Strain AGMB02718^T was isolated from swine feces collected from a swine farm located in the National Institute of Animal Science in Cheonan, Republic of Korea (36.93 N, 127.11 E). To isolate gut microbiota, the fecal samples were placed in anaerobic chamber (Coy Laboratory Products, Grass Lake, MI, United States) in an atmosphere consisting of 86% N₂, 7% CO₂, and 7% H₂. The stool was serially diluted to 10⁻⁴ in sterilized phosphate-buffered saline (PBS), and 100 μL of the diluted suspension was spread onto tryptic soy agar supplemented with 5% sheep blood (TSAB). After incubation for 72 h, single colonies were transferred to fresh TSAB. A convex, smooth, and circular colony of the strain AGMB02718^T was obtained and subjected to taxonomic analysis based on phenotypic, physiological, and phylogenetic characteristics. The isolate and reference strain, M. multiformis DSM 106860^T, were routinely cultured on TSAB plates at 37°C in an anaerobic chamber for 72 h and preserved at − 80°C in 10% (w/v) skim milk. E. coli DH5α was grown overnight in LB medium at 37°C with aeration. AGMB02718^T was deposited in the Korean Collection for Type Cultures (KCTC) and Guangdong Microbial Culture Collection (GDMCC).

For phylogenetic analysis, the 16S rRNA gene of strain AGMB02718^T was amplified by PCR from cell suspensions using the universal 16S rRNA bacterial primers 27F (5′-AGAGTTTGATCMTGGCTCAG-3′) and 1492R (5′-TACGGYTACCTTGTTACGACTT-3′). The amplified 16S rRNA gene was sequenced with the universal primers 518F (5′-CCAGCAGCCGCGGTAATACG-3′) and 805R (5′-GACTACCAGGGTATCTAATC-3′) (BIOFACT, Daejeon, Republic of Korea). The sequenced 16S rRNA gene was analyzed using BLAST searches in the EzBioCloud database (http://www.ezbiocloud.net; Yoon et al., 2017) and GenBank/EMBL/DDBJ databases to identify taxa related to strain AGMB02718^T. Multiple alignments were performed using CLUSTAL W (Thompson et al., 1997) and gaps were edited using BioEdit (Hall, 1999). Phylogenetic trees were constructed using Molecular Evolutionary Genetics Analysis 11 (MEGA 11) software (Tamura et al., 2021) with different algorithms based on neighbor-joining (NJ) (Saitou and Nei, 1987), maximum likelihood (ML) (Fitch, 1971), and maximum parsimony (MP) (Kolaczkowski and Thornton, 2004) methods, performed using 1,000 bootstrap values. In addition, the evolutionary distances of all trees were determined using the Kimura-2-parameters method set to default.

For the morphology analysis, strain AGMB02718^T was incubated anaerobically on TSAB plates for 72 h at 37°C. The cellular morphology was observed using phase-contrast microscopy (Eclipse 80i, Nikon, Tokyo, Japan), scanning electron microscopy (SEM; CX-200TA, Coxem, Daejeon, Republic of Korea), and transmission electron microscopy (TEM; CM-120, Philips, Amsterdam, Netherlands). For SEM analysis, the cultured cells were fixed with 4% paraformaldehyde and dehydrated with gradient ethanol solutions (10–100%). After fixation, samples were treated with isoamyl acetate and dried over hexamethyldisilazane (HMDS). Thereafter, the dried samples were coated with platinum using a sputter coater (SPT-20, Coxem, Daejeon, Republic of Korea) and observed by SEM. For the TEM analysis, the cells were fixed and applied to carbon-coated grids that had been glow-discharged for 3 min in air. Next, the grids were negatively stained with 1% uranyl acetate. Subsequently, the samples were observed using a TEM equipped with a lanthanum hexaboride cathode. Gram staining was performed using a commercial kit (Sigma-Aldrich, MA, United States) according to the manufacturer’s instructions. Various temperatures (15, 20, 25, 30, 37, 40, 45, and 50°C) were assayed to identify the growth range of the novel strain. The NaCl tolerance (0.5, 1, 2, 3, 4, and 5%, w/v) was measured on TSAB plates. The oxygen requirements for growth were examined by incubating strain AGMB02718^T under aerobic, microaerophilic (CO₂ incubator filled with, 10% O₂, 10% CO₂, and 80% N₂), and anaerobic (anaerobic chamber filled with 86% N₂, 7% CO₂, and 7% H₂) conditions for 72 h at 37°C. The activities of catalase and oxidase were tested by using a commercial reagent (bioMérieux, Marcy-l’Étoile, France) following the manufacturer’s instructions. Other biochemical features, such as enzyme activities, utilization of substrates, and acid production from carbohydrates, were determined by using API 20A, API ZYM, and Rapid ID 32A strips according to the manufacturer’s protocols (bioMérieux, Marcy-l’Étoile, France). Closely related strain types were tested under the same conditions.

The cellular fatty acid compositions of the strain AGMB02718^T and the reference strain were examined. After harvesting the cells grown for 72 h at 37°C on TSAB plates, the whole cellular fatty acids were obtained through saponification, methylation, and extraction as detailed in the MIDI/Hewlett Packard Microbial Identification System (Sasser, 1990). The cellular fatty acids obtained were analyzed using gas chromatography (6,890 N with a 7683 autosampler; Agilent Technologies, California, United States). Fatty acids were identified using the Sherlock Microbial Identification System in the Anaerobe database v 6.1. Quinones were extracted from 100 mg of freeze-dried cells using chloroform:methanol (2:1; v/v) and used for respiratory quinone analysis. This analysis was performed according to the method described by Komagata and Suzuki (1988). The crude quinone compounds were purified by thin-layer chromatography (TLC) using petroleum benzene:diethyl ether (9:1; v/v) on a silica gel 60 F254 plate (20 × 20 mm, Merck, NJ, United States). The quinones were detected by reverse-phase HPLC with a mixture of methanol:isopropyl ether (3:1; v/v) as the mobile phase (flow rate of 0.7 mL min⁻¹) by monitoring a UV detector at 270 nm.

Genomic DNA was extracted from AGMB02718^T using the phenol:chloroform:isoamyl alcohol method as described by Wilson et al. (1990). Whole-genome sequencing of the strain was performed using the PacBio Sequel System and Illumina NovaSeq technology, and filtered reads were assembled using Flye (ver. 2.9.2) and SPAdes (ver. 3.13.1) software, respectively. The assembled genome was annotated using the NCBI Prokaryotic Genome Annotation Pipeline (PGAP) (Tatusova et al., 2016). The predicted protein sequences were assigned to functional groups in the Clusters of Orthologous Groups (COG) using eggNOG 5.0. The metabolic pathways of strain AGMB02718^T were investigated using BlastKOALA, which is based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway database (Kanehisa et al., 2016). To analyze the genomic similarity between strain AGMB02718^T and the closely related taxa, the pairwise average nucleotide identity (ANI) values were calculated using ANI calculator¹ and Orthologous Average Nucleotide Identity Tool (OAT) software (Lee et al., 2016), and digital DNA–DNA hybridization (dDDH) was calculated using the Genome-to-Genome Distance Calculator (GGDC) version 3.0 (https://ggdc.dsmz.de/ggdc.php#; Meier-Kolthoff et al., 2013). Additionally, phylogenomic tree was constructed based on the up-to-date bacterial core gene (UBCG) consisting of 92 single-copy core gene sets and pipelines, as described by Na et al. (2018). The available genomes of related taxa were retrieved from draft or complete genome sequences in the NCBI database. To construct the phylogenomic tree, UBCGs were extracted from a whole-genome assembly using Prodigal (Hyatt et al., 2010) and hmmsearch (Eddy, 2011). The extracted 92 core genes were aligned using multiple alignments with Fast Fourier Transform (MAFFT) (Katoh and Standley, 2013). A phylogenomic tree was generated using FastTree (Price et al., 2010) and viewed using MEGA 11. A circular map of the AGMB02718^T genome was constructed using CGView software.²

The RAW264.7 cells were incubated in 24-well plates (3 × 10⁵ cells/well) with Dulbecco’s modified Eagle medium (DMEM, welgene, Gyeongsangbuk-do, Republic of Korea) supplemented with 10% fetal bovine serum and 1% antibiotic antimycotic solution (ABS, welgene, Gyeongsangbuk-do, Republic of Korea) for 24 h. Then, E. coli DH5α and AGMB02718^T were treated at a multiplicity of infection (MOI) of 0.1 (3 × 10⁴ CFU/well), 1 (3 × 10⁵ CFU/well), and 10 (3 × 10⁶ CFU/well) for 6 h. LPS was isolated from 1 × 10⁹ CFU of each bacterium using an LPS extraction kit (iNtRON Biotechnology, Gyeonggi-do, Republic of Korea) following the manufacturer’s instructions. The isolated LPS was subsequently treated at 0, 62.5, 125, and 250 μg/mL for 6 h, Afterward, the supernatant was collected for measurement. Cytokines (IL-6 and TNF-α) were quantified using ELISA (BD, NJ, United States), and the analysis was performed according to the manufacturer’s instructions.

Fifteen male C57BL/6 mice (5-week-old) were purchased from KOSA BIO (Seongnam, Republic of Korea). All mice were raised under specific pathogen-free conditions and housed with five heads per cage, allowing for spontaneous intake of food and water. Animals were kept under a 12 h light/dark cycle (light on 09:00–21:00) at room temperature (23 ± 2°C) and humidity (55 ± 10%). All mouse experiments were approved by the Institutional Animal Care and Use Committee of the Korea Research Institute of Bioscience and Biotechnology (approval number: KRIBB-AEC-22146) and all animals were cared for according to the guidelines for animal experiments of the Korea Research Institute of Bioscience and Biotechnology.

A colitis model was established as described by Chassaing et al. (2014). After 1 week of acclimation, C57BL/6 mice were randomly divided into three groups (n = 5/group): control (PBS only), dextran sulfate sodium (DSS) (DSS and PBS treatment), and AGMB02718 (DSS and bacterial treatment). AGMB02718 group mice were administered M. faecium AGMB02718^T (1 × 10⁸ CFU/100 μL/mouse) once a day for 1 week. Colonic inflammation was induced by the addition of 2.5% (w/v) DSS (MW: 40,000; MP Biomedicals, CA, United States) to drinking water ad libitum for 10 days. DSS was administered 10 days after the start of PBS or bacterial administration. Body weight, stool consistency, and presence of gross rectal blood were evaluated daily. Weight change was calculated using the following formula:

The severity of colitis was calculated daily based on the Disease Activity Index (DAI) score, first described by Cooper et al. (1993). On the 10th day after the initiation of DSS treatment, all mice were sacrificed after collection of serum, feces, and colon tissue. Feces and serum were stored at −80°C until analysis. After measuring the length of the resected colon, it was segmented and stained with hematoxylin and eosin (H&E).

The segmented colon tissue was fixed in 10% (v/v) neutral-buffered formalin (Sigma-Aldrich, MA, United States) and embedded in paraffin. The tissue was then cut into 5 μm thick sections and stained with H&E (BCC Biochemical, WA, United States). For histopathological analysis, at least three slides were randomly selected and graded in a blinded manner using a light microscope (Leica Microsystem, Wetzlar, Germany) with 4× and 20× objective lenses. The histological activity index (HAI) was measured based on the quality and dimensions of inflammatory cell infiltrates (1–5), epithelial changes (1–5), and overall mucosal structure (1–5) scores as previously described (Erben et al., 2014).

Intestinal permeability was determined using fluorescein isothiocyanate-conjugated dextran (FITC-dextran; average MW 4000, Sigma-Aldrich, MA, United States) assays. Mice were fasted overnight and FITC-dextran (400 mg/kg) dissolved in PBS was administered by gavage 4 h before blood collection. The FITC-dextran concentration in the serum was measured at 485/538 nm using a microplate reader (Fluoroskan FL; Thermo Scientific, MA, United States).

Fecal samples were immediately frozen and stored at −80°C freezer after collection. Total bacterial DNA was extracted from fecal samples (approximately 200 mg) using a QIAamp Fast DNA Stool Mini Kit (Qiagen, Hilden, Germany), following the manufacturer’s instructions. The V3–V4 region of the bacterial 16S rRNA gene was amplified using 341F (5′-CCTACGGGNGGCWGCAG-3′) and 805R (5′-GACTACHVGGTATCTAATCC-3′) primers from each sample following standard procedures at an external facility (Macrogen, Seoul, South Korea). Identical sequencing reads were combined through DADA2’s dereplication functionality, and the DADA2 (v1.18.0) sequence–variant inference algorithm was applied to each dataset (Callahan et al., 2016). A taxonomy was assigned to each amplicon sequence variant (ASV) based on BLASTN (v2.9.0+) of the NCBI 16S Microbial DB (Zhang et al., 2000). For multiple alignments between representative sequences of ASVs, MAFFT (ver. 7.475) was used (Katoh and Standley, 2013). Alpha (α) diversity was applied to analyze the complexity of species diversity for a sample through ASVs, Chao1, and Shannon indexes. Beta (β) diversity group analysis was performed using a Jaccard index method, and divergence between the groups was highlighted by principal coordinates analysis (PCoA). α and β diversity were calculated with QIIME2 (ver. 1.9.0) (Caporaso et al., 2010). The feature abundance was normalized using the relative abundance of each sample. Functional predictions based on 16S rRNA gene data were performed using Tax4Fun provided by MicrobiomeAnalyst (Asshauer et al., 2015; Dhariwal et al., 2017).

The complete 16S rRNA gene sequence of strain AGMB02718^T contained 1,461 bp and was submitted to GenBank (accession number: OM971902). Comparative sequence analysis using the 16S rRNA gene sequences determined that strain AGMB02718^T belonged to the genus Mesosutterella and was closely related to Mesosutterella multiformis 4NBBH2^T (96.2%), Sutterella stercoricanis CCUG 47620^T (91.1%), Sutterella megalosphaeroides 6FBBBH3^T (91.0%), Parasutterella secunda YIT 12071^T (90.9%), Sutterella wadsworthensis WAL 7877^T (90.6%), and Sutterella parvirubra YIT 11816^T (90.4%). Based on the threshold of interspecies 16S rRNA gene sequence similarity percentages, which are between 95.0 and 98.7% (Beye et al., 2018), strain AGMB02718^T was considered a novel species in the genus Mesosutterella. The NJ, ML, and MP algorithms were used for phylogenetic tree analysis (Supplementary Figure S1). The phylogenetic tree reconstructed using NJ and ML algorithms revealed that strain AGMB02718^T formed a stable cluster with another strain of the genus Mesosutterella (Figure 1). The genera Sutterella and Parasutterella formed independent monophyletic branches. The phylogenetic results suggested that strain AGMB02718^T represents a novel species of the genus Mesosutterella. Therefore, M. multiformis DSM 106860^T was selected as the reference species for further comparative tests.

Phenotypic characteristics based on the optimal growth medium, morphology, temperature, and salt tolerance were investigated. Strain AGMB02718^T grew well in the TSAB medium. Cells were obligate anaerobe, Gram-negative, coccobacilli-shaped with a mean width of 0.4 μm and mean length of 1.25 μm (Supplementary Figure S2). Growth of strain AGMB02718^T occurred in the range of 25–45°C (optimal 37°C). In addition, the isolate was able to grow in up to 1.0% (w/v) NaCl (optimal, 0.5%). The strain AGMB02718^T was negative for catalase and oxidase activities. The results are presented in Supplementary Table S1. When compared to the reference strain in API tests, including API 20A, API ZYM, and Rapid ID 32A, strain AGMB02718^T showed differences in salicin, D-raffinose, and D-sorbitol activities in API 20A, and differences in esterase (C4), cystine arylamidase, naphthol-AS-BI-phosphohydrolase, N-acetyl-b-glucosaminidase, glutamyl glutamic acid arylamidase, and serine arylamidase activities in API ZYM and Rapid ID32A. The different characteristics of strain AGMB02718^T compared with those of the reference strain are presented in Table 1. All API test results for AGMB02718^T and the reference strain are listed in Supplementary Table S2.

The cellular fatty acid profiles (>1.0%) of strain AGMB02718^T and the reference strain are shown in Table 2. The major fatty acids (>10.0%) of strain AGMB02718^T were C_18:1 ω9c (24.0%), C_16:0 (21.4%), and C_16:0 DMA (11.6%), whereas the major fatty acids of the reference strain M. multiformis DSM 106860^T were C_18:1 ω9c (33.0%), C_16:0 (20.7%), and C_18:0 (14.6%). Differences in major and minor fatty acids differentiated strain AGMB02718^T from the closely related strain of the genus Mesosutterella (Table 2). The respiratory quinones of the strain AGMB02718^T were menaquinone 6 (MK-6) and methylmenaquinone 6 (MMK-6), similar to those of the reference strain (Supplementary Table S1).

Whole-genome sequencing was performed to analyze the genomic characteristics of strain AGMB02718^T. The assembled genome of the strain AGMB02718^T was a single circular chromosome with a coverage of 250.0×. The whole-genome sequence of the novel strain was deposited in NCBI under the accession number JAKZJU020000000 and visualized using the CGView tool (Supplementary Figure S3). Based on the genomic analysis of strain AGMB02718^T, the genomic GC content was 62.2%, the length of the genome was 2,606,253 bp (N50 value of 2,332,978 bp), which contained 2,278 protein-coding genes and 71 RNA genes, including five 5S rRNA, five 16S rRNA, five 23S rRNA, three non-coding RNA (ncRNA), and 53 tRNA genes (Supplementary Table S3). In addition, 1,801 protein sequences were functionally assigned to categories based on clusters of orthologous group (COG) assignments, revealing that the largest functional category of strain AGMB02718^T accounted for function unknown (14.7%), followed by energy production and conversion (10.6%), and amino acid transport and metabolism (9.6%) (Supplementary Table S4). The ANI values between strain AGMB02718^T and closely related species including M. multiformis 4NBBH2^T (GCA_003402575.1), S. megalosphaeroide 6FBBBH3^T (GCF_003609995.1), S. wadsworthensis DSM 14016^T (GCA_003315195.1), and S. parvirubra YIT 11816^T (GCA_000250875.1) were 72.1, 69.1, 67.1, and 68.8, respectively (S. stercoricanis CCUG 47620^T, P. secunda YIT 12071^T, and P. excrementihominis YIT 11859^T were excluded owing to a lack of whole-genome sequencing information). OrthoANI values based on the entire genome were 72.1–67.1% for the most closely related strains (Supplementary Figure S4). The dDDH relatedness values between strain AGMB02718^T and the reference strains were 20.9, 21.5, 20.8, and 19.6% (using Formula 2), respectively. The results for ANI, orthoANI, and dDDH between strain AGMB02718^T and closely related strains are listed in Supplementary Table S5. Moreover, a whole-genome-based phylogenomic tree constructed using UBCG (version 3.0) supported that the strain AGMB02718^T formed a phylogenetic lineage belonging to the genus Mesosutterella, consistent with the 16S rRNA-based phylogenetic tree (Figure 2). Because strain AGMB02718^T exhibited significantly low values for the dDDH (<70%) and ANI cutoff (95–96%) proposed for bacterial species delineation (Chun et al., 2018), these genetic analyses collectively classified strain AGMB02718^T as a novel strain within the Mesosutterella genus.

Differences in LPS structure play an important role in determining its toxicity. Owing to the difference in the structure of lipid A, a component of LPS, Sutterella spp. and E. coli have different pro-inflammatory properties (Hiippala et al., 2016). Lipid A from E. coli is hexa-acylated to activate the TLR4–Mal–MyD88 pathway and is highly toxic, whereas tetra- or penta-acylated lipid A, generated from Sutterella spp., signals through TLR4–TRAM-TRIF and is 100-fold less toxic (Park et al., 2009; Herath et al., 2011; Han et al., 2013). Additionally, E. coli and related Proteobacteria contain all the nine genes required for complete KLA biosynthesis (Opiyo et al., 2010). Sutterella spp.-derived lipid A genomes lack lpxL and lpxM genes needed for the hexa-acylation of lipids, resulting in penta-acylated or monophosphorylated forms of lipid A (Park et al., 2009; Herath et al., 2011; Han et al., 2013). Because Mesosutterella faecium AGMB02718^T belongs to the same family as Sutterellaceae, we expected that it would have a lipid A biosynthetic pathway similar to that of the genus Sutterella. Therefore, the sequenced genome of the strain AGMB02718^T was examined for LPS biosynthesis. Whole-genome mining of the strain AGMB02718^T was performed using BlastKOALA, and the following sections were predicted from the genome sequences. The genes associated with LPS biosynthesis, including lpxA (UDP-N-acetyl glucosamine acyltransferase [EC:2.3.1.129]), lpxC (UDP-3-O-acyl-N-acetylglucosamine deacetylase [EC:3.5.1.108]), lpxD (UDP-3-O-acyl-glucosamine N-acyltransferase [EC:2.3.1.191]), lpxH (UDP-2,3-diacylglucosamine hydrolase [EC:3.6.1.54]), lpxB (lipid-A-disaccharide synthase [EC:2.4.1.182]), lpxK (tetraacyldisaccharide 4′-kinase [EC:2.7.1.130]), waaA (lipid IV_A 3-deoxy-D-manno-octulosonic acid transferase [EC 2.4.99.12] or (Kdo)-lipid IV_A 3-deoxy-D-manno-octulosonic acid transferase [EC 2.4.99.13]), and lpxL (Kdo₂-lipid IV_A lauroyltransferase/acyltransferase [EC 2.3.1.241]) were detected; however, the lpxJ and lpxM genes, responsible for synthesizing KLA from lauroyl or stearoyl-Kdo₂-lipid IV_A, were absent (Supplementary Table S6). These results suggest that strain AGMB02718^T is unable to biosynthesize hexa-acylated lipid A (Figure 3).

The administration of mouse macrophages with E. coli (hexa-acylated lipid A) induced higher expression of inflammatory cytokines than Bordetella pertussis (penta-acylated lipid A), owing to the different forms of lipid A (Ernst et al., 2021). From the results of the whole-genome analysis, it can be expected that the strain AGMB02718^T produces LPS with low toxicity. To evaluate whether the strain AGMB02718^T has low toxicity, its inflammatory stimulatory activity was first examined using live bacteria. The result showed that the E. coli-treated group significantly induced pro-inflammatory cytokines (IL-6 and TNF-α) at all concentrations of MOIs, whereas the strain AGMB02718^T-treated group did not stimulate macrophages at a similar level as the non-treated group at all MOIs (Figure 4A). In addition, the inflammatory stimulation activity was examined using LPS isolated from the bacteria, and it was determined that the LPS of strain AGMB02718^T was less toxic than that of E. coli (Figure 4B).

A non-excessive inflammatory response by intestinal mucosal microbes is important to induce an optimal host immune response (Pandiyan et al., 2019). This is related to the characteristics of the family Sutterellaceae, including the genera Sutterella, Mesosutterella, and Parasutterella, which are commensal bacteria that adhere to the intestine (Hiippala et al., 2016). Tetra- or penta-acylated lipid A plays an important role in the host immune response by not exceeding the inflammatory response threshold owing to its attenuated toxicity (Di Lorenzo et al., 2020). This claim is supported by earlier findings showing that mice lacking immune regulation develop more severe DSS-induced colitis (Keubler et al., 2015). The results of the LPS toxicity test confirmed that AGMB02718^T-derived LPS treatment induced low levels of inflammation in vitro. Taken together, these studies suggest that the strain AGMB02718^T possessing all these characteristics may have ideal effects against IBD. To this end, we established a mouse model of DSS-induced colitis to test the in vivo activity of the AGMB02718^T strain. After pretreatment with bacteria or PBS, IBD was induced using DSS for 10 days (Figure 5A). Although the body weights of the mice in the DSS group were consistently reduced from day 2, the mice in the AGMB02718 group showed a significant improvement in weight loss, although there was an apparent decrease compared to the control group (non-DSS-induced). At the end of the experiment, the weight in the DSS, AGMB02718, and control groups was decreased by −18.21 ± 1.85%, −7.58 ± 2.61%, and increased by +5.03 ± 2.60%, respectively (Figure 5B). The DSS group had bloody stools with diarrhea, whereas mild diarrhea was observed in the bacterial treatment group; however, bloody stools were prevented (Figure 5C). Treatment with strain AGMB02718^T significantly reduced colonic architecture disruption compared with DSS treatment (Figure 5D). Consistent with these results, the histopathology scores of the AGMB02718 group were significantly lower than those of the DSS-treated group (7.40 ± 1.50 vs. 10.60 ± 1.85) (Figure 5E). A common feature of the DSS-induced colitis model is a decrease in colon length and an increase in the DAI. Although the DSS treatment decreased the colon length of the mice (4.15 ± 0.31 cm) compared to the control group (5.86 ± 0.52 cm), the strain AGMB02718^T treatment reversed this effect (5.14 ± 0.66 cm) (Figure 5F). The DAI values were evaluated based on the physiological signs (weight loss, diarrhea, and occult/gross bleeding) induced by 2.5% DSS treatment for 10 days, showing that increased DAI scores in the DSS group (9.00 ± 1.00) were markedly suppressed in the AGMB02718 group (5.25 ± 1.48) (Figure 5G). In addition, intestinal permeability was investigated using FITC-dextran analysis. As a result, the permeability was decreased in the mice of the AGMB02718 group (5.55 ± 1.48 μg/mL) compared to the DSS group (19.18 ± 7.29 μg/mL), suggesting that the intestinal barrier function was improved after treatment with strain AGMB02718^T (Figure 5H). Collectively, it was confirmed that administration of M. faecium AGMB02718^T prevented body weight and colon length reduction and protected and maintained overall intestinal integrity, suggesting that strain AGMB02718^T can mitigate DSS-induced colitis symptoms. However, there are still several limitations to conclusively stating that this preventive effect is entirely due to low-toxicity LPS. To overcome this limitation, further studies are needed, such as the direct administration of LPS derived from M. faecium AGMB02718^T at an in vivo level, investigation of various immune cells including macrophages in the colon, and investigation of inflammation levels in the colon. Nevertheless, there still exists the possibility of alternative mechanisms for IBD prevention through the administration of M. faecium AGMB02718^T.

To further investigate the protective effect of strain AGMB02718^T administration in a DSS-induced mouse model, we evaluated changes in the gut microbiota in mouse feces. Fecal samples were collected from each group (n = 5) after DSS treatment (day 10). A total of 452,589 clean reads (control = 142,265; DSS = 160,271; and AGMB02718 = 150,053) were obtained, and 3,071 identified ASVs (control = 1,064; DSS = 935; and AGMB02718 = 1,072) were clustered. For microbiota diversity analysis, alpha-diversity between groups was compared and analyzed using ASVs, Chao1, and Shannon. Alpha diversity is decreased in patients with CD or UC (Sokol and Seksik, 2010). The fecal microbiota of mice administered only DSS (DSS group) showed reduced alpha-diversity compared to that of the control group, as previously reported; however, interestingly, the reduced alpha-diversity was slightly recovered in the group administered the strain AGMB02718^T (Figure 6A). Next, a PCoA analysis based on the Jaccard index was performed to compare beta diversity in the gut microbiota. As shown in Figure 6B, ANOSIM analysis indicated that the Control group was separated from the other two groups, and the AGMB02718 group and DSS group were marginally changed (R-value = 0.60889, p value = Not significant).

It has been known that the relative abundance of bacteria taxa was altered in individuals with IBD and in DSS-induced mouse (Zuo and Ng, 2018). The abundance of taxa at the phylum, class, family, and genus levels differed between groups (Figure 6C; Supplementary Figure S5). Specifically, at the phylum level, the abundance of Proteobacteria increased in the DSS group compared to the control group as in a previous study, whereas this dysbiosis was reversed by AGMB02718^T administration (Figure 6C). In particular, the abundance of Gammaproteobacteria increased in the DSS group (22.32%) compared to the control group (0.06%), but the abundance of these bacteria reduced after AGMB02718^T treatment (5.36%) (Figure 6D). To further support the effect of strain AGMB02718^T on maintaining intestinal microbial homeostasis, we continued to explore the specific bacterial genera showing stable changes during IBD, in which the abundances of some bacterial genera including Duncaniella, Muribaculum, and Phocea decrease, but the abundances of Enterococcus, Ruminiclostridium, and Streptococcus increase compared with those of a healthy model (Teitelbaum and Triantafyllopoulou, 2006; Bian et al., 2019; Dobranowski et al., 2019; Zhou et al., 2020; Forster et al., 2022; Pisani et al., 2022; Thomas et al., 2022). This imbalance in the gut microbiota induced by IBD occurred in DSS-induced mice; however, AGMB02718^T administration alleviated DSS-induced dysbiosis (Figure 6E). Additionally, functional profiling of the gut microbial communities in each group was predicted based on 16S rRNA sequencing data using tax4fun analysis. Predictions of LPS synthesis genes for each group were conducted through KEGG pathway analysis. As a result, the gene lpxJ, associated with toxicity in LPS, was significantly enriched in the DSS group compared to the other two groups. Conversely, excluding lpxD among the non-toxicity-related genes, there was almost no difference in lpxA and lpxB between the three groups (Figure 6F). Meanwhile, according to reports, gut dysbiosis can lead to gut leakage, damaging the epithelial barrier and allowing gut contents to come into contact with the host, thus inducing an inflammatory response (Walker and Lawley, 2013). These research findings underscore the importance of maintaining gut integrity in suppressing additional inflammatory reactions. Furthermore, based on existing reports, due to the characteristic of the family Sutterellaceae adhering to the mucosa, more strains of the Sutterella genus are found in mucosa covered with mucus than in those without (Tajima et al., 2022). In summary, the destruction of the gut structure induces the detachment of Sutterella from the gut mucosa and triggers colonic inflammation. These facts provide a comprehensive explanation for our findings of increased colon permeability (Figure 5H) and the presence of Sutterellaceae in the feces of the DSS group (Supplementary Figure S6). Because we did not investigate the gut mucosal microbiota, we cannot definitively claim whether M. faecium AGMB02718^T has colonized in the gut. However, considering the significant decrease in gut permeability in the AGMB02718 group, and the almost non-detection of Sutterellaceae in fecal metagenome analysis, we can indirectly infer the possibility of M. faecium AGMB02718^T colonized in the gut.

Taken together, the gut microbiota analysis suggests that the administration of strain AGMB02718^T may prevent DSS-induced colitis by regulating the gut bacterial community.

In this study, strain AGMB02718^T was isolated from swine feces and identified as a novel member of the genus Mesosutterella based on phenotypic, physiological, phylogenetic, and biochemical analyses. Interestingly, whole-genome sequencing analysis revealed that the genome of the isolate does not include genes responsible for the high toxicity of LPS, suggesting that strain AGMB02718^T may produce low-toxicity LPS and consequently help maintain host immune homeostasis, as previously mentioned. Thus, we investigated this hypothesis using a DSS-induced colitis and found that strain AGMB02718^T can protect the colitis induced by DSS. In addition, the isolate contributed to the maintenance of host immune balance and intestinal microbial community homeostasis. This study identified a new member of the genus Mesosutterella that may be used as a next-generation probiotic to control IBD.

Mesosutterella faecium sp. nov. (fae’ci.um. L. fem. Gen. pl. n. faecium, derived from feces).

Colonies grown on TSAB plate are white, circular (0.5–1 mm in diameter), smooth, and convex after a 72 h culture at 37°C. Cells (0.4 μm in width and 1.0–1.5 μm in length) are coccobacilli-shaped, strictly anaerobic, Gram stain-negative, non-motile, flagella-lacking, catalase- and oxidase-negative. Growth occurs at 25–45°C (optimal 37°C) and with 0.5–1% NaCl (optimal 0.5%). Strain AGMB02718^T is positive for the following enzyme activities in the API ZYM and Rapid ID 32A strips: mannose fermentation, raffinose fermentation, nitrate reduction, alkaline phosphatase, arginine arylamidase, proline arylamidase, leucyl glycine arylamidase, phenylalanine arylamidase, leucine arylamidase, pyroglutamic acid arylamidase, tyrosine arylamidase, alanine arylamidase, glycine arylamidase, histidine arylamidase, glutamyl glutamic acid arylamidase, and serine arylamidase. However, it is negative for urease, arginine dihydrolase, α-galactosidase, β-galactosidase, β-galactosidase 6-phosphate, α-glucosidase, β-glucosidase, α-arabinosidase, β-glucuronidase, β-N-acetyl-glucosaminidase, glutamic acid decarboxylase, α-fucosidase, and indole production. In the API 20A strips, all tests are negative. The major fatty acids detected in AGMB02718^T are C_18:1ω9c, C_16:0, and C_16:0 DMA. Respiratory quinones are MK-6 and MMK-6.

GenBank accession numbers for the 16S rRNA gene and whole-genome sequences of strain AGBM02718^T are OM971902 and JAKZJU020000000, respectively. The strain was deposited in the Korea Collection for Type Culture (KCTC 25541^T) and Guangdong Microbial Culture Collection Center (GDMCC 1.2717^T).