Targeting Aggressive B-cell Lymphomas through Pharmacological Activation of the Mitochondrial Protease OMA1 Open Access

Malignant cells and tumors require robust homeostatic regulation to adjust the metabolic demands of rapid proliferation with nutrient availability and accumulation of metabolic end products to survival (1). To meet these demands, tumors cells rely on endogenous stress survival pathways (2). On the other hand, utilization of stress signaling pathways also creates dependencies that present therapeutic opportunities.

Several mitochondria quality control (MQC) pathways serve to maintain the integrity of mitochondria. MQC pathways sense changes in mitochondrial oxidative phosphorylation, membrane potential, proteostasis, and translation of mitochondrial DNA encoded proteins (3, 4). MQC pathways signal to cytosolic pathways that control the ATF4-Integrated Stress Response (ISR), an adaptive gene expression program controlling amino acid biosynthesis and transport, redox homeostasis, and enhanced protein folding (3, 5). The ATF4-ISR facilitates adaptation to the tumor microenvironment and can be pharmacologically targeted (6, 7). At the same time, persistent ISR activation can sensitize cells to apoptotic stimuli by altering the levels of pro- and antiapoptotic proteins (8–10).

Herein, we describe the preclinical pharmacology of a novel class of pyrazolo-thiazole derivatives (hereafter BTM-compounds), first described for their ability to induce cell growth arrest and cell death in a wide variety of solid and hematopoietic tumor cell lines (11). We demonstrate that select BTM compounds specifically hyperactivate the ISR through activation of the MQC protease OMA1, thereby shifting the equilibrium from ATF4-ISR controlled prosurvival effects towards tumor cell death. Sensitivity to BTM compounds is suppressed by the mitochondrial protein FAM210B, which is minimally expressed across multiple subtypes of diffuse large B-cell lymphoma, providing a therapeutic opportunity for OMA1 activators.

The structures and synthesis of the compounds used in this article are described in the following US patents:. US10537558B2 (12) describes the synthesis of BTM-3528 (compound 197) and BTM-3532 (Compound 201a). US 20190345152a1 (13) describes the synthesis of BTM-3566 (Compound 5).

BJAB (DSMZ Cat# ACC757 RRID:CVCL_5711) and OCI-Ly7 (DSMZ Cat# ACC 688, RRID:CVCL_1881) cells were obtained from the German Collection of Microorganisms and Cell lines. HCT-116 cells (ATCC Cat# CCL-247, RRID:CVCL_0291), SU-DHL-10 cells were (ATCC Cat# CRL-2963, RRID:CVCL_1889), and HeLa cells (ATCC Cat # CCL-2, RRID:CVCL_0030) were purchased from the American Type Culture Collection. All cell lines were tested as Mycoplasma negative by the vendors. The cells were brought in as tested from ATCC or DSMZ, expanded and then batches placed into storage. Any batch of cells was passaged for approximately 3 months then replaced with a fresh stock from frozen storage.

Screening of tumor cell lines was performed by Crown Bioscience. Cells were plated at a starting density of 4 × 10³ cells/well and incubated for 24 hours. BTM-3528 was prepared as a 10× solution of test article with a final working concentration of 30 μmol/L of test article in media with nine 3.16-fold serial dilutions. Following the addition of BTM-3528, the plates were incubated for an additional 96 hours at 37°C with 5% CO₂. Final cell numbers were determined using the Cell-Titre Glo assay (Promega; Catalog#: G7571). The absolute IC₅₀ curve was fitted using a nonlinear regression model with a sigmoidal dose response. Activity area (AUC) for each compound was determined by calculating the integrated area bounded for each dose response curve fit. The AUC reflects both the magnitude of effect (maximal inhibition) and potency (IC₅₀).

To quantify apoptosis, BJAB were cultivated using RPMI (ATCC Cat # 30–2001) with 15% fetal bovine serum in a 96-well format and treated with BTM compounds for the indicated time intervals. Cells were washed twice with ice-cold PBS and resuspended in 1× Binding Buffer (BD Biosciences Catalog# 51–66121E) at a concentration of 1 × 10⁶ cells/ml. To 0.5 × 10⁵ cells, 2.5 μL of Annexin V-APC (BD Biosciences Catalog# 550474) or Annexin V-FITC (BD Bioscience Catalog# 556419) were added and incubated for 15 minutes at room temperature in the dark. Cells were washed once with Binding Buffer and the pellets were resuspended in 100 μL Binding Buffer containing either 2 μL Propidium Iodide (50 μg/mL) or 2 μL DAPI (1 mg/mL). Cells were analyzed on a Cytoflex S (Beckman Coulter).

The human colon adenocarcinoma cell line HCT-116 was used to evaluate the effects of BTM compounds on gene expression. To fully evaluate the effects of the compound on cell-cycle controlled genes, cells were synchronized prior to BTM compound treatment. HCT-116 cells grown in McCoys 5a Media (ATCC Cat # 30–2007) supplemented with 10% fetal bovine serum and penicillin/streptomycin were first blocked in the S-phase by treatment with thymidine. After 24 hours, the thymidine containing media was removed and replaced with media containing nocodazole to block cells in the M-phase in a high degree of synchrony. Cells were then released into G₁ either in complete medium or complete medium plus 10, 1, or 0.1 μmol/L BTM-3528. Cells were harvested at five time points: 1, 2, 4, 6, and 8 hours after release into the G₁ phase, and mRNA extracted for Illumina RNA sequencing (RNA-seq). Three replicates of each concentration and time point along with time point specific controls (i.e., cells without compound) were collected for RNA-seq.

