In poultry, the probiotic Bacillus subtilis 29784 (Bs29784) sustains intestinal health, enhancing animal resilience and performance through the production of the bioactive metabolites hypoxanthine (HPX), Niacin (NA), and pantothenate (PTH). Here, using enterocyte in vitro models, we determine the functional link between these metabolites and the three pillars of intestinal resilience: immune response, intestinal barrier, and microbiota. To this end, we evaluated in vitro Bs29784 vegetative cells, spores, and metabolites capacity to modulate global immune regulators (using HT-29-NF-κB and HT-29-AP1 reporter cells), intestinal integrity (HT-29-MUC-2 reporter cells and Caco-2 cells), and cytokine production (Caco-2 cells). Finally, we simulated chickens’ intestinal fermentations to determine the effect of Bs29784 metabolites on the microbiota and their fermentation profile.

Bs29784 vegetative cells reduce the inflammatory response more effectively than spores, indicating that their benefit is linked to metabolic activity. To assess this hypothesis, we studied individually Bs29784 metabolites. The results show that each metabolite had different beneficial effects. PTH and NIA reduced the activation of the proinflammatory pathways AP1 and NF-κB. HPX upregulated mucin production by enhancing MUC2 expression. HPX, NA, and PTH increased cell proliferation. PTH and HPX increased epithelial resilience to an inflammatory challenge by limiting permeability increase.

In cecal fermentations NA increased acetate, HPX increased butyrate while PTH increased acetate, butyrate, and propionate. In ileal fermentations, PTH increased butyrate. All molecules lead changes to microbiota explaining the different fermentation patterns. Altogether, we show that Bs29784 modulates intestinal health by acting on the three lines of resilience via its secreted metabolites.

All experiments were conducted according to the European Union Guidelines of Animal Care and legislation governing the ethical treatment of animals, and investigators were certified by the French government to conduct animal experiments. The Center for Expertise and Research in Nutrition facilities (Malicorne, France) are in accordance with the agreement no. C 03 159 4 of 6 November 2008, relative to experimentation on vertebrate living animals (European regulation 24/11/86 86/609 CEE; Ministerial decree of 19 April 1988). Twelve 28-days old broilers chickens were euthanized and their cecal and ileal contents were collected and pooled and immediately stored at -80°C.

Cecal and ileal chicken microbial fermentation was performed in Hungate tubes in anaerobic buffer prepared according to Davies et al. (2000). Four technical replicates were performed by condition. The anaerobic buffer was composed of 5 solutions (A, B, C, D and E) prepared individually: solution A (per Liter: 5.7 g Na₂HPO₄, 6,2g KH₂PO₄ and 0,6g MgSO₄-7H₂O), solution B (per Liter: 4 g NH₄HCO₃ and 35 g NaHCO₃), solution C (per 10mL: 132 mg CaCl₂-2H₂O, 100 mg MnCl₂-4H₂O, and 80 mg FeCl₃-6H₂O), solution D (per Liter: 0.1 % resazurin) and solution E (per 100mL: 4ml de NaOH 1M, 625mg de Na₂S). The anaerobic buffer was assembled in anaerobic condition the day of the experiment (gazed with a mixture of CO₂ and N₂): 0,01 % of solution A, 25,3% of solution B, 25,3% of solution C, 0,1 % of solution C, 49,29% of ultra-pure water (18,2mΩ) and autoclaved in the presence of 0,5g/L of L-cysteine (reducing agent). On the day of the experiment, the buffer was further reduced by adding 4% (v/v) of solution E before adding the cecal inoculum (final pH: 7.5 and Eh: -150mV). The cecal content was mixed at 5% (w/v) with the anaerobic buffer and 10 mL of the slurry was transferred in Hungate tubes. All treatments were added at 1mM. The fermentation was then performed in water bath under constant agitation (200 rpm) at 39°C for 24h. At the end of the fermentation, samples were collected to quantify microbiota composition (Illumina sequencing, Genoscreen, France). 

Sequencing of 16s was done on Illumina platform at GenoScreen using the Metabiote kit on the V3-V4 16s hypervariable region. Briefly, DNA was extracted, normalized and the multiplex library was prepared for Illumina Miseq Paired-end 2x300 bases. Quality control of the sequencing was performed using a mock community (15 bacterial and 2 archaeal strains) including in the sequencing run. The primer and index were identified (100% homology) and removed to create demultiplexed fastq files. The fastq files were quality trimmed at Q30 at the end of the read, the reads were then paired assembled with minimum 30bp alignment at 97% homology using Qiime. The demultiplexed, quality trimmed and assembled read were then clustered using DADA2 software.