Integrated Somatic Mutation and DNA Methylation Analysis Reveal Genomic and Epigenomic Co-Evolution In Follicular Lymphomas

Follicular lymphoma (FL) is a B-cell lymphoma whose cytogenetic hallmark is the translocation t(14;18)(q32;q21) which juxtaposes the BCL2 oncogene to the immunoglobulin heavy chain locus (IGHV). FLs maintain key features of normal germinal center reactions, such as ongoing somatic hypermutation (SHM) of IGHV genes and selection for a functional B-cell receptor. SHM is mediated by activation-induced cytidine deaminase (AID) leading to single nucleotide exchange in IGHV genes and to a much lesser extent in non-IG genes. It was our objective to investigate the clonal evolution of t(14,18) FL from primary to relapse tumors simultaneously on several genetic and epigenetic levels.

We studied paired primary and relapsed tumors from 33 patients with t(14;18)-positive FL (76 samples: 25 pairs; 6 trios and 2 quadruples). In a core set of 19 patients we performed Sanger and next generation sequencing of the clonal V_HD_HJ_H rearrangements of the IGHV locus.

We performed deep sequencing of 9 genes (BCL2, BCL6, MYC, RHOH, PAX5, IRF4, C2TA, REL and PIM1) targeted by aberrant SHM (aSHM) in lymphoma for 69 FL samples including the complete core set. We furthermore analyzed mutations in the coding regions of 10 candidate driver genes of lymphomagenesis (BCL2, MLL2, CREBBP, TNFRSF14, EZH2, EP300, MEF2B, BCL6, MYC, TP53) by deep sequencing. In addition to genetic analyses, evolutionary patterns of DNA-methylation were addressed by Illumina 27k arrays.

We found strong evidence for ongoing selection against replacement mutations in the IGHV genes both in complementarity determining regions and framework regions, consistent with ongoing dependence of FL on a functional B-cell receptor and stimulation by antigens.

Using mean normalized Hamming distance as a quantitative measure for the evolutionary divergence of paired samples (IGHV-divergence) and analyzing phylogenetic trees we classified the patterns of evolution into 3 categories: “No Evolution” (shared IGHV sequences in primary and relapse tumors), “Sequential Evolution” (relapse sequences emerge out of primary ones), “Divergent Evolution” (sequences of primary and relapse sample appear disjoint).

We observed a mutation frequency of 62.0 per 100 kb in aSHM targets. These mutations were strongly enriched at the WRCY/RGYW target motifs characteristic for the SHM/AID machinery (OR=3.4; p<0.001). The most frequently altered locus was BCL2 affected in 68 of 69 samples (with 315.3 mutations per 100 kb) likely due to the recombination of the translocated allele into the IGH locus.

We detected 197 single nucleotide variants (SNV) affecting the candidate driver genes (4.4 mutations per 100 kb) of which 145 were protein-changing. The genes most frequently affected were CREBBP (52 SNVs in 43/69 samples) and MLL2 (54 in 39/69). High incidence of CREBBP mutations and the high mutated allele frequencies of these mutations suggest that CREBBP deregulation is an early event in FL lymphomagenesis.

We defined the aSHM-divergence as the proportion of observed aSHM mutations not present in both samples and found a significant correlation between the IGHV-divergence and the aSHM-divergence in the non-IG aSHM targets (r=0.724 [0.40-0.88]). We calculated a measure for divergence in DNA-methylation within patients by the proportion of all paired CpGs showing a methylation difference of at least 25%. DNA-methylation-divergence was found to correlate with IGHV-divergence (r=0.516 [0.24-0.72]) indicating that evolutionary divergence also affects the level of DNA-methylation.

Our observations demonstrate correlated evolutionary changes on all genomic and epigenomic levels in FL. This process most likely originates from an ongoing germinal center reaction with a functional BCR driving FL cells, sustaining a continuously active AID-machinery leading to addition of IGHV-mutations and to aberrant mutations in non-IG aSHM-target sites. The drifts in DNA-methylation patterns might be a consequence of additional mutations found in histone modifying genes such as CREBBP. The individual pattern of tumor evolution is likely to impact prognosis and clinical decision making which needs to be investigated in larger series.

(Acknowledgment: BMBF/PTJ 0315452)

No relevant conflicts of interest to declare.