Skip to main content
NCBI home page
NPJ Science of Food logoLink to NPJ Science of Food
. 2026 Jul 20;10:230. doi: 10.1038/s41538-026-00932-2

Lactobacillus fermentum 2-14 mitigates the cytotoxicity induced by methylglyoxal by activating the AMPK-autophagy signaling axis

Junjun Gao 1, Yuqin Chen 1, Junfeng Xiao 1, Siyu Guo 1, Manli Li 1, Yapeng Guo 1, Lei Wang 2, Mingye Peng 1, Yang Chen 1,✉, Mengzhou Zhou 1,✉
PMCID: PMC13401601  PMID: 42476998

Abstract

Methylglyoxal (MGO), a reactive dicarbonyl formed during the food processing, induces oxidative stress, inflammation and apoptosis. However, there are little effective methods to reduce MGO-induced cytotoxicity. Here, we screened 35 lactic acid bacteria and identified Lactobacillus fermentum 2-14 as the most effective strain in restoring Caco-2 cell viability and reducing LDH cytotoxicity under MGO challenge. Mechanistically, L. fermentum 2-14 attenuated MGO-induced ROS accumulation, apoptosis and inflammatory responses, and promoted autophagy, as indicated by increased LC3 puncta and autolysosome formation using an RFP–GFP–LC3 reporter. Using integrative transcriptomics and metabolomics, we further suggested that L. fermentum 2-14 activates the AMPK pathway by increasing the level of pyruvate in Caco-2 cells. Supplementing with pyruvate partially mimicked the protective effect in an AMPK- and autophagy-dependent manner. Collectively, our findings indicate that L. fermentum 2-14 mitigates MGO cytotoxicity via a pyruvate–AMPK–autophagy axis, supporting the development of probiotic-based strategies to counter food-derived dicarbonyl stress.

Subject terms: Biochemistry, Cell biology, Microbiology

Introduction

Methylglyoxal (MGO) is an intermediate in the Maillard reaction and is generated through a pathway closely related to various food processing techniques1. During food storage and heat treatment, reducing sugars and amino compounds condense with carbonylamine to form a Schiff base, which is then converted into 1-deoxyglucuronide (1-DG) and 3-deoxyglucuronide (3-DG) via the Amadori rearrangement reaction and subsequently degraded into MGO via Strecker degradation2,3. Furthermore, the caramelization and oxidative cleavage of monosaccharides (such as those found in honey), the lipid peroxidation of unsaturated fatty acids, and the activation of the glycolytic bypass in microbial metabolic activities (particularly in fermented foods) all significantly contribute to the production of MGO4,5. Metabolic kinetic studies have shown that the total daily dietary intake of MGO by an individual is approximately 0.04–0.3 mmol. More than 10% of this amount accumulates in tissues such as vascular endothelium, hepatic parenchymal cells, and renal tubular epithelial cells in the form of advanced glycation end products (AGEs)6. Previous studies have shown that MGO triggers Caspase-dependent apoptosis, induces endoplasmic reticulum (ER) stress by activating the unfolded protein response (UPR), increases the production of Reactive oxygen species (ROS) by activating NADPH oxidase, and promotes inflammatory effects7–12. Consequently, it is imperative to develop effective methods to mitigate MGO-induced cellular toxicity.

Over the past few decades, various substances have been found to reduce MGO-related cellular toxicity, including plant extracts13–32, amino acids and peptides33–35, persulfides and polysulfides36,37, anti-diabetic drugs7,38, AGEs inhibitors39, and other small molecule inhibitors40,41. Nevertheless, the industrial production of natural-origin inhibitors is limited by factors such as low extraction efficiency, the presence of organic solvents during extraction, high process costs, and pharmacokinetic defects such as a significant first-pass effect and poor membrane permeability42,43. While synthetic inhibitors offer advantages in terms of bioavailability and stability, their side effects and the risk of environmental pollution during production also restrict their use44. Lactic acid bacteria (LAB), which are the most commonly used probiotic group at present, are a type of probiotic that are widely colonized in the mouth, gastrointestinal tract, and female reproductive tract45. Previous studies have shown that LAB play an important role in regulating digestion and metabolism, maintaining immune system homeostasis, modulating neurological functions, and anti-tumor46–50. Several studies have reported that LAB may ameliorate MGO-induced cytotoxicity51–53. However, the underlying molecular mechanisms have yet to be explored.

Autophagy is a highly conserved intracellular degradation system in eukaryotes that plays a key role in maintaining cellular homeostasis and survival54–57. Treatment with MGO increases the expression of the autophagy markers LC3-II and P62, which partially antagonize MGO-induced apoptosis in renal cells58,59. Therefore, the application of autophagy modulators may offer new insights for interventions in MGO-related diseases. Several studies have demonstrated that LAB regulate autophagy to enhance the clearance of intracellular bacteria, inhibit cell apoptosis, mitigate intestinal infections, alleviate oxidative stress-induced intestinal damage, enhance goblet cell function, promote intestinal mucus secretion, and inhibit cancer cell proliferation60–63. However, it remains to be investigated whether LAB alleviate MGO-induced cytotoxicity by regulating autophagy.

Here, we found that MGO decreases cell viability and increases cytotoxicity in a concentration-dependent manner. Through screening, we discovered that Lactobacillus fermentum 2-14 alleviates the decrease in cell viability and the increase in cytotoxicity induced by MGO. Then, we found that Lactobacillus fermentum (L. fermentum) 2-14 mitigates MGO-induced oxidative stress, inflammation, and apoptosis. Furthermore, we uncovered that L. fermentum 2-14 activates autophagy in Caco-2 cells. Inhibiting autophagy significantly reduced the protective effects of L. fermentum 2-14 against MGO-induced oxidative stress, inflammation, and apoptosis. Utilizing transcriptomics and metabolomics analyses, we found that pyruvic acid derived from L. fermentum 2-14 activates the Adenosine 5‘-monophosphate (AMP)-activated protein kinase (AMPK) pathway associated with autophagy by promoting the production of pyruvic acid. Supplying pyruvic acid alone also inhibited MGO toxicity, but this effect was significantly reduced when the AMPK pathway and autophagy were blocked. Hence, our work establishes a theoretical foundation for the development of new natural MGO inhibitors, offering valuable insights into the role of LAB in human health and the development of dietary supplements.

Results

L. fermentum 2-14 mitigates the decrease in cell viability and the increase in cytotoxicity induced by MGO

To determine the optimal concentration of the MGO treatment, we exposed Caco-2 cells to various concentrations of MGO and assessed the impact on cell viability and cytotoxicity. We found that Caco-2 cell viability decreases with increasing concentration of MGO (Fig. 1A). Additionally, we found that there was no significant change in cytotoxicity compared to the control at lower MGO concentrations (0.4 and 0.8 mM). However, at MGO concentrations of 1.2 mM or greater, cytotoxicity increased significantly with the rise in MGO concentration (Fig. 1B). These results suggest that MGO treatment decreases cell viability and increases cytotoxicity in a concentration-dependent manner. In subsequent experiments, we chose a concentration of 2.4 mM for MGO treatment, which resulted in an approximately 50% decrease in cell viability.

Fig. 1. MGO reduces cell viability while increasing cytotoxicity in a concentration-dependent manner.

Fig. 1

Open in a new tab

A Cell viability of Caco-2 cells under different concentrations of MGO treatment. B LDH cytotoxicity of Caco-2 cells under different concentrations of MGO treatment. Data are presented as mean ± SEM from at least three independent experiments, with six replicate wells per condition in each experiment. n.s., non-significant; *p < 0.05, ****p < 0.0001.

