Gut microbial metabolites as key mediators in inflammatory bowel disease

Abstract

BACKGROUND

Inflammatory bowel disease (IBD) is characterized by inflammation of the intestinal mucosa, increased intestinal permeability, and impaired immune regulation, all of which are accompanied and aggravated by gut microbial dysbiosis. Disturbed microbial composition contributes to the pathogenesis of IBD by altering the production of microbial metabolites, which may either alleviate or exacerbate IBD progression. The main feature of metabolic dysbiosis in IBD is impaired microbial synthesis of biologically active compounds characteristic of eubiosis, including short-chain fatty acids. The use of a combination of these metabolites in IBD patients appears promising due to its potential synergistic effects on host health and microbiota composition.

AIM

To evaluate the effect of bacterial metabolites composition (butyric, propionic, valeric acids) on in vitro and in vivo IBD models.

METHODS

We used Caco-2 cells exposed to lipopolysaccharides and Mucin-2 knockout mice as in vitro and in vivo IBD models. Caco-2 cells were exposed to lipopolysaccharides with metabolites for 24 hours. Muc2^-/- mice were given metabolites via oral gavage daily for 1 week. C57BL/6 mice were used as healthy controls. Immune cells were analyzed by flow cytometry; intestinal barrier integrity by fluorescein isothiocyanate-dextran transport; cytokine expression by real-time quantitative polymerase chain reaction; cytokine content by ELISA; and the mouse metabolome and microbiome composition by gas chromatography-mass spectrometry and 16S rRNA metagenomic sequencing.

RESULTS

Butyric acid decreased Caco-2 cell monolayer integrity and intestinal permeability in Muc2^-/- mice; propionic and valeric acids decreased gene expression and the concentration of pro-inflammatory cytokines in Caco-2 cells and in the Muc2^-/- mouse intestine; and the combination of these metabolites increased cell viability and possessed anti-inflammatory and intestinal barrier-strengthening properties. Metabolites and their combination increased the number of regulatory T cells and anti-inflammatory M2 peritoneal macrophages. Metabolite treatment increased gut microbiota biodiversity and Bacillota abundance in Muc2^-/- mice, while reducing the abundance of Thermodesulfobacteriota and Pseudomonadota, and decreased the levels of lactate, long-chain fatty acids, and 2-hydroxybutyric acid, which are markers of metabolic dysbiosis in a disturbed microbiota and dysregulated metabolism in inflamed host cells. We revealed 2-hydroxybutyric acid’s biological activity toward IBD progression: It increased intestinal permeability, inflammation, and the number of pro-inflammatory M1 macrophages; decreased cell viability and the number of regulatory T cells.

CONCLUSION

We have shown that certain bacterial metabolites may promote IBD development, while others may have a therapeutic effect, especially when used in combination.

Key Words: Inflammatory bowel disease; Gut microbial metabolites; Butyrate; Propionate; Valerate; 2-Hydroxybutyrate; Synergistic regulatory activity; Dysbiosis; Microbe-host cross-talk; Metabiotic concept

Core Tip: Gut microbial dysbiosis is a driving factor in inflammatory bowel disease (IBD) pathogenesis, mainly through a reduced capacity of the dysbiotic microbiota to synthesize beneficial metabolites, such as short-chain fatty acids, or through increased production of pathogenic metabolites, such as hydrogen sulfide, reactive nitrogen species, and lactate. We found elevated levels of 2-hydroxybutyric acid in Muc2^-/- mice’s blood and revealed its ability to worsen IBD progression in the models of the disease. Butyric, propionic, and valeric acids, especially when used in combination, had a therapeutic effect in IBD through their anti-inflammatory and immunomodulatory potential and their gut microbiota-restoring properties.

INTRODUCTION

Inflammatory bowel disease (IBD) is a multifactorial, genetically determined autoimmune disease characterized by chronic intestinal inflammation, thinning of the mucus layer, and increased permeability of the intestinal epithelium[1]. The mucus layer (composed mainly of secretory mucin 2), epithelial cells sealed by tight junctions, and mucosal immune cells play key roles in forming a barrier against external factors and in maintaining symbiotic relationships with commensal bacteria. In IBD, aberrant immune responses, such as inappropriate activation of macrophages and regulatory T cells (Tregs), along with compromised epithelial barrier integrity, lead to the uncontrolled secretion of pro-inflammatory cytokines and sustained mucosal injury[2]. Inflammatory signals in IBD promote a pro-inflammatory M1 macrophage phenotype over an anti-inflammatory M2 phenotype, and Th17 predominance over Treg lymphocytes[3,4]. Metabolism of M1 macrophages and Th17 lymphocytes is shifted to anaerobic glycolysis, which is characterized by low oxygen consumption, high glucose consumption, and high lactate release[2]. This leads to increased intestinal oxygen levels, which may contribute to the growth of potentially pathogenic facultative anaerobic bacteria and to the reduction of beneficial obligate anaerobic bacteria, including Bacteroidota (Bacteroidetes) and Bacillota (Firmicutes), in IBD animals and patients[5]. Thus, the immune system shapes the gut microbiota to be diverse and dominated by obligate anaerobic bacteria, thereby ensuring that the microbiome provides benefits to the host by generating short-chain fatty acids (SCFAs) from the digestion of complex carbohydrates[4]. Propionate and butyrate, together with acetate, are the main SCFAs (more than 95% of all the SCFAs content in humans), while valerate is a much less abundant and less studied metabolite. SCFAs are major gut metabolites, which act through binding to G-protein-coupled receptors (GPCRs), peroxisome proliferator-activated receptors gamma, or inhibition of histone deacetylases, which causes anti-inflammatory activities on intestinal epithelial cells, macrophages, dendritic cells, and promotes Tregs differentiation, and via the AMP-activated protein kinase pathway upregulate the intestinal epithelial tight junction[4,6,7]. Moreover, butyric acid serves as an energy source for colonocytes, and propionate reduces lipogenesis. It lowers serum cholesterol levels[8], and valeric acid has been shown to alleviate mucosal inflammation in IBD, primarily through anti-inflammatory effects on macrophages[9].

IBD is associated with decreased populations of SCFAs-producing bacteria and reduced SCFAs production in both IBD patients and colitis mice, while supplementation with SCFAs-producing bacteria or SCFAs showed a positive effect on colitis[10]. These changes of inflammation-induced gut microbiota composition (taxonomic dysbiosis) lead to altered microbial metabolism (metabolic dysbiosis), which are accompanied by quantitative and qualitative changes in the blood metabolome composition[11]. In this regard, studies of the metabolome in IBD, where intestinal dysbiosis plays a significant role, are of particular interest.

In recent years, there has been a trend towards an increase in the incidence of IBD among older people. Age-related chronic inflammation (“inflammaging”), decreased intestinal barrier function, and declining immune tolerance to the gut microbiota create the preconditions for the development of IBD in older adults[12]. Taking into account the increasing aging of the population, healthcare systems will face the rising burden of IBD, which requires new drugs that are particularly relevant. The search for treatment aims not only to reduce inflammation and improve damaged barrier function but also to restore dysbiotic changes in the composition and metabolic activity of the gut microbiota; this is a task of great importance.

Previously, we showed a synergistic effect of a combination of certain bacterial metabolites (carboxylic and amino acids), which were autostimulators of Escherichia coli M-17 growth, in normalizing the blood metabolomic composition of mice. This metabolite combination decreased the concentrations of 2-hydroxybutyric (2-HB), 2,3-dihydroxybutyric, and lactic acids, which were elevated in IBD patients. It increased the concentration of valeric acid and amino acids - alanine, aspartic acid, proline, isoleucine, threonine, phenylalanine, and α-aminoadipic acid, whose levels were reduced in IBD patients[13]. The combined positive effects of bacterial metabolites on host health reflect an evolutionarily developed symbiosis between hosts and bacteria. These emergent properties should be taken into account when developing a metabiotic formulation for treating IBD. Another reason for using a combination rather than a single metabolite is the concentration-dependent effects of SCFAs on host health. For example, butyrate promotes intestinal barrier function at low concentrations (≤ 2 mmol/L) but may disrupt it at high concentrations (≥ 5 mmol/L) by inducing apoptosis[14]. Therefore, we hypothesized that, rather than increasing metabolite concentrations to enhance the effect, using a combination of metabolites may be a promising approach for treating IBD.

Previously, we have shown that in vitro and in vivo models of IBD exhibit similar trends in intestinal permeability and inflammatory response parameters, adequately reflect the disease’s pathogenesis, and effectively complement each other[15]. Such IBD models, both in vitro and in vivo, enable a better assessment of the anti-inflammatory and immunomodulatory potential, as well as gut microbiota-restoring properties, of new therapeutic agents, including probiotics, prebiotics, and metabiotics. In this study, we used Caco-2 cells exposed to either dextran sulfate sodium (DSS) or lipopolysaccharides (LPS) as in vitro IBD models and mucin 2 knockout mice predisposed to colitis as in vivo IBD model to compare combined biological activity of butyric, propionic and valeric acids with biological activity of individual metabolites on inflammation and permeability of the intestinal epithelial barrier, on gut microbiome and blood metabolome composition in wild-type C57BL/6 mice (parental strain) and Muc2^-/- mice on C57BL/6 genetic background.

MATERIALS AND METHODS

Reagents

Butyric, propionic, and valeric acids (BPV), 2-HB, LPS (from Escherichia coli O111: B4), fluorescein isothiocyanate (FITC)-dextran (4 kDa MW), phenazine methosulfate, hematoxylin, eosin, and alcian blue were purchased from Sigma-Aldrich (Darmstadt, Germany). Dulbecco’s Modified Eagle medium (DMEM), L-glutamine, penicillin/streptomycin, and phosphate buffered saline (PBS) were purchased from Biolot (St. Petersburg, Russia), and fetal bovine serum was purchased from HyClone Laboratories (North Logan, UT, United States). DSS (40 kDa MW) and XTT were purchased from BioChemica, PanReac AppliChem (Darmstadt, Germany). A highly specific ELISA kit for human interleukin (IL)-8 was obtained from OOO “Citokin” (Limited Liability Company) (St. Petersburg, Russia). Mouse IL-1β and Mouse IL-10 ELISA kits were purchased from Cloud-Clone Corp. (Wuhan, China). ExtractRNA reagent, MMLV kit, and primers for quantitative polymerase chain reaction (PCR) were purchased from Evrogen (Moscow, Russia), and DNase was purchased from Thermo Scientific (Waltham, MA, United States). PCR mix was purchased from Syntol (Moscow, Russia). Mice serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels were detected using kits purchased from Olvex Diagnosticum (St. Petersburg, Russia). Anti-mouse antibodies for flow cytometry were purchased from BioLegend (San Diego, CA, United States).

Cell culture and grouping

The study was conducted in the Institute of Experimental Medicine (St. Petersburg, Russia). The human colon adenocarcinoma cell line Caco-2 was obtained from the Russian Cell Culture Collection (Institute of Cytology, Russian Academy of Sciences, St. Petersburg, Russia). Caco-2 cells were cultured in DMEM medium supplemented with 100 mL/L fetal bovine serum, L-glutamine, and penicillin/streptomycin in 75 cm² cell culture flasks (Jet Biofil, Guangzhou, China) at 37 °C, 50 mL/L CO₂. To assess the effects of DSS and metabolites on cell viability, Caco-2 cells were cultured in 96-well plates (Costar, Corning, NY, United States) in complete DMEM medium. On the 10^th day after seeding, cells were divided into 3 groups and followed by 24 hours treatment, as shown below: (1) Serum-free DMEM for the control group; (2) Serum-free DMEM containing 3% DSS for the IBD group (DSS-induced colitis); and (3) Serum-free DMEM containing 3% DSS with different metabolites for the IBD treatment group. After 24 hours, cell viability was assessed using the XTT assay. For assessment effect of LPS and metabolites on barrier integrity [determination of transepithelial electrical resistance (TEER) and/or apparent permeability coefficient for monolayer permeability to FITC-dextran - P_app values for Caco-2 cell monolayer], expression levels of tight junction genes (ZO-1, Claudin-1) and pro-inflammatory cytokines [IL-8, tumor necrosis factor-α (TNF-α)], and IL-8 secretion in medium, Caco-2 cells were cultured on transwell support (with 1 μm pore size, 0.3 cm² growth area) suitable for a 24-well culture plate (Sarstedt, Nümbrecht, Germany) for 17 days. On the 18^th day after seeding, cells were divided into 3 groups and followed by 24 hours treatment: (1) Serum-free DMEM for the control group; (2) Serum-free DMEM containing 10 μg/mL LPS for the IBD group (LPS-induced colitis); and (3) Serum-free DMEM containing 10 μg/mL LPS with different metabolites for the IBD treatment group. For Caco-2 cells, the following metabolite concentrations were used: 1 mmol/L for butyric acid, 0.5 mmol/L for propionic, valeric, and 2-HB acids. Final metabolite concentrations in BPV combination were the same as when used separately. After 24 hours, TEER and/or P_app were measured, and the medium was collected to determine IL-8 content; total RNA was isolated from the cells to determine gene expression.

