A flexible high-throughput cultivation protocol to assess the response of individuals' gut microbiota to diet-, drug-, and host-related factors

Abstract

The anaerobic cultivation of fecal microbiota is a promising approach to investigating how gut microbial communities respond to specific intestinal conditions and perturbations. Here, we describe a flexible protocol using 96-deepwell plates to cultivate stool-derived gut microbiota. Our protocol aims to address gaps in high-throughput culturing in an anaerobic chamber. We characterized the influence of the gas phase on the medium chemistry and microbial physiology and introduced a modular medium preparation process to enable the testing of several conditions simultaneously. Furthermore, we identified a medium formulation that maximized the compositional similarity of ex vivo cultures and donor microbiota while limiting the bloom of Enterobacteriaceae. Lastly, we validated the protocol by demonstrating that cultivated fecal microbiota responded similarly to dietary fibers (resistant dextrin, soluble starch) and drugs (ciprofloxacin, 5-fluorouracil) as reported in vivo. This high-throughput cultivation protocol has the potential to facilitate culture-dependent studies, accelerate the discovery of gut microbiota-diet-drug-host interactions, and pave the way to personalized microbiota-centered interventions.

Keywords: 16S rRNA gene sequencing; anaerobic chamber; ex vivo; fecal microbiota; fibers; in vitro; microbial ecology; personalized microbiota response; short-chain fatty acids; xenobiotics.

Figure 1

Overview of the high-throughput cultivation protocol and effect of the anaerobic chamber atmosphere on the medium’s chemistry and microbial physiology; (A) overview of the step-by-step cultivation protocol adapted for the anaerobic chamber setup; in a 96-deepwell plate, heat-stable/sensitive supplements (1) are mixed with HCl-titrated bYCFA (2) and inoculated with diluted feces (3); (B) characterization of pH dynamics of CO₂ and N₂-flushed bYCFA (uninoculated) in a 96-deepwell plate in an anaerobic chamber; (C) relationship between pH, aqueous NaHCO₃ concentration, and gaseous CO₂ levels (~10% in the anaerobic chamber, ~100% in gas-tight tube) in bYCFA after 48-h equilibration; (D) change of pH over time upon titration of bYCFA with HCl, with subsequent pH increase over 48 h; the resulting titration curve (at time 0 h) can be used to adjust the pH to different values; (E) pH and growth (OD₆₀₀) during 48-h batch fecal cultivation in bYCFA (initial pH 6.8 and 5.8; containing 4 g/l NaHCO₃ and 3 g/l resistant dextrin) titrated with HCl immediately before inoculation (donor CDE; 1% inoculation of 10⁻⁴ fecal dilution); data are presented as the mean and standard deviation of three technical replicates; all pH dynamic analyses were conducted in the 60 central experimental wells to circumvent the edge effect (Supplementary File 1, Step 4).

Figure 2

Impact of the cultivation technique (96-deepwell plates vs. gas-tight tubes) on the metabolism and composition of ex vivo cultures (donor ABX and BCY) cultivated in bYCFA for 48 h, 37 °C, in the presence of resistant dextrin (3 g/l), soluble starch (3 g/l), or H₂O as control (technical triplicates); (A) experimental setup for comparing the plate- and tube-based techniques; all analyses were performed after 48h cultivation; (B) correlation of observed ASVs; points represent means, and bars represent standard deviations; (C) PCoA plot of Bray–Curtis distances with 95% confidence ellipses (calculated separately for each donor) for the two cultivation techniques; (D) taxa that are significantly different between each fiber (dextrin, starch) and H₂O (P < .05); each data point corresponds to the median difference of the clr-abundance (genus-level) between plate and tube; (E) correlation of the total metabolite production; total metabolites were calculated by summing the medium-corrected concentrations of organic acids (SCFA and intermediates); (F) correlation of individual metabolites; points represent means of blank corrected concentrations and bars represent standard deviations.

Figure 3

Formulation of a growth medium maintaining the microbial composition of healthy adult feces in ex vivo cultures during high-throughput cultivation (48 h, 37 °C) in bYCFA medium; (A) experimental setup for determining the optimal C-source mix to maintain the composition of the original feces in cultures; all analyses were performed after 48-h cultivation; (B) number of observed ASVs and Shannon-Index in feces and cultures with different C-sources; (C) Bray–Curtis distances between the feces and the respective culture; (D) visual representation of the community composition (top) and the shared ASVs between feces and cultures (bottom); taxonomic relative abundance values are provided in Supplementary File 2; (E) evaluation of the impact of additional N-source supplementation (amicase and yeast extract) on Bray-Curtis distances and on the difference in Enterobacteriaceae clr-abundance as compared to the feces; all analyses were performed on pooled samples from three independent replicates for each donor microbiota; a total of 11 distinct donors were used, i.e. AXZ, BFW, GCB, JHB, KPJ, SUT, TLM, WPL, KHG, OIP, and ZTR; multiple comparisons across all conditions were performed using a t-test and P-values were corrected using the Holm–Bonferroni method; statistically significant results are marked by stars, with ^* indicating P ≤ .05, ^**P ≤ .01, and ^***P ≤ .001.

Figure 4

Treatment with dietary fibers (donor n = 8) or drugs (donor n = 7) elicits physiologically relevant changes in the ex vivo cultures’ metabolism and community composition (bYCFA, 48 h, 37 °C); (A) experimental setup to test the fiber-specific response of fecal cultures. All analyses were performed after 48-h cultivation; (B) SCFA production by fecal cultures when bYCFA was supplemented with resistant dextrin (3 g/l), soluble starch (3 g/l), or no C-source (H₂O as control); (C) median clr-differences of taxa significantly differed between each fiber (dextrin, starch) and H₂O; genera significantly different (P ≤ .05) are labeled and located above the dotted red line; (D) experimental setup to test the effect of drugs (i.e. omeprazole, ciprofloxacin and 5-FU) as compared to the vehicle control (DMSO or H₂O) in bYCFA supplemented with 6C + Muc; all analyses were performed after 48h cultivation; (E) number of observed ASVs as compared to the vehicle control; (F) Bray–Curtis distances between controls and the drug-treated cultures; (G) Clr-differences of the 15 most abundant families; significantly altered families, as compared to the control, are marked by stars; all the analyses were performed on pooled samples from incubations for each donor microbiota; multiple comparisons across all conditions were performed using a t-test, and P-values were corrected using the Holm–Bonferroni method; statistically significant results are marked by stars, with ^* indicating P ≤ .05, ^**P ≤ .01, ^***P ≤ .001, and ns, non-significant.