YAP, PRC2 and MYC contribute to the transcriptional repression of tumor-suppressor genes in NSCLC
Previous studies showed that YAP and EZH2 synergize to repress the transcription of lineage-specific genes and onco-suppressor genes in NSCLC, Schwann cells, trophoblast and intestine, and that YY1 (Yin Yang 1) transcription factor may bridge YAP and EZH2 at the promoter of cell-cycle regulators (17–20). MYC was also proposed to cooperate with EZH2 or YY1 to repress the transcription of onco-suppressor genes in cancer (21–23) while cooperating with YAP/TAZ mainly to enhance the transcription of oncogenes (24)].
Considering the oncogenic impact of PTEN loss on NSCLC tumorgenicity and its possible downregulation by transcriptional and post-transcriptional mechanisms, we examined its promoter region using the Cistrome Data Browser and found a clear co-localization of TEAD4, YY1 and MYC onto its promoter in the A549 NSCLC cell line (Fig. 1A). TEAD1-4 transcription factors are known to bind YAP/TAZ on both activated and repressed genes promoters. Therefore, despite the absence of YAP track in the Cistrome DB of A549 cells, we assume that YAP binds PTEN promoter since a YAP peak overlapping with TEAD4, YY1 and MYC is present in the mesothelioma cell line H2052 (Fig. 1A). The presence of YAP, MYC and YY1 on PTEN promoter was validated by chromatin immunoprecipitation (ChIP) in the H1299 NSCLC cell line (Fig. 1B). Furthermore, co-immunoprecipitation (Co-IP) and immunofluorescence (IF) experiments in H1299 cells revealed physical interaction and cellular co-localization of endogenous YAP with TEAD1, pMYC, EZH2, YY1 as well as with LaminA/C, a structural protein associated with the inner nuclear membrane where the epigenetically repressed chromatin is clustered (25) (26) (Fig. 1C-J, S8). Interestingly, pMYC also co-immuno-precipitates, co-localizes and interacts in the nucleus with Lamin A/C (Fig. 1C, F, S8). We obtained similar patterns of YAP distribution and co-localization with Lamin A/C, pMYC, YY1 and H3K27me3 in H1975 cells (Fig. S1). Notably, H3K27me3 is enriched at the nuclear lamina (Fig. S1D) with a ring-like pattern, similar to YAP (Fig. S1A).
Proximity ligation assay (PLA) in H1299 confirmed in situ the interaction of YAP with EZH2, YY1, pMYC, and lamin A/C as well as the interaction between pMYC and lamin A/C (Fig. 1K).
By matching published lists of genes upregulated in YAP- or TAZ-depleted cells (GSE151200) (27), which are possibly target genes transcriptionally repressed by YAP or TAZ, with those that are bound by MYC (ChIP-Atlas) and either EZH2 (28) or H3K27me3 (GSE29611) in A549 cells, we obtained two distinct lists of genes (Table S1). These lists were obtained by applying pval < 0.05 and padj < 0.3. The heatmaps and Venn diagram shown in Fig.S2A, B show that TAZ/EZH2/MYC co-represses a higher number of genes (N = 165) compared to YAP/EZH2/MYC (N = 50 genes). Moreover, 20 genes are commonly repressed by both YAP and TAZ, indicating that almost half of YAP repressed genes (40%) are also repressed by TAZ. However, only 12% of genes repressed by TAZ are also repressed by YAP (Fig.S2A, B, Table S1). Genes set enrichment analysis (GSEA) of genes repressed by either YAP or TAZ and co-bound by MYC and EZH2/H3k27me3 highlights apical junction, apical surface, EMT, genes downregulated by KRAS activation and IL2-STAT5 signaling, as the most significant pathways affected by these proteins (Fig. S2C, Table S1, sheet 3). Collectively, these data suggest that the transcriptional repression mediated by YAP, TAZ, MYC and EZH2 may be a general mechanism, and supports the emerging notion that YAP and TAZ may have both overlapping and distinct functions not only in the transcriptional activation but also in the transcriptional repression of different subsets of genes (29).
