All research performed complied with the ethical regulations of the regions where the studies were performed. All mouse studies were performed under German and European animal welfare regulations and approved by local authorities (LAGeSo). All dog studies were performed in accordance with the Organization for Economic Co-operation and Development (OECD) principles of good laboratory practice as accepted by regulatory authorities throughout the European Union, USA (Food and Drug Administration (FDA) and Environmental Protection Agency) and Japan (Ministries of Health, Labor and Welfare, Agriculture, Forestry and Fisheries and Economy, Trade and Industry), as well as other countries that are signatories to the OECD Mutual Acceptance of Data agreement.

Proteins, BH3-derived peptides and compounds

MCL1 protein (amino acids 173–321), used for in vitro biological assays (surface plasmon resonance (SPR), HTRF and X-ray crystallography), was expressed and purified as previously described using a bacterial expression system49. Bcl-2 and Bcl-XL protein for HTRF assays was purchased from BPS Bioscience. BH3-derived peptides (for example, Noxa and Bad) were purchased from Biosyntan with the following sequences: Noxa, biotin-PEG2-PEG2-PAELEVE-Nva-ATQLRRFGDKLNFRQKLL-amide; Bad, biotin-PEG2-PEG2-NLWAAQRYGRELRR-Nle-SDEFVDSFKK-amide. Compounds 1 and 2 and BRD-810 were synthesized as described in WO 2017198341 (compound 1), WO 2019096905 (compound 2) and WO 2020236556 (BRD-810). The MCL1 inhibitors AZD5991 (ref. 55), AMG176 (ref. 41) and S64315 (ref. 43) were made in house according to published methods. Etoposide and cisplatin were purchased from Selleckchem.

Biochemical and biophysical assays

Time-resolved fluorescence energy transfer (TR-FRET) assays were used to measure BRD-810-mediated dose-dependent inhibition of MCL1 and Noxa (BH3-derived peptide) binding as previously described in detail (reference patent). Briefly, MCL1 protein served as the protein receptor and Noxa served as the tracer ligand, a compound dose range of 0.1–20,000 nM was tested in duplicate across the same microtiter plate and the assay was initiated by the addition of 2 µl of a 2.5-fold concentrated MCL1 protein solution (1 nM final concentration) in aqueous assay buffer to the compounds in the assay plate. The fluorescence emission at 620 nm and 665 nm after excitation at 330–350 nm was measured using Rubystar (BMG Lab Technologies), Pherastar (BMG Lab Technologies) or Viewlux (PerkinElmer) plate readers. IC50 values were determined by regression analysis based on a four-parameter (4-PL) equation (minimum, maximum, IC50 and Hill; y = max + (min − max)/(1 + (x/IC50) Hill) using the Screener Software (Genedata). For selectivity testing with Bcl-XL and Bcl-2, similar protocols were used with the following modifications: Bcl-XL protein receptor with BAD peptide as the tracer ligand or Bcl-2 protein receptor with BAD peptide as the tracer ligand.

For SPR assays measuring BRD-810 binding to MCL1 protein, Biacore T200 or S200 (Cytiva Life Sciences) instruments were used as previously described (patent WO 2019096922). For SPR measurements, MCL1 protein was immobilized using standard amine coupling and serial dilutions of compound (eight dilution steps, typically ranging from 0.2 nM up to 1,000 nM) were injected over immobilized protein. The double-referenced sensorgrams were fit to a simple reversible Langmuir 1:1 reaction mechanism as implemented in the Biacore T200 and S200 evaluation software.

