Samples from juvenile zebrafish fed the three different diets were sampled from individual separate tanks. Amplicon libraries of the V4 region of the 16S rRNA were generated from the cDNA synthetized from single gut and water samples, using barcoded and modified F515-806R primers [86 (link)]. The PCRs were performed in triplicate, purified, and quantified as previously described [29 (link)]. Purified PCR amplicons were pooled in an equimolar mix and sent for library preparation and sequencing using the Illumina NovaSeq 6000 S2 PE150 XP technology at Eurofins Genomics Germany GmbH (Eurofins Genomic, Ebersberg, Germany). Raw paired-end reads were analyzed using the standard parameters of NG-Tax 2.0 [66 (link)], with the exception of using 100 bp as the forward and reverse read length, as implemented in Galaxy [2 (link)], to obtain Amplicon Sequence Variants (ASVs). Taxonomy was assigned to ASVs using the Silva_132 database [67 (link)]. Two synthetic “mock communities” with known compositions were amplified and sequenced as positive controls and a no-template control was also included as a negative control [68 (link)]. The distribution of reads per sample and the variance in ASVs were assessed and Alpha- and Beta-diversity measurements were performed using R v4.1.2 and RStudio [43 ], using packages ggplot2, [89 ], ape, [61 (link)], plyr, [93 ], vegan, [59 ], RColorBrewer, [58 ], reshape2, [90 ], scales [91 ], data.table, [19 ], microbiome, [42 ], dplyr, [92 ], phyloseq, [55 (link)], ggdendro, [84 ] and DT [95 ]. The analysis yielded 17,203,234 high-quality reads. We excluded one sample (54 dpf butyrate diet) because it had 2 reads only and we kept all the other samples (> 30.000 reads). Rarefaction curves for all samples reached a plateau, indicating that sufficient sequencing depths was achieved (data not shown). For the calculation of alpha-diversity indices, data was rarefied against the sample containing the lowest number of reads (31,814 reads). Redundancy analysis (RDA) and principal component analysis (PCA) were performed with Canoco v5.15 [9 ] using analysis type “constrained” or “unconstrained”, respectively. Response variables were log-transformed with the formula log(10,000*relative_abundance + 1). RDA p-values were determined through permutation testing (500 permutations). Boxplots were generated using Prism v.9.0.0 (GraphPad Software, San Diego, California USA). Cytoscape v3.9.1 [75 (link)] was used to visualize the diet-specific co-occurrence of ASVs based on their relative abundances. Additional data handling and format conversions were done in Python (https://www.python.org/ ).
Free full text: Click here
