香京julia种子在线播放

Ethical approval was granted by the University of Calgary Animal Care Committee (Protocol#AC15-0079) and conformed to the Guide to the Care and Use of Experimental Animals. Obesity was induced in 8-week-old female Sprague-Dawley rats using a 10-week feeding period with a high fat/high sucrose diet (HFS, 39% fat and 44% sucrose, Dyets #102412), the composition of which has been previously published (17). Given that Sprague-Dawley rats can be obesity-prone or obesity-resistant when fed a HFS diet (18), we chose the top n = 45 best weight gainers from the n = 135 rats included in the diet-induced obesity protocol (Figure 1). The selected obese female rats were bred with male Sprague-Dawley rats and randomly allocated to one of three groups of n = 15 each throughout pregnancy and lactation: (1) WTR (HFS + Water); (2) APM (HFS + aspartame) (5–7 mg/kg, Fluka, Ottawa, ON, Canada); (3) STV (HFS + stevia) (2–3 mg/kg Rebaudioside A, Sigma-Aldrich, Oakville, ON, Canada). Aspartame and stevia were administered in the drinking water as described (9). These doses were selected within the acceptable daily intake (aspartame: 40 mg/kg, stevia: 4 mg/kg) (19) established by the Food Directorate of Health Canada and reflect physiological daily intake in humans (20). To minimize the effects of varying litter size, litters were culled to 10 offspring (5 females + 5 males) at birth. Offspring were weaned at 3 weeks of age and fed a control diet (AIN-93, Dyets, Bethlehem, PA, USA) until 18 weeks of age. Rats were fed ad libitum and housed in a temperature- and humidity-controlled room with a 12-h light/dark cycle. Body weight was measured weekly. Body composition of the dams was measured at weaning and the offspring measured at 3 and 18 weeks of age using dual-energy X-ray absorptiometry (DXA) (Hologic QDR 4500; Hologic Inc., Marlborough, MA, USA). Each litter was considered as n = 1 and male and female offspring were housed separately in plastic cages with wood-chip bedding following weaning at 3 weeks of age.

Distal jejunum and ileum tissue samples, and cecal digesta were collected at terminate points of dams (at weaning) and offspring (at 3 and 18 weeks of age) and immediately stored at −80°C until use. Total bacterial DNA was isolated from cecal digesta using FastDNA spin kit for feces (MP Biomedicals, Lachine, QC, Canada), pre-treated with bead-beating (MP Biomedicals, Lachine, QC, Canada) for 40 s × 3 times. Purified DNA was quantified using Quant-iT™ PicoGreen™ dsDNA Assay Kit (Invitrogen, Burlington, ON, Canada) and diluted to 20 ng/μl for 16S rRNA gene sequencing and 50 ng/μl for shotgun metagenomic sequencing.

To assess the contributions of maternal and offspring factors to cecal microbiota variations, genomic DNA from 105 samples were randomly selected for 16S rRNA gene sequencing, corresponding to 19 dams at weaning, 36 offspring at 3 weeks of age, and 50 offspring at 18 weeks of age. Microbial 16S rRNA gene tags were sequenced on Illumina MiSeq platform (2 × 300 bp) by amplifying the hypervariable V3–V4 regions (Centre for Health Genomics and Informatics, Calgary, AB, Canada), and analyzed in QIIME2 platform (21) (QIIME2 2020.2). The quality controlled 16S rRNA were filtered and denoised using DADA2 in QIIME2 (21) (QIIME2 2020.2). Frequency of sequence variants less than 10 were discarded. Taxonomy was assigned to sequence variants using Silva 138 database as reference. Sequence variants classified as genus Lactobacillus were additionally aligned to Genome Taxonomy Database release 95 (GTDB, https://gtdb.ecogenomic.org/) to reflect the current taxonomy of Lactobacillaceae (22) with the best BLAST hits.

