Activation of gut metabolite ACSL4/LPCAT3 by microplastics in drinking water mediates ferroptosis via gut–kidney axis

Abstract

The environmental pollutant Benzo[a]pyrene (BaP) is commonly found in the environment, with microplastics (MPs) acting as the primary carriers of BaP into living organisms, increasing its availability in the body. However, the specific pathways and mechanisms through which MPs carrying pollutants cause kidney damage are not fully understood. This study aimed to investigate the routes and mechanisms of kidney injury in mice to low concentrations of both MPs and BaP. The combination of polystyrene (PS) and BaP disrupted lipid metabolism in the kidneys, leading to a form of cell death known as ferroptosis. However, this effect was not observed in HK-2 cells in vitro, indicating a cell-specific response. Interestingly, in HIEC-6 cells, both PS and BaP directly induced ferroptosis. These findings confirm that exposure to both PS and BaP can disrupt metabolic homeostasis in the kidneys, contributing to kidney dysfunction and cell death.

Polystyrene and Benzo[a]pyrene induce ferroptosis in the kidney by disrupting intestinal metabolites, contributing to kidney disfunction.

Introduction

In recent years, the production of plastic waste has increased due to the convenience and overuse of plastics accompanied with human activities. These wastes with very limited degraded, could continuously break down into micro~nano scale small, namely microplastics (MPs), in various environments. The presence of MPs has already been detected in various foods^1–3, it is estimated that approximately 33 billion tonnes of plastic waste will be produced by 2050⁴. Moreover, MPs have been found in various parts of the human body, including the digestive tract⁵, lungs⁶, blood⁷, urine⁸, faeces⁹, and even placenta¹⁰. As plastic waste continues to be discharged, the threat to human health posed by MPs is increasing and imminent.

MPs can act as carriers for environmental pollutants, including heavy metals, polycyclic aromatic hydrocarbon (PAHs), and other hazardous substances, into organism^11,12. PAHs, which are persistent pollutants, are more likely to be adsorbed by MPs and enter the body through the respiratory and digestive tracts. Benzo(a)pyrene (BaP) is one of the most prevalent and toxic PAHs found in the environment and has been shown to be highly biotoxic^12–14. Recent studies have shown that MPs are capable of transporting BaP into organisms. MPs taken orally into the GI tract are not fully absorbed by the intestinal epithelium and the unabsorbed portion is transported across the intestinal vascular barrier to all parts of the body, leading to systemic inflammation. The specific mixture toxicities of both MPs and BaP in combination with organisms are yet to be clearly established. Some studies suggest that there may be antagonistic effects of MPs and BaP on organisms, while others suggest significant synergistic effects when they are co-administered to organisms. It is noted that the outcomes may depend on MPs’ type, size, and adsorption capacity for BaP as well as exposure time in experiments.

The digestive system is where external contaminants are first encountered, and the kidneys play a crucial role in waste metabolism in the body. Research has shown that the absorption of MPs and BaP from the digestive tract can cause significant damage to both the gut and kidneys^15,16. A recent study by Liang et al., found that polystyrene (PS), one of the commonly used plastics, can lead to kidney injury by disrupting the intestinal barrier and activating the C5a/C5aR pathway¹⁷. BaP has also been linked to oxidative stress in the kidneys, resulting in injury¹⁸.

Ferroptosis, a unique form of iron-dependent cell death, has been identified as a potential mechanism of kidney damage¹⁹. Excessive reactive oxygen species accumulation in kidney tubular cells can disrupt intracellular glutathione levels, leading to kidney ferroptosis²⁰. Increased intracellular iron levels and lipid peroxidation are also implicated in tissue damage²¹. Although, it remains unclear whether chronic exposure to MPs and BaP through the digestive tract can cause kidney damage via the gut-kidney axis.

A preliminary finding by our team did indicate kidney injury by the two contaminants in combination, which triggers further investigation. This study elucidated the toxic effects and mechanisms of two pollutants in subject animals by validating key genes and metabolites. The results show that by disrupting metabolites and gut flora in mice, co-exposure to PS and BaP leads to ferroptosis in the kidneys.

Results

PS characterisation

SEM analysis revealed that the PS particles were uniformly dispersed and did not aggregate, as evidenced by their regular smooth spherical shape (Supplementary Fig. 1a, b). Zeta potential measurements indicated that the PS particles had a high zeta potential, suggesting good stability (Supplementary Table 1)²².

PS and BaP induce kidney damage in mice

To monitor the health of the mice during drug administration, body weights and organ coefficients were measured and recorded on a weekly basis. The body weight of the mice did not change significantly (Fig. 1a). However, the kidney index was significantly elevated (P = 0.0001) in the 6th week (Fig. 1b). Biomarkers of renal function were tested, including serum creatinine (CRE), blood urea nitrogen (BUN), uric acid (UA), albumin and nitric oxide (NO) (Fig. 1c). Among the measured parameters, only CRE (P = 0.012), and UA (P < 0.0001) showed significant differences from the control group, but only in the Mix group. Only in the mixed group was renal function impaired. We then analysed the histological structure of the kidneys.

Fig. 1. Kidney damage has been found in mice exposed to PS and BaP.

a Changes in body weight of the mouse over time, n = 10. b Changes in the kidney index in mice, n = 10. c Renal function changes in mice, n = 5. d Histological changes were observed in the mouse kidney. Cellular vacuolisation was indicated by red arrows, ballooning degeneration by yellow arrows, inflammatory cells were present as shown by black arrows, glomerular atrophy and detachment by red triangles, and tubulolysis by black triangles. 200× scale bar is 100 μm, 400× scale bar is 40 μm, n = 4. All differences identified via one-way ANOVA followed by the Tukey post hoc test. * indicates p < 0.05 compared with the control (0 μg/mL) group.

The kidney in the control group exhibited normal structure, with full glomeruli and tubular structure (Fig. 1d). In the PS group, there was a slight vacuolisation phenomenon and balloon-like degeneration. The BaP group showed infiltration of inflammatory cells, and the glomeruli were atrophied and detached. The Mix group exhibited severe vacuolar phenomenon, inflammatory cell infiltration, tubulolysis, and severe balloon-like degeneration of the glomeruli. Furthermore, the solvent control group did not cause any harm to the mice.

