OA-16: The rs9344 G risk allele upregulates CCND1 expression through t(11;14) and PAX5 in multiple myeloma

Introduction: Multiple myeloma (MM) is an incurable plasma cell malignancy characterized by recurrent primary cytogenetic abnormalities such as t(11;14)(q13;q32), resulting in upregulation of CCND1. While the G allele of rs9344 (c.723G>A, p.Pro241) in exon 4 of the CCND1 gene is the most significant susceptibility allele for t(11;14) by GWAS, the mechanism of how rs9344 contributes to MM remains unknown.

Methods: We studied 1,359 patients with MM including two independent and ancestrally diverse cohorts (CoMMpass and Tempus). We analyzed whole exome and RNA-seq data and performed a multivariate generalized linear model (GLM). We also analyzed ATAC-seq and ChIP-seq data of the GM12878 normal lymphoblastoid cell line (heterozygous for rs9344 G/A) and MM patient samples. The t(11;14)+ U266 and KMS12 PE cell lines were used for biological confirmation. QRT-PCR was used to assess the expression of CCND1 in U266 cells before and after PAX5 knockdown (KD) with siRNAs and KMS12 PE engineered cell lines with G>A substitution at rs9344. P-values < 0.05 (two-side) were considered statistically significant.

Results: The rs9344 G risk allele was associated with t(11;14) MM (OR=1.88, P-value =3.2x10-5 in CoMMpass and OR=1.54, P-value =0.002 in Tempus cohorts). In cells heterozygous for rs9344 (A/G), the G allele occurred in cis with the t(11;14) in 80.34% (95%CI: 64.92–98.32%) of cases demonstrating a biological preference for the G allele in t(11;14). Among t(11;14) cases, the G allele was associated with a significant increase of H3K27ac, H3K4me3, H3K4me1 and chromatin accessibility (P-value=9.2 x10-3-7.3 x10-16) demonstrating an allele-specific regulatory role for rs9344 within the CCND1 locus. Further, using ENCODE ChIP-seq data from the GM12878 cell line, we identified a PAX5 binding site within CCND1, with the binding motif being one base-pair centromeric to rs9344. The G allele was associated with increased PAX5 signal relative to the A allele (P-value =1.09E-01).
We reason that the increased binding of PAX5 to the G allele contributed to increased expression of CCND1. This was supported by the observations of the increased PAX5 expression in t(11;14) relative to non-t(11;14) MM patient samples, and the increased CCND1 expression within the t(11;14) subgroup with the G allele: 21.98% (95%CI:18.06-25.8%, P-value =1.3 x 10-26) in CoMMpass and 14.48% (95%CI:9.32-19.64%, P-value =5.2 x 10-8) in Tempus. PAX5 KD in U266 cells resulted in a reduction in CCND1 expression and conversion of AA to GG using CRISPR/Cas9 in KMS12 PE cells was associated with an increase in CCND1 expression, confirming the associations of t(11;14) with rs9344 (P-value ≤ 0.002) and with CCND1 (P-value ≤ 1.2x10-143).

Conclusions: We demonstrate a novel enhancer involving rs9344 and its interaction with PAX5 together with t(11;14) accounting for 98% (95%CI:86.51–100%) of CCND1 expression. Future studies will explore how this association contributes to increased risk of t(11;14) MM.