Obesogenic polystyrene microplastic exposures disrupt the gut-liver-adipose axis

Abstract

Microplastics (MP) derived from the weathering of polymers, or synthesized in this size range, have become widespread environmental contaminants and have found their way into water supplies and the food chain. Despite this awareness, little is known about the health consequences of MP ingestion. We have previously shown that the consumption of polystyrene (PS) beads was associated with intestinal dysbiosis and diabetes and obesity in mice. To further evaluate the systemic metabolic effects of PS on the gut-liver-adipose tissue axis, we supplied C57BL/6J mice with normal water or that containing 2 sizes of PS beads (0.5 and 5 µm) at a concentration of 1 µg/ml. After 13 weeks, we evaluated indices of metabolism and liver function. As observed previously, mice drinking the PS-containing water had a potentiated weight gain and adipose expansion. Here we found that this was associated with an increased abundance of adipose F4/80⁺ macrophages. These exposures did not cause nonalcoholic fatty liver disease but were associated with decreased liver:body weight ratios and an enrichment in hepatic farnesoid X receptor and liver X receptor signaling. PS also increased hepatic cholesterol and altered both hepatic and cecal bile acids. Mice consuming PS beads and treated with the berry anthocyanin, delphinidin, demonstrated an attenuated weight gain compared with those mice receiving a control intervention and also exhibited a downregulation of cyclic adenosine monophosphate (cAMP) and peroxisome proliferator-activated receptor (PPAR) signaling pathways. This study highlights the obesogenic role of PS in perturbing the gut-liver-adipose axis and altering nuclear receptor signaling and intermediary metabolism. Dietary interventions may limit the adverse metabolic effects of PS consumption.

Since the middle of the 20th century, there has been an explosion of plastics production and their usage. Current estimates suggest that 364 million tons of plastics are produced annually and that this number will triple by 2050 (Rubio et al., 2020). Despite recycling efforts, vast amounts are disposed, ending up in the water systems and landfills. Given exposure to natural forces and/or chemical processes, these discarded polymers are slowly degraded into particles of smaller sizes of the micro- and nano-scale. In addition, the aquatic and terrestrial ecosystems have also become contaminated with micro- and nano-plastics, specifically engineered in this size range for manufacturing or commercial purposes. As these small plastic particles are ubiquitous in the environment, they have found their way into water supplies and the food chain (Cole et al., 2013; Duis and Coors, 2016; Rillig, 2012). As such, human exposure is inevitable, as evidenced by their detection in stool samples, blood, breast milk, and the lungs (Jenner et al., 2022; Leslie et al., 2022; Liu et al., 2023a; Schwabl et al., 2019). Thus, there is accumulating evidence that environmental microplastics (MP) contamination is pervasive and that MP uptake occurs in humans. These issues will remain ongoing concerns for the foreseeable future, given the widespread use of plastics in the modern world, their continued deposition in the environment, and their accelerated breakdown owing to globally increasing temperatures and UV exposures.

The physiological responses to MP exposure have been largely studied in aquatic species or rodents. Among commonly described outcomes are the induction of inflammatory responses (Brown et al., 2001; Farhat et al., 2011; Fuchs et al., 2016; Prietl et al., 2014), the generation of reactive oxygen species (Deng et al., 2017; Lu et al., 2016), altered metabolic homeostasis (Prata et al., 2020; Wu et al., 2019), neurotoxicity (Jin et al., 2022), and changes in microbiome composition and/or metabolism (Jin et al., 2019; Lu et al., 2018). Despite this awareness, if and how these responses contribute to adverse health outcomes in humans remain uncertain. In addition to the inherent toxicity of MP themselves, these particles can potentially bind microorganisms or chemicals from the environment. Their consumption can thus effectively deliver these xenobiotics in high concentration to distal tissues (Prata et al., 2020; Yee et al., 2021). Understanding the breadth of health outcomes resulting from MP consumption, the mechanisms thereof, and the identification of mitigation strategies are priority research areas.

In keeping with these goals, we have recently found that mice consuming water containing polystyrene (PS) beads had an increase in adiposity and indications of an insulin-resistant phenotype (Zhao et al., 2022). Thus, these PS MP can be classified as obesogens and endocrine-disrupting chemicals (EDCs). Obesogens are defined as chemicals that elicit increased white adipose tissue mass, whereas EDCs may interfere with any aspect of hormone action (Heindel et al., 2017, 2022; Lustig et al., 2022). EDCs fall under the broader umbrella of metabolism-disrupting chemicals (MDCs). The MDC hypothesis postulates that environmental chemicals have the ability to promote metabolic changes that can result in obesity, type 2 diabetes, or nonalcoholic fatty liver disease (NAFLD) (Heindel et al., 2017). The mechanisms of MDCs, EDCs, and obesogens were recently reviewed (Heindel et al., 2017, 2022), and include alterations of the intestinal microbiome, microbial metabolites, and nuclear receptor signaling acting on the gut-liver-adipose tissue axis. Nevertheless, the mechanism(s) by which PS induces obesity remains unclear. Furthermore, it is unknown if the adipose depots are dysfunctional, if other organ systems are also impacted by PS ingestion, and if dietary interventions can mitigate these effects. Although we have several publications detailing the epidemiological associations between monomeric styrene exposures and human liver injury/NAFLD (Cave et al., 2011; Wahlang et al., 2021, 2023; Werder et al., 2020), little is known about the potential role of PS in NAFLD. To further this understanding, we supplied mice with unsupplemented drinking water or that containing 0.5 or 5 μm PS beads and analyzed several indices of adipose, liver, gut, and vascular physiology. In addition, we tested whether supplementation with delphinidin, a berry anthocyanin, which improves insulin resistance in models of obesity (Lee et al., 2017; Park et al., 2019; Saulite et al., 2019), inhibits preadipocyte differentiation (Rahman et al., 2016), and protects from fat diet-induced microbiome dysbiosis (Cremonini et al., 2019) could protect from the adverse outcomes associated with PS exposure.

