Dynamic BH3 profiling identifies pro-apoptotic drug combinations for the treatment of malignant pleural mesothelioma


Measuring drug-induced apoptotic priming on fresh MPM patient samples

Targeted agents that evoke an early death signal measured by DBP have been shown to be efficacious in vivo20,21,22,23. We therefore hypothesized that by identifying drug combinations that prime MPM cells ex vivo, we could identify drug/s that would be efficacious in vivo. The schematic of our approach is shown in Fig. 1a. While HTDBP has the capacity to test thousands of drugs simultaneously, given the potentially limited patient tissue in this context, we focused on drugs (and their combinations) clinically relevant (currently in clinical trials or approved in the clinic) to thoracic malignancies in the interest of most efficient clinical translation (Supplementary Table 1).

Fig. 1: Clinically relevant oncology combination screen (CROCS) on primary MPM patient samples using HTDBP to identify hits.
figure 1

Fresh primary MPM patient samples were dissociated, treated with CROCS and HTDBP carried out. Cells were analyzed by immunofluorescence microscopy and Z-score was calculated to identify drug/drug combinations that prime tumor cells. a Schematic showing the workflow for measuring drug-induced priming using HTDBP on primary MPM patient samples. Created with BioRender.com. b Graphs show mean Z-score for each drug treatment (carried out in duplicate), for each primary MPM patient sample (MPS). Each individual dot represents a different drug treatment (single agent or drug-drug combination). A red dot represents a hit with a Z-score ≥ 3 with no replicate <1.5. Black dots are non-hits. c Graph showing the mean Z-score correlation between two tumor samples from the same patient (MPS:L and MPS:M) using one-tailed Spearman ranked test (p value = <0.0001). MPS:L is a tumor from the 7th rib and MPS:M is a pleural tumor. d Venn diagram showing the overlap between the CROCS HTDBP hits (red dots in part b) for MPS:L and MPS:M.

Prolonged ex vivo culture has been shown to change tumor cell characteristics such as drug sensitivity and gene expression22,24,25. A major advantage of DBP is that it requires primary cells to be cultured for less than 24 h, because it measures early changes in apoptotic signaling which occur before frank apoptosis20,21,22,23. We first set out to see if these standard tissue culture condition had any effect on cell viability. Over 24 h luminescence increased indicating increase in ATP production (Supplementary Fig. 1A).

To identify drug combinations that prime primary MPM cells ex vivo, cells are treated in our CROCS (clinically relevant oncology combination screen), and immunofluorescence microscopy based HTDBP is carried out to identify hits (Fig. 1). For HTDBP to be carried out on limited cells, we first had to determine the optimum BIM BH3 peptide concentration to use in the assay for each sample, chosen to be the EC10 for MOMP (10% cytochrome c release) in untreated cells. At this concentration, drug-induced priming can be sensitively captured. We received 13 freshly resected MPM patient tumors, which we dissociated to produce a single cell suspension and seeded on 384-well plate/s. The next day, a BIM BH3 peptide titration on untreated cells to calculate BIM EC10 was performed (Supplementary Fig. 1B), followed by HTDBP on CROCS-treated primary MPM cells.

Drug-potentiated peptide-induced loss of cytochrome c was quantified as the difference in percentage of cytochrome c positive cells between DMSO-treated and drug-treated wells; we call this delta priming %. Each treatment (single agent and combination) is present as an experimental duplicate. To test repeatability, we tested the correlation between experimental duplicates (delta priming % n1 Vs. delta priming % n2). All MPM patient samples showed significant correlation between delta priming % replicates (Supplementary Fig. 2A), even when the number of drugs used was low due to low number of cells (mesothelioma patient sample (MPS):D, E, F and G; Supplementary Table 2). Exact drugs used in CROCS for each patient sample is shown in Supplementary Table 3.

