Systematic evaluation of antimicrobial food preservatives on glucose metabolism and gut microbiota in healthy mice

Antimicrobial preservatives (APs), both synthetic and biogenic, are widely used to inhibit microbial spoilage and extend food shelf life. Given their antibacterial properties and emerging evidence that various food additives can alter gut microbiota and host metabolism, APs may perturb gut ecosystems and metabolic homeostasis after ingestion. Prior studies suggest certain APs can dysregulate gut microbiota in vitro and in animal models, but a systematic evaluation across a diverse set of APs in healthy hosts is lacking. It is also unclear whether biogenic APs are inherently safer than synthetic ones. This study systematically evaluates the chronic effects of 11 commonly used synthetic and biogenic APs on gut microbiota and glucose metabolism in wild-type healthy mice to address these gaps and assess potential metabolic consequences.

Limited prior work indicates APs can alter gut microbiota composition and function. Sulfites inhibited growth of beneficial gut bacteria in vitro. Potassium sorbate changed gut microbiota composition and metabolism in zebrafish, and nitrate exposure induced dysbiosis and metabolic disorder in tadpoles. In rodents, a mixture of sodium benzoate, sodium nitrite, and potassium sorbate induced dysbiosis in both wild-type and human microbiota-associated Nod2−/− mice. Nitrated meat products altered gut microbiota, behavior, and brain gene expression in rats. Beyond APs, other food additives like emulsifiers and artificial sweeteners have been shown to adversely affect gut microbiota and induce glucose intolerance in mice. However, no systematic study had compared a broad panel of APs, nor clarified whether biogenic APs are safer than synthetic counterparts.

Animals and treatments: Wild-type C57BL/6J male mice (8–10 weeks) were randomized into control and AP-treated groups (four mice/cage, three cages/group) after 1 week acclimation. Control received normal diet for 8 weeks; AP groups received the same diet supplemented with one of 11 APs at three times the acceptable daily intake (ADI) for 8 weeks to simulate high-level consumer exposure. Food intake and body weight were monitored; no significant differences were observed after 8 weeks. Glucose metabolism: Oral glucose tolerance tests (OGTTs) were performed after 2 and 8 weeks of treatment (2 g/kg glucose by gavage). Blood glucose was measured at 0, 15, 30, 60, 90, and 120 min (n=6 per group). The 2-h area under the curve was compared. Gut microbiota profiling: After 8 weeks, fecal samples underwent 16S rDNA V4 region sequencing (Illumina). Data were processed with QIIME2 (Deblur for ASVs; taxonomy via Greengenes). Alpha diversity (Shannon, Pielou’s evenness, observed OTUs) and beta diversity (Weighted UniFrac) were computed. Group differences were assessed by PERMANOVA with FDR correction (n=9 per group). Taxonomic composition was compared using ANOVA with Dunnett’s multiple comparisons. Targeted metabolomics: Fecal targeted metabolomics was performed by UPLC-MS/MS following derivatization, with internal standards and QC. Statistics included PCA and OPLS-DA; differential metabolites were selected by VIP>1.0 and p<0.05, visualized in heatmaps, and subjected to KEGG pathway analysis (n=6 per group). Gene expression and hormones: Colonic mRNA expression of Gcg, Pcsk1, and Pcsk2 was quantified by qPCR (2^−ΔΔCt, GAPDH control; n=6). Serum glucagon and insulin were quantified by ELISA on the final day (n=8). Statistics: Two-tailed t-tests for two-group comparisons; one-way ANOVA with Dunnett post hoc test for multiple groups; PERMANOVA for beta diversity with FDR<0.05 considered significant. Software: GraphPad Prism 8.21 and iMAP for metabolomics analyses.