Human cell line xenograft models were established using SU-DHL-10 cells. Cells were grown in RPMI1640 (ATCC Cat# 30–2001) supplemented with 15% fetal bovine serum and penicillin/streptomycin. Cells were harvested by centrifugation and resuspended in cold 50% serum-free medium: 50% Matrigel to generate a final concentration of 2.50E+07 trypan-excluding cells/mL. Female Envigo SCID beige mice (C.B-17/IcrHsd Prkdc^scidlyst^bg-j) were implanted subcutaneously high in the right axilla on day 0 with 5 × 10⁶ cells/mouse. Mice were randomized into groups based on tumor volume with a mean tumor burden for each group of 150 mm³. BTM-3566 was prepared as a solution in dosing vehicle containing 5% NMP, 15% PEG400, 10% Solutol, and 70% D5W. The final dose concentration was 4 mg/mL, and the dose volume was 5 μL/gram. All mice were dosed by oral gavage once daily for 21 days. Tumor volume and body weights were determined every third day. All mice were dosed according to individual body weight on the day of treatment.

For patient derived xenograft models, all tumors were sourced from Crown Bio. Models were established in female mice with an average body weight of 25 grams. Balb/c nude were sourced from GemPharmatech Co., Ltd (LY0257); NOD SCID mice from Shanghai Lingchang Biotechnology Co., Ltd (LY2214, LY2264, LY2345, LY3604, LY6701) or NPG/NOD/SCID from Beijing Vital Star Biotechnology Co, Ltd (LY6933, LY6934) mice. Each mouse was inoculated subcutaneously in the right flank region with fresh tumor derived from mice bearing established primary human cancer tissue. Mice were randomized into vehicle or treatment groups with a mean tumor burden of 200 mm³. All mice were dosed once daily by oral gavage for 21 days. Tumor volume and body weights were determined three times per week.

All animal studies were conducted under strict ethical and animal welfare guidelines following an approved IACUC protocol from the host institutions. Although the studies were not conducted in accordance with the FDA Good Laboratory Practice regulations, 21 CFR Part 58, all experimental data management and reporting procedures were in strict accordance with applicable industry practices Guidelines and Standard Operating Procedures.

HCT-116 cells were plated in a 35 mm dish containing a 14 mm, uncoated coverslip (MatTek). The next day, cells were stained with McCoy media containing 200 nmol/L Mitotracker green (MTG) and 15 nmol/L of tetramethyl rhodamine, ethyl ester (TMRE) for 40 minutes. Cells were then treated with 3 μmol/L BTM-3528 or 4 μmol/L FCCP in McCoy 5a growth media containing 15 nmol/L TMRE, but lacking MTG, and incubated for 3 hours. Imaging was perfumed using a 63× objective and in a Zeiss LSM880 confocal microscope with Airyscan mode.

Cells were plated in black clear-bottom 96-well plate and stained with McCoy media containing 200 nmol/L MTG and 15 nmol/L of TMRE for 40 minutes. Cells were then treated with 3 μmol/L BTM-3528 or 4 μmol/L FCCP in McCoy containing 15 nmol/L TMRE, but lacking MTG, and incubated for 30 minutes. MTG and TMRE signal was obtained by imaging in two wavelengths (MTG – Excitation 460–490 nm and Emission 500–550 nm; TMRE – Excitation 560–580 nm and Emission 585–605 nm) using a PerkinElmer Operetta microscope system with 40× lens. Five fields of each well were analyzed and TMRE/MTG fluorescence ratio was determined by image processing with Harmony software.

Mitochondrial morphology was assessed using Fiji/ImageJ software and a Trainable Weka Segmentation plugin. Mitochondrial membrane potential, TMRE/MTG fluorescence ratio was calculated from segmented mitochondrial structures obtained by MTG channel. One-way ANOVA and Tukey multiple comparison test were used for statistical analysis; P values ≤0.05 (*) were considered significantly different.

Respirometry assays were run on a Seahorse Extracellular Flux Analyzer (Agilent). HCT-116 cells were seeded at 14,000 cells/well using XF96 well microplates and incubated overnight (37°C and 5% CO₂) in McCoy's 5a Modified Medium culture medium with 10% FBS. Before the respirometry assay, cells were washed with assay medium: DMEM with 10 mmol/L glucose, 2 mmol/L glutamine, 1 mmol/L pyruvate, 5 mmol/L HEPES, and 10% FBS (pH 7.4). BTM compounds were tested at a final concentration of 3 μmol/L, and cells were either acutely treated during the assay or pretreated for 4 hours before the assay. In pretreatment experiments, compounds were added in complete medium and incubated at 37°C and 5% CO₂. Compounds injected during the assay included 2 μmol/L oligomycin, 1 μmol/L FCCP, and 2 μmol/L of antimycin A and rotenone. Upon completion of each respirometry assay, the cells were stained with 1 μg/mL Hoechst and cell number was measured with an Operetta High-Content Imaging System. The respirometry well level data (pmol O₂/min) was normalized to cell number per well (pmol O₂/min/10³ cells) in each assay.

