Splice-Transcript Isoforms in AML
Acute Myeloid Leukemia (AML) is a heterogeneous disease marked by the rapid expansion of malignant myeloid cells in the bone marrow, disrupting normal blood production. The phenotypic diversity among AML patients is driven by extensive genetic and epigenetic variability, leading to a spectrum of clinical subtypes. Each subtype correlates with specific molecular abnormalities that define distinct clinical courses and may reveal unique therapeutic opportunities.
Dysregulation of the splicing machinery is a hallmark of leukemia, in particular of AML with myelodysplasia-related changes (AML-MRC), and contributes to cancer heterogeneity and progression. This study aims to investigate the heterogeneity in splicing across AML genetic subtypes using a combination of long-read and short-read RNA sequencing.
We used Iso-Seq technology to sequence 31 AML across seven WHO2022-defined AML subtypes, creating a database of high-quality known and previously unreported transcript isoforms. Hence, we identified isoforms expressed in at least three leukemia samples and investigated their presence in healthy bone marrow and sorted myeloid progenitor subsets. Isoforms were annotated with SQANTI3 and quantified using short-read RNA-Seq data from 429 AML patients, processed with Salmon. Through differential transcript usage (DTU) analysis, we identified isoforms preferentially expressed in specific AML subtypes.
Our analyses yielded 76,981 high-quality isoforms, of which 48% were previously unreported. Thus identified isoforms featured 21,193 novel exon-exon splice junctions, yet analysis of RNA-Seq data of healthy bone marrow and sorted myeloid progenitor samples showed that 87% of these junctions are also found in healthy tissues.
Using short-read data from 429 AML samples, we performed a DTU analysis that identified 7,969 isoforms with preferential usage across AML subtypes, ranging from 151 isoforms in AML with in-frame bZIP CEBPA mutations (AML-CEBPA) to 2,105 in AML-MRC. From these preferentially used isoforms, 2,452 were previously unreported, ranging from 50 isoforms in AML-CEBPA to 669 in AML-MRC. Of all preferentially used and previously unreported isoforms, 13 were not detected in healthy bone marrow and myeloid progenitor samples.
Using long-read RNA sequencing we generated a comprehensive alternative splicing landscape of AML. With aid of this database and 429 previously sequenced AML samples, we uncovered thousands of isoforms that are preferentially used by particular AML subtypes, providing novel aetiological insights. In addition, after careful inspection, we identified various previously unreported preferentially used isoforms that are absent in healthy samples, which may have potential relevance as source for neoantigen discovery.