In order to screen for LAB capable of mitigating MGO-induced cytotoxicity, we selected 35 different strains from 6 genera. We co-incubated the LAB with Caco-2 cells for 6 hours, after which the cells were treated with MGO for a further 24 hours. Cell viability and cytotoxicity were then detected. As shown in Fig. 2A, L. fermentum 2-8, 2-10 and 2-14 were found to significantly attenuate the MGO-induced decrease in cell viability. Meanwhile, L. fermentum 2-10 and 2-14 were found to significantly reduce MGO-induced cytotoxicity (Fig. 2B). In conclusion, L. fermentum significantly mitigates the decrease in cell viability and the increase in cytotoxicity induced by MGO. L. fermentum 2-14 was chosen for further research due to its superior protective properties.

Fig. 2. L. fermentum 2-14 significantly mitigates the decrease in cell viability and the increase in cytotoxicity induced by MGO.

Fig. 2

Open in a new tab

A Cell viability of Caco-2 cells under MGO and different LAB strains + MGO treatment. B LDH cytotoxicity of Caco-2 cells under MGO and different LAB strains + MGO treatment. LDH cytotoxicity of Caco-2 cells after MGO treatment with or without different LAB strains. Caco-2 cells were pre-incubated with LAB for 6 hours, followed by MGO treatment for 24 hours. Data are presented as mean ± SEM from at least three independent experiments, with six replicate wells per condition in each experiment. Statistical significance was analyzed using one-way ANOVA followed by an appropriate post hoc test. n.s., non-significant; ****p < 0.0001.

L. fermentum 2-14 mitigates MGO-induced apoptosis, oxidative stress, and inflammation

As mentioned above, MGO treatment leads to apoptosis, the accumulation of ROS, and inflammation64–66. To investigate whether L. fermentum 2-14 attenuates MGO-induced apoptosis, oxidative stress, and inflammation, we performed Terminal Deoxynucleotidyl Transferase-mediated dUTP Nick-End Labeling (TUNEL, an apoptosis marker) and examined the ROS and inflammatory factor protein levels after MGO and L. fermentum 2-14 treatment. Our results showed that L. fermentum 2-14 treatment reduces apoptosis of MGO-treated Caco-2 cells (Fig. 3A, B). As shown in Fig. 3C, L. fermentum 2-14 mitigates MGO-induced increase in ROS. In addition, we observed a significant increase in the pro-inflammatory factor interleukin-6 (IL-6) and a significant decrease in the anti-inflammatory factor interleukin-10 (IL-10) following MGO treatment. Pre-treating Caco-2 cells with L. fermentum 2-14 rescued the MGO-induced increase in IL-6 levels and decrease in IL-10 levels (Fig. 3D, E). Thus, our data demonstrate that L. fermentum 2-14 mitigates MGO-induced apoptosis, oxidative stress, and inflammation.

Fig. 3. Pre-treating Caco-2 cells with L. fermentum 2-14 rescues apoptosis, oxidative stress, and inflammation induced by MGO.

Fig. 3

Open in a new tab

A TUNEL staining level of Caco-2 cells after control, positive control, L. fermentum 2-14, MGO, and L. fermentum 2-14 + MGO treatment. B Quantification of TUNEL staining level after different treatments. C The level of ROS production after MGO and L. fermentum 2-14 + MGO treatment compared with control. The protein level of IL-6 (D) and IL-10 (E) after MGO and L. fermentum 2-14 + MGO treatment compared with control. n.s., non-significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

L. fermentum 2-14 activates autophagy in Caco-2 cells

Next, we investigated the mechanism by which L. fermentum 2-14 mitigates MGO-induced cytotoxicity. Previous studies have shown that autophagy prevents MGO-induced apoptosis59,67. We therefore explored whether L. fermentum 2-14 activates autophagy. To achieve this, we transfected the autophagy-related protein LC3, which was double-labeled with GFP and RFP, into Caco-2 cells and then detected the fluorescent protein level after treatment with L. fermentum 2-14. Interestingly, we found that treatment with L. fermentum 2-14 for 6 h increased increased the intensity of LC3-GFP immunostaining compared with control (Fig. 4A, B). In addition, we examined the transcription and protein levels of the autophagy-related genes using RT-qPCR and ELISA, respectively. Our results showed that the expression of the autophagy-related genes LC3, Beclin-1 and ATG16L1 increased significantly after L. fermentum 2-14 treatment for 6 h (Fig. 4C, D). In contrast, the transcription and protein levels of P62, a negative regulator of autophagy, decreased after L. fermentum 2-14 treatment (Fig. 4C, E). In summary, our data demonstrate that L. fermentum 2-14 activates autophagy.

Fig. 4. L. fermentum 2-14 activates autophagy.

Fig. 4

Open in a new tab

Representative images (A) and quantification (B) of LC3-GFP staining after different treatments. C The relative mRNA expression of autophagy-related gene after control and L. fermentum 2-14 treatment. The protein levels of LC3 (D) and P62 (E) after control and L. fermentum 2-14 treatment. F Representative images of tandem mRFP-GFP-LC3 fluorescence after different treatments. Yellow puncta (GFP+ and RFP+, yellow arrowhead) indicate autophagosomes, and red-only puncta (RFP+ only, red arrowhead) indicate autolysosomes. G The number of autophagosomes and autolysosomes per cell after control, positive control, and L. fermentum 2-14 treatment. n.s., non-significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

The final step of autophagy is the formation of autolysosomes, which requires the fusion of autophagosomes and lysosomes68,69. Since autolysosome environments are acidic, once the autolysosomes form in Caco-2 cells transfected with RFP-GFP-LC3, the GFP is quenched and only the RFP remains70. Therefore, the ratio of red to yellow fluorescence can be used to determine whether autolysosomes have formed. As shown in Fig. 4F, G, treating Caco-2 cells with L. fermentum 2-14 significantly increased the number of autophagosomes and autolysosomes compared with the control group. In conclusion, our results indicate that L. fermentum 2-14 activates autophagy and promotes the formation of autolysosomes.

Autophagy is essential for the alleviating effect mediated by L. fermentum 2-14

Next, we wondered whether autophagy is necessary for L. fermentum 2-14-mediated protection. To this end, we administered chloroquine (Cq), the classic autophagy inhibitor that prevents the fusion of autophagosomes and lysosomes71, alongside L. fermentum 2-14 treatment. Much to our delight, our results showed that inhibiting autophagy reduced cell viability (Fig. 5A) and increased cytotoxicity (Fig. 5B). Additionally, we found that the TUNEL staining levels increased after Cq administration (Fig. 5C, D). We also examined the expression of genes related to apoptosis. We found that the expression of the pro-apoptotic genes, Bad, Bax, Caspase-3 and Caspase-9 increased following MGO treatment. The protein levels of Bad, Bax and Caspase-9 also increased after MGO treatment. Pre-treating Caco-2 cells with L. fermentum 2-14 decreased the transcription levels of Bad, Bax and Caspase-9, as well as the protein levels of Bad, Caspase-3 and Caspase-9 (Fig. 5E, F, H, I and Supplementary Fig. 1). Co-administering Cq with L. fermentum 2-14 decreased the expression of Bad and Caspase-3 (Fig. 5E, F, H and I). On the contrary, the expression of Beclin-2, an anti-apoptosis gene, decreased after MGO treatment. Pre-treating Caco-2 cells with L. fermentum 2-14 increases the expression of Beclin-2. Inhibiting autophagy using Cq leads to a decrease in the expression of Beclin-2 (Fig. 5G, J). Thus, autophagy is necessary to increase cell viability and decrease cytotoxicity and apoptosis mediated by L. fermentum 2-14.