We labeled a drug or drug combination as a “hit” in our assay when the treatment scored a mean (of duplicates) Z-score ≥ 3, with neither replicate having a Z-score below 1.5 (red dot Fig. 1b and Supplementary Fig. 2A; gray dashed line marks Z-score = 1.5). All drug combinations and single agents for all MPM patient samples are shown in a heatmap in Supplementary Figure 3, blue is a hit and yellow is a non-hit. This reveals there are clearly some common hits between the patient samples. To assess if chemical vulnerabilities correlated with disease subtype in MPM patient samples, we compared epithelioid and biphasic top hits (only 1 sarcomatoid sample, so no comparisons were made). There was no difference in top hits between epithelioid and biphasic histologies which may be due to the biphasic containing epithelioid cells (Supplementary Table 4). All MPM patient samples had oncopanel profiling carried out. No tier 1 or tier 2 aberrations which are associated with clinical or potential clinical significance respectively were observed in any of the patient tumor samples (Supplementary Table 5). Therefore, we cannot conclude if chemical vulnerabilities are associated with clinically significant aberrations in the patient samples profiled here. All MPM patient clinical data is available in Supplementary Table 6.

We had the chance to study paired MPM samples from the same patient. One was from a tumor found adherent to the 7th rib (MPS:L) and the other was a tumor from the pleura at a different location. (MPS:M). This gave us an interesting opportunity to ask the question of whether the two distinct but local tumors from the same patient had similar chemical vulnerabilities. The CROCS HTDBP results from both tumors had strong correlation (r = 0.74, Fig. 1c and Supplementary Fig. 2B) and 89 of the hits overlapped (MPS:L had 104 hits and MPS:M had 95 hits; Fig. 1d), suggesting that these two distinct tumors had the same chemical vulnerabilities.

BH3 mimetics navitoclax and S63845 but not venetoclax, enhance apoptotic priming in combination with PI3K/AKT/mTOR pathway inhibitors in primary MPM cells

Representative immunofluorescence microscopy images in a MPM epithelioid (MPS:A), biphasic (MPS:C) and sarcomatoid (MPS:J) patient sample are shown in Fig. 2a and Supplementary Figs. 4–6. These images show the difference between top hits (highest drug-induced priming), non-hits (no drug-induced priming) and DMSO-control under the microscope.

Fig. 2: BCL-xL and MCL-1 antagonism enhance priming with PI3K/AKT/mTOR pathway inhibitors in MPM.
figure 2

a Primary epithelioid MPM cells were treated as previously described in Fig. 1, in technical duplicate. Representative immunofluorescence microscopy images from MPM patient sample A (MPS:A). Images taken at tenfold magnification. Hoechst 33342 used to stain DNA (blue) and identify the number of cells/well. Pan-cytokeratin-488 antibody (green) used to identify epithelioid cells (parent population). From the parent population, cytochrome c positive cells % was determined using cytochrome c-647 antibody (red). DMSO treatment is a negative control for cytochrome c loss. Non-hit is a drug treatment that didn’t score above the hit threshold (no drug-induced priming). Top hit is the drug treatment that scored the highest mean Z-score (highest drug-induced priming) for this patient sample, first red dot in Fig. 1b. Scale bar is 100 µm. b Schematic showing drug targets for BH3 mimetics, BH3 peptides and PI3K/AKT/mTOR pathway inhibitors used in this paper. All drugs (not BH3 peptides) shown here are included in CROCS list except A-1331852 (because it’s not in clinical trials yet). BH3 peptides are peptides derived from the BH3 domain of the pro-apoptotic BH3-only Bcl-2 family members and are used in BH3 profiling/DBP assay. c Graph showing mean Z-score for top ten most common hits and hits with highest mean across n = 13 MPM patient samples. d Graph showing mean Z-score for navitoclax, S63845 or venetoclax in combination with PI3K/AKT/mTOR pathway inhibitors (combinations and single agents), across n = 13 MPM patient samples.

Schematic showing BH3 mimetics and PI3K pathway inhibitors/targets, used in this manuscript are show in Fig. 2b. Two papers published recently from Arulananda et al. showed the importance of BCL-xL antagonism in the treatment of MPM and how combining with MCL-1 antagonist is highly efficacious in MPM cell lines26,27. The most common hits and hits that caused the highest amount of drug-induced priming across all the MPM patient samples are shown in Fig. 2c, Supplementary Tables 7 and 8. BH3 mimetics, specifically navitoclax (BCL-2, BCL-xL, and BCL-w antagonist) and S63845 (MCL-1 antagonist) in combination with one another, or a PI3K/AKT/mTOR pathway inhibitor, commonly primed primary MPM cells. However, the BH3 mimetic venetoclax (BCL-2 antagonist) did not, suggesting that antagonism of BCL-xL and MCL-1 is more important in primary MPM cells in combination with one another or with PI3/AKT/mTOR pathway inhibitors (Fig. 2d). The entire CROCS HTDBP results for each MPM patient is found in Supplementary Data file 1.

