Safety and Next-Generation Probiotic Potential of Mitsuokella multacida AGMB01124 Newly Isolated from Healthy Swine Feces

Abstract

Mitsuokella multacida AGMB01124 isolated from the feces of healthy swine was evaluated for its safety and potential as a next-generation probiotic (NGP). These bacteria have been increasingly reported to be positively correlated with short-chain fatty acids (SCFA), which are a beneficial energy source in the gut environment. We investigated the potential application of M. multacida AGMB01124 as an NGP by examining its genomic features, probiotic properties, cytotoxicity, oral toxicity, and its effect on SCFA levels in feces. Genome analysis did not identify any virulence-associated factors, and the strain harbored two antibiotic resistance genes (tetW/N/W and ermB). Furthermore, endoglucanase D (CelD) exhibited differences in four surface-exposed amino acid residues, three of which were located near the active site and one within the signal peptide region, suggesting a role in modulating catalytic function. Phenotypically, the strain exhibited susceptibility to most antibiotics recommended by Clinical & Laboratory Standards Institute and was non-hemolytic, non-cytotoxic. The strain demonstrated some tolerance to acidic and bile conditions, strong self-aggregation, co-aggregation with pathogens, and low cell surface hydrophilicity. Oral toxicity studies indicated no adverse effects on clinical signs, body weight, diet intake, relative organ weights, hematology, serum biochemistry, histopathology, or harmful enzyme activities. Short-chain fatty acid concentrations were significantly higher in the M. multacida AGMB01124-administered group than in the control group, suggesting potential contributions to host gut health and metabolic regulation. Therefore, this study provides preliminary evidence supporting the safety of the M. multacida AGMB01124 strain and its potential application as a candidate NGP.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12602-026-10925-y.

Introduction

Probiotics were defined by the Food and Agriculture Organization of the United Nations/World Health Organization in 2001 as “live microorganisms that provide health benefits to their hosts when administered in appropriate amounts” [1]. They are broadly classified into traditional probiotics and next-generation probiotics (NGPs), based on their use in promoting general gut health or alleviating specific disease conditions, respectively [2]. Although the precise mechanisms through which probiotics exert their effects in the gastrointestinal (GI) tract remain to be fully elucidated, they are recognized as key contributors to the metabolism of the gut environment [3]. Their beneficial roles include facilitating substrate digestion, modulating the immune system, and enhancing energy recovery within the GI tract [4]. Furthermore, supplementation with beneficial microbes contributes to maintaining host health by supporting immune function, shaping the gut microbial community, preventing colonization by pathogenic organisms, fermenting non-digestible dietary fiber, and producing vital nutrients such as vitamins and short-chain fatty acids (SCFAs) [5]. Disruptions in these microbial communities, characterized by reduced diversity, loss of beneficial bacteria, and overgrowth of pathogens, are associated with various inflammatory and infectious diseases [6]. In particular, microbial imbalance (dysbiosis) is closely linked to the development and progression of GI disorders [7]. GI disorders such as irritable bowel syndrome, inflammatory bowel disease, ulcerative colitis, Crohn’s disease, and diarrhea are often marked by impaired gut microbial balance and compromised digestive function [4, 8]. Several commensal bacteria, including Akkermansia muciniphila, Eubacterium hallii, Butyricicoccus pullicaecorum, and Faecalibacterium prausnitzii, have been identified as NGPs with the potential to restore intestinal homeostasis [9]. In this context, NGPs have emerged as attractive therapeutic candidates for alleviating GI disorders through targeted modulation of gut microbiota and enhancement of digestive function [10].

Traditional probiotics such as Lactobacillaceae and Bifidobacterium species are generally recognized as safe (GRAS) owing to their long history of use in fermented foods and dairy products. In contrast, NGPs, which are commensal bacteria recently identified in the GI tract, require safety assessments for application in biopharmaceutical and nutraceutical products due to the lack of historical documentation and regulatory precedent [11, 12]. To address these challenges, whole-genome sequence (WGS) analysis of NGPs has been rapidly expanded, offering valuable insights for predicting comprehensive characteristics such as safety and functional genomic features [13]. European food safety authorities are also mandating WGS analysis of microorganisms to monitor the presence of potential risks, such as virulence factors, antibiotic resistance genes, and harmful metabolic pathways [14]. However, WGS-based safety assessments alone are insufficient and should be complemented with phenotypic evaluations, including in vitro and in vivo experiments [15]. The theoretical safety concerns associated with NGPs therapy include systemic infections, metabolic and toxic disorders, transmission of virulence and antibiotic resistance genes, and immune hyperactivation in susceptible individuals [16, 17]. Based on safety data from preclinical animal studies, several NGPs, including Roseburia intestinalis [18], E. hallii [19], and A. muciniphila [20], are currently under investigation for their therapeutic potential.

M. multacida is a Gram-negative, obligate anaerobe, non-spore forming, non-motile, rod-shaped, and fermentative bacterium belonging to the phylum Bacillota, Class Negativicuts, order Selenomonadales, Family Selenomonadaceae. It is widely distributed in the rumen and GI tract of homothermic animals [21, 22]. Additionally, Bai et al. [23] have demonstrated that M. multacida in the GI tract of Tibetan swine can contribute to the regulation of the host energy metabolism by efficiently utilizing carbon sources such as cellulose, galactose, and fructose to produce SCFAs, including propionate and butyrate. Furthermore, Luo et al. [24] have reported that the abundance of M. multacida significantly increased in growing pigs fed a diet low in amylose/amylopectin ratio, accompanied by an increase in acetate and butyrate concentrations. M. multacida has been found to be of lower abundance in the gut microbiota of keratitis patients than healthy individuals [25], potentially reflecting a depletion of SCFA-producing bacteria. Although some studies have suggested a positive correlation between M. multacida and SCFA concentrations, evidence supporting its safety and probiotic properties as an NGP remains limited.

In this study, we investigated novel NGP strains from commensal bacteria and conducted a comprehensive genetic and phenotypic safety evaluation to further explore previously uncharacterized commensal bacteria for expanding the field of NGP research. In particular, the present study aimed to evaluate the safety and probiotic potential of M. multacida AGMB01124 isolated from healthy swine stool through in silico, in vitro, and in vivo studies. A total of 100 commensal bacteria were isolated and identified from the feces of healthy Korean cattle and swine (Supplementary Table 1). These isolates were screened for digestive enzyme activity to evaluate their functional potential as NGP candidates (Supplementary Table 2). Five strains (AGBM 00339, 00927, 01124, 01585, and 02846) exhibiting notable digestive enzyme activity were selected, and their galactosidase activity was subsequently evaluated (Supplementary Table 3). Among them, the M. multacida AGMB01124 strain isolated from swine feces was selected as a promising NGP candidate based on its enzyme activity profile (Supplementary Fig. 1). We then performed genome analysis, probiotic property assessment, and animal toxicity testing to evaluate its safety and potential probiotic properties.

Materials and Methods

Isolation and Identification of Commensal Bacteria from Korean Cattle and Swine Stool Samples

The commensal microbes were isolated from healthy Korean cattle and swine stools, which were obtained from the National Institute of Animal Science (Cheonan, Republic of Korea). Experimental procedures were conducted in an anaerobic chamber (Coy Laboratory Products, Grass Lake, MI, USA), containing an atmosphere of 90% nitrogen, 5% carbon dioxide, and 5% hydrogen. The stool samples were serially diluted to 10^− 6 in phosphate-buffered saline (PBS) and incubated for 2–3 days at 37 °C by spreading 100 µL on tryptic soy agar containing 5% sheep blood (TSAB). The single colony was then streaked onto a new TSAB plate and incubated under the same conditions. The isolated bacteria from stool samples were identified by 16 S rRNA gene sequencing using primers 785 F (5′-GGATTAGATACCCTGGTA-3′) and 907R (5′-CCGTCAATTCMTTTRAGTTT-3′). The identified strain was preserved in 10% (w/v) skim milk at − 80 °C until further experiments.

Whole-Genome Sequencing of M. Multacida AGMB01124

To obtain genomic DNA (gDNA) of strain AGMB01124 to be sequenced, gDNA was extracted from 1 mL of exponentially growing AGMB01124 culture (OD₆₀₀ = 0.8–1.0) using the G-spin™ Genomic DNA Extraction Kit (Intron Biotechnology, South Korea), following the manufacturer’s instructions. DNA quality was verified according to the standard quality control procedures recommended by Oxford Nanopore Technologies (ONT) before sequencing. For whole-genome sequencing, library preparation was conducted using the SQK-LSK109 kit (ONT, UK), and the resulting library was loaded onto a primed FLO-MIN106 R9.4 Spot-On Flow Cell. Sequencing was conducted on the MinION Mk1B platform, and the generated FAST5 files were base-called using MinKNOW software (v22.12.7) with settings matched to the library preparation protocol. High-quality reads were assembled using Trycycler (v0.5.3) [26], and genome annotation was performed with Prokka (v1.14.6) [27]. Functional assignments of predicted open reading frames were conducted using InterProScan (v5.64–96.0.0) [28], while functional categorization and pathway annotation were conducted using the COGNIZER platform and the KEGG Automatic Annotation Server (KAAS), respectively [29]. Final genome visualization was achieved through Artemis (v16.0.17) [30] and GenVision Pro (v17.4).

