Marfan syndrome (MFS), due to FBN1 mutations, is associated with thoracic aortic aneurysm (TAA). The TGFß and other signalling pathways have been implicated in TAA formation, however, regulators of cell signalling in MFS are yet to be defined. We hypothesized that distinct microRNA (miRNA) expression profiles occur in MFS aorta and that these can illuminate novel cell signalling targets contributing to TAA. MiRNAs were extracted from MFS (n=15) and controls (n=11) ascending aortic tissue, for discovery open-array panel profiling of miRNA expression. Individual miRNA expression profiles were compared between MFS and controls by random forest machine learning algorithm (Python). The miRNAs identified with a significant importance value were selected for gene target prediction (score >90) through miRDB, followed by KEGG analysis. Any miRNA with more than 2000 predicted targets in the genome was excluded. P=0.05 was used as a cut-off value for KEGG analysis. The random forest classifier identified 53 miRNAs with differential expression between MFS and controls, contributing to the model (predictive accuracy 73%). The five most significant miRNAs were miR-181a-3p, miR-144-3p, miR-545-5p, miR-27a-5p and miR-181c-3p. 2984 unique target genes were identified from 53 significant miRNAs. KEGG analysis of these target genes identified multiple pathways as having potential differential regulation by miRNAs between MFS and controls. The most significant and relevant pathways identified were PI3K/AKT (p<0.00001) and the downstream FOXO (p<0.0004) pathways. Others include the MAPK (p<0.00001), mTOR (p<0.00002), RAP1 (p<0.005), ErbB (p<0.003) and RAS (p<0.001) pathways, in addition to TGFß (p<0.01). Mapping miRNA targets to KEGG pathways identified potential regulatory points, e.g., miR-29c-3p affecting FBN1, AKT3, PTEN (within the TGFß and PI3K/AKT pathways); miR-19a-3p affecting RAF1 , FMR1, RIN2 (within the MAPK, fragile X and adherens junction pathways); miR-27a-5p affecting LTBP1 (within the TGFß pathway). In conclusion, machine learning analysis identifies differential expression of multiple miRNAs in aortic tissue from MFS, which map to cell signalling pathways implicated in the pathogenesis of TAA in MFS.
Read full abstract