Polymorphisms in microRNA target sites modulate risk of lymphoblastic and myeloid leukemias and affect microRNA binding

MicroRNAs (miRNAs) are a class of small (~22 nucleotide) nocoding RNAs that are potent regulators of gene expression in animals and plants. In animals miRNAs bind to target sequences (usually located in the 3′ untranslated region [3′UTR]) in messenger RNAs (mRNAs) and act by negatively regulating gene expression. This binding requires complementarity between the nucleotides 2-8 of miRNA (so called “seed” region) and the target mRNA [1]. To date more than 2500 mature human miRNAs have been identified [2] and they are predicted to regulate over 60% of human protein-coding genes [3]. This regulatory network can be very complex as one miRNA may potentially regulate several mRNAs, and a given mRNA may possess in its sequence binding sites for several miRNAs.

Since miRNAs control a wide variety of biological processes, including proliferation, apoptosis and differentiation, dysfunctions of the miRNA regulatory network may contribute to tumorigenesis. MiRNAs can act as both oncogenes and tumor suppressors. Examples of oncogenic miRNAs that are amplified or overexpressed in cancers include miR-17-92 cluster, miR-21, miR-155 and miR-372/373, while tumor suppressor miRNAs commonly deleted or with reduced expression in cancers are represented by miR-15a and miR-16-1, miR-34 family, let-7 family and miR-29 [4]. MiRNAs play a crucial role in normal hematopoiesis by controlling the differentiation of hematopoietic stem cells into different types of mature blood cells, while deregulation of miRNA networks has been linked to hematological malignancies [5]. Aberrant miRNA expression profiles have been observed in leukemias and lymphomas, and for several miRNAs there is experimental evidence for their functional involvement in leukemogenesis [6, 7]. Specific miRNA expression signatures can accurately discriminate different leukemia subtypes and are often of great prognostic relevance [8‐12]. Changes in miRNA expression may result from genomic and epigenetic alterations or the impairment of miRNA biogenesis pathway [13]. In addition, polymorphisms in miRNA genes or miRNA target sites (miRSNPs) can modify miRNA action. While polymorphisms in miRNA genes are relatively rare, SNPs in miRNA-binding sites in target genes are more frequent. Several studies have shown that SNPs in miRNA target sites enhance or weaken the interaction between miRNA and its target transcripts and are associated with cancers and other diseases [14, 15]. In leukemia, however, so far only one study associated a SNP in the 3′UTR of the NPM1 gene with adverse outcome in acute myeloid leukemia [16].

Considering that miRNAs have been shown to play an essential role in leukemogenesis and that SNPs in miRNA-binding sites in target genes have been associated with various cancers, in this study we aimed at identifying miRSNPs associated with leukemia risk and assessing the impact of these miRSNPs on miRNA binding to target transcripts.

In this study we show that polymorphisms in microRNA-binding sites (miRSNPs) modulate leukemia risk and influence binding of miRNAs to target transcripts. To our knowledge this is only the second study reporting a relevance of miRSNPs in leukemia. Recently, Cheng et al. [16] identified a SNP in the 3′UTR of the NPM1 gene that, although present with similar frequency in the control group, was associated with adverse outcome and shorter survival in patients with AML. They further showed that this SNP created an illegitimate binding site for miR-337-5p, which reduced levels of NPM1 mRNA and protein. MiRSNPs in PLA2G2A, IL-16 and NOD2 were also studied in acute leukemia but no significant differences in the genotype frequencies between leukemia patients and control group were detected [23].

Our study identified five miRSNPs associated with risk of different leukemia types, which is in line with findings in other tumors showing that polymorphisms in miRNA-binding sites may predispose to cancer [21, 24‐26]. Moreover, we observed a significant trend for an increasing ALL and CML risk with the growing number of risk genotypes, indicative of a possible additive effect of the identified miRSNPs. This finding suggests that a panel of miRNA-binding site polymorphisms could be of clinical utility as markers of leukemia risk. To verify this possibility, miRSNPs reported in this study should be tested in prospective studies on independent patient groups.

In chronic myeloid leukemia SNPs in ARHGAP26 and IRF8 showed significant association with an increased risk of CML, however only for the uncorrected p-value. ARHGAP26 (GRAF) belongs to a family of Rho GTPase activating proteins and is a tumor suppressor acting by negatively regulating RhoA, a small GTP-binding protein with a growth-promoting effect in RAS-mediated malignant transformation [27, 28]. Abnormal methylation of the ARHGAP26 promoter and downregulation of the ARHGAP26 mRNA was observed in acute myeloid leukemia and myelodysplastic syndrome [29, 30]. In our study, the C allele of ARHGAP26 _rs187729 T > C, associated with an increased risk of CML, created a new binding site for miR-18a-3p which decreased the protein expression by 34%. MiR-18a is a component of the oncogenic miR-17-92 cluster which is often amplified in aggressive B-cell lymphomas [31] and increased expression of miR-18a was correlated with a shorter overall survival in diffuse large B-cell lymphoma patients [32]. Our results demonstrate that the functional polymorphism in the 3′UTR of the tumor suppressor ARHGAP26 creates an illegitimate binding site for miR-18a-3p, which may affect protein levels and contribute to an increased risk of CML.

