Genome-Wide Aberrant Splicing in Patients with Acute Myeloid Leukemia (AML) Indetifies Potential Novel Targets

AML is characterized by specific chromosomal translocations along with a number of other recurrent mutations. However, it is also becoming clear that AML cells have frequent epigenetic abnormalities that also contribute to transformation. Alternative splicing (AS) is an epigenetic event that is used in normal cells to generate protein diversity from single genes. Using genome-wide screening approaches, we have shown that alternative splicing of numerous genes is aberrant in most cases of AML, and that several common myeloid cell surface proteins may have novel and potentially antigenic epitopes as a result.

Using Affymetrix Human Exon 10ST arrays we performed a genome-wide analysis of AS in bone marrow (BM) aspirates from two cohorts (training cohort including 40 patients and test cohort including 41 patients) of patients with AML (total 81 patients) and CD34⁺ BM progenitor cells from 8 normal donors (NDs). After principle component analysis, 6 patients were removed from the test group, because they were outliers. In final array analysis for the test group we analysed 35 patients. Aberrant splicing of several selected genes was confirmed through cloning and sequencing analysis followed evaluating expression patterns of cloned novel splice variant transcripts in 193 patients compared to NDs. To evaluate the clinical significance of AS in AML we determined if there were any correlations between novel aberrant splice variant expression and clinical features of patients.

We found that AS is a common event in AML involving many genes in any given patient. Overall, we identified ∼550 genes that were significantly spliced in patients compared to normal CD34⁺ cells. Three genes encoding myeloid cell surface proteins, NOTCH2, CD13, and FLT3, were selected for further study. By cloning and sequencing analysis we found that novel splice variants of the selected genes are the results of either exon skipping, usage of cryptic 5' or 3' splice sites on exons and/or partial retention of an intron. We also found that these splice variants encode proteins, as assessed by expressing splice variant transcripts in the HEK293 cell line. To identify the most frequently expressed variants of NOTCH2, FLT3 and CD13 in AML, we performed expression profiling of these transcripts and their variants in an independent validation group of 193 AML patients. These analyses identified NOTCH2-Va (74%), FLT3-Va (50%) and CD13-Va (60%) as the most frequently expressed variants in AML. Expression of these novel splice variants, other than NOTCH2-Vb, was undetectable in BM CD34⁺ cells, normal monocytes or neutrophils. To examine whether expression of novel variants are associated with disease evolution, we performed longitudinal monitoring of expression levels of NOTCH2-FL, FLT3-FL, and CD13-FL and their splice variants in paired samples of 13 patients taken over the course of the disease. Our study suggests the “splicing profile” normalizes in remission and recurs when patients relapse. Correlation analysis among NOTCH2, CD13 variants and FLT3 splice variants, and clinical features of patients with AML, showed that splice variant NOTCH2-Va expression is significantly negatively associated with overall survival (OS) of patients (P=0.008). Our study also showed that deregulated expression of NOTCH2-Va identifies two subgroups of patients within the intermediate-risk profile patients. We also observed significantly inferior OS of the patients that overexpressed NOTCH2-Va within this group (P=0.007). Thus, NOTCH2-Va expression can be used to stratify patients with cytogenetically determined intermediate-risk profiles. Overall, our study demonstrates that genome-wide study of AS can be used to discover new prognostic markers, and potentially creates new surface antigenic epitopes that could be targeted in AML.

No relevant conflicts of interest to declare.