Fig. 5. Inhibition of autophagy leads to decreased cell viability, increased cytotoxicity and apoptosis.

Fig. 5

Open in a new tab

A The cell viability of Caco-2 cells after control, MGO, L. fermentum 2-14 + MGO, L. fermentum 2-14 + MGO+Cq treatment. B The cytotoxicity of Caco-2 cells after different treatments. Quantification (C) and representative images (D) of TUNEL staining after different treatments. Normalized Bad (E), Caspase-3 (F), and Beclin-2 (G) mRNA levels after different treatments. The protein levels of Bad (H), Caspase-3 (I), and Beclin-2 (J) after different treatments. n.s., non-significant; *p < 0.05, **p < 0.01, ***p < 0.001, *****p < 0.0001.

Our previous results have shown that L. fermentum 2-14 inhibits MGO-induced oxidative stress and inflammation (Fig. 3C–E). To investigate whether autophagy is necessary in this process, we measured the activity of antioxidant enzymes, the production of ROS, the concentration of malonaldehyde (MDA) in Caco-2 cells, and the expression of inflammatory factors. Incubating Caco-2 cells with MGO was found to decrease the activity of catalase (CAT), superoxide dismutase (SOD), and glutathione peroxidase (GSH-Px), while increasing the production of ROS and MDA compared with the control. L. fermentum 2-14 treatment rescued the phenotypes induced by MGO. Inhibiting autophagy by using Cq reduces the activity of antioxidant enzymes and promotes ROS and MDA production (Fig. 6).

Fig. 6. Inhibition of autophagy reduces the activity of antioxidant enzymes and increases the production of ROS and MDA.

Fig. 6

Open in a new tab

The enzyme activity of CAT (A), SOD (B), GSH-Px (C) after control, L. fermentum 2-14, MGO, L. fermentum 2-14 + MGO, L. fermentum 2-14 + MGO+Cq treatment. D The level of ROS production after different treatments. E The concentration of MDA in Caco-2 cells after different treatments. n.s., non-significant; *p < 0.05, **p < 0.01, ****p < 0.0001.

In addition, we found that MGO administration promoted the transcription of the pro-inflammatory genes IL-1β, IL-6 and TNF-α compared with the control group. Following MGO treatment, the protein levels of IL-1β and IL-6 increased, while the protein level of the anti-inflammatory factor IL-10 decreased. Pre-incubating Caco-2 cells with L. fermentum 2-14 reduced the transcription levels of IL-1β, IL-6 and TNF-α, as well as decreasing protein levels of IL-1β and IL-6. Conversely, IL-10 expression increased following L. fermentum 2-14 treatment. Blocking autophagy increased the relative mRNA levels of IL-1β, IL-6 and TNF-α, while reducing the relative mRNA level of IL-10. The transcription and protein levels of IL-10 decreased following Cq treatment (Fig. 7 and Supplementary Fig. 2). In summary, our results demonstrated that autophagy is essential for the mitigating effects of L. fermentum 2-14 on oxidative stress and inflammation.

Fig. 7. Inhibition of autophagy promotes inflammation.

Fig. 7

Open in a new tab

The transcription (A–C) and protein (D–F) levels of IL-1β (A, D), IL-6 (B, E) and IL-10 (C, F) after control, L. fermentum 2-14, MGO, L. fermentum 2-14 + MGO, L. fermentum 2-14 + MGO+Cq treatment. n.s., non-significant; *p < 0.05, **p < 0.01, ***p ﹤ 0.001, *****p < 0.0001.

L. fermentum 2-14 mitigates MGO-induced cytotoxicity via promoting the production of pyruvic acid to activate AMPK signaling pathway associated with autophagy

To investigate how L. fermentum 2-14 activates autophagy to mitigate the cytotoxicity induced by MGO, we first performed transcriptome sequencing. Both principal component analysis (PCA, Fig. 8A) and gene expression heatmaps (Fig. 8B) of transcriptome sequencing revealed significant variations in gene expression across the control group, the MGO group and the L. fermentum 2-14 + MGO group. Good parallelism is observed within each group. According to the Venn plot (Fig. 8C), of the examined genes, 858 are specifically expressed in the control group, 383 in the MGO group, and 331 in the L. fermentum 2-14 + MGO group. Additionally, MGO treatment was found to result in the upregulation of 525 genes and the downregulation of 580 genes compared with the control group. Pre-treating with L. fermentum 2-14 caused 1054 genes to be upregulated and 931 genes to be downregulated, when compared with the MGO group (Fig. 8D, F). Then, we performed the Gene Ontology (GO) annotation and the Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. We found that 1217, 2448, and 2059 genes are involved in the biological process, cellular component, and molecular function classes, respectively, in MGO-treated Caco-2 cells compared with the control group. In L. fermentum 2-14-pretreated Caco-2 cells, 2906, 3561 and 1896 genes were involved in the biological process, cellular component and molecular function classes, respectively, compared with the MGO group (Supplementary Fig. 3A). Finally, KEGG enrichment analysis revealed significant enrichment of the AMPK signaling pathway in the differentially expressed genes between the MGO and L. fermentum 2-14 + MGO groups (Fig. 8G, Supplementary Fig. 3B). By analyzing changes in the TPM values of multiple genes in the AMPK signaling pathway across the three treatment groups, we discovered that genes closely related to mitochondrial function, lipid metabolism and gluconeogenesis are significantly more highly expressed in the L. fermentum 2-14 + MGO groups (Fig. 8H). In conclusion, L. fermentum 2-14 may mitigate MGO-induced cytotoxicity by regulating the AMPK signaling pathway. This regulatory effect involves enhancing mitochondrial function and improving lipid and glucose metabolism, thereby improving the cell’s ability to adapt to and resist energy stress.

Fig. 8. AMPK signaling pathway is activated by L. fermentum 2-14.

Fig. 8

Open in a new tab

A The PCA analysis of the control group, the MGO group, and the L. fermentum 2-14 + MGO group. B The heatmap for gene expression of different samples. C The Venn plot of the three treatment group. D The Volcano plot showing differences in gene expression levels between the control group and the MGO group. E The Volcano plot showing differences in gene expression levels between the MGO group and the L. fermentum 2-14 + MGO group. F The number of differential genes. G The KEGG enrichment analysis of MGO vs L. fermentum 2-14 + MGO. H The TPM value of genes in the AMPK signaling pathway between the three treatments.

We then performed metabolomics analysis. Both PCA (Fig. 9A) and metabolite heatmap (Fig. 9B) revealed significant variations in metabolites across the control group, the MGO group and the L. fermentum 2-14 + MGO group. Good parallelism is observed within each group. According to the Venn plot (Fig. 9C), of the examined metabolites, 1 is specifically existed in the control group, 11 in the MGO group, and 3 in the L. fermentum 2-14 + MGO group. Additionally, MGO treatment was found to result in the upregulation of 63 metabolites and the downregulation of 243 metabolites compared with the control group (Supplementary Fig. 4A, B). Pre-treating with L. fermentum 2-14 caused 154 metabolites to be upregulated and 200 metabolites to be downregulated, when compared with the MGO group (Fig. 9D, Supplementary Fig. 4B). KEGG enrichment analysis revealed significant enrichment of the AMPK signaling pathway between the MGO and L. fermentum 2-14 + MGO groups (Fig. 9E, Supplementary Fig. 4C). The abundance of metabolites nicotinamide adenine dinucleotide (NAD+, Fig. 9F) and pyruvic acid (Fig. 9G), which are associated with the AMPK signaling, significant increase after L. fermentum 2-14 treatment.

