Work Package 1 – Identify the most relevant subunits of SWI/SNF associated to CCA
Here, we first performed in silico analysis of 3 independent cohorts of CCA patients in order to identify the most frequent mutated chromatin remodelling factors. This analysis was carried out considering all chromatin remodelling complexes described in the literature, such as SWI/SNF, PRC2, PRC1, INO80, TIP60, ISWI, and NuRD. In line with previous reports, we found that mutations on genes associated to SWI/SNF complex are most prevalent, varying from 45% to 88%. Within this complex, ARID1A, SMARCA4 and PBRM1 showed the highest mutation frequency. We also identified that low expression of the SWI/SNF members leads to a poorer prognosis in patients.
In order to determine the functional role of SWI/SNF on the maintenance of CCA, we performed a cell-based high-throughput shRNA knockdown screening, querying all known members of this family. In total, we investigated a library of 70 different lentiviral shRNA clones, targeting 18 genes, with proper validation of RNA targeting efficiency and specificity.
Given the redundancy of the SWI/SNF complex, where partners have usually overlapping roles in different contexts, our results show that some of them are dispensable for the proliferation of the cell. However, ARID1A and SMARCA4 showed a poorer proliferation rate when compared to their control counterpart. We further investigated 7 different cell lines in order to confirm whether this is a phenotype specific to CCA.
Based on these findings, we were interested in observing the transcriptional landscape of CCA cells in the absence of ARID1A or SMARCA4. We carried out gene expression microarray analysis in our CCA cells panel. We found 1552 and 2174 differentially expressed genes common to all cell lines, respectively (Figure 3B and C). Among those genes, there was a significant representation of DNA repair-associated genes.
Work Package 2 – Characterize target candidates in the context of homologous recombination (HR) repair
There are mainly two exclusive and competing pathways in this system – homologous recombination (HR) repair and non-homologous end joining (NHEJ). We further sought to investigate the HR repair dynamics in CCA cells by observing early response markers, phosphorylated H2AX (γH2AX) and ATM (pATM). We observed SWI/SNF-knocked-down CCA cells presented a lower level of basal γH2AX compared to its control counterpart. However, when damage was induced with ionizing radiation (IR) those cells exhibited a comparatively higher DNA damage signalling, and for a prolonged time. Noteworthy, we also knocked-down potential redundant partners, in order to observe their effect. The increase in γH2AX marks was followed by an equally higher increase in pATM. Phosphorylation of ATM is associated to a specific type of DNA repair, the HR pathway. Therefore, from these results we reasoned that HR downstream factors would also be affected in the absence of SWI/SNF components. Indeed, we observed lower levels of RAD51 in IR-treaded cells. Interestingly, we found no significant difference in levels of the NHEJ marker, 53BP1, in IR-treated cells in the absence of SWI/SNF components. Therefore, we further investigated the role of HR repair in the maintenance of CCA.
In excess of evaluating the initial HR signalling we further investigated the overall transcriptional effect in the absence of ARID1A or SMARCA4 in CCA cell lines in the context of DNA repair. Gene expression on DNA repair-associated pathways was found enriched in the absence of ARID1A or SMARCA4. Gene set enriched analysis showed a negative correlation of HR repair-associated pathway to ARID1A-depleted cells, indicating that this pathways is dysregulated in the absence of SWI/SNF. Similar association was indicated in SMARCA4-depleted cells. Interestingly, despite the high heterogeneity of CCA cells, the DNA repair pathway was equally disrupted in all cell lines tested.
Next, we adapted a gene reporter assay to CCA cell lines in order to evaluate the efficiency of HR repair. Here, depletion of SMARCA4 led to a drastic higher efficiency of repair, whilst the absence of ARID1A showed no significant difference. That discrepancy might be due to the role of those factors. ARID1A is an accessory subunit for the SWI/SNF complex and share homology and function with ARID1B and potentially ARID2. Hence, the expression of those factors might be regulated in order to rescue the loss of ARID1A. On the other hand, SMARCA4, and SMARCA2, are the catalytic subunit of the complex with DNA-dependent ATPase activity. Although they also have equally exclusive functions, they are both required for full activation of DNA damage response, corroborating with our results.
In summary, on this work package we highlighted the impact of SWI/SNF on the DNA damage response and repair in CCA cells. Furthermore, the homologous recombination repair showed to be the pathway of preference in the absence of a functional complex, indicated by the loss of the catalytic subunit SMARCA4.