RNA-seq raw count data, FASTQC, and library manifests have been supplied as supplemental information. The R-Code used to perform the bioinformatic analysis is available upon request from the authors. All other data are available in the main text or the supplementary materials. BTM3528 and BTM3566 are available upon request through a materials transfer agreement (MTA) with Bantam Pharmaceutical.

We previously described a series of pyrazolo-thiazole compounds that displayed anti-proliferative activity across a range of hematopoietic and solid tumor cell lines (11). We further improved this class of compounds by optimizing substituents around the pyrazolo-thiazole core. The medicinal chemistry approach was guided by the potency of derivatives to induce cell death in the human lymphoma cell line BJAB while sparing non-transformed B-cells. Two active lead compounds, BTM-3528 and BTM-3566, emerged from this effort. Both exhibited potent in vitro activity. BTM-3532, a structurally related but biologically inactive compound (Fig. 1A) was also identified for use as a chemical control. We assessed the activity of BTM-3528 and BTM-3566 in a diverse panel of 99 tumor cell lines, including 19 hematopoietic lines and 80 solid tumor lines. BTM-3528 and BTM-3566 demonstrated equivalent activity across solid tumor cell lines, with the greatest activity observed in lung, colorectal and pancreatic cancer and much lower or no activity in cell lines derived from brain, skin, or breast tumors (Fig. 1B and C; Supplementary Table S1). Hematopoietic tumor lines were broadly responsive to both BTM compounds (Fig. 1B and C; Supplementary Table S1), with the most sensitive being Burkitt lymphoma and DLBCL lines (Fig. 1C; Supplementary Table S1). Both BTM-3528 and BTM-3566 exhibited potent, dose-dependent activity against DLBCL lines of diverse genotypes, including MYC-rearranged (“double” and “triple” hit) lymphomas, with >90% growth inhibition and IC₅₀ values of 0.16–0.57 μmol/L (Supplementary Table S2). BTM-3566 induced rapid apoptosis in lymphoma lines, with nearly complete killing at 24 hours (Fig. 1D; Supplementary Fig. S1A), whereas the inactive compound BTM-3532 had no effect on cell viability (Supplementary Fig. S1B). Lymphoma cell death induced by BTM-compounds was accompanied by caspase-3/7 activation and cytochrome c release from mitochondria to the cytosol (Figs. 1E; Supplementary Fig. S1C and S1D). Cell death was dependent on the BCL2 effector protein BAX, whose deletion abrogated BTM-induced apoptosis in BJAB lymphoma cells (Supplementary Fig. S1E). These data demonstrate that BTM compounds trigger the canonical intrinsic apoptotic pathway in lymphoma cells, whereas solid tumor lines do not undergo apoptosis but rather exhibit cell growth inhibition associated with G₁ arrest (Fig. 1E; Supplementary Fig. S2A and S2B).

BTM-3566 has favorable pharmacokinetic properties and potent in vivo activity in human cell line and patient-derived xenograft models. To investigate the pharmacokinetic properties of BTM-3566, we performed intravenous/oral crossover studies in mice (Fig 2A–C). Bioavailability was >90% and the terminal half-life of 4.4 to 6.6 hours was acceptable for oral, once daily dosing.

The therapeutic activity of BTM-3566 was assessed in a human xenograft model using the double-hit lymphoma DLBCL tumor line SU-DHL-10 (Fig 2D). At the 10 mg/kg dose, we observed delayed tumor growth that was lost with further dosing (Fig. 2E, top). At doses at or above 20 mg/kg, BTM-3566 treatment resulted in complete responses (CR; defined as no palpable tumor) in all animals by 10 days of dosing and maintained for 21 days of dosing. To assess whether BTM treatment would induce durable responses, animals were followed for 30 additional days after cessation of dosing. Thirty-day tumor-free survival was maintained in 40% of animals dosed with 20 mg/kg and 60% of animals dosed with 30 mg/kg BTM-3566 (Fig. 2E top). Body weight loss was dose-dependent but <10% at the 20 mpk dose level (Fig. 2E bottom). In the 30 mg/kg dose group, two of 10 mice exceeded 20% body weight loss, necessitating an unscheduled dose holiday. Weight loss in both groups was reversible with cessation of dosing. (Fig. 2E bottom).

Having established the 20 mg/kg dose as effective and well tolerated in the SU-DHL10 model, we next tested BTM-3566 in 9 human DLBCL patient-derived xenograft (PDX) models representing ABC and GCB DLBCL-subtypes with high-risk genotypes (Fig. 2F-H). Treatment with 20 mg/kg BTM-3566 resulted in CR in all 3 mice from 6 of 9 PDX models. When pooling the 27 mice in the treatment arms of the nine models, CR was observed in 66% (19/27), partial response (PR) occurred in another four mice, 2 mice had stable tumors, and two had progressive disease. In summary, the single-agent overall response rate (CR + PR) was 85.2% with all models having at least one animal exhibiting full or partial regression (Fig. 2H).

