Introduction APOBEC mutational signatures are one of the most important mutational signatures in newly diagnosed multiple myeloma (NDMM). APOBEC mutagenesis primarily works through two enzymes, APOBEC3A (A3A) and APOBEC3B (A3B). While these genes are expressed in most NDMM patients, their influence on APOBEC mutagenesis appear to be extremely heterogeneous, with some patients having no detectable mutations and others having high contribution (hyper-APOBEC). Methods To decipher the mechanism of APOBEC mutagenesis, we interrogated 752 low coverage long insert whole genomes, 723 whole exomes (WES) and 767 RNA sequencing (RNA-Seq) from NDMM patients enrolled in the CoMMpass study. Results APOBEC mutational activity was identified in 416/723 (57.5%) patients using WES. Of these, 41 (5.8%) had MAF/MAFB translocations. Overall, 48/723 (6.6%) patients were defined as hyper-APOBEC (13 without MAF/MAFB events). We used differential expression (DE) analysis between three groups: hyper-APOBEC with MAF/MAFB (HA_TRA: n=31), hyper-APOBEC without MAF/MAFB (HA_NORM: n=8), and non-hyper-APOBEC without MAF/MAFB (WT: n=510). 534 genes were significantly associated with hyper-APOBEC independently from MAF/MAFB events. Importantly, A3A and A3B expression were higher in hyper-APOBEC compared to other cases (logFC=2.38, logFC=1.58 respectively). Using Spearman's correlation, we identified 50 significantly correlated genes (r-squared>0.18; p<0.01) with A3B. Most of the genes belonged to the APOBEC-associated group defined by the DE and were enriched for cell cycle/proliferation activity, suggesting a link between highly proliferative disease and hyper-APOBEC. To validate this finding, we evaluated the link between these APOBEC-associated genes and hyper-APOBEC among ICGC breast cancers, a tumor known to have high prevalence of APOBEC mutagenesis. Overall, 35/50 of these genes showed similar correlation between hyper-APOBEC breasts and MM. Interestingly, most of these genes appear to be negatively controlled by E2F4 (DREAM complex), and positively controlled by E2F1, FOXM1, and MYBL2 (i.e. cell cycle genes). In line with mouse models (Roelofs et al., eLife, 2020), our data confirmed that highly proliferating tumors suppress the DREAM complex leading to induction of A3B. Overall, these data suggest a strong link between proliferation and A3B expression and mutagenesis. However, not all high proliferating tumors had high APOBEC mutagenesis. To better investigate this aspect, we compared 139 known MM genomic events to identify potential differences between the hyper-APOBEC group (HA) and the rest of the cohort (WT). HA_TRA and HA_NORM showed similarities, including a highly complex genomic profile with enrichment for 1q gain/amp (p < 0.02), 13q (p < 0.05) and 16q deletions (p<0.01). Interestingly, 5/13 (38.4%) of HA_NORM samples harbor CCND1 translocation. In terms of gene expression, both groups showed enrichment for GEP70 indicating a high-risk transcriptional profile apart from MAF/MAFB overexpression. Finally, HA_TRA and HA_NORM showed similar poor outcomes compared to WT. To further validate the association between APOBEC and complex genomic profile, we explored again the ICGC breast WGS, observing a strong correlation between homologous recombination deficiency (HRD) and hyper-APOBEC (p < 0.05). To validate the relationship between the DREAM complex and APOBEC, we pharmacologically inhibited FOXM1 (FDI6, 20uM) and E2F4/6 (HLM00074, 20uM) in 8226, U266 and MM1.S multiple myeloma cell lines and showed increased A3B levels upon 72h of E2F4/6 inhibition. Simultaneously, FOXM1 inhibition confirmed A3B downregulation. Conclusion Overall, our data support a model where, APOBEC genes are variably expressed in all NDMM, but only patients with high proliferation rate and high level of complexity acquire mutations (Figure 1).
Read full abstract