A significant association with increased CML risk was also observed for a SNP in IRF8. IRF8 (ICSBP) is a transcription factor that regulates expression of genes stimulated by interferons and is essential for the differentiation of myeloid, dendritic and B-lymphoid lineages [33]. Its tumor suppressor activity is supported by mice with a null mutation of IRF8 developing a CML-like syndrome [34] and lack of IRF8 expression in patients with chronic and acute myeloid leukemia [35]. The mechanism for the role of IRF8 downregulation in the pathogenesis of CML is the regulation by IRF8 of several apoptosis-related genes [36, 37]. We did not observe any effect of the IRF8 3′UTR on the protein expression in the presence of miR-330-3p, irrespective of the allele. However, it is possible that this SNP affects binding of another miRNA or influences other regulatory functions of the 3′UTR. The potential functional significance of IRF8 _rs10514611 C > T remains to be elucidated.

In pediatric acute lymphoblastic leukemia we identified three miRSNPs modulating the ALL risk, in ETV6, PML and TLX1. The tumor suppressor ETV6 (TEL) is a transcription factor (repressor of translation) with a crucial role in the embryonic development and hematopoietic regulation [38, 39]. Translocations involving the ETV6 locus (12p13) are a frequent event in leukemia and myelodysplastic syndrome. They result in fusions with numerous partners and contribute to leukemogenesis by several pathogenic mechanisms [40]. In our study, the C allele of ETV6 _rs1573613 T > C, associated with an increased risk of ALL, weakened binding of miR-34c-5p and miR-449b-5p, which resulted in higher protein levels (respectively 17% and 33% higher than for the T allele). The result is in accordance with the effect predicted by the in silico analysis, however it does not fit in the model of ETV6 as tumor suppressor. A polymorphism associated with an increased leukemia risk would be expected to decrease rather than increase expression of a protein with a tumor suppressor function. Two cases of ETV6 amplification have been described: one in B lymphoblastic leukemia [41] and the other in myelodysplastic syndrome [42]. Also in our study one patient with common-ALL had an additional chromosome 12 in her blast cell kariotype, and this girl was also a homozygote for the C allele of ETV6 _rs1573613 T > C. It is possible then that ETV6 can play a dual role of both a tumor suppressor gene and an oncogene, although a mechanism underlying the oncogenic activity of ETV6 remains to be revealed. Among the known rearrangements involving ETV6 there are a few in which the 3′UTR of ETV6 is preserved. For example the MN1-ETV6 fusion protein acts as a transcriptional activator whereas ETV6 is a repressor, and the PAX5-ETV6 fusion protein is an aberrant transcription factor affecting both the PAX5 and the ETV6 pathways [40]. Increased expression of those fusion proteins resulting from the weaker binding of miR-34c-5p and miR-449b-5p to the 3′UTR with the C allele of ETV6 _rs1573613 T > C could intensify their aberrant action. It is also feasible that the miRSNP in ETV6 could act in trans. Recently a concept of “competing endogenous RNAs” (ceRNA) has been developed which assumes that messenger RNAs, transcribed pseudogenes and long non-coding RNAs compete for miRNA binding, intertwined in a large regulatory network [43]. Thus, presence of the C allele would release the miRNAs otherwise bound by the wild-type 3′UTR and increase the pool of miR-34c-5p and miR-449b-5p available for other targets. Indeed, members of the miR-34 family have been reported to be overexpressed in childhood ALL [44, 45].

A significant association with elevated ALL risk was also observed for a SNP in TLX1. TLX1 (HOX11) is a transcription factor that plays a crucial role in embryonic development and in the genesis of the spleen [46]. Translocations involving the TLX1 locus (10q24) and its increased expression occur in a significant proportion of T-ALL, supporting the oncogenic role of TLX1[47] and suggesting that TLX1 expression in adult tissues is tightly controlled. We did not observe any effect of the TLX1 3′UTR on the protein expression in the presence of miR-492 for either allelic variant. Hence, the potential functional role of TLX1 _rs2742038 C > T remains to be revealed.

The A allele of PML _rs9479 G > A was associated with reduced risk of both pediatric ALL and acute myeloid leukemia in adults. PML is a transcription factor and tumor suppressor that controls cell growth and apoptosis [48]. Translocation involving the PML locus (15q22) resulting in the expression of a fusion protein PML-RARα is found in the majority of acute promyelocytic leukemia cases [49]. Moreover, a partial or complete loss of the PML protein expression has been observed in several solid tumors [48], highlighting its role in cancerogenesis. In this study the A allele of PML _rs9479 G > A enhanced binding of miR-510-5p resulting in 19% decrease in the protein expression as compared to the G allele. However, lower expression caused by the A allele, which had a protective effect in our study, does not fit in the model of PML as a tumor suppressor. Increased expression of PML was observed in tumor cells of Hodgkin lymphoma [50] and hepatocellular carcinoma [51, 52] suggesting an oncogenic role of PML but its functional significance is unknown. It is more plausible that the protective role of the A allele could be attributed to sequestering miR-510-5p from its other targets. Little is known about miR-510-5p function and its validated target genes but its oncogenic role was shown in breast cancer where overexpression of miR-510 increased tumor growth in vivo [53].

We are aware of some limitations of our study. The groups of leukemia patients were small, so the results should be treated as preliminary and need to be replicated in larger cohorts. Also, although we restricted our analysis to the miRNAs that are expressed in bone marrow, white blood cells or leukemias and lymphomas, the actual interaction of the analyzed miRNA-mRNA pairs in leukemic cells should be confirmed. Nonetheless, our study demonstrates that microRNA-binding site polymorphisms influence leukemia risk by interfering with the miRNA-mediated regulation of gene expression and underscores the significance of the variability in the 3′UTRs in leukemia.