Kidney metabolic disorders caused by PS and BaP

To further investigate the specific mechanisms of renal injury, we analysed the transcriptome and metabolome of the kidney. The control group was compared with the experimental group (PS group, BaP group, Mix group), and 1002, 3347 and 2560 DEGs were detected, respectively (Supplementary Fig. 2a). The differential genes were then analysed by GO and KEGG, and in the GO analysis, the PS, BaP, and Mix groups were significantly enriched in entries for redox processes, cofactor binding, lipid binding and ferric iron binding, respectively, compared to the control group (Fig. 2a–c). KEGG analyses showed that pathways related to iron metabolism, including iron metabolism, fat digestion and absorption, and cholesterol metabolism, were significantly enriched only in the Mix group (Fig. 2d–f). This suggests that the combined effect of the two significantly increases the risk of lipid peroxidation in the body, which may directly lead to lipid deposition in tissues, thereby disrupting lipid metabolism and iron metabolism homoeostasis, and ultimately leading to cellular ferroptosis. Enrichment analyses of differential genes also identified key genes such as Alox15, Apoa4, Slc27a, and Lpcat3 as collectively involved in and inducing cellular ferroptosis in the kidney (Supplementary Fig. 2b, c).

Fig. 2. Imbalance of the metabolic homoeostasis of the kidney.

a Gene Ontology (GO) analysis for kidney transcriptome in Con and PS groups. b Gene Ontology (GO) analysis for kidney transcriptome in Con and BaP groups. c Gene Ontology (GO) analysis for kidney transcriptome in Con and Mix groups. d Kidney transcriptional analysis with KEGG in Con and PS groups. e Kidney transcriptional analysis with KEGG in Con and BaP groups. f Kidney transcriptional analysis with KEGG in Con and Mix groups, n = 4.

We also investigated and analysed the metabolites of the kidney. The results of the PCA of the experimental and control groups showed that there was a significant difference in the composition of the differential metabolites, as well as a significant difference in the metabolites between the experimental and control groups (Supplementary Fig. 2d–f). In addition, we also compared the key renal differential metabolites with the HMDB database, and most differential metabolites were enriched in lipid and lipid-like molecules, which accounted for 42.96%. This is in agreement with the transcriptome results (Supplementary Fig. 2g). KEGG analysis of differential metabolites showed that differential metabolites in the PS group were mainly enriched in pathways related to inflammation, including pathways such as retrograde endocannabinoids, whereas differential metabolites in the BaP and Mix groups were mainly enriched in pathways related to ferroptosis, including ferroptosis, and glycerophospholipid metabolism (Supplementary Fig. 3a–c). This is in agreement with our transcriptome results. Therefore, a cluster analysis of key metabolites involved in lipid metabolism and ferroptosis in the control and mixed groups showed that key lipid metabolites such as arachidonic acid, adrenoic acid, and stearic acid were significantly elevated in the mixed group, a phenomenon that also suggests that these lipid metabolites are involved in renal lipid peroxidation and play a key role in the process of renal ferroptosis (Supplementary Fig. 4).

To further explore the association of differential genes and differential metabolites in mouse kidney after exposure to exogenous pollutants, we performed co-analysis (Fig. 3). The results showed that the PS group co-enriched the glutathione metabolism pathway, the BaP group co-enriched the fat digestion and absorption pathway and fatty acid biosynthesis pathway, and the Mix group co-enriched the iron metabolism pathway, unsaturated fatty acid biosynthesis pathway and steroid biosynthesis pathway. As a result, pathways related to lipid metabolism, iron metabolism, and kidney injury were enriched for differential genes and differential metabolites. This further confirms that PS and BaP together cause renal damage, leading to a dysregulation of renal lipid metabolism homoeostasis and iron metamorphosis.

Fig. 3. The combined analysis of the transcriptome and metabolome.

a The transcriptome and metabolome of the kidney were co-analysed with KEGG in Con and PS groups. b The transcriptome and metabolome of the kidney were co-analysed with KEGG in Con and BaP groups. c The transcriptome and metabolome of the kidney were co-analysed with KEGG in Con and Mix groups, n = 4 for transcriptome and n = 6 for metabolome.

Kidney ferroptosis caused by MPs and BaP

To further verify the effect of PS and BaP on renal ferroptosis, we performed in vivo validation experiments on mouse kidney using ferrostatin-1 (Fer-1 group) as the inhibitor group in the Mix group to verify whether the renal injury was due to lipid deposition induced iron-dependent cell death. The oil red O results showed that there was a large amount of lipid deposition only in the Mix group and that renal lipid deposition was significantly improved by the addition of Fer-1 (Fig. 4a). The TEM results showed that in the control group, the mitochondrial cristae structure was clearly visible. The mitochondria were arranged in an orderly manner without the presence of lipid droplets. The PS group showed mitochondrial swelling, slight damage to mitochondrial cristae structure and the presence of some lipid droplets. BaP group showed broken mitochondrial cristae, swollen and vacuolated mitochondria. In the Mix group, mitochondria showed significant swelling, deposition of large numbers of lipid droplets, loss of mitochondrial cristae structure, mitochondrial vacuolisation and enhanced mitochondrial membrane density (Fig. 4b). In the Fer-1 group, mitochondria were swollen and vacuoles were restored compared to Mix. A small amount of lipid droplet deposition remained.

Fig. 4. Ferroptosis occurs in the kidney.

a Lipid deposition in the kidney. 100× scale bar is 200 μm, 200× scale bar is 100 μm, 400× scale bar is 40 μm, n = 4. b Changes in the organelles of the kidney. Mitochondrial swelling is indicated by red arrows, damage to mitochondrial cristae structure by yellow arrows, lipid droplet deposition by black arrows, mitochondrial swelling and vacuolisation by yellow triangles, and increased mitochondrial membrane density by red triangles, n = 3. c Glutathione levels in the kidney were altered, n = 5. d Levels of lipid peroxidation in the kidney, n = 5. e, f The levels of protein expression for ACSL4 and LPCAT3 were examined, n = 4. g Expression of the key genes for ferroptosis, n = 6. All data were expressed as the mean ± SD. All differences identified via one-way ANOVA followed by the Tukey post hoc test. *indicates p < 0.05 compared with the control (0 μg/mL) group.

