All the experimental procedures were approved by the Committee for Animal Research, Kyoto Prefectural University of Medicine, Japan (approval number: M2023-87, M2022-92). Six-week-old C57BL/6J (wild-type) male mice were purchased from Shimizu Laboratory Supplies (Kyoto, Japan) and housed in a pathogen-free controlled environment. We used littermates born simultaneously in a mouse supply facility. The mice were housed one per cage and fed a normal diet (ND; 345 kcal/100 g, fat kcal 4.6%; CLEA, Tokyo, Japan) for a duration of six weeks, commencing at 6 weeks of age. Paired feeding was performed by supplying equal amounts of the feed. Mice in the MPs exposure group were administered carboxyl group-modified fluorescent polystyrene particles (F-K1 050; 0.45–0.53 μm polystyrene COOH; Green Fluorescent Protein (GFP) fluorescence; Merck, Germany) were dissolved in water at 1000 μg/L and 5000 μg/L under the condition of free water intake (n=10) (16, 17). Water with dissolved MPs was sonicated at 20 kHz for 15 minutes and the water was changed every 3 days, and the amount of water lost was measured to estimate the amount of water consumed (13). Body weight was assessed weekly. Upon reaching 12 weeks of age, the mice were euthanized by administering 0.3 mg/kg of medetomidine, 4.0 mg/kg of midazolam, and 5.0 mg/kg of butorphanol, after overnight fasting (18) and sacrificed under specific-pathogen-free conditions ( Figure 1A ).

Five days before the mice were sacrificed, at the age of 12 weeks, mice were orally administered a solution of dextran labeled with fluorescein isothiocyanate (FITC; 25 mg/mL) at a dose of 20 mL/kg after a 4-hour fast. Blood samples were obtained through retro-orbital puncture, both prior to and 3 hours after oral gavage. The obtained blood samples were promptly subjected to density gradient centrifugation at 1500 × g for 15 minutes. The plasma fraction was harvested and diluted with phosphate-buffered saline (PBS) at a 1:2 ratio. Plasma levels of dextran were estimated by measuring luminescence using an Orion L microplate luminometer at an excitation wavelength of 490 nm and emission wavelength of 520 nm (Berthold Detection Systems, Pforzheim, Germany). A total of 10 samples were analyzed. Due to the challenges in distinguishing the FITC signal from FITC-dextran and the GFP signal from GFP-MPs through absorbance spectrometry, we assessed the GFP signal in the serum prior to administration and the FITC-dextran signal 3 hours after administration. The difference in plasma signal before and after FITC-dextran administration was calculated as a ratio to the control group and used as an indirect index of intestinal permeability (19).

Blood samples were collected from fasted mice via the portal vein during sacrifice, and blood aspartate aminotransferase (AST), alanine aminotransferase (ALT), total cholesterol, triglyceride (TG), and non-esterified fatty acid (NEFA) levels were measured. Biochemical examinations were performed using the FUJIFILM Wako Pure 18 Chemical Corporation (Osaka, Japan).

Jejunum and colon were excised from mice, promptly immersed in 10% buffered formaldehyde, and allowed to fix for 24 hours at 22°C. Subsequently, they were embedded in paraffin, sectioned into 4-µm-thick slices, and stained with HE and periodic acid Schiff (PAS) stain in Carnoy’s solution.

Liver tissue was obtained, fixed in 10% buffered formaldehyde, embedded in paraffin, sectioned into 4-µm-thick slices, and subjected to staining with hematoxylin and eosin. Oil Red O stock solution was formulated in isopropanol (0.25 g/100 mL) and subjected to heating at 100°C for 10 minutes. The liver sections, fixed with 4% paraformaldehyde for 30 minutes and rinsed with PBS, were subsequently immersed in a 60% dilution of Oil Red O stock solution in distilled water for 30 minutes. Stained sections were further washed with PBS until the background was clear. Images of stained sections were captured using a fluorescence microscope (BZ-X810; Keyence). Villus height/width and crypt depth were quantified using HE-stained sections at five distinct sites per slide for each group of 10 animals using Image J (Version 1.53 k, NIH, Bethesda, MD, USA). Goblet cells (PAS+) were counted and expressed as the mean number of goblet cells (PAS+) per 10 crypts using Image J, following a previous study (20).