Genome-Based Safety Profiling of M. Multacida AGMB01124

Comparative genomic analysis was performed to assess the genomic relatedness between Mitsuokella strains by calculating average nucleotide identity (ANI) values using FastANI (v1.32) [31]. The resulting distance matrix was used to construct a phylogenetic tree, which was visualized in RStudio (v4.2.1) with base R packages [32]. To explore the pan-genomic structure, strain AGMB01124 was compared with publicly available M. multacida and M. jalaludinii genomes retrieved from the NCBI database, using Roary (v3.11.2) for pan-genome clustering [33]. The outputs were visualized using Phandango (v.1.3.0) [34] and RStudio for comparative representation. Additionally, virulence-associated genes were screened using a virulence factor database [35], while antimicrobial resistance determinants and potential horizontal gene transfer (HGT) events were identified through the Resistance Gene Identifier tool (v6.0.2) with reference to the Comprehensive Antibiotic Resistance Database (CARD; v3.2.7) [36] and further examined using HGTphyloDetect [37].

In Vitro Probiotic Characteristics of M. Multacida AGMB01124

Profile of Carbohydrate Utilization

The carbohydrate utilization profile of M. multacida AGMB01124 was assessed using the API 50 CH and API 50 CHL medium according to the manufacturer’s instructions (BioMérieux, Lyon, France). M. multacida AGMB01124 was resuspended in 10 mL of PBS and adjusted to a 2.0 McFarland standard. The suspension was then inoculated into test strips of API 50 CH and incubated at 37 °C for 48 h. Carbohydrate utilization patterns were confirmed in duplicate.

Auto-aggregation

To assess the adherence ability of epithelial cells of M. multacida AGMB01124, the auto-aggregation was performed according to the method described by Zhang et al. [38], with slight modifications. Briefly, bacterial cells were collected through centrifugation at 3,264 × g for 10 min at 4 °C, washed twice with PBS, and resuspended to adjust to 10⁸ CFU/mL in PBS. The suspension of 4 mL was vortexed for 5 s and incubated for 3, 6, 12, and 24 h at 37 °C in anaerobic conditions. The assay was performed using three independent replicates (n = 3). Then, the supernatant was measured at 600 nm using a spectrophotometer, and auto-aggregation was calculated as follows:

Whereas, A₀ = Initial absorbance at 0 h; A₁ = Final absorbance after 3, 6, 12, and 24 h.

Co-aggregation

The co-aggregation was determined using the method by Hojjati et al. [39], with minor modifications. M. multacida AGMB01124 and pathogens, including Escherichia coli O157:H7 (ATCC43889), Salmonella Typhimurium UK1 (ATCC 68169), Staphylococcus aureus (KCCM 11335), and Listeria monocytogenes (KCCM 40307), were incubated for 24 h in tryptic soy broth. Bacterial cells were harvested using centrifugation at 3,264 × g for 10 min at 4 °C, washed twice with PBS, and adjusted to 10⁸ CFU/mL in PBS. M. multacida AGMB01124 and each pathogen were mixed in equal volumes (2 mL), vortexed for 1 min, and incubated at 37 °C for 2 h under anaerobic conditions. The experiment was conducted in triplicate. The absorbance was determined at 600 nm using a spectrophotometer. Co-aggregation was calculated as follows:

Whereas, A_mix = Absorbance of the mixture; A₁ = Absorbance of the M. multacida AGMB01124; A₂ = Absorbance of the pathogens.

Hydrophobicity

To evaluate adhesion capacities to various solvents of M. multacida AGMB01124, cell surface hydrophobicity was conducted by following the method described by Yasmin et al. [40], with slight modifications. The bacterial cells were pelleted after centrifugation at 3,264 × g for 10 min at 4 °C, washed twice in PBS, and resuspended in the same buffer adjusted to 10⁸ CFU/mL. Then, 3 mL of bacterial suspension was mixed with 1 mL of different solvents such as chloroform, ethyl acetate, hexane, hexadecane, toluene, and xylene, and vortexed for 2 min. The mixtures were incubated for 20 min to separate the two phases (organic and aqueous phases), and the aqueous phase was gently collected. This analysis was performed using three replicate experiments (n = 3). The absorbance was measured at 600 nm to calculate the cell surface hydrophobicity using the following equation:

Whereas, A₀ = Initial absorbance at 0 h; A₁ = Final absorbance.

Acid and Bile Salt Tolerance

The tolerance of M. multacida AGMB01124 to acid and bile salts was determined according to the method described by Dowarah et al. [41], with minor modifications. After incubation for 48 h, the bacterial cells were harvested through centrifugation (3,264 × g, 10 min, 4 °C) and washed twice with PBS. The harvested bacterial cells were inoculated into bile salts (0.1, 0.2, and 0.3%) or pH (2, 3, and 4) adjusted tryptic soy broth to a concentration of 10⁸ cells/mL and incubated at 37 °C for 2 or 6 h, respectively. The broth cultures were serially diluted 10-fold in PBS, and 0.1 mL of each dilution was spread onto a TSAB plate, which was incubated at 37 °C for 48 h. Microbial counts were determined using plates containing fewer than 300 colonies, and the results were expressed as log CFU/mL. All procedures were performed using three independent samples.

In Vitro Safety Assessment of M. Multacida AGMB01124

Hemolytic Activity

Hemolysis test was conducted according to the method described by Kim et al. [42]. The hemolytic activity of M. multacida AGMB01124 was performed using TSAB plates, which were streaked onto the plates and then incubated for 48 h at 37 °C. After incubation, the hemolysis was assessed for white or transparent zones surrounding the bacteria (β-hemolysis), an irregular or partial green zone (α-hemolysis), and an absent clearing zone (γ-hemolysis), of which only γ-hemolysis is considered safe. S. aureus KCCM 11,335 was used as a positive control to compare the hemolytic activity.

Antibiotic Resistance

For antimicrobial susceptibility testing, an overnight culture of AGMB01124 was spread onto Brucella agar with hemin (5 µg/mL), vitamin K₁ (1 µg/mL), and sheep blood (5% v/v). ETEST^® strips (bioMérieux, France) containing various antibiotics were then applied to the surface, followed by incubation at 37 °C for 12 h. For cefotetan and ceftriaxone, the agar dilution method was employed: 10⁵ CFU of the test strain was spotted onto the supplemented Brucella agar containing 8 mg/L of each antibiotic and incubated under the same conditions as the ETEST^® assay. After incubation, inhibition zones or colony morphology were examined, and minimum inhibitory concentrations (MICs) were determined. The assay was performed in duplicate, and the higher MIC value obtained from the two replicates was reported as the final MIC. Interpretive criteria were applied based on the cut-off values proposed by the Clinical & Laboratory Standards Institute (CLSI) [43].

Cytotoxicity

For cytotoxicity evaluation, the human colorectal epithelial cell line Caco-2 was cultured in Dulbecco’s modified Eagle’s medium (Welgene, South Korea) supplemented with 10% fetal bovine serum (GW Vitek, South Korea) and 1× penicillin-streptomycin (Welgene, South Korea) under standard cell culture conditions (37 °C, 5% CO₂, humidified atmosphere). Cells were seeded in 12-well plates at a density of 4 × 10⁵ cells/well and allowed to differentiate for 10 days, with media replacement every two days. After differentiation, AGMB01124 were applied to the wells at concentrations of 10⁷, 10⁸, and 10⁹ CFU/well in fresh Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum and incubated for 24 h. Cytotoxicity was assessed using the lactate dehydrogenase assay (EZ-LDH, DoGenBio, South Korea) on culture supernatants, according to the manufacturer’s protocol. Basal media-treated cells were used as negative controls, and E. coli O157:H7 (ATCC 48395) served as the positive control for cytotoxic effects. This analysis was tested in triplicate (n = 3).

In Vivo Oral Toxicity Study of M. Multacida AGMB01124

Animal and Experimental Design

Oral toxicity experiments were performed using 6-week-old male and female C57BL/6 mice weighing 20 ± 2 g and maintained under controlled environmental conditions (22 ± 2 °C, 60 ± 10% humidity, 12-h light/dark cycle). All mice were provided with standard rodent chow (Orient Bio, Rodents Diet 5053) and drinking water ad libitum. The mice were divided by sex, and male and female mice were randomly assigned into four groups each (n = 5 per group). The animals were acclimatized for one week prior to the experiment. The groups were administered daily for four weeks with PBS (control) or M. multacida AGMB01124 at 10⁷ (Low-dose group), 10⁸ (Middle-dose group), and 10⁹ (High-dose group) CFU. On day 29, the animals were sacrificed via carbon dioxide asphyxiation after 4 h fast, and blood samples were collected in K₂EDTA-treated and heparinized tubes. Organ samples such as kidneys, liver, and spleen were weighed, and colon samples were measured for length and stored in conical tubes containing 10% formalin.