Materials and methods

 

 

Animals and treatment

Male C57BL/6 mice at 12 weeks of age were purchased from Jackson Laboratories and housed (5 per cage) in AALAC- and USDA-accredited facilities at University of Louisville (UofL). After 1 week of acclimation, randomly selected cages (2 cages, n = 10 mice per treatment) were supplied with unsupplemented water (filtered, autoclaved water) or that containing either “0.5 μm” (size range = 0.4–0.6 μm) or “5 μm” (size range = 4.5–4.9 μm) PS beads (Creative Diagnostics; Shirley, New York) at a concentration of 1.0 µg/ml. These doses and bead sizes induced a significant weight gain and adiposity in our prior study (Zhao et al., 2022), and given an average water consumption of 3–5 ml per day per mouse, are in a range of estimated human exposures (Cox et al., 2019; Senathirajah et al., 2021). Fresh water solutions were prepared every other day, and before making these water supplies, stock PS beads were sonicated for 30 min prior to dilution. On those days when fresh water was not supplied to mice, existing water bottles were agitated to limit potential bead settling. The mice had access to their water supplies ad libitum and were maintained in these conditions for 13 weeks. Body weights were recorded every 3 weeks during the treatment protocol and a body mass analysis was obtained by Lunar (Chicago, Illinois) PIXlmus Dual Energy X-ray Absorptiometer (DEXA) scanning at the same time intervals. At the end of the treatment duration, mice were euthanized by sodium pentobarbital (150 mg/kg body weight) injection. Blood was obtained by cardiac puncture using 0.2 M EDTA as an anticoagulant and epididymal white adipose (eWAT), perivascular adipose tissue (PVAT), and liver tissues were also collected at this time. In some experiments, mice drinking PS-containing water received i.p. injections of delphinidin chloride (Cayman Chemicals) (20 mg/kg) or vehicle (20% DMSO) twice weekly for 4 weeks. Prior to water supplementation and at the 4 weeks time point, we obtained body weights and a body mass analysis. At the end of 4 weeks, the mice were euthanized and tissues collected as above. All animal procedures were approved by the University of Louisville Institutional Animal Care and Use Committee (no. 23265).

Flow cytometry

To measure adipose immune cell infiltrate, the epididymal fat pad was removed and 200 mg was minced and enzymatically digested with 0.5 mg/ml of type II collagenase (Sigma-Aldrich) on a gentleMACS Dissociator (Miltenyi Biotec) for 42 min at 37°C. Dissociated tissue was washed with phosphate-buffered saline (PBS) containing 1% bovine serum albumin (PBS/BSA) and strained through a 70-µm filter. After washing, the cells were incubated for 10 min at 4°C with anti-mouse FC blocking reagent (Miltenyi Biotec: 130-092-575) and then stained for 30 min at 4°C with an antibody cocktail which included anti-CD45-BV510 (Miltenyi: 103138, 0.1 µl/test), anti-CD3-BV711 (Miltenyi: 100241, 3 µl/test), anti-CD4-AF700 (Miltenyi: 100430, 1 µl/test), anti-CD8a-AF488 (Miltneyi: 100723, 0.3 µl/test), anti-CD11c-BV421 (Milteny: 117330, 3 µl/test), anti-F4/80-PE (Miltenyi: 123110, 1 µl/test), and anti-CD301 (Biolegend: 145708, 1 µl/test). The cells were then washed, centrifuged, and the cell pellets were resuspended in 250 µl of PBS/BSA. Flow cytometric data acquisition was performed on a Cytek Aurora at medium speed for 2 min. Analysis of cell populations was performed utilizing FlowJo software (BD Biosciences) and fluorescence minus one controls.

Plasma analytes

Plasma was obtained from whole blood samples by centrifugation at 400 × g for 20 min and frozen until ready for analysis. Levels of adiponectin, resistin, leptin, PAI-1, ICAM-1, P-selectin, ALT, and AST were measured using a Luminex MagPix instrument and a specific reagent kit (Millipore). E-selectin was measured using an ELISA kit (R&D Systems).

Biochemical analysis and histology

PVAT and eWAT tissues harvested from the mice were pulverized and homogenates prepared in cold RIPA buffer (50 mM Tris-HCL, 0.15 M NaCl, 0.1% SDS, 0.5% sodium deoxycholate, 1% NP-40) with protease and phosphatase inhibitors. Equal protein amounts were separated by SDS-PAGE, transferred to PVDF membranes (BioRad) and probed with GAPDH (Cell Signaling: 2118S, 1:1000) and β-catenin (Cell Signaling: no. 9587S, 1:1000) or Fam3b (R&D Systems: no. AF2867, 1:1000) antibodies. Images were acquired using the ChemiDoc XRS imaging system (BioRad).

