Mutations in the miR-142 gene are not common in myeloproliferative neoplasms

Recent data indicate that MIR142 is the most frequently mutated miRNA gene and one of the most frequently mutated noncoding elements in all cancers, with mutations occurring predominantly in blood cancers, especially diffuse large B-cell lymphoma (DLBCL) and follicular lymphoma. Functional analyses show that the MIR142 alterations have profound consequences for lympho- and myelopoiesis. Furthermore, one of the targets downregulated by miR-142-5p is CD274, which encodes PD-L1 that is elevated in many cancer types, including myeloproliferative neoplasms (MPNs). To extend knowledge about the occurrence of MIR142 mutations, we sequenced the gene in a large panel of MPNs [~ 700 samples, including polycythemia vera, essential thrombocythemia, primary myelofibrosis (PMF), and chronic myeloid leukemia], neoplasm types in which such mutations have never been tested, and in panels of acute myeloid leukemia (AML), and chronic lymphocytic leukemia (CLL). We identified 3 mutations (one in a PMF sample and two others in one CLL sample), indicating that MIR142 mutations are rare in MPNs. In summary, mutations in MIR142 are rare in MPNs; however, in specific subtypes, such as PMF, their frequency may be comparable to that observed in CLL or AML.

Our recent cumulative study showed that although these mutations are distributed over the entire MIR142 gene (i.e., the sequence coding for the stem-loop structure of the pre-miR-142 precursor, including flanks), they mostly cluster around the miR-142-3p seed 4 and are strongly enriched in hematologic malignancies, which suggests that the mutations are functionally deleterious. Functional analysis of specific mutations (n.55A > G and n.58G > C) located in the miR-142-3p seed revealed that the mutations affect the miRNAs generated from both arms, resulting in loss of the ability of miR-142-3p to recognize/downregulate its targets and in lowering the level of miR-142-5p. More specifically, it was shown that by releasing repression of ASH1L (target of miR-142-3p), MIR142 mutations cause sustained HOXA9/A10 expression and alteration in the differentiation of hematopoietic progenitors (myeloid cell lines expansion, impaired erythropoiesis, and T cell lymphopenia), ultimately contributing to leukemic transformation 5 . It was also shown in the mouse model that mutations in MIR142 synergize with mutations in Idh2 to initiate AML 5,6 . It may be expected that deleterious MIR142 mutations also affect other well-documented hematologic targets of miR-142-3p or miR-142-5p, including RAC1 (Rac Family Small GTPase 1) 7 , PROM1 encoding CD133 antigen 8,9 , TNFRSF13C encoding B cell-activating factor receptor (BAFF-R) 10 , SOCS1 (Suppressor of Cytokine Signaling 1) 11,12 , PTEN (Phosphatase And Tensin Homolog) 13,14 , IL6 (Interleukin 6) 15 and CD274 encoding programmed death-ligand 1 (PD-L1) 14,16,17 which is an important immune checkpoint molecule that is elevated in many cancers, including different types of MPNs 18,19 . In light of results to date, miR-142 has emerged as an antitumor immunity-regulating factor important in different types of blood cancers, mostly lymphoid neoplasms. www.nature.com/scientificreports/ In this study, to extend and complement knowledge about the occurrence and/or frequency of MIR142 mutations in hematologic cancers, we sequenced MIR142 in large panels of MPNs [including polycythemia vera (PV), essential thrombocythemia (ET), primary myelofibrosis (PMF), and chronic myeloid leukemia (CML)], blood disorders for which such mutations have never been tested, and in AML and CLL, for which the mutations have been reported. Additional molecular analyses. The analysis of the JAK2V617F mutation and BCR-ABL transcript was performed using real-time quantitative allele-specific RQ-PCR, and multiplex RT-PCR, respectively, as described before 20,21 . The mutations in CALR exon 9, MPL exon 10, SRSF2 exon 1, U2AF1 exons 2 and 6, IDH1 exon 4, IDH2 exon 4, and ASXL1 exon 13 were screened with the use of the established high-resolution melting (HRM) assays and verified by Sanger sequencing as described before [22][23][24][25][26][27][28] .

Methods
To verify whether two mutations found in the CLL-2423 sample are located in one or two different alleles, we sequenced individual colonies obtained from the sample. Briefly, MIR142 was amplified by PCR with the following primers MIR142_F 5'-TAA CTA CAG CGG CCG CAT CTC CGA AGC CCA CAG TAC , MIR142_R 5'-TCC ACT ACG GAA TTC CGG ACA GAC AGA CAG TGCAG introducing the NotI and EcoRI restriction enzyme sites, respectively; PCR product was digested, gel purified, ligated into the pCDH-CMV-MCS-EF1-copGFP-T2A-Puro expression plasmid (CD513B-1, System Biosciences, Mountain View, CA, USA), and propagated in MAX Efficiency™ E. coli Competent Cells (Invitrogen, Carlsbad, CA, USA). Individual colonies (N = 10) after MIR142 PCR amplification were sequenced using the Sanger method as described above. The sequencing revealed 6 clones with both identified mutations and 4 clones with the wild-type sequence.