To elucidate the mechanism of action of the active BTM compounds, we next examined the transcriptional changes induced by compound treatment. We chose the colon cancer cell line HCT-116 for this experiment because DLBCL cell lines rapidly undergo apoptosis upon exposure to BTM compounds. We performed RNA-seq in HCT-116 cells treated with BTM-3528 at 10, 1, or 0.1 μmol/L for 1, 2, 4, 6, or 8 hours, and vehicle-treated cells (see Supplementary Table). Treatment with BTM-3528 induced rapid changes in gene expression, with the first differentially expressed genes appearing at the 4-hour time point (Fig 3A; Supplementary Fig. S3A and S3B). Gene set enrichment analysis revealed upregulation of genes associated with the ATF4-linked integrated stress response (Supplementary Fig. S3C). Expression of ATF4 itself was upregulated by BTM-3528 treatment and many of the most upregulated genes at each time point were direct ATF-4 target genes (Fig. 3A, in red font).

Using qPCR, we confirmed that BTM-3528 increased the expression of ATF4 target genes in responsive solid and hematopoietic tumor lines including HCT-116, the chronic myelogenous leukemia cell line HAP1, and two DLBCL cell lines SU-DHL-2 and SU-DHL-10 (Fig. 3B). In contrast, ATF4-regulated genes were not induced in normal human lung fibroblasts. Consistent with the RNA-seq data, transcriptional responses to BTM-3528 were selective for ATF4- but not ATF6- or ERN1/XBP1–regulated transcripts associated with ER stress (Fig. 3B). Finally, we tested the ability of BTM-3566 to induce the eIF2α-ATF4-ISR in vivo in the SU-DHL-10 xenograft model of DLBCL. Mice were dosed with 3, 10, and 30 mg/kg per day of BTM-3566 for 5 days. Six hours following drug administration, the expression of ATF4 target genes (ASNS, DDIT3, TRIB3, and CDKN1A) in tumors increased in a dose-dependent manner (Fig. 3C). Both the reduction in tumor volume and the increase in ATF4 target gene expression for ASNS, CDKN1A, TRIB3, and DDIT3 correlated with overall BTM-3566 exposure in the mouse (Supplementary Fig. S4).

To confirm activation of the ATF4 regulated ISR, we determined whether phosphorylation of eIF2α and nuclear localization of ATF4 protein were increased in HCT-116 and BJAB DLBCL cells treated with BTM-3528. Western blotting demonstrated a time-dependent increase in eIF2α phosphorylation (Fig. 3D), while immunofluorescent staining of nuclear ATF4 revealed a dose- and time-dependent accumulation of ATF4 protein in the nuclei of HCT-116 cells treated with BTM-3528 (Fig. 3E; Supplementary Fig. S5). Induction of ATF4 by BTM-3528 occurred within 1 hour following treatment with an EC₅₀ of 0.19 μmol/L, consistent with the early upregulation of ATF4 target genes observed in our RNA-seq data.

ATF4 regulated gene expression is controlled by phosphorylation of the translation initiation factor eIF2α by one of four kinases: EIF2-AK1 (HRI), EIF2AK2 (PKR), EIF2AK3 (PERK), and EIF2AK4 (GCN2; refs. 14–16). Each of these kinases responds to separate cell stress conditions: HRI is activated by heme-deprivation, protein misfolding and mitochondrial stress; PERK is activated by the ER unfolded protein response and ER stress; GCN2 is activated by amino acid deprivation leading to accumulation of deacylated tRNA; PKR is activated by the presence of double-stranded RNA. Gene set enrichment implicated HRI, GCN2, and PERK as potential drivers of BTM-3528 mediated ATF4 upregulation (Fig. 3F). Because ER stress also activates the transcription factors ATF6 and XBP1S independently of the activation of PERK, we determined whether ATF6 and XBP1 target genes were affected by BTM-3528 treatment. We found that genes regulated by ER stress and the transcription factors ATF6 or XBP1 (17) were not upregulated by treatment with BTM-3528 (Fig. 3G, top and middle). These data suggest that exposure to BTM-3528 does not lead to PERK activation or ER stress. Using a similar methodology, we found that the transcriptional changes induced by BTM-3528 treatment were highly similar to those observed in the context of mitochondrial stress (Fig. 3G, bottom; ref. 5). Therefore, we hypothesized that mitochondrial dysfunction leading to HRI activation might be upstream of ATF4-ISR induction by BTM-3528.

Having established that BTM compounds trigger the ATF4-integrated stress response together with upregulation of genes associated with mitochondrial stress, we next asked whether BTM compounds induce mitochondrial dysfunction. Several recent studies have shown that agents that depolarize mitochondria and block mitochondrial ATP synthesis activate the mitochondrial protease OMA1. Increased OMA1 activity induces the cleavage of OPA1, a dynamin-like protein that controls inner membrane fusion and cristae shape, leading to mitochondrial fragmentation (18, 19). Consequently, we generated OMA1-knockout HCT-116 cells using CRISPR editing and performed live imaging of cells stained with Mitotracker Green (MTG) to assess the effects of BTM-3528 on mitochondrial structure in wild-type (WT) and OMA1^−/− cells treated with 3 μmol/L BTM-3528. BTM-3528 induced mitochondrial fragmentation, as indicated by the reduction in both mitochondrial aspect ratio and form factor in WT cells, but not in OMA1^−/− cells (Fig. 4A and B).