Liver and ileum sections harvested from the exposed mice were washed in PBS and fixed in 10% neutral buffered formalin for 48 h. Tissues were then dehydrated in 75% ethanol, embedded in paraffin, and sectioned into 5 µm slices with a Microtome (Leica Biosystem). To assess morphology, these sections were stained with modified Mayer’s hematoxylin and eosin-Y w/phloxine (H&E) (Epredia). Images were captured using an Aperio GT 450—automated, high-capacity pathology slide scanner (Leica Biosystems).

Additional liver sections were homogenized in 50 nM NaCl and lipids extracted using chloroform: methanol (2:1) as described (Bligh and Dyer, 1959). Hepatic lipids were recovered from the lower chloroform layer and then evaporated to dryness. Levels of cholesterol, triglycerides, and free fatty acid were measured using commercial kits (Infinity Liquid Stable Reagents, Thermo Fisher) according to the manufacturer’s instructions and normalized to liver weight. We also measured glutathione (GSH) and glutathione disulfide (GSSG) by HPLC (Jones and Liang, 2009). Hepatic sections were homogenized in 5% perchloric acid and 0.2 M boric acid using 10 µM γ-glutamyl glutamate as an internal standard. These preparations were centrifuged, and the resulting supernatant was derivatized with iodoacetate and dansyl chloride to produce S-carboxymethyl and N-dansyl derivatives, whereas the resulting pellet was reconstituted in sodium hydroxide for protein quantification using BioRad DC protein assay. The S-carboxymethyl and N-dansyl derivatives were analyzed on an Arc HPLC equipped with a Spherisorb NH2 analytical column and a W2475 fluorescence detector (Waters Corporation, Milliford, Massachusetts). Concentrations of cysteine (Cys), cystine (CySS), GSH, and GSSG were determined by comparing with the internal standard and normalized to total protein.

Targeted gene expression analysis

To determine gene expression changes in adipose tissue, total RNA was first isolated from pulverized samples using a miRNeasy kit (Qiagen) and concentrations were determined using a NanoDrop instrument (Thermo Scientific). cDNAs were prepared using the RT² first strand kit (Qiagen) and then levels IL-6 (Mm00446190_m1), Mcp-1 (Mm00441242_m1), Tnfα (Mm00443258_m1), Dkk1 (Mm00438422_m1), Wnt1 (Mm01300555_g1), Gata-3 (Mm00484683_m1), Bmp-7 (Mm00432102_m1), Ucp1 (Mm01244861_m1), Akr1b7 (Mm00477605_m1), and Fam3b (Mm00508056_m1) were determined by rtPCR on a Quant Studio 5 (Thermo Fisher) using specific primers and reagents (Thermo Fisher). Fold changes were calculated using the ΔΔC_T method (Livak and Schmittgen, 2001) and Hprt (Mm03024075_m1) as a reference gene.

RNAseq analysis

To perform the adipose transcriptomic analysis, RNA was prepared from tissue sections using a miRNeasy kit (Qiagen). Concentrations were determined using a NanoDrop instrument (Thermo Scientific). RNAseq and data analysis utilizing the Kyoto Encyclopedia of Genes and Genomes (KEGG), Gene Ontology (GO) classifications, and the Reactome Database was performed by Novogene. Section of the harvested liver was processed the same way, but RNAseq analysis was performed by the Center for Genetics and Molecular Medicine at UofL. Hepatic core pathway analysis was performed using ingenuity pathway analysis (IPA) software (Qiagen, Germantown, Maryland) to determine overrepresented pathways based on differentially expressed genes (DEGs) for each comparison. Significance for upregulated and downregulated hepatic genes were set at a q-value <0.05 with a log₂ fold change = 0. Pathways were defined to be overrepresented if −log(p-value) >1.3.

Bile acid analysis

Bile acids were analyzed using liquid chromatography mass spectrometry (LC-MS). A Thermo Q Exative HF Hybrid Quadrupole-Orbitrap Mass Spectrometer coupled with a Thermo DIONEX Ultimate 3000 HPLC system (Thermo Fisher Scientific, Waltham, Massachusetts) was used in the analysis, as previously described (He et al., 2021). In brief, liver and cecum samples were homogenized with 80% MeOH to extract bile acids using a ratio of 1 mg of sample per 10 μl of 80% MeOH. The homogenized samples were then centrifugated at 18 000 × g at 4°C for 15 min and the supernatant (400 μl) was transferred to a clean tube. After evaporating the MeOH using a SpeedVac and overnight lyophilization, the dried samples were reconstituted with 200 μl water (pH ≥ 8.5). A solid-phase extraction was then applied to purify and enrich the bile acids using OASIS HLB cartridges (Waters, Milford, Massachusetts) as described with minor modifications (He et al., 2019). The reconstituted sample (200 μl) was loaded on the cartridge after its re-activation and re-equilibration. The cartridge was then washed with 1 ml H₂O at pH ≥ 8.5. Then, three 150 μl aliquots of 70% acetonitrile with pH ≥ 8.5 were loaded to elute the bile acids. The total 450 μl eluate was lyophilized overnight. The dried extracts were finally redissolved in 50 μl initial mobile phase buffer, 19.2% acetonitrile with 1 mM ammonium acetate (pH = 4.15), for LC-MS analysis.