Biomarkers of lipid peroxidation and ferroptosis in the kidney were then examined. It was found that T-GSH was significantly decreased in the Mix group. Oxidised glutathione (GSSG) was significantly increased and reduced glutathione (GSH) was significantly decreased in the Mix group. There were no significant changes in the PS and BaP groups. GPX activity was significantly suppressed in the PS, BaP and Mix groups (Fig. 4c). Lipid peroxidase (LPO) and malondialdehyde (MDA) in the kidney showed a significant increase in the PS and Mix groups, but there was no significant change in the BaP group. Although the Fer-1 group was still higher than the control group, the difference with the control group was not statistically significant (Fig. 4d). The level of Fe²⁺ was also significantly increased in the kidneys of the Mix group, whereas the level of Fe²⁺ was reduced after the addition of Fer-1. From the above results, it was clear that it was only in the Mix group that severe lipid peroxidation and, ultimately, ferroptosis occurred in the kidneys. Therefore, for the validation of ferroptosis biomarkers, we selected kidneys from the control, Mix and Fer-1 groups. We found that protein expression of ACSL4 and LPCAT3 was significantly upregulated (Fig. 4e, f, Supplementary Fig. 11), whereas gene expression of SLC7A11, SLC3A2 and GPX4 was significantly downregulated (Fig. 4g). This indicated that the antioxidant system of the organism was still active with high expression of GPX4 and low expression of System Xc-. Furthermore, ACSL4, LPCAT3, ALOX15, SLC39A8 and SLC39A14 genes were markedly increased. This indicates that the uptake of PS and BaP does indeed lead to renal ferroptosis. We then carried out a joint analysis of the key genes and the key metabolites. The results showed that, especially in the Mix group, lipid metabolites such as arachidonic acid and epinephrine were significantly and positively correlated with ACSL4, LPCAT3, ALOX15, SLC39A, SLC39A14, whereas antioxidant-related metabolites such as ascorbic acid and vitamin A were significantly and negatively correlated with GPX4, SLC7A11 and SLC3A2 (Supplementary Fig. 5a–c).

PS and BaP induce lipid peroxidation in kidney cells

The question of whether the damage to the kidneys was direct was also a matter for investigation. We selected HK-2 cells as the cell line for in vitro validation. According to the results of the CCK8 experiments, the 200 μg/mL Mix and the 1 μmol/mL BaP groups were selected for the subsequent validation assay (Fig. 5a, b). The Mix group significantly inhibited the cell viability at 48 h (Fig. 5c), whereas the group Fer-1 had a significant recovery effect. The levels of reactive oxygen species (ROS), iron ions and lipid peroxidation of the HK-2 cells were also measured. Interestingly, lipid peroxidation and ROS were significantly increased in HK-2 (Fig. 5d–f), but there was no significant change in Fe²⁺. In addition, the levels of ROS and lipid peroxidation were also significantly improved in the Fer-1 group, but they were still significantly different from those in the control group. Curiously, the proteins ACSL4 and LPCAT3 did not show a significant increase in the Mix group (Fig. 5g, h, Supplementary Fig. 10). There was also no statistical significance between the mRNA levels of ACSL4, LPCAT3, ALOX15, SLC39A8 and SLC39A14 in the Mix group. In HK-2 cells, GPX4, SLC7A11 and SLC3A2 were significantly expressed. This may be due to the fact that lipid peroxidation caused by oxidative stress still occurred in HK-2 when stimulated with PS and BaP, although no direct ferroptosis occurred (Fig. 5i).

Fig. 5. Induction of lipid peroxidation by PS and BaP in HK-2 cells.

a HK-2 cell viability treated with PS for 48 h, n = 6. b HK-2 cell viability treated with BaP for 48 h, n = 6. c HK-2 cell viability treated with PS, BaP and Fer-1 for 48 h, n = 5. d ROS assay for HK-2 cells for 48 h, n = 6. e, f Assays for lipid peroxidation and Fe²⁺ were conducted on HK-2 cells. The C11-BODIPY molecule exhibits a reduction in red fluorescence and an increase in green fluorescence when oxidised. The scare bar is 50 μm, n = 6. g, h The levels of protein expression for ACSL4 and LPCAT3 were examined, n = 3. i Expression of the key genes for ferroptosis, n = 6. Differences identified via one-way ANOVA followed by the Tukey post hoc test. Gene expression data were analysed using unpaired t-tests.* indicates p < 0.05 compared with the control (0 μg/mL) group.

Impairment of intestinal barrier by PS and BaP leading to inhibition of intestinal tight junction protein expression

Numerous studies have shown that oral administration of microplastics causes damage to the digestive tract²³. While our previous study found evidence of renal ferroptosis, in vitro experiments showed that PS and BaP were unable to directly induce ferroptosis in renal cells. Indirect damage to the kidney via the gut-kidney axis pathway may explain this. To further understand the routes and mechanisms by which oral PS and BaP cause kidney damage, we studied intestinal permeability and intestinal barrier damage. Results showed that serum LPS, DAO and FITC levels were significantly increased in PS and Mix groups (Fig. 6a–c). Of these, the damage was significantly higher in the Mix group than in all other groups. The expression of claudin-1, occludin and ZO-1 genes in the intestine was significantly suppressed in the Mix group (Fig. 6d–f). The immunofluorescence results also showed that the expression of all three of the above-mentioned tight junction proteins was significantly reduced in the small intestinal tissues of the Mix group (Fig. 6g, h).

Fig. 6. The intestinal barrier can be damaged by PS and BaP.

a, b, c Indicators of the permeability of the intestine, n = 5. The scare bar is 200 μm. d, e, f Expression of genes related to tight junctions in the intestine, n = 6. g, h Localisation and quantification of intestinal tight junction proteins, n = 4. i Organisational changes in intestine. Yellow arrows indicate slight jagged damage to the intestinal villi, red arrows indicate cellular vacuoles, and red triangles indicate severe erosion of the intestinal villi, n = 4. j Changes in the ultrastructure of the intestinal tract. Yellow arrows indicate broken mitochondrial cristae, red arrows indicate swollen mitochondria, yellow triangles indicate dissolved mitochondria, red triangles indicate loss of mitochondrial cristae, n = 3. Differences identified via one-way ANOVA followed by the Tukey post hoc test. Gene expression data were analysed using unpaired t-tests.* indicates p < 0.05 compared with the control (0 μg/mL) group.

Small intestine histological and organelle damage was assessed. HE staining showed intact intestinal mucosa and well-aligned intestinal villi in the normal control group, with no structural damage or inflammatory cells. The intestinal mucosa of PS and BaP groups exposed alone was morphologically intact, with slight villous loss and vacuolisation. In contrast, the Mix group showed severe villous erosion. The damage to the upper part of the intestinal villi was particularly severe and accompanied by a large number of vacuolation phenomena (Fig. 6i). Transmission electron microscopy results showed that the PS, BaP and Mix groups all showed different degrees of intestinal mitochondrial cristae disruption and swelling, and the Mix group also showed mitochondrial lysis and cristae disruption (Fig. 6j). We found a significant increase in intestinal permeability and damage to the intestinal barrier after oral administration of a composite form of PS and BaP, which was also accompanied by the appearance of large numbers of inflammatory cells in the intestinal tissue. These results suggest that co-stimulation of PS and BaP may damage the intestinal barrier by inducing intestinal inflammation in mice, ultimately leading to severe leaky gut²⁴.