Mucin-2 knockout mice housing and grouping

The study was conducted in the Scientific Research Institute of Neuroscience and Medicine (Novosibirsk, Russia). Muc2^+/- mice were generated by crossing male Muc2^-/- on C57BL/6 genetic background mice to C57BL/6 females. Muc2^-/- mice (n = 31) and Muc2^+/+ (n = 5) were generated by crossing of Muc2^+/- mice. Muc2^-/- mice were used as an in vivo model of IBD; Muc2^+/+ mice (hereinafter designated as C57BL/6 mice) were used as a healthy control. All procedures were conducted in accordance with Russian legislation, Good Laboratory Practice standards (directive No. 267 from 19.06.2003 of the Ministry of Health of the Russian Federation), inter-institutional bioethical committee guidelines, and the European Convention for the protection of vertebrate animals used for experimental and other scientific purposes. All procedures were approved by the local ethics committee of the Scientific Research Institute of Neuroscience and Medicine, protocol No. 4 (16.03.2023). All animals had SPF status, which was tested quarterly in accordance with the Federation of European Laboratory Animal Science Associations’ recommendations, and they tested negative for the recommended pathogens. Genotypes of the offspring were determined using PCR and Muc2-specific gene primers using the DNA extracted from mouse tails at the age of 10 days old. Adult 12-week-old mice were housed in same-sex groups of 2-4 individuals in ventilated cages (Animal Care Systems, Centennial, United States). The housing conditions were: A 12 hours/12 hours light/dark photoperiod at 22-24 °C, 30%-60% humidity; food (BioPro, Novosibirsk, Russia) and water were provided ad libitum. The animal protocol was designed to minimize pain or discomfort to the animals. Before the experiment, Muc2^-/- mice were divided into a control group, mice that received oral gavage with PBS (n = 11) and an experimental group, mice that received oral gavage with various metabolites (n = 4). For 7 days, mice received either 200 μL of PBS or metabolite solutions in PBS via oral gavage daily. This treatment duration was selected based on our previous experiments[13]. For animal experiments, the following metabolite concentrations were used: 3 mg/mL for butyric and propionic acids; 1.5 mg/mL for valeric and 2-HB acids. Final metabolite concentrations in the BPV combination were the same as when used separately. When choosing concentrations, we relied on approximate physiological concentrations of SCFAs in the intestine[16,17]. On the 8^th day, mice received FITC-dextran via oral gavage. After 3.5 hours, blood was collected by orbital sinus puncture, fecal samples were collected for metagenomic analysis, and mice were euthanized by cervical dislocation. Then, the mice were decapitated, and blood samples were collected for metabolomic and biochemical analysis. For flow cytometry analysis, peritoneal macrophages were collected from the peritoneal cavity, and mesenteric lymph nodes (LNs) were excised. Descending colon samples were collected for histology, cytokine ELISA, and real-time PCR.

Measurements of Caco-2 cell viability

Determination of non-toxic metabolite concentrations and the minimum effective cytotoxic concentration of DSS was carried out using the XTT viability test. Caco-2 cells were seeded into 96-well plates in complete DMEM medium at a concentration of 2 × 10⁴ cells/well, 100 μL per well. When they reach 80% confluency, cells were exposed to different concentrations of metabolites (0.1-10 mmol/L) for 24 hours. For the DSS experiment, Caco-2 cells were seeded in 96-well plates in complete DMEM medium at a concentration of 2 × 10⁴ cells/well, with 100 μL per well. After cells reached confluence and differentiated (10 days after seeding), cells were incubated with DSS at 1%-7% and DSS with metabolites at nontoxic concentrations for 24 hours. DSS and metabolites were dissolved in culture media and filter-sterilized using a 0.45 μm filter. After 24 hours treatment, 50 μL of a solution containing XTT (1 mg/mL) and phenazine methosulfate (3 mg/mL) in a 400:1 ratio was added to each well and incubated for 2 h at 37 °C. Cell viability was detected by measuring absorbance at 450 nm. Serum-free DMEM was used as the blank; the control group was assumed to have 100% cell viability.

Assessment of barrier integrity of Caco-2 cells

The effect of LPS or LPS with metabolites on the integrity of the Caco-2 cell monolayer was assessed by the value of TEER and P_app. Caco-2 cells were inoculated onto transwell supports at a concentration of 1 × 10⁵ cells/well in 400 μL of medium per well, and 900 μL of medium was added to the basolateral chamber. On the 18^th day after seeding, LPS and metabolites were added to the cells for 24 hours. Before and after the experiment, the integrity of the monolayer was assessed by measuring TEER and P_app. TEER was determined using a voltammeter, as described in our previous study[15]. The resistance of the monolayer was calculated using the formula TEER = (R - R₀) × S, where S is the effective membrane area (0.3 cm²), R is the resistance of the insert with cells, Ω, and R₀ is the resistance of the insert without cells, Ω. Measurement of cell monolayer permeability was assessed by the paracellular transport of 4 kDa FITC-dextran. 0.4 mL of FITC-dextran dissolved in DMEM at 8 mg/mL was added to the apical chamber, and 0.9 mL of DMEM was added to the basolateral chamber. After 2.5 hours of incubation, the basolateral medium was collected to measure the fluorescence intensity. The apparent permeability coefficient P_app (cm/second) was calculated according to the formula: P_app = Q/∆t × 1/(A × C₀ ), where ∆Q/∆t is the change in the concentration of FITC-dextran in the basolateral chamber over time ∆t (9000 seconds), A is the area of the membrane of the insert on which the cells were cultured (0.3 cm²), C₀ is the initial concentration of FITC-dextran (8 mg/mL). The P_app values for the empty insert were set to 100% permeability. If the TEER was higher than 900 Ω × cm² and the P_app value was lower than 4 × 10^-8 cm/second (< 0.1% transmittance relative to an empty insert), cells were considered acceptable for further experiments[15].

Real-time PCR

Total RNA was isolated from Caco-2 cells and colon samples using the Extract RNA reagent via phenol-chloroform extraction and treated with DNase. Reverse transcription was performed with 1 μg of RNA using the MMLV RT kit. Real-time PCR was performed using RT-PCR Kit with the following primers for genes in Caco-2 cells (Table 1) and following primers for genes in Muc2^-/- mice (Table 2).

Table 1 Primers for genes in Caco-2 cells.

Gene	Forward primer	Reverse primer
TJP1 (ZO-1)	5’-GAGCCTAATCTGACCTATGAACC-3’	5’-TGAGGACTCGTATCTGTATGTGG-3’
CLDN1 (Claudin 1)	5’-TGGTGTTGGGTAAGAGGTTGT-3’	5’-CCAGGTACGAATTTGGTCAGG-3’
CXCL8 (IL-8)	5’-GCAGCTCTGTGTGAAGGTGCAGTTT-3’	5’-CTCAGCCCTCTTCAAAAACTTCTCC-3’
TNF (TNF-α)	5’-TCTCGAACCCCGAGTGACAA-3’	5’-TATCTCTCAGCTCCACGCCA-3’
ACTB (β-actin)	5’-AGCGGGAAATCGTGCGTGAC-3’	5’-TCCATGCCCAGGAAGGAAGG-3’

Table 2 Primers for genes in Muc2^-/- mice.

Gene	Forward primer	Reverse primer
Foxp3	5’-AGGAGACATCCATCAGGGCT-3’	5’-TGCTGCTTTAGGTGGAGTGG-3’
Tnf (TNF-α)	5’-CCCTCACACTCAGATCATCTTCT-3’	5’-GGCACCACTAGTTGGTTGTCTTT-3’
Il1b (IL-1β)	5’-AAGGAGAACCAAGCAACGACAA-3’	5’-AACTCTGCAGACTCAAACTCCA-3’
Il6 (IL-6)	5’-GCTACCAAACTGGATATAATCAGGA-3’	5’-CCAGGTAGCTATGGTACTCCAGAA-3’
Il10 (IL-10)	5’-AAGGCAGTGGAGCAGGTGA-3’	5’-CTATGCAGTTGATGAAGATGTCA-3’
Tgfb1 (TGF-β)	5’-TGACGTCACTGGAGTTGTAC-3’	5’-AAGAGCAGTGAGCGCTGAAT-3’
Tubb5 (β-tubulin 5)	5’-TGAAGCCACAGGTGGCAAGTAT-3’	5’-CCAGACTGACCGAAAACGAAGT-3’

Amplification and detection were completed using CFX96 Touch™ Real-Time PCR Detection System (Bio-Rad Laboratories Inc., Hercules, CA, United States). The total reaction volume was 20 μL per well in a 96-well plate. Gene expression was normalized to the level of mRNA of the housekeeping gene β-actin for Caco-2 cells, ΔCt = 2 ^ (Ct_ACTB mRNA - Ct_{gene of interest} mRNA), and β-tubulin for mice, ΔCt = 2 ^ (Ct_Tubb5 mRNA - Ct_{gene of interest} mRNA).

ELISA measurements

After 24 hours of exposure of Caco-2 cells to LPS or LPS with metabolites, and assessment of barrier integrity, medium from the upper chamber was collected to determine the pro-inflammatory IL-8 content by ELISA. To measure cytokine levels in the mouse colon, a descending colon sample was homogenized in liquid nitrogen, 100 μL PBS was added, and then the samples were centrifuged at 12000 rpm for 15 minutes at 4 °C. Cytokine concentration in the supernatant was measured using the Mouse IL-1β ELISA kit and Mouse IL-10 ELISA kit according to the manufacturer’s recommendations. Cytokine concentration was normalized to total protein, measured as described by Bradford, and expressed as pg of cytokine per mg of total protein.

Histological analysis of colon samples

Descending colon samples were fixed in 10% neutral buffered formalin and embedded in paraffin. For analysis, 5-μm-thick sections were prepared using a Rotary 3003 PFM microtome (PFM Medical AG, Germany), deparaffinized, and stained with hematoxylin and eosin to assess the overall tissue morphology and with Alcian blue to identify goblet cell secretion. Images were taken with a Leica DM 750 microscope and HI PLAN 40×/0.65 objectives, using a Leica ICC 50 color camera (Leica Microsystems, Wetzlar, Germany).

Intestinal permeability assay

Intestinal permeability of Muc2^-/- and C57BL/6 mice was measured by quantifying the concentration of 4 kDa FITC-dextran in the blood. 100 μL FITC-dextran (20 mg/mL in PBS) was administered by oral gavage using a steel feeding tube. After 3.5 hours, 200 μL of blood was collected by orbital sinus puncture. Blood was diluted with 100 μL PBS containing 0.5% heparin and centrifuged at 3000 rpm for 15 minutes at 4 °C. 100 μL of supernatant was applied to a 96-well plate, and FITC fluorescence (485 nm/535 nm) was measured using Synergy 2 microplate photometer (BioTek Instruments Inc., Winooski, Vermont, United States). Baseline blood plasma fluorescence was determined in mice after oral gavage with PBS and subtracted from the fluorescence obtained after FITC-dextran gavage. FITC-dextran concentrations were determined from standard curves generated by serial dilutions of FITC-dextran. The amount of FITC-dextran in the blood was expressed as μg/mL.

Flow cytometry analysis

Analysis of lymphocytes in mesenteric LNs and peritoneal macrophages was performed by flow cytofluorometry. Mesenteric LNs were dissected out of the mesentery and placed in a 1.5 mL tube containing 100 μL of cold PBS. The tube was kept on ice. LNs were homogenized using a homogenizer, and the suspension was then filtered through a cell filter with a 70 μm pore diameter (BD Falcon, United States). Then the cell suspension was washed 2 times in PBS with 2% bovine serum albumin, centrifuged at 1500 rpm at +4 °C for 7 minutes, fixed, and permeabilized. Lymphocytes were divided into 2 groups. The first group was stained for CD45-Pac.blue, CD3-FITC, CD25-Apc, Foxp3-PE, CD4-perc/Cy5.5. The second group was stained for CD25-PE/Cy7, CD3-FITC, CD4-Percp/Cy5.5, CD8-APc, CD45-Am cyan, CD19-PE. Incubation with anti-mouse antibodies was performed for 30 minutes in the dark at room temperature, and samples were then analyzed to determine the percentage of Tregs (CD4⁺CD25⁺Foxp3⁺) among CD4⁺ lymphocytes using a BD FACSC anto™ II flow cytometer (BD Biosciences, Erembodegem, Belgium). Macrophages were collected from the abdominal cavity of mice using a syringe, added to wells of a 24-well plate in DMEM for 1 hour, and those that did not attach were discarded. The remaining macrophages were collected and divided into 2 groups - the first was stained for surface markers - CD45-Pac.blue, CD80-PE/Cy7, CD209-PE; the second group was permeabilized and stained for internal markers: CD45-Pac.blue, Arginase-1-APC, iNOS-PE in order to identify M1 (positive for CD80, iNOS) and M2 macrophages (positive for CD209 and Arginase-1)[18,19]. Incubation was carried out for 30 minutes in the dark at room temperature, then the samples were analyzed using a BD FACSCanto™ II flow cytometer.