Group-based pool samples were prepared by mixing identical volumes of each individual sample from the same group. To obtain the full MS data for metabolite quantification, samples were analyzed in ESI⁻ mode. The pooled samples were analyzed to obtain the MS/MS spectra using different normalized collision energies (20%, 40%, and 60%) for metabolite identification, utilizing both in-house database and Compound Discover software (version 3.2, Thermo Fisher Scientific, Inc., Germany). MetSign software was used to process LC-MS data for cross-sample alignment and normalization (Wei et al., 2011). One-way analysis of variance (ANOVA) was performed followed by post hoc Tukey’s multiple comparison with p-value < .05 considered statistically significant for all tests.

Statistical analysis

Statistical analyses were performed using GraphPad Prism version 9.5.1 for Windows (GraphPad Software Inc., La Jolla, California). Data are expressed as mean ± SD. Data were initially assessed for normality using the D’Agostino & Pearson test as well as homoskedasticity using Barlett’s test. Based on results from these tests, a comparison of central tendencies was performed employing one way ANOVA followed by a Tukey’s post hoc test (for parametric and homoscedastic data) or Kruskal-Wallis test followed by Dunn’s multiple comparison test (for nonparametric and homoscedastic data), or Brown-Forsythe and Welch ANOVA test followed by Dunnett post hoc test (for parametric but heteroskedastic data). t tests were used to compare between 2 groups. p < .05 was considered statistically significant.

Results

 

Polystyrene consumption and adiposity

A major finding of our previous study (Zhao et al., 2022) and repeated in the current treatment was that consumption of 0.5 µm PS beads led to a potentiated weight gain compared with control mice after 13 weeks of exposure (Supplementary Figure 1), and that was primarily due to adipose tissue expansion. To determine if this tissue was also dysfunctional, we initially examined eWAT immune cell infiltrate by flow cytometry. In this analysis, we observed an increase in the total number of infiltrated macrophages (positive F4/80 staining) in mice consuming the 0.5 µm PS beads compared with mice drinking unsupplemented water (Figure 1A). Mice consuming the 5 µm PS beads did not demonstrate this increase. Although there was no difference in the abundance of anti-inflammatory M2 macrophages between the groups (Figure 1B), there was a trending, but nonsignificant (p = .06) increase in the abundance of pro-inflammatory M1 macrophages in mice consuming the 0.5 µm beads (Figure 1C). In addition, there was no difference in the T lymphocyte (positive CD3 staining) infiltration between the groups (Figure 1D). To further analyze adipose tissue function, we assessed the levels of plasma adipokines in that group demonstrating adipose inflammation (0.5 µm beads). However, we observed no differences in levels of adiponectin, resistin, leptin, or PAI-1 between this group and the group consuming normal water (Figure 2). Finally, we measured the expression levels of genes indicative of adipose inflammation. Although we only observed minor changes in the expression of IL-6, Mcp-1, and TNFα in eWAT, there was a mild upregulation of Mcp-1, and TNFα in PVAT (Table 1).

Figure 1.

Adipose tissue immune cell infiltrate. White adipose tissue was collected from mice drinking unsupplemented water (water) or that containing either 0.5 or 5 µm polystyrene beads (PS) beads at euthanasia, digested, and analyzed for the presence of F4/80 macrophages (A), M2 macrophages (B), M1 macrophages (C), and CD3⁺ lymphocytes (D) by flow cytometry.*p < .05. n = 8–10.

Figure 2.

Plasma adipokines. Blood was collected from mice drinking unsupplemented water (water) or that containing 0.5 µm PS (PS) beads at euthanasia, the plasma was isolated and then analyzed for the presence of adiponectin (A), resistin (B), leptin (C), and PAI-1 (D). n = 10.

Table 1.

Adipose gene expression changes

Gene	Abbreviation	eWAT fold change (SE) N	PVAT fold change (SE) N
Inflammation
Interleukin 6	IL-6	0.76 (0.08) n = 9	ND
Monocyte chemoattractant protein 1	Mcp-1	1.04 (0.15) n = 8	1.21 (0.22) n = 5
Tumor necrosis factor alpha	Tnfα	0.92 (0.20) n = 9	3.02 (1.04) n = 5
Wnt-1 signaling
Dickkopf-related protein 1	Dkk1	1.19 (0.24) n = 6	2.12 (0.54) n = 4
Wingless-related MMTV integration site 1	Wnt1	0.78 (0.09) n = 9	ND
GATA-binding protein 3	Gata-3	1.88 (0.31) n = 8	1.63 (0.32) n = 5
Browning
Bone morphogenic protein 7	Bmp-7	1.88 (0.37) n = 8	1.06 (0.16 n = 5)
Uncoupling protein 1	Ucp1	1.54 (0.44) n = 9	0.91 (0.09) n = 5

Listed are those fold changes in mice consuming the 0.5 µm beads relative to control mice. ND, not determined.

Adipose expansion is regulated in part through the Wnt/β-catenin signaling pathway, which inhibits adipocyte differentiation (Chen and Wang, 2018; Christodoulides et al., 2009; Prestwich and Macdougald, 2007). To better clarify the role of this pathway in stimulating adiposity in our model of exposure, we examined levels of β-catenin in PVAT and eWAT collected from mice consuming PS-containing water or unsupplemented water. As mice consuming the 5 µm PS beads did not show a significant, potentiated weight gain during this exposure time course (Supplementary Figure 1), they were excluded from this analysis. For mice consuming the 0.5 µm PS beads, we found no changes in β-catenin levels between the treatment groups in both adipose tissue types (Figure 3), suggesting the presence of an intact Wnt-β-catenin signaling pathway. Consistent with this observation, we found only minor perturbations in the expression of genes regulating Wnt-1 signaling (Dkk1, Wnt-1, and Gata-3) (Table 1) as determined by rtPCR. Some genes involved in adipose browning (Ucp-1, Bmp-7) showed mild elevation in eWAT (Table 1). Finally to determine if PS ingestion in our mouse model of exposure promoted endothelial activation, we assessed the abundance of some indicative plasma markers. However in this analysis, we found that samples from both the control (normal water) and PS-supplemented mice had similar levels of ICAM-1, P-selectin, and E-selectin (Supplementary Figure 2).

