Disruptive ARID1A Mutations in Follicular Lymphoma Impair DNA Repair Efficiency and Are Associated with Favorable Outcome in Patients Receiving First-Line Immunochemotherapy

Follicular lymphoma (FL) is among the most common lymphomas worldwide and considered incurable for the majority of patients who present with advanced disease. FL is a clinically and molecularly heterogeneous disease. We have recently established a clinicogenetic risk model (m7-FLIPI) that integrates the mutation status of 7 genes, including ARID1A, into a predictive algorithm for improved risk stratification for failure-free survival in patients receiving first-line immunochemotherapy (Pastore et al., Lancet Oncology 2015). When adjusted for FLIPI and ECOG performance status, the presence of ARID1A mutations correlated with longer failure-free survival (HR 2.85, 95% confidence interval (CI) 1.12-7.27; p=0.049) in patients receiving first-line R-CHOP. The underlying molecular mechanism for better treatment outcome of patients with ARID1A mutated FL is unclear.

ARID1A is a member of SWI/SNF nucleosome remodeling complex. Evolving evidence indicates that the SWI/SNF complex functions as a tumor suppressor and is implicated in DNA damage repair. ARID1A is the most commonly mutated SWI/SNF complex member in FL: in our series, we identified 44 ARID1A mutations in 304 FL cases (14%). These mutations were primarily heterozygous and disruptive (66%), another 6 cases harbored splice site mutations. Primary patient samples and lymphoma cell lines that harbored ARID1A mutations (e.g., SU-DHL5, WSU-FSCCL, Namalwa) were had lower ARID1A protein expression as compared to ARID1A wild type (wt) t(14;18)-positive lymphoma cell lines (OCI-Ly1, -Ly8, DB).

Non-homologous end joining (NHEJ) has been reported to be the predominant pathway of DNA double strand break (DSB) repair in FL (Koues et al., Immunity 2015) and is indispensible for successful VDJ rearrangement during early B-cell maturation.To analyze the impact of Arid1a disruption on B-cell maturation in vivo, we used a conditional Arid1a knock-out mouse model (Arid1a^f/f; Gao, PNAS 2008). 8-10 week old Mx1-Cre⁺ Arid1a ^f/f  mice and control (Mx1-Cre^- Arid1a^f/f) mice were treated with poly(IC). Bone marrow (BM) were harvested from these animals 14 days later. Conditionally deletion of Arid1a (n=3) resulted in severely impaired B-cell maturation in BM cells affecting Hardy fractions B/C though F. E.g., flow cytometric analysis showed that the frequency of fraction C (B220⁺ CD43⁺ HSA⁺ BP-1⁺) was greatly reduced in BM from Mx1-Cre⁺ Arid1a ^f/f  vs in the control group (0.1 % vs 1.2 %, n=3), consistent with the inability to successfully complete VDJ recombination.

As a proof-of-principle experiment we assayed the impact of Arid1a loss on DSB repair efficiency using mouse embryo fibroblasts (MEFs) from either wt or conditional Arid1a knock-out mice. MEFs were irradiated with 2 Gy 36 hrs after retroviral transduction with Cre and stained for γH2A.X and 53BP1 to assay for NHEJ. In three independent and blinded experiments, evaluation of >100 cells per condition demonstrated significantly more double-positive cells (>6 foci/cell) in Arid1a^-/- MEFs (43+/-14%) compared to both Arid1a^f/f MEFs (9+/-1%) and Cre-exposed wt MEFs (15+/-5%) at 16 hrs post irradiation, whereas there was no significant difference for Arid1a^f/for wt MEFs with or without Cre at baseline (all <10%), and 1 hr post irradiation (all >50%).

ARID1A haplodeficient lymphoma cell lines (SU-DHL5, WSU-FSCCL, Namalwa) were 10-100 fold more sensitive to the DSB inducing doxorubicin treatment compared to ARID1A wt lymphoma cell lines (OCI-Ly1, -Ly8, DB). Tet-induced shRNA knock-down (via pTRIPZ vectors) of endogenous ARID1A in two t(14;18)-positive lymphoma cell lines with wt ARID1A (OCI-Ly1 and OCI-Ly8) doubled the number of NHEJ foci by 53-BP1 staining in two independent experiments. Furthermore, shRNA knock-down of ARID1A resulted in increased sensitivity to doxorubicine (IC50 <10 nM) in OCI-Ly8 cell lines as compared to 25 nM for scrambled shRNA.

We conclude that ARID1A is recurrently and significantly mutated in FL. Most mutations are disruptive and result in functionally relevant ARID1A protein loss. ARID1A disruption slows the repair kinetics of NHEJ and delays the clearance of DSB both in  a genetically clean MEF model and in t(14;18)-positive lymphoma cell lines. As a result, ARID1A loss might underlie the genomic instability of FL, but -at the same time- sensitize tumors to DNA damaging agents and irradiation.