Breast cancer cell lines with differing BRCA1/2 status demonstrate differential sensitivity to PRMT5 Inhibition
We treated a panel of seven breast cancer cell lines with different BRCA1/2 status (wildtype or mutant) (Table 1) with the PRMT5 inhibitor, GSK3326595, at doses ranging from 0 to 50 µM for five (MDAMB231 and SUM149PT) or seven (HCC1806, MDAMB468, MDAMB436, MX-1, HCC1428) days and measured cell viability using an MTT assay. We observed that there was a difference in sensitivity to PRMT5 inhibition (Fig. 1A), with the BRCA1/2 wildtype (WT) cell lines HCC1806, MDAMB468, and MDAMB231 being the most sensitive with IC values less than 2 µM. MDAMB436 (BRCA1 mutant) and MX-1 (BRCA1/2 mutant) lines were less sensitive to PRMT5 inhibition with IC values between 12.6 µM and 16.2 µM, respectively, while SUM149PT (BRCA1 mutant) and HCC1428 (BRCA2 mutant) were resistant to PRMT5 inhibition with IC values greater than 50 µM. This difference in sensitivity to PRMT5 inhibition was not related to the protein expression level of PRMT5 in these cell lines (Supplemental Fig. 1). Next, we sought to test the efficacy of GSK3326595 on PRMT5 enzyme inhibition in our panel of seven cell lines treated with a low concentration (40 nM) and a high concentration (5 µM) of GSK3326595 for five (MDAMB231 and SUM149PT) or seven (HCC1806, MDAMB468, MDAMB436, MX-1, HCC1428) days. We measured the global protein levels of symmetric dimethylarginine (SDMA) in whole cell protein lysates using western blotting (Fig. 1B) and observed that SDMA levels were reduced in all the cell lines starting at the low concentration of 40 nM. We calculated fold-changes for each cell line by dividing their IC values by 40 nM and noted a difference of 21-fold between the dose required to inhibit PRMT5 enzyme activity and the dose required to induce cytotoxicity for the most sensitive BRCA1/2 WT cell line (HCC1806), while a difference of 315-1250-fold was noted for the BRCA1/2 mutant lines.
Overall, this indicated that though the cell lines showed differential sensitivity to PRMT5 inhibition, the most sensitive cell lines were BRCA1/2 WT.
While PARP inhibitors have been approved for treatment of BRCA1/2 mutant breast cancer patients, the majority of breast cancer patients are BRCA1 WT, thus often lacking targeted therapies. Since the most sensitive cell lines to PRMT5 inhibition in our panel were BRCA1 WT, we sought to evaluate whether combining the PRMT5 inhibitor (GSK3326595) with a potent PARP inhibitor (Talazoparib) would be cytotoxic. First, we determined the IC values of Talazoparib in our panel of cell lines by treating the cells with a range of 0-50 µM for five (MDAMB231 and SUM149PT) or seven (HCC1806, MDAMB468, MDAMB436, MX-1, HCC1428) days and measured cell viability using an MTT assay. As expected, the BRCA1 mutant cell lines, MDAMB436 and SUM149PT were the most sensitive to PARP inhibition with IC of 0.39 and 4.49 nM, respectively (Table 1). BRCA1/2 WT (HCC1806, MDAMB468 and MDAMB231) and BRCA1/2 mutant (MX-1) cell lines were less sensitive to PARP inhibition (IC between 11 and 34 nM) while the BRCA2 mutant cell line (HCC1428) was resistant to PARP inhibition with an IC greater than 50 µM (Table 1).