Mitochondrial fragmentation is associated with cleavage of proteins that are required for maintenance of a fused state. OPA1 has 5 isoforms detected by Western blot (bands a-e) and isoform content is sensitive to OMA1 activation (20). OMA1 activation cleaves the long isoforms of OPA1 (L-OPA1) to short OPA1 isoforms (S-OPA1), resulting in mitochondrial fragmentation (18, 19). L-OPA1 levels were reduced by more than 90% in parental HCT-116 cells, but not in OMA1^−/− HCT-116 cells following 4-hour incubation with 3 μmol/L BTM-3528 or BTM-3566, but not inactive BTM-3532 (Fig. 4C and D). L-OPA1 levels were significantly reduced as early as 30 minutes after treatment with BTM-3528 or BTM-3566 (Fig. 4E and F). As noted above, BTM compounds do not induce apoptosis in HCT-116 cells (Fig. 1F), indicating that the observed extent of OMA1 activation and OPA1 cleavage are not sufficient to induce apoptosis.

Previous studies have demonstrated that a decrease in mitochondrial ATP production, either achieved by depolarization/uncoupling or by blocking electron transfer (18, 21, 22), can induce OMA1 activation. Therefore, we determined whether BTM compounds could depolarize mitochondria or directly inhibit electron transfer. WT and OMA1^−/− HCT-116 cells were treated with BTM-3528, BTM-3566, or the inactive analog BTM-3532 for 3 hours. Live cells stained with TMRE, a dye reporting on real time changes in mitochondrial membrane potential, and MTG, a dye used to visualize mitochondria mass in a membrane potential independent manner. Co-staining with MTG is used to control for changes in TMRE staining that are not caused by differences in membrane potential (i.e., changes in mitochondrial size or mitochondria moving outside the focal plane). While the mitochondrial uncoupler FCCP strongly reduced the TMRE/MTG ratio, BTM compounds had no effect, consistent with a lack of mitochondrial depolarization (Fig. 4G). To further evaluate effects on mitochondrial function, we determined whether acute changes in basal oxygen consumption rate (OCR) occurred after treatment. After 30-minute exposure to BTM-3528 or BTM-3566 in WT cells, no significant changes occurred in basal oxygen consumption in WT or OMA1^−/−cells (Fig. 4H and I). ATP-linked respiration and bioenergetic efficiency were mildly decreased and further reduced by 4 hours of treatment in WT but not in OMA1^−/− cells (Fig. 4I and J; Supplementary Fig. S6A and S6B). Thus, effects on respiration from BTM-3528 and BTM-3566 are dependent on the presence of OMA1 but do not result from direct inhibition of electron transport or uncoupling of mitochondrial respiration.

Having established that BTM compounds induce OMA1 activation and OPA1 cleavage without acting as inhibitors of mitochondrial electron transport, we next sought to define the effector pathway leading to lymphoma cell death. In addition to cleavage of OPA1, activation of OMA1 is also known to induce the cleavage of the mitochondrial protein DELE1, leading to release of cleaved DELE1 into the cytosol, where it binds to and activates HRI (23, 24). To investigate the role of OMA1, DELE1, and OPA1 as effectors of compound activity in DLBCL, we engineered clonal BJAB cell lines with homozygous deletions of OMA1 or DELE1, or where the OMA1-cleavage site in L-OPA1 was deleted (OPA1^ΔS1/ΔS1). Consistent with our results in HCT-116 cells, BTM-3566 failed to induce L-OPA1 cleavage in OMA1^−/− BJAB cells (Fig 5A). Processing of L-OPA1 after exposure to BTM-3566 was also blocked in OPA1^ΔS1/ΔS1 cells, whereas the deletion of DELE1 had no effect on the processing of L-OPA1 (Fig. 5A). Activation of the integrated stress response by BTM-3566, as indicated by phosphorylation of eIF2α and induction of ATF4, was lost in OMA1- and DELE1-knockout cells but preserved in OPA1^ΔS1/ΔS1 cells unable to undergo OMA1-dependent OPA1 cleavage (Fig. 5A). In agreement with these findings, deletion of either OMA1 or DELE1 protected BJAB cells from BTM-3566 and BTM-3528 induced apoptosis, whereas OPA1^ΔS1/ΔS1 BJAB cells remained fully sensitive (Fig. 5B; Supplementary Fig. S7A–S7C).

Activation of HRI has been reported to be contingent on proteolytic cleavage of DELE1 to a shorter form (23–25). As OMA1 dependent cleavage of OPA1 is occurring in as little as 30 minutes, we performed a time course experiment to determine whether levels of DELE1 (DELE1-L) are reduced following treatment with BMT-3566, BJAB cells were transduced with a lentivirus expressing a DELE1-ALFA-tag construct to facilitate detection of DELE1 protein. BTM-3566 treatment resulted in a loss of DELE1-L by 60 minutes of treatment to a greater degree than that of a positive control, oligomycin. BTM-3532 treatment had no effect on the levels of DELE1-L Notably, the time course for reduction in levels of DELE1-L appeared to trail that of OPA1-L, with little to no loss of DELE1-L by 30 minutes (Fig. 5C).