Additionally, to evaluate the severity of NAFLD, we computed the NAFLD activity score (NAS) (21), a widely recognized standard used for assessing non-alcoholic steatohepatitis (NASH) severity and measuring changes in NAFLD. The scoring system comprises 14 histological attributes, of which the following four were semi-quantitatively assessed: steatosis (0–3), lobular inflammation (0–2), hepatocellular ballooning (0–2), and fibrosis (0–4).

The jejunum of mice subjected to a 16-hour fast was excised and promptly cryopreserved using liquid nitrogen. Both jejunum and liver specimens were homogenized in ice-cold QIAzol Lysis Reagent (Qiagen, Venlo, Netherlands), followed by isolation of total RNA according to the manufacturer’s instructions. A cDNA library was constructed using the TruSeq^® Stranded mRNA kit (Qiagen, Carlsbad, CA, USA). Paired-end sequencing was conducted on an Illumina NovaSeq6000 platform (n = 3). The comprehensive mRNA expression profiles of amino acids, fatty acids, and glucose transporters were visualized through volcano plots and heat maps.

The jejunum and liver of mice fasted for 16 hours were excised and immediately frozen in liquid nitrogen. Total RNA was extracted for the microarray analysis.

Total RNA was reverse transcribed using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA) for first-strand cDNA synthesis using an oligonucleotide dT primer and random hexamer priming, according to the manufacturer’s recommendations. The reverse transcription reaction was carried out for 120 minutes at 37°C, followed by termination for 5 minutes at 85°C. Quantitative reverse transcription polymerase chain reaction (RT-qPCR) was used to quantify the mRNA expression levels of Tnfa, Il6, Il1b, Il22, Pept1, Cd36, and Sglt1 in the jejunum and Tnfa, I16, Ilib, and Mcp1 in the liver. RT-qPCR was performed using the TaqMan Fast Advanced Master Mix (Applied Biosystems, Waltham, MA, USA) according to the manufacturer’s instructions. The cycling conditions were as follows: 1 cycle for 2 min at 50°C and 20 s at 95°C, followed by 40 cycles of 1 s at 95°C and 20 s at 60°C.

The relative expression levels of each targeted gene were normalized to the threshold cycle (Ct) values of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and quantified using the comparative Ct 2^−ΔΔCT method. Signals originating from HFD-fed mice were assigned a relative value of 1.0. Six mice from each group were assessed, and RT-qPCR was conducted in triplicates for each sample.

To avert blood contamination in the jejunum, we conducted systemic perfusion using heparinized saline before the harvesting or rinsing of the tissue with PBS. The acquired samples were preserved in cold RPMI supplemented with 2% FBS until required. Upon euthanasia, mononuclear cells from the intestinal lamina propria (LPL) were isolated using the Lamina Propria Dissociation Kit (130-097-410; Miltenyi Biotec, Germany) following the manufacturer’s guidelines. The resultant cell pellets were subsequently suspended in 5 mL of 40% Percoll^® solution. This cellular suspension was delicately layered onto the upper segment of centrifuge tubes containing a lower layer of 5 mL of 80% Percoll^®. Density gradient centrifugation was conducted at 420 × g for 20 minutes, and the mononuclear cells situated in the intermediary layer were meticulously extracted using a 1 mL pipette and then washed with 2% FBS/PBS (22).

Hepatic lymphocytes were isolated through mechanical dissection. Initially, the liver was excised from the euthanized mice, delicately filtered through a 200-gauge stainless-steel mesh, and suspended in Roswell Park Memorial Institute 1640 medium supplemented with 100 mL/L fetal calf serum (FCS, 10%). The suspension of liver cells was centrifuged at 1,500 rpm. The resulting pellet was then suspended in a 40% Percoll solution, layered onto an equivalent volume of 60% Percoll solution, and subjected to centrifugation at 2,000 rpm for 20 minutes at 22°C. Subsequently, cells were retrieved from the Percoll interface (buffy coat), pelleted through centrifugation, and rinsed twice with PBS containing 10% FCS prior to use (23).