General Observations

The general observations for oral administration of M. multacida AGMB01124 during the experimental period included skin, fur, respiration, movement, behavior, tremors, convulsions, salivation, diarrhea, sleep, and mortality. Furthermore, body weight, feed consumption, and water intake of all mice were recorded.

Relative Organ Index

The relative organ index was calculated by weighing the kidney, liver, and spleen of the mice as follows.

Each organ index was calculated from five independent animals (n = 5 per group).

Hematological and Serum Biochemical Analysis

Hematological analysis was performed on blood samples collected in K2EDTA tubes using an automated hematology analyzer (XN-V; Sysmex Corporation, Kobe, Japan). The analyzed parameters for whole blood were as follows: red blood cell (RBC), hemoglobin (HGB), hematocrit (HCT), mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), mean corpuscular hemoglobin concentration (MCHC), platelet count (PLT), and white blood cell (WBC). Blood samples were obtained from five individual animals per group (n = 5).

Serum biochemical analysis was prepared by clotting blood in heparin-treated tubes at room temperature and then collecting serum with centrifugation (2000 × g, 10 min, 4 °C). Serum biochemical parameters were conducted using an automated clinical chemistry analyzer (CHEM NX500; Fujifilm, Tokyo, Japan) with the following indices: albumin (ALB), alkaline phosphatase (ALP), alanine aminotransferase (ALT), aspartate aminotransferase (AST), calcium (Ca), creatinine (CRE), γ-glutamyltransferase (GGT), glucose (GLU), high-density lipoprotein cholesterol (HDL-C), inorganic phosphorus (IP), total bilirubin (TBIL), total cholesterol (TCHO), triglyceride (TG), total protein (TP), and uric acid (UA). Serum was obtained from five animals in each group (n = 5).

Histopathological Analysis

The liver, kidney, spleen, and colon were collected from each mouse (n = 5 per group) and stored in 10% neutral formalin solution at room temperature for 24 h. Subsequently, tissue samples were processed through standard histological procedures: washing, graded ethanol dehydration, xylene clearing, and paraffin embedding. Using a rotary microtome, 3–5 μm-thick paraffin sections were obtained. These sections were deparaffinized in xylene, rehydrated through graded ethanol, and stained with hematoxylin and eosin (H&E) according to standard protocols. The stained slides were scanned for digital imaging using a slide scanner (ZEN 3.4 blue edition, Carl Zeiss Co., Ltd., Germany). Toxicity severity and type were classified into five grades (Grade 1–5) according to the NCI General Toxicity Criteria [44].

Harmful Enzyme Activity of the Cecum

The activity levels of harmful enzymes, including β-glucosidase and β-glucuronidase were measured in cecal samples from mice orally administered the M. multacida AGMB01124 strain for four weeks, using a slightly modified method described by Śliżewska et al. [45]. Approximately 0.2 g of the cecal sample was homogenized with liquid nitrogen, followed by the addition of 1 mL of PBS (0.01 M, pH 7.4) and mixed using a vortex mixer. The sample was then centrifuged at 13,000 × g for 20 min at 4 °C. Subsequently, 50 µL of supernatant was mixed with 100 µL of PBS and 100 µL of a 20 mM substrate solution (either 4-nitrophenyl β-D-glucopyranoside or 4-nitrophenyl β-D-glucuronide) and incubated at 37 °C for 1 h. After incubation, 250 µL of 0.5 N NaOH was added to the samples to stop the reaction, and absorbance was measured at 410 nm using a spectrophotometer. The protein concentration in the cecal samples was determined using a bicinchoninic acid assay protein kit. Enzyme activity was expressed as the amount of p-nitrophenol (µM) released during 1 h of reaction per mg of protein (µMh/mg), with five mice analyzed per group (n = 5).

SCFA Concentrations in Stool Samples Using Gas Chromatography–Mass Spectrometry (GC/MS)

The SCFA analysis of stool samples was performed after four weeks of oral administration of M. multacida AGMB01124, using a slightly modified method based on Zhang et al. [46]. Stool samples were diluted 10-fold with high-performance liquid chromatography-grade water and homogenized for 20 s at 6,500 rpm using a Bead Ruptor 24 (Omni International Inc., Kennesaw, GA, USA). The homogenized samples were centrifuged at 13,000 × g for 30 min at 4 °C, and 100 µL of the supernatant was transferred to a new tube containing 10 µL of 5 M hydrochloric acid. Subsequently, 100 µL of anhydrous diethyl ether was added and mixed with a vortex for 5 s. The mixture was placed on ice for 5 min and centrifuged at 10,000 × g for 5 min to separate the organic layer. The organic layer containing SCFAs was transferred to a tube containing sodium sulfate (Na₂SO₄) to completely remove the remaining aqueous solution. After two additional extractions using anhydrous diethyl ether, 100 µL of the extract was transferred to a GC vial. Derivatization of SCFAs was performed by mixing 5 µL of N, O-bis(trimethylsilyl)trifluoroacetamide, followed by incubation at 70 °C for 40 min and further incubation at 37 °C for 2 h before GC-MS analysis. Quantitative analysis of SCFA was conducted using a GC-MS analyzer (QP 2010 Ultra; Shimadzu Corporation, Kyoto, Japan) equipped with a DB-5MS UI capillary column (30 m × 0.25 mm × 0.25 μm film thickness; J&W Scientific, CA, USA). Helium (99.999%) was used as the carrier gas at a constant flow rate of 1 mL/min, with a 1 µL sample injected in split mode (10:1). The injector, ion source, and interface temperatures were maintained at 260 °C, 230 °C, and 280 °C, respectively. To analyze SCFAs, the GC oven temperature was programmed to start at 40 °C for 2 min, then increased to 150 °C at a rate of 15 °C/min, held for 1 min, and finally raised to 300 °C at 30 °C/min and held for 5 min. Ionization was conducted at 70 eV using electron impact. The mass spectrometer operated in full scan mode (40–400 m/z), with data acquired at a rate of 12.8 scans per second. SCFAs were identified using SIM mode, with target ions detected at m/z 117 (acetic acid), 131 (propionic acid), 145 (isobutyric and butyric acids), and 159 (isovaleric and valeric acids). The qualifier ion for all SCFAs was m/z 75, and m/z 117 was used additionally for those with four or five carbon atoms. The SCFAs were identified and quantified by comparing their retention times and mass spectral data with those of external standard solutions. Stool samples from five mice per group were analyzed (n = 5).

Statistical Analysis

The experimental results were presented as mean ± standard deviation. Statistical comparisons between groups were analyzed by two-tailed t-test and one-way analysis of variance with Tukey test using Minitab software (Minitab LLC, Pennsylvania, USA). SCFA data were further analyzed using MetaboAnalyst 6.0 (https://www.metaboanalyst.ca/MetaboAnalyst/ModuleView.xhtml) to perform principal component analysis (PCA).

Results

Genome-Based Identification of M. Multacida AGMB01124

The general genomic characteristics of the M. multacida AGMB101124 strain are illustrated in Fig. 1. The genome comprises a circular double-stranded DNA molecule measuring 2,461,174 base pairs, with a GC content of 58.2%. A total of 2,253 coding sequences (CDSs) were annotated, along with 75 tRNA genes and 21 rRNA genes. For taxonomic and comparative genomic evaluation, both ANI-based phylogenetic analysis and pan-genome profiling were performed. The ANI-based phylogenetic tree constructed using 75 representative Mitsuokella strains (including M. multacida and M. jalaludinii) revealed four distinct clusters corresponding to species-level groupings (Cluster M-I and M-II: M. multacida, Cluster J-I, J-II: M. jalaludinii) (Fig. 2A). AGMB01124 was included in the Cluster M-I and showed the highest genomic similarity (97.77% ANI) to M. multacida SRR14812371 (Accession No. GCA_946411775.1), affirming its classification as M. multacida. In the pan-genome analysis involving AGMB01124 and other 19 M. multacida genomes of the Cluster M-I from the NCBI database, a total of 6,616 genes were identified (Fig. 2B). Among them, 1,030 genes constituted the core genome shared by all strains, 5,586 were accessory genes present in multiple but not all strains, and 2,673 were strain-specific genes, of which 187 were uniquely found in M. multacida AGMB01124.

Fig. 1.

Genome map of Mitsuokella multacida AGMB01124. The outermost ring displays all annotated open reading frames (ORFs) on both strands, with colors representing their respective clusters of orthologous groups (COG). The inner ring, shown with red peaks, illustrates the GC content variation across the genome. Between these two rings, sky blue arrows indicate the positions of rRNA operons, while orange arrows mark the locations of tRNA genes. A legend describing the color scheme for COG functional categories is provided

Fig. 2.