Metabolomic analysis

Metabolomic analysis was conducted at the Federal State Budgetary Institution “Scientific and Clinical Center of Toxicology named after Academician S.N. Golikov of the Federal Medical and Biological Agency” (St. Petersburg, Russia). For metabolomic analysis, blood samples were collected from C57BL/6 and Muc2^-/- mice that received oral gavage with PBS (n = 4), butyric acid (n = 4), propionic acid (n = 4), valeric acid (n = 4), and a BPV combination (n = 4). The analysis was carried out, as previously[11], by gas chromatography-mass spectrometry (GC-MS) using a GCMS-QP2010 Plus (Shimadzu, Kyoto, Japan) instrument on a TR-5MS analytical capillary column 30 m long (Thermo Fisher Scientific, Waltham, MA, United States), with an internal diameter of 0.25 mm and a stationary phase film thickness of 0.25 μm. For analysis, 0.1 mL of blood serum was mixed with 0.5 mL of acetonitrile (Kriokhrom, St. Petersburg, Russia) for 3 minutes in a vortex mixer, and then the sample was centrifuged (3 minutes at 13000 rpm). The clear supernatant was collected, and the sediment was washed with 0.5 mL of acetonitrile. Both solutions were combined, 0.4 mL was collected and transferred to a vial, and dried in a nitrogen stream using a microcompressor until a dry residue was obtained (20-30 minutes). Then trideuteromethanol ester and tridecanoic acid dissolved in methanol were added to the dry residue as an internal standard, dried, and the silylating agent N,O-bis(trimethylsilyl)trifluoroacetamide was added and incubated for 2 minutes at 80 °C. 30 μL of methylene chloride were added to the cooled test sample, and 1 μL of the resulting solution was injected into the chromatograph injector. Data processing was performed using GC-MS solution software (Shimadzu, Kyoto, Japan). Identification of the components of the studied samples was performed using the electronic library of mass spectra NIST08 (National Institute of Standards and Technology, United States, http://www.nist.gov). Based on a comparison of the peak areas of the identified metabolites and a standard compound of known concentration, the conditional (hypothetical) concentrations of all metabolites were calculated. To visualize metabolomics data, the median levels for each metabolite were standardized by the z-score method. The formula used to calculate the z-score was z = (x - μ)/σ, where x, μ, and σ correspond to specific scores, means, and standard deviations, respectively. The computed Z score was used to plot a heatmap.

Metagenomic analysis

Metagenomic analysis was carried out using the equipment of the resource center “Genomic Technologies, Proteomics and Cell Biology” at the Federal State Budgetary Scientific Institution “All-Russian Research Institute of Agricultural Microbiology” (St. Petersburg, Russia). The composition of the microbiota was investigated using metagenomic sequencing of 16S rRNA V3-V4 × 10000 on the Illumina MiSeq 250PE platform (San Diego, CA, United States). For metagenomic analysis, feces were collected from C57BL/6 and Muc2^-/- mice, which received oral gavage with PBS (n = 4), butyric acid (n = 4), propionic acid (n = 4), and valeric acid (n = 4). The analysis was carried out, as previously[20] and consisted of the following stages: DNA extraction from samples, preparation of amplicon libraries of 16S rRNA gene fragments of all provided samples with universal primers F515/R806 for the variable region of the 16S rRNAv3-v4 gene, analysis of the nucleotide sequence of the obtained PCR fragments by high-throughput sequencing using Illumina technology to obtain at least 10000 reads for each library, primary bioinformatics processing of the obtained data, including determination of the taxonomic structure of the bacterial and archaeal microbiome in the biological material samples. To visualize metagenomic data, median levels for each bacterium were used to plot a heatmap.

Statistical analysis

Statistical analysis and data visualization were performed using GraphPad Prism software (version 10.3.1, GraphPad Software, United States). The normality of the data was assessed by the Shapiro-Wilk test. Not normally distributed data were processed using the nonparametric Mann-Whitney U-test to assess the reliability of differences between control and IBD groups; Kruskal-Wallis test with post hoc Dunn’s multiple comparisons test, corrected with Benjamini-Hochberg procedure to assess the reliability of differences between metabolite groups and control. Differences between median values were considered significant at P < 0.05. All in vitro experiments were performed in three or four independent experiments, and all in vivo experiments were performed in one or two independent experiments.

RESULTS

Effects of different concentrations of metabolites on Caco-2 cells’ viability

Determination of non-toxic concentrations of metabolites, cytotoxic concentration of DSS, and the effect of metabolites on cell viability against the background of the damaging effect of DSS were assessed using the XTT test (Figure 1). Caco-2 had the highest (optimal) cell viability at metabolite concentrations of 1 mmol/L for butyric acid and 0.5 mmol/L for propionic and valeric acids (Figure 1A). These concentrations were used for subsequent experiments on Caco-2 cells. The final metabolite concentrations in the BPV combination were the same as those selected separately. Cell viability was reduced at a metabolite concentration of 10 mmol/L (P < 0.05).

Figure 1 Viability of Caco-2 cells (% of control). A: Caco-2 cells treated with various concentrations (0-10 mmol/L) of metabolites; B: Caco-2 cells treated with different concentrations (0-7%) of dextran sulfate sodium (DSS); C: Caco-2 cells treated with metabolites together with 3% DSS, 3% DSS considered as 100% cell viability. Values are shown as medians and interquartile range (from min to max). All assays were performed in four independent experiments, each consisting of 5-10 biological replicates per group. ^aP < 0.05, ^bP < 0.01 vs the control, and ^cP < 0.05 vs cells treated with butyric, propionic, and valeric acids combination, Kruskal-Wallis test with post hoc Dunn’s test, corrected with Benjamini-Hochberg procedure. DSS: Dextran sulfate sodium; But: Butyric acid; Prop: Propionic acid; Val: Valeric acid; BPV: Combination of butyric, propionic, and valeric acids.

DSS-treated Caco-2 cells as an in vitro model of IBD

3%-7% of DSS decreased Caco-2 viability by 23%-32% (P < 0.05); therefore, 3% DSS was chosen for the IBD cell model (DSS-induced colitis) (Figure 1B). BPV, added to the cell culture medium with 3% DSS, increased Caco-2 viability by 5% (P < 0.05) compared to the control (Caco-2 cells cultured with 3% DSS) (Figure 1C). Using a combination of metabolites resulted in a greater improvement in cell survival (10%, P = 0.005) than using each metabolite individually.

LPS-treated Caco-2 cells as an in vitro model of IBD

Our previous study showed that 10 μg/mL LPS (LPS-induced colitis) decreased TEER values (Figure 2A) by 57% (P = 0.0159) and increased the permeability for FITC-dextran in the Caco-2 cell monolayer (Figure 2B) by 38% (P = 0.0079), increased the expression of IL-8 and TNF-α by 2.8 and 2.3 times (P = 0.0411, P = 0.0079), decreased the expression of ZO-1 and Claudin-1 by 54 and 53% (P = 0.0303, P = 0.0260) (Figure 2C), increased the secretion of IL-8 by 27 times (P = 0.0079) compared to the control (Figure 2D), indicating that in vitro LPS-induced colitis adequately reproduces the main factors of IBD pathogenesis[15].

Figure 2 Effect of 10 μg/mL lipopolysaccharides on indicators of the intestinal barrier integrity and inflammation in Caco-2 cells. A: Transepithelial electrical resistance; B: Apparent permeability coefficient (P_app) for monolayer permeability to fluorescein isothiocyanate-dextran; C: Gene expression of tight junction proteins and pro-inflammatory cytokines; D: Interleukin-8 content in culture medium. Values are shown as medians and interquartile range (from min to max). All assays were performed in four independent experiments, each comprising 5-6 biological replicates per group. ^aP < 0.05, ^bP < 0.01 vs the control, Mann-Whitney U test. TEER: Transepithelial electrical resistance; P_app: Apparent permeability coefficient; LPS: Lipopolysaccharides; IL-8: Interleukin-8; TNF-α: Tumor necrosis factor-α.

Effects of metabolites on barrier integrity and inflammation in Caco-2 cells

To assess the influence of metabolites on intestinal barrier integrity, the paracellular flux of FITC-dextran was measured (Figure 3A). Treatment of cells with butyric acid decreased paracellular permeability for FITC-dextran in Caco-2 cells by 23% (P = 0.0410), while using BPV combination resulted in a more substantial improvement in the paracellular permeability (by 31%, P = 0.0012) compared to control (Caco-2 cells treated with 10 μg/mL LPS). To confirm which effect of metabolites on improvement of paracellular permeability was due to the change in expression of tight junction proteins, the mRNA levels of ZO-1 and Claudin-1 were determined by real-time PCR. Butyric acid increased ZO-1 and Claudin-1 expression by 2.5 times (P = 0.0173) and 1.7 times (P = 0.0383) respectively, while BPV combination, as in the case of the influence on paracellular permeability, had a greater effect on increasing ZO-1 and Claudin-1 expression (by 2.7 times, P = 0.0026 and 2.3 times, P = 0.0016, compared to control) than the butyric acid alone. To evaluate the effect of metabolites on inflammation, we assessed IL-8 and TNF-α gene expression and IL-8 secretion (Figure 3B and C) in Caco-2 cells. Even though all metabolites and particularly their combination significantly reduced the expression of proinflammatory cytokines, at the protein level, only propionic, valeric acids, and BPV combination exerted a significant anti-inflammatory effect, they reduced IL-8 secretion by 27% (P = 0.0014), 20% (P = 0.0087), and 18% (P = 0.0465), respectively. According to this data, when used alone, butyric acid was more effective at strengthening the epithelial barrier, whereas propionic and valeric acids significantly reduced inflammation. BPV combination possessed both anti-inflammatory and epithelial barrier-strengthening properties.

Figure 3 Effect of metabolites with 10 μg/mL lipopolysaccharides on the barrier integrity and inflammation in the Caco-2 cells. A: Apparent permeability coefficient (P_app) for monolayer permeability to fluorescein isothiocyanate-dextran; B: Interleukin-8 content in culture medium; C: Gene expression of tight junction proteins and pro-inflammatory cytokines. Values are shown as medians and interquartile range (from min to max). All assays were performed in three independent experiments, each with 4-9 biological replicates per group. ^aP < 0.05 vs the control (10 μg/mL lipopolysaccharides), ^bP < 0.01 vs the control (10 μg/mL lipopolysaccharides), ^cP < 0.05 vs lipopolysaccharides + combination of butyric, propionic, and valeric acids group, ^dP < 0.01 vs lipopolysaccharides + combination of butyric, propionic, and valeric acids group, Kruskal-Wallis test with post hoc Dunn’s test, corrected with Benjamini-Hochberg procedure. P_app: Apparent permeability coefficient; LPS: Lipopolysaccharides; But: Butyric acid; Prop: Propionic acid; Val: Valeric acid; BPV: Combination of butyric, propionic, and valeric acids; IL-8: Interleukin-8; TNF-α: Tumor necrosis factor-α.

Muc2^-/- mice as an in vivo model of IBD

Inflammation of the colon is often associated with reduced colon length[21], and the descending colons of Muc2^-/- mice were shorter (by 12%, P = 0.0048) than those from C57BL/6 mice (Figure 4A and B). Histological sections of descending colon (Figure 4C) show that goblet cells are poorly distinguishable and mucin secretion is depleted; crypt architecture of the colon is disrupted, and the number of cells per crypt is increased (crypt hyperplasia) in Muc2^-/- mice. Muc2 deficiency causes constitutive inflammation, characterized by increased intestinal permeability, as assessed by the FITC-dextran assay, and by expression of pro-inflammatory cytokines in the colon (Figure 4D and E). Intestinal permeability of Muc2^-/- mice was 5.8 times higher (0.67 μg/mL vs 0.11 μg/mL of FITC-dextran in blood, P = 0.0025) compared to C57BL/6 mice (Figure 4D). We observed strong up-regulation of the expression of pro-inflammatory cytokines in the colon in Muc2^-/- compared to C57BL/6 mice: IL-1β (a 12-fold increase, P = 0.0082), TNF-α (a 7-fold increase, P = 0.0159), and decreased anti-inflammatory IL-10 expression (by 66%, P = 0.0112) (Figure 4E). Muc2^-/- mice showed increase in intestinal level of IL-1β (by 60%, P = 0.0095), while level of IL-10 in colon of Muc2^-/- mice was lower (by 41%, P = 0.0381) compared to C57BL/6 mice (Figure 4F). Lowered expression level of anti-inflammatory IL-10 in Muc2^-/- mice was accompanied by suppressed Foxp3 expression (by 59%, P = 0.0120), transcriptional factor of Tregs, which are known to produce IL-10. When examining the immune status of Muc2^-/- mice, we found a decrease (by 20%, P = 0.0176) in the percentage of Tregs (CD4⁺CD25⁺Foxp3⁺) in the mesenteric LNs of mice compared to C57BL/6 mice (Figure 4G).

Figure 4 Muc2 knockout mice as an in vivo model of inflammatory bowel disease - increased intestinal permeability and colon inflammation. A and B: Differences in colon length between wild-type mice (C57BL/6) and mice with mucin 2 gene knockout (Muc2^–/–) mice; C: Hematoxylin-eosin (top panel) and alcian blue (bottom-panel)-stained colonic sections (10 × magnification) of C57BL/6 and Muc2^-/- mice; D: Intestinal permeability of C57BL/6 and Muc2^-/- mice, expressed in μg/mL of fluorescein isothiocyanate-dextran in blood; E: Expression of pro- and anti-inflammatory cytokines and Foxp3 transcriptional factor in colon of C57BL/6 and Muc2^-/- mice; F: Intestinal content of inflammatory interleukin-1β and anti-inflammatory interleukin-10 in C57BL/6 and Muc2^-/- mice; G: Regulatory T cells content in mesenteric lymph nodes in C57BL/6 and Muc2^-/- mice; H: Ratio of M1 (CD80, iNOS) and M2 (CD209, Arginase-1) peritoneal macrophages in C57BL/6 and Muc2^-/- mice. All assays were performed in two independent experiments, each consisting of 4-11 biological replicates (mice) per group. ^aP < 0.05, ^bP < 0.01 vs the control (C57BL/6 mice), Mann-Whitney U test. FITC: Fluorescein isothiocyanate; Treg: Regulatory T cells; IL: Interleukin; TNF-α: Tumor necrosis factor-α; TGF-β: Transforming growth factor β.