Fig. 9. Metabolomics analysis of the control, MGO and L. fermentum 2-14 + MGO treatment group.

Fig. 9

Open in a new tab

A The PCA analysis of the control group, the MGO group and the L. fermentum 2-14 + MGO group. B Metabolite heatmap of different samples. C The Venn plot of three treatment group. D The volcano plot showing differential metabolites for the MGO and the L. fermentum 2-14 + MGO group. E The KEGG enrichment analysis of MGO vs L. fermentum 2-14 + MGO. F The abundance of NAD+ in Caco-2 cells after MGO and L. fermentum 2-14 treatments. G The abundance of pyruvic acid in Caco-2 cells after MGO and L. fermentum 2-14 treatments. H, Correlation analysis of differentially expressed genes and differentially expressed metabolites with L. fermentum 2-14 alleviating MGO cytotoxicity. n.s., non-significant; *p < 0.05, **p < 0.01.

Finally, we investigated how differentially expressed genes and differential metabolites affect the mitigation of MGO-induced cytotoxicity by L. fermentum 2-14. As shown in Fig. 9H, there is a strong positive correlation between cell viability and various biological markers, including LC3B, Beclin-2, IL-10, CAT, SOD, GSH-Px, SIRT1, PCK2, PPARG, CREB5, PFKFB3, IRS2, NAD+, and pyruvic acid. Additionally, Beclin-2, CAT, SOD, GSH-Px, SIRT1, PCK1, PPARG, IRS2, and pyruvic acid are significantly negatively correlated with apoptosis and cytotoxicity, while P62, Bad, Bax, Caspase-3, Caspase-9, IL-1β, IL-6, TNF-α, NF-κB, ROS, MDA, and FASN are significantly positively correlated with them. These findings suggest that apoptosis may be triggered by inflammation and oxidative stress. Furthermore, the differential metabolite NAD+ showed a significant positive correlation with cell viability, LC3B, Beclin-2, IL-10, SIRT1, CREB5, and IRS2. Meanwhile, pyruvic acid exhibited a significant positive correlation with cell viability, Beclin-2, IL-10, CAT, GSH-Px, SIRT1, PCK2, PCK1, PPARG, CREB5, PFKFB3, and IRS2 (Fig. 9H).

Our previous results implied that L. fermentum 2-14 increases the production of NAD+ and pyruvic acid in Caco-2 cells, thereby activating the AMPK signaling pathway related to autophagy. To verify our hypothesis, we first added the AMPK signaling pathway inhibitor Dorsomorphin dihydrochloride (Dd) and then examined the cell viability of the Caco-2 cells. We found that inhibiting the AMPK signaling pathway blocks the mitigating effect of L. fermentum 2-14 on the decrease in cell viability induced by MGO (Fig. 10A), indicating that the AMPK signaling pathway is essential for L. fermentum 2-14 to alleviate the decrease in cell viability induced by MGO. Although adding NAD+ alone increased the viability of Caco-2 cells slightly but significantly, it did not mitigate the decrease in cell viability caused by MGO (Fig. 10B). However, adding pyruvic acid mitigates the MGO-induced decrease in cell viability (Figs. 10C, D). This mitigating effect is impaired when autophagy (Fig. 10C) or AMPK signaling pathway (Fig. 10D) is inhibited. In conclusion, our data indicate that L. fermentum 2-14 increases pyruvic acid production in Caco-2 cells, thereby activating the AMPK signaling pathway and autophagy to mitigate MGO-induced cytotoxicity.

Fig. 10. Pyruvic acid increases cell viability via the AMPK signaling pathway and autophagy.

Fig. 10

Open in a new tab

A–D The cell viability of Caco-2 cells after indicated treatments. n.s., non-significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Discussion

This study investigated the toxic effects of MGO on Caco-2 cells and the mitigating role of L. fermentum 2-14. Our results demonstrated that MGO induces apoptosis, oxidative stress, and inflammatory responses, while L. fermentum 2-14 significantly alleviates this damage by reducing ROS/MDA accumulation, enhancing antioxidant enzyme activities (SOD, CAT, GSH-Px), and modulating the IL-10/IL-1β balance. Mechanistically, L. fermentum 2-14 promotes autophagy—evidenced by increased LC3, Beclin-1, and ATG16L1 expression and P62 degradation—thereby enhancing autophagosome and autolysosome formation. This protective effect was found to be autophagy-dependent, as inhibiting autophagy reversed the anti-apoptotic and anti-inflammatory benefits. Furthermore, transcriptomic and metabolomic analyses revealed that L. fermentum 2-14 stimulates pyruvic acid production, which activates the AMPK signaling pathway to drive autophagy.

Probiotic effects are often strain-dependent, and our screening results similarly indicate that only a subset of LAB strains provide robust protection against MGO-induced cytotoxicity. These findings align with prior reports showing that selected LAB strains (e.g., L. plantarum, and L. rhamnosus GG) can protect epithelial models such as Caco-2 from inflammatory or oxidative challenges and can reprogram host transcriptional responses72,73. LPS produced by Gram-negative bacteria is the primary cause of autophagy in intestinal epithelial cells. Similarly, LAB are of significant research interest in protecting rat intestinal epithelium from LPS-induced intestinal homeostasis disruption by inhibiting autophagy74. Furthermore, You and colleagues demonstrated that L. fermentum exerts a protective effect against DSS-induced intestinal damage by increasing the levels of the anti-inflammatory metabolite hercogenin in the intestine, which may facilitate the treatment of ulcerative colitis75.

Several limitations should be acknowledged. First, all experiments were performed in a single cancer-derived epithelial line (Caco-2), which lacks key components of the in vivo intestinal environment (mucus, immune cells, microbiota interactions, and physiological gradients)76. Accordingly, our conclusions should be restricted to this in vitro model, and translational claims should be stated conservatively. Future work combining orthogonal pharmacological tools with genetic perturbations (e.g., ATG5/ATG7 modulation) will strengthen pathway specificity. Future studies should extend validation to more physiologically relevant systems and barrier-function readouts, including TEER measurements, FITC–dextran permeability assays, and tight junction analyses (ZO-1, occludin, claudins). Intestinal organoids, differentiated IEC models, and co-culture systems incorporating mucus and immune components may better capture host–microbe interactions. Ultimately, in vivo testing in dietary MGO/AGE exposure models will be essential to assess safety, colonization and metabolite dynamics, and epithelial barrier/inflammatory outcomes. Collectively, our results provide a coherent functional and mechanistic framework for the protective role of L. fermentum 2-14 against MGO cytotoxicity in a Caco-2 model and support further exploration of probiotic–metabolite–AMPK/autophagy-based strategies to counter dietary dicarbonyl stress.

Materials and methods

Bacterial strains and cells

The 35 LAB strains used in this experiment were screened, isolated, and purified from human feces, then stored in our laboratory (see Supplementary Table 1). The strains were cultured in DeMan, Rogosa, and Sharpe (MRS) broth medium. The human colon adenocarcinoma cell line Caco-2 was purchased from the American Type Culture Collection Center (ATCC). The Caco-2 cells were cultured in complete essential medium (MEM, HyClone) containing 10% (v/v) fetal bovine serum (FBS, TIANHANG Biotechnology), 1% non-essential amino acids (Gibco), and 1% (w/v) penicillin-streptomycin solution (Biosharp Biotechnology). The cell lines were cultured at 37 °C in a humidified environment with 5% CO2.