Results and discussion
To determine the occurrence of mutations in MIR142 in MPNs, we analyzed a collection of 672 samples consisting of ET (n = 321), PV (n = 174), CML (n = 107), and PMF (n = 70). Additionally, we included CLL (n = 210) and AML (n = 47) samples, diseases for which the mutations in MIR142 have been reported 4,29-32 . Screening was performed by Sanger sequencing of a PCR-amplified DNA fragment encompassing the MIR142 gene (defined as described above). Altogether, we identified only 3 mutations: one mutation [chr17:58331257G>T[-]; n.62C>A] not reported before was identified in a PMF sample (PMF_19 patient; 1/70); two other mutations [chr17:58331286A>G[-], n.33T>C and chr17:58331260A>G[-], n.59T>C] were identified in one CLL sample (CLL_2423 patient; 1/210). The n.62C>A mutation is located in the miR-142-3p post-seed region in a symmetric one-nucleotide-bulge structure ( Fig. 1A and C). Although the mutation does not induce any apparent change in the predicted secondary structure of the miRNA precursor, it may influence miRNA:target interactions as well as the stability of the miRNA itself 33,34 . Both mutations identified in the CLL sample are located in two established hotspots (n.33T>C, n.59T>C) and were previously described in CLL 29 . Both mutations are in the mature miRNA duplex, specifically in the seed region of miR-142-3p and in the post-seed region of miR-142-5p (Fig. 1A). Cloning of the PCR product and subsequent sequencing of individual clones revealed that both mutations are located in the same allele (6 out of 10 sequenced clones). TargetScan analysis predicted that the n.59T>C mutation severely affects miR-142-3p target recognition, disabling recognition of 55% (139/250) of its targets and creating 79 new targets. Whereas, analysis of the n.33T>C mutation showed that it substantially affects the structure of the miRNA precursor (Fig. 1B)  www.nature.com/scientificreports/ and miRNA release. All identified mutations have increased functional weight scores resulting from a location in crucial miRNA precursor elements (based on miRMut criteria defined before) 4,35,36 . Briefly, the PMF_19 patient, a male aged 57 years old, was diagnosed with PMF grade 3 (MF3), with a bone marrow blast content of 3.5% and intermediate 2 risk of unfavorable disease outcome according to the Dynamic International Prognostic Scoring System. The patient had a normal karyotype, was negative for the BCR-ABL transcript, and was triple-negative for driver mutations (CALR exon 9, JAK2 exon 12 and 14, and MPL exon 10), which is characteristic of up to 10% of PMF cases, usually with a less favorable prognosis 39 . Additionally, molecular analysis of ASXL1 and U2AF1 genes revealed c.1904_1927dup (p. Glu635_Gly642dup) in the former and c.470A>G (p.Gln157Arg) in the latter, which are considered unfavorable prognostic factors 40 . These additional, considered non-driver mutations were shown to contribute to disease progression, and ineffective hematopoiesis, and are taken into account in the genetically inspired prognostic scoring system (GIPSS) of PMF 41,42 . Co-occurrence of ASXL1 and U2AF1 mutations with mutations in MIR142 has also been noted in AML 30 .
The CLL_2423 patient, a male aged 62 years old, was diagnosed with B cell CLL (IV stage according to Rai classification). No chromosome alteration was observed by nuclear FISH (including the chr17p/TP53 gene). After 6 cycles of immunochemotherapy (Rituximab, cyclophosphamide, dexamethasone protocol), complete , and red-circled symbols indicate mutations identified in this study. (B) Secondary (above) and 3D (below) RNA structures adopted by the n.33T>C and n.59T>C mutant found in CLL. The 3D mutant structure (blue) is aligned with the corresponding wild-type structure (black). The mutations are indicated in red. The 3D structures were generated using RNAComposer 38  www.nature.com/scientificreports/ hematologic remission was achieved in this case. Of interest, two events of further disease relapse were noted, the latter with coexisting hyperleukocytosis and TP53 deletion in 7% of nuclei tested. As mentioned above, it was noticed before that MIR142 mutations observed in AML coincide with and are functionally related to the R140Q hotspot mutation in IDH2 5,6 . However, in this study, we did not detect any IDH1/IDH2 mutation, generally observed in 4% of PMF cases 43 , in any of the two samples (PMF_19, CLL_2423) with MIR142 mutations.
Together, our results show that mutations in MIR142 are generally very rare in MPN but that their frequency (~ 1.5%) in PMF may be in the range of that observed in CLL or AML 4,[29][30][31][32] . Nevertheless, further studies are needed to determine the real contribution of MIR142 mutations to PMF. Despite the mostly negative nature of our results, this study complements knowledge about the occurrence of MIR142 mutations in blood cancers and pancancer in general. Additionally, as functional studies of MIR142 mutations were performed only in AML, in which the mutations are relatively rare, the confirmed recurrence of mutations in other hematologic cancers indicates that further studies are needed to fully understand the role of these mutations, especially in malignancies in which they are the most frequent, i.e., DLBCL and FL.
In summary, although we show that MIR142 mutations may occur in MPNs, particularly in PMF, the frequency is very low (~ 1.5%). We also confirmed the occurrence of MIR142 mutations in CLL, but similarly at a very low frequency.

Data availability
All data generated or analyzed during this study are included in this published article.