To further determine how the apoptotic signal generated in mitochondria is relayed, we investigated the role of each component of the ATF4-ISR. We created clonal BJAB lines with HRI or ATF4 knockout and a BJAB cell line with a homozygous knock-in of eIF2α^S49A/S52A, in which the serine residues phosphorylated by HRI are mutated to alanine, preventing phosphorylation of eIF2α by HRI. Following treatment with BTM-3528 or BTM-3566 compounds, neither phosphorylation of eIF2α nor increased ATF4 protein was observed in HRI^−/− and eIF2α^S49A/S52A cells. In contrast, phosphorylation of eIF2α was maintained in ATF4-deficient cells (Fig. 5D; Supplementary Fig S7E). HRI^−/− and eIF2α^S49A/S52A BJAB cells were protected from compound-induced caspase activation and apoptosis (Fig.5E; Supplementary Fig. S7D and S7F). ATF4^−/− cells were also protected from compound-induced apoptosis, but the effect was somewhat weaker at later timepoints (Fig. 5E; Supplementary Fig. S7F).

Known ISR effects downstream of eIF2α phosphorylation include downregulation of mRNA translation with concomitant decreases in the levels of short lived oncogenic proteins and increased μORF-mediated translation of ATF4 and other stress effectors (14). We investigated the impact of treatment on the abundance of short-lived oncogenic proteins essential for DLBCL cells and found that BTM-3528 treatment depleted c-MYC and Cyclin D3 (CCND3) proteins and reduced the expression of the anti-apoptotic protein MCL1 after 8 hours of treatment (Fig 5F; Supplementary Fig. S7B and S7C). To quantify cellular protein synthesis, we performed polysome gradients, which revealed that upon treatment with BTM-3566, polysomes from WT and ATF4^−/− BJAB cells exhibited a shift towards smaller size, indicating global attenuation of cap-dependent mRNA translation (Fig. 5G and J). In contrast, cells with knock-out of HRI or knock-in of eIF2α^S49A/S52A had preserved polysome size (Fig. 5H and I). In HRI^−/− or eIF2α^S49A/S52A BJAB cells, the effect of BTM-3566 on c-MYC, CCND3, and MCL-1 was fully rescued, whereas in ATF4^−/− cells, the expression of c-MYC and MCL-1 was largely preserved, but CCND3 was still lost (Fig 5F).

We sought to understand what cellular contexts most affected the activity of BTM-3528 and BTM-3566. Therefore, we extended our screen for compound activity to a larger panel of cell lines (Supplementary Fig. S8). Each cell line was treated in duplicate with a dose titration of BTM-3528 to establish the potency and magnitude of inhibition. The activity area (area under curve, AUC) for each dose response was then computed (Supplementary Table S3). The AUC response data were analyzed for correlation to genomic alterations or gene expression using data from the Cancer Cell Line Encyclopedia (26). No single mutations were found to be associated with compound activity in 284 cell lines for which both gene expression and genomic data were available.

We identified a striking relationship between expression of the mitochondrial protein FAM210B and response to BTM-3528 across all cell lines tested (Fig. 6A; Supplementary Tables S4 and S5, Supplementary Fig. S9). FAM210B expression levels were significantly lower in drug-sensitive cell lines (Fig. 6B). Notably, the most responsive cell lines were derived from B-cell lymphomas (DLBCL, Burkitt, and Mantle cell lymphoma) which also had the lowest levels of FAM210B expression (Fig. 6B and C). An analysis using the BioGPS database revealed that FAM210B expression levels in normal human blood cells are low compared to other normal tissues and that expression is lowest or absent in centrocytes, centroblasts, and naïve B-cells (Fig. 6D) (27). This suggests that low FAM210B levels found in DLBCL reflect the cell of origin and is not the result of a mutation or derangement in gene expression.

To determine whether FAM210B suppresses the effects of BTM-3528 and BTM-3566, we engineered BJAB cells to stably overexpress FAM210B-tGFP and compared them with wt and HRI^−/− cells. Bortezomib and FCCP were included as controls to assess the specificity of FAM210B on preserving cell viability when ATF4-ISR is induced via alternative methods of activation. Strikingly, FAM210B-tGFP expression completely suppressed the activity of BTM-3528 and BTM-3566 but had no effect on bortezomib or FCCP-induced cell death (Fig. 6E). Importantly, knock-out of HRI reduced the activity of both bortezomib and FCCP, confirming the common utilization of HRI in induction of the ATF4 ISR and subsequent apoptosis.

To determine the effects of FAM210B on OMA1 activation, we compared the ability of BTM-3528, 3566, and the inactive control BTM-3532 to activate OMA1 in the presence or absence of FAM210B-tGFP overexpression. As expected, L-OPA1 cleavage in WT HCT-116 cells was observed in the presence of BTM-3528 or BTM-3566 but not BTM-3532 (Fig. 6F). L-OPA1 cleavage did not occur in HCT-116 cells stably expressing FAM210B-tGFP (Fig. 6F), indicating that FAM210B acts upstream of OMA1 to suppress the effect of compounds on OPA1 cleavage. FCCP-induced cleavage of OPA1 was unaffected by FAM210B-tGFP. We also compared the effects of FAM210B expression on ATF4 protein expression using BTM-3566 and alternative inducers of the ATF4 ISR. FAM210B expression robustly suppressed ATF4 induction by BTM-3566 but not suppressed by amino acid starvation, tunicamycin, bortezomib, or ONC201 treatments (Supplementary Fig. S10), suggesting a unique mechanism. In addition, compound-dependent decreases in basal, ATP-linked, and maximal mitochondrial respiration were blocked in cells overexpressing FAM210B-tGFP (Supplementary Fig. S11A and B).