To characterize the immune state of peritoneal macrophages, we assessed the expression of key markers for classifying macrophages into pro-inflammatory (M1) and anti-inflammatory (M2) phenotypes. We measured CD80 and iNOS expression as markers of pro-inflammatory M1 macrophages, and Arginase-1 and CD209 expression as markers of the M2 macrophage subpopulation. Muc2^-/- mice demonstrated a notable shift in peritoneal macrophage polarization to pro-inflammatory M1 type (percentage of CD80 and iNOS-expressing macrophages were 2.1 and 3.3 times higher compared to C57BL/6 mice, P = 0.0317, P = 0.0238; percentage of CD209 and Arginase-1-expressing macrophages were 64% and 51% lower compared to C57BL/6 mice, P = 0.0286) (Figure 4H). This imbalance in Muc2^-/- mice between excessive secretion of the pro-inflammatory cytokines IL-1β and TNF-α and insufficient secretion of the anti-inflammatory cytokine IL-10, along with immune cell dysregulation (decreased Tregs and a shift toward M1 macrophage polarization), leads to an overactive inflammatory response and tissue damage.

Effects of metabolites on gut permeability and inflammation in the colon of Muc2^-/- mice

We showed in Caco-2 cells that 1 mmol/L (approximately 0.1 mg/mL) butyric acid, 0.5 mmol/L (approximately 0.04 mg/mL) propionic acid, and 0.5 mmol/L (approximately 0.05 mg/mL) valeric acid are non-toxic. The estimated oral LD50 (for a mouse weight of 20 g) is 40 mg for butyric acid, 27 mg for propionic acid, and 12 mg for valeric acid. For in vivo experiments, we chose 3 mg/mL (approximately 35-40 mmol/L) for butyric and propionic acids, 1.5 mg/mL (approximately 15 mmol/L) for valeric acid. The daily dose given to mice in a volume of 200 μL did not exceed 0.6 mg of every metabolite.

Although we chose doses that were known to be non-toxic (for healthy mice), we tested their toxicity based on the levels of ALT and AST in the blood of mice, as in IBD, impaired epithelial barrier function and mucus layer disruption greatly affect drug bioavailability, leading to unpredictable systemic exposure and possible toxicity. ALT and AST have been widely used as sensitive markers of possible tissue damage, particularly liver toxicity, for many years, both in non-clinical toxicology studies and clinical trials[22]. Metabolites in chosen concentrations didn’t alter ALT and AST levels (Figure 5A).

Figure 5 Effect of metabolites on gut permeability and inflammation in the colon of Muc2^-/- mice. A: Evaluation of metabolites toxicity by alanine aminotransferase and aspartate aminotransferase activity in blood; B: Effect of metabolites on intestinal permeability of Muc2^-/- mice, expressed in μg/mL of fluorescein isothiocyanate-dextran in blood; C: Effect of metabolites on n (%) of regulatory T cells in the mesenteric lymph nodes of Muc2^-/- mice; D: Effect of metabolites on expression of pro- and anti-inflammatory cytokines and Foxp3 transcriptional factor in colon of Muc2^-/- mice; E: Effect of metabolites on cytokine content in colon of Muc2^-/- mice; F: Effect of metabolites on expression of markers of pro-inflammatory M1 (CD80, iNOS) and anti-inflammatory M2 (CD209, Arginase-1) peritoneal macrophages. Values are shown as medians and interquartile range (from min to max). All assays were performed in one experiment, consisted of 4-11 biological replicates (mice) per group. ^aP < 0.05 vs the control (Muc2^-/- mice), ^bP < 0.01 vs the control (Muc2^-/- mice), ^cP < 0.05 vs Muc2^-/- mice treated with butyric, propionic, and valeric acids combination, Kruskal-Wallis test with post hoc Dunn’s test, corrected with Benjamini-Hochberg procedure. ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; FITC: Fluorescein isothiocyanate; Treg: Regulatory T cells; But: Butyric acid; Prop: Propionic acid; Val: Valeric acid; BPV: Combination of butyric, propionic, and valeric acids; IL: Interleukin; TNF-α: Tumor necrosis factor-α.

Among metabolites, only butyric acid was able to improve gut barrier integrity in vivo, it decreased gut permeation for FITC-dextran in Muc2^-/-mice by 62% (0.25 μg/mL vs 0.67 μg/mL of FITC-dextran in blood, P = 0.0069) compared to Muc2^-/-mice without treatment (Figure 5B). Butyric and propionic acids were shown to increase Tregs in the mesenteric LNs by 65% (P = 0.0151) and 48% (P = 0.0337) compared to Muc2^-/-mice without treatment, while BPV combination had a greater effect on increasing Tregs (by 83%, P = 0.0059) (Figure 5C). Although all metabolites and their combinations reduced the expression of the pro-inflammatory cytokine IL-1β in the colon of Muc2^-/- mice, at the protein level, only propionic acid, valeric acid, and the BPV combination exerted an anti-inflammatory effect (Figure 5D and E). Propionic acid reduced IL-1β content by 40% (1.8 pg/mg vs 3.0 pg/mg in Muc2^-/-mice without treatment, P = 0.0094), valeric acid reduced IL-1β content by 37% (1.7 pg/mg vs 3.0 pg/mg, P = 0.0291), BPV combination reduced IL-1β content by 51% (1.5 pg/mg vs 3.0 pg/mg, P = 0.0041). Propionic acid and BPV combination increased anti-inflammatory IL-10 both expression (by 2 times, P = 0.0256, P = 0.0130) and content in the colon of Muc2^-/-mice (5.3 pg/mg for propionic acid and 4.3 pg/mg for BPV combination vs 2.4 pg/mg in Muc2^-/-mice without treatment, P = 0.0020, P = 0.0426). Only propionic acid increased Foxp3 expression (by 5 times, P = 0.0054), and only valeric acid decreased TNF-α expression in Muc2^-/-mice colon (by 79%, P = 0.0181).

After all metabolites treatment, Muc2^-/- mice showed signs of M2 macrophages predominance over M1 macrophages (Figure 5F). Butyric and valeric acids exerted the greatest impact on macrophage polarization. After butyric acid administration in Muc2^-/- mice, the percentage of CD209-expressing macrophages was 2.1 times higher (P = 0.0249), and the percentage of CD80-expressing macrophages was 71% lower (P = 0.0090) compared to Muc2^-/- mice without treatment. After valeric acid administration, Muc2^-/- mice demonstrated an increase in Arginase-1-expressing macrophages (by 2.9 times, P = 0.0439) and a decrease in iNOS-expressing macrophages (by 80%, P = 0.0482) compared to Muc2^-/- mice without treatment.

Effects of metabolites on the fecal metagenome in Muc2^-/- mice

There was a substantial difference in the fecal microbiota between Muc2^-/- mice and healthy controls. Metagenomic analysis of mouse fecal samples revealed a decrease in alpha diversity (the difference in the number of sequences between groups) of intestinal bacteria in Muc2^-/- mice, consistent with reduced bacterial biodiversity in patients with IBD compared with healthy individuals[23]. Specifically, in the C57BL/6 mouse group, there were 575 sequences with non-zero read values; in Muc2^-/- mice, there were 315 sequences (of 1554 total). Metabolites were shown to increase the alpha diversity of intestinal microbiota (for propionic acid, n = 442; for butyric acid, n = 388; for valeric acid, n = 340 reads). The beta diversity (the interindividual variability among mice in each group, taken as standard deviation) in Muc2^-/- mice was 3 times higher than in healthy controls (σ = 42 vs 14). Within metabolite groups, mice became more similar to one another in microbiota composition (σ = 8 for propionic acid, 24 for butyric acid, and 26 for valeric acid). Thus, the metagenome of Muc2^-/- mice was not only ecologically depleted but also showed marked differences among individuals, which may be considered a feature of IBD[23,24]. Metabolites were shown to restore a disturbed gut microbiota, enriching bacterial species diversity and reducing differences between mice, thereby bringing the microbiota of Muc2^-/- closer to that of healthy mice.

Based on the taxonomic results, Bacillota and Bacteroidota were the most predominant phyla in mice (Figure 6A). The major gut microbial abundances (median values) observed in feces samples of C57BL/6 mice were Bacteroidota (46%), Bacillota (40%), Verrucomicrobiota (2.3%), Actinomycetota (Actinobacteria) (2.3%), Pseudomonadota (Proteobacteria) (0.7%), Thermodesulfobacteriota (0.5%), and followed by other taxa (< 1% of each). There was no difference in microbial abundances at the phylum level between C57BL/6 and Muc2^-/- mice, including Bacillota (40%), Bacteroidota (36%), and Actinomycetota (2.3%). But the abundances of Verrucomicrobiota (7.6%), Thermodesulfobacteriota (3%), and Pseudomonadota (1.3%) were higher in Muc2^-/- than in C57BL/6 mice (3-, 5-, and 2-fold, respectively; P = 0.0286). Usually, patients with IBD have a lower Bacillota/Bacteroidota ratio[25]. Surprisingly, a slightly lower abundance of Bacteroidota and no difference in Bacillota abundance were observed in Muc2^-/- mice in comparison with C57BL/6 mice, which is partly in agreement with previous findings[26]. After BPV treatment, Muc2^-/- mice demonstrated increased Bacillota and decreased Thermodesulfobacteriota and Pseudomonadota abundance. BPV increased Bacillota abundance up to 64% (P = 0.0250), 55% (P = 0.0114), and 57% (P = 0.0440), respectively. An increased Bacillota/Bacteroidota ratio may be considered a marker of IBD treatment effectiveness[25]. All metabolites decreased the abundance of Thermodesulfobacteriota and Pseudomonadota by 38%-82% (P = 0.0286).

Figure 6 The abundance of the main bacterial phyla in all studied groups of mice. A: Sum of the median level of reads for each phylum presented as cumulative bar chart; B: Abundance of the main bacterial families, genus and species in Bacillota phylum presented as heat map; C: Abundance of the main bacterial families, genus and species in Bacteroidota phylum presented as heat map; D: Abundance of the main bacterial genus and species in Actinomycetota phylum (pink underline), Verrucomicrobiota phylum (black underline), Thermodesulfobacteriota phylum (orange underline), Pseudomonadota phylum (yellow underline) presented as heat map. Values out of range are marked in dark blue color: 20% abundance for Lachnospiraceae family, 12% abundance for Muribaculaceae family, 10% abundance for Akkermansia genus. The metagenomic assay was performed in a single experiment with 4 mice per group. ^aP < 0.05 between Muc2^-/- mice vs the control (C57BL/6 mice), Mann-Whitney U test; ^bP < 0.05 between Muc2^-/- mice treated with metabolites vs the control (Muc2^-/- mice without treatment), ^cP < 0.01 between Muc2^-/- mice treated with metabolites vs the control (Muc2^-/- mice without treatment), Kruskal-Wallis test with post hoc Dunn’s test, corrected with Benjamini-Hochberg procedure. But: Butyric acid; Prop: Propionic acid; Val: Valeric acid.

We found elevated levels of gram-negative potential pathobionts in Muc2^-/- mice, including Akkermansia, Desulfovibrio, Bilophila, and Turicimonas. Increased concentrations of bacterial antigens, resulting from a high abundance of Gram-negative bacteria, may drive colitis progression by enhancing aberrant LPS signaling and mucosal inflammation in IBD[27].

Among Bacillota

Within Bacillota phylum, we identified Lactobacillaceae (15%), Lachnospiraceae (15%), Clostridiaceae (6%), Natronincolaceae (6%), Oscillospiraceae (2%), Eubacteriaceae (1%), and Erysipelotrichaceae (1%) families (Figure 6B). Many species within the phylum Bacillota exert beneficial effects because they encode a broad spectrum of enzymes for hydrolyzing different complex carbohydrates with the production of SCFAs[4]. Taxonomic intestinal dysbiosis in patients with IBD is characterized by a decrease in the abundance of anti-inflammatory microorganisms, such as Bacillota, especially butyrate-producing bacteria (BPB) from clostridial clusters IV and XIVa (families Oscillospiraceae and Lachnospiraceae, respectively)[28]. Bacillota are the major players in SCFA production, and in the microbial biotransformation of bile acids (BAs), and dysbiosis-induced deficiency of SCFAs and secondary BAs (especially the Lachnospiraceae family) in IBD patients may promote inflammation[29].