Next, to evaluate the efficacy of combining GSK3326595 and Talazoparib, we treated our panel of seven cell lines with a maximum dose of approximately twice the IC dose for each drug and serially diluted it two-fold (MDAMB468, HCC1806, MDAMB436, SUM149PT, HCC1428) or four-fold (MDAMB231, MX-1). At the end of the five (MDAMB231, SUM149PT) or seven (MDAMB468, HCC1806, MDAMB436, HCC1428, MX-1) day treatment, cell viability was measured using an MTT assay, and synergy was calculated using Combenefit Software. We observed significant synergy between GSK3326595 and Talazoparib in all the cell lines tested (Fig. 2A). Interestingly, when we used the SYN_MAX metric from the Combenefit Software to compare the strength of maximum synergy between the cell lines, we observed that the highest synergy scores were obtained for the BRCA1/2 WT cell lines, MDAMB468 and MDAMB231, while the lowest synergy scores were obtained for the BRCA1 mutant SUM149PT and MDAMB436 cell lines (Fig. 2B). Since the BRCA1 WT cell lines were also the most sensitive to PRMT5 inhibition alone, we wanted to determine if that could be an indicator for synergy with PARP inhibitors. To determine this, we performed a Pearson correlation between the maximum synergy score for each cell line and its respective IC values for GSK3326595 (Fig. 2C). The Pearson correlation coefficient was negative (r = -0.5666) but was not statistically significant (p-value = 0.18), suggesting that there is a trend indicating that higher PRMT5 sensitivity correlates with obtaining a higher synergy with PARP inhibitors.
To understand mechanistically whether PRMT5 inhibition was altering the expression of genes involved in DNA damage response, we examined the gene expression of several known genes involved in homologous recombination (HR) and non-homologous end-joining (NHEJ) pathways upon PRMT5 inhibition in the two most sensitive BRCA1/2 WT cell lines, HCC1806 and MDAMB468. ATM, BRCA1, and BRCA2 mRNA were significantly downregulated in MDAMB468 after 24 h of PRMT5 inhibition, but these genes were either upregulated (BRCA1, BRCA2) or showed no change (ATM) in HCC1806 (Supplementary Fig. 2A). XRCC6, an NHEJ gene, was also downregulated in both cell lines but only at the higher PRMT5 inhibitor dose (5 µM) (Supplementary Fig. 2A). The only HR gene that was consistently upregulated in a dose-dependent manner and both cell lines after PRMT5 inhibition was RAD50 (Supplementary Fig. 2A). However, we did not observe a statistically significant increase in RAD50 protein level (Supplementary Fig. 2B, C).
These results suggested that the higher synergies observed between PRMT5i and PARPi in BRCA1/2 WT cells were not mediated through a consistent dysregulation of any of the tested genes in the HR and NHEJ pathways.
In our panel of breast cancer cell lines, we observed that there was a difference in sensitivity to PRMT5 inhibition based on BRCA1/2 status (Fig. 1A; Table 1). To determine whether BRCA1 expression was solely responsible for this difference in sensitivity, we tested a pair of isogenic ovarian cancer cell lines, UWB1.289 (BRCA1 mutant) and UWB1.289 + BR1 (BRCA1 complemented). Upon treating these cells with a range of 0-50 µM GSK3326595 for seven days, we observed that the parental cell line, UWB1.289, was more sensitive to PRMT5 inhibition with an IC of 2.38 µM compared to the BRCA1 complemented cell line, whose IC was 18.53 µM (Fig. 3A). Additionally, we treated both cell lines with two different doses of GSK3326595 (40 nM and 5 µM) and observed that the SDMA mark, indicative of PRMT5 enzyme activity, was decreased at the low concentration of 40nM in both cell lines (Fig. 3B), similar to the panel of breast cancer cell lines.
Next, we evaluated the effect of combining PRMT5i with PARPi in this pair of isogenic cell lines to determine whether BRCA1 expression was contributing to the synergy between these two inhibitors. We treated both cell lines with a maximum drug dose of approximately twice the IC dose for each drug and serially diluted them two-fold. At the end of a seven-day treatment, we measured cell viability using an MTT assay and calculated drug synergy using the Combenefit Software. Combining PRMT5i and PARPi in these cell lines demonstrated synergy (Fig. 3C) in both cell lines, similar to our observation in the panel of breast cancer cell lines (Fig. 2A). However, when we used the SYN_MAX metric to compare the strength of synergy between the UWB1.289 and UWB1.289 + BRCA1 cell line, we observed that UWB1.289 had a higher maximum synergy score compared to UWB1.289 + BRCA1 (Fig. 3D). This is in contrast to our panel of breast cancer cell lines, where the highest synergies were observed in BRCA1 WT cell lines (Fig. 2A).