Malignant pleural mesothelioma PDX’s chemical vulnerabilities overlap with MPM patient samples

To validate HTDBP as an approach to identify efficacious hits we wanted to confirm a top hit that primes the primary MPM cells ex vivo, would also be efficacious in vivo in a MPM PDX model. Before we could start an in vivo efficacy study with a top hit, we wanted to confirm that MPM PDX models recapitulated the pattern of drug sensitivity seen in MPM patient samples. We created three MPM PDX models of biphasic (CPDM_0011x, CPDM_0106x) and sarcomatoid (CPDM_0184x) histological subclasses. While generating MPM PDX’s, two epithelioid models developed lymphoma, likely because of the Epstein-Barr virus positivity status of the patient28. Therefore, we confirmed each model was to be devoid of lymphoma (hCD45 negative) with lineage match (STR fingerprinting) of the patient tumors, and whole genome copy number profiling to confirm the presence of tumor cells. We then carried out the same approach as in the MPM patient samples in MPM PDX tumors (Supplementary Fig. 7A). Malignant pleural mesothelioma PDX tumors were implanted into the right and left flank of immunocompromised SCID-bg mice. Tumors took between 40 and 120 days to grow enough material for CROCS HTDBP (~1000 mm³; Fig. 3a). All MPM PDX tumors showed good correlation between replicates (Supplementary Figure 7B). Seventy-nine percent of the MPM PDX hits were hits in the MPM patient samples (Fig. 3b and Supplementary Data files 2, 3). Heatmap in Fig. 3c shows all the drug treatments across all MPM PDX tumors profiled and Fig. 3d shows individual Z-score graph for each tumor (red dot is a hit in the assay). The top hits (Fig. 3e, f and Supplementary Tables 9 and 10) were also hits in MPM patient samples, confirming that MPM PDX models demonstrate similar chemical vulnerabilities to primary MPM patient samples. The top two common hits in primary MPM patient (Fig. 2d) and MPM PDX (Fig. 3e) samples were (1) navitoclax plus S63845 and (2) navitoclax plus AZD8055 (mTORC1/2 inhibitor).

Fig. 3: Clinically relevant oncology combination screen (CROCS) on MPM PDX tumors using HTDBP to identify hits.
figure 3

Malignant pleural mesothelioma PDX tumors were dissociated, treated with CROCS and HTDBP carried out. Cells were analyzed as previously described in Fig. 1. a Malignant pleural mesothelioma PDX tumors (CPDM_0011x, CPDM_0106x and CPDM_0184x) were implanted subcutaneously in the right and left flank of immunocompromised SCID-bg mice. When the right (RF) and/or left flank (LF) tumor reached 1000 mm3 mice were sacrificed and tumors harvested for CROCS HTDBP. b Venn diagram show the overlap between the CROCS HTDBP hits for MPM patient and PDX samples. c Heatmap showing ranked (highest at the top) Z-score across MPM PDX tumors harvested in (a), for each drug treatment. Blue represents a hit with a Z-score ≥ 3 and yellow are non-hits. d Graphs show mean Z-score for each drug treatment, for individual MPM PDX tumor harvested in (a). One tumor for CPDM_0106x and CPDM_0184x models and four tumors for CPDM_0011x model. Each individual dot represents a different drug treatment (single agent or drug-drug combination). A red dot represents a hit with a Z-score ≥ 3 with no replicate <1.5. Black dots are non-hits. e Graph showing mean Z-score for top ten most common hits and f hits with highest mean across n = 6 MPM PDX tumors.

We did a tolerance study in non-tumor bearing immunocompromised SCID-bg mice, with navitoclax plus S63845 or navitoclax plus AZD8055, at doses well tolerated as single agents. Unfortunately, all the mice in the navitoclax plus S63845 combination arm died within 4 h of treatment. However, navitoclax plus AZD8055 was well tolerated in vivo (Supplementary Fig. 8A, B). We therefore decided to pursue navitoclax plus AZD8055 to see if this was an efficacious combination in vivo in the MPM PDX model, CPDM_0011x because we had the most data generated for this model (Fig. 3).

