Canine Mucosal Artificial Colon: development of a new colonic in vitro model adapted to dog sizes

Abstract

Differences in dog breed sizes are an important determinant of variations in digestive physiology, mainly related to the large intestine. In vitro gut models are increasingly used as alternatives to animal experiments for technical, cost, societal, and regulatory reasons. Up to now, only one in vitro model of the canine colon incorporates the dynamics of different canine gut regions, yet no adaptations exist to reproduce size-related digestive parameters. To address this limitation, we developed a new model of the canine colon, the CANIne Mucosal ARtificial COLon (CANIM-ARCOL), simulating main physiochemical (pH, transit time, anaerobiosis), nutritional (ileal effluent composition), and microbial (lumen and mucus-associated microbiota) parameters of this ecosystem and adapted to three dog sizes (i.e., small under 10 kg, medium 10-30 kg, and large over 30 kg). To validate the new model regarding microbiota composition and activities, in vitro fermentations were performed in bioreactors inoculated with stools from 13 dogs (4 small, 5 medium, and 4 large). After a stabilization period, microbiota profiles clearly clustered depending on dog size. Bacteroidota and Firmicutes abundances were positively correlated with dog size both in vitro and in vivo, while opposite trends were observed for Actinobacteria and Proteobacteria. As observed in vivo, microbial activity also increased with dog size in vitro, as evidenced from gas production, short-chain fatty acids, ammonia, and bile acid dehydroxylation. In line with the 3R regulation, CANIM-ARCOL could be a relevant platform to assess bilateral interactions between food and pharma compounds and gut microbiota, capturing inter-individual or breed variabilities. KEY POINTS: • CANIM-ARCOL integrates main canine physicochemical and microbial colonic parameters • Gut microbiota associated to different dog sizes is accurately maintained in vitro • The model can help to move toward personalized approach considering dog body weight.

Keywords: Body weight; Breed; In vitro gut model; Large intestine; Microbiota; Mucus.

Fig. 1

Overview of experimental design in the CANIM-ARCOL model. Once adapted to three dog sizes conditions, the CANIM-ARCOL was inoculated with fecal samples from 13 healthy dogs (n = 4 small in green, n = 5 medium in yellow and n = 6 large in orange) and fermentations were run for 21 days. Age and weight of all dogs involved in the study (males and females are represented by square and circle, respectively) were plotted, and significant differences were analyzed by one-way ANOVA (mean ± SD, single asterisk: p < 0.05; triple asterisk: p < 0.001; quadruple asterisks: p < 0.0001). Samples were regularly collected in the atmospheric phase, in the luminal compartment and from mucin beads to monitor microbiota composition and metabolic activities.

Fig. 2

Impact of dog sizes on α and β-diversity of microbial populations in the CANIM-ARCOL model Fermentations were performed in the CANIM-ARCOL under three dog size conditions. Lumen and mucus-associated microbiota composition were analyzed by 16S rRNA Metabarcoding and diversity indexes were calculated based on ASV table. Only stabilized points (from days 10 to 21) are represented. Alpha diversity (observed ASVs and Shannon index) is represented as box plots in the luminal medium (a) and mucin-beads (b), with significant differences indicated by different letters (p < 0.05). Redundancy analysis (RDA) two-dimension plot visualizations reported bacterial community β-diversity, showing the effects of fermentation time (c), dog size (e), or donor effect (f). Size effect was removed (partial-RDA) to visualize the impact of colonic microenvironment, i.e., luminal medium or mucin beads (d). For luminal samples only, corresponding SCFA concentrations were added as environmental variables and RDA was recalculated accordingly (g). Samples from luminal medium are represented in circles while mucin beads are in squares. Numbers correspond to dog ID.

Fig. 3

Impact of colonic microenvironment and dog size on microbiota composition in the CANIM-ARCOL. Fermentations were performed in the CANIM-ARCOL under three dog size conditions. Lumen and mucus-associated microbiota compositions were analyzed by 16S rRNA Metabarcoding and differential analysis were further performed at the ASV level. Only stabilized points (from days 10 to 21) are represented. sPLS-DA analysis was performed to generate loading plots of the 15 most contributing ASVs between the luminal medium and mucin beads -all size confounded- (a) and between sizes -whatever the microenvironment- on component 1 (d) and 2 (e). Bars are colored according to the group in which the median abundance is maximal, which for each ASV, the relative abundancy indicated in grey. Species annotations are provided when a sequence identity percentage higher than 97% was identified using BLAST (given into bracket). Venn diagrams based on ASV repartition were also generated on both luminal medium (b) and mucin beads (c). ASV numbers and corresponding percentages (sequence number of ASV over total sequence number) are indicated.

Fig. 4

Microbiota composition of luminal medium and mucin beads at the family level. Fermentations were performed in the CANIM-ARCOL under three dog size conditions. Lumen and mucus-associated microbiota composition were analyzed by 16S rRNA metabarcoding at the family level. The most 30 abundant families are represented.

Fig. 5

Impact of three dog sizes on microbiota composition at the phylum and family levels. Fermentations were performed in the CANIM-ARCOL under three dog size conditions. Lumen and mucus-associated microbiota composition were analyzed by 16S rRNA metabarcoding. Significant impacts of dog sizes on each bacterial population are indicated at the phylum (a) and family (b) levels (one-way ANOVA, double asterisks: p < 0.01; triple asterisks: p < 0.001; quadruple asterisks: p < 0.0001).

Fig. 6

Impact of dog sizes on microbiota activity in the CANIM-ARCOL. Fermentations were performed in the CANIM-ARCOL under three dog size conditions. Samples were regularly collected from atmospheric phase to determine total gas production in milliliters (a) and gas composition in relative percentages depending on dog size conditions (b) or type of gas (c). The three main short-chain fatty acids (d, e, f), the six major branched chain fatty acids (g, h, i), ammonia (j), and main primary and secondary bile acids (k, l, m) were quantified in the luminal medium. Results are expressed as mean daily concentrations in mM ± SD (d, f, g, i, k, m) or relative percentages (e, h, l). All stabilized points (from 10 to 21 days) are represented for gas and SCFA, while only end points (from 18 to 21 days) are kept for BCFA, ammonia and BA. BA, bile acids; BA I, primary bile acids; BA II, secondary bile acids; BCFA, branched-chain fatty acids; CA, cholic acid; CDCA, chenodeoxycholic acid; CH₄, methane; CO₂, carbon dioxide; DCA, deoxycholic acid; H₂, dihydrogen; I-LCA, Isoallo-3-ketocholate; LCA, lithocholic acid; N₂, nitrogen; O-LCA, 3-oxolithocholic/dehydrolithocholic acid; O₂, dioxygen; SCFA, short-chain fatty acids. Statistical differences are indicated by letters or single asterisk: p < 0.05, double asterisks: p < 0.01; triple asterisks: p < 0.001; quadruple asterisks: p < 0.0001 (one-way ANOVA).