To reconstruct cecal microbiome of offspring, 36 genomic DNAs from offspring at 3 weeks of age (weaning) were randomly selected for shotgun sequencing that included 12 samples (6 female and 6 males) from each dietary group. Shotgun metagenomic sequencing was conducted on NovaSeq 6000 (2 × 300 bp, Centre for Health Genomics and Informatics, Calgary, AB, Canada) with NEB Ultra II library preparation protocol (New England Biolabs, Ipswich, MA, USA) to generate 400 GB of data (800 M reads). Due to the high percentage of the unidentified rodent microbiome and the high functional heterogeneity in microbial community at the strain level, we adopted de novo assembly-based metagenomic reconstruction to understand the complexity of functional alterations in offspring's gut microbiota (23, 24). Adapters were checked and removed by Trimmomatic (25). Quality-controlled reads were assembled into contigs/scaffolds by samples using default parameters in IDBA_UD (26). Long contigs (> 3,000 bp) were further clustered into bins using MaxBin2 (27) based on G + C content and abundance of contigs through Expectation-Maximization algorithm. The 16S rRNA gene sequences data and shotgun metagenomic sequences data generated during this study are available at the National Center for Biotechnology Information (BioProject PRJNA675294) https://www.ncbi.nlm.nih.gov/biosample.

In total, 188 representative bins were chosen by de-replicating highly similar genomes using dRep (28), which refers to metagenomic assembled genomes (MAGs). Genomes completeness and contamination were estimated by examining a set of ubiquitous genes using CheckM (29) (Supplementary Table 1).

Phylogenetic trees of MAGs were built with PhyloPhlAn 3.0 (30) by aligning to a set of 400 core proteins. Species-, Genus- and Family-level taxonomic labels were assigned to genomes when Mash (31) average genomic distance calculated from at least 100 makers was < 5%, 15% and 30% to the closest reference set, respectively (30). The phylogenetic tree was generated with Randomized Axelerated Maximum Likelihood (RAxML) method and visualized by iTOL (32). Raw reads were mapped back to MAGs to calculate relative abundance using CoverM with “relative_abundance” formula in “genome” model.

To estimate the contribution of the cecal microbiome to microbial-derived energy harvest, metabolic enzymes were selected to retrieve pathways for microbial degradation of lactose and production of lactate, succinate, acetate, propionate, and butyrate (Supplementary Table 2). Query sequences for the same enzyme were selected from different bacteria and pre-aligned to minimize the overlap between BLAST hits for the same enzyme. Amino acid sequences of biochemically characterized key enzymes were blasted against amino acid database build from the open reading frames (ORF) of 188 assembled genomes (Prodigal 2.0) (33). An alignment coverage of > 70%, e value ≤ 1e-5 and amino acid identity of ≥40% or higher (34) were used as cut-off values for BLAST. The sums of mapped reads (%) for all the positive hits for the same enzymes were calculated as their potential capacities to produce the corresponding metabolites in each sample.

Total RNA was extracted from jejunum and ileum tissue using Trizol (Invitrogen, Carlsbad, CA) and reverse-transcribed into cDNA using SuperScriptII RT as previously described (35). Quantitative PCR was performed on BioRad iCycler (Bio-Rad, Hercules, USA) with primer sets LCT (forward, 5′-AATCTTCTTGGCTGGGAATGG-3′; reverse, 5′-CCTTGAGCACCTCGTTGA TG-3′) and GATA4 (forward, 5′-AATCTTCTTGGCTGGGAATGG-3′; reverse, 5′-CCTTGA GCACCTCGTTGATG-3′) targeted at lactase gene (lct) and regulatory gene (gata4). The 18S gene was amplified as reference gene for relative quantification with primer set (forward, 5′-TGACTCAACACGGGAAACC-3′; reverse, 5′-TCGCTCCACCAACTAAGAAC-3′). The relative expression ratio (R) of targets were calculated based on primer efficiency (E) and the Ct deviations as follow: R = ((Etarget)Δ Ct target (control-sample) / (E_ref)^{Δ Ct ref (control−sample)} (36).