Navitoclax plus AZD8055 is efficacious in vivo in MPM CPDM_0011x PDX model

Mice bearing MPM PDX CPDM_0011x tumors were randomized into 5 arms once tumors reached between 150 and 250 mm3: Arm (1) Vehicle, (2) navitoclax-only, (3) AZD8055-only, (4) navitoclax plus AZD8055 and (5) venetoclax-only. Venetoclax was used as a negative control as it did not cause a significant amount of drug-induced priming in MPM patient and PDX tumors. Both navitoclax and venetoclax were dosed at 100 mg/kg qd for 21 days and AZD8055 was dosed at 16 mg/kg qd for 21 days.

Consistent with HTDBP results, navitoclax plus AZD8055 combination significantly reduced tumor volume compared to all other arms (Fig. 4a; navitoclax plus AZD8055 tumor volume Vs. all other arms = p < 0.0001). Mouse survival was based on time taken to reach the predefined endpoint 5 times initial tumor volume (5xITV). Navitoclax plus AZD8055 increased mouse survival by 30 days from day 19 (vehicle) to day 49 and survival was significantly longer compared to every other arm of the study (Fig. 4b; p < 0.0001). All mice in the vehicle, navitoclax-only, venetoclax-only and 4/7 mice in AZD8055-only, reached the predefined tumor volume endpoint of the study ≤ day 21, indicating the aggressive nature of CPDM_0011x PDX model (Fig. 4c). Relative tumor burden for navitoclax plus AZD8055 combination arm remained significantly low while mice were dosed, compared to all other arms (Fig. 4d). These data determine navitoclax plus AZD8055 combination is efficacious in vivo in CPDM_0011x PDX model, validating HTDBP as an approach to identify efficacious drug combinations ex vivo in MPM.

Fig. 4: In vivo efficacy study validates HTDBP as an approach to identify efficacious hits in MPM.
figure 4

Malignant pleural mesothelioma PDX model CPDM_0011x, was implanted subcutaneously in the right flank of immunocompromised mice and when tumors reached 150–250 mm3, randomized into 5 treatment groups: vehicle (red, n = 8 mice), 100 mg/kg/qd navitoclax (blue, n = 7 mice), 16 mg/kg/qd AZD8055 (green, n = 7 mice), 100 mg/kg/qd navitoclax plus 16 mg/kg/qd AZD8055 (purple, n = 9 mice), or 100 mg/kg/qd venetoclax (black, n = 7 mice). Mice dosed for 21 days and sacrificed when tumors reached the endpoint of 5 times the initial tumor volume (5xITV) (ad). Mice bearing CPDM_0011x was administered with one dose of indicated drug treatment and 24 h later tumors were harvested and dissociated and used for western blotting analysis (e) or iBH3 profiling to measure mitochondrial priming (f). a Graph showing mean tumor growth over 21 days of dosing. Error bars represent ± standard deviation for 7–9 mice per treatment group. P value is <0.0001 for navitoclax plus AZD8055 combination treatment compared to all other treatment groups using a 2-way ANOVA multiple comparisons test. b Kaplan–Meier survival curve, mice were sacrificed at 5xITV endpoint. Navitoclax plus AZD8055 combination treatment is statistically significant compared to every other treatment group (p value is <0.0001) based on a Log-rank (Mantle–Cox) test. c Graph showing the individual tumor volume for each mouse for each treatment group over time. Orange line indicates the dosing period (21 days). d Bar graph showing relative tumor burden calculated for each treatment group relative to the mean tumor volume of the vehicle-control group on day 7 and day 14 respectively. Data represents the mean and shows corresponding data points with error bars ± standard deviation. Significance calculated using 2-way ANOVA multiple comparisons. P values stated on the graph. e Representative immunoblots (n = 3) of cell lysates of CPDM_0011x tumor cells after 24-h treatment in vivo for MCL-1, S6, phosphoSer235/236-S6 (pS6), PARP, cleaved PARP (Cl.PARP), cleaved caspase 3 (CC3) and β-Actin (loading control). f Intracellular BH3 profiling was performed on CPDM_0011x tumor cells after 24 h treatment in vivo. Cells analyzed by flow cytometry for pan-cytokeratin/vimentin positive cells (parent population) and cytochrome c positive cells %. Graphs represent the mean of three mice ± standard deviation. Table represent BIM area under curve (AUC) ± 95% confidence intervals (CI) and significance is determined according to one-tailed unpaired t test comparing vehicle control to treatment. *P < 0.05 was considered statistically significant.