Interestingly, we observed a notable increase in the facultative anaerobic lactic acid-producing genera Lactobacillus, Ligilactobacillus, and Limosilactobacillus from the family Lactobacillaceae in Muc2^-/- mice compared to C57BL/6 mice (4.5% vs 0.1%, 7.7% vs 2.5%, 3.0% vs 0.03%, respectively; P = 0.0286). Clinical studies indicate that the abundance of Bifidobacterium and Lactobacillus decreases in the intestinal microbiota of IBD patients, but there is evidence that their abundance increases in patients with active IBD. The Lactobacillus genus is phylogenetically diverse and contains over 100 species. It was shown that some of them may exacerbate DSS-induced colitis and induce an inflammatory response in healthy tissue cultured ex vivo[30]. Lactobacillus abundance decreased after butyric and valeric acids administration in Muc2^-/- mice (up to 0.3%, P = 0.0279). Alkaliphilus genus from family Natronincolaceae was decreased in Muc2^-/- mice compared to C57BL/6 mice (0.1% vs 0.4%, P = 0.0286), and butyric acid caused a great (up to 6% abundance, P = 0.0276) increase in Alkaliphilus quantity. At present, there is limited information on the role of Alkaliphilus in IBD. There is evidence of its probiotic properties, as the abundance of the genus Alkaliphilus increased after probiotic administration, and it has also been reported to produce SCFAs[31]. Not only do patients with IBD have lower numbers of Butyricicoccus bacteria (family Oscillospiraceae) in their stools, but oral administration of Butyricicoccus pullicaecorum (B. pullicaecorum) has also been shown to attenuate trinitrobenzenefulfonic-induced colitis in rats, and the supernatant of B. pullicaecorum exerts anti-inflammatory and epithelial-strengthening effects on Caco-2 cells stimulated by TNF-α and interferon-γ. These effects are due to B. pullicaecorum ability to produce high concentrations of butyrate[23]. The gut microbiome of Muc2^-/- mice was characterized by the complete absence of this bacterium, and propionic acid administration resulted in B. pullicaecorum colonization of Muc2^-/- mice gut.

Clostridium genus (family Clostridiaceae) is well known for its beneficial species that produce SCFAs and convert primary BAs to secondary BAs, which attenuate inflammation, energize intestinal epithelial cells, and strengthen the intestinal barrier, for example, Clostridiaceae butyricum and Clostridiaceae scindens[32]. Even though most Clostridium species are commensal bacteria, Clostridium genus also includes pathogenic species, like Clostridium difficile (C. difficile), Clostridium perfingens, and Clostridium botulinum. Clostridium spp. are depleted in IBD patients[28]. The abundance of Clostridium genus in the healthy C57BL/6 mice was 1% of the total microbial community, while none of Muc2^-/- mice had these bacteria. Butyric and propionic acids increased Clostridium genus abundance (0.35%, P = 0.0386; 0.41%, P = 0.0250). We found a dramatically increased abundance of Clostridium porci (C. porci), another member of the Clostridium genus, in Muc2^-/- mice (up to 6% of the total microbiome), whereas in C57BL/6 mice, C. porci abundance was 0.2%. All metabolites reduced C. porci abundance to the C57BL/6 mouse level (up to 0.4%, P = 0.0428). C. porci was first identified in 2020 in the intestinal microbiota of pigs; it is also present in the human gut microbiota and is more prevalent in adults with depression than in healthy controls. It is still little known about the pathogenicity of C. porci and its difference from other Clostridium species, but one report has shown a pediatric case of C. porci bacteremia[33].

Butyrate-producing Eubacterium coprostanoligenes (E. coprostanoligenes, family Eubacteriaceae) level was 2 times lower in Muc2^-/- mice compared to C57BL/6 mice (1.0% vs 0.48%, P = 0.0286); after propionic acid treatment, its abundance increased (up to 1.8% of total microbiome, P = 0.0025). E. coprostanoligenes, like others members of genus Eubacterium, is known for its ability to transform cholesterol to coprostanol, thus removing cholesterol from the gut and systemic circulation[34]. E. coprostanoligenes was shown to promote mucin 2 secretion by goblet cells, thereby fortifying the integrity of the intestinal mucus barrier[35].

Although there was no difference between Muc2^-/- and C57BL/6 mice in Oscillospiraceae and Lachnospiraceae abundance, all of the metabolites tended to increase their level in Muc2^-/- mice. The prevalence of Lachnospiraceae and Oscillospiraceae in Muc2^-/- mice without treatment was 8% and 0.5%, respectively; after butyric and propionic acids administration Lachnospiraceae abundance was increased up to 20% (P = 0.0472); after valeric acid administration, Oscillospiraceae abundance was increased up to 2.8% (P = 0.0070). Lachnospiraceae family members are BAs 7-dehydroxylating bacteria, they express enzymes crucial for secondary BAs synthesis, which are uncommon among intestinal bacteria[36].

Flintibacter butyricus (family Oscillospiraceae) is a strictly anaerobic, widespread murine bacterium that can produce butyrate when growing in the presence of amino acids glutamine and glutamate[37]. Flintibacter butyricus abundance was reduced in Muc2^-/- mice compared to C57BL/6 mice (0.1% vs 0.7%, P = 0.0286). Butyric acid increased its level by 8 times (up to 0.8%, P = 0.0070), compared to untreated Muc2^-/- mice. Propionic and valeric acids also stimulated the growth of Neglectibacter timonensis (family Oscillospiraceae) up to 1.3% (P = 0.0143) and 1.2% (P = 0.0347) compared to Muc2^-/- mice without treatment (0.3%).

IBD-associated taxonomic dysbiosis is characterized by a decrease in the number of anti-inflammatory BPB bacteria, Roseburia spp. and Blautia spp. from Lachnospiraceae family[28]. Blautia level was 2 times lower in Muc2^-/- mice compared to C57BL/6 mice (0.006% vs 0.02%, P = 0.0286), and valeric acid significantly increased its level up to 2.2% (P = 0.0054). Blautia is a dominant genus in the human gut microbiota; its abundance ranges from 4%-5% of the total microbiota in healthy individuals to 3%-9% in IBD patients. Moreover, the total abundance of Blautia spp. was higher in male patients than in healthy men and tended to be lower in female patients than in healthy women[20]. At the genus level, Blautia abundance in Muc2^-/- and C57BL/6 mice was negligible to have any impact on host health, but at species level Blautia hominis and Blautia intestinalis abundance reached 3.3% and 2.9% of the total microbial community in Muc2^-/-, while none of C57BL/6 mice had these bacteria. After butyric and propionic acids administration, none of Muc2^-/-mice had Blautia hominis and Blautia intestinalis, while after valeric acid treatment, the level of these bacteria remained the same as in Muc2^-/- mice without treatment. Both Blautia overabundance and deficiency may represent functional (metabolic) dysbiosis in IBD[20,28], which may depend on differences in Blautia composition at the species or strain level. In the case of Blautia overabundance in IBD, we assume that the microbiota attempts to compensate for the lack of SCFAs but fails to deliver metabolites effectively under pathologically altered colonic conditions[20].

Interestingly, none of C57BL/6 and Muc2^-/- mice had Roseburia in their microbiota, but after butyric and propionic acids administration, Roseburia abundance increased in Muc2^-/- mice to 0.3% (P = 0.0372) and 0.4% (P = 0.0296), respectively. Clostridium scindens (C. scindens, Clostridiaceae) was significantly depleted in Muc2^-/- mice compared to C57BL/6 mice (0.07% vs 2.5%, P = 0.0286). In Muc2^-/- mice treated with propionic acid C. scindens abundance increased to 0.8% respectively (P = 0.0119). C. scindens inhibits the growth of C. difficile through synthesis of secondary BAs using 7a-hydroxysteroid dehydrogenase - enzyme, which is critical for secondary BAs biosynthesis. Very few intestinal bacteria possess a complete secondary BA synthesis pathway[36]. Thus, SCFAs may be used as therapy for C. difficile infection, as they effectively stimulate C. scindens growth.

There are conflicting data on changes in the abundance of Lachnospiraceae in IBD. It was reported that the reduced abundance of Erysipelotrichaceae and Lachnospiraceae families in TNBS-treated mice[10] and the increased abundance of Erysipelotrichaceae family in Muc2^-/- mice[26] were also evidence that increased Erysipelotrichaceae abundance was causally related to lower risks of IBD[38]. We found 2 murine species - Dubosiella newyorkensis (human homologue Clostridium innocuum) and Faecalibaculum rodentium (human homologue Holdemanella biformis), whose abundance was increased (up to 3.0%, P = 0.0068 and 8.5%, P = 0.0182) after butyric acid treatment, although none of C57BL/6 mice or Muc2^-/- mice had these bacteria. It was shown that Dubosiella newyorkensis inhibited DSS-induced inflammation in mice by inducing Foxp3⁺ Tregs and ameliorating mucosal barrier injury by producing SCFAs, especially propionate and lysine[38]. Similarly, Faecalibaculum rodentium was shown to reduce inflammation by stimulating the development of Tregs in the colon, and a decrease in Faecalibaculum levels was associated with the emergence of colorectal cancer[39].

Abundance of Eubacteriales (the order level) which included Clostridiaceae, Eubacteriaceae, Lachnospiraceae, Oscillospiraceae families, was decreased in Muc2^-/- mice (2% vs 11% in C57BL/6 group, P = 0.0286), and propionic acid increased the amount of Eubacteriales compared to Muc2^-/- mice without treatment (up to 5.8%, P = 0.0025).

Among Bacteroidota

Some bacteria of Bacteroidota, as well as Bacillota, are capable of fermenting indigestible carbohydrates to generate beneficial SCFAs, but many other members of Bacteroidota may exhibit pro-inflammatory properties by influencing pro-inflammatory cytokine production[25]. As Gram-negative bacteria, Bacteroidota endotoxins can also contribute to IBD. IBD patients exhibited an elevation in gut abundance of Bacteroidota compared to healthy individuals[10]. Within the Bacteroidota phylum we discovered Muribaculaceae (32%), Prevotellaceae (9%), Bacteroidaceae (2%), and Odoribacteraceae (1%) families (Figure 6C). Muribaculaceae, which are a more abundant family in mice than in other species, contained several genera such as Muribaculum, Duncaniella, and Sangeribacter. We found that at the family level Muribaculaceae, which is a dominant bacterial group in the mouse gut, there was no difference between Muc2^-/- and healthy mice (12% of total microbiota), while propionic and valeric acids reduced Muribaculaceae bacteria (3.7%, P = 0.0280 and 2.6%, P = 0.0025 compared to 12% in Muc2^-/- mice without treatment); however, at genus level Muribaculum was decreased in Muc2^-/- mice (0.8% compared to 1.9% in C57BL/6 mice, P = 0.0286), and butyric acid increased its level up to 2% (P = 0.0426). At species level, Muribaculum intestinale was also depleted in Muc2^-/- mice (0.2% compared to 2.6% in C57BL/6 mice, P = 0.0286). Our results are consistent with the data on the reduction of Muribaculum in murine gut microbiome after DSS treatment. After oral gavage of gut bacterial consortium producing anti-inflammatory secondary BAs, Muribaculum level was restored[40]. Although bacteria from the family Muribaculaceae are claimed to have probiotic properties via their ability to produce SCFAs from mucin glycans and exogenous polysaccharides, there is evidence that Muribaculaceae may promote inflammation via kynurenine metabolites signaling[41]. Another genus from the Muribaculaceae family, Duncaniella, was equally represented in all groups of mice, but Duncaniella dubosii was decreased in Muc2^-/- mice (0.3% compared to 2.2% in C57BL/6 mice, P = 0.0286), and butyric acid increased its level compared to Muc2^-/- mice without treatment (up to 0.9%, P = 0.0347). Duncaniella is abundant in the mouse intestine and plays a protective role in DSS-induced colitis[42]. But there is an example of pathogenic species from Duncaniella genus - Duncaniella muricolitica, which enhances inflammatory responses after epithelial damage and thus contributes to the development of IBD[43]. Sangeribacter muris (S. muris) was also depleted in Muc2^-/- mice (2.9% compared to 5.0% in C57BL/6 mice), and propionic acid restored its level to that of C57BL/6 mice. S. muris was the most dominant in the mouse gut microbiota in the phylum Bacteroidota; it was present in all mice of all groups. There is evidence that the beneficial species S. muris is associated with protection against inflammation in IBD[43].

Among Bacteroidaceae family, Bacteroides pectinophilus (B. pectinophilus) and Phocaeicola vulgatus (P. vulgatus) species were decreased in Muc2^-/- mice (0.002% and 0.08% compared to 0.7% and 0.2% in C57BL/6 group), and propionic acid increased level of B. pectinophilus up to 0.2% (P = 0.0262), and level of P. vulgatus up to 0.9% (P = 0.0143). There are indications that these bacteria may have probiotic potential: B. pectinophilus was negatively correlated with inflammatory markers, markers for insulin resistance or dyslipidemia[44]; and P. vulgatus was modulating host inflammatory responses through SCFAs production (especially butyric and propionic acids) and significantly attenuated symptoms of DSS-induced colitis in mice[45]. Otherwise, Bacteroides acidifaciens (B. acidifaciens) abundance was increased in Muc2^-/- mice compared to C57BL/6 mice (0.7% vs 0.3%, P = 0.0286), and propionic and valeric acids decreased its level (up to 0.2%, P = 0.0384 and 0.1%, P = 0.0113, respectively). Just like Akkermansia muciniphila (A. muciniphila), B. acidifaciens is also a mucin-degrading bacterium, and its high abundance in the context of colitis could exacerbate inflammation via degradation of the mucosal barrier[46].

Odoribacteraceae family has been found to be less abundant in Muc2^-/- mice (0.01% vs 0.4% in C57BL/6 group, P = 0.0286), and butyric and valeric acids restore its level (up to 0.4%, P = 0.0279; and up to 0.8%, P = 0.0089, respectively). It was shown that a member of the Odoribacteraceae family, Odoribacteraceae splanchnicus, may produce secondary BAs with antimicrobial properties, such as isoallolithocholic acid[47]. Isoallolithocholic acid enhances the differentiation of anti-inflammatory Treg cells by facilitating the formation of a permissive chromatin structure at the Foxp3 promoter region[48].