Cocrystallization of MBP–MCL1 bound to BRD-810

Crystallization was accomplished using our previously reported method47 with several modifications to provide a rapid, reliable crystallization system with improved cocrystallization success rates for low-solubility compounds. Briefly, 2.5 μl of solubilized BRD-810 solution (1 mM in 20% PEG400, 40% methyl prednisolone and 40% DMSO) was placed on the side of a thin-walled PCR tube. To improve compound solubilization, this drop was rapidly flushed with 7.5 μl of maltodextrin-binding protein (MBP)–MCL1 protein (10 mg ml−1 in 2 mM maltose, 20 mM HEPES pH 7.5, 200 mM NaCl, 1% glycerol and 2 mM DTT), mixed thoroughly and allowed to incubate for 15 min on ice. Then, 3 μl of this protein–compound mix was combined with 3 μl of precipitant (seed) solution (16% PEG3350, 50 mM magnesium formate, 20 mM HEPES pH 7.5 and 1:10,000 diluted microseeds) in an EasyXtal 15-well DropGuard crystallization tool. This completed mixture was sealed and allowed to equilibrate against 400 μl of 1.5 M NaCl. Protein crystals appeared within 1 day and continued to grow for approximately 7 days. A single crystal with dimensions of approximately 40 μM × 80 μM × 200 μM was then harvested directly from the crystallization drop in a nylon cryoloop and flash-frozen by rapidly plunging into liquid nitrogen. X-ray diffraction datasets were collected from frozen single crystals at the ALBA synchrotron and processed and refined with the programs xia2 and DIALS from the CCP4 program suites59,60,61 (Supplementary Table 2). Automated refinement and ligand identification were performed using DIMPLE and a known MBP–MCL1 protein as a model (Protein Data Bank (PDB) 4WGI). Iterative model rebuilding and refinement were performed using the programs Coot and Phenix62,63. Figures were generated using PyMOL and MOE. All software used for data processing, refinement and modeling were accessed using SBGrid64.

Cell lines

The following cell lines were used in this study: RKO (colon carcinoma; American Type Culture Collection (ATCC), CRL-2577), HMC1-8 (breast cancer; Japanese Collection of Research Bioresources, JCRB0166), MV-4-11 (AML; ATCC, CRL-9591), THP-1 (AMP; ATCC, TIB-202), MOLP-8 (multiple myeloma; DMSZ, ACC569), AMO-1 (multiple myeloma; German Collection of Microorganisms and Cell Cultures (DSMZ), ACC538), SU-DHL-4 (DLBCL; DSMZ, ACC495), SU-DHL-5 (DLBCL; DSMZ, ACC571), SU-DHL-10 (DLBCL; DSMZ, ACC576), KMS-12-BM (multiple myeloma; DSMZ, ACC551), RPMI 8226 (multiple myeloma; DSMZ, ACC402), OPM-2 (multiple myeloma; DSMZ, ACC50), JJN-3 (plasma cell leukemia; DSMZ, ACC541), KMS-12-PE (multiple myeloma; DSMZ, ACC606), DMS114 (lung cancer; ATCC, CRL2066), A-427 (lung cancer; ATCC, HT-B53), HCC-1187 (breast cancer; ATCC, CRL-2322), SNU398 (liver cancer; ATCC, CRL-2233), HCC-2157 (breast cancer; ATCC, CRL-2340), PA-1 (ovarian cancer; ATCC, CRL-1572), NCI-H82 (lung cancer; ATCC, HTB-175), SNU-16 (gastric cancer; ATCC, CRL-5974) and A-431 (melanoma; ATCC, CRL-1555). Authentication of cell lines was performed using karyotyping and PCR analysis with species-specific primers. Authentication was performed together with Mycoplasma testing before cell stocks were frozen. Cell stocks were used for a maximum of 3 months before a fresh vial was thawed.

Cellular assays

BAK;BAX double-knockout cells were generated using the CRISPR–Cas9 system. Briefly, HMC1-8 cells were infected with lentivirus made from pLX-311-Cas9-Blast plasmid and packaging plasmid (Genetic Perturbation Platform (GPP), Broad Institute) to establish a stable cell line expressing Cas9. Blasticidin (5 μg ml−1) was used for positive selection. Then, those cells were infected with lentivirus made from packaging plasmids and a single guide RNA (sgRNA) vector (pXPR-003, GPP, Broad Institute) containing an sgRNA sequence (BAK, CACCGGTTGATGTCGTCCCCGATGA; BAX, CACCGAGCGAGTGTCTCAAGCGCAT). Virus production and infection were based on the Broad Institute GPP protocols (https://portals.broadinstitute.org/gpp/public/resources/protocols). Pooled cells were used in the viability and caspase activity experiments.

