm) shift calculated with a DSF assay S16

Table S3
Kinetic constants calculated from Surface Plasmon Resonance S17

Table S4
Data processing statistics for the crystallographic structure of BMX in complex with inhibitor JS24 S17

Table S5
Refinement statistics for the crystallographic structure of BMX in complex with inhibitor JS24 S19 S3 Figure S1. Structure of the analogues prepared. The modified regions in each analogue are highlighted in red.     Figure S6. Distance evolution, derived from 0.5 µs MD simulations, between quinoline ring of JS24 (a) and JS27 (b) and methyl groups of Ala443, Val431, and Leu543 of the receptor. Distance evolution between the side chain of Tyr491 and the aromatic phenyl-sulfonamide ring (JS24) or piperazine ring (JS27) is also shown. In all cases, the centre of mass of the rings were considered in the calculations.
The combined organics were washed with brine, dried over MgSO4 and evaporated to dryness to afford the title compound as an orange solid (3.73 g; 100% yield).
The solution was cooled to 0ºC, and NaOH (1 M) was slowly added. The mixture was stirred for 15 min at rt. H2O was added and the phases were separated. The aqueous phase was further extracted with DCM (3x). The combined organics were washed with brine, dried over MgSO4 and taken to dryness to afford the title compound as a yellow solid (1.75 g; 75% yield).
Compound 7n was isolated as a yellow solid (285 mg; 95% yield
Compound 7o was isolated as a yellow solid (215 mg; 77% yield
Fe (224 mg; 4.0 mmol; 6 equiv.) and NH4Cl (215 mg; 4.016 mmol; 6 equiv.) in H2O (20 mL) were added and the mixture heated to reflux. After 2h, TLC analysis (MeOH:DCM 1:9) showed that the reaction was completed. The hot mixture was filtered through a Celite pad and the pad further washed with EtOH and MeOH:DCM (2:8). The solvent was evaporated and the crude partitioned between H2O and EtOAc. The phases were separated and the aqueous phase was further extracted with EtOAc (3x). The combined organics were washed with brine, dried over MgSO4 and taken to dryness to afford the title compound as an off-white solid (330 mg; 100% yield).

1(2H)-yl)phenyl)acrylamide (9a; BMX-IN-1)
A stirred solution of 8a (90 mg; 0.191 mmol) in dry THF (20 mL) was cooled in an ice-bath to -10ºC for 20 min. DIPEA (133 µL; 0.765 mmol; 4 equiv.) was added and the mixture stirred for 10 min at a temperature < 4 ºC. After 10 min, acryloyl chloride was added and the mixture further stirred at -10ºC for 10 min. and then 1h at rt. THF was then evaporated and the crude redissolved in EtOAc and washed three times with NaHCO3 (4%). The organics were dried over MgSO4 and taken to dryness. The crude was applied in a silica column and eluted with a gradient from 100:0 to 96:4 in DCM:MeOH. The desired fractions were collected and taken to S48 dryness to afford the title compound as a white solid. (20 mg; 20% yield

In silico cLog and LogS:
cLogP and LogS were calculated using SwissADME software. 3 cLog P is a consensus value obtained as the arithmetic mean of five freely available predictive models (XLOGP3, 4 WLOGP, 5 MLOGP, 6,7 SILICOS-IT 8 and iLOGP 9 ) and LogS is the arithmetic mean of two topological methods (ESOL model 1 and Ali et al. 2 ).

Dynamic Light Scattering:
Dynamic light scattering (Zetasizer Nano S, Malvern, UK) was used to determine compound colloidal aggregation. The particle sizes were measured at 25 °C. A 10 mM stock solution of test compounds was prepared in DMSO, following dilution to deionized and filtered water to obtain an analyte solution of 10 μM (0.1% DMSO). Colloidal aggregation was measured through sequential dilutions at 10 μM, 1 μM and 0.1 μM.

