RUNX1 regulates miRNA expression in AML
To identify potential roles of RUNX1 in AML, the differentially expressed miRNAs were identified from 14 TCGA AML samples by using DESeq2 algorithms. Compared the expression in AML samples with wildtype RUNX1, in those with mutant RUNX1 28 miRNAs were differently expressed (DEmiRNAs)with significance, in which 16 DEmiRNAs were up-regulated and 12 DEmiRNAs were down-regulated (Fig. 1 and Supplemental 2, significant differentially expression miRNAs and raw miRNAs expression array). Some miRNAs such as hsa-miR-1304, hsa-miR-139, hsa-miR-151a, hsa-miR-1515b, hsa-miR-4473, hsa-miR-873, hsa-miR-874, hsa-miR-877 were differential expressed in patients with NPM1+/FLT3+ AML versus control(34). hsa-miR-126 was involved in AML pathogenesis with cell cycle, proliferation, survival, and differentiation(35, 36).hsa-miR-551b was found to be upregulated in AML(37).
Significant Correlation of RUNX1 miRNAs with the Characteristics and Survival of Patients with AML
To explore the correlation of DEmiRNAs with RUNX1 in AML, the association of miRNAs and RUNX1 in expression was determined by Spearman algorithms. (Supplemental 3, significant correlation miRNAs to RUNX1, p < 0.005). We found 6 miRNAs were significantly correlated to RUNX1 in DEmiRNAs. hsa-miR-551b, has-miR-4473, and has-miR-873 were positive to RUNX1, hsa-miR-6877, has-miR-30c-2, and has-miR-30c-1 were negative to RUNX1. As shown in table I, by Cox regression analyses, the expression of 3 miRNAs (hsa-miR-873, hsa-miR-551b, and hsa-miR-4473) of these six miRNAs was correlated to patients’ overall survival (OS, p < 0.005), and patients with low expression had better treatment outcome (Fig. 2). These findings revealed that these three miRNAs could be involved in chemo-resistance by interfering expression of target mRNAs, and confer poor prognosis in AML.
RUNX1 regulates the MAPK signaling pathway through miRNAs
Check the previous IDs of hsa-miR-4473, hsa-miR-873 and hsa- miR-551b in miRBase database, we used hsa-miR-873 as hsa-miR-873-5p, and hsa-miR-551b as hsa-miR-551b-3p. miRNAs regulated gene expression at the post-transcriptional level by binding to 3’ UTR of the target mRNAs. To illustrate the mechanism on chemo-resistance of those 3 miRNAs associated with OS, miRWalk3.0 was applied to predict target genes of hsa-miR-4473, hsa-miR-551b-3p, and hsa-miR-873-5p. By target regions within 3’UTR and more than 0.95 of score, 4633 target genes of these three miRNAs were identified. The KEGG pathway enrichment of these target genes was analyzed with miRWalk3.0 (Supplemental 4, predictions result of DEmiRNAs from miRWalk3.0.). As presented in Fig. 3A-B, with the highest scores, MAPK signaling pathway(hsa04010), Ras signaling pathway(hsa04014), and Rap1 signaling pathway(hsa04015) were found by these enrichment assays. The target genes of three miRNAs were most significantly enriched in the MAPK signaling pathway. The target genes in MAPK pathway signaling were shown in Supplemental 5, graph of miRNA-target genes mapping in the MAPK signaling pathway. Total 95 miRNA-target genes by the red font mapping in the MAPK signaling pathway from KEGG Mapper, and these genes were involved in ERK, JNK, and p38 MAP kinase pathways (red color). Bioinformatics target prediction and pathway analysis results indicated RUNX1 could regulate MAPK signaling pathway through miRNAs, which implied that MAPK signaling could modulate chemo-response of AML patients with RUNX1 mutations.
RUNX1 regulates the MAPK signaling pathway directly
To investigate the control of RUNX1 on mRNA expression, genes associated with RUNX1 expression in AML carrying wild-type RUNX1 were gathered from GEO dataset GSE106291, and the most significant associated genes(p<0.05) were used for KEGG enrichment analysis (Supplemental 6, KEGG enrichment of GSE106291 and row wild-type RUNX1 expression array and Supplemental 7, dot plot of KEGG enrichment of RUNX1 correlate genes in GSE106291). It was noted that the MAPK signaling pathway was the highest enriched pathway, too. One hundred and eighty-two RUNX1-related genes were regulating MAPK signaling pathway (hsa04010, pvalue=0.0071). With RUNX1 ChIP-seq data of human K562 cell line in ENCODE (Fig. 4A), it was found that 35 genes of the optimal and conserved peaks with annotation were enriched to MAPK signaling pathway (Fig. 4B). Meanwhile, in human CD34+ cells there were 9727 genes within whose promoters there were RUNX1 binding motifs (TGTGGT/TGTGTT) (38). Among these genes, 172 genes were participating in MAPK signaling pathway (figure 4B). These results supported RUNX1 could transcriptionally control genes’ expression directly, which were components of MAPK signaling pathway.
As it was shown above, by CHIP-seq data of K562 and human CD34 cells, many genes of MAPK signaling pathway were recognized and potentially controlled by RUNX1. In the RUNX1-related genes, 61.90% genes mediated MAPK signaling pathway. More than 60% genes, which had RUNX1 peaks within 10kb of their promoters (CD34 cell and K562 cell) were involved in MAPK signaling pathway. In comparison, there was less than 30% genes genome-wide with RUNX1 peaks in their promoters. In DemiRNAs target genes, there were 95 genes involved in MAPK signaling pathway, more than 60% of genes (59/95) of which were overlapped with RUNX1-related genes. Further, 63 of 95 target genes of DemiRNAs had RUNX1 binding peaks in their promoters. It meant that RUNX1 could regulate their 63 genes’ expression indirectly and/or directly. Among these four subsets, there were 44 shared genes (Fig. 5 and Supplemental 8, RUNX1 related genes target to MAPK signaling pathway). Log-rank test of these 44 genes found three of them were significantly related to survival (Fig. 5D, p<0.01), including MEF2C, TGFBR2, and CACNB3. It indicated that RUNX1 regulated MAPK signaling pathway indirectly (miRNAs) and directly (mRNA).