TCGA data shows differential roles for BET isoforms in glioma based on IDH status
We first delved into TCGA data via the UCSC Xena browser using the low-grade glioma and glioblastoma (GBMLGG) cohort dataset. GBMLGG data was segregated based on IDH1/2 status, null data was excluded, and overall survival was plotted based on median BET isoform (BRD2/3/4) expression levels. We found that enhanced BRD4 expression was associated with a statistically significant decrease in overall survival only in IDHmut patients, whereas this decrease in survival was not seen in IDHwt patients based on BRD4 expression (Fig. 1A, B). Interestingly in contrast to the BRD4 TCGA data, we found that BRD3 expression only influenced IDHwt patient survival, with BRD3 expression associated with a statistically significant increase in overall survival (Fig. 1C, D). BRD2 expression levels did not correlate with a significant increase or decrease in overall survival in either IDHwt or IDHmut glioma (Fig. 1E, F). Overall, these data suggest that increased BRD4 activity results in enhanced malignancy only in IDHmut glioma, whereas increased BRD3 activity results in enhanced survival only in IDHwt glioma. Further, this suggests that IDHmut glioma would be particularly sensitive to BET inhibition and supports investigation of BET inhibitors as a therapeutic strategy for the treatment of IDHmut glioma.
Bromodomain inhibition by JQ1 elicits delayed cytotoxicity only in IDHmut glioma
Next, we performed Muse cell viability assays with the BET inhibitor JQ1 on our IDHwt (0827, 0923) and IDHmut (0905, BT142) cell lines in vitro. To note is that all four cell lines were cultured in defined stem cell media and maintained as glioma stem-like cells. In JQ1 dose-response assays conducted with a 10 day endpoint, our IDHmut cells exhibited a much more pronounced cytotoxic response to JQ1, with IDHmut cells displaying a statistically significant reduction in cell viability relative to IDHwt cells. Reduction in cell viability ranged from 54–72% in IDHmut glioma cells whereas the IDHwt counterpart only had a reduction in cell viability by 10–31% (Fig. 2A; Fig. S1A). Time course assays utilizing 250 nM JQ1 at timepoints of 3, 7, and 10 days showed that a statistically significant cytotoxic response is delayed to 10 days incubation with JQ1 (Fig. 2B; Fig. S1A-C). Further analysis via western blot supported an apoptotic mechanism of cell death in IDHmut cells (Fig. 2C). Phase-contrast microscopy images showed our cells as large and numerous neurospheres in our vehicle treatments, but at 250 nM JQ1 over 10 days we saw preferential reduction in sphere size, sphere number, and an increase in phase-dark single cells in our IDHmut (0905, BT142), but not in our IDHwt (0827, 0923) cultures (Fig. 2D). Overall, these data indicate that IDHmut cells display a profound delayed cytotoxic/apoptotic response that may not be observed in short-term cytotoxicity/apoptosis assays.
JQ1 early effects are primarily antiproliferative with a modest sensitivity associated with IDH status
Next, we focused on investigating the early effects of JQ1-mediated BET inhibition. We observed that at 3 days incubation with JQ1 there was very little reduction in cell viability irrespective of IDH status (Fig. 3A; Fig. S1C). We then performed BrdU incorporation assays to investigate the antiproliferative effects mediated by JQ1 at this early time point with our IDHwt/mut glioma cells. Notably, we did not see a significant antiproliferative response in our IDHmut cells compared to their wild-type counterpart treated with 250 nM JQ1 for 3 days, with a 50% reduction in BrdU + cells for our IDHwt glioma cultures (0827, 0923) compared to approximately 63% reduction in BrdU + cells for our IDHmut glioma cultures (0905, BT142, TB096) (Fig. 3B, C; Fig. S2A, B). At 2000 nM JQ1 for 3 days, IDH status was associated with a statistically significant, yet modest, decrease in cellular proliferation with a 64% and 80% reduction in BrdU + cells compared to controls for IDHwt and IDHmut glioma cells, respectively (Fig. 3B, C; Fig. S2A, B).