By disrupting the microbial composition of the gut, PS and BaP contribute to lipid peroxidation

In order to further understand the specific mechanisms of the effects on the kidney after intestinal injury, we investigated and analysed the changes in the diversity and the flora of the intestinal micro-organisms. It was found that polystyrene significantly reduced bacterial diversity, as evidenced by the significantly lower Chao and Shannon indices of intestinal α-diversity in the PS and Mix groups (Supplementary Fig. 6). It is interesting to note that the PS group had a more pronounced effect on the gut flora than the Mix group. However, the results of PCA and PCoA showed a significant difference between the microbial communities of the control and Mix groups and a higher similarity to the microbial communities of the BaP and PS groups exposed to BaP and PS alone. At the phylum level, the abundance of Bacteroidota, Firmicutes, Desulfobacterota and Campylobataer was significantly altered (Fig. 7a). At the order level, PS and BaP mainly altered the abundance of Oscillospiraces, Saccharimonadales and Desulfovibrionales (Fig. 7a, b). It showed that the PS and BaP groups disrupted the intestinal barrier through a decrease in beneficial bacteria and an increase in harmful bacteria, leading to intestinal lipid accumulation. Furthermore, Firmicutes/Bacteroidota was significantly increased in the Mix group (Fig. 7c–e), also a biomarker of lipid dysregulation²⁵. The levels of GSH, GPX, MDA, LPO and Fe²⁺ were then examined in the gut. T-GSH was significantly reduced in the Mix group, whereas GSSG was significantly increased and GSH was significantly reduced in the Mix group only (Fig. 7f–h). GPX levels were significantly inhibited in all experimental groups (Fig. 7i). MDA and LPO were both in the PS group and Mix group significantly higher than the control levels (Fig. 7j, k). In addition, there was no significant change in Fe²⁺ levels in any of the groups (Fig. 7l). This indicates that a significant amount of lipid peroxidation occurred in the small intestine. In order to further confirm whether the lipid peroxidation that occurs in the small intestine leads to ferroptosis. We examined the relevant genes and proteins. Interestingly, there was no statistical significance in the protein expression of ACSL4 and LPCAT3 in the control and mix groups (Fig. 7m, n, Supplementary Fig. 9). Furthermore, the GPX4 and SLC39A8 genes were significantly upregulated, which could be due to the fact that oxidative stress leads to iron accumulation within the body, but not at a level that would lead to intestinal ferroptosis. In contrast, there were no significant changes in the expression of other genes (Fig. 7o). This suggests that it is not ferroptosis, but rather intestinal lipid peroxidation that is induced by PS and BaP.

Fig. 7. PS and BaP can disturb intestinal microbes, resulting in lipid peroxidation.

The genes related to ferroptosis are associated with lipid metabolites. a, b Changes in the abundance of microorganisms, n = 6. c, d, e Analysis of changes in the abundance of the Bacteroidota and Firmicutes, n = 6. f, g, h Dynamic changes in intestinal glutathione levels, n = 4. i, j, k, l Altered levels of Fe²⁺ and intestinal lipid peroxidation, n = 4. m, n In ferroptosis, the key protein is altered, n = 3. o Expression of the key genes for ferroptosis, n = 6. Differences identified via one-way ANOVA followed by the Tukey post hoc test. Gene expression data were analysed using unpaired t-tests.* indicates p < 0.05 compared with the control (0 μg/mL) group.

PS and BaP induce cellular ferroptosis in small intestine epithelial cells

We also wanted to know why ferroptosis occurs in the kidney but not in the intestine. We carried out further in vitro validation using the small intestinal epithelial cell line HIEC-6. In the following experiments, we were surprised to find that the Mix group stimulated Fe²⁺ accumulation and lipid peroxidation in HIEC-6 (Fig. 8a, b, Supplementary Fig. 12). Whereas in Fer-1 group, PS and BaP induced lipid peroxidation and Fe²⁺ accumulation were significantly ameliorated. We then examined the protein and gene expression of HICE-6. The protein levels of ACSL4, LPCAT3 were significantly increased (Fig. 8c, d). In addition, GPX4, SLC7A11 and SLC3A2 genes were significantly downregulated, and gene expression of ACSL4, LPCAT3, ALOX15, SLC39A8 and SLC39A14 were all significantly higher than that of control (Fig. 8e). This provides strong evidence that PS and BaP directly induce ferroptosis in intestinal epithelial cells, but do not induce ferroptosis in the intestine in vivo. This may be related to the function of the small intestine. Direct ferroptosis was not induced in the small intestine, which is an important organ of immunity and would be involved in the mucosal immunity of the body²⁶.

Fig. 8. Ferroptosis occurs in HIEC-6 cells.

a Assays for lipid peroxidation and Fe²⁺ were conducted on HIEC-6 cells. The C11-BODIPY molecule exhibits a reduction in red fluorescence and an increase in green fluorescence when oxidised. The scare bar is 50 μm, n = 4. b The levels of protein expression for ACSL4 and LPCAT3 were examined, n = 3. c Expression of the key genes for ferroptosis, n = 6. d Examination of the blood and the fatty acid content of the cells, n = 3. Differences identified via one-way ANOVA followed by the Tukey post hoc test. Gene expression data were analysed using unpaired t-tests.* indicates p < 0.05 compared with the control (0 μg/mL) group.

Intestinal metabolites lead to ferroptosis of the kidney via the blood circulation

In this study, it was found that in mice, ferroptosis occurred in the kidney, but not in the intestine. The opposite was found in in vitro experiments. We hypothesise that harmful metabolites and bacteria excreted from the gut via the gut-kidney axis damage the kidneys. Metabolites in the blood, tissues and two cell types of mice were studied. Elevated levels of several polyunsaturated fatty acids were detected in the blood of mice in the mixed group compared with the control group. Arachidonic acid was significantly higher than the other fractions (Fig. 8d), and similar results were observed in HIEC-6 cells. In addition, arachidonic acid levels were also significantly elevated in the Mix group in intestinal and kidney tissues (Supplementary Fig. 7), however, the metabolites did not change significantly in HK-2 cells (Fig. 8d). To further confirm the intestinal and renal correlation, we performed a joint analysis of intestinal flora and renal metabolites, which showed that the bacterial communities significantly and positively correlated with arachidonic acid in both control and experimental groups included pathogenic bacteria such as Bacteroidales, Burkholderiales and other pathogens, whereas beneficial bacteria such as Lactobacillales showed a significant decreasing trend (Supplementary Fig. 8). The above results also support our hypothesis.