Although the level of genus Prevotella and Prevotellamassilia timonensis (P. timonensis) from family Prevotellaceae did not differ significantly between Muc2^-/- and C57BL/6 mice, after treatment with metabolites, the level of these bacteria decreased significantly and was even lower than in healthy mice. Prevotella abundance in Muc2^-/- group was 3.7%, butyric and valeric acids decreased to 0.7% and 0.6% (P = 0.0428). P. timonensis abundance in Muc2^-/- group was 1%, and propionic and valeric acids decreased its level (to 0.003%, P = 0.0087 and 0.007%, P = 0.0140). It is known that Prevotella genus aggravate local and systemic inflammation via reduction of SCFAs and IL-18 and increase in IL-1β production, while P. timonensis, through sialidase secretion, cleaves sialic acids and degrades mucin, thereby increasing intestinal barrier permeability[49].

Abundance of Bacteroidales (the order level) which included all identifiable bacteria of the phylum Bacteroidota, tended to be higher in Muc2^-/- mice (6.7% vs 3.7% in C57BL/6 group), and propionic acid reduced amount of Bacteroidales (up to 1.1%, P = 0.0089). As pointed out earlier, some members of the Bacteroidales order may contribute to etiology of IBD (inflammation, epithelial disruption). Although there are many beneficial bacteria in the phylum Bacteroidota, a decrease in their numbers may indicate IBD recovery.

Among Pseudomonadota

IBD is characterized by a notable shift to a higher abundance of Pseudomonadota - the phyla which includes pathogenic pro-inflammatory species[25]. Within Pseudomonadota phylum we discovered only 2 species - Parasutterella excrementihominis (P. excrementihominis) and Turicimonas muris (Figure 6D). The genus Turicimonas contains a single species, the pro-inflammatory Turicimonas muris (T. muris). Turicimonas positively correlates with IBD severity, including histopathological scores and levels of the pro-inflammatory cytokines IL-1β, IL-6, and TNF-α[50]. T. muris level was 7 times higher in Muc2^-/- mice compared to C57BL/6 mice (0.3% vs 0.04%, P = 0.0286), and all of the metabolites tend to decrease its level to that in healthy mice, but only propionic acid reduced T. muris abundance to 0.002% of total microbiota (P = 0.0067). There is evidence that Parasutterella might be associated with chronic intestinal inflammation. P. excrementihominis was increased in the stool of patients with irritable bowel syndrome[51]. There was no difference in P. excrementihominis level between Muc2^-/- and C57BL/6 mice, but propionic acid reduced its level compared to Muc2^-/- without treatment (0.1% vs 0.3%, P = 0.0426).

Among Thermodesulfobacteriota

Desulfovibrio porci and Bilophila wadsworthia are sulfate-reducing bacteria (SRB) of the phylum Thermodesulfobacteriota, their level in Muc2^-/- mice was higher than in C57BL/6 mice (2.5% vs 0.5%; 0.5% vs 0.1%, P = 0.0286; Figure 6D). Muc2^-/- mice after butyric acid treatment demonstrated a lowered level of Bilophila wadsworthia (up to 0%, P = 0.0045) and a reduced level of Desulfovibrio porci (up to 0.7%, P = 0.0426). Several studies have reported an increase in SRB in IBD[52]. Production of hydrogen sulfide (H₂S) has been linked to IBD, as it damages the gut epithelium’s mucus layer by breaking disulfide bonds that link mucin monomers, thereby disrupting barrier function. H₂S may trigger antibiotic resistance and may contribute to the onset of colorectal cancer[53]. A commonly used drug for the treatment of IBD, mesalamine, suppresses the growth of SRB and inhibits fecal sulfide production[52]. The lack of SCFAs increases the pH, which favors the growth of SRB. Higher concentrations of SCFAs lower pH, which favors the growth of methanogens over SRB. H₂S may increase the oxygenation in colonocytes by inhibiting β-oxidation of butyrate, and lead to the inhibition of obligate anaerobes that produce SCFAs. Excessive H₂S causes oxidative stress and energy starvation, which may lead to colonocyte death, penetration of the epithelial barrier by the intestinal microbes and their direct interaction with the mucosal immune system. The resulting inflammation leads to further disruption of the gut barrier, decreased butyrate oxidation, decreased mucosal sulfide detoxification, and the subsequent perpetuation of inflammation[54]. Given that Desulfovibrio species, which are generally described as strictly anaerobic organisms, can grow under microaerobic conditions[55], increased oxygenation caused by SRB may promote Thermodesulfobacteriota abundance.

Among Actinomycetota

Within Actinomycetota phylum we discovered Coriobacteriaceae (0.5%) and Eggerthellaceae (2%) families (Figure 6D). Bifidobacterium spp. were absent in mice, consistent with previous data[56]. Parvibacter caecicola (P. caecicola) level from the family Coriobacteriaceae was 5 times higher in Muc2^-/- compared to C57BL/6 mice (0.5% vs 0.1%, P = 0.0286), and propionic acid reduced P. caecicola level to 0.08% (P = 0.0426). P. caecicola is considered an aerotolerant pathobiont that is abundant when mice are treated with DSS or in TNF^ΔARE mice with elevated TNF levels or in mice deprived of indole-3-carbinol (AhR ligand - ameliorating chronic intestinal inflammation) in their diet[57,58]. Adlerkreutzia muris from the family Eggerthellaceae was less abundant in Muc2^-/- compared to C57BL/6 mice (0.1% vs 0.7%, P = 0.0286), and propionic acid increased Adlerkreutzia muris level to 0.3% (P = 0.0426). A substantial decrease in Adlercreutzia was found in feces in IBD patients. Adlerkreutzia genus metabolizes isoflavones, phenolic compounds with antimicrobial and anti-inflammatory properties, so a reduction in abundance may promote inflammation[59].

In Muc2^-/- mice we observed a decrease of Raoultibacter abundance (Eggerthellaceae family) compared to C57BL/6 mice (0.003% vs 0.1% of total microbiota, P = 0.0286), and butyric and valeric acids increased its level to values that were higher than those in the control C57BL/6 mice (up to 0.3% for butyric acid, P = 0.0345; up to 1.0% for valeric acid, P = 0.0032). Raoultibacter may interact with the host immune system through BAs pathways that can modulate inflammation in IBD. In the gut, host-derived primary BAs are metabolized by some bacteria including Raoultibacter and Clostridium species to the secondary BAs 3-oxo-lithocholic acid, which inhibits Th17 cell differentiation by blocking the function of the nuclear hormone receptor RORγt1[60]. It was also shown that Crohn’s disease patients treated with mesalazine, azathioprine, and infliximab, have an increased abundance of Raoultibacter[61].

Among Verrucomicrobiota

Akkermansia, a genus in the phylum Verrucomicrobiota, in Muc2^-/- mice was one of the most numerous genera of bacteria (up to 10% of the total fecal microbiota). Surprisingly, Muc2^-/- mice had an increased abundance of the mucin-degrading Akkermansia genus compared to C57BL/6 mice (7.6% vs 2.3%, P = 0.0286), and metabolites administration didn’t affect its level (Figure 6D). Although the Akkermansia we found in mice has not been classified to the species level, it is most likely A. muciniphila, a bacterium with probiotic properties, as this is the only Verrucomicrobiota species observed in mice[56]. A. muciniphila can degrade the intestinal mucin layer with SCFAs production, thus promoting differentiation of Foxp3⁺ Tregs in the host colon, relieving colitis[62]. The abundance of anti-inflammatory A. muciniphila was shown to be decreased in fecal samples from both experimental colitis mice and patients with IBD[62]. On the contrary, there is conflicting data on the increase in Akkermansia abundance in Muc2^-/- mice compared to Muc2^+/+ mice, and in mice or patients with colorectal cancer compared to healthy controls[26]. Recent studies have found that A. muciniphila is not strictly anaerobic and can tolerate a small amount of oxygen. Under microaerobic conditions, A. muciniphila growth rate could increase[62]. Reduced abundance of BPB could increase Akkermansia abundance in Muc2^-/- mice, because butyrate via GPCRs receptors and peroxisome proliferator-activated receptors gamma signaling pathway drives the metabolism of surface colonocytes towards mitochondrial β-oxidation of fatty acids with high oxygen consumption, which creates conditions for epithelial hypoxia[4]. The mucosal symbiont A. muciniphila was found to play an important role in maintaining intestinal barrier function and suppressing inflammation through the expansion of Tregs[63]. However, A. muciniphila may act as a pathobiont, overconsuming mucins which leads to thinning of the mucus layer and increased inflammation[26].

Effects of metabolites on the blood metabolome in Muc2^-/- mice

Gut microbiota dysbiosis is caused not so much by changes in the microbiome structure, but by disturbances in its metabolism; and the metabolome is a more significant predictor of dysbiosis than the taxonomic composition of the microbiome. Microbial function in IBD is more severely affected than microbiota composition: Changes affected 12% of metabolic pathways, compared to 2% of microbial genera. The most significant changes were observed in the metabolic pathways related to oxidative stress, carbohydrate metabolism, amino acid biosynthesis, and nutrient transport and absorption[28]. In addition to changes in microbial metabolism, which alter blood metabolite levels, the host’s endogenous metabolism is also altered. Thus, changes in the composition and metabolic activity of the microbiota in IBD and metabolic disorders caused by chronic inflammation are accompanied by quantitative and qualitative changes in the blood metabolome.

According to GC-MS data, 85 compounds were identified in the blood serum of all groups of mice, of which 36 compounds with non-zero values in all groups were selected. The level of almost all identified metabolites of microbial and endogenous origin in the blood of Muc2^-/- mice differed considerably from that in C57BL/6 mice (Figure 7). In serum of Muc2^-/- mice levels of the following metabolites were increased compared to C57BL/6 mice: In the hydroxycarboxylic acids group - lactic acid (84.8 vs 54.9 relative units), 2-HB (0.75 vs 0.53 relative units) and 3-hydroxybutyric acid (3-HB) (13.7 vs 8.2 relative units) (Figure 7A); in the long-chain fatty acids (LCFA) group - elaidic (25.4 vs 20.7 relative units), stearic (25.0 vs 20.2 relative units), trans-palmitoleic (7.8 vs 5.8 relative units), palmitic (89.5 vs 61.3 relative units), and myristic (2.6 vs 1.6 relative units) acids (Figure 7B); in amino acids group - isoleucine (3.0 vs 0.10 relative units) (Figure 7C); and also cholesterol (81.1 vs 28.5 relative units), phosphoric acid (12.8 vs 1.4 relative units), and urea (294.6 vs 0.90) (Figure 7D; for all comparisons P = 0.0286). And, on the contrary, the following metabolites were reduced in serum of Muc2^-/- mice compared to C57BL/6 mice: In the hydroxycarboxylic acids group - erythronic acid (0.46 vs 1.24 relative units) (Figure 7A); in the LCFA group - pentadecanoic (0.01 vs 0.07 relative units) and gondoic acids (0.75 vs 1.2 relative units) (Figure 7B); in the amino acids group - aspartic acid (0.10 vs 0.85 relative units), aminomalonic acid (0.16 vs 1.2 relative units), α-ketoisocaproic acid (0.10 vs 0.79 relative units), serine (0.14 vs 0.66 relative units), leucine (0.90 vs 3.9 relative units), threonine (0.32 vs 1.3 relative units), alanine (0.47 vs 1.5 relative units), methionine (0 vs 0.26 relative units) (Figure 7C); in group of other metabolites - α-d-glucopyranose or glucose (19.8 vs 60.8 relative units) and a-glycerophosphate (2.0 vs 5.1 relative units) (Figure 7D) (for all comparisons P = 0.0286). The concentrations of the other metabolites did not differ between Muc2^-/- and C57BL/6 mice.

Figure 7 Heatmap of the changes in metabolite concentration in the blood serum of all studied groups of mice. A: Hydroxycarboxylic acids; B: Long-chain fatty acids; C: Amino acids and their derivatives; D: Other metabolites. Red indicates an increase in metabolite concentration in the blood; green indicates a decrease. The metabolomic assay was performed in one experiment, consisted of 4 mice per group. ^aP < 0.05 between Muc2^-/- mice vs the control (C57BL/6 mice), Mann-Whitney U test; ^bP < 0.05 between Muc2^-/- mice treated with metabolites vs the control (Muc2^-/- mice without treatment), ^cP < 0.01 between Muc2^-/- mice treated with metabolites vs the control (Muc2^-/- mice without treatment), ^dP < 0.05 between Muc2^-/- mice treated with butyric, propionic and valeric acids vs Muc2^-/- mice treated with butyric, propionic and valeric acids combination, Kruskal-Wallis test with post hoc Dunn’s test, corrected with Benjamini-Hochberg procedure. But: Butyric acid; Prop: Propionic acid; Val: Valeric acid; BPV: Combination of butyric, propionic and valeric acids.

IBD exhibits chronic low-grade inflammation and dysregulated lipid metabolism. Previously, we have shown that patients with IBD have elevated levels of LCFA, 2-HB, 3-HB, and lactic acid, and decreased levels of a-glycerophosphate, alanine, threonine, and aspartic acid[1,11,64], which is consistent with the data from this study.