Oncogenic MYC cooperates with YAP and TAZ to repress transcription of PTEN
We have noticed that PTEN was not present in the heatmaps of Fig. S2A and Table S1, possibly because the analysis included upregulated genes in response to single KD of either YAP or TAZ. Indeed, when YAP and TAZ were concomitantly depleted in two NSCLC cell lines, H1299 and H1975, a strong upregulation of PTEN was observed, concurrent with a decrease in CYCD1 (cyclin D1) transcript, a cell cycle regulator that is negatively modulated by PTEN (Fig. 2A, B, S9A) (30). Importantly, the expression of CYCD1 and PTEN are anti-correlated in NSCLC patients (Fig. 2C). KD of PTEN robustly upregulated CYCD1 and rescued CYCD1 levels in siYAP/TAZ double KD cells (Fig. 2D, S9B). Conversely, PTEN overexpression reduced CYCD1 levels and increased the fraction of cells at the G1/S transition in two other NSCLC cell lines, H1975 and H358, with no evident effects in H1299 cells (Fig. S2D, E, S12B) (30).
To evaluate the impact of MYC on PTEN expression, we downregulated its expression by siRNA in H1299 and H1975 cells, and observed a significant upregulation of PTEN only in H1975 cells (Fig. 2E, F, S9C). However, colony formation was inhibited in both cell lines upon MYC depletion (Fig. 2G), suggesting that MYC elicits its oncogenic function through several mechanisms beyond PTEN downregulation. Upon MYC depletion, we also observed a cell line-specific de-repression of TGFBR2 and p21, previously shown to be repressed by YAP, TAZ and EZH2, implying a similar repression mechanism for those genes (Fig. 2E, F, S9C) (18). Although depletion of MYC alone did not affect PTEN expression in H1299 cells, we found that MYC overexpression in YAP/TAZ depleted H1299 cells could rescue PTEN, p21 and TGFBR2 repression, as well as a partial rescue of the colony-forming ability (Fig. 2H, I, S9D) implying a role for MYC in PTEN repression also in H1299 cells. In LUAD patients, upregulation of EZH2, MYC and YY1 was associated with downregulation of PTEN expression, both at the protein and the RNA level. This association was specifically observed in patients harboring somatic alterations in the Hippo pathway and MYC-related pathways (Fig. S3A, B). Moreover, the expression level of PTEN protein negatively correlates with its putative transcriptional repressors TAZ, EZH2 and MYC while it positively correlates with pYAPS127 (the YAP phosphorylated form that is retained in the cytoplasm), as well as with the onco-suppressive transcripts TGFBR2, CDKN1A (p21) and CDKN2B (p15), which were previously shown to be repressed by YAP/TAZ and EZH2 (Fig. S3C) (18). Finally, Kaplan Mayer survival analysis shows a better prognosis for LUAD patients expressing high levels of PTEN, especially in the earlier stages of the disease, supporting its role as an early prognostic onco-suppressor (Fig. S3D).
EZH2 and MYC inhibition activates YAP and TAZ targets
The oncogenic role of EZH2 and MYC in the transcriptional repression of onco-suppressors suggests that their inhibition would reactivate those onco-suppressors. Several specific inhibitors of EZH2, which catalyzes tri-methylation of histone H3 at Lys 27 (H3K27me3), have been developed, and Tazemetostat has been approved by the FDA for epithelioid sarcoma (31), diffuse large B-cell lymphoma and relapsed or refractory follicular lymphoma (32). While Tazemetostat is effective for the above cancer types, it is ineffective as monotherapy for other malignancies, especially for solid cancers, where cells show either innate or acquired resistance to this compound (33). Huang and coworkers showed that Tazemetostat treatment of solid tumors induced a decrease in global H3K27me3, as expected, which was accompanied by an increase in total H3K27ac. It was, therefore, proposed that an oncogenic transcriptional reprogramming mediated by the interaction of the epigenetic protein MLL1 with the p300/CBP complex onto several oncogenes induces global H3K27ac upregulation upon H3K27me loss, driving the hyper activation of genes belonging to oncogenic pathways that counteract the de-repression of onco-suppressor pathways (33).