Discussion

In recent years, more and more plastic waste has been released into the environment, carrying other harmful substances into the human body^27–29. Studies have shown that oral ingestion of contaminants induces dysbiosis of the gut microflora and disorders of lipid metabolism by damaging the intestinal barrier^30,31. The intestine plays a crucial role as the first barrier to exposure to exogenous contaminants in the oral cavity. However, the pathways and specific mechanisms by which renal injury occurs across the intestinal barrier are still largely unknown. In the present study, we report on the pathways and mechanisms by which PS and BaP induce renal injury through the gut-kidney axis. We found that PS and BaP damage the intestinal vascular barrier by inhibiting the expression of intestinal tight junction proteins ZO-1, claudin-1 and occludin, altering intestinal microbial community abundance and disrupting lipid metabolism. The severe damage to the intestinal barrier results in thinning of the intestinal wall, leading to an increase in intestinal metabolites, including polyunsaturated fatty acids, which translocate to the kidneys via the gut-kidney axis. The kidney is damaged by activating ACSL4/LPCAT3. This eventually results in ferroptosis, which is caused by abnormalities in renal lipid and iron metabolism.

Previous studies have shown that, with bottled water being the main source of microplastics in the body, the average human ingests ~100–5000 mg of plastic per week³². Based on an average body weight of 60 kg, this equates to ~0.24–11.9 mg/kg/day per person. In this study, we performed long-term environmental concentrations of PS and BaP administered via drinking water, with a mouse drinking 5 mL of water per day³³, and we had initial concentrations of PS and BaP in the water of 4 mg/L and 4 μg/L, respectively, with final concentrations entering the body via the mouth of 1 mg/kg/day and 1 μg/kg/d, respectively. This is comparable to the low levels and patterns of plastic exposure observed in humans.

Orally administered MPs and BaP come into direct contact with the apical portion of intestinal epithelial cells and are absorbed in small amounts by intestinal villus cells³⁴. The unabsorbed MPs, which are broken into irregular fragments by gastric juices, cause mechanical damage to the gastrointestinal tract³⁵. This leads to thinning of the intestinal wall and damage to the intestinal barrier by inhibiting the expression of intestinal tight junction proteins³⁶. This ultimately induces the development of a “leaky gut” effect, which promotes intestinal inflammation. Damage to the intestinal vascular barrier and leaky gut leads directly to the entry of MPs and BaP across the intestinal vascular barrier into the circulation, increasing the risk of systemic inflammation³⁷.

In this study, the expression of claudin-1, occludin and ZO-1 proteins and genes was significantly suppressed in the Mix group, while serum LPS and DAO and FITC were significantly higher than in the control group. This is a good indication that the intestinal barrier was significantly damaged. Microplastics carry environmental contaminants into the body and BaP exposure results in increased sensitivity and tissue destruction in the intestine³⁸. We also observed histological damage in the small intestine in our HE and TEM results. This included phenomena such as eroded villus and swollen mitochondria. Research has shown that gut flora is closely associated with gut health³⁹. In our experiments, the abundance of harmful flora in the Mixed group was significantly higher than in the control group and the group exposed to PS and BaP alone. The Firmicutes and the Bacteroides, as the two most abundant phyla in the gut flora, were significantly associated with lipid accumulation in mice⁴⁰. Furthermore, the ratio of former to latter is an important biomarker of lipid accumulation⁴¹. The Mix group had an increase in Firmicutes and a decrease in Bacteroides, and the Firmicutes/Bacteroides was significantly higher than in all other groups. The accumulation of lipids is a direct cause of intestinal inflammation and permeability. At the phylum level, Desulfobacterota and Campylobacter were significantly increased, in addition to changes in Firmicutes and Bacteroides. It has been shown that Desulfobacterota, as a sulfate anaerobe, induces intestinal inflammation and leads to pathological changes in intestinal tissue structure by disrupting the intestinal barrier^42–44. As a pathogenic bacterium, Campylobacter also causes significant damage to intestinal tissue and induces systemic inflammation⁴⁵. At the family level, Muribaculaceae produce propionate by fermenting indigestible carbohydrates, which is absorbed by intestinal epithelial cells and used as an energy source. SCFAs improve mucosal barrier function by increasing intestinal mucin and immunoglobulin A production, and propionate activates anti-inflammatory function by regulating T-cell homoeostasis⁴². Elevated levels of Rikenellaceae have been associated with lipid overaccumulation⁴³. Thus, PS and BaP treatment may reduce propionate and SCFAs production by decreasing the abundance of Muribaculaceae, further disrupting the intestinal barrier and inducing the onset of intestinal inflammation and lipid peroxidation. In addition, the decrease in Ruminococcaceae may be due to the fact that Ruminococcaceae is a major consumer of plant polysaccharides, which are drastically reduced in patients with lipid peroxidation⁴⁴. Lactobacillaceae is considered to be a key participant in host metabolic homoeostasis⁴⁵, and the Mix group showed a significant increase, indicating that mice under the combined effect of PS and BaP metabolic homoeostasis imbalance under PS and BaP. Gut microbiota dysbiosis has been shown to be closely associated with lipid dysbiosis⁴⁶. Lipid metabolism and lipid peroxidation were also associated with most of the microflora whose abundance we found to be altered⁴⁶.

It has been reported that oral administration of microplastics and PAHs leads to intestinal activation of antioxidant defenses, including a dramatic increase in antioxidant levels of GPX, GSH to scavenge excess reactive oxygen radicals, lipid peroxides and so on, generated through oxidative stress, such as superoxide anion, hydrogen peroxide, MDA and LPO. However, when the body is under prolonged stress, antioxidant levels are depleted. This is indicated by a decrease in GPX, GSH. Lipid peroxidation occurs when the body is unable to continue to produce antioxidant products. Lipid peroxidation has a negative correlation with health⁴⁷. In the present study, treatment with PS and BaP resulted in an imbalance of intestinal T-GSH homoeostasis. MDA, LPO and GSSG were significantly elevated. GPX, and reduced GSH was significantly decreased. This was an indication that the intestinal antioxidant capacity was suppressed and the levels of intestinal lipid peroxidation were elevated. Fe²⁺ is taken up by SLC39A8s and SLC39A14, and excess Fe²⁺ undergoes a Fenton’s reaction-induced ROS overload, which ultimately contributes to the development of ferroptosis⁴⁸. However, while significant lipid peroxidation-induced ferroptosis occurred in HIEC-6, we found that typical ferroptosis did not occur in intestinal tissues. This may be due to the complex environment of the intestine in vivo. Intestines are important immune barriers that are activated to remove excess toxins when threatened⁴⁹. While the intestinal epithelial cells are structurally simple, exogenous stimuli directly reduce the viability of the cells, induce cellular lipid peroxidation and Fe²⁺ accumulation, and ultimately lead to cell death.

