Construction of a TF-miRNA-gene feed-forward loop network predicts biomarkers and potential drugs for myasthenia gravis

Abstract

Myasthenia gravis (MG) is an autoimmune disease and the most common type of neuromuscular disease. Genes and miRNAs associated with MG have been widely studied; however, the molecular mechanisms of transcription factors (TFs) and the relationship among them remain unclear. A TF-miRNA-gene network (TMGN) of MG was constructed by extracting six regulatory pairs (TF-miRNA, miRNA-gene, TF-gene, miRNA-TF, gene-gene and miRNA-miRNA). Then, 3/4/5-node regulatory motifs were detected in the TMGN. Then, the motifs with the highest Z-score, occurring as 3/4/5-node composite feed-forward loops (FFLs), were selected as statistically significant motifs. By merging these motifs together, we constructed a 3/4/5-node composite FFL motif-specific subnetwork (CFMSN). Then, pathway and GO enrichment analyses were performed to further elucidate the mechanism of MG. In addition, the genes, TFs and miRNAs in the CFMSN were also utilized to identify potential drugs. Five related genes, 3 TFs and 13 miRNAs, were extracted from the CFMSN. As the most important TF in the CFMSN, MYC was inferred to play a critical role in MG. Pathway enrichment analysis showed that the genes and miRNAs in the CFMSN were mainly enriched in pathways related to cancer and infections. Furthermore, 21 drugs were identified through the CFMSN, of which estradiol, estramustine, raloxifene and tamoxifen have the potential to be novel drugs to treat MG. The present study provides MG-related TFs by constructing the CFMSN for further experimental studies and provides a novel perspective for new biomarkers and potential drugs for MG.

Figure 1

The basic characteristics of TMGN with all six types of regulatory pairs (miRNA–gene, miRNA–TF, TF–miRNA, TF–gene, gene–gene and miRNA–miRNA). (A) A global of the network of TMGN. TFs, miRNAs and genes are colored green, orange and blue, respectively. (B–E) The basic features of the network include degrees, clustering coefficients, topological coefficients and neighborhood connectivity of the network. (F) A sub-network with MYC as the central node. (G) A sub-network with ESR1 as the central node. TMGN: TF–miRNA–gene network; TF: transcription factor.

Figure 2

3-node, 4-node and 5 node motifs. (A) 15 different types of 3-node regulatory motifs. (B) 29 different types of 4-node regulatory motifs. (C) 39 screened different types of 5-node regulatory motifs. The motifs are composed of miRNAs, TFs, and target genes and their Z-scores and p value s are presented. Orange triangles represent miRNAs, green rounds represent TFs and purple squares represent genes. An arrow ending in a circle represents the regulatory relationship, while, an arrow ending in a “T” represents the repression relationship. Six types of relationships involved in these motifs: miRNA–gene (miRNA represses gene expression); miRNA–TF (miRNA represses TF gene expression); TF–miRNA (TF regulates miRNA expression); TF–gene (TF regulates gene expression); gene–gene (gene and gene interaction); and miRNA–miRNA (miRNA and miRNA interaction). Motifs surrounded by red squares (3-node, 4-node and 5-node composite feed-forward loop) as significant motifs were merged to form 3-node, 4-node and 5-node regulatory sub-networks, respectively. TF: transcription factor; FFL: feed-forward loop.

Figure 3

3/4/5-node composite FFL motif-specific sub-network (CFMSN). (A) A global view of 3/4/5-node composite FFL motif-specific sub-network (CFMSN). Orange triangle represents miRNA, green round represents TF and blue square represents gene. A line ending in a circle represents the regulatory relationship, while, a line ending in a “T” represents the repression relationship. (B–D) The examples of 3/4/5 node composite FFL. (E–H) The basic features of the network include degrees, clustering coefficients, topological coefficients and neighborhood connectivity of the network. TF: transcription factor; FFL: feed-forward loop.

Figure 4

Gene Ontology and KEGG pathway analysis using genes and targets of miRNAs in CFMSN. (A) The top 10 pathways enriched by genes (left) and target genes of miRNAs (right). Pathways colored in red are common ones in both left and right. (B) The top 10 GO terms enriched by genes (left) and target genes of miRNAs (right). GO terms colored in red are common ones in both left and right. (C) A part of PI3K-Akt signaling pathway. (D) A part of Hepatitis B pathway. Genes with a yellow background is the risk genes from CFMSN. GO: Gene Ontology; KEGG: Kyoto Encyclopedia of Genes and Genomes; CFMSN: 3/4/5-node composite FFL motif-specific sub-network.

Figure 5

CFMSN and associated drugs. (A) A sub-network of TF in CFMSN and selected drugs. (B) A sub-network of genes in CFMSN and selected drugs. (C) A sub-network of miRNAs in CFMSN and selected drugs. (D) A global view of CFMSN and selected drugs through TFs, genes and miRNAs in CFMSN. (E) A sub-network of selected drugs targeted with screened TFs, genes and miRNAs. Orange triangle represents miRNA, green round represents TF, blue square represents gene and red “V” represents drug. (F) Part of pathway 07,226: Progesterone, androgen and estrogen receptor agonists/antagonists. Drugs colored in red are from 21 drugs we selected. CFMSN: 3/4/5-node composite FFL motif-specific sub-network; TF: transcription factor.

Figure 6

A model illustrating the misregulated processes interferred by FFLs composed of TFs, miRNAs and genes. TFs and miRNAs cooperatively mediate the pathway dysregulation which may be implicated in cell survival and apoptosis regulation mechanisms in MG. TF: transcription factor; FFL: feed-forward loop; MG: myasthenia gravis.