The hiPS cell-derived cardiomyocytes used in this study were commercially available (Ncyte, Ncardia, cat. no. Nc-C-BRCM). They were thawed and cultured according to Ncardia’s instruction manual and maintained with Ncardia’s proprietary cardiomyocyte culture medium (Ncardia, cat. no. Nc-M-CMCM-250). The cells were seeded at a density of 15,000 cells per well in 96-well plates for all subsequent assays. BRD-810 and doxorubicin (Selleckchem, cat. no. S1208) were dissolved as 1 mM and 10 mM stock concentrations in DMSO, respectively, and were diluted accordingly to the desired test concentrations used in the cells (0.1 and 1 μM for BRD-810 and 10 μM for doxorubicin). Compounds were added 7 days after seeding the cells and the DMSO solvent did not exceed 0.1% (v/v) for each well. The 96-well plates with cells were maintained in an incubator at 37 °C in 5% CO2 for the duration of the experiment.

Immunoprecipitation and western blot analysis were performed as previously described using antibodies for Mcl1 (Cell Signaling Technology, clone D35A5, cat. no. 5453; 1:1,000), BAK (Cell Signaling Technology, clone D2D3, cat. no. 6947; 1:1,000) and Bim (Cell Signaling Technology, clone C34C5, cat. no. 2933; 1:1,000)65. The activity of caspases 3 and 7 was determined in multiple cell lines (SU-DHL-4, SU-DHL-5, SU-DHL-10, HMC1-8 parental, HMC1-8 Cas9 control, and HMC1-8 BAK;BAX knockout) upon treatment with different compounds using the Caspase-Glo 3/7 reagent from Promega (cat. no. G8092). Cell lines were plated in culture medium (RPMI 1640; Gibco, cat. no. 22400-089) supplemented with 10% FBS at a density of 3,300 cells in 30 pl per well in a sterile, solid, black, flat-bottom polystyrene tissue culture (TC)-treated 384-well microplate (Corning, cat. no. 3571) using a Multidrop Combi reagent dispenser. As a control, medium without cells was also added to the plate. Cells were incubated in a humidified incubator at 37 °C overnight. On the next day, the cells were treated with compounds (stock solution, 10 mM in DMSO) using the HP D300 digital dispenser in a concentration range of 3.3 × 10−5 M (33,000 nM) to 5 × 10−9 M (5 nM) in a single-dot curve with at least 16 dilutions and a DMSO concentration of 0.33%. Rim wells were excluded. The cells were incubated for 3 h in a humidified incubator at 37 °C. After this incubation, 30 pl of Caspase-Glo 3/7 reagent (Promega, cat. no. G8092) was added to each well using the Multidrop Combi reagent dispenser, followed by 1-h incubation at 37 °C. Finally, luminescence was read at 0.1 ms with a gain of 3,000 using the PHERAstar FS microplate reader (BMG Labtech).

Gene expression was quantified in a breast cancer cell panel using qPCR. Total RNA was extracted using the RNeasy Mini kit (Qiagen) and complementary DNA was generated using the SuperScript III first-strand synthesis kit (Thermo Fisher). TaqMan probe-based qPCR (Thermo Fisher) was carried out using a TaqMan fast advanced master mix and TapMan primer + probe sets for BCL2L1 Hs00236329_m1 and BAK Hs00832876_g1 (both from Thermo Fisher). qPCR was measured on a LightCycler 480 (Roche) real-time PCR system. Gene expression was normalized to GAPDH expression levels measured with the primer + probe set Hs02786624_g1 (Thermo Fisher).

Cell viability assays

The impact of compounds on the proliferation of different cell lines was assessed using the CellTiter-Glo luminescent cell viability reagent from Promega (cat. no. G7573). The different cell lines were plated in culture medium (RPMI 1640; Biochrom, cat. no. FG 1215]) supplemented with 10% FCS (Biochrom, cat. no. S 0415) at a density of 3,300 cells (for suspension cells) or 800 cells (for adherent cells) in 30 pl per well in a sterile, solid, black, flat-bottom polystyrene TC-treated 384-well microplate (Corning, cat. no. 3571) using a Multidrop Combi reagent dispenser. In parallel, cells were plated in a reference (day 0) plate for time zero determination. Cells were incubated in a humidified incubator at 37 °C overnight. On the next day, cells were treated with compounds (stock solution, 10 mM in DMSO) using the HP D300 digital dispenser in a concentration range of 3.3 × 10−5 M (33 µM) to 5 × 10−9 M (5 nM) in a single-dot curve with at least 16 dilutions and a DMSO concentration of 0.33%. Rim wells were excluded. The cells were incubated for 72 h in a humidified incubator at 37 °C. The day-0 plate was measured by adding 30 pl per well of CellTiter-Glo luminescent cell viability reagent (Promega, cat. no. G7573) to time zero wells in the reference plate followed by a 10-min incubation and luminescence reading at 0.1 ms using the PHERAstar FS microplate reader (BMG Labtech).

