Introduction: Familial eosinophilia (FE) is a rare inherited condition characterized by lifelong idiopathic hypereosinophilia (absolute eosinophil count >1.5 x 109/L) in multiple family members with or without clinical manifestations. Although the gene underlying FE in one large kindred was mapped to chromosome 5q31-33 using genomewide linkage analysis, targeted sequencing of genes in this region, including interleukin 5 (IL5), was unrevealing. Despite more recent studies demonstrating selective dysregulation of IL5 transcription, the underlying genetic abnormality in FE has remained elusive to date.
Methods: Whole genome sequencing (WGS) was performed using peripheral blood mononuclear cell (PBMC) DNA from 5 affected and 1 unaffected family members of the above-described kindred followed by targeted Sanger sequencing of candidate variants. After identifying a single nucleotide variant common to all affected family members in a noncoding region upstream of IL5, RNA sequencing (RNAseq), including long-read PacBio sequencing, was performed to examine IL5 gene expression and to identify IL5 isoforms. To predict potential transcription factor binding sites, we applied motif analyses around the variant and performed chromatin immunoprecipitation (ChIP)-PCR to assess chromatin occupancy of the predicted transcriptional regulators (HMGA1, FOXO3) and of RNA Polymerase II (Poll II). The variant of interest was introduced into a B-lymphoblast cell line (NCI-BL1184) via CRISPR/Cas9, after which clones derived from single cells were compared with respect to IL5 transcription (via quantitative PCR) and chromatin occupancy of HMGA1, FOXO3, and Pol II (by ChIP-PCR). Chromatin marks associated with active gene expression at enhancers (H3K4me1, H3K27Ac) and promoters (H3K4me3) were assessed by ChIP-PCR. Integration of genomewide ChIP-Seq, ATAC-Seq and gene expression analyses are underway to further assess the role of the variant as the primary driver of aberrant IL5 expression.
Results: WGS revealed a novel single nucleotide variant in all 5 affected family members located 13.19 kb from the transcription start site (TSS) of the IL5 gene and 250 bp from the TSS of the RAD50 gene. Targeted Sanger sequencing confirmed the presence of this variant in 4/4 affected and 0/4 unaffected family members. Bulk RNA-seq revealed abnormally long transcripts spanning the entire intergenic region between the variant and the IL5 TSS in affected PBMCs, while long-read RNA-seq identified novel full-length IL5 isoforms, including one containing all 5 exons of the IL5 gene. By contrast, RAD50 transcription was unaffected by the variant. In silico motif analysis predicted that the variant generates potential binding sites for transcriptional regulators, including the High Mobility Group A1 (HMGA1) chromatin regulator and the forkhead transcription factor, FOXO3. ChIP-PCR confirmed robust chromatin binding of both HMGA1 and FOXO3 at the variant site in PBMCs from affected family members, but not in PBMCs lacking the variant. Introduction of the single nucleotide variant into B-lymphoblast cells via CRISPR/Cas9 recapitulates the findings in affected family members, with up-regulation of IL5 expression and protein levels and no effects on IL4, IL13, or RAD50 gene expression. In edited lymphoblasts with the variant, HMGA1 chromatin binding and that of H3K4me3/H3K27Ac are enriched in the 200 bp region encompassing the variant, although FOXO3 occupancy was not enriched in this setting nor were there changes in occupancy of control histone H3 in that region.
Conclusions: We describe a previously unknown mechanism of human disease: a non-coding single nucleotide germline variant that results in a de novo enhancer and up-regulation of IL5 production in a family with autosomal dominant hypereosinophilia. Orthogonal methods indicate that this variant enables HMGA1 to bind DNA, recruit active histone marks, and induce IL5 expression from a de novo TSS. This finding was validated by CRISPR/Cas9-mediated introduction of the variant into a B cell lymphoblast cell line that does not normally produce IL-5. This mechanism highlights the importance of assessing non-coding regions in the study of Mendelian disorders and may be relevant to other human diseases.
No relevant conflicts of interest to declare.
Asterisk with author names denotes non-ASH members.