Recurrent Mutations in EGR2 Direct Specific Epigenetic Reconfiguration in Chronic Lymphocytic Leukemia

Epigenetic alterations are universal in cancer and are important in establishing the malignant phenotype. Dissection of the factors that shape the tumor-specific epigenome may reveal insight into key aspects of tumorigenesis and therapeutic resistance. In chronic lymphocytic leukemia (CLL), we have previously found that broad changes in epigenetic patterns co-occur with the evolution of genetic alterations. We have also uncovered that aberrant patterning of DNA methylation in CLL involves excessive activity of a defined group of transcription factors (TFs), including the early growth response (EGR) TF family. Recent work has further revealed that recurrent mutations in EGR2 are associated with exceptionally poor clinical outcomes in CLL. The basis for the adverse association of EGR2 mutations in CLL is unclear.

To explore the role of EGR2 mutations in CLL, we initially performed genome-wide DNA methylation analysis using Illumina arrays on CLL patients harboring EGR2 mutations (n=27) compared to EGR wild-type cases (n=265). We found that the three most common recurrent mutations, occurring at amino acid positions E356K, H384N and D411H within the DNA binding domain, are each associated with an exclusive subset of tumor-specific hypomethylated CpG sites. A search for TF sequence motifs at these loci revealed a strong enrichment of novel derivative EGR2 motifs that differ only marginally (usually by a single nucleotide) from the canonical EGR2 recognition sequence. Each recurrent mutation led to specific enrichment of a different derivative EGR2 motif. Furthermore, the canonical (wild-type) recognition sequence was not enriched, suggesting that mutations re-localize binding activity to derivate sequence motifs rather than simply altering binding affinity. Luciferase enhancer, proximity ligation and electrophoretic mobility shift assays confirmed that each EGR2 mutant protein specifically binds and enhances transcriptional activity only when the matched EGR2 derivative recognition motif is present. These results establish that derivative motif sequences may function as novel cryptic enhancers in the presence of the cognate EGR2 mutant TF.

We performed multiomics profiling (DNA methylome, ATAC-seq, ChIP-seq and RNA-seq) to examine the nature of the epigenetic reconfiguration and the phenotypic impact of individual EGR2 mutations. Whole genome bisulfite sequencing of E356K- and H384N-mutated CLL samples (n=4 each) was used to reveal the full complement of recurrent differentially methylated regions (DMRs) across the genome, and recapitulated the mutually-exclusive pattern of DMRs between mutations. Overlaying DMRs with data from ChIP-seq and ATAC-seq experiments in the same samples revealed the nature of EGR2 mutation-specific chromatin reconfiguration to be remarkably mutation-specific. For E356K, hypomethylated DMRs are often associated with foci of accessible chromatin, EGR2 binding, and flanked by gains of H3K4me1 and H3K27ac, indicative of the acquisition of active enhancer function. Conversely, H384N mutations generated fewer DMRs and mainly directed the deposition of H3K4me1 only, indicative of gain of poised enhancers at these loci. RNA-sequencing analyses revealed that a subset of epigenetically reconfigured regions was associated with mutation-specific altered gene expression, and differences were virtually always associated with proximal gene activation. E356K and H384N mutations displayed highly differential gene expression patterns, with E356K exhibiting a greater impact on gene expression. Integrated analyses indicated that E356K mutations may specifically involve activated Notch signaling, revealed by the aberrant activation of Notch target genes and the mutual exclusivity of NOTCH1 mutations, further highlighted by enriched co-mutation of NOTCH1 in H384N-mutated CLL.

Together these findings provide an exceptional example of the precise role that a singular TF may play in programming the epigenetic landscape. As there are no known TFs that naturally bind derivative EGR2 motifs, these mutant proteins provide insight into aberrant enhancer generation and the phenotypic impact of (re)directed TF binding in a human disease setting. Although these recurrent mutations are presently only known in CLL, these findings provide insight into the mechanisms that may surround other gain-of-function TF activity in various malignancies.