IRX3 Oncogene Activation in T-ALL through Enhancer Hijacking Caused By fto Intron 8 Deletions

Abstract

IRX3 is a homeobox transcription factor essential for normal embryonic development but with no detectable expression across the normal T-lineage. In contrast, aberrant expression of IRX3 is observed in ~50% of patients with T cell acute lymphoblastic leukemia (T-ALL). Prior studies have confirmed that IRX3 acts as a potent oncogene through its ability to immortalize HSPCs in myeloid culture, and induce lymphoid leukemias in vivo (Somerville et al. 2018). To date, the mechanism of transcriptional activation for IRX3 in T-ALL is unknown. We have previously identified somatically acquired variants of the noncoding genome that robustly activate oncogenes in T-ALL (Mansour et al. 2014; Rahman et al. 2017). Therefore, we investigated whether aberrations of the noncoding genome are a cause of dysregulated IRX3 expression. Two candidate cis-regulatory elements (cCREs) for IRX3 were identified by analyzing enhancer-promoter loops mapped by HiChIP from the IRX3 positive CUTLL1 T-ALL cell line. The first was situated 3' to IRX3 within FTO intron 8, and the second 5' to IRX3 within the CRNDE/IRX5 locus spanning distances of 0.22 Mb and 0.65 Mb respectively. Generation of a rank ordered gene list by comparing IRX3 negative and IRX3 positive primary T-ALLs (n=118) by RNA-Seq, showed that the expression of genes specifically within these loci were positively correlated with IRX3 mRNA levels, with FTO at rank 64, and IRX5 at rank 109 out of 19,464 total genes. This suggested the presence of a long-range regulatory axis in T-ALL encompassing FTO, IRX3, and IRX5. By analyzing publicly available datasets for genetic aberrations within these loci, we identified 12 T-ALL genomes, 11 of primary patient origin and 1 cell line (ALL-SIL) with heterozygous deletions impinging exclusively on the cCRE within FTO intron 8. No mutations were discovered at the CRNDE/IRX5 locus. Matched expression data were available for 5/12 T-ALL cases, with all 5 exhibiting supraphysiological levels of IRX3 mRNA, including 1 case with detectable monoallelic expression, consistent with the hypothesis that these deletions activate gene expression in cis. The aggregated deletions mapped a minimal region of ~155kb within FTO intron 8 that included both a CTCF and MYB binding site. We designed a ddPCR to delineate simultaneous copy number calls at these two sites in a larger cohort of primary leukemias. Heterozygous deletions were present in 6.2% (10/161) of pediatric and 1.5% (2/137) of adult T-ALL patients at diagnosis. Notably, 4/12 cases only had deletion of the CTCF site with the neighboring MYB site unaffected, thereby implicating the CTCF site mechanistically. To assess causality between the deletions and IRX3 dysregulation, we used CRISPR/Cas9 genome-editing to disrupt the CTCF and MYB sites individually or together in the FTO (wt/wt) IRX3 negative PF-382 T-ALL cell line. Deletion of CTCF and MYB sites together, and CTCF site alone, led to a ≥4-fold increase in IRX3 expression, whereas MYB site deleted clones had no detectable impact on IRX3 expression. Furthermore, chromatin looping data from the IRX3 proximal promoter by UMI-4C between T-ALL cell lines with (n=3), and without (n=3) ∆FTO intron 8 identified a significant increase in promoter contacts with the CRNDE/IRX5 locus positioned 5' to IRX3. Remarkably, this locus harbors a super-enhancer determined by rank ordering of all enhancers (rank 113/23,737 total enhancers), and an actively transcribed noncoding RNA (CRNDE) during normal T cell development. Thus, these data reveal that activation of IRX3 in T-ALL is caused by deletion of a CTCF site within FTO intron 8 and consequential long-range enhancer hijacking. We postulate that similar deletions of the noncoding genome especially those that impinge on CTCF sites, may cause dysregulated oncogene expression over vast genomic distances in both hematological and solid malignancies.