(A) Average nucleotide identity (ANI) analysis and (B) pan-genome analysis of AGMB01124 and Mitsuokella sp. (A) The phylogenetic tree was constructed based on ANI-derived distance values (1 − ANI), with a scale bar of 4.0 indicating approximately 4% genome dissimilarity between taxa. M. multacida AGMB01124 is highlighted in red. The tree is divided into four distinct clusters, designated as Cluster M-I, M-II, including M. multacida spp., and Cluster J-I, J-II, including M. jalaludinii spp. (B) The presence/absence matrix displays gene clusters (columns) across all genomes (rows) belonging to Cluster M-I. Blue blocks indicate the presence of a gene cluster in a given genome. The left-side line graph represents the number of strain-specific (unique) gene clusters found exclusively in each genome. The gene matrix is sorted according to gene frequency from core to accessory regions

Based on this genome information, a comprehensive in silico safety analysis of M. multacida AGMB01124 was performed using multiple bioinformatics tools. First, a virulence factor search was conducted against the virulence database under relaxed thresholds (> 60% identity, > 60% coverage, and E-value < 1e − 10), which revealed no putative virulence-associated genes within the genome. In parallel, screening against the CARD confirmed that there are two genes that encode tetracycline-resistant ribosomal protection protein (tetW/N/W) and erythromycin resistance methylase 23 S ribosomal RNA methyltransferase (ermB), suggesting the necessity of subsequent in vitro validation to confirm the phenotypic expression of these resistance genes (Table 1). HGT potential was further assessed, revealing two putative gene acquisition events in which AGMB01124 acted as the recipient, and one gene was predicted to have been donated (Table 2). These included acquired genes encoding the translation initiation factor (infA, AGMB01124_01663) and cold shock protein (cspB, AGMB01124_02040), as well as a donated gene encoding the 30 S ribosomal protein S18 (rpsR, AGMB01124_02198). None of these genes was associated with virulence or antimicrobial resistance.

Table 1.

Predicted antibiotic resistance genes in M. multacida AGMB01124 based on CARD

AROa term	AMRb gene family	Drug class	Resistance mechanism	% Identity of Matching region	% Length of reference sequence
tet(W/N/W)	Tetracycline-resistant ribosomal protection protein	Tetracycline antibiotic	Antibiotic target protection	92.33	100.00
ermB	Erm 23 S ribosomal RNA methyltransferase	Macrolide antibiotic, lincoamide antibiotic, streptogramin antibiotic, streptogramin A antibiotic, streptogramin B antibiotic	Antibiotic target protection	98.78	100.81

^aARO, Antibiotic-resistant organism

^bAMR, Antimicrobial resistance

Table 2.

Horizontal gene transfer analysis of M. multacida AGMB01124

Locus tag	Gene	Product	Predicted donors/recipients (Reference sequence source)	Virulence factor	Antibiotic resistance
As recipient
AGMB01124_01663	infA	Translation initiation factor	Exiguobacterium sp. (GCF_000023045.1)	NAa	NA
AGMB01124_02040	cspB	Cold shock protein	Kroppenstedtia pulmonis (GCF_013265585.1)	NA	NA
As donor
AGMB01124_02198	rpsR	30 S ribosomal protein S18	Desulfotomaculum reducens (GCF_000016165.1)	NA	NA

^aNA, not applicable

In Vitro Probiotic Characteristics of M. Multacida AGMB01124

Profile of Carbohydrate Utilization

To evaluate the carbohydrate utilization profile of M. multacida AGMB01124 isolated from swine feces, its ability to ferment 49 different carbohydrates was assessed using the API kit (Supplementary Table 4). M. multacida AGMB01124 fermented 22 out of 49 tested carbon sources. These included seven monosaccharides (L-Arabinose, D-Ribose, D-Xylose, D-Galactose, D-Glucose, D-Fructose, and D-Mannose), six disaccharides (D-Maltose, D-Lactose, D-Melibiose, D-Saccharose, D-Trehalose, and D-Turanose), three polysaccharides (D-Melezitose, D-Raffinose, and Amidon), five sugar alcohols (Inositol, D-Mannitol, D-Sorbitol, D-Arabitol, and L-Arabitol), and one secondary amide (N-Acetylglucosamine).

Auto-/Co-Aggregation and Hydrophobicity

The auto-aggregation ability of M. multacida AGMB01124 was examined at 3, 6, 12, and 24 h (Fig. 3A). The auto-aggregation rate of M. multacida AGMB01124 was observed to increase in a time-dependent manner with 14.01 ± 1.86, 31.65 ± 1.80, 69.38 ± 0.63, and 84.97 ± 1.29% at 3, 6, 12, and 24 h, respectively. The co-aggregation capacity of M. multacida AGMB01124 with the pathogenic bacteria E. coli O157:H7 ATCC43889, S. Typhimurium UK1 ATCC 68,169, S. aureus KCCM 11,335, and L. monocytogenes KCCM 40,307 is shown in Fig. 3B. The co-aggregation rate of M. multacida AGMB01124 against E. coli O157:H7 ATCC43889, L. monocytogenes KCCM 40,307, S. aureus KCCM 11,335, and S. Typhimurium UK1 ATCC 68,169 revealed 18.24 ± 1.43, 44.97 ± 2.00, 54.70 ± 3.09, and 35.06 ± 1.48, respectively. The co-aggregation capacity of M. multacida AGMB01124 against S. aureus KCCM 11,335 was significantly higher than that of E. coli O157:H7 ATCC43889, L. monocytogenes KCCM 40,307, and S. Typhimurium UK1 ATCC 68,169. The hydrophobicity of M. multacida AGMB01124 for different solvents, including chloroform, ethyl acetate, n-hexane, n-hexadecane, toluene, and xylene, is shown in Fig. 3C. The hydrophobicity of chloroform, ethyl acetate, n-hexane, n-hexadecane, toluene, and xylene was 53.55 ± 0.78, 35.35 ± 0.52, 23.62 ± 2.72, 12.71 ± 0.36, 19.01 ± 5.25, and 17.89 ± 1.48% respectively, and was below 50% excluding chloroform.

Fig. 3.

Probiotic properties of M. multacida AGMB01124 isolated from the stool of healthy swine. (A) Auto-aggregation, (B) Co-aggregation, (C) Hydrophobicity, (D) Acid tolerance, (E) Bile tolerance, (F) Cytotoxicity, and (G) Hemolytic activity. Results are presented as mean ± SD. N.D: Not detected (detection limit: 1 log CFU/mL). Bars with different superscript letters ^(a-d) indicate significant differences at p < 0.05 based on one-way ANOVA with Tukey’s post hoc test. * p < 0.05 and *** p < 0.001 evaluated using a two-tailed t-test compared with the control

Acid and Bile Salt Tolerance

The acid and bile salt tolerance of M. multacida AGMB01124 under simulated GI conditions was shown in Fig. 3D and E. According to the findings obtained, acid tolerance at pH 2, 3, and 4 after 2 h of incubation decreased by 5–8 log CFU/mL compared to the control (8.60 ± 0.09 log CFU/mL) with viable cell counts of < 1.00 ± 0.00, 2.59 ± 0.30, and 3.41 ± 0.25 log CFU/mL, respectively. Bile salt tolerance after 6 h of cultivation at concentrations of 0.1%, 0.2%, and 0.3% resulted in viable cell counts of 9.44 ± 0.19, 6.86 ± 0.20, and 5.06 ± 0.02 log CFU/mL, respectively, representing a reduction of 2–5 log CFU/mL compared to the control (11.64 ± 0.13 log CFU/mL).

In Vitro Safety Assessment of M. Multacida AGMB01124

Hemolytic Activity

The hemolytic activity of M. multacida AGMB01124 isolated from swine feces was evaluated after 48 h of incubation in TSAB plates under anaerobic conditions (Fig. 3G). S. aureus KCCM 11,335 was used as a positive control for β-hemolysis. M. multacida AGMB01124 strain exhibited no hemolytic activity (γ-hemolysis), as evidenced by the absence of greenish discoloration (α-hemolysis) and clear zones (β-hemolysis) around the colonies.

Antibiotic Resistance and Cytotoxicity in Intestinal Epithelial Cells

To experimentally support the genome-based safety predictions, in vitro assays were performed to assess antibiotic susceptibility and cytotoxicity. Antibiotic susceptibility test based on CLSI guidelines revealed that strain AGMB01124 was sensitive to tetracycline, moxifloxacin, chloramphenicol, cefotetan, and ertapenem, but exhibited resistance to ampicillin (Table 3, Supplementary Fig. 2). Notably, no corresponding ampicillin resistance genes were detected in the CARD, indicating that the observed resistance may be attributable to unidentified genetic mutations rather than the acquisition of known resistance determinants, suggesting that the strain may still be considered safe for practical applications. Moreover, genome-based analysis revealed that AGMB01124 harbored erythromycin resistance methylase (Table 1), and its erythromycin resistance phenotype was also confirmed via in vitro susceptibility test (Supplementary Fig. 2). In cytotoxicity evaluation using Caco-2 cells, AGMB01124 demonstrated no cytotoxic effects, in contrast to the strong cytotoxic response elicited by the positive control strain E. coli O157:H7 ATCC 48,395 (Fig. 3F), further supporting its safety profile.