Among LCFA, saturated fatty acids (margaric, palmitic, myristic, stearic acids), trans fatty acids (elaidic, trans palmitoleic acid), and ω-6 polyunsaturated fatty acids (linoleic, arachidonic) exhibit pro-inflammatory effects, while monounsaturated omega-9 fatty acids (gondoic, oleic acids), ω-3 polyunsaturated fatty acids, and pentadecanoic acid display anti-inflammatory actions. Increased levels of LCFA, both saturated palmitic acid (C16:0) and monounsaturated palmitoleic acid (C16:1) and its trans-isomer trans-palmitoleic acid (trans-C16:1), may indicate an increased intensity of fatty acid biosynthesis and increased lipogenesis, which may be due to various reasons, including chronic hypoxia and dysbiosis caused by chronic inflammation in the intestine[64]. De novo lipogenesis index (palmitic acid/Linoleic acid ratio) was 1.6 times higher in Muc2^-/- mice compared to C57BL/6 mice (1.53 vs 0.94, P = 0.0286; Mann-Whitney U test). Also, Muc2^-/- mice were characterized by an increased ratio of saturated stearic acid to essential ω-6 polyunsaturated linoleic acid (by 38%, 0.43 vs 0.31, P = 0.049) and an increase in the ratio of stearic to oleic acid concentrations (by 30%, 4.59 vs 3.52, P = 0.0286) compared to C57BL/6 mice. These indicators are considered potential biomarkers of intestinal inflammation[1]. Only propionic acid lowered serum levels of most pro-inflammatory LCFA, while BPV combination increased anti-inflammatory oleic, gondoic, and pentadecanoic acids (P < 0.05). Besides, propionic acid reduced lipogenesis index (0.77 vs 1.53, P = 0.0280), while BPV combination decreased stearic to oleic acid ratio (0.41 vs 4.59, P = 0.0013).

Pentadecanoic acid (C15:0), an odd-chain saturated fatty acid, was lower in Muc2^-/- mice compared to C57BL/6 mice, and propionic acid and BPV combination restored its level to healthy control values. Pentadecanoic acid has garnered recent attention for its potential anti-inflammatory properties, particularly in inflammation-related diseases. Pentadecanoic acid may be synthesized by gut bacteria from propionic acid[64].

Interestingly, the serum cholesterol level in Muc2^-/- mice was 3 times higher than in C57BL/6 mice. This may be due to the reduction of cholesterol-reducing microbes in the Muc2^-/- mice gut (E. coprostanoligenes, Clostridium)[34]. Only propionic acid reduced (by 85%, P = 0.033) the serum cholesterol level, and it was the most potent activator of the growth of cholesterol-reducing bacteria. Also, propionic acid has been shown to exert hypolipidemic and hypocholesterolemic effects via an immunomodulatory pathway, increasing Tregs and IL-10 levels in the intestine and thereby downregulating intestinal cholesterol transporters[65]. As propionic acid was the most potent activator of IL-10 secretion in Muc2^-/- mice intestine, it also explains the hypocholesterolemic action of propionate.

As expected, Muc2^-/- mice had increased levels of lactic acid in the blood. This is not only due to increased lactic acid bacteria in Muc2^-/- mice gut, but also to an inflammation-induced switch from butyrate β-oxidation in colonocytes to anaerobic glycolysis[4] (due to decreased BPB levels and deficiency of butyrogenic substrates), resulting in lactate formation and increased epithelial oxygenation. Lactate, a product of glycolysis, can contribute to the M1 macrophage phenotype polarization by promoting NLRP3 inflammasome activation[66]. Butyric acid administration returns colonocyte and immune cells metabolism to β-oxidation, which may lead to a lower level of lactic acid in blood (35.6 vs 84.8 relative units in Muc2^-/- mice without treatment, P = 0.0360); also, butyric acid was shown to decrease Lactobacillus level in Muc2^-/- mice, which may also be a cause of the lowered level of lactate in blood. Consequently, Muc2^-/- mice had lower glucose and a-glycerophosphate (an intermediate in glycolysis) level in blood due to high glucose consumption which is accompanied by anaerobic glycolysis in inflamed colonocytes. BPV combination increased glucose levels in Muc2^-/- mice by 5-fold (95.8 vs 19.8 relative units in Muc2^-/- mice without treatment; P = 0.0007). SCFAs may serve as substrates for the production of new glucose (gluconeogenesis).

Muc2^-/- mice had a markedly higher 2-HB level in blood than C57BL/6 mice, which may be an indicator of the development of oxidative stress, one of the leading pathophysiological factors of IBD, and increased microbial production of 2-HB. All metabolites, individually and in combination, reduced 2-HB to an undetectable level (P < 0.05). Muc2^-/- mice had higher 3-HB levels in blood than C57BL/6 mice. 3-HB is one of the ketone bodies that acts as an alternative energy source during low glucose and is formed as a product of β-oxidation of fatty acids. Also, in Muc2^-/- mice, we found decreased levels of leucine, an exclusively ketogenic amino acid, and of α-ketoisocaproic acid, a leucine metabolite. We hypothesized that through increased leucine utilization under low-glucose conditions, Muc2^-/- mice produced higher levels of 3-HB. As a source of fatty acids, which can be metabolized in 3-HB, BPV combination increased levels of 3-HB by 3.3 times (46.0 vs 13.7 relative units in Muc2^-/- mice without treatment, P = 0.0345). In microorganisms, 3-HB mainly serves as a substrate for the synthesis of polyhydroxybutyrate, which is an important source of carbon and energy for bacteria, it increases the chances of survival of the microorganism in times of starvation and during the action of stress factors. Though mainly 3-HB has energetic functions, it also exhibits signaling and regulatory properties. 3-HB, as well as other SCFAs, can act via GPCRs, and as inhibitor of histone deacetylases, thus suppressing the pathogenesis of IBD[67].

Muc2^-/- mice had reduced levels of erythronic acid compared to healthy mice, and the BPV combination increased its level (12.4 vs 0.46 relative units in Muc2^-/- mice without treatment, P = 0.0092). Erythronic acid is an endogenous metabolite of carbohydrates, and also may be considered as a microbial biologically active metabolite, as its level increased in response to probiotic therapy[64]. Furthermore, erythronic acid was negatively correlated with the abundance of P. excrementihominis[68], which is known to be associated with chronic inflammation in the intestine[51]. We confirmed negative correlation between erythronic acid and P. excrementihominis, as propionic acid reduced P. excrementihominis level in Muc2^-/- mice fecal samples, while the BPV combination increased erythronic acid in Muc2^-/- mice blood. The BPV combination also significantly increased glycolic acid levels in Muc2^-/- mice blood (4.2 vs 2.4 relative units in Muc2^-/- mice without treatment, P = 0.028). Glycolic acid of bacterial origin possesses antibacterial activity against pathogens, thus showing anti-inflammatory effects in IBD[1].

In IBD, amino acid metabolism is considerably altered due impaired absorption in the intestine and increased demand by immune cells and gut microbiota[69]. Previously we found reduced serum levels of amino acids in IBD patients[11]. Increased amino acids turnover results in accelerated urea synthesis in the liver, and gut bacteria, particularly from Pseudomonadota phylum, which exhibit increased expression of urease-utilizing gene pathways, utilize host urea releasing ammonia, influencing dysbiosis and inflammation through increased reactive nitrogen species production[70]. Muc2^-/- mice had higher urea levels in blood than C57BL/6 mice, and butyric acid reduced its level (23.4 vs 294.6 relative units in Muc2^-/- mice without treatment, P = 0.0280). It is particularly important to note the decrease in blood levels of threonine and serine in Muc2^-/- mice, which are necessary for intestinal mucosal protein synthesis. Malnutrition in Muc2^-/- mice, due to impaired nutrient absorption, alter the mucus layer by decreasing intestinal mucin synthesis. Muc2^-/- mice can compensate for Muc2 loss by upregulation of other secretory mucins, such as MUC6[71]. Butyric and valeric acids increased threonine (1.9 and 1.8 vs 0.32 relative units in Muc2^-/- mice without treatment, P = 0.0055, P = 0.0135) and serine (1.3 and 0.90 vs 0.14 in Muc2^-/- mice without treatment, P = 0.0092, P = 0.0199) level in Muc2^-/- mice. Isoleucine was the only amino acid, which level was elevated in Muc2^-/- mice compared to C57BL/6 mice. It has been previously shown that IBD patients have a notably increased plasma level of isoleucine compared with healthy controls, and isoleucine gavage administration aggravates the progression of colitis in mice presumably through impairment of intestinal barrier function and exacerbation of intestinal inflammatory response by nuclear factor-κB signaling[72]. Propionic acid reduced levels of isoleucine (0.48 vs 3.0 relative units in Muc2^-/- mice without treatment, P = 0.0164).

2-HB acid, as a marker of IBD, possesses biological activity toward IBD progression

We have repeatedly observed an increase in blood 2-HB levels in patients with ulcerative colitis[1,11,64]. 2-HB is considered a biomarker for many diseases, including insulin resistance, type 2 diabetes, cancer, and mitochondrial diseases, which are characterized by increased lipid oxidation and oxidative stress[73]. High levels of 2-HB are produced in response to oxidative stress as a metabolic byproduct in glutathione synthesis from cystathionine to cysteine. It is known that 2-HB is derived from the conversion of α-ketobutyrate catalyzed by lactate dehydrogenase. α-ketobutyrate, in turn, is a product of threonine and methionine catabolism, as well as of the anabolic pathway of glutathione, and is catabolized by pyruvate dehydrogenase and branched-chain α-ketoacid dehydrogenase complexes.

Besides, gut microbiota could also influence 2-HB blood levels in Muc2^-/- mice, since 2-HB-producing bacteria are mainly distributed in Bacillota and Pseudomonadota[73]. We found a dramatically increased level of C. porci in Muc2^-/- mice (up to 6% of all bacteria in Muc2^-/- mice), and its level was reduced to the level of healthy mice after the metabolite administration. The genus Clostridium was shown to be able to synthesize 2-HB using all possible metabolic pathways (aspartate, methionine, threonine, and 2-aminobutyric acid pathway)[73]. According to the metabolomic analysis, levels of methionine, threonine, and aspartic acid were reduced in Muc2^-/- mice, which may indicate their use for the synthesis of 2-HB. Such a common increase of 2-HB in various diseases, in particular in ulcerative colitis, may indicate the possible involvement of this metabolite in the pathogenesis of IBD. To test this hypothesis, we investigated the effect of 2-HB on all the above-mentioned factors of the pathogenesis of IBD.

2-HB at a concentration of 0.5 mmol/L (approximately 0.05 mg/mL) decreased the viability of DSS-stimulated Caco-2 cells by 6% (P = 0.0259), increased paracellular permeability for FITC-dextran in the cell monolayer by 31% (P = 0.0159), and IL-8 secretion by 45% (P = 0.0260; Figure 8A and B). 2-HB decreased ZO-1 expression by 32% (P = 0.0303) and increased IL-8 expression by 3.3 times (P = 0.0381; Figure 8C and D). In the absence of DSS or LPS, 2-HB had no effect on viability, inflammation, or monolayer permeability (data not shown). These results indicate that 2-HB exhibits biological activity in specific contexts, it may exacerbate LPS-induced inflammation or DSS-induced injury without affecting intact cells.

Figure 8 Biological activity of 2-hydroxybutyric acid in the pathogenesis of inflammatory bowel disease on Caco-2 cell model. A: Viability of Caco-2 cells treated with 2-hydroxybutyric acid together with 3% dextran sulfate sodium; B: Apparent permeability coefficient (P_app) for monolayer permeability to fluorescein isothiocyanate-dextran; C: Gene expression of pro-inflammatory cytokine interleukin-8 and tight junction protein ZO-1; D: Interleukin-8 content in culture medium. Values are shown as medians and interquartile range (from min to max). All assays were performed in three independent experiments, each of which consisted of 5-11 biological replicates per group. ^aP < 0.05 vs the control, Mann-Whitney U test. DSS: Dextran sulfate sodium; LPS: Lipopolysaccharides; P_app: apparent permeability coefficient; 2-HB: 2-hydroxybutyric acid; IL-8: Interleukin-8.

2-HB at 1.5 mg/mL didn’t alter ALT and AST levels in Muc2^-/- mice (data not shown); this concentration was chosen for the subsequent experiment. 2-HB oral administration to Muc2^-/- mice resulted in a 77% increase in IL-1β expression (P = 0.0424) and in 3.2-fold increase in TNF-α expression (P = 0.0286) in the colon; intestinal level of IL-1β was increased by 53% (P = 0.0381; Figure 9A and B). 2-HB decreased number of Tregs (CD4⁺CD25⁺Foxp3⁺) in the mesenteric LNs by 21% (P = 0.0286) and increased level of markers of pro-inflammatory M1 macrophage - CD80 by 72% (P = 0.0317; Figure 9C and D).

Figure 9 Biological activity of 2-hydroxybutyric acid in the pathogenesis of inflammatory bowel disease in Muc2^-/- mice. A: Gene expression of pro-inflammatory cytokines in colon; B: Intestinal content of inflammatory interleukin-1β; C: n (%) of regulatory T cells in the mesenteric lymph nodes; D: Effect of 2-hydroxybutyric acid on CD80 expression in peritoneal macrophages. Values are shown as medians and interquartile range (from min to max). All assays were performed in one experiment, consisted of 4-11 biological replicates (mice) per group. ^aP < 0.05 vs the control, Mann-Whitney U test. Treg: Regulatory T cells; 2-HB: 2-hydroxybutyric acid; IL: Interleukin; TNF-α: Tumor necrosis factor-α.