Molecular analysis of parallel samples via western blot showed decreased activation of p-ERK at 3 days incubation with 250 nM JQ1, though these effects were not seen in 0827 (Fig. 3D). Additionally, total PDGFRα levels were downregulated in 0827 and 0923, but no effect on activated p-PDGFRα was seen at this timepoint (Fig. 3D). Effects on c-MYC were not associated with IDH status, with downregulation only seen in 0827 and BT142 GSCs (Fig. 3D).
Exogenous R132H IDH1 is sufficient to recapitulate delayed cytotoxicity in response to JQ1
To investigate whether R132H IDH1 is sufficient to sensitize an IDHwt cell line to BET inhibition, we inserted a dox-inducible R132H IDH1 transgene into our 0923 GBM cell line. Expression of the R132H IDH1 allele was validated via western blot and elevated 2-HG levels were validated via an enzymatic microplate assay (Fig. 4A, B). We performed a 10-day dose-response cell viability assay with our dox-inducible 0923-R132H-IDH1 cell line +/- doxycycline and discovered that induction of R132H IDH1 expression was sufficient to sensitize 0923-R132H-IDH1 to the cytotoxic effects of JQ1. In our dose response assay, JQ1 reduced cell viability by 50% for the 0923-R132H-IDH1 + dox treatment group compared to vehicle controls, whereas cell viability was reduced by only 20% without doxycycline treatment (Fig. 4C; Fig. S3). Similar to our endogenous IDHmut cell lines, the cytotoxic response to JQ1 was delayed past 3 days treatment for 923-R132H-IDH1 + dox, but maximal cytotoxicity appeared to occur earlier at 7 days of treatment (Fig. 4D; Fig. S3). In Annexin V apoptosis assays, R132H IDH1 transgene expression significantly enhanced the apoptotic response to JQ1 from 31–56% at 250 nM and from 45–72% at 1000 nM (Fig. 4E; Fig. S4). Doxycycline treatment itself did not elicit any cytotoxic effects at the utilized dose of 0.2 ng/mL. Overall, these data indicate that exogenous R132H IDH1 is sufficient to potentiate the cytotoxic and apoptotic responses mediated by JQ1 in an IDHwt GBM background.
Other molecular alterations at late time point
We further delved into the molecular responses of IDHwt/mut glioma in response to JQ1 at our late timepoint, 10 days. In contrast to the early 3 day time point, at 10 days JQ1 treatment all cell lines exhibited substantial downregulation of total PDGFRα, and this downregulation was also accompanied by decreased levels of active p-PDGFRα (Fig. 5A). This is especially interesting when considering that only our endogenous IDHmut cells recommend PDGF supplementation in the cell culture media. ERK was also downregulated in all cell lines analyzed, though intriguingly with our IDHmut glioma cells we only see decreased activation of ERK1, but not ERK2 (Fig. 5A). Additionally, we had a more consistent downregulation of c-MYC irrespective of IDH status at 10 days JQ1 treatment, though interestingly expression of c-MYC in BT142 was not observed at this time point (Fig. 5A). Overall, these data support further investigation into the differential role that PDGFRα activation and c-MYC downregulation play in the cytotoxic/antiproliferative response to JQ1.
Additional molecular analysis also showed that certain transcriptionally active epigenetic marks, such as H3K27Ac, H3K14Ac, and H3S10Ph, are downregulated in our IDHmut cells but in their IDH counterparts are conversely upregulated (Fig. 5B). This is interesting as it highlights unique epigenetic responses to BET inhibition in IDHmut glioma, which may owe to the unique epigenetic landscapes in IDHwt vs IDHmut glioma.
Lastly, we observe downregulation of mTORC1/2-related proteins including RAPTOR and RICTOR, with associated decreases in mTOR and S6K phosphorylation in response to 250 nM JQ1 for 10 days (Fig. 5C). To note is that this molecular response is not restricted to IDHmut cell lines. Considering that JQ1-mediated effects on mTORC function have not been reported in the literature, future research in this area could yield a more comprehensive characterization of the molecular effects mediated by BET inhibition.