For viability evaluation according to nuclear count (high-content imager), cells were fixed with 4% PFA (Thermo Fisher, cat. no. 28908) and were stained with DAPI dye. Following DAPI staining, each plate was imaged with an ImageXpress Micro confocal high-content imaging system (Molecular Devices). Using the MetaXpress image analysis software (Molecular Devices), the number of nuclei per well was calculated for all wells. Cell viability was expressed as the number of nuclei in the well divided by the average number of nuclei of vehicle control (DMSO). Data from three replicate wells of each condition were averaged. Bar graphs represent the average and error bars represent the s.d. of the averages.

The levels of extracellular LDH were evaluated using the commercially available LDH Glo assay (Promega, cat. no. J2380). The assay was performed according to the manufacturer’s instructions and luminescence was recorded using a CLARIOstar (BMG) plate reader.

ELISA assay for MCL1–BAK or MCL1–BIM interaction

On day 1, RKO colon cancer cell line cells were plated at 0.8 million cells per ml (100 µl per well) in 96-well flat-bottom TC-treated plates (Corning). MCL1 antibody (Santa Cruz, sc-12756) was diluted 200-fold (final concentration, 1 µg ml−1) in carbonate buffer (pH 9.6; Thermo Fisher) and 50 µl of diluted antibody was added to each well of high-bind ELISA plates (SARSTEDT). Each plate was tapped to make sure liquid covered the entire bottom of the wells and then incubated at 37 °C overnight. On the second day, MCL1 antibody was washed from the ELISA plate. Then, 250 µl of Odyssey blocking buffer (PBS) (Li-Cor) was added to each well, incubated at room temperature for at least 1 h and then washed once with 250 µl of 1× PBST. BRD-810 was added to plates with RKO cells using a HP Tecan D300e compound dispenser in a threefold dilution series (highest dose 30 µM), with ten doses per compound in quadruplicate. After incubation at 37 °C for the indicated times, plates with RKO cells were gently washed once with 100 µl of PBS per well and then 50 µl of CHAPS buffer (50 mM Tris-CI pH 7.4, 150 mM NaCl, 1% CHAPS, 1 mM EDTA, 1 mM EGTA, complete protease inhibitors (Roche) and PhosSTOP (Roche)) was added to each well. Plates were rocked for 1 h at 4 °C and then 45 µl of cell lysate from each well was transferred to ELISA plates coated with MCL1 antibody. Plates were incubated overnight in the cold room with rocking. On the third day, ELISA plates were washed once with 250 µl of 1× PBST. Biotinylated anti-BAK (Santa Cruz, sc-1035) or anti-Bim (Cell Signaling, 2933) antibody was added and incubated for 2 h. Wells were washed once with 250 µl of 1× PBST. Streptavidin poly(horseradish peroxidase) (Thermo Fisher) was diluted to 20 ng ml−1 in Odyssey blocking buffer plus 0.05% Triton X-100 and 100 µl was added to each well of the ELISA plate. Plates were incubated at room temperature for 1 h with rocking and then washed with 100 µl of 1× PBST three times. Each SuperSignal ELISA Femto maximum-sensitivity substrate (Thermo Fisher) was added to a 50-ml tube and mixed and then 100 µl of mixed substrate was added to each well. Plates were shaken for 1 min and then luminescence was measured using an Envision plate reader (HP). The signal of each well was normalized by the no-compound control and no-cell control. The IC50 was calculated using GraphPad Prism software.

Bioinformatic analysis

To identify a subset of baseline genomic features of the cell lines that best explained the measured MCL1 inhibitor sensitivity (area under the curve (AUC)), a variational approximation to a Bayesian linear regression variable selection model was used, implemented in the ‘varbvs’ R package66. Five different values for the ‘logodds’ hyperparameter were used, ranging from 10−5 to 10−1 and were shown as the posterior mean regression coefficients plotted against the Pearson correlation between each feature and the AUC.