We previously showed that H1299 and H1975 cells are resistant to Tazemetostat, with an IC50 of 24 µM and 36 µM, respectively (18). We also observed a reduced number of colonies in H1299 but not in H1975 or h358 cells at micromolar doses of Tazemetostat (Fig. S4A). In these cells, depletion of EZH2 by siRNA or treatment with 2µM Tazemetostat increased the expression of YAP and TAZ and their transcriptional targets CTGF, ANKRD1 and CYR61 (Fig. 3A-C, S10A). These effects were also observed in H1299 and H1975 cells upon treatment with a lower dosage of Tazemetostat (1µM, Fig. S4B, S12C, D). Moreover, Tazemetostat treatment induces a dose-dependent increase of global H3K27ac that counteracts the H3K27me3 decrease (Fig. 3D, S10B, C). This epigenetic switch was observed previously in other Tazemetostat-resistant cells but not in sensitive cell lines (33). We found that global levels of YAP and TAZ are increased upon Tazemetostat treatment in a dose-dependent fashion, especially at high doses and more strongly in H1975 cells (Fig. 3D, S10B, C) which are more resistant to Tazemetostat than H1299 (Fig. S4A) (18). In accordance, both in lung adenocarcinoma (LUAD) and in lung squamous cell carcinoma (LUSC) patients, the EZH2 mRNA has a negative correlation with CTGF, ANKRD1 and CYR61 transcripts (Fig. 3E).
Based on these results, we speculated that the upregulation of YAP- and TAZ- oncogenic signature may associate with the acquired resistance of H1299 and more strongly of H1975 cells to Tazemetostat treatment. Strikingly, the knockdown of YAP and TAZ increased the sensitivity of H1975 cells to very low doses of Tazemetostat (Fig. S4C) compared to Tazemetostat alone, which did not affect colony formation even at higher doses (up to 8µM, Fig. S4A). Interestingly, MYC depletion or inhibition by MYCi975 also increased the expression of YAP and TAZ targets in H1299, H1975 and H358 cells (Fig. 3F-G, S4D-E, S10D-E, S12E, S13A) consistent with previous reports in liver and breast cancer models (24, 34). Furthermore, the three cell lines were resistant to micromolar doses of MYCi975 as shown by the colony formation assay (Fig. S4F). In contrast, treatment with JQ1, a BRD4 inhibitor that indirectly inhibits oncogenic MYC function (35), induced opposite effects in different cell lines, increasing the expression of total YAP, TAZ, CTGF and ANKRD1 in H1299 but decreasing them in H1975 cells (Fig. S4G, S13B). This might be related to the broad effects of JQ1, which not only inhibits MYC but also other oncogenic pathways through epigenetic mechanisms. Notably, it was previously shown that combination of Tazemetostat with JQ1 inhibits MLL1-mediated H3K27ac accumulation onto oncogenes, and reverses the resistance to EZH2 inhibitors in different cellular models (33). However, considering the different effects of JQ1 treatment in our models, we decided to inhibit MYC directly with MYCi975 and hypothesized that combined inhibition of YAP/TAZ together with MYC and EZH2 may be the best strategy to prevent activation of YAP/TAZ-mediated oncogenic pathways and concurrently de-repressing onco-suppressive pathways.
