ARID1A facilitates KRAS signaling-regulated enhancer activity in an AP1-dependent manner in colorectal cancer cells

Abstract

Background: ARID1A (AT-rich interactive domain-containing protein 1A) is a subunit of the BAF chromatin remodeling complex and plays roles in transcriptional regulation and DNA damage response. Mutations in ARID1A that lead to inactivation or loss of expression are frequent and widespread across many cancer types including colorectal cancer (CRC). A tumor suppressor role of ARID1A has been established in a number of tumor types including CRC where the genetic inactivation of Arid1a alone led to the formation of invasive colorectal adenocarcinomas in mice. Mechanistically, ARID1A has been described to largely function through the regulation of enhancer activity.

Methods: To mimic ARID1A-deficient colorectal cancer, we used CRISPR/Cas9-mediated gene editing to inactivate the ARID1A gene in established colorectal cancer cell lines. We integrated gene expression analyses with genome-wide ARID1A occupancy and epigenomic mapping data to decipher ARID1A-dependent transcriptional regulatory mechanisms.

Results: Interestingly, we found that CRC cell lines harboring KRAS mutations are critically dependent on ARID1A function. In the absence of ARID1A, proliferation of these cell lines is severely impaired, suggesting an essential role for ARID1A in this context. Mechanistically, we showed that ARID1A acts as a co-factor at enhancers occupied by AP1 transcription factors acting downstream of the MEK/ERK pathway. Consistently, loss of ARID1A led to a disruption of KRAS/AP1-dependent enhancer activity, accompanied by a downregulation of expression of the associated target genes.

Conclusions: We identify a previously unknown context-dependent tumor-supporting function of ARID1A in CRC downstream of KRAS signaling. Upon the loss of ARID1A in KRAS-mutated cells, enhancers that are co-occupied by ARID1A and the AP1 transcription factors become inactive, thereby leading to decreased target gene expression. Thus, targeting of the BAF complex in KRAS-mutated CRC may offer a unique, previously unknown, context-dependent therapeutic option in CRC.

Keywords: AP1; ARID1A; BAF complex; Colorectal cancer; Enhancers; KRAS; MEK/ERK pathway; Transcriptional regulation.

Conflict of interest statement

The authors declare that they have no competing interests.

Figures

Fig. 1

ARID1A protein loss in human cancer samples is consistent with mutation rates. ARID1A is mutated in up to ~ 12% colorectal cancers from different colorectal cancer cohorts [3, 4] (a). Representative immunohistochemical staining images show clear ARID1A staining in the rectal mucosa and positive tumors while negative tumors show no staining at all (b). The University Medical Center Gӧttingen rectal cancer cohort showed 81.7% strongly positive ARID1A staining (++) and 3.7% showed weakly positive staining (+) while 14.6 % of the samples were negative for ARID1A staining (c)

Fig. 2

The loss of ARID1A impairs proliferation of a subset of colorectal cancer cell lines. To mimic ARID1A-deficient colorectal cancer, we used CRISPR/Cas9-mediated gene editing to delete exon 5 of the ARID1A gene from four colorectal cancer cell lines HCT116, DLD1, COLO320DM, and HT29. Regions flanking exon 5 were targeted by two guide RNAs leading to an out of frame gene product. Genotyping PCRs show a shorter product of 761 bp in the ARID1A knockout (KO) cells (a, middle panel). This led to a complete loss of the ARID1A protein from the KO cells (a, lower panel). HSC70 is used as a loading control. Representative graphs show that the proliferation of the HCT116 and DLD1 cell lines is significantly impaired by the deletion of ARID1A, whereas the proliferation of COLO320DM and HT29 is unaffected by this knockout (b). Error bars represent standard deviation between the replicates. n for proliferation assay = 2, *p < 0.05 unpaired t test