The stained cells were analyzed using a FACSCanto II, and the resulting data were processed using FlowJo version 10.9.0 (TreeStar, Ashland, OR, USA). The following antibodies from eBioscience (San Diego, CA, USA) were used for gating of M1 and M2 macrophages: APC-CD45.2 (17045482; clone: 104, 1/50), PE-F4/80 (12480182; clone: BM8, 1/50), APC-Cy7-CD11b (47011282; clone: M1/70, 1/50), FITC-CD206 (MA516870; clone: MR5D3, 1/50), and PE-Cy7-CD11c (25011482; clone: N418, 1/50) (24) ( Supplementary Figure S1 ).

The following antibodies from eBioscience were used for gating ILCs: biotin-CD3e (100304; clone: 145-2C11, 1/200), biotin-CD45R/B220 (103204; clone: RA3–6B2, 1/200), biotin-Gr-1 (108404; clone: RB6-8C5, 1/200), biotin-CD11c (117304; clone: N418, 1/200), biotin-CD11b (101204; clone: M1/70, 1/200), biotin-Ter119 (116204; clone: TER-119, 1/200), biotin-FceRIa (134304; clone: MAR-1, 1/200), FITC-Streptavidin (405202; 1/500), PE-Cy7-CD127 (135014; clone: A7R34, 1/100), Pacific Blue-CD45 (103116; clone: 30-F11, 1/100), PE-GATA-3 (clone: TWAJ, 1/50), APC-RORγ (clone: AFKJS-9, 1/50), and Fixable Viability Dye eFluor 780 (1/400) (25, 26) ( Supplementary Figure S2 ).

Short-chain fatty acid (SCFA) concentrations were assessed using gas chromatography–mass spectrometry (GC/MS) with an online solid-phase extraction (SPE) technique. In the SPE-GC system SGI-M100 (AiSTI Science, Wakayama, Japan), the sample was introduced into the vial and placed on the autosampler tray, and subsequently, SPE and injection into the GC/MS system were executed automatically. Solid phase stratification was accomplished using flash-SPE ACXs (AiSTI Science). For sample processing, a 50 µL aliquot of each extract was deposited onto the solid phase and then subjected to rinsing with a 1:1 mixture of acetonitrile and water. Following this step, the samples were dehydrated using acetone, impregnated with a 4 μL N-tert-butyldimethylsilyl-N-methyltrifluoroacetamide (MTBSTFA)-toluene solution (1:3 ratio) and eluted using hexane post-derivatization on the solid phase. The injection into the GC/MS system was carried out using a programmed temperature vaporizer (PTV) injector, LVI-S250 (AiSTI SCIENCE), with an initial temperature of 150°C for 0.5 minutes, gradually increasing to 290°C at a rate of 25°C/minute, and held for 16 min. The samples were injected into a capillary column, Vf-5ms (30 m × 0.25 mm (inner diameter) × 0.25 μm (membrane thickness); Agilent Technologies). The initial column temperature was 60°C for 3 minutes, increased by 10°C/minute to 100°C, further increased by 20°C/minute to 310°C, and finally held at 310°C for 7 minutes. Sample injection was performed in split mode at a ratio of 20:1. Each SCFA was identified using scan mode (m/z 70–470) (n=10) (27).

Microbial DNA was isolated from frozen appendicular fecal samples using the QIAamp DNA Feces Mini Kit (Qiagen, Venlo, Netherlands), in accordance with the manufacturer’s protocols. Subsequently, the V3-V4 region of the 16S rRNA gene was amplified from the DNA using a bacterial universal primer set (341F and 806R). PCR was conducted using EF-Taq (Korea, Solgent), with 20 ng of genomic DNA serving as the template in a 30 µL reaction mixture. The thermocycling parameters were as follows: initial activation of Taq polymerase at 95°C for 2 minutes, followed by 35 cycles of denaturation at 95°C, annealing at 55°C, and elongation at 72°C for 1 minute each, and a final extension step at 72°C for 10 minutes. The PCR products were purified on a multiscreen filter plate (Millipore Corp., Billerica, MA, USA). A MiSeq sequencer (Illumina, CA, USA) was used for 16S rRNA sequencing according to the manufacturer’s instructions (Macrogen, Seoul, Korea). For quality filtering of the sequences, QIIME version 1.9.1 (28) was used. Barcodes or primers with scores of less than 75% were excluded from the files. Operational taxonomic units (OTUs) with 97% sequence similarity were determined using the UCLUST algorithm (29). Moreover, BLAST (UNITE database v9.0, released in 2023. Available online: https://unite.ut.ee) was applied for the taxonomic classification of 16S rRNAs, using the UNITE sequence set of the Greengenes core set aligned with UCLUST and ITS.

