Abstract
Next generation sequencing (NGS) and single nucleotide polymorphism arrays (SNP-A) contribute to more precise identification of the spectrum of somatic mutations and karyotypic abnormalities in myeloid neoplasms. The diversity of individual defects and their combinations corresponds to the tremendous clinical heterogeneity. Identification of key driver genes remains a fundamental component to understanding the immense data generated from this technology and the contributions to leukemogenesis.
In this project, we evaluated 1200 cases of MDS and AML. Somatic mutations of AT rich interactive domain 2 (ARID2) were found in myelodysplastic syndrome (MDS), myeloproliferative neoplasms (MPN), primary acute myeloid leukemia (pAML) or secondary AML (sAML). All ARID2 mutations occurred in either frameshift (at p.S1489 and p.T1645) or nonsense (at p.E65, p.S154 and p.Q1637) configurations, consistent with loss of function. We have identified a total of 5 mutant cases, all of somatic origin. Study of clonal architecture elucidated that ARID2 mutations were ancestral events in 50% of mutant cases as defined by variant allelic frequencies. By SNP-A, a commonly deleted region on chr.12q was defined by mapping minimally affected lesions in 9 patients with MDS, MPN, sAML or pAML. Haploinsufficient expression of ARID2 was confirmed by expression array analysis in the cases with del(12q), which is compatible to the frequent incidence of heterozygous ARID2 loss-of-function mutations. Del(12q) was more frequent in high-risk phenotype with higher blast counts. In addition, significantly lower expression of ARID2 was also observed in 22 out of 183 patients with diploid 12q. Interestingly, amplification of locus was not found in any of the cases studied by SNP-A. Altogether, such lesions of defective ARID2 were pathogenic in more than 10% of cases with myeloid neoplasms.
ARID2 is encoding one of the components of SWI/SNF complex and involved in chromatin remodeling. Therefore, we stipulate that other genes which function together with ARID2 might be affected with somatic mutations or deletions. Further analyses demonstrated the presence of other somatic mutations and deletions affecting SWI/SNF complex, including ACTL6B (N=53) and SMARCD3 (N=66). While SWI/SNF complex lesions were mutually exclusive, concomitant subclonal mutations in such affected cases were commonly observed in RAS pathway genes, including K/NRAS, NF1 and PTPN11. To the contrary, ARID1B, which negatively regulates chromatin remodeling, is predominantly activated in the cohort with similar phenotype. While germline mutations of multiple genes in SWI/SNF complex are reported to be associated with Coffin-Siris syndrome, no family or past history characteristic of this congenital disorder was observed in our cohort.
Further clues into the function of ARID2 in myeloid neoplasms came from studying splicing dysfunction in U2AF1 mutant cases. Deep RNA sequencing in the cases with U2AF1 mutations (p.S43F and p.Q157P), showed a consistent loss of ARID2 exon 8 (predominantly noted in sAML). Two types (whole and partial) of exon skipping led to frameshift in the transcript creating a stop codon. Targeted RT-PCR confirmed the transcriptome sequencing results in primary bone marrow samples of the cases with U2AF1 but not in the corresponding wild-type cases. These results are compatible with the genetic finding that spliceosomal mutations were not concomitantly observed in the cases with SWI/SNF complex defects, suggesting misspliced transcript with nonsense decay consequences is enough pathogenic to preclude additional spliceosomal mutations.
To validate functional consequences of ARID2 loss, knockdown experiment using ARID2-shRNA transduction in K562 and HL60 cell lines were performed. Knockdown of ARID2 generally demonstrated cell cycle arrest in G2 phase prior to entry into the S-phase, which was partly caused by decreased expression of CDKL3 and CCND3. Hb staining with Benzidine showed impairment of erythroid differentiation in ARID2 knockdown K562, which was confirmed by cytological examination.
In sum, multiple mechanisms of defective ARID2 including somatic mutations, haploinsufficiency and phenocopy due to spliceosomal mutations can be involved in ARID2-mediated leukemogenesis. Together with the other components, novel lesions of SWI/SNF complex constitute a group of tumor suppressor genes in myeloid neoplasms.
No relevant conflicts of interest to declare.
Author notes
Asterisk with author names denotes non-ASH members.
This feature is available to Subscribers Only
Sign In or Create an Account Close Modal