Multidimensional scaling:
BTK and BMX ligands, as well as the annotated activities, were collected from ChEMBLv24. Data was pre-processed as previously reported 10 and the CATS2 descriptors were calculated 11 (MOE, chemical computing group implementation). Dimensionality was reduced through multidimensional scaling algorithm.

Artificial Membrane Permeability (PAMPA):
The PAMPA Evolution™ instrument was used to determine permeability, at Pion Inc. In PAMPA, a sandwich is formed such that each composite well is divided into two chambers, separated by a 125 µm thick microfilter disc (0.45 µm pores), coated with Pion GIT-0 phospholipid mixture. The effective permeability, Pe (x10 -6 cm/sec), of each compound was measured at pH 6.8 in the donor compartment using low-binding, low UV Prisma buffer. The drug-free acceptor compartment was filled with acceptor sink buffer containing a scavenger at After 60 min, the fluorescence transfer is measured at λex=337 nm, λem=620 nm and λem=665 nm using a microplate reader (Rubystar, BMG). The enzyme activity is determined by dividing the signal measured at 665 nm by that measured at 620 nm (ratio). The results are expressed as a percent inhibition of the control enzyme activity.

BTK Biochemical Activity Assay:
BTK kinase activity (IC 50 ) was performed at DiscoverX through a radiometric assay with the respectively) and for ROX reference dye (580nm and 623nm, respectively). The melting temperatures were obtained by taking the midpoint of each transition.

Surface Plasmon Resonance:
The kinetic and affinity parameters of protein-compound interaction were evaluated by SPR.
Experiments were carried out in a Biacore 4000 instrument ( (1% -3%) was performed to account for variations in bulk signal and to achieve high-quality data. Each compound was injected over immobilized His 6 -BMX for 220 s (30 µL min -1 ; association phase) followed by 600 to 2000 s of buffer flow (dissociation phase) at a maximum concentration of 0.5 µM or 10 µM for high and low affinity binders, respectively, and diluted five times in 2-fold dilution series. All sensorgrams were processed by first subtracting the binding response recorded from the control surface (reference spot), followed by subtracting the buffer blank injection from the reaction spot. All datasets were fit to a simple 1:1 Langmuir S67 interaction model with the provided Biacore 4000 Evaluation software, to determine kinetic rate constants (k on , k off ) or steady-state affinity (K Dss ).
Kinetic characterization using an ADP-GLO TM kinase assay: The BMX enzyme system (Catalog. V4512) and the ADP-Glo TM kinase assay were purchased from Promega Corporation (USA). In each kinase reaction, the concentration of BMX was set to 4.5 ng/µL. The peptide substrate Poly ( Crystallization, data collection, and structure determination of BMX catalytic domain in complex with the inhibitor JS24: The guidelines for plasmid construction and vector cloning are described by Muckelbauer et al.. 13 The expression of BMX protein using Sf-9 cells, as well as the purification process, were optimized to improve sample quality at the end of purification, in order to increase the likelihood of protein crystallization. 14  and COOT. 20 Initial structure refinement was undertaken with REFMAC. 21 The stereochemical restraint dictionary for ligand JS24 was created with JLIGAND 22 and the ligand was manually fitted into the electron density using COOT. Refinement was continued with PHENIX, 23 alternating with manual model editing in COOT between refinements against s A -weighted 2|F o |-|F c | and |F o |-|F c | electron density maps. In the final refinement cycles, hydrogen atoms were added and refined in calculated positions, Translation-Libration-Screw rigid-body anisotropic atomic displacement parameters were refined, water molecules added automatically and the relative weights between the crystallographic and stereochemical energy terms optimized.