AZD8055 reduces MCL-1 protein levels in MPM PDX cells

To investigate the in vivo mechanism of action of the navitoclax plus AZD8055 combination in MPM, CPDM_0011x PDX tumor bearing mice were treated with one dose of either: (1) vehicle, (2) navitoclax-only, (3) AZD8055-only or (4) navitoclax plus AZD8055. Twenty-four hours later tumors were harvested, dissociated, and used for Western blotting analysis. PARP cleavage and caspase 3 cleavage were observed, consistent with an in vivo apoptotic mechanism of tumor cell death. MCL-1 is a known resistance biomarker for navitoclax treatment29 and AZD8055 has been previously shown to reduce MCL-1 protein levels30. AZD8055-only decreased MCL-1 protein levels compared to untreated in the CPDM_0011x PDX cells but had no effect on the apoptotic biomarkers cleaved PARP and cleaved caspase 3 suggesting a reduction in MCL-1 levels alone is not sufficient to induce apoptosis in this model (Fig. 4e). Note that the navitoclax plus AZD8055 combination arm demonstrated preserved MCL-1 levels at the 24-h time point. This is possibly because tumor cells that most downregulated MCL-1 via inhibition of mTOR pathway (AZD8055) in combination with navitoclax have undergone apoptosis (high levels of cleaved PARP and cleaved caspase 3, Fig. 4e) resulting in removal of dead cells from the tumor by phagocytosis. Phospho-S6 is used as a pharmacodynamic biomarker for AZD8055 activity and levels are reduced in arms treated with AZD8055 (Fig. 4e).

Drug-induced mitochondrial priming is recapitulated in vivo

Our identification of the navitoclax plus AZD8055 combination was predicated on their ability to induce apoptotic priming ex vivo. To investigate whether drug-induced priming also occurred in vivo, we treated mice bearing CPDM_0011x PDX tumors with either vehicle, navitoclax-only, AZD8055-only, or navitoclax plus AZD8055. Twenty-four hours later we carried out flow cytometry based iBH3 profiling31 on dissociated tumors. Cells were permeabilized and incubated with a range of BIM BH3 peptide concentrations for 1 h, then fixed and stained for cytochrome c. Cytochrome c release was plotted against BIM dose response and BIM AUC (area under curve) and BIM EC50 was calculated to determine relative drug-induced priming. The higher the AUC and lower the BIM EC50, the more primed the tumor cells are because they released more cytochrome c in response to BIM BH3 peptide. All drug treatments significantly primed CPDM_0011x cells in vivo compared to vehicle control (Fig. 4f and Supplementary Table 11). Notable, navitoclax-only and in combination with AZD8055 caused 50% and 44% cytochrome c release, respectively, even without BIM BH3 peptide added (UnT x-axis) in CPDM_0011x PDX cells (Fig. 4f), This suggests that at this the 24-h time point, the effect of the treatment has gone beyond priming and apoptosis had been activated, consistent with immunoblot results in Fig. 4e. Note that single agent navitoclax was nonetheless lacking in efficacy in vivo, likely due to the heterogenous response of CPDM_0011x tumors to navitoclax as a single agent (primed 2/4 of CPDM_0011x tumors ex vivo), highlighting the need for combinations to overcome resistance to single agent navitoclax.

BCL-xL antagonism drives MCL-1 dependency and AZD8055 drives mitochondrial sensitivity to the BAD BH3 peptide in MPM cell lines

To gain more detailed insight into the molecular mechanism behind the efficacy of navitoclax plus AZD8055, we investigated anti-apoptotic dependencies after treatment with navitoclax or AZD8055 at 24-h, in a panel of MPM cell lines (H2052, JMN, JMN1B and MSTO-211H). We hypothesized that navitoclax treatment would increase MPM cells dependency to MCL-1 when BCL-xL is antagonized, because BCL-xL and MCL-1 are the dominant anti-apoptotic members and compensate for one another in MPM26,27. To validate changes are due to BCL-xL and not BCL-2 antagonism, we also treated cells with single agent A-1331852 (selective BCL-xL antagonist) or venetoclax (selective BCL-2 antagonist). We used DBP to identify anti-apoptotic dependencies at baseline and changes after 24-h drug treatment. Cells were permeabilized and incubated with a panel of BH3 peptides, including sensitizer BH3 peptides which show specificity in their binding to anti-apoptotic family members and therefore reveal anti-apoptotic dependencies (Fig. 2b). None of the MPM cell lines showed individual anti-apoptotic dependencies at baseline (red bar DMSO-control; Fig. 5a). However, navitoclax and A-1331852 showed significant increased response to MS1 BH3 peptide (MCL-1 antagonist32) indicating an increased dependency to MCL-1. Venetoclax treatment had no effect on MS1 BH3 peptide (Supplementary Fig. 9), suggesting that BCL-2 antagonism doesn’t affect MCL-1 dependency in MPM. Treatment with S63845 increased anti-apoptotic dependency to BCL-xL (HRK BH3 peptide; Supplementary Figure 9) suggesting that MCL-1 antagonism drives BCL-xL dependency.

