MicroRNAs mediate liver transcriptome changes upon soy diet intervention in mice

Abstract Soy‐based diets have triggered the interest of the research community due to their beneficial effects on a wide variety of pathologies like breast and prostate cancer, diabetes and atherosclerosis. However, the molecular details underlying these effects are far from being completely understood; several recent attempts have been made to elucidate the soy‐induced liver transcriptome changes in different animal models. Here we used Next Generation Sequencing to identify a set of microRNAs specifically modulated by short‐term soy‐enriched diet in young male mice and estimated their impact on the liver transcriptome assessed by microarray. Clustering and topological community detection (CTCD) network analysis of STRING generated interactions of transcriptome data led to the identification of four topological communities of genes characteristically altered and putatively targeted by microRNAs upon soy diet intervention.


| INTRODUCTION
Soy-based/rich diets have been associated with decreased risk of breast and prostate cancer, type 2 diabetes and atherosclerotic pathologies, an effect attributed to soy proteins and/or associated isoflavones. 1 Several transcriptomic studies have investigated the effect of soy components on vertebrate physiology and metabolism, emphasizing the central role of the liver in mediating these effects. [2][3][4] However, nothing is known about the impact of soy diet on liver microRNA expression and their role in mediating the soy-associated transcriptome changes. In the present study, we used the Next Generation Sequencing (NGS) and microarray to assess the expression of micro-RNAs and mRNAs in the liver of adult male mice fed for 4 weeks a soy-rich diet. By combining miRWalk3.0 prediction and STRING algorithms with a clustering and topological community detection (CTCD) approach, we analysed and characterized the functional gene communities impacted by microRNAs upon soy diet intervention.

| Dietary intervention
Two groups of three 12 weeks old male mice housed in Udel ® polysulphone cages, on a 12 hour light-dark cycle were fed ad libitum granulated regular chow (Cantacusino Institute, Bucharest) and granulated soy-enriched chow (25% soy bean) for 28 days. On day 28, the animals were sacrificed and approximately 0.5 g of liver tissues have been collected, immediately, immersed in RNAlater stabilization solution (Qiagen) and stored at −80°C until its further use.  Mapped data were filtered using a cut-off of four reads, normalized using the DESeq2 geometric mean-based method, log2-transformed, followed by calculation of differential miRNA expression using a False Discovery Rate (FDR) of 5% (Benjamini & Hochberg).

| Bioinformatics analysis
Target prediction for the differentially expressed microRNAs was computed using the miRWalk3.0 platform (adjusted binding probability > 0.95). 6 The functional interactions between the genes found deregulated upon soy-diet intervention were retrieved using the STRING platform (https://string-db.org//; medium confidence interaction score) and further analysed using a CTCD approach which assumes both modularity-class and force directed layout clustering, as previously described. [7][8][9][10] In our graph representation, each vertex represents a gene/protein and each edge stands for all types of gene interactions (ie either up-and down-regulation) between two genes/proteins. Modularity classes are indicated by assigning a distinct colour to each community and associate with distinct biological functions. 11 3 | RESULTS

| The soy-enriched diet
Despite the biochemical changes triggered by soy addition (File S1; Supplementary Table S1), the overall change of the soy-enriched diet energetic value is minimal (+6.1%), which is reflected by the lack of significant weight gain in the soy-fed group compared to control (14.2% vs 6.8%, P = 0.224).

| Network analysis
In order to understand the biological significance of our transcriptome data, we used a CTCD approach to analyse the functional protein-protein interactions network of the differentially expressed genes generated by STRING10.0 algorithms (http://string-db.org/).
We identified five gene communities, partially overlapping with, but also complementing the STRING functional networks (File S1; Supplementary  (Figure 1).

| DISCUSSION
Our microarray analysis identified transcriptome changes suggestive for an alteration of the liver defense against xenobiotics: Cyp4a14  F I G U R E 1 Network analysis: Analysis of liver microRNA-mRNAs relationships upon soy diet intervention based on a clustering and topological community detection procedure applied to STRING data generated from the differential gene expression network. Vertex/gene positions are assigned by employing force-directed network layout Force Atlas 2. The distinct colours indicate modularity classes associated with functional properties (Bluecatalytic activity community, Turquoise -Iron metabolism community, Red -Egfr (nuclear receptor) community, Pink -Esr1/ RXR (nuclear receptor) community, Green -Xenobiotic/fatty acid metabolizing community, Yellow -Glutamine/Glutathione metabolizing community). mmu-miR-145a-5p and mmu-miR-455-3p miRWalk3.0 predicted targets are marked with a star and outlined in tabular form SECLAMAN ET AL.