The data for 16S rRNA gene sequences and shotgun metagenomic sequences, including relative abundance of 16S rRNA gene sequence variants, weighted UniFrac distances, relative abundance of assembled genomes and the corresponding prediction of metabolite production were compared using Kruskal-Wallis rank-sum test in R version 4.0.0 (2020-04-24). Statistical significance of ANOSIM was determined through 999 permutations between dietary groups. The R value between 0 and 1 reflects the dissimilarity between the groups and was calculated as follows: R = difference of mean rank (all distances between groups – all distances within groups) / (N(N-1)/4). The data for host parameters, including daily weight gain, body fat (%), liver weight, and bone mineral density, and Log₁₀ transformed gene expression were analyzed using linear mixed-effects (LME) models in R version 4.0.0 0 (2020-04-24). Dietary treatment was treated as fixed factor; rat was considered as experimental unit and its random effect was removed. Correlations between genome, metabolite production and host parameters were analyzed by Spearman's rank method. Bonferroni-adjusted p < 0.05 were considered significant.

Consumption of low-dose aspartame and stevia showed limited influence on the overall structure of cecal microbiota in dams (Figures 2A1,B1,C1; Supplementary Table 3) but significantly altered (p < 0.001) cecal microbiota of their 3-week old offspring (Figures 2A2,B2,C2; Supplementary Table 3). The differences at 3 weeks gradually dissipated from weaning into adulthood (Figures 2A2,B3,C3; Supplementary Table 3). Dams and 18-week-old offspring shared more similarities in cecal microbiota (Figure 2A1; Supplementary Table 3). Significant litter effects (male and female from same dam) were observed in 3-week old offspring, as indicated by the weighted UniFrac distances (Figure 2A2, R = 0.53, p = 0.001) and a higher number of highly similar (>99.0% average nucleotide identity) bacterial genomes that were shared between offspring from the same dam (Figure 2A3, p < 0.001). Litter effects decreased in 18-week-old offspring (Figure 2A2, R = 0.07, p = 0.0741). No sex difference was observed in 3-week and 18-week (p > 0.05) old offspring. Therefore, subsequent metagenomic analyses focused on microbiome alterations in 3-week old offspring without stratification by sex.

In total, 188 genomic bins were recovered from 36 3-week old offspring cecal samples, with an average size of 3.23 Mb and an average N50 of 74,759 bp (Supplementary Table 1). The high quality of reconstructed bacterial genomes enabled genome-based analysis of metabolic traits of intestinal microbial communities. Figure 3 depicts the phylogeny of 188 metagenomic assembled genomes. Five bacterial phyla were assigned to these 188 assembled genomes, including 1 assigned to Verrucomicrobia, 7 identified as Actinobacteria, 10 as Proteobacteria, 23 assigned to Bacteroidetes, and 137 identified as Firmicutes. Of the 188 assembled genomes, 149 (79%) were assigned to genus or species level with an average amino acid identity of > 95% to the closest references calculated from at least 100 protein sequences. The relative abundance of each genome, referring to the percentage of mapped reads, was further transferred to Z-scores for visualization (Figure 4). In total, 92 genomes showed differential relative abundance (p < 0.05) between dietary treatments (Supplementary Figure 1).

To estimate the contribution of the cecal microbiome to microbial-derived energy harvest, pathways for microbial degradation of lactose and for production of lactate, succinate, acetate, propionate, and butyrate were reconstructed (Figure 4; Supplementary Table 1). Intracellular GH2 β-galactosidases BbgI, LacM, LacZ and Lacdeb, specifically distributed in Lachnospiraceae, Lactobacillaceae and Clostridiales were determined as the major metabolic enzymes hydrolysing milk-derived lactose in rats (Figures 4A1–A5). The relative abundance of these β-galactosidases significantly decreased in the stevia and/or aspartame groups compared to water control (Figures 4A1,A2,A4,A5, p < 0.05). Phospho-β-galactosidase LacG (Figure 4A3) showed low abundance and no differences between sweeteners and water control groups in the cecal microbiota. The abundance of enzymes for pyruvate to lactate conversion (Figure 4B) was significantly reduced (p < 0.05) in the aspartame group compared to the stevia and water control.