Figure 3.

β-catenin levels. Lysates were prepared from eWAT (A) and PVAT (B) samples collected from mice drinking unsupplemented water (H₂O) or that containing 0.5 µm polystyrene beads (PS) and then analyzed for the presence of β-catenin and GAPDH by Western blotting. Also illustrated is normalized group data (C). n = 4.

Treatment with the anthocyanin, delphinidin, mitigates polystyrene-induced metabolic effects

As the plant- and berry-derived anthocyanin, delphinidin has anti-adipogenic effects, we next tested if this compound could protect against PS-induced metabolic effects. We performed this analysis in mice supplied with the 0.5 µm PS beads, because this size demonstrated the greater effects on obesity (Zhao et al., 2022). When mice were supplied with PS-containing water and injected with delphinidin, we observed an attenuated weight gain compared with those mice receiving PS-water and injected with vehicle (Figure 4A). Similarly, the delphinidin-injected mice had decreased adiposity compared with the vehicle-injected mice (Figure 4B). To gain insight into the protective effects of delphinidin, we performed an RNAseq analysis on adipose tissue collected from the treated mice. DEGs based upon a fold change are listed in Supplementary Table 1. Relative changes in the most upregulated DEG (Akr1b7) and downregulated DEG (Fam3b) were confirmed by rtPCR (Supplementary Table 1). Decreased Fam3b levels in the samples from mice receiving delphinidin were also confirmed by Western blot (Supplementary Figure 3). We also performed a KEGG pathway analysis of this RNAseq data. Among the top pathways downregulated by delphinidin treatment were those involved in cyclic adenosine monophosphate (cAMP) and peroxisome proliferator-activated receptor (PPAR) signaling (Supplementary Figure 4). Highest among the upregulated pathways were those involved in “cytokine-cytokine receptor interactions,” “phagosome,” and “osteoclast differentiation” (Supplementary Figure 5). Other pathways regulated by delphinidin were related to endocrine disruption (eg, glucagon signaling and insulin secretion), metabolism disruption (eg, linoleic acid and arachidonic acid metabolism; “glycolysis and gluconeogenesis”; “taurine and hypotaurine metabolism”; “protein digestion and absorption”), and “bile secretion.”

Figure 4.

Effects of delphinidin injection. Mice were supplied with unsupplemented water (U; n = 4) or that containing polystyrene beads (PS) and injected with either vehicle (Veh; n = 8) or delphinidin (Del; n = 8) as indicated. After 4 weeks of this treatment, we measured weight gain (A) and percent body fat (B).

Polystyrene consumption and hepatic effects

Many MDCs and obesogens have previously been associated with steatotic liver disease and NAFLD. Therefore, liver phenotyping was performed to investigate hepatic steatosis, inflammation, and metabolism in response to PS exposure. We observed that liver histology was unchanged by either 0.5 or 5 µm PS exposure (Figure 5A). Specifically, neither lipid droplets nor necroinflammatory foci were seen on H&E stain. However, we did observe that the liver:body weight ratios were significantly decreased in the 0.5 µm (4.56 ± 0.22, p = .0165) and 5 µm (4.33 ± 0.50, p = .0005) PS exposure group compared with control (5.01 ± 0.29) (Figure 5B). Consistent with the former observation, hepatic triglycerides were unchanged by the PS exposures (Figure 5C). However, both 0.5 µm (2.34 ± 0.33 mg/g liver; p = .005) and 5 µm (2.21 ± 0.15 mg/g liver; p = .0006) PS exposure groups demonstrated significantly increased hepatic cholesterol compared with the control group (1.89 ± 0.16 mg/g liver) (Figure 5D). Plasma alanine transaminase and aspartate transaminase activities were unchanged between exposure groups (Supplementary Figure 6), again consistent with the histology.

Figure 5.

Liver phenotyping. Livers were collected from mice exposed to control water, 0.5 µm PS, or 5 µm PS and sections stained with H&E. Illustrated are representative images captured at 10× magnification. Scale bar: 300 µm (A). Also illustrated are: the liver:body weight ratios (B); levels of hepatic triglycerides (C); and levels of hepatic cholesterol levels (D) in the control and PS exposure groups. *p < .05.

mRNA-Seq was performed in liver samples. There were 283 DEGs in the 0.5 µm PS versus water control comparison, with 203 of these upregulated and 80 downregulated (Figure 6A). The most significantly upregulated genes were Steap4, Tnfaip3, and Ccl2, whereas the most significantly downregulated genes were Tpm2, Trib3, and Pcp4l (Figure 6B and Supplementary Tables 2 and 3). There were fewer DEGs (n = 49) in the 5 µm PS versus water control comparison. Forty-one DEGs were upregulated and 8 DEGs were downregulated (Figure 6A). The most significantly upregulated DEGs were Firre, Fam193b, and Mug-ps1, whereas the most downregulated were Slpi, Ctgf, and Sgk1 (Figure 6B and Supplementary Tables 4 and 5). Five DEGs (Rev1, Brpf1, Cnnt2, Cavin2, and Pea15a) were common to both PS treatment groups (Figure 6C).