Tumors acquire the ability to overcome normal growth control mechanisms. Mitochondria play a crucial role in this process as central hubs controlling cell survival in response to proapoptotic stimuli that activate the intrinsic apoptotic pathway. The novel pharmacology described here presents an opportunity to impact a broad set of tumors, particularly hematologic malignancies irrespective of cell of origin or genomic background. Importantly, we have demonstrated that the therapeutic activity of BTM-compounds appears to span a diversity of genetic backgrounds in DLBCL and other B-cell lymphomas that differ in cell-of-origin, presence of MYC rearrangements and other genetic alterations.

Collectively, our data suggest a model in which the response to BTM compounds are controlled by the mitochondrial proteins OMA1, DELE1, and FAM210B (See Supplementary Fig. S12). OMA1 is a protease that, once activated, cleaves various mitochondrial proteins including OPA1 and DELE1. The genetic data supporting this model clearly demonstrates the requirement of OMA1 and DELE1 gene expression for BTM-3566 drug activity in BJAB cells. The functional genetic data supports a model in which cleavage of DELE1 by OMA1 lead to the downstream effector activity of the pathway. Yet the molecular mechanism linking OMA1-mediated proteolysis of DELE1 to outcome is not clear. The apparent temporal differences in reduction of OPA1-L and DELE1-L levels and induction of the ATF4 ISR complicate interpretations of a lineal temporal relationship between OMA1 mediated DELE1 cleavage and HRI activation. It is possible that other aspects of DELE1 synthesis and maturation may also be affected by drug activity. For instance, impeding import of DELE1-L into the mitochondria may lead to activation of HRI in the absence of OMA1 cleavage (25). Further work is necessary to fully comprehend the effects BTM compounds may have on DELE1 protein import into the mitochondria and any relationship with OMA1 proteolytic activity and therapeutic outcome.

Induction of apoptosis by BTM compounds in DLBCL is linked to activation of HRI and phosphorylation of eIF2α resulting in inhibition of global translation and increased ATF4-dependent transcription. The apoptotic program is Bax dependent and is correlated with activation of caspase activity and release of cytochrome c. The exact molecular context that makes DLBCL uniformly sensitive to induction of the ATF4 ISR and apoptosis by BTM-compounds are not fully understood. Persistent eIF2α phosphorylation can sensitize cells to apoptotic stimuli by reducing the levels of antiapoptotic proteins such as MCL1, BCL2 and XIAP (8–10, 28) as well as upregulating expression of proapoptotic genes such as TRB3 (29) and BIM (30). ATF4 target gene such as DDIT3 (CHOP) are also known to promote cell death through a variety of mechanisms in the context of the ER-stress response (31–33) It is notable that BTM compounds are uniformly effective in DLBCL lines that have differing sensitivities to venetoclax or variable expression of components of the intrinsic apoptotic pathway, including MCL1, BCL2, and BCL-XL (34–36). Additional studies are needed to identify factors distinguishing BTM-compound sensitive from resistant tumors and whether particular agents capable of inducing apoptosis (e.g., the BCL2 inhibitor venetoclax) are likely to synergize with BTM-compounds in specific contexts.

Cleavage of OPA1 and fragmentation of the mitochondrial network are prominent outcomes of BTM-compound activity. OPA1, a dynamin-like like protein involved with fusion of the inner mitochondrial membrane, is intimately involved in maintenance of cristae structure and mitochondrial morphology (19, 37, 38). Controlled OMA1-dependent proteolysis of L-OPA1 to S-OPA1 has been implicated as an early event associated with apoptosis (18, 39–42). Our findings suggest that BTM-compound activation of OMA1-dependent OPA1 cleavage is neither necessary nor sufficient to activate apoptosis. This is particularly true in solid tumor lines where cell growth arrest, but not apoptosis, is the outcome. This suggests that BTM-compound induced cleavage of OPA1 and fragmentation of the mitochondrial network are not universally capable of inducing cellular apoptosis and tumor regression, at least in the contexts tested here.

The precise mechanism by which BTM-compounds induce OMA1 activation remains unclear. The compounds had no effect on mitochondrial electron transport or membrane potential suggesting an alternative mechanism for activation of OMA1. That mitochondria are central to the pharmacologic effects of BTM-compounds is reinforced by the observation that there is an inverse correlation between cell line sensitivity and expression levels of the mitochondrial protein FAM210B. Low FAM210B levels are associated with a greater degree of compound activity and DLBCL lines do not express FAM210B to any appreciable degree. Overexpression of FAM210B in both BJAB and HCT-116 cells conferred resistance to BTM-compound induced activation of the ATF4 ISR while having no effect on other inducers of the ISR (tunicamycin, nutrient deprivation and bortezomib) or compounds that affect mitochondrial membrane potential (the uncoupler FCCP) or mitochondrial proteostasis (the CLPP1 activator ONC201). This data suggests that multiple pathways leading to changes in mitochondrial function exist that, when pharmacologically perturbed, lead to activation of the ATF4 ISR but only a FAM210B-regulatable process is affected by BTM-compounds.