Concomitant inhibition of EZH2, MYC and YAP/TAZ maximizes the de-repression of onco-suppressors while preventing the activation of oncogenes
We next depleted EZH2, MYC and YAP/TAZ by RNA interference (RNAi) either alone or in combination. We found that the triple knock-down (KD) of EZH2, MYC and YAP/TAZ led to stronger de-repression of onco-suppressor genes (TGFBR2, PTEN, p21) while reducing the activation of YAP and TAZ targets, which was observed upon double EZH2/MYC KD (Fig. 4A, S11A). Consistent with these results, we found that triple KD increased the binding of active RNAPol2 onto the PTEN promoter but decreased the binding onto the CTGF and CYR61 promoters (Fig. 4B). Functionally, the triple KD had stronger anti-tumorigenic effects compared to single or double KD in NSCLC cell lines, as shown by the reduced colony formation potential (Fig. 4C) and by the apoptotic phenotype demonstrated by AnnexinV/PI staining and by cleaved casp3 (Fig. 4AD-E, S5A, S11B). Conversely, KD of PTEN reduced the percentage of apoptotic cells already 24h post-transfection (Fig. S5B) and the level of cleaved PARP/casp3 (Fig. S5C, S14). These results suggest that the increased apoptosis observed in response to triple KD is possibly mediated, at least in part, by the increase in PTEN and the concurrent decrease of YAP and TAZ. Moreover, PTEN KD increased colony formation in H1299, H1975 and H358 cells (Fig. S5D). To further confirm the KD results, we treated H1299 and H1975 cells with low doses of MYCi975 (36), Tazemetostat and Dasatinib (an indirect YAP and TAZ inhibitor) (37). Low doses of the combined three inhibitors reduced the viability of both cell lines and de-repressed onco-suppressor genes while suppressing reactivation of YAP and TAZ oncogenes (Fig. S6A, B, S15) consistent with the KD experiments.
Knockdown of PTEN restores the tumorigenic properties of MYC, EZH2 and YAP/TAZ depleted NSCLC cells.
To further demonstrate the link between the three transcriptional regulators: MYC, EZH2 and YAP/TAZ and the tumor suppressor PTEN, we downloaded a list of transcripts and proteins positively or negatively correlated to PTEN in LUAD patients from the lung adenocarcinoma TCGA, Firehose Legacy study (Table S2). Gene set enrichment analysis (GSEA) of anti-correlated proteins highlighted E2F targets, mTORC1 signaling, and MYC targets as significantly anti-correlated to PTEN, consistent with the roles of PTEN in cell proliferation, cell growth, and cell metabolism (Table S2, Fig. 5A).
To demonstrate that inhibition of PTEN expression through the concerted function of MYC, EZH2, and YAP/TAZ contributes to NSCLC tumorigenicity, we knocked down PTEN together with MYC, EZH2, and YAP/TAZ (defined as “MEY”). MEY KD induced PTEN and p21 reactivation and concurrently inhibited the PI3K/AKT/mTOR pathway as shown by reduced pAKT levels (Fig. 5B, S12A). These effects were accompanied by reduced mitochondrial respiration (Fig. 5C), which was previously shown to be essential for early events of lung tumorigenesis (38). Strikingly, the concomitant KD of PTEN could partially rescue these cellular changes (Fig. 5B, C, S7, S12A).
Altogether, these observations are summarized in Fig. 5D: in cancer context, onco-suppressors are repressed while oncogenes are hyper-activated due to the aberrant functions of EZH2, MYC and YAP/TAZ. Treatment with Tazemetostat (or with MYC inhibitors) as a single agent, reactivates onco-suppressors on one hand but concomitantly may induce stronger activation of oncogenic signaling, including the activation of YAP/TAZ signature, as a mechanism of acquired resistance. Targeting of YAP/TAZ oncogenic signature may overcome the acquired resistance to Tazemetostat and/or MYC inhibitors. Combined targeting of EZH2, MYC and YAP/TAZ can, therefore, effectively de-repress expression of tumor suppressors and concurrently repress the expression of oncogenic pathways, to ultimately inhibit tumor growth.