Introduction:

ARID1A is an essential component of the BAF complex, a large SWI/SNF chromatin-remodeling complex which is implicated in many cancer types. Roughly 20% of bladder cancers (BC) contain genomic alterations in ARID1A, making it an attractive target for both mechanistic exploration and potential therapeutic inhibition in BC. However, the biological consequences of ARID1A loss in BC remain to be elucidated. Based on the observation that the frequency of ARID1A alterations increases with increasing stage of disease, we hypothesize that ARID1A deficiency primes an invasive transformation in BC, thus increasing the likelihood of disease progression and metastases. We used CRISPR/Cas9-mediated gene editing to knockout (KO) the ARID1A gene in several BC cell lines, thus providing a platform for cellular function assays. We show that ARIDA1 loss potentiates an invasive phenotype in these cell lines, giving evidence for its transformative role in the development of higher stage and metastatic BC.

Methods:

The Cancer Genome Atlas (TCGA) was leveraged to determine the frequency and nature of alterations in ARID1A across tumor types. The Cancer Cell Line Encyclopedia (CCLE) was then used to query all available BC cell lines for mutations in chromatin-remodeling genes. CRISPR/Cas9-mediated gene editing was then used to KO ARID1A in three cell lines with wild-type (WT) ARID1A (T24, SW1710, and 5637). Loss of ARID1A protein was confirmed using Western blot. Baseline proliferation studies (Sartorius Incucyte Live Cell Imaging) and invasion studies (scratch assay) were performed on the KO and WT cell lines to define how ARID1A loss impacts these cellular properties.

Results:

Query of TCGA revealed ARID1A alterations present in 20% of BC sample genomes.  CCLE was queried for BC cell lines with a broad spectrum of mutational profiles and we chose to utilize three BC cell lines (SW1710, T24, and 5637) with wild-type ARID1A expression (Figure 1A). CRISPR-Cas9 directed KO of ARID1A yielded two ARID1A-deficient clones for SW1710 and one clone each for T24 and 5637 (Figure 1B). The proliferation rates for ARID1A WT and KO cell line pairs were analyzed and showed no significant differences (Figure 2A). A scratch assay was performed with the cell lines in triplicate, with each ARID1A-deficient cell line showing a trend towards more rapid wound closure than its matched WT cell line (WT=31.5% vs. KO=53.5% for SW1710, p=0.15; WT=17.2% vs. KO=52.0% for T24, p=0.07; WT=10.4% vs. KO=13.9% for 5637, p=0.59) (Figure 2B).

Conclusion:

To explore the increased frequency of ARID1A alterations in metastatic BC compared to non-invasive tumors, we developed model cell lines of ARID1A-deficient BC using CRISPR-Cas9 gene editing. We compared the relative invasiveness of the lines using a scratch assay. All lines trended towards a more invasive phenotype following ARID1A KO, which was especially pronounced in T24. Taken together, these results suggest that ARID1A deficiency in BC may prime tumors for a more invasive phenotype, thus explaining the higher frequency of ARID1A alterations in higher stage and metastatic samples. Future studies include probing the downstream pathways which may explain the invasive phenotype of ARID1A KO and using mouse xenograft models to define ARID1A mutations in vivo.

Funding: N/A