Table 3.

Antimicrobial susceptibility of M. multacida AGMB01124 isolated from healthy swine stool using CLSI guidelines

Guideline	Antibiotics	Breakpoint (mg/L)b	MIC (mg/L)
CLSIa	Ampicillin	≥ 2	24
	Tetracycline	≥ 16	8
	Moxifloxacin	≥ 8	6
	Chloramphenicol	≥ 32	0.75
	Cefotetan	≥ 64	< 8
	Ceftriaxone	≥ 64	< 8
	Erythromycin	-	> 256

^a Clinical and Laboratory Standards Institute

^b Breakpoint value was suggested for Gram-negative anaerobes

In Vivo Oral Toxicity Study of M. Multacida AGMB01124

General Observation

In an oral toxicity study, normal male and female mice were administered M. multacida AGMB01124 strain daily via oral gavage at three different doses: 10⁷ (low-dose, LD), 10⁸ (middle-dose, MD), and 10⁹ CFU (high dose, HD) for four weeks. No deaths related to oral administration were observed during the experimental period (Supplementary Table 5). Furthermore, no clinical signs were observed in any of the control and treatment groups (n = 5 per group).

Assessment of Body Weight, Daily Feed Consumption, and Daily Water Intake

The results of body weight, feed consumption, and water intake following the daily administration of M. multacida AGMB01124 to normal male and female mice in the experiment groups are presented in Supplementary Figs. 3 and 4. During the four-week experimental period, no significant differences in body weight were observed between the control and treatment groups of male and female mice. Similarly, daily feed consumption and water intake did not differ significantly between all groups.

Measurement of Relative Organ Weights and Colon Lengths

The relative organ weights and colon lengths of normal male and female mice orally administered M. multacida AGMB01124 daily for four weeks are presented in Supplementary Figs. 5 and 6. The relative organ weights of the kidney (male: 1.31–1.35%, female: 1.25–1.27%), liver (male: 4.17–4.82%, female: 4.27–4.71%) and spleen (male: 0.23–0.28%, female: 0.31–0.35%) did not differ significantly between the control and treatment groups of normal male and female mice, respectively (p > 0.05). Additionally, colon length showed no significant differences among the groups in normal male and female mice.

Hematological and Serum Biochemical Analysis

Hematological parameters such as red blood cell (RBC), hemoglobin (HGB), hematocrit (HCT), mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), mean corpuscular hemoglobin concentration (MCHC), platelet count (PLT), and white blood cell (WBC) in normal male and female mice orally administered M. multacida AGMB01124 strain daily for four weeks are presented in Table 4. In male mice, the HD group (46.80 ± 0.44 fL) showed a significant reduction in MCV level compared with the control group (48.08 ± 0.94 fL). In female mice, the HCT value in the control group was 47.76 ± 1.26%, but it decreased significantly to 41.14 ± 5.68% in the MD group. In contrast, the MCHC value in the control group was 30.44 ± 0.34 g/dL but increased to 31.34 ± 0.75 g/dL in the MD group.

Table 4.

Hematological parameters of normal male and female mice orally administered M. multacida AGMB01124 at different concentrations for four weeks

Sex	Parameter	CON	LD	MD	HD
Male	RBC (×106 cells/µL)	9.55 ± 0.59	9.56 ± 0.74	9.47 ± 0.72	9.16 ± 1.54
	HGB (g/dL)	13.84 ± 0.84	13.82 ± 1.05	13.62 ± 0.96	13.04 ± 2.07
	HCT (%)	45.90 ± 3.25	45.54 ± 3.63	45.36 ± 3.20	42.88 ± 7.37
	MCV (fL)	48.08 ± 0.94a	47.64 ± 0.26ab	47.94 ± 0.86ab	46.80 ± 0.44b
	MCH (pg)	14.50 ± 0.07	14.46 ± 0.09	14.40 ± 0.12	14.26 ± 0.32
	MCHC (g/dL)	30.20 ± 0.47	30.36 ± 0.25	30.04 ± 0.32	30.48 ± 0.64
	PLT (×106 cells/µL)	1322.00 ± 194.72	1388.60 ± 150.96	1388.00 ± 392.98	1115.50 ± 191.94
	WBC (×103 cells/µL)	3.27 ± 1.39	4.79 ± 1.18	5.36 ± 1.52	5.61 ± 2.34
Female	RBC (×106 cells/µL)	9.98 ± 0.34	9.33 ± 0.60	8.72 ± 1.25	9.67 ± 0.34
	HGB (g/dL)	14.54 ± 0.44	13.64 ± 0.88	12.90 ± 1.83	13.82 ± 0.53
	HCT (%)	47.76 ± 1.26a	44.44 ± 3.12ab	41.14 ± 5.68b	45.78 ± 1.76ab
	MCV (fL)	47.82 ± 0.50	47.64 ± 0.59	47.22 ± 0.44	47.34 ± 0.48
	MCH (pg)	14.54 ± 0.15ab	14.64 ± 0.19ab	14.82 ± 0.44a	14.28 ± 0.19b
	MCHC (g/dL)	30.44 ± 0.34b	30.70 ± 0.25ab	31.34 ± 0.75a	30.20 ± 0.14b
	PLT (×103 cells/µL)	1346.50 ± 212.55	1020.67 ± 351.91	1056.00 ± 566.44	1356.00 ± 311.63
	WBC (×103 cells/µL)	3.87 ± 0.77ab	2.47 ± 0.79b	2.60 ± 1.38b	5.32 ± 0.68a

Results are presented as mean ± SD of five mice

Results with different superscript letters^(a-b) within rows are significantly different at P < 0.05

LD, Low-dose (10⁷ CFU); MD, Middle-dose (10⁸ CFU); HD, High-dose (10⁹ CFU)

Table 5 showed the results of serum biochemistry parameters, including albumin (ALB), alkaline phosphatase (ALP), alanine aminotransferase (ALT) aspartate aminotransferase (AST), calcium (Ca), creatinine (CRE), γ-glutamyltransferase (GGT), glucose (GLU), high-density lipoprotein cholesterol (HDL-C), inorganic phosphorus (IP), total bilirubin (TBIL), total cholesterol (TCHO), triglyceride (TG), total protein (TP), and uric acid (UA). In male mice, TBIL values were significantly decreased in the LD, MD, and HD groups (0.76 ± 0.24, 0.88 ± 0.39, and 0.88 ± 0.17 mg/dL, respectively) compared with the control group (1.30 ± 0.17 mg/dL). The TCHO value in the control group was 63.00 ± 2.92 mg/dL, whereas it significantly increased to 76.60 ± 2.61 mg/dL in the LD group. In contrast, no significant differences were observed in serum biochemical indicators in female mice.

Table 5.

Serum biochemistry parameters of normal male and female mice orally administered M. multacida AGMB01124 at different concentrations for four weeks