Caco-2 cells preparation

Once the Caco-2 cells in the culture dishes had attached with 80–90% efficiency, the cells were harvested and counted using a haemocytometer. The cells were then adjusted to a density of 10⁵ cells/mL. A 100 μL cell suspension was seeded into each well of a 96-well cell culture plate, and the edges of the wells were treated with PBS. The cells were cultured overnight at 37 °C in a 5% CO₂ incubator until they reached 80%-90% attachment efficiency, after which they were treated with MGO (Sigma) and/or LAB.

MGO treatment

Media containing 0.4, 0.8, 1.2, 1.6, 2.0, 2.4, 2.8, 3.2, and 3.6 mM MGO were added. After 24 hours of culture, the viability and cytotoxicity of the Caco-2 cells were assessed. Six replicate wells were set up for each experimental group, and six control and blank wells were also established.

LAB treatment

The activated LAB were inoculated at a concentration of 2% (v/v) into 10 mL of MRS medium and cultured at 37 °C for over 18 hours. 1 mL of the bacterial culture was then centrifuged at 8000 g for 12 min at 4 °C. The resulting supernatant was discarded, and the bacterial pellet was resuspended in phosphate-buffered saline (PBS, Biosharp Biotechnology). This suspension was then centrifuged again under the same conditions three times. Finally, the bacterial pellet was resuspended in an antibiotic-free complete medium, and the bacterial suspension was diluted to 10⁹ CFU/mL for subsequent experiments.

Caco-2 cells were digested using an antibiotic-free complete medium (penicillin-streptomycin solution), and then inoculated into 96-well plates for 24 hours. After removing the initial medium from the wells of the LAB-treated and LAB groups, 100 μL of the LAB suspension (10⁹ CFU/mL) was added. An equal volume of MEM was added to the blank and MGO groups. The cells were then incubated in a 5% CO₂ incubator at 37 °C for a further 6 hours. After this time, the medium was discarded. Then, 100 μL of MEM containing 2.4 mM MGO was added to the LAB-treated and MGO groups. An equal volume of MEM was added to the LAB and blank groups. The cells were then incubated for 24 hours.

Cell viability and cytotoxicity assay of Caco-2 cells

For cell viability experiments, 100 μL of CCK-8 solution (Beyotime Biotechnology) diluted in MEM was added to the Caco-2 cells treated with MGO and LAB. The plate was then incubated at 37 °C for 2 hours and the absorbance at 450 nm was measured using a multi-mode microplate reader to calculate cell viability. Cell viability (% control) = (absorbance of experimental group – background of experimental group)/(absorbance of control group – background of control group).

For cytotoxicity experiments, add 100 μL of the LDH assay working solution (Beyotime Biotechnology) to each well of a 96-well plate. Mix well, and then incubate on a horizontal shaker with slow shaking at room temperature (approximately 25 °C) away from light for 10–30 minutes. Add 20 μL of termination solution to each well, mix well, and then measure the absorbance at 450 nm to calculate the cytotoxicity. LDH cytotoxicity (% control) = (absorbance of experimental group – background of experimental group)/(absorbance of control group – background of control group).

Apoptosis assay of Caco-2 cells

After 24 hours of complete cell inoculation in 12-well plates, the cells were treated as described above. The cells were fixed in 4% paraformaldehyde (Beyotime Biotechnology) for 30 min and permeabilized in 0.1% Triton X-100 (Beyotime Biotechnology) for 10 min. Then the cells were washed with PBS solution 2-3 times. The TUNEL kit (Novozymes Bioscience and Technology) was used to label apoptotic cells. 4’, 6-diamidino-2-phenylindole (DAPI, Beyotime Biotechnology) was used to stain the cell nuclei for 5 min. Three or more randomly selected fields of view were used for image acquisition by fluorescence microscope, and three independent biological replicates and a positive apoptosis-inducing control were set up in each group of experiments. Three independent biological replicates were set up for each set of experiments and a positive control for apoptosis induction was established. The images were analyzed using Image J to measure the average fluorescence intensity.

Oxidative stress assay in Caco-2 cells

For ROS quantification, the Caco-2 cells were treated with MGO and/or LAB as described above. Then the cells were collected by trypsinization (Biosharp Biotechnology) and washed three times with pre-cooled PBS. Reactive Oxygen Species Assay Kits (Beyotime Biotechnology) were used to quantify the content of ROS in Caco-2 cells. Absorbance was measured at EX = 488 nm and EM = 525 nm using a fluorescence spectrophotometer.

For antioxidase activities measurement, cells were washed once with pre-cooled PBS. 200 μL of prepared radioimmunoprecipitation (RIPA, Beyotime Biotechnology) lysis buffer was added to each well. After an ice bath and shaking for 10 min, the lysates were collected into centrifuge tubes and centrifuged at 11,000 g for 15 min at 4 °C. The supernatant was collected for further testing. Protein concentrations in the test samples were determined using a bicinchoninic acid (BCA) protein assay kit (Beyotime Biotechnology), followed by measurement of superoxide dismutase (SOD), catalase (CAT), and glutathione peroxidase (GSH-Px) activities using the corresponding assay kits (Shanghai Biotechnology). Malondialdehyde (MDA) content was determined according to the instructions of the assay kit (Beyotime Biotechnology).

Real-Time quantitative Polymerase Chain Reaction (RT-qPCR) analysis

Primers used in this study for RT-qPCR were designed with Primer Premier 6 and are listed in Supplementary Table 2. Total RNA was extracted using TRIzol reagent (Vazyme), and cDNA was synthesized using a commercial reverse transcription kit (Takara). RT-qPCR was performed using ChamQ Universal SYBR qPCR Master Mix (Vazyme) on an Applied Biosystems QuantStudio 3 system. The thermal cycling conditions were as follows: 95 °C for 30 s; 40 cycles of 95 °C for 10 s and 60 °C for 30 s, with fluorescence acquisition at the annealing/extension step; followed by melting curve analysis at 95 °C for 15 s, 60 °C for 60 s, and 95 °C for 15 s. GAPDH was used as the internal control, and relative gene expression was calculated using the 2^−ΔΔCt method.

Enzyme-Linked Immunosorbent Assay (ELISA) analysis

Treated Caco-2 cells were washed twice using pre-cooled sterile PBS. The lysis system containing protease inhibitors (RIPA buffer: PMSF) was prepared at a 100:1 (v/v) ratio, and 200 μL of pre-cooled lysate was added to each well. Place the culture plate on ice and perform staged lysis: initial static lysis for 5 min, shaking at 50 rpm on a horizontal shaker for 3 min, then secondary static lysis for 2 min. During this time, collect the adherent cells using a sterile cell scraper in a fixed direction (scrape off in a single clockwise direction). The lysate mixture was transferred to a pre-cooled centrifuge tube and centrifuged at 12,000 rpm for 15 min at 4 °C, and the supernatant was carefully transferred to a labeled freezing tube. Protein quantification was performed using a BCA kit, and the IL-1β, IL-10, LC-3, P62, Bad, Bax, Caspase-3, Caspase-9, Beclin-2, IL-6, NF-κB, and TNF-α content in Caco-2 cells were detected according to the standard procedure of the ELISA kit (SenBeiJia Biological Technology), and the data were finally read at 450 nm.

