2.9. DNA extraction and Illumina MiSeq sequencing
Faecal DNA was isolated as described(Rodriguez-Nogales et al., 2017). The resultant sequences were quality-filtered, clustered, and taxonomically allocated against the SILVA database with 97% similarity threshold using the QIIME software package (Version 1.9.1) (Knight Lab, San Diego, CA, USA). The resulting abundance was used to compute the total bacterial diversity in an equivalent manner.
DNA from fecal contents was isolated following the procedure described by Rodríguez-Nogales et al . 2017(Rodriguez-Nogales et al., 2017). Total DNA from faecal samples was PCR amplified using primers targeting regions flanking the variable regions 4 to 5 of the bacterial 16S rRNA gene (V4-5), gel purified, and analyzed using multiplexing on the Illumina MiSeq machine. PCR reactions from the same samples were cleaned and then normalized using the high-throughput Invitrogen SequalPrep 96-well Plate kit. Later, a library from the samples was made fluorometrically to be quantified fluorometrically before sequencing.
The resulting sequences were completed, quality-filtered, clustered, and taxonomically assigned on the basis of 97% similarity level against the RDP (Ribosomal Database Project) by the QIIME software package (Version 1.9.1) (Knight Lab, San Diego, CA, USA). Sequences were selected to estimate the total bacterial diversity of the DNA samples in a comparable manner and were trimmed to remove barcodes, primers, chimeras, plasmids, mitochondrial DNA and any non-16S bacterial reads and sequences <150 bp.
Alpha diversity (α-diversity) indices and bacterial abundance data were compared using Kruskal–Wallis test followed by pairwise Mann–Whitney U comparison. Resulting p-values were corrected by the Bonferroni method. Analysis of α-diversity was performed on the output normalized data, which were evaluated using Mothur. The biomarkers for both species taxonomic analysis and functional pathways via calculation of the linear discriminant analysis (LDA) score among different phenotype groups were calculated by LEfSe (linear discriminatory analysis (LDA) effect size) (Version 1.0). Principal coordinate analysis (PCoA) was performed to identify principal coordinates and visualize β-diversity in complex multidimensional data of bacteriomes from different groups of mice. Differences in β-diversity were tested by permutational multivariate analysis of variance (PERMANOVA) using the web-based algorithm tool Microbiome Analyst (Dhariwal et al., 2017). The data are expressed as the mean ± standard error of the mean (SEM). Experimental data were analysed in GraphPad Prism 8 (GraphPad Software, Inc., La Jolla, CA, USA) by one-way or two-way ANOVA or Pearson correlation. Data with p<0.05 were considered statistically significant. Metabolic phenotypes were obtained by genera classification according to their primary fermentation products as acetate, butyrate, lactate, or other producers using Bergey’s Manual of Systematic Bacteriology(Boone, Castenholz, & Garrity, 2001). The genera with unknown or ambiguous fermentative products were excluded. Major genera were classified according to the dominant fermentation end-product(s).
Hierarchical clustering and heat maps depicting the metabolic parameters, patterns of abundance and log values were constructed within the “R” statistical software package (version 3.6.0; https://www.r-project.org/) using the ”pheatmap”, “heatmap.2” and “ggplots” packages. Spearman’s correlations of bacterial taxa with metabolic parameters and KEGG metagenomic functions were calculated in “R”. Co-occurrence networks between taxa and functions were calculated by using the open-source software Gephi (https://gephi.org/) to find differential associations caused by similar alterations in the proportion of different taxa and their predicted functions between different groups of mice. Modularity-based co-occurrence networks were analysed at a Spearman’s correlation cut off 0.7 andp<0.01 ; the selected correlation data were imported into the interactive platform, Gephi (version 0.9.2; https://gephi.org), and the following modularity analyses and keystone node identification were conducted within Gephi.