Figure 6.

Hepatic differential gene expression. Illustrated is a bar graph depicting the number of differentially expressed genes (DEGs) in liver (mRNA-Seq) in the 0.5 µm PS versus water and the 5 µm PS versus water comparison groups (A). Listed are the log2-fold change values of the top 10 most upregulated and downregulated DEGs in the comparison groups by log2-fold change (B). A Venn diagram depicts the relationships between the DEGs in the comparison groups. Breakout boxes indicate the top 10 most upregulated (red) and downregulated (blue) regulated DEGs in the exposure groups (C). Cut-off values are q-value < 0.05. n = 10.

Next, IPA constructed the hepatic GO processes enriched by the DEGs associated with the PS exposures (Supplementary Figure 7: complete list; Figure 7: top pathways). Consistent with the overall larger number of DEGs, there were more enriched pathways in the 0.5 μm PS exposure group compared with the 5 μm exposure group. Interestingly, “FXR/RXR activation” and “LXR/RXR activation” were enriched by both bead sizes. Other selected pathways enriched only by 0.5 μm PS exposures included: “glutathione-medicated detoxication”; “LPS/IL1 mediated inhibition of RXR function”; “hepatic cholestasis”; and other processes related to xenobiotic metabolism (eg, AHR, PXR, NRF2) as well as cytokine signaling, the glucocorticoid receptor, and linoleate metabolism (Figure 7 and Supplementary Figure 7). The unique pathways enriched by the 5 μm PS exposures included several related to carbohydrate metabolism (“sucrose degradation” and “glycolysis”) and cytokine signaling.

Figure 7.

Pathway analysis of liver differentially expressed genes (DEGs). DEGs (liver mRNA-Seq) were analyzed using ingenuity pathway analysis (IPA) software and the top enriched pathways for the comparison groups (0.5 µm PS vs water and 5 µm PS vs water) are provided. Two processes were predicted to be enriched in both exposure groups. n = 10.

Effect of polystyrene exposure on bile acid and glutathione metabolism

Given the predicted alterations in FXR, PXR, and LXR signaling and the previously reported impact of PS on the intestinal microbiome (Zhao et al., 2022), bile acids were measured in liver and cecum. As expected, PS exposures were associated with significant changes in bile acids. In liver, 0.5 μm PS exposures (vs water control) significantly decreased the abundance of taurochenodeoxycholic acid (TCDCA), fold change (FC = 0.57, p = .02), and taurohyodeoxycholic acid (THDCA, FC = 0.33, p < .05). Also observed were nonsignificant trends toward reductions in 4 other bile acids (p = .06–.11) with an increasing, but nonsignificant trend in only 1 bile acid (p = .13) (Supplementary Table 6). These bile acids reduced to varying degrees of statistical significance by 0.5 μm PS exposures included several formed via microbial metabolism, including 4 which were taurine conjugates (Supplementary Table 6). Exposure to 5 μm PS significantly increased (vs water control) the abundance of hyodeoxycholic acid (HDCA, FC = 1.93, p = .0002) and taurohyocholic acid (THCA, FC = 2.02, p = .03) (Supplementary Table 6). When we compared changes in the 0.5 μm PS bead exposures relative to the 5 μm bead exposures, we found that 5 bile acids were significantly reduced and that another 5 bile acids were reduced but were trending toward statistical significance (p < .10) (Supplementary Table 6). There were no statistically significant increases or trends toward an increase in hepatic bile acids when comparing 0.5 μm versus 5 μm PS exposures. Although PS exposure altered cecal bile acid abundance, those bile acids were different from those impacted in the liver. Apocholic acid (ACA, FC = 2.17, p = .005) increased in abundance in the 0.5 μm PS group compared with the water control group, whereas ursodeoxycholic acid (UDCA, FC = 1.87, p = .03) and ɣ-muricholic acid (ɣ-MCA, FC = 3.11, p = .03) increased in abundance in the 5 μm group compared with the water control. Finally, because exposure to 0.5 μm PS enriched liver pathways related to glutathione-mediated detoxication and NRF2, hepatic glutathione levels were measured. Reduced (GSH) and oxidized (GSSG) glutathione, as well as the GSH/GSSG ratio (all normalized to total liver protein) were unchanged by exposure to either size PS bead exposure (Supplementary Figs. 8A–C). However, hepatic protein content was increased in mice consuming the 5 μm PS beads (Supplementary Figure 8D).

Polystyrene consumption induced no changes to the ileum

Intestinal H&E staining was performed to investigate any abnormalities to the ileum in response to PS exposure. Ileum histology showed no pathological changes in response to either 0.5 or 5 µm PS bead exposure (Supplementary Figure 9). There was no significant qualitative difference between any of the exposure groups and the control group. The villous and crypt architecture were intact with no inflammation or evidence of mucosal injury.

Discussion

Despite a growing awareness of ubiquitous MP contamination, there is only a rudimentary understanding of the biological responses to acute or chronic exposure to these particles and the underlying mechanisms thereof. There is even less known with regards to the long-term health consequences of exposure. To begin to address some of these issues, we used a mouse model of exposure and characterized the impact of PS consumption on aspects of adipose, vascular, gut, and liver physiology. In addition, we determined if a dietary intervention, supplementation with the flavonoid, delphinidin, could moderate these responses.