In order to determine whether sequential treatment of PRMT5i and PARPi would be more efficacious than combination treatment, we pre-treated the cells with 4 µM of GSK3326595 (equivalent to 2 X IC dose in UWB1.289) for 4 h, 8 h, and 24 h and subsequently treated with Talazoparib for seven days. We observed that the IC of PARP was decreased by twofold in both UWB1.289 and UWB1.289 + BRCA1 cells when pre-treated with GSK3326595 (Fig. 3E). This indicated that inhibiting PRMT5 first was sufficient to sensitize the cells to subsequent PARP inhibition, in this isogenic pair of ovarian cancer cell lines. The similar antiproliferative and synergistic effects seen for short (4-hour) non-overlapping exposure, as that after 7 days of combined exposure, suggest that a non-canonical mechanism may be involved in the early stages, which could ultimately lead to the changes in DDR gene expression observed at later time points.
To determine which DNA damage repair pathway was being regulated by GSK3326595 treatment to sensitize these cells to PARP inhibition, we analyzed γ-H2AX, RAD51, and 53BP1 foci as markers for DNA double-strand breaks, HR-mediated repair, and NHEJ-mediated repair, respectively. We observed that there was no significant increase in γ-H2AX foci after PRMT5 inhibition in either UWB1.289 or UWB1.289 + BRCA1 at any of the time points tested (Supplementary Fig. 3A, B, and Supplementary Fig. 4, 5). Additionally, we did not observe any significant changes in RAD51 (Supplementary Fig. 3C, D and Supplementary Fig. 4, 5) or 53BP1 foci (Supplementary Fig. 3E, F and Supplementary Fig. 4, 5), except a modest decrease in RAD51 foci after 4 h treatment in UWB1.289 cells (Supplementary Fig. 3C) and an increase in 53BP1 after 24 h treatment in UWB1.289 + BRCA1 (Supplementary Fig. 3F).
Together, these results suggested that although short-term inhibition of PRMT5 was sufficient to sensitize both UWB1.289 and UWB1.289 + BRCA1 to PARP inhibition. Mechanistically, this was not mediated by inducing DNA damage as measured with known markers at these early time points.
Recent evidence suggests that cancers with MTAP deficiency are vulnerable to PRMT5 inhibition, leading to the development of a new class of PRMT5 inhibitors that are MTA-cooperative and are in clinical trials for MTAP-deleted cancers. Therefore, we sought to evaluate the effect of a potent MTA-cooperative PRMT5 inhibitor, TNG908, in combination with PARP inhibitors in MTAP WT and MTAP null (MTAP) breast cancer cell lines.
We first evaluated the sensitivity of our panel of breast cancer cell lines by treating them with a range of 0-10 µM of TNG908. As expected, MDA-MB-231, an MTAP cell line (Supplementary Fig. 1), was the most sensitive to TNG908 with an IC of 1.83 µM (Table 1), while MDA-MB-436, an MTAP WT cell line (Supplementary Fig. 1), was the least sensitive with an IC > 10 µM (Fig. 4A). Surprisingly, SUM149PT cells were not as sensitive (IC = 9 µM) to TNG908 despite being an MTAP cell line (Supplementary Fig. 1). Then, we examined the efficacy of TNG908 to inhibit PRMT5 activity in all cell lines by measuring global SDMA levels and observed that both 40 nM and 5 µM treatments decreased SDMA levels in all cell lines (Fig. 4B). When we combined TNG908 with Talazoparib and evaluated the efficacy of this combination, we saw that, similar to the GSK-Talazoparib combination, TNG908 also synergized with Talazoparib in all cell lines (Fig. 4C), and the maximum synergy was observed in the MDA-MB-468 (MTAP WT) and MDA-MB-231 (MTAP) cell lines (Fig. 4D). To determine whether sensitivity to TNG908 could be an indicator of synergy with PARP inhibition, we performed a Pearson Correlation using the maximum synergy score (SYN_MAX metric) for each cell line with its respective IC values of TNG908 (Fig. 4E). Similar to the correlation between IC values of GSK3326595 and the maximum synergy scores (Fig. 2C), we obtained a negative correlation coefficient of r = -0.8990, which was also statistically significant (p-value < 0.05, Fig. 4E). This suggested that cell lines that were more sensitive to PRMT5 inhibition using TNG908 were indicative of obtaining high synergy with PARP inhibitors.
Together, these results demonstrated that though MTAP cells are more sensitive to PRMT5 inhibition using the MTA-cooperative inhibitor, the synergy between PRMT5i and PARPi does not depend on the MTAP status of the cells.