Changes in the intestinal flora lead to reduced levels of SCFAs and increased production of unsaturated fatty acids⁵⁰. An increase in polyunsaturated fatty acids not only induces a local inflammatory response in the body, but also leads to severe lipid peroxidation and ultimately ferroptosis⁵¹. In this study, PS and BaP significantly elevated gut metabolites including arachidonic acid and linoleic acid. Among these, arachidonic acid was found to play a key role in the development of ferroptosis. Kidney injury has been shown to occur via a gut-kidney axis⁵². Arachidonic acid was found to be produced at high levels in the gut, kidneys, cells and blood of the mixed group of subjects. The high production of this fatty acid can exacerbate intestinal inflammation and damage the intestinal barrier, leading to leaky gut, where arachidonic acid passes through the intestinal vascular barrier and into the bloodstream, eventually reaching the kidneys and eventually causing kidney damage, which can lead to ferroptosis¹⁷. All this suggests that the Mix group has significant synergistic kidney toxicity.

The kidney is one of the most important sites of metabolism, and the deposition of lipids is an important manifestation of tissue damage⁵³. When the body is exposed to exogenous stimuli, the antioxidant system is activated in response to ROS generated by the external stimuli to protect against reactive oxygen species⁵⁴. When treated with PS and BaP alone, the renal antioxidant system is activated in response to increased ROS levels due to exogenous stimulation. However, excessive ROS production leads to depletion of the antioxidant system and causes antioxidant levels to decrease significantly⁵⁴. In the in vivo experiments, renal GSH levels were markedly depressed, with marked increases in MDA, LPO and Fe²⁺ levels. This is because intestinal metabolism of arachidonic acid, linoleic acid and other polyunsaturated fatty acids enter the kidney with the circulation and directly stimulate the kidney to generate more reactive oxygen species (ROS). The excess ROS directly inhibit the antioxidant system, resulting in a significant decrease in GPX and significant downregulation of GPX4. The ROS that are not scavenged accumulate in the kidney, causing lipid peroxidation and inducing the production of more unsaturated fatty acid metabolites⁵⁵. It has been shown that polyunsaturated fatty acids, and in particular arachidonic acid, are a key substrate in the development of ferroptosis⁵⁶. We have also observed that arachidonic acid activates a high level of ACSL4/LPCAT3 expression, which in turn induces a significant level of ALOX15 expression and causes an increase in PE-AA-O-OH. In this experiment, Fe²⁺ is also taken up into the body through the action of SLC39A8 and SLC39A14, and the accumulation of Fe²⁺ is one of the most important reasons for the induction of the appearance of ferroptosis. On the other hand, treatment with PS and BaP leads to inhibition of the Xc^- system, which reduces cystine uptake in vivo, leading to reduced substrate for glutathione synthesis. This results in decreased T-GSH levels, increased GSSG and depleted GPX4, leading to GPX4 depletion. In addition, TEM showed disappearance of mitochondrial cristae, more lipid droplets and increased mitochondrial membrane density in the kidney. All of these are important features and biomarkers for ferroptosis to occur⁵⁷. Interestingly, despite significant lipid peroxidation, no increase in Fe²⁺ was observed in HK-2 cells. GPX4 showed a trend towards a significant increase. ACSL4, LPCAT3, ALOX15, SLC39A8, SLC39A14 also showed an opposite trend to renal tissue. The metabolites from HK-2 also did not show an increase in arachidonic acid. This suggests that in the in vivo environment, renal injury results from an excess of arachidonic acid metabolised in the gut and not directly from the toxicity of PS and BaP. The direct nephrotoxicity of PS and BaP results from the activation of the antioxidant system by exogenous stimuli, leading to oxidative stress and the generation of excess ROS. Therefore, ferroptosis of the kidney by PS and BaP is not direct.

In conclusion, the present study showed that the treatment with PS and BaP alone resulted in lipid peroxidation in the intestine and kidney, and the kidney also showed a small amount of lipid deposition in the group treated with PS alone. However, co-treatment with PS and BaP altered the intestinal flora of the mice and induced high production of arachidonic acid in the intestine. This is mainly achieved through the high expression of ALOX15, which is activated by ACSL4/LPCAT3 upregulation. Arachidonic acid plays a role in kidney injury through the gut-kidney axis transfer, ultimately leading to kidney ferroptosis via the production of PE-AA-O-OH. This study highlights the combined toxicity of PS and BaP in mammals and uncovers the specific pathways and mechanisms by which exogenous pollutants harm the gut-kidney axis in animals.

Materials and methods

Characterisation of MPs

PS suspended in water was purchased from Tianjin Besler Chromatography Technology Research Centre (Tianjin, China), and BaP (99.99% purity) was supplied by Sigma Corporation (USA). According to the manufacturer, the average diameter of PS microspheres was 100 nm. BaP was dissolved with DMSO, in which the BaP group for in vivo experiments was diluted with ultrapure water, and the BaP group for in vitro experiments was diluted with basal medium. In addition, diluted BaP was co-incubated with PS for 48 h at room temperature⁵⁸, which was performed on an oscillator. PS was confirmed for particle size and morphology by scanning electron microscopy(SEM). The PS surface zeta potential was detected using a Malvern Zetasizer Nano ZSE. The microplastic particles were dispersed in deionised water and treated with ultrasonic vibration for 30 min to allow complete suspension and dispersion before use.

