Comprehensive Evaluation of MicroRNA Genes and Gene Expression Using Next Generation Sequencing in a Patient with Acute Myelogenous Leukemia.

Abstract 271

MicroRNAs (miRNAs) are small non-coding RNAs that regulate gene expression and have been implicated in the pathogenesis of human cancer. Most current studies utilize array-based or quantitative reverse-transcription-polymerase chain reaction (RT-PCR) approaches to measure miRNA expression. However, these approaches do not interrogate all known (or predicted) miRNAs and are unable to detect mutations in miRNAs. Herein, we use “next generation” sequencing approaches to comprehensively assess miRNA expression and to identify genetic variants of miRNA genes in a patient with AML. This patient (UPN 933124) was a 56 year old female with FAB M1 AML. Routine cytogenetics revealed a normal 46 XY karyotype, and high-resolution comparative genomic hybridization studies revealed no somatic copy number alterations at a resolution of ∼5kb. We previously reported the sequence of genic regions in the cancer genome of this patient (Nature 456:66, 2008).

Massively-parallel sequencing of small RNAs isolated from the myeloblasts of patient 933124 was performed using the ABI SOLiD sequencing platform. As a control, we also analyzed pooled RNA isolated from CD34+ bone marrow cells of 4 healthy volunteers. In each case, RNA was size fractionated (corresponding to RNAs of 19-26 nucleotides in length) to enrich for miRNAs prior to sequencing. A total of 28 ×10⁶ sequence reads from AML 933124 and 20×10⁶ reads from the pooled normal CD34+ cells were obtained. Expression of 498 and 458 known miRNAs were detected in AML 933124 and CD34 cells, respectively. MiR-233 was the most highly expressed miRNA in both AML 933124 and CD34; remarkably, it represented 47.3% of all miRNA reads in AML 933124. To determine if extremely high miR-223 expression is a consistent feature of AML, we performed real time RT-PCR for miR-223 in an additional 23 AML samples and in CD34+ cells from 4 healthy donors. Of note, to avoid underestimating miR-233 expression, significant dilution of the RNA samples was required to ensure that the RT-PCR assay for miR-223 was in the linear range. Compared with normal CD34+ cells, miR-223 expression in AML 933124 was increased 6.8-fold. However, increased miR-223 expression was not a common feature of AML, with only 2 of 23 AML samples (both therapy-related AML) showing increased miR-223 expression. A large number of sequence reads mapped to unannotated regions of the genome. Using an in-house program developed to view predicted RNA structure, more than 10 novel putative miRNAs were identified, some of which were differentially expressed in AML 933124 compared with normal CD34+ cells.

To detect genetic variants of miRNA genes, we designed 454-amplification and sequencing primers to sequence all 695 miRNA genes that were in the Sanger miR database (version 12.0). We sequenced approximately 200 bp flanking the mature miRNA (total ∼400 bp per miRNA gene) to ensure that mutations affecting the primary miRNA were detected. To differentiate germline polymorphisms from somatic mutations, genomic DNA from both leukemic blasts and a skin biopsy from this patient were sequenced. Average sequence coverage depth was 52.2X, and 95% of miRNAs had at least one supporting read. Thirteen single nucleotide variants and 3 Indels in miRNA genes were detected. All were present in the skin DNA sample, suggesting that they represent germline polymorphisms. Finally, we analyzed the previously generated whole genome sequence for this AML genome for genetic variants in the 3'-untranslated region of all coding genes. A single somatic mutation in the 3'-UTR of TNFAIP2 was detected. This mutation is predicted to disrupt the binding of several expressed miRNAs. However, no recurrent mutations in the 3'-UTR of TNFAIP2 were detected in an additional 180 patients with AML. Thus, the contribution of this somatic mutation to the pathogenesis of AML is unclear.

These data demonstrate the feasibility of ‘next generation' sequencing technologies to identify novel miRNAs, accurately measure mature miRNA expression, and identify both somatic and germline genetic variants of miRNA genes in primary cancer. Using this platform, studies are underway to comprehensively characterize miRNAs in additional human AML samples.