Transcriptome analysis

Caco-2 cells treated with MGO and/or LAB were collected, and RNA extraction and sequencing were performed by Majorbio Biomedical. The data were analyzed on the free online platform of Majorbio Cloud Platform (www.majorbio.com). After obtaining the read counts of the genes, the DESeq2 algorithm was used for differential gene expression analysis between groups. Genes with p-adjust 0.05 and |log2fold change (FC)| > 1 were considered differentially expressed. Then, the differentially expressed genes (DEGs) were subjected to functional annotation and enrichment analysis.

Metabolome analysis

Fifty mg of sample was accurately weighed, and a 6 mm diameter grinding bead was added to 400 μL of extraction solution (methanol: water, 4:1) containing four internal standards. The samples were ground for 6 min (−10 °C, 50 Hz) on a frozen tissue grinder and extracted via cryosonication for 30 min (5 °C, 40 kHz). The samples were incubated to stand at − 20 °C for 30 min, centrifuged for 15 min (13,000 g, 4 °C), and the supernatant was pipetted into an injection vial with an internal cannula for analysis.

In addition, 20 μL of supernatant was pipetted from each sample separately and mixed as a quality control (QC) sample. The Liquid Chromatograph-Mass Spectrometer (LC-MS) analysis of the sample was conducted on a SCIEX UHPLC-Triple TOF system equipped with an ACQUITY UPLC HSS T3 column (Waters) at Majorbio Bio-Pharm Technology. The pre-treatingof LC/MS raw data was performed by Progenesis QI (Waters) software. Internal standard peaks and all false positive peaks were removed from the data matrix, de-redundant and peak-pooled. At the same time, the metabolites were identified by searching databases, including the Human Metabolome Database (HMDB) (http://www.hmdb.ca/), Metlin Database (https://metlin.scripps.edu/), and Majorbio Database. Venn diagram was used to calculate the number of common and unique metabolites in different groups, and partial least squares-discriminant analysis (PLS-DA) was used to establish a regression model between metabolite expression and sample category, so as to predict sample category. An orthogonal PLS-DA (OPLS-DA) model was constructed to obtain variable importance in the projection (VIP) value, and FC ≥ 1, VIP ≥ 1, and p-value ≤ 0.05 were used as criteria to screen for significant differential metabolites. Finally, the enrichment analysis of metabolites with significant differences was conducted based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. Paired Spearman rank correlations (|r|> 0.7, P < 0.05) were calculated for cell viability, cytotoxicity, apoptosis, autophagy genes LC3B and P62, apoptotic genes Beclin-2, Bad, Bax, Caspase-3, Caspase-9, inflammatory genes IL-1β, IL-6, IL-10, TNF-α, NF-kB, antioxidant indexes ROS, MDA, CAT, SOD, GSH-Px as well as correlation with the 12 differentially expressed genes and 2 differential metabolites in the AMPK pathway for correlation analysis by using R 3.5.0, and then correlation heatmaps were plotted using Origin.

Statistical analysis

All experiments were performed in at least three replications for each strain or condition. The measured data were expressed as mean ± SEM. Before parametric analysis, normality and homogeneity of variance were verified using the Shapiro–Wilk test and Levene’s test, respectively. Data were analyzed by independent samples t-test, one-way ANOVA, or Duncan’s analysis according to experimental grouping and experimental purpose using IBM SPSS Statistics 26.0, and plots were generated using GraphPad Prism 9.0. A p < 0.05 was considered statistically significant. n.s., non-significant; p < 0.05, *p < 0.01, **p < 0.001, ***p < 0.0001.

Supplementary information

41538_2026_932_MOESM1_ESM.pdf (440.1KB, pdf)

Revised Supplementary Figures and Tables.

Acknowledgements

This research was supported by National Natural Science Foundation of China (No. 32272299), the National Key Research and Development Program of China (No. 2024YFF1106100), and the China Postdoctoral Science Foundation under Grant Number 2025M772605.

Author contributions

Junjun Gao: Writing – review & editing, Writing – original draft. Yuqin Chen: Data curation. Junfeng Xiao, Siyu Guo: Investigation, Formal analysis. Manli Li, Yapeng Guo, Lei Wang, Mingye Peng: Supervision, Software. Yang Chen, Mengzhou Zhou: Writing – review & editing, Funding acquisition.

Data availability

The transcriptome data have been deposited in the NCBI Sequence Read Archive (SRA) under the BioProject accession number PRJNA1441130. The metabolome data have been deposited in the EMBL-EBI Metabolights repository under the accession number MTBLS14104.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Contributor Information

Yang Chen, Email: 20231134@hbut.edu.cn.

Mengzhou Zhou, Email: zmzkelvin@hbut.edu.cn.

Supplementary information

The online version contains supplementary material available at 10.1038/s41538-026-00932-2.