Sex	Parameter	CON	LD	MD	HD
Male	ALB (g/dL)	2.98 ± 0.19	2.78 ± 0.19	2.70 ± 0.23	2.83 ± 0.05
	ALP (U/L)	261.60 ± 15.27	283.00 ± 30.98	267.00 ± 39.60	265.75 ± 14.48
	CRE (mg/dL)	0.42 ± 0.16	0.36 ± 0.12	0.30 ± 0.14	0.23 ± 0.02
	Ca (mg/dL)	10.92 ± 1.30	11.18 ± 0.45	11.54 ± 0.55	11.05 ± 0.25
	GGT (U/L)	1.40 ± 0.55	1.00 ± 0.00	1.20 ± 0.45	1.00 ± 0.00
	GLU (mg/dL)	262.75 ± 26.58	281.80 ± 25.91	260.20 ± 31.37	287.50 ± 5.78
	AST (U/L)	85.67 ± 49.81	86.80 ± 26.34	73.00 ± 27.22	82.25 ± 14.86
	ALT (U/L)	31.00 ± 8.72	31.20 ± 6.61	28.50 ± 8.10	30.50 ± 3.00
	HDL-C (mg/dL)	64.00 ± 2.74	75.40 ± 3.51	62.80 ± 12.28	73.75 ± 6.65
	IP (mg/dL)	8.16 ± 2.11	7.32 ± 0.72	8.62 ± 0.70	8.23 ± 1.07
	TBIL (mg/dL)	1.30 ± 0.17a	0.76 ± 0.24b	0.88 ± 0.39b	0.88 ± 0.17b
	TCHO (mg/dL)	63.00 ± 2.92b	76.60 ± 2.61a	67.00 ± 9.19ab	71.25 ± 2.75ab
	TG (mg/dL)	135.20 ± 12.87ab	144.20 ± 18.46a	108.80 ± 23.69b	106.50 ± 16.90b
	TP (g/dL)	5.62 ± 0.19	5.48 ± 0.29	5.34 ± 0.27	5.43 ± 0.10
	UA (mg/dL)	2.28 ± 0.83	1.64 ± 0.42	1.95 ± 0.21	1.88 ± 0.41
Female	ALB (g/dL)	3.36 ± 0.46	3.28 ± 0.28	3.20 ± 0.26	3.22 ± 0.26
	ALP (U/L)	329.40 ± 26.81	368.50 ± 17.25	332.33 ± 36.83	369.50 ± 39.20
	CRE (mg/dL)	0.23 ± 0.06	0.21 ± 0.04	0.20 ± 0.01	0.24 ± 0.08
	Ca (mg/dL)	11.18 ± 0.83	10.82 ± 0.16	10.93 ± 0.40	11.22 ± 0.39
	GGT (U/L)	1.20 ± 0.45	1.00 ± 0.00	1.00 ± 0.00	1.00 ± 0.00
	GLU (mg/dL)	263.20 ± 35.79	292.60 ± 26.67	285.33 ± 28.57	278.00 ± 21.46
	AST (U/L)	101.50 ± 43.07ab	134.67 ± 17.24a	79.00 ± 30.51ab	77.60 ± 36.87b
	ALT (U/L)	25.40 ± 4.16	33.80 ± 3.35	27.33 ± 5.03	26.00 ± 5.83
	HDL-C (mg/dL)	63.00 ± 7.48	58.40 ± 12.34	65.33 ± 2.08	60.20 ± 12.07
	IP (mg/dL)	8.44 ± 1.53	9.14 ± 0.41	7.23 ± 0.74	7.46 ± 1.45
	TBIL (mg/dL)	1.14 ± 0.65	1.12 ± 0.35	1.03 ± 0.31	0.60 ± 0.48
	TCHO (mg/dL)	66.40 ± 5.86	67.20 ± 10.92	70.33 ± 0.58	63.80 ± 11.73
	TG (mg/dL)	97.00 ± 13.71	101.80 ± 17.48	123.33 ± 11.72	118.20 ± 27.13
	TP (g/dL)	5.34 ± 0.34	5.26 ± 0.11	5.37 ± 0.15	5.06 ± 0.15
	UA (mg/dL)	2.12 ± 0.68	2.10 ± 1.04	1.60 ± 0.44	1.42 ± 0.58

Results are presented as mean ± SD of five mice

Results with different superscript letters^(a-b) within rows are significantly different at P < 0.05

LD, Low-dose (10⁷ CFU); MD, Middle-dose (10⁸ CFU); HD, High-dose (10⁹ CFU)

Histopathological Analysis

Histological analysis was conducted on the liver, kidney, spleen, and colon of mice orally administered M. multacida AGMB01124 for four weeks. Histopathological examination revealed no lesions such as infection, necrosis, iron deposition, and bilirubin pigmentation in the liver, kidneys, spleen, and colon of male and female mice (Supplementary Figs. 7 and 8). According to NCI Common Toxicity Criteria, no significant histopathologic changes were observed in the liver, kidney, spleen, and colon of the control and treatment groups, with all tissues classified as Grade 5 (data not shown).

Harmful Enzyme Activity of the Cecum

Several harmful enzymes, such as β-glucosidase and β-glucuronidase, catalyze the release of carcinogenic aglycons from heterocyclic aromatic amines, polycyclic aromatic hydrocarbons, and certain bile acids in the GI tract [45]. Therefore, we investigated the effects of oral administration of M. multacida AGMB01124 on these enzymes in the cecum (Fig. 4). In normal male mice, β-glucosidase activities were 1.90 ± 0.29, 2.20 ± 0.32, and 1.82 ± 0.54 µMh/mg in the control, LD, and MD groups, respectively, with no significant differences observed among the groups. However, β-glucosidase activity significantly decreased to 1.34 ± 0.28 µMh/mg in the HD group (p < 0.05). Similarly, β-glucuronidase activity was significantly lower in the HD group (0.77 ± 0.09 µMh/mg) than in the control (1.11 ± 0.16 µMh/mg). In normal female mice, β-glucosidase and β-glucuronidase activities ranged from 0.95 to 1.11 µMh/mg and 1.55–1.71 µMh/mg, respectively, with no significant differences observed among the groups.

Fig. 4.

Harmful enzyme activities, including β-glucosidase and β-glucuronidase in the cecum of normal male (A, B) and female (C, D) mice orally administered M. multacida AGMB01124 at different concentrations for four weeks. (A, C) β-glucosidase and (B, D) β-glucuronidase. LD, Low-dose (10⁷ CFU); MD, Middle-dose (10⁸ CFU); HD, High-dose (10⁹ CFU). Results are presented as mean ± SD of five mice. ** p < 0.01 evaluated using a two-tailed t-test compared with the control

SCFA Analysis of Stool Samples

To investigate the effect of M. multacida AGMB01124 on SCFAs, which are beneficial metabolites in the GI tract of mice, we measured the concentrations of acetic, propionic, butyric, isobutyric, valeric, and isovaleric acid in fresh stool samples collected after four weeks of oral administration. In male mice, the treatment group showed a significant increase in the concentrations of propionic, butyric, isobutyric, valeric, and isovaleric acid compared with the control group (Fig. 5A–F). Consequently, total SCFA content also significantly increased, with the LD group exhibiting the highest levels (Fig. 5G). Principal coordinate analysis (PCA) based on SCFAs content revealed clear separation between the control and treatment groups, and PERMANOVA confirmed significant differences (F = 10.516, R² = 0.66349, p = 0.001) (Fig. 5H). In female mice, the SCFA concentration increased in all treatment groups for acetic acid, in the LD group for propionic acid, in the LD and MD groups for butyric and isobutyric, and valeric acids, and in the MD group for isovaleric acid (Fig. 6A–F). Overall, the total SCFA levels were highest in the LD group, with a dose-dependent decrease observed across treatment groups (Fig. 6G). PCA analysis confirmed distinct clustering of all groups, and PERMANOVA results indicated statistically significant differences (F-value = 16.769, R-squared = 0.7587, p-value = 0.001) (Fig. 6H).

Fig. 5.

SCFA concentrations in the stool of normal male mice orally administered M. multacida AGMB01124 at different concentrations for 28 consecutive days. (A) Acetic acid, (B) Propionic acid, (C) Butyric acid, (D) Isobutyric acid, (E) Valeric acid, (F) Isovaleric acid, (G) Total SCFAs, and (H) PCA plot. LD, Low-dose (10⁷ CFU); MD, Middle-dose (10⁸ CFU); HD, High-dose (10⁹ CFU). Results are presented as mean ± SD of five mice. * p < 0.05, ** p < 0.01, and *** p < 0.001 evaluated using two-tailed t-test compared with the control

Fig. 6.

SCFA concentrations in the stool of normal female mice orally administered M. multacida AGMB01124 at different concentrations for 28 consecutive days. (A) Acetic acid, (B) Propionic acid, (C) Butyric acid, (D) Isobutyric acid, (E) Valeric acid, (F) Isovaleric acid, (G) Total SCFAs, and (H) PCA plot. LD, Low-dose (10⁷ CFU); MD, Middle-dose (10⁸ CFU); HD, High-dose (10⁹ CFU). Results are presented as mean ± SD of five mice. * p < 0.05, ** p < 0.01, and *** p < 0.001 evaluated using two-tailed t-test compared with the control

Discussion

Recently, NGPs have been reported to play an important role in promoting host health by enhancing the intestinal epithelial barrier, modulating the gut microbiota composition, and improving digestive function [9, 20, 47]. Additionally, GI disorders such as irritable bowel syndrome, inflammatory bowel disease, ulcerative colitis, Crohn’s disease, and diarrhea have been associated with alterations in gut microbiota, including microbiome dysbiosis, the presence of harmful bacteria, and the loss of beneficial bacteria [4, 8]. Moreover, beneficial metabolites such as SCFAs produced by commensal bacteria are gaining attention as a natural and effective strategy for recovering from GI disorders due to their ability to supply energy to intestinal epithelial cells and inhibit the growth of pathogenic microorganisms [9]. Under normal physiological conditions in mammals, nutrients are broken down by various enzymes such as amylase, lipase, and protease, and are mainly digested in the small intestine [48]. However, dietary fiber such as resistant starch and non-starch polysaccharides is not digested in the small intestine due to the lack of specific enzymes required for their breakdown, and thus passes into the large intestine [49]. Commensal bacteria are most abundant in the large intestine, where they ferment non-digested carbohydrates to produce straight-chain SCFAs (acetic, propionic, butyric, and valeric acids), while proteolytic fermentation of amino acids generates branched-chain SCFAs such as isobutyric and isovaleric acids [50, 51]. Therefore, we selected the M. multacida AGMB01124 strain through the screening of cellulase, amylase, proteinase, lipase, and β-galactosidase activities in 100 species of commensal microorganisms isolated from the stool of healthy Korean cattle and swine.

