Characterization of Causal Variants at the GATA3 Loci Associated with Susceptibility to Ph-like Acute Lymphoblastic Leukemia

Acute lymphoblastic leukemia (ALL) is the most common cancer in children and there is growing evidence of inherited susceptibility to this type of leukemia. Recent genome-wide association studies have definitively identified at least 6 genomic loci with common variants that influence ALL risk, often in an age- and/or subtype-dependent fashion. In particular, we had previously reported germline intronic variants at the GATA3 locus that specifically associated with the risk of developing Philadelphia chromosome-like ALL (Ph-like ALL) and also inferior outcomes of ALL therapy (Nature Genetics 45:1494). However, the identity of the causal variant(s) that functionally alters GATA3 activity and drives the association with the susceptibility to Ph-like ALL remained unknown.

To comprehensively identify ALL risk variants at the GATA3 locus, we first performed targeted sequencing of a ~27 Kb genomic region encompassing GATA3 exons and introns in 5,008 children with ALL enrolled on Children's Oncology Group and St. Jude Children's Research Hospital frontline trials. Of 1048 variants with sufficient sequencing quality, rs3824662 showed the strongest association with susceptibility to Ph-like ALL (P=1.2x10^-8) and another 2 variants also reached statistical significance after corrected for multiple testing (P <1x10^-5). However, multivariate analysis conditioning on rs3824662 revealed no independent signals, thus pointing to a single causal variant responsible for the association signal at this locus. This linkage disequilibrium block also overlapped with a chromatin segment epigenetically active in the hematopoietic cells, as inferred from histone modification marks.

Next, we set out to systemically characterize potential functions of this regulatory DNA element and how its activity is influenced by rs3824662 genotype. The ALL risk variant A allele at rs3824662 was associated with higher enhancer activity in the reporter gene assay, and elevated levels of H3K4Me1 and H3K27Ac marks (for active chromatin) relative to the wildtype C allele in lymphoblastoid cells. ATAC-seq data from a representative panel of B-ALL cell lines confirmed a more open chromatin conformation linked to the A allele at this locus, e.g., in Ph-like ALL. By CRISPR-cas9 genome editing, we introduced the A allele at rs3824662 into the wildtype lymphoblastoid cell line GM12878. Engineered GM12878 cells with variant genotype (AA or CA) showed 3.7- and 3.8-fold increase in GATA3 expression as compared to isogenic cells with wildtype genotype, with a concomitant increase in H3K27Ac and H3K4Me1 modification at this locus, indicating that the risk allele at rs3824662 positively regulates GATA3 transcription.

Ectopic expression of GATA3 in a non-Ph-like ALL cell line (Nalm6), induced a global change in gene transcription profile that resembles the Ph-like expression signature (P=0.0004), with particularly notable upregulation of CRLF2 . Chromosomal rearrangement-mediated CRLF2 overexpression is a defining molecular abnormality in up to 50% of Ph-like ALL cases. In fact, using GATA3 ChIP-seq, we identified 5 regions upstream of CRLF2 enriched for binding to this transcription factor in ALL cell line models, with binding sites located proximal to CRLF2 rearrangement hotspots observed in patients with Ph-like ALL. Inducible overexpression of GATA3 led to upregulation of CRLF2 in a time-dependent manner; conversely, downregulation of GATA3 suppressed CRLF2 transcription, further suggesting that GATA3 may function as transcriptional regulator of CRLF2 . In mouse hematopoietic cell BaF3, GATA3 overexpression resulted in elevated phosphorylation of JAK2 and STAT5, downstream of CRLF2 . Co-expression of GATA3 and JAK2^R683G were sufficient to induce cytokine-independent growth and leukemic transformation of BaF3 cells, in a fashion similar to co-expression of CRLF2 and JAK2^R683G. Interestingly, the addition of CRLF2 ligand, TSLP, potentiated transformation of BaF7 cells expressing GATA3 and JAK2^R683G . Taken together, these data suggested that the GATA3 -mediated activation of CRLF2 - JAK signaling is a plausible mechanism by which it contributes to the pathogenesis of Ph-like ALL.

In conclusion, our results indicate that rs3824662 causally regulates a cis-acting enhancer for GATA3 transcription and influences Ph-like ALL leukemogenesis by regulating CRLF2-JAK signaling.