Rat cecal microbes (90.43%) were generally able to produce acetate through acetate kinase (AK) (not shown) and due to the multiple mapping of identical reads, the average of mapped read ratio for acetate was higher than 130%. Variable bacterial types and abundance were involved in acetate production, but no difference was observed in total abundance between dietary groups (Figure 4A). Very few bacteria produce both propionate and butyrate as major products of carbohydrate fermentation (Supplementary Figure 1). Firmicutes genomes were identified as the major butyrate producers. Of these butyrate producers, 15 of 28 butyryl: acetate CoA-transferase carrying genomes and 10 of 17 butyrate kinase carrying genomes were significantly enriched in both aspartame and stevia groups or specifically enriched in aspartame (p < 0.05). Of the three pathways for propionate production, the succinate pathway, dominated by Akkermansia muciniphila and 14_Bacteroidaceae unclassified, was the most abundant (30 of 46 propionate-producing genomes). This pathway was significantly enriched in the aspartame group. Of the 30 succinate-propionate producers, 9 genomes were enriched by aspartame and stevia or only by aspartame (4 of 9). The propanediol pathway for propionate production was less abundant in the stevia group although the production of 1,2-propanediol was not different between dietary groups. Genes coding for the acrylate pathway were rare in the rat cecal microbiome. Production of cobalamin (vitamin B12), a co-factor required by multiple metabolic processes including propionate production, also showed variable abundances in aspartame and/or stevia offspring.

To further explore the influence of non-nutritive sweeteners on the cecal microbiome of offspring, genomes with high relative abundance and significant differences between groups were selected as representatives for further correlation analysis (Figure 5A; Supplementary Table 4). The genome of 14_Bacteroidaceae unclassified represented the most abundant microbe that was enriched in both aspartame and stevia groups; the genome of 1_Akkermansia muciniphila represented bacteria specifically enriched by aspartame; microbes reduced by aspartame were represented by 48_Limosilactobacillus reuteri and 49_Ligilactobacillus murinus. The other genomes, including 176_Eubacterium spp., 151_Blautia spp., 152_Blautia spp., and 175_Lachnospiraceae unclassified, showed variable differences between dietary treatments. Jejunal and ileal expression of lactase and its regulator GATA4 can directly affect the availability of lactose for fermentation in the cecum. The expression of these genes was therefore examined as a potential upstream factor shaping the microbiome, however no significant differences between dietary treatments were found (Figure 5B). Host parameters, including daily weight gain, body fat, liver weight, and bone mineral density were selected for correlation analysis (Figure 5C). Maternal consumption of aspartame increased daily weight gain in offspring (Figure 5C). Aspartame and stevia significantly increased body fat % and liver weight but decreased bone mineral density in 3-week old offspring (Figure 5C).

Correlations between bacterial genomes, bacterial metabolites, and between host parameters depict microbial-host interactions in weanling pups (Figure 5D). Two distinct bacterial consortia were identified, showing negative correlations with each other and consistently opposite correlations with metabolite production and host parameters. The propionate- and butyrate-producing consortium, represented by 1_Akkermansia muciniphila, 14_Bacteroidaceae unclassified, 175_Lachnospiraceae unclassified, and 176_Eubacterium spp., was enriched in offspring of sweetener-consuming dams, and was positively correlated with offspring's daily body weight gain, liver weight and body fat but negatively correlated with bone mineral density (Figure 5D). Conversely, the lactate-producing consortium, including 48_Limosilactobacillus reuteri, 49_Ligilactobacillus murinus, 151_Blautia spp., and 152_Blautia spp., which degrade lactose into lactate, were negatively correlated to the host parameters mentioned above except bone mineral density (Figure 5D). Significant correlations between gene expression of lactase/GATA4 gene and selected bacterial genomes and host parameters were only observed between jejunal lct/gata4 with butyrate formation (Figure 5D).