Based on enzyme screening results, functional annotation was performed on two enzymes (cellulase and β-galactosidase) that exhibited significantly higher activities in M. multacida AGMB01124 (Supplementary Tables 2 and 3). For β-galactosidase, all annotated sequences from publicly available M. multacida genomes were identical, suggesting that the observed activity is not attributable to strain-specific sequence variation. In contrast, endoglucanase D (CelD), a type of cellulase, was identified in 20 of 29 M. multacida genomes and classified into two distinct sequence types. This classification was supported by multiple sequence alignment, which revealed several amino acid differences across the two groups; among them, four residues (positions 19, 260, 337, and 383) were located on the protein surface, suggesting potential structural or functional relevance (Supplementary Fig. 9). Both types share a canonical domain structure consisting of a conserved C-terminal catalytic domain, an N-terminal domain, and a signal peptide (Supplementary Fig. 10A). However, AlphaFold3-based structural predictions revealed four amino acid residues (positions 19, 260, 337, and 383) that differ between the two types and are exposed on the protein. Among these, three (260, 337, and 383) are located near the active site and one (19) within the signal peptide region (Supplementary Fig. 10C–E). Structural superposition of the two types (CelD_I and CelD_II) demonstrated that three of these residues (260, 337, and 383) are positioned near or within the predicted active site (Supplementary Fig. 10B), indicating a high likelihood that they contribute to functional divergence in catalytic activity. Previous studies have shown that the presence of specific amino acids near the active site can significantly alter catalytic performance: lysine may enhance electrostatic stabilization of the transition state [52, 53], while glycine may confer increased local flexibility and facilitate favorable rearrangement of the active site geometry [54]. These substitutions may contribute to strain-specific functional differences in cellulase activity. Furthermore, comparative analysis of whole-genome-based phylogeny revealed that CelD types were non-randomly distributed across phylogenetic clusters (Supplementary Fig. 10F). Strains within Cluster M-I, including AGMB01124, predominantly harbored the same CelD type (CelD_I), characterized by K260, G337, G383, and A019. This suggests that sequence-level divergence in key residues may be associated with functional differentiation across phylogenetic lineages.

In vitro studies were conducted to evaluate the safety and probiotic characteristics of M. multacida AGMB01124 by assessing its carbohydrate utilization profile, antimicrobial susceptibility, auto- and co-aggregation abilities, surface hydrophobicity, acid and bile salt tolerance, and hemolytic activity. M. multacida AGMB01124 was characterized for its carbohydrate utilization ability using the API 50 CH kit and digested the following carbon sources: L-arabinose, D-xylose, D-galactose, D-glucose, D-fructose, D-mannose, D-maltose, D-lactose, D-melibiose, D-saccharose, D-trehalose, D-turanose, D-melezitose, D-raffinose, amidon, inositol, D-mannitol, D-sorbitol, D-arabitol, L-arabitol, and N-acetylglucosamine. The AGMB01124 strain was defective in the fermentation of carbon sources, including glycerol, erythritol, D-arabinose, L-xylose, D-adonitol, methyl-β-D-xylopyranoside, L-sorbose, L-rhamnose, dulcitol, methyl-α-D-mannopyranoside, methyl-α-D-glucopyranoside, amygdalin, arbutin, esculin ferric citrate, salicin, D-cellobiose, inulin, glycogen, xylitol, gentiobiose, D-lyxose, D-tagatose, D-fucose, L-fucose, potassium gluconate, potassium 2-ketogluconate, and potassium 5-ketogluconate. Overall, the M. multacida AGMB01124 strain was found to metabolize 22 carbon sources, including monosaccharides, disaccharides, polysaccharides, sugar alcohol, and secondary amide. These results showed that M. multacida AGMB01124 could grow by utilizing different types of carbohydrates in the GI environment.

The GI tract harbors approximately 10 to 100 trillion commensal microorganisms (bacteria, fungi, archaea, viruses, and protozoa), and alterations in microbiome composition can majorly affect host health [55]. Aggregation and hydrophilicity are considered desirable biological traits of probiotic microorganisms because they influence various functions such as adhesion, colonization, and biofilm formation [56]. Auto-aggregation behavior is the gathering of the same microbial strains to increase the adhesion of intestinal epithelial cells [57]. Co-aggregation is also a beneficial attribute of probiotic strains due to its capability to inhibit the surface colonization and adhesion of pathogens on host mucosal cells [58]. Microbial surface hydrophobicity can be assessed using different organic solvents, including xylene, toluene, hexadecane, hexane, ethyl acetate, and chloroform [59]. Non-polar solvents such as xylene, toluene, hexadecane, and hexane are used to reflect the hydrophobic or hydrophilic nature of the cell membrane [60]. Polar solvents such as chloroform and ethyl acetate are used to assess the ability of cell walls as electron donors (basic) and electron acceptors (acidic), respectively [59]. In our study, M. multacida AGMB01124 achieved a high level of auto-aggregation (84.97 ± 1.29% after 24 h) and showed over 40% co-aggregation capacity with pathogens such as S. aureus and L. monocytogenes. M. multacida AGMB01124 showed a low hydrophobicity of the cell membrane, with values below 30% in non-polar solvents such as xylene, toluene, hexadecane, and hexane. Furthermore, it exhibited greater affinity for chloroform (an acidic solvent and electron donor) than for ethyl acetate (a basic solvent and electron acceptor). The hydrophobicity of traditional probiotics has been reported to range from 6.1 to 76.0% for Limosilactobacillus reuteri, 11.3 to 60.7% for Lacticaseibacillus casei subsp. paracasei, and 20.4 to 87.4% for Bifidobacterium thermophilum [57]. Previous studies have revealed significant differences in hydrophilicity even among strains belonging to the same species of traditional probiotics, indicating that the health benefits to the host are mediated not only by hydrophilicity but also by various mechanisms acting in the GI environment. These results showed that M. multacida AGMB01124 isolated from swine feces had remarkable auto-aggregation ability and co-aggregation with pathogenic bacteria such as S. aureus and L. monocytogenes. Furthermore, it was confirmed that the cell membrane exhibited low hydrophobicity and the cell wall played an electron donor rather than an electron acceptor.

To assess survival in the GI tract, potential probiotic strains are typically evaluated under stress conditions (e.g., low pH and bile salt), which reflect the environments of gastric juice and intestinal fluid [61–63]. The GI simulation model employs various media, pH levels, incubation times, and additives such as bile salt and pepsin to estimate the survival of candidate probiotics under extreme conditions [63]. In our study, the acid tolerance of M. multacida AGMB01124 was below the detection limit (< 1 log CFU/mL) under the lowest acidic conditions (pH 2), while survival rates of 30.09% and 39.69% were observed at pH 3 and 4, respectively. Bile tolerance of M. multacida AGMB01124 showed survival rates of 81.13%, 58.95%, and 43.48% in media containing 0.1%, 0.2%, and 0.3% bile salt, respectively. These results are consistent with previous studies. Lim et al. [64] have reported that 39 out of 57 Lactobacillaceae and Streptococcus strains were viable at pH 2.5, while 12 out of 43 Bifidobacterium strains survived at pH 3. Similarly, Ruiz-Moyano et al. [65] have found that 116 lactic acid bacteria strains isolated from artisanal soft cheese did not survive when exposed to pH 2.5, but all strains survived at pH 3. The pH of gastric juice ranges from 1.5 in a fasting state to 4.5 after food intake, and gastric transit can take up to 3 h [66]. These conditions are commonly used as initial parameters to evaluate the viability of probiotics. In the small intestine, bile salts secreted into the duodenum at concentrations of approximately 0.3% play a key role in lipid digestion [67]. In addition to their digestive function, bile salts exhibit antimicrobial activity by disrupting the lipid components of bacterial cell membranes [68]. Therefore, probiotic candidates are required to tolerate both acidic conditions and bile salts to survive in the GI tract. In summary, these results suggest that M. multacida AGMB01124 is resistant to pH 3 and 0.3% bile salts, which indicates its potential as a probiotic candidate.

We evaluated the in vitro safety of M. multacida AGMB01124 by performing hemolysis, antibiotic resistance, and cytotoxicity assays. Hemolytic evaluation of probiotics is essential because some strains, such as Enterococcus and Streptococcus, can cause hemolysis-related disorders such as anemia and edema [42]. Hemolytic activity is classified into three types: α-hemolysis (green zone around the colony), β-hemolysis (clear zone around the colony), and γ-hemolysis (no change around the colony). M. multacida AGMB01124 showed no change around the colony, indicating that it was non-hemolytic. Antibiotic resistance was evaluated using ampicillin, tetracycline, moxifloxacin, chloramphenicol, cefotetan, and ertapenem in accordance with CLSI guidelines, and erythromycin was additionally evaluated based on genome-based prediction. The M. multacida AGMB01124 strain harbored the tetracycline resistance gene tet (W/N/W) and exhibited phenotypic resistance to tetracycline. Conversely, although no genes associated with ampicillin resistance were detected, the strain exhibited phenotypic resistance to ampicillin. This inconsistency between genotypes and phenotypes suggests the involvement of alternative resistance mechanisms, such as efflux pumps, outer membrane permeability, or unspecified genetic factors not included in the currently available resistance gene database [69, 70]. None of the identified resistance genes were associated with mobile genetic elements, and HGT analysis did not predict that the strain would act as a donor for horizontal gene transfer. Furthermore, M. multacida AGMB01124 exhibited no cytotoxicity in contrast to the high cytotoxicity observed with E. coli O157:H7. Therefore, in vitro safety evaluation demonstrated that M. multacida AGMB01124 was non-hemolytic, non-cytotoxic, and susceptible to most antibiotics tested in accordance with CLSI guidelines. These findings support the potential of M. multacida AGMB01124 as a safe NGP candidate.

In vivo oral toxicity assessment using suitable animal models has been recognized as the standard method for confirming the safety of novel probiotic strains. According to Punarvasu et al. [71], animal toxicity studies considered body weight, daily feed consumption, daily water intake, and relative organ weights as important indicators for evaluating the harmful effects of a substance [72]. Additionally, lesions such as necrosis, fibrosis, and inflammation within the tissue are useful for evaluating the toxicity of treatment substances [73, 74]. In this study, no differences were observed in clinical signs, mortality, body weight, daily food consumption, daily water intake, relative organ weights (kidney, liver, and spleen), colon length, or histopathological examination between the control group and the M. multacida AGMB01124 administration group, indicating no evidence of harmful effects. Furthermore, hematological parameters are employed to assess inflammation and infectious diseases, while serum biochemical parameters are utilized to evaluate glucose levels and organ function [75]. Previous studies have reported reference ranges for hematological and serum parameters [76–78], and the results of this study were within normal values in both the treatment and control groups. This suggests that the administration of the M. multacida AGMB01124 strain did not adversely affect physiological functions, including immune and organ-related responses.

The activity of enzymes such as β-glucosidase and β-glucuronidase is closely associated with the presence of harmful microorganisms in the GI environment [45] and is considered a biological indicator of adverse effects of the gut microbial community [79]. Previous studies have reported that these enzymes may act as potential mediators in the development of colon cancer. Probiotics have been suggested to regulate the gut microbial community and lower GI pH, thereby inhibiting the activity of these enzymes [79, 80]. In the present study, administration of a high dose (10⁹ CFU) significantly reduced the activity of these two harmful enzymes in male mice, suggesting a potential sex- and dose-dependent response. Therefore, these findings indicated that M. multacida AGMB01124 showed no adverse effects in the in vivo oral toxicity evaluation, including clinical signs, dietary intake, relative organ weights, hematological and serum biochemical profiles, histopathological observations, and harmful enzyme activity.

Commensal bacteria in the GI tract convert complex nutrients such as cellulose, pectin, hemicellulose, lignin, and mucin into simple sugars, which are then fermented to produce straight-chain SCFAs, including acetic, propionic, butyric, and valeric acids [50, 81]. SCFAs produced by the gut microbiota play a crucial role in maintaining host health and homeostasis, as well as influencing the development of various diseases and disorders [82, 83]. Thus, the absence of beneficial bacteria in the gut can lead to an increase in pathogenic bacteria and a reduction in SCFA levels. For example, several studies have reported changes in the gut microbiota and decreased SCFA levels in patients with Parkinson’s disease [84], Crohn’s disease [85], and irritable bowel syndrome [86] compared with healthy individuals. This suggests that probiotics are important not only for maintaining gut homeostasis but also for their potential in the treatment of specific diseases. In this study, the administration of M. multacida AGMB01124 significantly increased the SCFA levels in the treatment groups compared with the control group, regardless of sex. However, SCFA concentrations decreased in a dose-dependent manner, suggesting that low concentrations may exhibit optimal efficacy. Previous studies have reported that M. multacida is associated with butyrate production [87]. Furthermore, Bai et al. [23] have reported that M. multacida was enriched in the propionic and butyric acid metabolic pathways. Additionally, Luo et al. [24] indicated that M. multacida is associated with increased levels of acetic and butyric acid in the ileum of pigs. Acetic, propionic, and butyric acid have been extensively researched as the major SCFAs produced in the gut by the commensal microbiome. They provide energy to colon epithelial cells and contribute to maintaining gut function and regulating the immune response [88]. Conversely, valeric, isobutyric, and isovaleric acids are generally generated in relatively small amounts, and their concentrations in feces, associated microbial producers, and health effects are not well understood [89]. In a previous study, intraperitoneal administration of valeric acid was reported to be negatively associated with inflammatory markers in a DSS-induced colitis mouse model [90]. Additionally, oral administration of valeric acid has been reported to attenuate tissue damage and enhance intestinal epithelial integrity in radiation-exposed mouse models [91]. Isobutyric acid and isovaleric acid are branched-chain SCFAs produced by the microbial fermentation of branched-chain amino acids, primarily valine and leucine. Their production is modulated by dietary amino acid availability, host metabolic processes, and the composition and functional capacity of the gut microbiota [51]. In a pig model of DSS-induced colitis, sodium isobutyrate supplementation alleviated diarrhea, bloody stools, weight loss, and colonic shortening, and increased the levels of SCFAs and branched-chain SCFAs in the feces [92]. In steers administered isovalerate, it has been reported to enhance cellulolytic enzyme activities, while reducing protease activity and suppressing methanogenic bacteria [93]. These findings suggest that increases in SCFA levels, including both straight-chain and branched-chain forms, may contribute to beneficial metabolic and physiological effects in the gut environment. Therefore, M. multacida AGMB01124 exhibits functional metabolic effects in the gut, supporting its potential as a next-generation probiotic candidate.

Conclusion

In this study, the safety and probiotic potential of M. multacida AGMB01124, isolated from healthy swine feces, were comprehensively evaluated using in silico, in vitro, and in vivo studies. Genome analysis revealed the absence of virulence factors and the presence of two antibiotic resistance genes without mobile genetic elements. This strain showed potential probiotic properties, including non-hemolysis and non-cytotoxicity, tolerance to some acidic and bile conditions, strong aggregation ability, and the ability to ferment various carbohydrates. Oral toxicity evaluation revealed no adverse effects in terms of clinical signs, hematological and biochemical parameters, histopathological examinations, and harmful enzyme activities. Additionally, M. multacida AGMB01124 significantly increased SCFA concentrations, suggesting a functional contribution to gut metabolic health. Therefore, these findings demonstrate that M. multacida AGMB01124 exhibits probiotic characteristics and a safety profile, supporting its potential as an NGP. Further studies using disease-specific animal models are needed to investigate the efficacy of this strain.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Seo-Joon Yeom: Conceptualization, Data curation, Formal analysis, Methodology, Validation, Visualization, Writing – original draft, Writing – review & editing. Jihye Yang: Conceptualization, Data curation, Formal analysis, Methodology, Validation, Visualization, Writing – original draft, Writing – review & editing. Ki-Nam Yoon: Data curation, Formal analysis, Writing – original draft. Sang-Su Kim: Data curation, Formal analysis, Writing – original draft. Min-Ji Cho: Data curation, Formal analysis, Writing – original draft. Jong-Heum Park: Data curation, Formal analysis, Methodology, Validation, Writing – original draft. Beom-Seok Song: Data curation, Formal analysis, Methodology, Validation, Writing – original draft. Joo Hyung Park: Data curation, Formal analysis, Writing – original draft. Yeung Bae Jin: Data curation, Formal analysis, Writing – original draft. Young-Min Kim: Supervision, Validation, Writing – review & editing. Seung-Hwan Park: Supervision, Validation, Writing – review & editing. Ju Huck Lee: Supervision, Validation, Writing – review & editing. Hyeun Bum Kim: Supervision, Validation, Writing – review & editing. Ju-Hoon Lee: Methodology, Supervision, Validation, Writing – original draft, Writing – review & editing. Jae-Kyung Kim: Conceptualization, Methodology, Supervision, Writing – original draft, Writing – review & editing.

Funding

This work was supported by the Korea Atomic Energy Research Institute (KAERI) Institutional Program (Project No. 523320-26).

Data Availability

The data that support the findings of this study are available from the corresponding author upon reasonable request. The genome sequence of M. multacida AGMB01124 is available in the GenBank database under the accession number CP195933.

Declarations

Ethics Approval

The animal experiments were approved by the Committee on the Use and Care of Animals at the Korea Atomic Energy Research Institute (KAERI; approval no. KAERI-IACUC-2024-004).

Competing interests

The authors declare no competing interests.