Consistent with our earlier study (Zhao et al., 2022) and a recent report (Huang et al., 2023), we found that mice consuming PS beads had a potentiated weight gain, which seemed to be largely driven by an increase in fat mass. In the current study, we also observed an accumulation of F4/80⁺ macrophages in eWAT. Although we did not see a statistically significant accumulation of either the classically activated, pro-inflammatory M1 macrophages or the alternatively activated, anti-inflammatory M2 macrophages, there was a tending increase of the former (p = .06). Despite this increase in WAT macrophage abundance, we did not detect an increase in pro-inflammatory gene expression in this tissue nor were plasma adipokines increased in the PS-fed animals. Thus, although we observed an increase in adiposity, this tissue did not appear to be dysfunctional at this PS-treatment time course. Nevertheless, given the trending increase of pro-inflammatory M1 macrophages, we speculate that prolonged PS feeding or that a second “hit” (eg, high fat diet [HFD]) may lead to widespread adipose inflammation, and dysfunction, potentially contributing to frank CVD (eg, atherosclerosis) in animal models or humans who chronically ingest high levels of MP. Indeed, some recent studies report metabolic disturbances, systemic inflammation, and liver pathology in MP-exposed mice placed on a high-fat diet (Lee et al., 2023; Li et al., 2022; Shiu et al., 2022; Wang et al., 2023) although it is difficult to compare results between the studies given differences in MP dose, size, and mode of delivery.

The mechanism(s) driving the increased weight gain and adiposity in PS-fed mice remain unclear. However, as we observed no downregulation of β-catenin levels in either WAT or PVAT, this study at least suggests that disruption of Wnt-β-catenin signaling does not appear to be a major mechanism. Supporting this contention, we also observed only modest changes in the expression of genes regulating Wnt signaling in WAT samples. Furthermore, genes promoting adipose browning showed a modest increase, suggesting an adipose distribution favoring the white phenotype was not strongly established. Another potential mechanism involves gut dysbiosis with altered intestinal microbial metabolites as suggested by our previous study (Zhao et al., 2022) and in a more recent report (Huang et al., 2023). A clearer definition of the involvement of this mechanism requires further study.

An increase in obesity is a risk factor for, and often associated with, fatty liver disease (Joshi-Barve et al., 2015). Although we did see an increase in hepatic cholesterol in our PS-exposed mice, neither histologic steatosis, injury nor inflammation were observed. To our knowledge, this is the first report of an obesogen also associated with insulin resistance that did not cause NAFLD (Zhao et al., 2022). This does not exclude the possibility that PS MPs could worsen diet-induced NAFLD, as the mice in the present study were fed a chow diet. This was indeed the case in a recent publication investigating the interaction between PS MPs and HFD in leaky gut syndrome and NAFLD (Okamura et al., 2023). Interestingly, that manuscript postulated that HFD-induced leaky gut potentiated the systemic metabolic impacts of PS acting primarily at the level of the intestine. Likewise, some other recent studies report liver pathology when PS-treated animals were placed on a high-fat diet (Chen et al., 2023; Li et al., 2022; Shiu et al., 2022).

Despite the absence of hepatic steatosis in our model, there was ample evidence that PS disrupted liver metabolism. Both PS bead sizes were associated with enrichment in hepatic LXR and FXR signaling. These findings were accompanied by increased liver cholesterol and altered liver/cecal bile acids. Primary bile acids are synthesized from cholesterol by the liver and may then be conjugated and secreted into the intestinal lumen as bile. Cholesterol and bile acid homeostasis are regulated, in part, by nuclear receptors including liver X receptor (LXR) and the farnesoid X receptor (FXR). Secondary bile acids are formed by chemical modification by the gut microbiome. Bile acids undergo enterohepatic circulation between the intestinal lumen and liver. Specific bile acids may agonize or antagonize the FXR providing feedback control of their metabolism. Increasingly, roles for bile acids in the control of obesity, diabetes, lipid metabolism, and NAFLD have been determined (Chavez-Talavera et al., 2017). Interestingly, although both PS bead sizes enriched FXR and LXR signaling pathways, PS had size-dependent effects on liver and cecal bile acids. Several of the bile acids associated with the PS exposures are reported in the literature to be ligands for or alter the expression of the FXR (eg, TCDCA, HDCA, and ɣ-MCA), sometimes impacting body weight gain or NAFLD (Liu et al., 2023b; Song et al., 2020; Xie et al., 2023). The 0.5 µm PS exposure, which was associated with obesity, insulin resistance, and altered microbiome (Zhao et al., 2022), tended to decrease liver bile acids although only 2 of these reached statistical significance. The 5 µm PS exposure did not cause obesity or hepatic bile acid depletion. Two hepatic bile acids were significantly increased by 5 µm PS. The cecal levels of UDCA, long used as a therapy for liver disease, were increased by 5 µm PS. We postulate that metabolites generated by a PS-regulated intestinal microbiome could contribute to the observed obese phenotype by impacting the gut-liver-adipose axis through differential nuclear receptor activation.