S70
Each BMX molecule was divided into 4 rigid-body segments, estimated from the TLSMD server 24 using the isotropic atomic displacement parameters from a previous refinement run.
The final refinement was carried out to 2.0 Å against the STARANISO dataset and the statistics are included in ESI (Table S5). Figures were prepared with PYMOL. 25

Molecular Dynamics simulations on BMX covalently linked to JS24 and JS27:
The X-ray structure of JS24 bound to BMX was used as starting structure in the MD simulations. Parameters for ligands JS24 and JS27 were generated with the antechamber module of Amber18, 26 using the general Amber force field (GAFF), 27 with partial charges set to fit the electrostatic potential generated with HF/6-31G(d) by RESP. 28 The charges were calculated according to the Merz-Singh-Kollman scheme using Gaussian 09. 29 The ff14SB force field, which is an evolution of the Stony Brook modification of the Amber 99 force field (ff99SB), 30 was used for the protein. Each complex was immersed in a water box with a 10 Å buffer of TIP3P water molecules. 31 The system was neutralized by adding explicit counter ions.
A two-stage geometry optimization approach was performed. The first stage minimizes only the positions of solvent molecules and ions, and the second stage is an unrestrained minimization of all the atoms in the simulation cell. The systems were then gently heated by incrementing the temperature from 0 to 300 K under a constant pressure of 1 atm and periodic boundary conditions. Harmonic restraints of 30 kcal·mol -1 were applied to the solute, and the Andersen temperature coupling scheme 32 was used to control and equalize the temperature.
The time step was kept at 1 fs during the heating stages, allowing potential inhomogeneities to self-adjust. Water molecules are treated with the SHAKE algorithm such that the angle between the hydrogen atoms is kept fixed. Long-range electrostatic effects are modelled using the particle-mesh-Ewald method. 33 An 8 Å cutoff was applied to van der Waals interactions. Each system was equilibrated for 2 ns with a 2-fs time-step at a constant volume and temperature of S71 300 K. Production trajectories were then run for additional 0.5 µs under the same simulation conditions.

Target Engagement Intracellular Kinase Assay:
In-cell target engagement was performed at the Reaction Biology Corporation (USA) using the NanoBRET TM technology. Very briefly, HEK296 cells, purchased from ATCC, were transfected with BMX and treated in duplicate with test compounds, BMX-IN-1 or JS25, and with the reference compound Dasatinib, for 1 hour of incubation. Compounds were diluted 10 times with 3-fold dilution, starting at 1 µM. Curve fits were performed only when % NanoBret signal at the highest concentration of compounds was less than 55%. The IC 50 values were determined using the GraphPad Prism 8 (USA).

Cellular activity Assay: LNCaP and PC-3 Cell Growth Assay
Cells were seeded in white, opaque-bottom 96-well plates at 5000 cells/well (LNCaP) or 2000 cells/well (PC-3) in a total volume of 100 µL of culture media. Serial diluted compounds (2fold) in 100 µL media were added to the cells 24 hours later. After 96 hours incubation cellular viability was assessed by CellTiter-Glo ® (Promega) according to the manufacturer's instructions. The values were normalized to vehicle and IC 50 was calculated using GraphPad Prism 8 (USA).

Cellular activity Assay: Multi cell line Growth Assay
Compound activity was profiled against 14 human cell lines from different tissues in a 384well format, opaque white assay plates at 500-1000 cells per well using a semi-automated system. Cells were incubated at 37 ºC and 5% CO 2 . Compound stocks were plated in a 384well format in 11-point and 2-fold concentration ranges. Compounds were pin-transferred into duplicate assay plates and incubated for72h. ATP levels were assessed by CellTiter-Glo ® (Promega) according to the manufacturer's instructions. The values were normalized to vehicle and GI 50 was calculated using GraphPad Prism 8. When ambiguous fit was observed curves were top (100%) and bottom (0%) constrained and GI 50 was determined with 4-P least squares fit. In these cases SD is not calculated by GraphPad Prism 8.

Propidium iodide assay:
LNCaP cells were seeded in 24 well-plates at 8000 cells/well in in a total volume of 500 µL tubulin antibody (Cell signaling Technology) were used in the Western blot. Band intensity was measured in ImageJ.

Accession Codes:
The final refined coordinates and observed structure factors for The PDB access code for the structure of JS24 bound to BMX, were submitted to the Worldwide Protein Data Bank with accession code 6I99.