In vivo safety studies in dog

All dog studies were performed in accordance with the OECD principles of good laboratory practice as accepted by regulatory authorities. Purpose-bred, naive beagle dogs (6–9 months old; Supplier Marshall UK for Charles River) were provided with an intravenous port and were individually fitted with a custom-designed dog jacket, including a protective collar (if necessary), and acclimatized to them for a minimum of 5 days before the treatment period started. Dogs were randomized into four groups: vehicle group (five male and five female), 3 mg kg−1 BRD-810 group (three male and three female), 8 mg kg−1 BRD-810 group (three male and three female) and 15 mg kg−1 BRD-810 group (five male and five female). During the treatment period, BRD-810 was infused once weekly at 3, 8 or 15 mg kg−1 for 4 weeks. At the end of the 4-week treatment period, two male and two female dogs from the vehicle group and the 15 mg kg−1 group were admitted to the recovery period. Full toxicology analysis was performed including cardiac toxicity evaluation.

Cardiac toxicity evaluation

ECGs were measured using a Cardiofax ECG 9620 electrocardiograph (Nihon Kohden Europe). Heart tissue for histopathology was collected at the end of the study and embedded in paraffin, sectioned, mounted on glass slides, stained with hematoxylin and eosin and evaluated histopathologically by a board-certified veterinary pathologist blinded to treatment groups.

For the troponin I assay, supernatants (50 μl per well) were collected from each well of the 96-well plates in culture and were immediately frozen at −80 °C. On the day of the assay, the supernatant samples were thawed at room temperature and were used to perform the troponin I AlphaLISA assay (PerkinElmer, cat. no. CUSM86458000EASBR1) according to the manufacturer’s instructions. The standard curve of the assay was generated using a nonlinear regression 4-PL sigmoidal curve fit and the LLOD (lower level of detection) was calculated at 1.8 pg ml−1 cTnI. Positive control and test samples with a notable cardiotoxic effect (exceeding the 1,000 pg ml−1 threshold value) were found within the dynamic range of the assay.

In vivo efficacy and assays

BRD-810 efficacy was tested in xenograft models of multiple myeloma (MOLP-8 and AMO-1), DLBCL (SUD-HL-10), liver cancer (SNU398), lung cancer (A-427) and AML (THP-1). Human tumor cells (5–10 × 106 suspended in Matrigel) of the respective cancer type were injected subcutaneously into 6–8-week-old immunocompromised mice (female CB17/SCID mice (C.B-Igh-1b/IcrTac-Prkdcscid), Taconic). Once the primary tumor growth was established (>30 mm2) animals were randomized and assigned to study groups. Animals received intravenous treatment with BRD-810 at the indicated doses and schedules or vehicle control. The treatment response to BRD-810 versus control groups was assessed by a determination of tumor area using a caliper. Animals were killed once they were close to reaching the maximum bearable tumor size of 250 mm2. The tumor size was not exceeded. The principles of such xenograft studies were previously summarized67.

For in vivo caspase 3 measurements in the MOLP-8 model, tumors were sampled from mice at different time points following a single intravenous dose of BRD-810 and protein was extracted. Western blot analysis was performed using human-specific antibodies for total caspase 3 (Cell Signaling Technology, clone 3G2, cat. no. 9668; 1:1,000) and cleaved caspase 3 (Abcam, clone E83-77, cat. no. ab32042; 1:1,000). Western blot bands were quantified using densitometric analysis.

Statistics and reproducibility

No statistical methods were used to predetermine sample sizes but our sample sizes are similar to those reported in previous publications41,55. Sample sizes for dog studies were chosen according to the FDA International Council for Harmonization of Technical Requirements for Pharmaceuticals for Human Use S9 guideline. In vitro experiments were performed with at least three biological replicates and repeated in independent experiments. All attempts at replication were successful. For in vivo studies, mice were randomized according to the ‘tumor size matched distribution’ method before the start of treatment to ensure that each group started with an approximately equal mean tumor size. Male and female beagle dogs were randomized separately to achieve similar group mean body weights while ensuring that litter mates were homogeneously distributed across all groups where possible. The person performing histopathology analyses was blinded to the study. The persons performing the in vivo mice studies were blinded to the expected outcome and mode of action of the compounds tested. For all other studies, data collection and analysis were not performed blind to the conditions of the experiments. No data were excluded from the analyses. Statistics were performed on values from independent experiments. All statistical calculations were performed with GraphPad Prism 10. Data distribution was assumed to be normal but this was not formally tested. Error bars represent the s.d. Box plots are shown with the median and quartiles as described in the figure legends.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.