• Glucose intolerance: After 2 weeks, significant hyperglycemic effects were observed for synthetic APs sodium benzoate (p=0.0073), potassium sorbate (p=0.0191), ethylparaben (p=0.0424), sodium nitrate (p=0.0340), sodium propionate (p=0.0024), and the biogenic AP nisin (p=0.0147). After 8 weeks, sodium benzoate (p=0.0179) and sodium propionate (p=0.0027) remained significant, and all biogenic APs—natamycin (p=0.0402), nisin (p=0.0005), lysozyme (p=0.0109), and ε-polylysine (p=0.0070)—significantly increased blood glucose. Potassium sorbate, sodium nitrate, and ethylparaben effects were transient; sodium benzoate and sodium propionate effects were sustained. All biogenic APs showed time-dependent hyperglycemic effects; nisin induced the most pronounced glucose intolerance.
• Gut microbiota alterations: PERMANOVA on Weighted UniFrac distances showed significant shifts vs control for sodium benzoate (p=0.0051), potassium sorbate (p=0.0440), ethylparaben (p=0.0051), lysozyme (p=0.0094), nisin (p=0.0051), and ε-polylysine (p=0.0124), indicating AP-induced dysbiosis irrespective of origin. Nisin produced the most distinct community and increased alpha diversity (Shannon, Pielou’s evenness, observed OTUs).
• Taxonomic changes: Across APs, Actinobacteria decreased. Verrucomicrobia increased in most AP groups, especially with lysozyme (p=0.0030) and ε-polylysine (p=0.0035). Proteobacteria was enriched by sodium benzoate, ethylparaben, sodium nitrate, sodium propionate, natamycin, nisin, and lysozyme. At the genus level, Bifidobacterium decreased with sodium benzoate (p=0.0052), sodium diacetate (p=0.0240), sodium nitrate (p=0.0349), natamycin (p=0.0139), lysozyme (p=0.0002), and ε-polylysine (p=0.0105). Akkermansia increased with lysozyme (p=0.0030) and ε-polylysine (p=0.0035). Coriobacteriaceae decreased with all APs (notably sodium benzoate p=0.0086 and potassium sorbate p=0.0396). Helicobacter was enriched by most APs. Nisin specifically decreased Bifidobacterium (p<0.0001), Coriobacteriaceae (p=0.0019), and Allobaculum (p<0.0001), and enriched Oscillospira (p<0.0001), S24-7 (p=0.0197), Clostridiales (p=0.0003), Ruminococcaceae (p=0.0036), and Lactobacillus (p=0.0072).
• Metabolomics: Nisin significantly altered fecal metabolites (clear separation by PCA and OPLS-DA; R2Y=0.858, Q2Y=0.405). Fifty-eight metabolites changed: taurocholic acid was upregulated; 57 metabolites were downregulated, including multiple amino acids, carbohydrates, fatty and organic acids, benzenoids, benzoic acids, bile acids, imidazoles, indoles, phenols, pyridines, and short-chain fatty acids. Amino acids showed the most marked decreases. KEGG analysis highlighted altered pathways: phenylalanine/tyrosine/tryptophan biosynthesis and metabolism, D-glutamine and D-glutamate metabolism, branched-chain amino acid biosynthesis, alanine/aspartate/glutamate metabolism, and aminoacyl-tRNA biosynthesis.
• Microbiota–metabolite correlations: Allobaculum, Bifidobacterium, and an unknown Coriobacteriaceae genus positively correlated with downregulated metabolites; Oscillospira, unknown Clostridiales genera, and an unknown Ruminococcaceae genus negatively correlated. Taurocholic acid correlated negatively with Bifidobacterium and positively with Oscillospira, unknown Clostridiales, and unknown Ruminococcaceae.
• GLP-1 axis and hormones: In colon, Pcsk1 expression decreased with nisin (p=0.0149), suggesting impaired proglucagon processing to GLP-1. Serum glucagon increased (p=0.0123) and insulin showed an increasing trend (p=0.0919), indicating perturbed glucoregulatory hormone secretion potentially contributing to glucose intolerance.

Chronic exposure to commonly used APs induced glucose intolerance in healthy mice, paralleling prior findings with other food additives like artificial sweeteners. The dysbiosis observed by 16S rDNA profiling indicates that APs, regardless of synthetic or biogenic origin, can shift gut community structure and key taxa linked to metabolic health. Biogenic APs were not uniformly safer; notably, nisin elicited the strongest and most consistent effects on glucose tolerance and microbiota, increasing diversity while depleting beneficial taxa (e.g., Bifidobacterium, Coriobacteriaceae) and enriching taxa associated with dysbiosis. With nisin, marked reductions in fecal amino acids and altered metabolic pathways were detected, aligning with the recognized role of gut microbiota in shaping intestinal amino acid pools. Given amino acid–mediated stimulation of GLP-1 release from L-cells, reduced amino acids and lowered Pcsk1 expression suggest diminished GLP-1 production, with concomitant increases in glucagon and a trend toward higher insulin, perturbing glucoregulatory balance and promoting hyperglycemia. While these data support a microbiota–metabolite–hormone axis in AP-induced metabolic disturbance, causality requires validation (e.g., fecal microbiota transplantation). The dosing at 3× ADI modeled high-exposure consumers and revealed heterogeneity among APs in both microbiota and metabolic impacts, emphasizing the need for nuanced safety evaluation beyond origin (synthetic vs biogenic).

This systematic evaluation demonstrates that several antimicrobial food preservatives, both synthetic and biogenic, can induce glucose intolerance and perturb gut microbiota in healthy mice. Biogenic preservatives are not inherently safer; nisin produced the most pronounced alterations in microbiota composition, fecal metabolites (notably widespread amino acid reductions), and GLP-1–related hormonal regulation, potentially mediating impaired glucose homeostasis. Future work should include causal testing of microbiota involvement (e.g., fecal microbiota transplantation), mechanistic dissection of the GLP-1 pathway and insulin resistance, dose–response and mixture effects, and translational studies to assess human relevance.

Causality between AP-induced microbiota changes and glucose intolerance was not established; the authors note that fecal microbiota transplantation studies are needed to validate the microbiota’s role. Mechanistic links to insulin resistance were not assessed and require further exploration. The study modeled high-level consumer exposure using 3× ADI dosing, which may not reflect typical intake across populations.