Kyoto Encyclopedia of Genes and Genomes (KEGG) ortholog abundances were predicted using Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt2) software (30).

The relative abundance of phyla within different groups was assessed using one-way ANOVA with a Holm–Šídák multiple-comparison test. Alpha diversity, representing diversity within individual samples, was evaluated using the Chao1 (31), Shannon (32), and Gini–Simpson indices (33).

The relative abundance of bacterial genera between groups was assessed using linear discriminant analysis (LDA) coupled with effect size measurements (LEfSe) (http://galaxy.biobakery.org/, accessed November 16, 2023) (34). Using a normalized relative abundance matrix, taxa exhibiting statistically significant differences in abundance were discerned using LEfSe, with the effect size of these features being determined via LDA. The significance level was set at p ≤ 0.05 using the Wilcoxon rank-sum test, and an effect size threshold of 2 was applied to all biomarkers.

Data were analyzed using JMP software (version 14.0; SAS, Cary, NC, USA). One-way ANOVA with Holm–Šídák’s multiple comparisons test was used to compare the results of different groups. Data are presented as mean ± SD values. Statistical significance was set at P < 0.05. Figures were created using the GraphPad Prism software (version 10.1.2; San Diego, CA, USA).

From 8 to 14 weeks of age, body weight, and food and water intake were monitored in ND, ND + MP 1000 µg/L, and ND + MP 5000 µg/L mice. There were no discernible disparities in body weight between mice in these groups ( Figure 1B ).

There were no differences in food or water intake between the three groups ( Figures 1C, D ). Intestinal permeability was evaluated using fluorescent-labeled dextran. There existed no significant difference in plasma FITC expression between ND mice, ND + MP 1000 µg/L mice, and ND + MP 5000 µg/L mice ( Figure 1E ). We also assessed the immunostained area of the small intestine for claudin-1, a crucial membrane protein constituting tight junctions. While there was a tendency for the Claudin-1 staining area to decrease, no statistically significant difference was discerned among ND mice, ND + MP 1000 μg/L mice, and ND + MP 5000 μg/L mice ( Figures 2A, F ).

We investigated serum levels of hepatic enzymes and lipids. There were no significant differences in serum AST, ALT, and total cholesterol levels between ND mice, ND + MP 1000 µg/L mice, and ND + MP 5000 µg/L mice, but TG and NEFA were significantly higher in ND + MP 1000 µg/L and ND + MP 5000 µg/L mice than in ND mice ( Figures 1F–J ).

Representative histological images of the jejunum are shown in Figure 2A . The villus height and width were significantly lower in ND + MP 1000 μg/L and ND + MP 5000 μg/L mice than in ND mice ( Figures 2B, C ); therefore, the decrease in villus height and width due to MPs tended to be smaller with increasing concentrations. Furthermore, crypt depth was reduced in ND + MP1000 µg/L and ND + MP 5000 µg/L mice than in ND mice ( Figure 2D ). The total number of goblet cells was determined using the PAS-stained images. The number of goblet cells was significantly lower in ND + MP 1000 µg/L and 5000 µg/L mice than in ND mice ( Figure 2E ).

We also measured the number of cells involved in innate immunity in the intestinal LPL using FACS. Although there were no significant changes in ILC1, ILC3, and M1/M2 macrophage ratios in the intestinal LPL due to MPs exposure ( Figures 2G–I ), the number of NK cells was higher in ND + MP 5000 µg/L mice than in ND mice ( Figure 2J ).

The concentration of SCFA in the feces was measured using GC/MS and was expressed as the ratio to the mean concentration in the feces of ND mice. SCFA such as acetic acid ( Figure 2K ), propanoic acid ( Figure 2L ), and butanoic acid ( Figure 2M ) were not significantly different between any of the groups.

Differences in the expression of several transporters and metabolism-related genes in the small intestine were determined using microarray analysis. The expression of inflammation-related genes such as FAM107a, Il4ra, Ager, Mboat4, Il10, and Zbtb16 was higher in ND + MP 5000 µg/L mice than in ND mice ( Figure 3A ).