Fig. 5: Understanding the molecular mechanism of navitoclax plus AZD8055 combination in vitro in MPM.
figure 5

Analysis of drug-induced priming and anti-apoptotic dependencies after treatment with navitoclax, AZD8055 and a BCL-xL specific antagonist, A-1331852 in MPM cell lines (H2052, JMN, JMN1B and MSTO-211H). Overall priming is measured by BIM or PUMA, whereas HRK, MS1, FSI and venetoclax are specific for BCL-xL, MCL-1, BFL-1, and BCL-2 dependency respectively (a, b). Drug treatment with navitoclax, AZD8055 and navitoclax plus AZD8055 combination in MPM cell lines (ce). a, b Cells treated with 1 µM navitoclax, A-1331852 or AZD8055 and then DBP carried out (n = 3). Cytochrome c positive cells % (cytochrome c released = 100 − cytochrome c positive cells %) was measured using immunofluorescence microscopy, on permeabilized cells after 1 h incubation with indicated BH3 peptide concentration. Data are presented as bar graphs as mean values with error bars (±standard deviation). We calculated significance using a two-way ANOVA multiple comparisons test to DMSO-control (n = 3). c Cells treated with indicated concentration of drug/s for 3 days. Fourteen days from initial drug treatment cells were fixed and stained with crystal violet and area confluency and growth rate was calculated. Data presented as violin plots with median as solid line and quartiles as dashed line. We calculated significance using ANOVA multiple comparisons test to DMSO-control (n = 3). ac P values are shown on the graph. d Representative Immunoblots (n = 3) of MPM cell line lysates after 24-h treatment with indicated drug concentrations for, phosphoSer473-AKT (pAKT), AKT, MCL-1, BCL-xL, BCL-2, BAK, BAX, BIM, PUMA, and β-Actin (loading control). e 24 h after treatment with indicated drug concentration, BCL-xL and MCL-1 were immunoprecipitated in H2052 cells (n = 2) and BIM complexes were determined by western blotting analysis (Input total cell lysate; IP, immunoprecipitated fraction; IgG1 (immunoglobulin isotype 1; MCL-1 isotype) and IgG3 (immunoglobulin isotype 3; BCL-xL isotype) control; S, supernatant).

Navitoclax mimics the BAD BH3 peptide in that it antagonizes BCL-2, BCL-xL, and BCL-W. We hypothesized that efficacy of the combination might result from a sensitization of mitochondria to the BAD BH3 peptide by AZD8055. Dynamic BH3 profiling revealed that AZD8055 treatment (24-h) significantly increased the response of all the MPM cell lines to the BAD BH3 peptide (Fig. 5b). These experiments highlight the importance of BCL-xL and MCL-1 in MPM for survival, antagonizing one, directly or indirectly, results in increased dependence on the other.

Navitoclax plus AZD8055 is efficacious in vitro recapitulating efficacy in vivo in MPM

Having observed mechanisms that might explain the effectiveness of the combination in cell lines, we next confirmed that we could recapitulate the in vivo efficacy of navitoclax plus AZD8055, in vitro in these MPM cell lines. After optimizing dosing and timing (Supplementary Fig. 10 and Supplementary Table 12), MPM cell lines were treated for 72 h with either DMSO-control, 1 µM navitoclax, 30 nM AZD8055 or 1 µM navitoclax plus 30 nM AZD8055 combination. Compounds were washed off 3 days later, and colonies were left to grow until day 14. Then colonies were simultaneously fixed and stained, and area confluency and growth rate were calculated (Fig. 5c). All MPM cell lines significantly reduced area confluency and growth rate in the combination compared to either drug as a single agent and DMSO-control. These data confirm that navitoclax plus AZD8055 is efficacious in vitro in the panel of MPM cell lines used to investigate this mechanism.

