Structure of the gut microbiota from Winnie and C57BL/6 mice
16S rRNA gene metabarcoding was used to characterize the structure of the bacterial communities from stool samples of WN and WT mice 4–16 weeks old. WT and WN fecal samples showed a similar Shannon diversity index (2.975 and 3.316, respectively).
The 16S Metagenomics (Version: 1.1.0) pipeline was used to determine taxonomic classification. Heat maps show relative abundance of phyla (Fig. 1A) and most represented (> 1%) genera (Fig. 1B) and species (Fig. 1C) in WT and WN samples. The full list of phyla, genera, and species are shown in Tables S2, S3, and S4, respectively. At phylum level, only few differences between the two microbiota could be observed. Firmicutes, Bacteroidetes, Proteobacteria and Verrucomicrobia were the most abundant phyla, accounting for approximately 92.7% and 94.4% relative abundances in WT and WN, respectively (Fig. 1A). In the WN sample, it was possible to note a lower relative abundance of Bacteroidetes, Actinobacteria, Candidatus Saccharibacteria, Cyanobacteria, Deferribacteres and Synergistetes, and a greater abundance of Firmicutes, Tenericutes, Acidobacteria, and Chloroflexi compared to WT. Aquificae and Spirochaetes were detected only in WT, while Poribacteria, Gemmatimonadetes and Thermotogae were only found in WN. Relative abundances of Proteobacteria and Verrucomicrobia were similar in WT and WN.
Greater differences than phylum level could be observed when microbial communities were analyzed at the genus level (Fig. 1B). Eight genera, i.e., Muribaculum, Barnesiella, Bacteroides, Clostridium (cluster XIVa), Alistipes, Prevotella, Akkermansia, Falsiporphyromonas were the most abundant, accounting for 51.07% and 51.09% relative abundances in WT and WN, respectively. Although the cumulative abundance of these genera was similar between the two microbiota (51.07% vs. 51.09%), important differences could be detected in their relative distribution. In the WN sample it was possible to note a lower relative abundance of Muribaculum, Barnesiella, Falsiporphyromonas, and a greater abundance of Bacteroides, Clostridium (cluster XIVa), Alistipes, Akkermansia. Relative abundances of Prevotella were similar in WT and WN.
Among the less abundant genera, Eubacterium, Intestinimonas, Alkaliphilus, Anaeroplasma, Paraprevotella, Ruminococcus, Lachnospiracea incertae sedis, Flavonifractor were more represented in WN compared to WT, while Odoribacter, Eisenbergiella, Saccharibacteria incertae sedis, Nitritalea, Mucispirillum, Clostridium (group IV), Falsiporphyromonas, Turicibacter, and Rikenella were less represented. Relative abundances of Desulfovibrio and Helicobacter were similar in WT and WN.
These differences at the genus level are reflected at the species level, although caution should be used in the interpretation of the data at the species level due to the inherent limitations of the metabarcoding analysis. Here, the most abundant four species, i.e., Muribaculum intestinale, Akkermansia muciniphila, Barnesiella intestinihominis and Falsiporphyromonas endometrii, accounted for approximately 27.26% and 25.19% relative abundances in WT and WN, respectively (Fig. 1C). Relative abundances (within the sample) of these four species were similar in the two microbiomes.
In contrast, marked differences were observed among several less represented species. Specifically, Eubacterium coprostanoligenes, Eubacterium hallii, Clostridium polysaccharolyticum, Clostridium saccharolyticum, Flintibacter butyricus, Alistipes onderdonkii, Alistipes obesi, Alistipes shahii, Desulfovibrio piger, Bacteroides acidifaciens, Anaeroplasma bactoclasticum, and Paraprevotella clara were more abundant in WN compared to WT. In contrast, Barnesiella viscericola, Parabacteroides distasonis, Saccharibacteria incertae sedis, Turicibacter sanguinis, Alistipes senegalensis, Rikenella microfusus, Nitritalea halalkaliphila, Mucispirillum schaedleri, and Odoribacter laneus were less abundant. Relative abundances of Culturomica massiliensis and Parabacteroides merdae were similar in WT and WN.
It may be interesting to note the different distribution of species within the same genus, namely: Barnesiella intestinihominis (similar in WT and WN) and Barnesiella viscericola (much less abundant in WN); Parabacteroides merdae (similar in WT and WN) and Parabacteroides distasonis (slightly less abundant in WN); Alistipes onderdonkii (slightly more abundant in WN), Alistipes obesi and Alistipes shahii (much more abundant in WN), and Alistipes senegalensis (less abundant in WN).