The expression of glucose transporter genes, such as Ptpn1 was higher, whereas that of Rbpj and Slc2a10 was lower ( Figure 3B ) in ND + MP 5000 µg/L mice than in ND mice. The expression of amino acid transporter genes, such as Ppp1r15a, Slc1a4, and Slc7a1, was higher, whereas that of Rbpj and Slc2a10 was lower in ND + MP 5000 µg/L mice than in ND mice ( Figure 3C ). The expression of fatty acid transporter gene Trpv4 was higher whereas that of Pparg and Thy1 was lower in ND + MP 5000 µg/L mice than in ND mice ( Figure 3D ).

Gene expression in the small intestines of each group was compared using real-time PCR. The expression of genes associated with inflammation, such as Tnfa and Il6, was higher ( Figures 3E, F ), while that of Il22 related to intestinal immunity (35) was significantly lower ( Figure 3H ) in ND + MP 5000 µg/L mice than in ND mice. No significant differences were observed in the expression of Il1b associated with inflammation and Pept1, Cd36, and Sglt1 related to nutrient absorption, in the small intestine among any of the groups ( Figures 3G, I-K ).

Liver weight / body weight was not significantly different between any of the groups ( Figure 4A ). Liver histology was analyzed to investigate the effects of MPs on the liver. Representative histological images of the liver are shown in Figure 4B . The NAFLD activity score and Oil Red staining area were higher in ND + MP 5000 μg/L mice than in ND mice and ND + MP 1000 μg/L mice. However, there were no significant differences between ND + MP 1000 μg/L mice and ND mice ( Figures 4C, D ).

The number of cells involved in innate immunity in the liver was measured by FACS. There were no significant differences in the number of ILC1 and ILC3, the ratio of M1/M2 macrophages, or NK cells due to MPs exposure between any of the groups ( Figures 4E–H ).

The expression of genes related to inflammation and hepatic steatosis in the liver of each group was compared using real-time PCR. Compared to ND mice, the expression of inflammation-related genes Tnfa and Il6 was not different in ND + MP 1000 µg/L mice but significantly higher in ND + MP 5000 µg/L mice ( Figures 5A, B ). In contrast, no differences were observed in the expression of inflammation-related genes Il1b and Mcp1 between any of the groups ( Figures 5C, D ). Further, the expression of hepatic steatosis-related genes Scd1 and Fasn in any of the groups, while that of Elovl6 was higher in ND + MP 5000 µg/L mice than in ND mice ( Figures 5E–G ).

We investigated the relative abundance of gut microbiota using 16s rRNA sequencing ( Figure 6A ). There was a decrease in the relative abundance of phylum Bacteroides between ND mice and ND + MP 5000 µg/L mice. There were no differences in OTUs between ND mice, ND + MP 1000 µg/L mice, and ND + MP 5000 µg/L mice ( Figure 6B ). However, in terms of alpha diversity indices, the Shanon index was significantly lower, and the Chao1 index and Gini-Simpson index showed a decreasing trend in ND + MP 5000 µg/L mice than in ND mice. No differences were observed in the alpha diversity indices between ND mice and ND + MP 1000 µg/L mice ( Figures 6C-E ).

Furthermore, taxa with considerably different abundances between ND mice and ND + MP 1000 µg/L mice, and ND mice and ND + MP 5000 µg/L mice were identified using the LEfSe algorithm. Five taxa (including the genus Paraprevotella, the family Paraprevotellaceae, and the family Erysipelotrichaceae) were significantly over-represented, while 22 taxa (including the genus Akkermansia, the family Verrucomicrobiaceae, the genus Prevotella, the family Prevotellaceae, the family Rikenellaceae, the order RF32, and the genus Desulfovibrio) were under-represented in ND + MP 1000 µg/L mice than in ND mice ( Figure 6F ). Taxa (including the family Coriobacteriaceae, the genus Biophilia, the genus Allobaculum, and the phylum Firmicutes) were over-represented, while seven taxa (including the order RF32 and the genus Desulfovibrio) were under-represented in ND + MP 5000 µg/L mice than in ND mice ( Figure 6G ).