The physiologic role of FAM210B appears to depend on cellular context. FAM210B is involved in high-capacity heme biosynthesis occurring during erythropoiesis, acting to increase the importation of iron into the mitochondria. FAM210B is not required for basal heme biosynthesis (43–45). FAM210B may have broader activity controlling cellular metabolism unrelated to heme biosynthesis but tied to its ability to act as a scaffold protein for other mitochondrial membrane or matrix proteins. Reduction in FAM210B expression using siRNA leads to increased mitochondrial respiratory capacity, decreased glycolysis and an aggressive metastatic tumor phenotype (46). FAM210B has also been described as a tumor suppressor, with lower FAM210B levels associated with poor prognosis in patients with renal, cervical, and lung cancers (46).

Multiple therapeutic strategies have been utilized to induce eIF2α phosphorylation and the ATF4 ISR as treatment for cancer. The 26S proteasome inhibitor bortezomib and the novel CLPP1 activator ONC201 rely on activation of HRI to phosphorylate eIF2α, with subsequent induction of ATF4 and CHOP (47–50). Our data provide further support for the induction of ATF4 ISR, in this case through pharmacologic activation of OMA1, as an orthogonal strategy for inducing cancer growth arrest and apoptosis.

A. Schwarzer reports grants and personal fees from Bantam Pharmaceutical during the conduct of the study. M. Kleppa reports grants from Bantam Pharmaceutical during the conduct of the study. S.D. Slattery reports other support from Bantam Pharmaceuticals during the conduct of the study. A. Cooper reports a patent for US10537558B2 issued, a patent for US20230076820A1 issued, a patent for WO2020243582 pending, and a patent for WO2020243584 pending. M. Hannink reports grants from Bantam Pharmaceutical during the conduct of the study. T. Hembrough reports personal fees, nonfinancial support, and other support from Bantam Pharmaceutical during the conduct of the study. J. Levine reports other support from Bantam Pharmaceutical outside the submitted work, as well as a patent for US10537558B2 issued and a patent for US20190345152a1 issued. M. Luther reports personal fees from Bantam Pharmaceutical during the conduct of the study, personal fees from Bantam Pharmaceutical outside the submitted work, as well as a patent for US10537558B2 issued, a patent for US20230076820A1 issued, and a patent for WO2019236966A3 pending; and Consultant and board member of Bantam Pharmaceutical. Michael Luther is an employee of Bantam Pharmaceutical and is an inventor on issued patents US10537558B2 and US20230076820A1 and patent application WO2019236966A3 associated with the information in this article. M. Stocum reports personal fees and other support from Bantam Pharmaceutical, LLC and other support from Personalized Medicine Partners, LLC outside the submitted work; and the work of Bantam Pharmaceutical, LLC has been funded by private investors with dual interests of bringing drugs to market that improve cancer patient care and profiting from investments in companies like Bantam who develop such products. L. Stiles reports other support from Bantam Pharmaceutical during the conduct of the study. D.M. Weinstock reports personal fees and other support from Bantam during the conduct of the study; grants from Daiichi Sankyo, grants and personal fees from Astra Zeneca, and personal fees from Verastem outside the submitted work; and Employee of Merck. M. Liesa reports grants from Bantam Pharmaceuticals during the conduct of the study; other support from Enspire Bio outside the submitted work. M. Kostura reports a patent for US010537558B2 issued, a patent for US20230076820A1 issued, and a patent for WO 2019/236966A3 pending. No disclosures were reported by the other authors.

A. Schwarzer: Conceptualization, investigation, visualization, methodology, writing–review and editing. M. Oliveira: Investigation, visualization, methodology. M.-J. Kleppa: Investigation, methodology. S.D. Slattery: Investigation, methodology. A. Anantha: Writing–review and editing. A. Cooper: Conceptualization, investigation, methodology. M. Hannink: Investigation, visualization, methodology, writing–review and editing. A. Schambach: Investigation, methodology. A. Dörrie: Investigation, methodology. A. Kotylarov: Investigation, methodology. M. Gaestel: Investigation, methodology. T. Hembrough: Writing–review and editing. J. Levine: Conceptualization, investigation. M. Luther: Project administration. M. Stocum: Project administration, writing–review and editing. L. Stiles: Investigation, methodology. D.M. Weinstock: Project administration, writing–review and editing. M. Liesa: Conceptualization, investigation, visualization, methodology, writing–review and editing. M.J. Kostura: Conceptualization, supervision, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.

The authors would like to thank Dr. Josh Rabinowitz of Princeton University for helpful scientific discussions, comments, review, and editing of this manuscript. Dr. Alexander van der Bliek, at the UCLA David Geffen School of Medicine UCLA for the HCT-116 OMA 1-/- cell line.

Bantam Pharmaceutical, Durham, N.C., U.S.

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.