Metabolic profiling of the gut microbiota from Winnie and C57BL/6 mice
Metabolic patterns of the microbial communities from stool samples of WN and WT were evaluated by BIOLOG system (Fig. 2, Fig. S1-S4).
The growth of the microbial communities was analyzed, at different time points (0, 24, 48, 72, 96, 168, 216 h), in the presence of 31 substrates including 9 carbohydrates, 10 carboxylic acids and phenols, 8 amino acids and amines, and 4 polymers, as previously detailed.
Among carbohydrates, WT and WN microbes were able to utilize all substrates, with the exception of the i-Erythritol (Fig. 2; Fig. S1). However, β-Methyl-D-Glucoside and D-Xylose appeared to be much better assimilated by WT microbes, while D-Cellobiose was better utilized by WN microbes. Among carboxylic acids and phenols, WT and WN microbial communities were able to grow efficiently only on D-Galactonic acid γ-lactone, Pyruvic acid methyl ester, and D-Galacturonic acid, while only the WT microbes were able to utilize D-Malic acid (Fig. 2; Fig. S2). When microbial growth was tested in the presence of amino acids and amines, an appreciable growth was detected only in the presence of L-Asparagine, L-Serine, and, to a lesser extent, L-Threonine, while only WT microbes were able to utilize Glycyl L-Glutamic acid (Fig. 2; Fig. S3). Among polymers, α-Cyclodextrin was not assimilated by WT and WN microbial communities; Glycogen, Tween 40, Tween 80 were utilized by both community without substantial differences (Fig. 2; Fig. S4).
Overall, these data show the usefulness of the Biolog EcoPlate system in metabolic phenotyping of the gut microbiota, highlighting interesting differences between WT and WN microbes. These were then investigated by matching the substrates utilized by WT and WN microbial communities in the Biolog EcoPlates with the substrates that were associated with the most abundant bacterial species (> 1%) in the WT and WN gut microbiota by the MetaCyc Metabolic Pathway Database (Fig. 3). Results demonstrated a good match for carbohydrate substrates, confirming the potential of the most abundant bacterial species in the WT and WN gut microbiota to assimilate all substrates, except for i-Erythritol.
A rather good correspondence was also found for assimilated amino acid and amine substrates (L-Asparagine, L-Serine, and L-Threonine), although MetaCyc revealed an ability to utilize also L-Arginine, L-Phenylalanine, and Putrescine, which was not revealed on Biolog EcoPlates under aerobic incubation conditions. As for carboxylic acids and phenols, MetaCyc confirmed a poor ability to assimilate D-Glucosaminic acid and Itaconic acid (not shown in Fig. 3), and a good ability to assimilate D-Galacturonic acid. MetaCyc revealed the possibility of also using α-Ketobutyric acid, 2-Hydroxybenzoic acid, 4-Hydroxybenzoic acid, which was not revealed on Biolog EcoPlates, although the ability to assimilate 2-Hydroxybenzoic acid and 4-Hydroxybenzoic acid was limited to a small number of species. MetaCyc failed, however, to reveal the utilization of D-Galactonic acid γ-lactone, Pyruvic acid methyl ester, and D-Malic acid.
Geno-cytotoxic and pro-inflammatory properties of the gut microbiota from Winnie and C57BL/6 mice assayed with Caco-2 cells
By using customized qPCR arrays (H96 PrimePCR®) (Fig. S5), a panel of 43 marker genes was used to evaluate the geno-cytotoxic and pro-inflammatory potential of the gut microbiota from WN and WT mice on differentiated intestinal Caco-2 cells. To this purpose, stool samples from the 4–16 weeks old mice were decanted to remove coarse debris, and then used to inoculate Caco-2 differentiated cell monolayer, seeded on multi-well plates 21 days before inoculation. Inoculated and non-inoculated cells were incubated and harvested at different time points. Fluorescence microscopy assay showed that, after 3 and 6 h of infection, the mean number of bacteria adhering to Caco-2 cells was about four and two-fold higher, respectively, when cells were incubated with WN stool sample compared to WT stool sample (Fig. 4A-B). To further investigate this evidence, we tried to detect and quantify the bacterial peptidoglycan in WT- and WN-infected cell extracts, as a direct measurement of Caco-2 adherent bacteria (Fig. 4C; Fig S6). Peptidoglycan detection and quantification, performed by immunoblot analysis, revealed that in Caco-2 cell extracts, treated with WN fecal samples, the peptidoglycan level was about 0.4-fold higher than that detected in Caco-2 cell extracts incubated with WT fecal samples (Fig. 4C; Fig S6).
The expression of the 43 marker genes was determined 9 h after inoculation, as at that time the pro-inflammatory response could be better assessed and the cells were still viable. Results demonstrated significant up-regulation of mRNA levels of 14 genes in the Caco-2 cells after 9 h of exposure to WN stool sample compared to the same cells exposed to WT stool sample. These genes include those coding for RAC-alpha serine/threonine-protein kinase (AKT1), breast cancer type 1 susceptibility protein (BRCA1), checkpoint kinase 1 (CHEK1), checkpoint kinase 2 (CHEK2), claspin (CLSPN), histone family member X (H2AFX), interleukin 1 beta (IL1B), double-strand break repair protein MRE11 (MRE11A), poly [ADP-ribose] polymerase 1 (PARP-1), cell cycle checkpoint protein RAD17, DNA repair protein RAD50, tumor protein P53 (TP53), and X-Ray Repair Cross Complementing 5 (XRCC5) protein Ku80 (Fig. 5A; Table S5). No significant changes were detected in the expression of the remaining 34 genes tested. It may be seen that most of up-regulated genes are involved in cell cycle regulation, and response to DNA damage, in addition to IL1B whose gene product plays a key role in the inflammatory response.
Geno-cytotoxic and pro-inflammatory properties of the gut microbiota from Winnie and C57BL/6 mice assayed with Caco-2 and PMA-differentiated THP-1 co-culture system
To better simulate the gut’s complex microbiome environment, a cell co-culture model was implemented using Caco2 cells and THP-1, a human monocytic cell line. Caco-2 cells were seeded in co-culture with THP-1 cells in the Transwell system, as previously described. To verify cell monolayer integrity, TEER measurement was performed and only wells with values up to 350 Ω cm2 were used as this is the threshold proposed for Caco-2 cell line . Moreover, the actin and E-cadherin immunostaining was performed on Caco-2 cells on filters after 9 h of infection. As shown in Figure S7, the epithelial integrity was conserved also after bacterial infection. Inoculated and non-inoculated cells were incubated and harvested at 9 h and 12 h.
In the co-culture system, where the Caco-2 cells expose only the apical domain to the microbes, the mRNA expression pattern of the 43 selected genes at 9 h in Caco2 cells (Fig. 5B; Table S6) was different compared to that observed with Caco-2 alone at the same time point (Fig. 5A). Indeed, all the genes up-regulated in the previous infection did not change their expression in this system, while we observed up-regulation of caspase-9 (CASP9), Toll-like receptor 3 (TLR3) and Toll-like receptor 4 (TLR4) mRNAs, and a down-regulation of IL1B mRNA after exposure to WN stool sample compared to the same cells exposed to WT stool sample. Gene expression was also evaluated after 12 h of exposure to the microbes, and we observed that the gene expression variations observed at 9 h were greatly enhanced in most of cases. Furthermore, compared to the Caco-2 cells exposed to WT stool sample, in the cells exposed to WN stool sample we observed up-regulation of Autofagy Related 7 (ATG7), ATM Serine/Threonine Kinase (ATM), BRCA1, Interferon Gamma (IFNG), MRE11A, Mitochondrially Encoded Cytochrome C Oxidase II (MT-CO2), Mechanistic Target Of Rapamycin Kinase (mTOR), Nuclear Factor Kappa B Subunit 1 (NFKB1), RAD17, RB Transcriptional Corepressor 1 (RB1), TAR DNA Binding Protein (TARDBP) and Tumor Necrosis Factor (TNF) mRNAs, and down-regulation of ABL Proto-Oncogene 1 (ABL1), AKT1, Annexin A5 (ANXA5), CHEK2, CLSPN and Ras-Related Protein Rab-7a (RAB7A) mRNAs (Fig. 5B).
The same analysis was performed on THP-1 cells stimulated by Caco-2 cells seeded on the filter support on top of THP-1 cells and exposed to the microbes. After 9 h of infection, an up-regulation of TLR3 mRNA was observed compared to control, while the CHEK2 mRNA could not be detected in THP-1 cells exposed to WN stool sample. After 12 h of infection, the mRNA levels of TLR3 returned comparable to control. Also, in this case the CHEK2 mRNA could not be detected in cells exposed to WN stool sample. Finally, ATM, BRCA1, IL1B, IL8, MDC1, MRE11A and TP53 mRNAs were not detected (Fig. 5C; Table S7).