Animal experiments

Five-week-old male Kunming mice (20 ± 2 g) were obtained from the Guangdong Experimental Animal Centre (Guangzhou, China) and maintained in a pathogen-free environment at 25 ± 2 °C, 50 ± 5% humidity, with a 12 h/12 h light/dark cycle. After a 1-week acclimatisation period, the mice were randomly divided into five groups: negative control (Con group) which was treated with pure water, solvent control (DMSO control, solvent control group), 400 μg/mL PS (PS group), 4 μmol/L BaP (BaP group), 400 μg/mL PS+BaP (Mix group), 40 mice were kept as biological replicates in each group, and 5 mice were kept in each cage. During the experiment, the drug was administered through drinking water that had been sonicated for 30 min before administration. Fresh drinking water was replaced daily for 56 days of exposure. Additionally, a ferroptosis inhibitor group of 400 μg/mL Mix + 5 mg/kg Ferrostatin-1 (Fer-1 group) was established based on the transcriptome and metabolome results⁵⁹. The drug was administered through intraperitoneal injection twice a week. Blood, intestines, kidney and intestinal contents were collected and mice were euthanised with CO₂ 24 h after the last day of the treatment. A portion of the intestine and kidney was fixed in 4% paraformaldehyde and 2.5% glutaraldehyde for histological analysis. Another portion was used for rapid freezing in liquid nitrogen and then stored at −80 °C for future analysis. The remaining samples were stored at −80 °C for backup purposes. We have complied with all relevant ethical regulations for animal use during the experiments. The Animal Care Committee of Haikou People’s Hospital approved all animal experimental procedures under the approval number SC20230241.

HE staining

Kidney and intestine tissues were prepared as wax blocks and cut into 5 μm sections, dehydrated through various concentrations of anhydrous ethanol and cleared with xylene, stained with H&E and examined by light microscopy.

Transmission electron microscopy

The tissues were dehydrated in 2.5% glutaraldehyde, fixed in cold ethanol and acetone, embedded and processed. Subsequently, the tissue blocks were cut into 90 nm sections using an ultrathin microtome. The sections were then stained with lead citrate and uranyl acetate before being observed using transmission electron microscopy.

Oil Red O staining

The kidney tissue was fixed and then sectioned before being treated with 60% isopropyl alcohol. It was subsequently stained with oil red O for 30 min, washed with PBS, and photographed using a light microscope for documentation purposes.

Biochemical indicators

Cre^60–63, BUN^60–63, UA^62,63, Albumin⁶¹, NO^64,65, LPS⁶⁶, Glutathione (GSH) (Nanjing Jiancheng, China), Malondialdehyde (MDA) (Nanjing Jiancheng, China), Lipid Peroxidase (LPO) (Nanjing Jiancheng, China), Total protein content (Nanjing Jiancheng, China), Fe²⁺ (Biosharp, China) were detected by kits.

Determination of intestinal permeability

An in vivo intestinal permeability assay using fluorescein isothiocyanate (FITC)-labelled dextran (MW4000, MCE, USA) was performed one day before the experiment ended. Briefly, mice were fasted for 4 h and then gavaged with FITC-dextran at a dose of 0.6 mg/g body weight. After 2 h, blood samples were collected from the inferior canthus and serum was collected. The fluorescence intensity in the serum was measured using a BIOTEK fluorescence spectrophotometer (USA).

Immunofluorescence

The intestine sections, which were 4 μm in size, were closed using 5% bovine serum albumin (BSA) after being incubated with diluted primary antibodies against claudin-1(1:3000, GB15032, China), occludin(1:3000, GB111401, China), and ZO-1(1:2000, GB111402, China) at 4 °C overnight. Subsequently, the sections were incubated with fluorescently labelled secondary antibody pairs for 1 h at 37°C. The nuclei were stained with DAPI under photoprotective conditions, and images were captured using a fluorescence microscope. The images were analysed using ImageJ software.

Sequencing of 16s rRNA

The fresh intestinal contents were kept refrigerated at -80°C and sent to NovoMajic. 16S rRNA analysis was used to determine the bacterial abundance in the intestine of mice after different treatments, using primers 151F (5’-TACGGRAGGCAGCAG-3’) and 806R (5’-AGGGTATCTAATCCT-3’), which amplified the V3–V4 variable region of the 16S rRNA gene for bacterial diversity analysis. The extracted genomic DNA was denatured at 98 °C for 1 min and 30 cycles were performed at 98 °C for 10 s, 50 °C for 30 s and 72 °C for 30 s. Finally, it was held at 72 °C for 5 min. The PCR products were detected by electrophoresis in 2% agar gel, and the PCR products that met the conditions were subjected to magnetic bead purification, and the target samples were collected for library construction and quantitative analysis by qPCR using NovaSeq6000 up-sequencing.

After sequencing, the data was first split according to the barcode sequence and PCR amplification primers to obtain the target sequence, and then the reverse primer sequence was matched using Cutadapt software. The spliced raw tags were filtered using fastp software (version 0.23.1) to obtain clean tags, and the filtered clean tags were processed to obtain effective tags, which were then processed according to DADT in QIIME 2 (version QIIME2- 202202) in the DADA software. 202202) in the DADA2 noise reduction module to obtain the final ASVs (Amolicion Sequence Variants) and feature table. Using the unweighted Unifrac distance matrix calculated in R, the unweighted Unifrac Principal Coordinate Analysis (PCoA) was performed to evaluate the beta diversity of the samples. Based on the R package, the ANOVA/Kruskal Wallis/T test statistical algorithm was used for the analysis of variance. Species abundance spectra were analysed for differences using LEfSe.

RNA-Seq

Transcriptome assays were performed on kidney tissues from control, PS, BaP, and Mix groups, each with 4 replicates. Transcriptome analyses were performed at Beijing Novogene Technology Co. RNA was extracted and used for library construction, followed by on-board testing of samples that met quality requirements. From RNA sample extraction to final data acquisition, including sample testing, library construction and sequencing, each step was performed qualitatively and quantitatively. Sequences were filtered by removing reads with splice contamination, removing reads with more than 10% unknown reads (N > 10%) and removing low-quality nucleotides (base mass value less than 5) accounting for more than 50% of the read length.

Novogene (China) used seven databases to achieve comprehensive annotation of gene functions. The software and parameters used in each database were NT: NCBI. e-value threshold was 1e^-5 (the first 10 comparison results were shown for each Unigene). nr, SwissProt, KOG, etc. Threshold was 1e^-5 (the first 10 comparison results were shown for each Unigene). Pfam (protein structural domain prediction): hmmer package, hmmscan. e-value threshold is 0.01; GO (protein annotation results based on NR and Pfam): blast2GO and novogene scripts. e-value threshold is 1e⁻⁶.

