Molecular Mechanisms of ARID5B-Mediated Genetic Susceptibility to Acute Lymphoblastic Leukemia

There is growing evidence for the inherited basis of susceptibility to childhood acute lymphoblastic leukemia (ALL). Genome-wide association studies have identified non-coding ALL risk variants at the ARID5B gene locus, but their exact functional effects and the molecular mechanism linking ARID5B to B-cell ALL leukemogenesis remain largely unknown.

We performed targeted sequencing of ARID5B in germline DNA of 5008 children with ALL. Variants were evaluated for association with ALL susceptibility using 3644 patients from the UK10K cohort as non-ALL controls, under an additive model. Cis-regulatory elements in ARID5B were systematically identified using dCas9-KRAB-mediated enhancer interference system enhancer screen in ALL cells. Disruption of transcription factor binding by ARID5B variant was predicted informatically and then confirmed using chromatin immunoprecipitation and coimmunoprecipitation. ARID5B variant association with hematological traits was examined using UK Biobank dataset. All statistical tests were 2-sided.

We identified 54 common variants in ARID5B statistically significantly associated with leukemia risk, all of which were noncoding. Six cis-regulatory elements at the ARID5B locus were discovered using CRISPR-based high-throughput enhancer screening. Strikingly, the top ALL risk variant (rs7090445, P = 5.57 × 10-45) is located precisely within the strongest enhancer element, which is also distally tethered to the ARID5B promoter. The variant allele disrupts the MEF2C binding motif sequence, resulting in reduced MEF2C affinity and decreased local chromosome accessibility. MEF2C influences ARID5B expression in ALL, likely via a transcription factor complex with RUNX1. Using the UK Biobank dataset (n = 349 861), we showed that rs7090445 was also associated with lymphocyte percentage and count in the general population (P = 8.6 × 10-22 and 2.1 × 10-18, respectively).

Our results indicate that ALL risk variants in ARID5B function by modulating cis-regulatory elements at this locus.