Small gene, big number, many effects

In this issue of Blood, Eisfeld and colleagues present strong evidence that miR-3151 and the host gene BAALC (brain and acute leukemia, cytoplasmic) concomitantly affect the outcome of patients with cytogenetically normal acute myeloid leukemia (CN-AML) by influencing significant signaling pathways.¹ 

Over the past decade, molecular oncology research has revealed that abnormalities in both protein-coding genes (PCGs) and noncoding RNAs (ncRNAs) can be identified in tumors and that the interplay between PCGs and ncRNAs is causally involved in the initiation, progression, and metastasis of human cancers.²  MicroRNAs (miRNAs), which are among the most studied ncRNAs, are small 19- to 25-nucleotide genes involved in the regulation of PCGs and other ncRNAs, and some are involved in AML pathogenesis (see table). MiRNAs are strongly conserved among distantly related organisms (including invertebrates, vertebrates, and plants), and a large fraction are located in introns of PCGs. A plethora of studies in recent years proved that miRNAs are involved in various biologic processes, including cell-cycle regulation, differentiation, development, metabolism, neuronal patterning, and aging.³  Initially, miRNAs were identified by standard cloning methods starting with RNA size separation, while recently a large number of miRNAs were discovered using various small-sequencing platforms.⁴  Such platforms allow the identification of genes with restricted cell-specific expression as well as with a low copy number per cell. Because of the sequential numbering system for miRNAs, the higher the number of an miRNA, the more recent the identification and therefore the higher the probability to be discovered by deep sequencing. This is the case of miR-3151, identified as expressed in melanocytes⁵  and childhood acute lymphoblastic leukemia.⁶ 

What makes the study by Eisfeld et al special and significant in the context of the more than 6000 publications on miRNAs in the past 10 years? First, the type of disease analyzed: AML is a deadly cancer with median survival of less than 3 years and has had minimal advances in improvement in survival duration in the last decade. Within AML, CN-AML comprises the largest subgroup, and it has variable outcomes, whereby gene mutations and gene expression signatures segregate patients in different prognostic categories.⁷  Despite recent treatment recommendations of the international expert panel, which consider the mutational profile of selected markers (CEBPA, NPM1, and FLT3-ITD),⁸  identification of new molecular markers and, more importantly, understanding of their role as potential targets for therapy may help to refine the existing risk-adapted stratification schemes and contribute toward development of individualized treatment strategies. The Ohio State University group has a long track record of significant contribution to the study of miRNA abnormalities in AML,⁹  and therefore this group is well-positioned to translate the bench findings for the advantage of AML patients. The large number of analyzed patients (ie, 179) from a homogenous subtype of AML, the de novo CN-AML, confers statistical strength to the identified new clinical correlations.

A second source of the current paper's significance is that the authors provide one of the few examples in which a host PCG and the miRNA encoded in one of its introns cooperate on the same pathways but are also involved in independent pathways that seemingly converge on the malignant phenotype. The BAALC gene is associated with lower response rates and shorter disease-free and overall survival in CN-AML.¹⁰  Although BAALC was recently shown to block myeloid differentiation in AML,¹¹  its functional role in AML pathogenesis and mechanisms by which high expression affects clinical outcomes are not entirely understood. In the study by Eisfeld and colleagues, AML patients classified as high miR-3151 expressors and those characterized as BAALC expressors share MN1, CD200, and MEIS as differentially expressed genes. Furthermore, the miR-3151–associated coding signature expression (meaning the PCG whose expression is directly, by sequence complementarity, or indirectly, by any other mechanism, regulated by these miRNAs) contains genes involved in major signaling pathways, such as cell-cycle control and ubiquitination. The authors proved that 2 members of the latter pathway, FBXL20 (F-box and leucine-rich repeat protein 20) and USP40 (ubiquitin-specific protease 40), are direct targets of miR-3151, which is per se a new finding.

Third, probably the most significant finding of this paper is that high expression of both miR-3151 and BAALC characterize the patients with the lowest (50%) rates of complete remission (CR) and significantly shorter disease-free and overall survival. In turn, patients with high expression of only 1 of those markers had intermediate outcomes, while the low expressors for both genes had the highest CR rates. Expression of BAALC was mainly associated with achievement of CR, while that of miR-3151 affected the outcomes once CR had been achieved (disease-free survival), suggesting that the 2 genes contribute to the biology of the disease through different mechanisms. Measuring both of these markers at diagnosis may help to identify upfront which CN-AML patients have an unfavorable prognosis and may require more aggressive or investigational therapies.

Finally, this article is important in particular for the scientists who have been involved in this field for years. One dogma based on empirical observation is that the miRNAs that are well-expressed and ubiquitously expressed would be the most significant players in a signaling pathway (as in the case of miR-21). The fact that a significant role is played by an miRNA from the 3000 series identified recently by deep-sequencing experiments at a modest expression level in a quite restricted panel of cells, including hematopoietic cells, suggests that the miRNA genomic galaxy has more hidden stars that will come into play in the future. This finding has also a practical consequence—it demonstrates that the initial screening studies on patients will have improved identification of potentially significant miRNAs if done by next-generation sequencing methods.

As with all important studies, the contribution by Eisfeld et al raises many questions that can boost further investigations: What are the mechanisms that regulate the expression of the BAALC and miR-3151 genes? Are these common and related to specific transcription factors? Are FBXL20 and USP40 the most important targets for the mechanism of response to therapy in AML involving miR-3151, or are there other more significant targets? Did the miR-3151 signature correlated with miR-3151 expressor target an overlapping spectrum of downstream PCGs? Did a 4-gene signature composed of the pair of miRNA/host genes and the targets FBXL20 and USP40 separate patients more or less distinctly than did the 2-gene signature? Certainly, these questions and others will not be left without answers for long.