References

  • 1.Nemet, I., Varga-Defterdarović, L. & Turk, Z. Methylglyoxal in food and living organisms. Mol. Nutr. Food Res.50, 1105–1117 (2006). [DOI] [PubMed] [Google Scholar]
  • 2.Thornalley, P. J. Dicarbonyl Intermediates in the Maillard Reaction. Ann. N. Y. Acad. Sci.1043, 111–117 (2006). [DOI] [PubMed] [Google Scholar]
  • 3.Bansal, S., Burman, A. & Tripathi, A. K. Advanced glycation end products: Key mediator and therapeutic target of cardiovascular complications in diabetes. World J. Diab14, 1146–1162 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Rabbani, N. & Thornalley, P. J. Methylglyoxal, glyoxalase 1 and the dicarbonyl proteome. Amino Acids42, 1133–1142 (2010). [DOI] [PubMed] [Google Scholar]
  • 5.Hellwig, M., Gensberger-Reigl, S., Henle, T. & Pischetsrieder, M. Food-derived 1,2-dicarbonyl compounds and their role in diseases. Semin. Cancer Biol.49, 1–8 (2018). [DOI] [PubMed] [Google Scholar]
  • 6.Ott, C. et al. Role of advanced glycation end products in cellular signaling. Redox Biol.2, 411–429 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Wang, G. et al. M Metformin prevents methylglyoxal-induced apoptosis by suppressing oxidative stress in vitro and in vivo. Cell Death Dis.13, 29 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Wan, R. et al. Methylglyoxal induces endothelial cell apoptosis and coronary microvascular dysfunction through regulating AR-cPLA2 signaling. Biochim. Biophys. Acta (BBA)-Mol. Basis Dis.1870, 167437 (2024). [DOI] [PubMed]
  • 9.Kim, S. et al. Apelin-13 inhibits methylglyoxal-induced unfolded protein responses and endothelial dysfunction via regulating AMPK pathway. Int. J. Mol. Sci.21, 4069 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Thornalley, P. J. Protein and nucleotide damage by glyoxal and methylglyoxal in physiological systems - role in ageing and disease. Drug Metab. Drug Interact.23, 125–150 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Paramita, D. & Wisnubroto, J. D. P. Effect of methylglyoxal on reactive oxygen species, KI-67, and caspase-3 expression in MCF-7 cells. Exp. Mol. Pathol.105, 76–80 (2018). [DOI] [PubMed] [Google Scholar]
  • 12.Schalkwijk, C. G. & Stehouwer, C. D. A. Methylglyoxal, a Highly Reactive Dicarbonyl Compound, in Diabetes, Its Vascular Complications, and Other Age-Related Diseases. Physiol. Rev.100, 407–461 (2020). [DOI] [PubMed] [Google Scholar]
  • 13.Wang, C.-H. et al. Protective evaluation of compounds extracted from root of Rhodiola rosea L. against methylglyoxal-induced toxicity in a neuronal cell line. Molecules25, 2801 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Zhang, D. et al. Protective effect of flavonoids against methylglyoxal-induced oxidative stress in PC-12 neuroblastoma cells and its structure–activity relationships. Molecules27, 7804 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Park, S. et al. Inhibitory effect of Arachis hypogaea (Peanut) and Its phenolics against methylglyoxal-derived advanced glycation end product toxicity. Nutrients9, 1214 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.He, Y. et al. EGCG attenuates the neurotoxicity of methylglyoxal via regulating MAPK and the downstream signaling pathways and inhibiting advanced glycation end products formation. Food Chem.384, 132358 (2022). [DOI] [PubMed] [Google Scholar]
  • 17.Wang, Y.-H., Yu, H.-T., Pu, X.-P. & Du, G.-H. Myricitrin alleviates Methylglyoxal-induced mitochondrial dysfunction and AGEs/RAGE/NF-κB pathway activation in SH-SY5Y cells. J. Mol. Neurosci.53, 562–570 (2014). [DOI] [PubMed] [Google Scholar]
  • 18.Cho, C.-H. et al. Therapeutic potential of Phlorotannin-Rich Ecklonia cava extract on methylglyoxal-induced diabetic nephropathy in in vitro model. Mar. Drugs20, 355 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Sompong, W., Cheng, H. & Adisakwattana, S. Ferulic acid prevents methylglyoxal-induced protein glycation, DNA damage, and apoptosis in pancreatic β-cells. J. Physiol. Biochem.73, 121–131 (2016). [DOI] [PubMed] [Google Scholar]
  • 20.Ji-Eun, K., Paras Man, P., Jang, S. & Yi, H.-K. Anti-inflammatory effect of luteoloside against methylglyoxal induced human dental pulp cells. J. Appl. Biomed.22, 33–39 (2024). [DOI] [PubMed] [Google Scholar]
  • 21.Seo, K., Seo, S., Han, J. Y., Ki, S. H. & Shin, S. M. Resveratrol attenuates methylglyoxal-induced mitochondrial dysfunction and apoptosis by Sestrin2 induction. Toxicol. Appl. Pharmacol.280, 314–322 (2014). [DOI] [PubMed] [Google Scholar]
  • 22.Huang, S.-M. et al. Inhibitory effect of vanillic acid on methylglyoxal-mediated glycation in apoptotic Neuro-2A cells. NeuroToxicology29, 1016–1022 (2008). [DOI] [PubMed] [Google Scholar]
  • 23.Wang, Y.-H., Han, Y.-P., Yu, H.-T., Pu, X.-P. & Du, G.-H. Protocatechualdehyde prevents methylglyoxal-induced mitochondrial dysfunction and AGEs-RAGE axis activation in Human lens epithelial cells. Eur. J. Pharmacol.738, 374–383 (2014). [DOI] [PubMed] [Google Scholar]
  • 24.Tang, D. et al. Pterostilbene prevents methylglyoxal-induced cytotoxicity in endothelial cells by regulating glyoxalase, oxidative stress and apoptosis. Food Chem. Toxicol.153, 112244 (2021). [DOI] [PubMed] [Google Scholar]
  • 25.Deng, Y. et al. Inhibitory effect of Ginkgo biloba seeds peptides on methylglyoxal-induced glycations. Food Chem. Toxicol.172, 113587 (2023). [DOI] [PubMed] [Google Scholar]
  • 26.Chen, W. et al. Lutein attenuated methylglyoxal-induced oxidative damage and apoptosis in PC12 cells via the PI3K/Akt signaling pathway. J. Food Biochem.46, e14382 (2022). [DOI] [PubMed] [Google Scholar]
  • 27.Angeloni, C. et al. Neuroprotective effect of sulforaphane against methylglyoxal cytotoxicity. Chem. Res. Toxicol.28, 1234–1245 (2015). [DOI] [PubMed] [Google Scholar]
  • 28.Suh, K. S., Chon, S., Jung, W. W. & Choi, E. M. Protective effects of sciadopitysin against methylglyoxal-induced degeneration in neuronal SK-N-MC cells. J. Appl. Toxicol.42, 274–284 (2021). [DOI] [PubMed] [Google Scholar]
  • 29.Zhang, S. et al. Umbelliferone protects against methylglyoxal-induced HUVECs dysfunction through suppression of apoptosis and oxidative stress. J. Appl. Toxicol.43, 490–499 (2022). [DOI] [PubMed] [Google Scholar]
  • 30.Suh, K. S., Chon, S. & Choi, E. M. Actein protects against methylglyoxal-induced oxidative damage in osteoblastic MC3T3-E1 cells. J. Sci. Food Agric.97, 207–214 (2016). [DOI] [PubMed] [Google Scholar]
  • 31.Suh, K. S., Chon, S. & Choi, E. M. Protective effects of piceatannol on methylglyoxal-induced cytotoxicity in MC3T3-E1 osteoblastic cells. Free Radic. Res.52, 712–723 (2018). [DOI] [PubMed] [Google Scholar]
  • 32.Suh, K. S., Chon, S. & Choi, E. M. Cytoprotective effects of xanthohumol against methylglyoxal-induced cytotoxicity in MC3T3-E1 osteoblastic cells. J. Appl. Toxicol.38, 180–192 (2017). [DOI] [PubMed] [Google Scholar]
  • 33.Schmoch, T. et al. Anserine reduces mortality in experimental sepsis by preventing methylglyoxal-induced capillary leakage. eBioMedicine114, 105644 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Lee, J. H., Subedi, L. & Kim, S. Y. Effect of Cysteine on Methylglyoxal-induced renal damage in mesangial cells. Cells9, 234 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Yılmaz, Z. et al. The effect of carnosine on methylglyoxal-induced oxidative stress in rats. Arch. Physiol. Biochem.123, 192–198 (2017). [DOI] [PubMed] [Google Scholar]
  • 36.Koike, S., Nishimoto, S. & Ogasawara, Y. Cysteine persulfides and polysulfides produced by exchange reactions with H2S protect SH-SY5Y cells from methylglyoxal-induced toxicity through Nrf2 activation. Redox Biol.12, 530–539 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Koike, S. et al. Polysulfides protect SH-SY5Y cells from methylglyoxal-induced toxicity by suppressing protein carbonylation: A possible physiological scavenger for carbonyl stress in the brain. NeuroToxicology55, 13–19 (2016). [DOI] [PubMed] [Google Scholar]