Fig. 3

Loss of ARID1A results in deregulated expression of MEK/ERK pathway target genes. Forty-eight genes are commonly downregulated in the HCT116 and DLD1 cell lines (a). Gene set enrichment analysis shows that gene sets containing targets of the MEK/ERK pathway are enriched in the parental condition (a). The genes EREG, F3, and JAG1 are highly expressed in cell lines with KRASG13D mutations as analyzed using on the Morpheus tool to analyze data from the Cancer Cell Line Encyclopedia (CCLE) [36] (b). The scale represents the minimum expression in a particular row (blue) to the maximum expression in that row (red). Treatment with 20 nM trametinib for 24 h leads to a reduction of phosphorylated ERK (pERK) (c). The expression of EREG, F3, and JAG1 was significantly reduced by trametinib treatment and ARID1A deletion. Treatment with trametinib in ARID1A-deficient cells did not lead to a further reduction of gene expression (d). ARID1A mutations are significantly mutually exclusive with KRAS mutations in the TCGA colorectal cancer patient cohort (e). The y-axis represents −log10 (p value) and the blue dots represent significantly mutually exclusive or co-occurring mutations. pERK and JUND levels remain unchanged in ARID1A-deficient cells. HSC70 is used as a loading control. The AP1 transcription network gene set [41] is enriched in the parental condition (f). qRT-PCRs were performed for biological triplicates and technical duplicates. Error bars represent the standard deviation between three biological replicates. Significance was calculated using an unpaired t test, *p < 0.05, **p < 0.01, ***p < 0.001

Fig. 4

ARID1A is localized to AP1-occupied enhancers. ARID1A co-localizes with the BAF complex subunits SMARCA4 and SMARCC1 as well as the active mark H3K27ac. The heatmaps represent all ARID1A binding sites ordered in descending order of ARID1A occupancy (a). ARID1A localizes mainly to regions that are distal to transcription start sites (TSS) with the maximum number of peaks between 5 and 500 kb of TSSs (b). ARID1A-bound enhancers were defined based on overlap of ARID1A with H3K27ac and ATAC-seq. From these regions, any annotated TSSs were subtracted and 3061 ARID1A-bound enhancers were identified (c). Motif analysis on ARID1A-bound enhancers shows enrichment for the Jun-AP1(bZIP) motif (d). ReMap analysis shows that several AP1 transcription factors such as FOSL1, FOSL2, FOS, and JUND are enriched on ARID1A-occupied regions identified in HCT116 cells (e). Co-localization of JUND and FOSL1 at ARID1A-bound enhancers in HCT116 cells is shown. The heatmaps represent the 3061 ARID1A-bound enhancers in descending order of ARID1A occupancy (f)

Fig. 5

The loss of ARID1A leads to a loss of enhancer activity. Six thousand seven hundred forty-one genes associated with ARID1A-bound enhancers were determined. Three hundred seventeen genes were also downregulated by the knockout of ARID1A. Among these were the EREG, F3, and JAG1 genes (a). The three MEK/ERK pathway target genes identified were downregulated by the deletion of ARID1A in both KRAS-mutated cell lines (HCT116 and DLD1) (b). At the genomic loci for these genes, we identified potential enhancers which are occupied by ARID1A, JUND, and H3K27ac within the same topologically associated domains (TAD). Vertical dotted lines represent TAD boundaries. The potential enhancers are also accessible as assessed by ATAC-seq and are transcriptionally active as assessed by PRO-seq (c). There is a significant reduction of H3K27ac at these enhancers upon the deletion of ARID1A (c, d). qRT-PCRs for gene expression and ChIP were run in biological triplicates and technical duplicates. The dotted lines represent the average ChIP-qPCR signal for the negative control IgG. Error bars represent the standard deviation between three biological replicates. Significance was calculated using unpaired t test, *p < 0.05, **p < 0.01, ***p < 0.001

Fig. 6

The attenuation of the MEK/ERK pathway leads to a loss of enhancer activity at ARID1A-bound enhancers. Upon treatment with 20 nM trametinib for 24 h, the occupancy of JUND was reduced at the potential enhancers identified (a). Moreover, the same treatment also leads to the reduction of ARID1A occupancy and a loss of the H3K27ac mark on these enhancers (b, c). ChIP was carried out in biological triplicates and qRT-PCRs were run in technical duplicates. The dotted lines represent the average ChIP-qPCR signal for the negative control IgG. Error bars represent the standard deviation between three biological replicates. Significance was calculated using unpaired t test, *p < 0.05, **p < 0.01, ***p < 0.001. In summary, we show that the loss of ARID1A has context-dependent functions in colorectal cancer. Mathur et al. describe that in a wild-type background, the loss of ARID1A is tumor suppressive and leads to the formation of invasive adenocarcinomas. This function is mediated through control of enhancer activity wherein there is a loss of the active H3K27ac mark upon the loss of ARID1A [21]. We show that in colorectal cancer cells which harbor KRAS mutations, ARID1A has a tumor-supporting function by facilitating transcription at enhancers downstream of the MEK/ERK pathway also partially through the placement of the H3K27ac mark (d). In the first case, synthetic lethal targets described in the literature may provide potential therapeutic targets [, , –54]. In the case described here, targeting the BAF complex itself using PROTACS could be an option