DISCUSSION

A shift in gut microbial composition contributes to the pathogenesis of IBD through altered production of microbial metabolites. Metabolic dysbiosis in IBD is associated with impaired microbial synthesis of SCFAs[10], secondary BAs production[34], and increased microbial production of H₂S[52], ammonia and nitric oxide[70], lactate[74], 2-HB[1,11,64], phenolcarboxylic, succinic, and 2-hydroxyisovaleric acids[1]. Chronic inflammation in the intestine also leads to significant metabolic disturbances in the host organism, primarily in immune cells and colonocytes[2,4]. Changes in microbial and host metabolism are reflected in the serum metabolome. These metabolites, both endogenous and microbial in origin, may serve as biomarkers of IBD and have biological activity that either drives disease progression or acts as a therapeutic agent. Metabiotics, a new class of therapeutic agents based on microbial metabolites, can correct taxonomic and metabolic dysbiosis, reduce inflammation, and increase the effectiveness of IBD treatment. We use the term “metabiotic” rather than “postbiotic”, since according to the International Scientific Association of Probiotics and Prebiotics, postbiotics are defined as preparations of inanimate microorganisms and/or their components that confer a health benefit. While neither microbe-derived nor chemically synthesized metabolites are living things and cannot have an “afterlife”, compositions consisting only of metabolites are not postbiotics[75].

In this study, we compared the biological activity of BPV, and their combination, against the main factors of pathogenesis in in vitro and in vivo models of IBD. Differentiated Caco-2 cells are considered the gold standard for modeling the intestinal epithelial barrier and inflammation. The effects of LPS and DSS on Caco-2 cells mimic the increased inflammation, increased intestinal barrier permeability, and cell damage caused by pathogens, pro-inflammatory cytokines, or toxic agents[76]. Muc2^-/-mice spontaneously develop chronic, progressive colitis due to a defective mucus barrier that increases bacterial contact with the intestinal epithelium, triggering immune overactivation and inflammation that closely resembles that in human ulcerative colitis[77].

After metabolite treatment, Caco-2 cells exhibited reduced IL-8 secretion, decreased monolayer permeability, and increased viability. Butyric acid treatment effectively reduced Caco-2 cell monolayer permeability; propionic and valeric acids reduced inflammation, and together, BPV exhibited synergistic anti-inflammatory, intestinal barrier-strengthening, and viability-improving effects on Caco-2 cells. In Muc2^-/-mice, only butyric acid improved gut barrier integrity, propionic acid increased IL-10 content in colon and Foxp3 expression in colon, valeric acid decreased TNF-α expression in colon and promoted the polarization of peritoneal macrophages in an anti-inflammatory manner (iNOS expression decreased whereas Arginase-1 expression increased), while BPV combination had a greater effect on increasing Tregs in LNs and decreasing IL-1β content in colon.

There are certain difficulties in dissecting macrophage phenotypes in vivo, where a spectrum of macrophage phenotypes exists. For example, the canonical M2 macrophage marker Arginase-1 labels only 24% of M2 macrophages and is very highly up-regulated in M2 but is still up-regulated in M1 cells[78]. Since M1 and M2 macrophages may not express all the markers used to define them, or may show overlap in marker expression, it is understandable that we detected changes in only two markers for each metabolite.

Butyric acid, in both Caco-2 cells and Muc2^-/-mice, had no effect on reducing pro-inflammatory cytokine levels (in multiple comparisons), but in pairwise comparisons with control, butyric acid had a significant effect. This may indicate that butyric acid, although effective on its own, had a weaker effect than other metabolites, as most of it could be used by colonocytes as an energy substrate, thereby reducing systemic impact.

There was a remarkable difference in the fecal microbiota and blood metabolome composition between Muc2^-/- and C57BL/6 mice (healthy control). We found the reduction of obligate anaerobic bacteria, especially BPB, in Bacillota phylum (Butyricicoccus, Clostridium, Eubacterium, Flintibacter, Blautia) and Bacteroidota phylum (Muribaculum, Duncaniella, Odoribacteraceae); and dysbiotic expansion of pathogenic Thermodesulfobacteriota (Bilophila, Desulfovibrio) and Pseudomonadota (Turicimonas) in Muc2^-/- mice. Also, in Muc2^-/- mice we discovered elevated Lactobacillaceae bacteria (including the genus Lactobacillus, Ligilactobacillus, Limosilactobacillus), and pathobiontic P. caecicola, which all can grow under aerophilic conditions. We also found that Muc2^-/- mice had elevated levels of mucin-degrading bacteria Akkermansia and B. acidifaciens, which could damage the intestinal mucosal barrier by overconsuming the mucus layer and contribute to inflammation and tumor growth[46]. Increased abundance of Akkermansia and B. acidifaciens during intestinal inflammation leads to higher release of less complex sugars (lactose, melibiose, raffinose, and galactinol) from mucins. Accumulation of these metabolites may lead to depletion of commensal bacteria mainly from the Bacteroidota and Bacillota phyla and expansion of pathogens such as Salmonella typhimurium and C. difficile in the inflamed gut[79]. Under IBD-like conditions, the bacterial community switches its metabolism to adapt and grow in the inflammatory environment. Some pathogenic bacteria can proliferate more readily during intestinal inflammation, adapting to the inflammatory microenvironment. For example, pathobionts can utilize unique nutrients such as ethanolamine and 1,2-propanediol, which are unavailable to commensal symbionts, thereby overcoming competition with commensals[80]. Inflammation in IBD shifts immune cells and colonocytes metabolism to glycolysis, which reduces their oxygen consumption[2,4], and increased intestinal oxygen level could be beneficial for pathogenic facultative anaerobic bacteria and harmful for obligate anaerobic bacteria, including Bacteroidota and Bacillota[5]. IBD is often associated with an increase in the abundance of facultative anaerobic Pseudomonadota[81] and microaerophilic Thermodesulfobacteriota[52], which indicates a disruption in anaerobiosis; reduced bacterial biodiversity[23], decreased populations of SCFAs-producing bacteria and SCFAs production[10], and lower levels of secondary BAs[34].

In the serum of Muc2^-/- mice, we found increased levels of lactic, 2-HB, 3-HB, LCFA, cholesterol, isoleucine, and urea, and reduced levels of α-ketoisocaproic, erythronic, pentadecanoic, gondoic acids, most of amino acids, glucose and a-glycerophosphate. These changes reflect metabolic dysbiosis of disturbed gut microbiota, and altered host metabolism during chronic inflammation. Increased lactate concentration in Muc2^-/- mice blood is a prominent example of IBD-associated gut metabolic dysbiosis characterized by increased abundance of lactic acid-producing bacteria and lack of lactate-utilizing bacteria. Beneficial lactate-utilizing bacteria from Bacillota phylum can utilize lactate and convert it to propionate or butyrate, while pathogenic Desulfovibrio spp. utilize lactate with H₂S production[74]. Although inflamed cells of the host organism also release an excess of lactate into the blood due to metabolic shift from β-oxidation to anaerobic glycolysis, most of the lactate in the blood is formed by intestinal microbiota, which was shown in germ-free mice[74].

After oral administration of metabolites, Muc2^-/- mice had a significant shift in microbiota and metabolite composition toward the healthy C57BL/6 mouse state. After SCFAs treatment, the gut microbiota of Muc2^-/- mice was characterized by increased bacterial biodiversity and increased abundance of anti-inflammatory Bacillota and decreased abundance of pathogenic Pseudomonadota and Thermodesulfobacteriota. Regulatory properties of SCFAs aimed at altering the composition of the intestinal microbiota to compensate for disease symptoms. These changes in microbiota composition after SCFAs treatment were reflected in blood metabolome composition.

Muc2^-/- mice after butyric acid administration had lowered levels of Lactobacillus and therefore, lowered levels of lactate in blood. Propionic acid significantly reduced serum cholesterol levels in Muc2^-/- mice, presumably by stimulating the growth of cholesterol-reducing bacteria E. coprostanoligenes and Clostridium spp. Propionic acid and BPV combination administration resulted in increased levels of anti-inflammatory pentadecanoic acid in Muc2^-/- mice blood, as gut bacteria can synthesize pentadecanoic acid from propionic acid. After BPV combination Muc2^-/- mice had increased levels of erythronic, glycolic acids, and 3-HB, which were shown to possess anti-inflammatory effects in IBD. All metabolites and the combination decreased 2-HB levels in the blood of Muc2^-/- mice. We hypothesized that this effect may be partly explained by metabolites inhibiting C. porci growth. However, this hypothesis requires experimental confirmation. In addition, the metabolites improved host cell metabolism, as evidenced by decreased blood levels of pro-inflammatory LCFA and lactic acid in Muc2^-/- mice, and increased levels of amino acids and glucose.

The results of our study are consistent with some earlier studies. Similarly, SCFA-producing bacterial strains reduced experimental ulcerative colitis severity through M2 macrophage polarization via JAK/STAT3/FOXO3 axis inactivation[41]. A 2024 systematic review of 29 clinical and translational studies confirmed a significant and consistent reduction in fecal butyrate, propionate, and acetate in IBD patients vs healthy controls, with dietary interventions producing the most consistent SCFA restoration[82]. Subsequently, another systematic review and meta-analysis further revealed a significant reduction in fecal SCFA levels in IBD compared to healthy controls (active IBD showed a greater decrease in butyrate, and ulcerative colitis showed a notable reduction in propionate). Dietary interventions in IBD patients led to increased SCFA levels, with butyrate showing the most improvement, suggesting the potential therapeutic value of SCFA[83]. Notably, valerate, the third component of our BPV combination, has received very little experimental attention. A single clinical study confirmed significantly lower valeric acid levels in IBD patients vs healthy controls[82]; however, its specific anti-inflammatory mechanisms and in vivo therapeutic potential have not been characterized in IBD animal models prior to the present work.

We also hypothesized that 2-HB may have biological activity toward IBD progression, it increased intestinal permeability, inflammation, and number of pro-inflammatory M1 peritoneal macrophages; decreased cell viability and number of Treg cells. Our data are consistent with previously obtained detailed evidence demonstrating that 2-HB promotes intestinal and systemic inflammation in both colorectal cancer and type 2 diabetes mice models by potentiating nuclear factor-κB signaling through a lactate dehydrogenase, A-dependent mechanism[84]. Further experiments are required to confirm these effects and elucidate the mechanisms of action of 2-HB in the context of IBD.

Despite the fact that Caco-2 cells exposed to LPS and Muc2^-/- mice are assumed as valuable models of human ulcerative colitis, effectively mimicking IBD pathogenesis due to spontaneous bacterial contact with the epithelium[15,76,77], leading to severe inflammation and altered metabolism, there are several limitations in translating results obtained on cell culture and Muc2^-/- mice to humans. Differences in immune responses[85], taxonomic composition of gut microbiota[86] and metabolic activity[87] between humans and mice, as well as small sample sizes in mouse studies (nevertheless according to “resource equation” method the value E, which is the degree of freedom of analysis of variance, should lie between 10 and 20, and our sample size n = 4 mice per group can be considered as adequate[88]) reduce the accuracy of translational efficacy. The human origin of Caco-2 cells, their close similarity to human intestinal epithelial cells in vivo, and their high reproducibility make them more suitable than animal models for screening the biological activity of therapeutic compounds. But being immortalized lines of human colorectal adenocarcinoma cells and not producing mucins, Caco-2 cells share similar disadvantages with Muc2^-/- mice, which are completely deficient in mucin 2 and predisposed to the early development of invasive adenocarcinoma in the colon and small intestine. Another limitation of the Caco-2 IBD model is the lack of microbiota and immune cell interactions, which are essential for accurately mimicking IBD pathophysiology. Therefore, further investigation of the metabiotic composition and biological activity across different experimental models of IBD will be necessary to confirm the results obtained. Also, in this study, we did not examine the effect of the BPV metabolite combination on the composition of the fecal microbiota in mice, but only examined the effect of composition on the serum metabolome. This is due to substantial differences between humans and mice in gut microbiota composition (up to 85% of microbial species found in mice are not found in humans)[89], which will complicate the translation of the results obtained to humans. Metabolomic approach, on the contrary, allows for the acquisition of a “fingerprint” of the microbiota’s functional status, and, given that metabolic markers are not species-restricted, metabolic phenotyping is more applicable to translational research in the context of human health and disease[90,91].

CONCLUSION

In summary, this study shows that a combination of SCFAs has a synergistic effect in normalizing metabolic dysbiosis, reducing inflammation, and strengthening the intestinal barrier, thus making the use of a combination of metabolites, rather than individual compounds, a promising approach for IBD treatment. Taken together, the present study advances beyond prior SCFA research by characterizing the specific and complementary contributions of three SCFAs in combination, identifying a novel pro-inflammatory role for 2-HB in IBD, and integrating these findings within a multi-omics framework that links host immune phenotype, microbiome composition, and metabolome in the same experimental animal cohort.

ACKNOWLEDGEMENTS

We would like to thank Utsal VA (Golikov Research Clinical Center of Toxicology of the Federal Medical and Biological Agency of Russia, St. Petersburg, Russia) for performing the gas chromatography-mass spectrometry measurements. The authors extend sincere appreciation to the staff of the resource center “Genomic Technologies, Proteomics and Cell Biology” at the Federal State Budgetary Scientific Institution “All-Russian Research Institute of Agricultural Microbiology” for performing the metagenomic analysis.