  • 38.McCormick, D. L. et al. Linagliptin Ameliorates Methylglyoxal-Induced Peritoneal Fibrosis in Mice. PloS one11, e0160993 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Figarola, J. L., Singhal, J., Rahbar, S., Awasthi, S. & Singhal, S. S. LR-90 prevents methylglyoxal-induced oxidative stress and apoptosis in human endothelial cells. Apoptosis19, 776–788 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Pintus, G. et al. Edaravone protects against methylglyoxal-induced barrier damage in human brain endothelial cells. PloS one9, e100152 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Tóth, A. E. et al. Compounds blocking methylglyoxal-induced protein modification and brain endothelial injury. Arch. Med. Res.45, 753–764 (2014). [DOI] [PubMed] [Google Scholar]
  • 42.Rao, T. et al. The pharmacogenetics of natural products: A pharmacokinetic and pharmacodynamic perspective. Pharmacol. Res.146, 104283 (2019). [DOI] [PubMed] [Google Scholar]
  • 43.Zhang, H., Wang, M. & Xiao, J. Stab. Polyphen. Food Process.102, 1–45 (2022). [Google Scholar]
  • 44.Liu, X. et al. Toxicological evaluation of advanced glycation end product Nε-(carboxymethyl)lysine: Acute and subacute oral toxicity studies. Regulat. Toxicol. Pharmacol.77, 65–74 (2016). [DOI] [PubMed] [Google Scholar]
  • 45.Sarao, L. K. & Arora, M. Probiotics, prebiotics, and microencapsulation: A review. Crit. Rev. Food Sci. Nutr.57, 344–371 (2015). [DOI] [PubMed] [Google Scholar]
  • 46.Lombardo-Hernandez, J. et al. An in vitro neurobacterial interface reveals direct modulation of neuronal function by gut bacteria. Sci. Rep.15, 25535 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Garbacz, K. Anticancer activity of lactic acid bacteria. Semin. Cancer Biol.86, 356–366 (2022). [DOI] [PubMed] [Google Scholar]
  • 48.de Moreno de LeBlanc, A. et al. Current review of genetically modified lactic acid bacteria for the prevention and treatment of colitis using murine models. Gastroenterol. Res. Pract.2015, 1–8 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Aldunate, M. et al. Antimicrobial and immune modulatory effects of lactic acid and short chain fatty acids produced by vaginal microbiota associated with eubiosis and bacterial vaginosis. Front. Physiol.6, 164 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Toshimitsu, T. Development of a lactic acid bacteria strain that suppresses chronic inflammation and improves glucose and lipid metabolism. Biosci. Microbiota Food Health42, 3–7 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Park, H.-Y. et al. Lactococcus lactis KF140 reduces dietary absorption of Nε - (Carboxymethyl)lysine in rats and humans via β-Galactosidase activity. Front. Nutr.9, 916262 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Li, Q. et al. Inhibition and mechanism of protein nonenzymatic glycation by Lactobacillus fermentum. Foods13, 1183 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Lin, S.-W. et al. Fermented Supernatants of Lactobacillus plantarum GKM3 and Bifidobacterium lactis GKK2 Protect against Protein Glycation and Inhibit Glycated Protein Ligation. Nutrients15, 277 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Parzych, K. R. & Klionsky, D. J. An overview of autophagy: morphology, mechanism, and regulation. Antioxid. Redox Signal.20, 460–473 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Yu, L., Chen, Y. & Tooze, S. A. Autophagy pathway: Cellular and molecular mechanisms. Autophagy14, 207–215 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Glick, D., Barth, S. & Macleod, K. F. Autophagy: cellular and molecular mechanisms. J. Pathol.221, 3–12 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Wei, Y., Liu, M., Li, X., Liu, J. & Li, H. Origin of the autophagosome membrane in mammals. BioMed. Res. Int.2018, 1–9 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Dafre, A. L., Schmitz, A. E. & Maher, P. Methylglyoxal-induced AMPK activation leads to autophagic degradation of thioredoxin 1 and glyoxalase 2 in HT22 nerve cells. Free Radic. Biol. Med.108, 270–279 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.Park, S.-H. et al. Autophagy functions to prevent methylglyoxal-induced apoptosis in HK-2 Cells. Oxid. Med. Cell. Longev.2020, 1–11 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 60.Wu, Y. et al. Probiotic Bacillus attenuates oxidative stress- induced intestinal injury via p38-mediated autophagy. Front. Microbiol.10 (2019). [DOI] [PMC free article] [PubMed]
  • 61.Goubet, A.-G. et al. Multifaceted modes of action of the anticancer probiotic Enterococcus hirae. Cell Death Differ.28, 2276–2295 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 62.Zhan, Y. et al. Lactobacillus plantarum P9 ameliorates dysfunctional immune and metabolic homeostasis along the gut–liver axis in a nonalcoholic fatty liver disease with low-grade colitis mouse model. Food Biosci.61, 104613 (2024). [Google Scholar]
  • 63.Engevik, M. A. et al. Bifidobacterium dentium fortifies the intestinal mucus layer via autophagy and calcium signaling pathways. mBio10, e01087–19 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 64.Vašková, Kováčová, J., Pudelský, G., Palenčár, J., Mič, D. & ková, H. Methylglyoxal Formation—metabolic routes and consequences. Antioxidants14, 212 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 65.Cha, S.-H., Hwang, Y., Heo, S.-J., Jun, H.-S. & Singhal, S. S. Diphlorethohydroxycarmalol attenuates methylglyoxal-induced oxidative stress and advanced glycation end product formation in human kidney cells. Oxid. Med. Cell. Longev.2018 (2018). [DOI] [PMC free article] [PubMed]
  • 66.Jarisarapurin, W., Kunchana, K., Chularojmontri, L. & Wattanapitayakul, S. K. Unripe Carica papaya Protects Methylglyoxal-Invoked Endothelial Cell Inflammation and Apoptosis via the Suppression of Oxidative Stress and Akt/MAPK/NF-κB Signals. Antioxidants10, 1158 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 67.Fang, L. et al. Autophagy protects human brain microvascular endothelial cells against methylglyoxal-induced injuries, reproducible in a cerebral ischemic model in diabetic rats. J. Neurochem.135, 431–440 (2015). [DOI] [PubMed] [Google Scholar]
  • 68.Lőrincz, P. & Juhász, G. Autophagosome-Lysosome Fusion. J. Mol. Biol.432, 2462–2482 (2020). [DOI] [PubMed] [Google Scholar]
  • 69.Tian, X., Teng, J. & Chen, J. New insights regarding SNARE proteins in autophagosome-lysosome fusion. Autophagy17, 2680–2688 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 70.Li, K. et al. High cholesterol induces apoptosis and autophagy through the ROS-activated AKT/FOXO1 pathway in tendon-derived stem cells. Stem Cell Res. Ther.11, 131 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 71.Levy, J. M. M., Towers, C. G. & Thorburn, A. Targeting autophagy in cancer. Nat. Rev. Cancer17, 528–542 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 72.Wei, C. X., Wu, J. H., Huang, Y. H., Wang, X. Z. & Li, J. Y. Lactobacillus plantarum improves LPS-induced Caco2 cell line intestinal barrier damage via cyclic AMP-PKA signaling. PLoS One17, e0267831 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 73.Hou, Y. et al. Transcriptomic responses of Caco-2 cells to Lactobacillus rhamnosus GG and Lactobacillus plantarum J26 against oxidative stress. J. Dairy Sci.102, 7684–7696 (2019). [DOI] [PubMed] [Google Scholar]
  • 74.Han, C. et al. The role of probiotics in Lipopolysaccharide-induced autophagy in intestinal epithelial cells. Cell Physiol. Biochem.38, 2464–2478 (2016). [DOI] [PubMed] [Google Scholar]
  • 75.You, X. et al. Lactobacillus fermentum inhibits the development of Colitis by regulating the intestinal microbiota and metabolites to enhance the intestinal barrier and decrease inflammatory responses. Int J. Mol. Sci.26, 9181 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 76.Sambuy, Y. et al. The Caco-2 cell line as a model of the intestinal barrier: influence of cell and culture-related factors on Caco-2 cell functional characteristics. Cell Biol. Toxicol.21, 1–26 (2005). [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

41538_2026_932_MOESM1_ESM.pdf (440.1KB, pdf)

Revised Supplementary Figures and Tables.

Data Availability Statement

The transcriptome data have been deposited in the NCBI Sequence Read Archive (SRA) under the BioProject accession number PRJNA1441130. The metabolome data have been deposited in the EMBL-EBI Metabolights repository under the accession number MTBLS14104.

Articles from NPJ Science of Food are provided here courtesy of Nature Publishing Group

RESOURCES