Bcl-2 family members levels after treatment with AZD8055 and/or navitoclax in mesothelioma cell lines

There is precedent for the PI3K/AKT/mTOR pathway affecting several Bcl-2 family members including BAD, BIM, BCL-2, MCL-1 and BAX33. Previously we showed that MCL-1 protein levels were reduced in MPM PDX cells in vivo after AZD8055 treatment (Fig. 4e). The effect of navitoclax and/or AZD8055 treatment on a panel of Bcl-2 family members was assessed by Western blotting analysis (protein) and qPCR (mRNA) in MPM cell lines. This data showed some changes common across many of the MPM cell lines, but there was heterogeneity in protein changes (Fig. 5d and Supplementary Fig. 11). We used phospho-AKT as a pharmacodynamic biomarker for mTOR pathway activity. MCL-1 protein levels significantly decreased after treatment with 1 µM AZD8055 in JMN, JMN1B and MSTO-211H cells consistent with what we observed in vivo. BIM levels were significantly increased after all treatment group compared to DMSO-control in H2052, JMN and MST-211H, and significantly increased in the combination group for JMN1B cells. The observed increase in BIM protein levels, offer a molecular explanation for the increase in drug-induced priming after treatment with navitoclax or AZD8055 measured by DBP in MPM cell lines (Fig. 5a, b, 0.3 µM BIM BH3 peptide). Overall, there were not any significant, common changes in relative mRNA levels across the MPM cell line panel after treatment, except for BCL-2 mRNA levels which increased significantly in JMN, JMN1B and MSTO-211H cells for some or all the treatment groups. This translated to a significant increase in protein levels in JMN and MSTO-211H cells but not JMN1B. H2052 cells did show a significant increase in BCL-2 protein levels even though no increase was observed at the mRNA level (Fig. 5d and Supplementary Fig. 11).

Bcl-xL antagonism sensitizes to the MCL-1 inhibitor triptolide in MPM cell lines

Several other drugs have been shown to downregulate MCL-1. Wei et al. identified the natural compound, triptolide, as a repressor of MCL-1 expression34 and Busacca et al. showed in MSTO-211H that ganetespib (HSP90 inhibitor) reduces MCL-1 levels35. Therefore, we wanted to assess the effect of BCL-xL antagonism with navitoclax or A-1331852 on these MCL-1 repressing compounds. We tested both compounds in H2052 and MSTO-211H cells and confirmed that triptolide nearly completely reduced MCL-1 levels at 24 h, however, ganetespib did not at the concentrations we used (Supplementary Fig. 12A, B). The discrepancy with ganetespib in MSTO-211H cells might be due to the time points used, 24 h used here Vs. 48 h used in Busacca et al.35. Based on these data we hypothesized that triptolide would significantly sensitize MPM cell lines to navitoclax or A-1331852 since MCL-1 were reduced so dramatically, while ganetespib would not. BCL-xL antagonism significantly sensitized to triptolide in all 4 MPM cell lines and ganetespib did not, except for H2052 cells (Supplementary Fig. 12C, D). Confirming that BCL-xL antagonism sensitizes to a variety of compounds that antagonize and/or reduce MCL-1 (S63845, AZD8055 and triptolide).

Navitoclax treatment increased abundance of complexes of MCL-1 and BIM and decreases abundance of complexes of BCL-XL and BIM

As previously described, BCL-xL and MCL-1 are the dominant anti-apoptotic proteins maintaining survival in MPM. We hypothesized that BIM bound to BCL-xL would be removed by navitoclax treatment and MCL-1 would bind and neutralize this newly available BIM. To directly compare the changes in BIM complexes with BCL-xL or MCL-1 after navitoclax treatment (single agent and combination), we performed immunoprecipitation of BCL-xL and MCL-1 and assessed BIM protein levels in H2052 and JMN1B cell lines. BIM was found in complexes with BCL-xL and MCL-1 at baseline but when cells were treated with navitoclax either as a single agent or in the combination there was a relative reduction in BIM in complexes with BCL-xL and an increase in BIM in complexes with MCL-1, consistent with our hypothesis (Fig. 5e and Supplementary Fig. 13A, B).



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *