The starting point was ecological. Traditional Thai fermented meat products such as naem, sai krok isan, and mum are produced through spontaneous or traditional fermentation in which meat, salt, spices, and indigenous microbial communities impose acidification, nutrient limitation, salt stress, and microbial competition on the bacteria present. These selective pressures favor well-adapted lactic acid bacteria. While naem and sai krok isan have previously been screened for traits such as bile salt hydrolase activity, acid and bile tolerance, and adhesion to Caco-2 cells, investigations of mum have focused mainly on fermentation population dynamics and species identification. Mum wua therefore represented an underexplored niche for asking whether food-adapted bacteria combine desirable phenotypes with reliable genome-based identity and interpretable metabolic signatures.

Taxonomy proved to be the study’s first technical challenge. From 45 presumptive lactic acid bacteria isolates, MUM8-1A was selected because it showed comparatively high cholesterol assimilation during preliminary screening. A Gram-positive, rod-shaped, catalase-negative strain that grew at 6 and 8 percent NaCl, at pH 3 and 9, and at 15 and 45 degrees Celsius, it initially looked like a member of the Lactiplantibacillus plantarum group. But 16S rRNA gene analysis was insufficient for definitive species delineation: the gene showed 99.93 percent similarity to Lactiplantibacillus pentosus DSM 20314 and 99.87 percent to L. plantarum ATCC 14917, values too close to separate these conserved taxa. The team sequenced the genome using a hybrid approach combining Oxford Nanopore long reads with Illumina short reads, assembled with Unicycler at approximately 563-fold coverage, and applied average nucleotide identity, digital DNA–DNA hybridization, and phylogenomics. With an ANI of 98.65 percent and a dDDH of 92.50 percent against L. plantarum ATCC 14917, but only 79.62 percent ANI against L. pentosus, the assignment as Lactiplantibacillus plantarum was resolved beyond 16S-based ambiguity.

The finished genome assembly, deposited under accession JCANWR000000000, was 3,286,841 base pairs long with a G+C content of 44.4 percent, an N50 of 1,739,192 base pairs, and 3,177 coding sequences identified by RAST annotation. CheckM estimated 99.02 percent completeness with 1.82 percent contamination. Probiogenomic profiling then connected genotype to the phenotypes measured in the laboratory. Genes encoding F1F0-ATPase components, acid stress responses, bile adaptation, and membrane protection provided molecular context for the strain’s gastrointestinal survival, while putative adhesins, mucus-binding proteins, LPXTG motif-containing surface proteins, sortase-dependent proteins, and moonlighting proteins such as GroEL and elongation factor Tu were consistent with its strong epithelial adhesion. The quorum-sensing gene luxS, linked in previous work to stress adaptation and host persistence, was also present. The authors emphasize that the contribution of individual genes to each phenotype was not directly validated.

The phenotypic results themselves were striking. Following the INFOGEST static digestion framework, MUM8-1A was exposed to simulated gastric fluid containing pepsin at 2,000 U/mL at pH 3.0 for 3 hours, then to simulated intestinal fluid with 10 mM bile and pancreatin equivalent to 100 U/mL trypsin activity for 5 hours. Its viable count fell from 9.12 to 8.48 log CFU/mL during the gastric phase and to 6.82 log CFU/mL after the intestinal phase, compared with 5.94 log CFU/mL for LGG under the same conditions. In the Caco-2 adhesion assay, MUM8-1A attached at 76.84 percent, significantly higher than LGG’s 68.42 percent. The strain also tested positive for bile salt hydrolase activity on taurodeoxycholic acid medium, and a choloylglycine hydrolase gene annotated as bsh was found in its genome, consistent with the phenotype. It removed 89.55 percent of cholesterol from supplemented broth after 24 hours, slightly exceeding LGG’s 85.04 percent.

Antimicrobial susceptibility testing produced the study’s most nuanced safety discussion. Using the ISO 10932:2010 broth microdilution method with EFSA microbiological cut-off values, MUM8-1A was susceptible to chloramphenicol, clindamycin, gentamicin, kanamycin, and tetracycline, but its minimum inhibitory concentrations for ampicillin and erythromycin, 4 and 2 mg/L respectively, exceeded the corresponding cut-offs of 2 and 1 mg/L. Genome screening complicated the interpretation: ResFinder detected no acquired antimicrobial resistance genes, and CARD found no perfect hits, only one strict hit corresponding to vanY, a D,D-carboxypeptidase involved in peptidoglycan metabolism that also occurs in probiotic L. plantarum genomes. KEGG annotation identified homologs associated with antimicrobial transport, cell-envelope responses, and multidrug efflux that may provide possible genomic context for the elevated MICs. The authors classify this as phenotype–genotype discordance rather than a confirmed intrinsic resistance mechanism, noting that similar patterns have been reported in other L. plantarum strains, including L. plantarum PMO 08, which showed an identical ampicillin MIC despite the absence of resistance genes detected by strict database searches.

Beyond the digestive and cholesterol-associated traits, MUM8-1A showed immunomodulatory and antioxidant activity in vitro. In PMA-differentiated THP-1 macrophage-like cells, both cell-free supernatant and heat-inactivated cells significantly reduced LPS-induced TNF-α production, with LGG showing a comparable response. Several IL-6 measurements fell below the ELISA quantifiable range and were reported as not detected, so TNF-α attenuation was the principal quantitatively supported cytokine response. In radical-scavenging assays, the cell-free supernatant reached 99.50 percent in the DPPH test, exceeding both LGG at 74.69 percent and 100 μg/mL L-ascorbic acid at 40.09 percent, though the authors caution that a matched sterile medium control was not included in the DPPH assay, so medium-derived contributions cannot be excluded. In the ABTS assay, the strain reached 87.53 percent, close to LGG’s 91.18 percent. The difference between the two assays reflects their distinct chemistries rather than contradiction.

Carbohydrate-active enzyme profiling revealed the genetic basis for metabolic flexibility. The conservative HMMER-based dbCAN approach identified 98 putative CAZyme genes, including 51 glycoside hydrolases, 39 glycosyltransferases, 5 carbohydrate esterases, 3 carbohydrate-binding modules, and 3 auxiliary activity enzymes, while broader DIAMOND-based searches found 224. The predominance of glycoside hydrolases suggests capacity to hydrolyze diverse oligo- and polysaccharide substrates, some of which could act as prebiotic compounds in the gut, while glycosyltransferases participate in biosynthesis of cell wall polysaccharides, exopolysaccharides, and capsular structures relevant to stress tolerance and adhesion.

Untargeted LC–MS/MS metabolomics added the final functional layer. Of 36,431 metabolite features in the positive ionization dataset, 23,113 were detected in at least one condition, with 10,697 unique to MUM8-1A-fermented broth; the negative dataset yielded 7,371 features with 1,749 unique to the fermented condition. Putatively annotated features included gamma-aminobutyric acid, indole and indole-3-lactic acid, pyroglutamate, the vitamins biotin, pantothenic acid, pyridoxine, and nicotinamide, nucleotide-turnover compounds such as adenosine and 5′-methylthioadenosine, and lipid-associated signals including palmitoleic and linoleic acid. Because analyte-specific confirmation with authentic standards was not documented, all assignments were treated as putative annotations rather than definitive identifications, and the authors interpret these metabolites as hypothesis-generating context linking extracellular remodeling to the observed antioxidant and immunomodulatory phenotypes, not as proven mediators.

The safety assessment closed the loop. PathogenFinder predicted MUM8-1A as non-human pathogenic with a probability of 0.199, comparable to the probiotic reference strains L. plantarum 299v and LGG. Major virulence determinants were absent, and although a hemolysin III-associated gene was identified, the strain showed no detectable hemolytic activity on sheep blood agar. No complete biogenic amine biosynthesis pathways for histamine, tyramine, cadaverine, putrescine, spermine, or tryptamine were found, no complete antimicrobial drug biosynthesis pathways or gene clusters were reconstructed, and no acquired resistance or virulence determinants were detected within three prophage regions or two putative plasmid-like sequences, neither of which was predicted to be self-transmissible. Two CRISPR arrays at evidence level 2 were found without cas genes and interpreted cautiously as possible remnants. The authors conclude that MUM8-1A is a metabolically versatile and functionally promising probiotic candidate from traditional Thai fermented food, while insisting that confirmation of host-beneficial effects would require appropriate in vivo or clinical evidence beyond the present study.

Subject of Research: Multi-Omics Characterization of Lactiplantibacillus plantarum MUM8-1A from Thai Fermented Beef Sausage: Probiotic Traits and Safety

Article Title: Multi-Omics Characterization of Lactiplantibacillus plantarum MUM8-1A from Thai Fermented Beef Sausage: Probiotic Traits and Safety

Article References: Kingkaew, E., Ratthanachot, N., Chaimongkol, A., Buahom, T., Maneeruttanarungroj, C., Mhuantong, W., Tansawat, R., Dejtisakdi, W., Woraprayote, W., Niwasabutra, K., Sitdhipol, J., Chaiyawan, N., Chaipanya, C., Phuengjayaem, S., Surachat, K., Ochaikul, D., Panngeun, T., & Tanasupawat, S. (2026). Multi-Omics Characterization of Lactiplantibacillus plantarum MUM8-1A from Thai Fermented Beef Sausage: Probiotic Traits and Safety. Journal of Agriculture and Food Research, Article 103389. https://doi.org/10.1016/j.jafr.2026.103389

Image Credits: AI Generated

DOI: 10.1016/j.jafr.2026.103389

Keywords: Multi-Omics, Characterization, Lactiplantibacillus, plantarum, MUM8-1A, Thai, Fermented, Beef, Sausage, Probiotic, Traits, Safety