Novel Non-Coding Candidate Drivers of Multiple Myeloma Linked to Plasma Cell Tumor Biology and B Cell Development

Introduction

Few mutations in non-coding elements, such as the TERT and FOXA1 promoter regions, have been shown to drive solid cancers. In multiple myeloma (MM), the significance and function of non-coding mutation hotspots remain unknown. Therefore, we conducted discovery analyses to identify significantly recurrent non-coding candidate drivers in MM. We then linked these first candidate drivers to the biological processes of precancerous and cancerous stages of MM, and characterized significant mutations in the promoter of ILF2 and in super-enhancer regions of B cells.

Methods

Whole-genome sequencing (WGS) data from 1,030 individuals with MM (n=812) and its precursor conditions (n=218 with monoclonal gammopathy of undetermined significance [MGUS] or smoldering multiple myeloma [SMM]) was analyzed homogeneously with the Cancer Genome Analysis pipeline from the Cancer Program of the Broad Institute of MIT and Harvard. Non-coding hotspots were discovered with the DIG method [Sherman 2022]. Candidate drivers were annotated for associated putative genes targets, mutational processes, and their presence was assessed in WGS data from publicly available normal naïve and memory B cells.

Results

Among 1,030 individuals, we identified 66 non-coding elements with significantly recurrent mutations across MGUS/SMM and MM (q<0.05). Of these, 15 were located in the 3’ and 5’ UTRs of coding genes, 13 in the promoters of coding genes, 1 in an enhancer region encompassing TXNDC5, 18 in lncRNAs/miRNAs and their promoters, 12 in immunoglobulin regions, and 7 were determined to be technical sequencing or mapping artifacts upon manual review.

Out of the 28 candidates potentially linked to a direct gene target (3’ UTR, 5’ UTR, Promoter), 17 were also found to be mutated in mature B cells and across all multiple myeloma (MM) subgroups (including the BCL6 promoter and 5’ UTR, BCL7A 5’ UTR, BACH2 promoter and 5’ UTR, etc.). By mapping mutational signatures, we confirmed that most of these mutations are enriched in AID signatures SBS9 and SBS85, which is consistent with off-target AID activity during B cell differentiation. Therefore, we conclude that they are less likely to drive MM.

Of the remaining 11 candidates, 6 were found in known MM drivers or key genes. In particular, the ILF2 promoter on chromosome 1q21.2 was found exclusively in the MAF subgroup of patients (9/33 [27%] MAF among the 177 deep WGS cases; Fisher’s Exact p=1.0x10-7). ILF2 promoter mutations were enriched with the APOBEC signature, specifically APOBEC-induced TpC>T and TpC>G mutations at two nucleotide positions on chr1q21.2 (chr1:153,671,247 and chr1:153,671,254). The 12 ILF2 promoter mutations from these 9 patients were mostly clonal (11/12 cases) and present across all disease stages (2 MGUS, 6 SMM, and 1 MM), suggesting that these mutations are early events in MM development.

To further validate these hotspots in MM, we examined CoMMpass low-pass WGS data for evidence of these mutations, even if present in a single read. We found these hotspot mutations in 26 patients among the 769 CoMMpass samples with at least two reads at one hotspot loci. Consistent with the deep WGS data, the MAF subgroup was overrepresented in CoMMpass with an ILF2 promoter hotspot mutation (13/26; Fisher’s Exact p=4.7x10-11).

Finally, we studied the downstream effects of these promoter mutations by testing ILF2 expression in CoMMpass cases with both gene expression data and sufficient genomic coverage at the ILF2 promoter hotspots (N=589). We found that tumors with ILF2 promoter mutations had higher ILF2 expression compared to wildtype cases, independent of 1q gains (p=0.018). Interestingly, three other genes (SNAPIN, S100A4, and S100A6) within the same topologically associated domain (TAD) were also significantly overexpressed in mutant cases (q<0.1 for all genes), suggesting that these mutations are associated with changes in the TAD or activation of enhancers that regulate multiple genes within the TAD.

Conclusion

Overall, we present the first comprehensive list of significantly mutated non-coding elements in MM, particularly in its precursor conditions. This list includes newly identified hotspots associated with known MM driver genes, and a catalog of B-cell-specific mutations that reflect the differentiation from B cells to plasma cells.