The 0.5 µm PS exposure was also associated with enrichment in pathways related with liver xenobiotic metabolism (eg, AHR, PXR, and NRF2), glucocorticoid receptor signaling, and linoleate metabolism. Because of the observed enrichment in NRF2-related pathways, we measured hepatic glutathione levels. The GSH:GSSG ratio was unchanged. IPA analysis showed that there was an enrichment in genes related to “glutathione-medicated detoxication,” but this result was driven by differential expression of glutathione S-transferases (data not shown), which would not necessarily alter the GSH:GSSG ratio. “LPS/IL1 mediated inhibition of RXR function” and “hepatic cholestasis” were also enriched by 0.5 µm PS. The gut provides a large reservoir for lipopolysaccharide (LPS), and metabolic endotoxemia has long been associated with diabetes and obesity (Regnier et al., 2021). LPS is another example of a gut-derived molecule that warrants additional investigation in PS toxicology. Although, not present histologically, the molecular observation of hepatic cholestasis could potentially be related to bile acids and FXR signaling, as FXR agonists are used clinically for the treatment of some cholestatic liver diseases. The 5 µm PS exposure was also associated with increased hepatic protein concentrations and enrichment in “sucrose degradation” and “glycolysis” pathways, evidence for disruption of normal liver protein and carbohydrate intermediary metabolism by these MPs.

In an attempt to mitigate the obesogenic effects of PS consumption, and better clarify the mechanism of PS-induced adipogenesis, we examined the efficacy of delphinidin supplementation. Consistent with the previously reported protective effects of delphinidin-containing extracts in high fat models (Daveri et al., 2018; Long et al., 2021; Sandoval et al., 2019), we observed that mice receiving purified delphinidin had an attenuated weight gain and lower percent body fat compared with PS-fed mice receiving vehicle injections. Although delphinidin and other anthocyanins may exert these effects through several means (Lee et al., 2017), it is interesting that the most upregulated gene in delphinidin-treated mice is Akr1b7 (Supplementary Table 1), which has reported anti-obesogenic properties (Volat et al., 2012). Furthermore, the most downregulated gene was Fam3b, which is upregulated in obesity and promotes hepatic lipogenesis and gluconeogenesis (Zhang et al., 2018). Additional KEGG analysis of our transcriptomic data suggests one protective mechanism promoted by delphinidin-treatment may be the downregulation of cAMP and PPAR signaling, which promote adipocyte gene expression and differentiation (MacDougald and Mandrup, 2002; Tang and Lane, 2012). One upregulated pathway as determined from KEGG analysis is that of osteoclast differentiation. Thus, it appears that although PS consumption promotes the differentiation of mesenchymal precursor cells into multiple lineages including preadipocytes and adipocytes, delphinidin skews differentiation into one favoring the osteocyte and myocyte lineages. Other pathways regulated by delphinidin were related to bile acids and taurine/hypotaurine metabolism. Interestingly, the 2 hepatic bile acids that were significantly reduced by 0.5 µm PS were taurine conjugates. Although the relationship between delphinidin and bile acids, in general, is unclear, there is some literature to suggest that delphinidin may regulate bile acids (Han et al., 2022). The potential mechanistic role of delphinidin acting through bile acid modulation to attenuate the obesogenic action of PS warrants future investigation.

One limitation of these studies is that we did not determine the distribution or accumulation of ingested PS beads in murine tissues. Thus, it is not clear if, as has been reported in the literature (Leslie et al., 2022), these beads can end up in circulation to directly access adipose and hepatic tissue. It is also not known if PS consumption disrupts redox balance or promotes systemic inflammatory responses to affect these tissues. Another limitation of these studies is the use of commercially obtained pristine beads. Natural exposure to MP is likely to be a mixture of plastic types and sizes, that have been subjected to weathering from environmental influences, and may be laden with additional chemicals and organisms. Nevertheless, our results support the growing concern of adverse health outcome resulting from MP exposures.

Conclusions

PS consumption promotes adiposity and is predicted to alter the gut-liver-adipose tissue axis, perhaps by altering nuclear receptor signaling and intermediary metabolism. Dietary interventions may limit the adverse metabolic effects of exposures to these food and water contaminants.

 

Dryad Digital Repository DOI: https://doi.org/10.5061/dryad.h44j0zpsj

Supplementary data

Supplementary data are available at Toxicological Sciences online.

Authors contributions

J.Z.: Conceptualization, Data curation, Formal analysis, Investigation, Writing—original draft. N.A.: Conceptualization, Data curation, Formal analysis, Investigation, Writing—original draft. D.G.: Investigation. M.M.: Data curation, Formal analysis, Investigation, Methodology. D.J.C.: Conceptualization, Methodology. A.E.: Data curation, Formal analysis, Writing—original draft. J.L.: Investigation. T.G.: Investigation. W.H.W.: Formal analysis, Methodology. M.B.: Formal analysis, Investigation. M.L.S.: Data curation, Formal analysis, Methodology, Validation. E.C.R.: Formal analysis, Software, Validation. R.X.: Data curation, Investigation. X.Z.: Methodology, Investigation, Validation. D.D.G.: Formal analysis M.C.C.: Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Writing—original draft. T.E.O.: Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Writing—original draft.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

National Institutes of Health (R01ES019217, R35ES028373, R01ES032189, P42ES023716, P30ES030283, P20GM113226, 5P50AA024337, T32ES011564, GM127607); the University of Louisville School of Medicine (G7021, G7016). This work is solely the responsibility of the grantees and does represent official views of the National Institutes of Health.

Data availability

Data associated with this manuscript are available at: https://doi.org/10.5061/dryad.h44j0zpsj.