Untargeted metabolomics by LC-MS

NovoMajic did an analysis of kidney tissue by means of metabolomics. The collected LC-MS data were pre-processed using XCMS software. Raw data files were converted to mz XML format and further processed using the XCMS, CAMERA and meta X toolboxes included in the R software. Secondary mass spectrometry information was matched and identified with internal standard databases and annotated with databases such as HMDB and KEGG. UPLC-Q-TOF/MS was used to collect serum sample data, and Principal Component Analysis (PCA), Supervised Partial Least Squares Discriminant Analysis (PLS-DA) and Orthogonal Projection of Latent Structures (OPLS-DA) were used to observe the cluster separation of each group after downscaling, and the results of the permutation test were used to determine the predictive ability of the model and whether the model was overfitted. Significantly different metabolites between the normal group and each of the modelled groups were screened with criteria of variable predictive importance (VIP) > 1 and p < 0.05. Changes in weekly differential metabolites were then analysed using heat maps and pathways were analysed using Metabo Analyst software version 5.0 (https://www.metaboanalyst.ca), based on which key pathways were screened for key differential metabolites for subsequent analyses.

Cell culture and cell viability assays

The human renal tubular epithelial cells (HK-2) and the human small intestinal epithelial cell line (HIEC-6) were purchased from Procell and Zhejiang Meisen Co, respectively. The HK-2 cells were cultured in MEM medium supplemented with 10% foetal bovine serum (v/v), 1% penicillin (100 U/mL) and streptomycin (100 mg/mL) (Procell, China), while the HIEC-6 cells were cultured in DMEM/F12 medium supplemented with 10% foetal bovine serum (v/v), 1% penicillin (100 U/mL) and streptomycin (100 mg/mL) (Zhejiang Mason Co. Ltd.). Both cell types were cultured in DMEM/F12 medium (Zhejian, China) in a humid atmosphere containing 5% CO₂.

Cell cultures containing concentrations of 0, 12.5, 25, 50, 100 and 200 μg/mL of PS and Mix, and cell cultures containing 0 (DMSO control), 0.01, 0.1, 1 and 10 μmol of BaP were added to the cells and cultured for 48 h. Cell viability was assessed using the CCK-8 kit (Tongren, Japan).

Detection of Reactive Oxygen Species (ROS)

Both types of cells were exposed to 200 μg/mL PS and Mix and 1 μmol BaP for 48 h. After removal of the medium, the cells were washed three times with PBS and incubated with DCFH-DA (10 μM) (Nanjing Jiancheng, China) at 37 °C for 30 min. The fluorescence intensity of the ROS in the cells was detected with the use of a fluorescent enzyme marker and a fluorescence light microscope.

Assay for cellular lipid peroxidation

HK-2 and HIEC-6 cells were inoculated into 24-well plates at a density of 1 × 10⁵. The group containing PS+BaP (Mix) at a concentration of 0, 200 μg/mL and 2 μmol/L Fer-1 was treated with drugs 24 h later. After 48 h of treatment, the cells were stained with C11-BODIPY (Invitrogen, D3861, USA) at 37 °C for 0.5 h. Staining was performed for 30 min. After washing with PBS, lipid peroxidation and antioxidant levels were observed using a confocal instrument (Olympus, Japan) with emission wavelengths of 581 and 500 nm.

Assay for cellular iron ion

HK-2 and HIEC-6 cells were inoculated at a density of 1 × 10⁵ in 24-well plates and incubated overnight at 37 °C. Following drug treatment, the cells were washed twice with PBS and stained with 10 μg/mL FeRhoNox-1 (MCE, USA) for 0.5 h. After washing with PBS, the stained cells were incubated in serum-free medium and then imaged using fluorescence microscopy (Olympus, Japan) with emission wavelengths of 540 and 575 nm. Imaging was performed to visualise the nuclei and Fe²⁺ analysis, and quantitative fluorescence analysis was conducted using ImageJ.

Real-time fluorescence quantification

Total RNA was extracted from kidney and intestinal tissues, as well as HK-2 and HIEC-6 cells, using the Trizol method (Invitrogen, USA). The RNA was then reverse transcribed into cDNA using a kit. mRNA expression levels were detected using the LightCycler 480 (Roche, Switzerland). Primer sequences were designed online using (https://www.sangon.com/login), and are shown in Supplementary Table 2. mRNA relative expression was determined by 2^−ΔΔCt, with β-actin used as an internal reference gene.

Western blot

Proteins were extracted from cells and tissues using RIPA lysate containing protease inhibitors and phosphatase inhibitors. The protein concentration was determined using the BCA method (Biyuntian, China). The protein samples were separated by SDS-PAGE gel electrophoresis, transferred to PVDF membranes, and incubated with primary antibodies for 12 h. The primary antibodies used were ACSL4(1:10000, 22401, China), LPCAT3 (1:6000, 67882, China), and β-actin (1:3000, WL01372, China). The PVDF membrane was then incubated with an HRP-coupled secondary antibody (1:5000, China) for 1 h. The results were imaged using an iBright 1500.

GC-MS

Analysis of tissue, serum and cellular metabolites using untargeted metabolomics and transcriptomics. Samples were collected from control, Mix and Fer-1 group mice for methyl esterification, and dissolved in n-hexane for on-line assay. To quantify the fatty acid content of the samples, C17:0 was used as the internal standard.

Statistics and reproducibility

All statistical analyses were performed with GraphPad Prism 9 software. Western blotting and immunohistochemistry were quantified using ImageJ software. Data were tested for normality using the Shapiro-Wilk test at p < 0.05. ANOVA followed by Tukey’s post-hoc test was used to compare differences between experimental groups. In all cases, p values less than 0.05 were considered significant: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Results of all data in this study were performed in 3–10 replicates and are presented as mean ± standard deviation.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

The manuscript was written through the contributions of all authors. All authors have approved the final version of the manuscript. Formal analysis, investigation, resources, data curation, writing-original draft and visualisation were performed by Yuting Zhang. Methodology, conceptualisation and writing - review and editing were performed by Kai Yin, Jiali Men, Yingai Zhang, Xue Li and Xiaobing Wang. Supervision was performed by Xiaoping Diao. Jing Yang participated in animal experiments. Project administration, funding acquisition and declaration interest statement were performed by Hailong Zhou.

Peer review

Peer review information

Communications Biology thanks Kunlong Duan and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editor: Kaliya Georgieva. A peer review file is available.

Data availability

All data supporting the findings of this study are available within the article and its Supplementary Information. Unprocessed blots are presented in Supplementary Fig. S9, Supplementary Fig. S10, Supplementary Fig. S11, Supplementary Fig. S12. All source data underlying the graphs and charts presented in the main figures are available in the Supplementary Data 1. Meanwhile, higher-resolution representative images of Supp. Figs. are provided as Supplementary Data 2. Sequencing data were submitted to the SRA database of NCBI (National Center for Biotechnology Information) under accession no: PRJNA1203385. All other data are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-025-07641-8.