C57BL/6 mice (6–8 weeks old) were purchased from the Chinese University of Hong Kong. The animal welfare and experiments strictly followed the procedures approved by the Ethics Review Committee of Macau University of Science and Technology. The mice were housed in a 12 h dark-light cycle facility and had free access to food (LabDiet, USA) and Milli-Q water. A total of 12 mice were equally divided into 2 groups, i.e., the control group and the HA group. HA was purchased from Aladdin (H107141, Shanghai, China). Mice were daily gavage with 50 mg/kg HA or Milli-Q water for 15 consecutive days. The body weight, food, and water consumption of the mice were recorded every 5 days. Fecal samples were collected individually on 0 and 15 days. All samples were stored at −80°C for later experiments. The experimental scheme was showed in Figure 1A. Mice were anesthetized and sacrificed with pentobarbital sodium.

The rotarod test was performed following our previous study (24). Briefly, mice were subjected to a 1-day learning period at a constant speed of 15 rpm, then scored at an accelerating speed of 20 rpm. The retention time on the rotarod was recorded from three consecutive attempts. The total number of falls from the rotarod within 15 min was recorded for each mouse. The rotarod test was conducted every five days.

The fecal DNA was extracted using QIAamp DNA Stool Mini Kit (QIAGEN, Germany) followed the manufacturer's manual, and analyzed for highly conserved ERIC regions with a pair of ERIC primers sequences: ERIC1 (5′-ATGTAAGCTCCTGGGGATTCAC-3′) and ERIC2 (5′-AAGTAAGTGACTGGGGTGAGCG-3′). PCR conditions were set following the previous description (25). ERIC-PCR products were isolated in 2% agarose gel at 100 V for 45 min. Bands were visualized with the Gel Doc XR+ system and digitized for clustering analysis through SIMCA-P 14.0 tool (Umetrics, Umea, Sweden) with a confidence level 95% (p < 0.05).

DNA samples were sequenced for 16S rRNA gene using Illumina MiSeq platform (Illumina, San Diego), targeting the V3–V4 region by barcoded 515F and 806R universal primers and processed as previously described (26). In short, dual-index barcodes and Illumina sequencing adapters were used to join the reads by a limited PCR cycle. After purification with Agencourt AMPure beads (Agencourt, USA), Nextera XT protocol was used for library normalization. And then, samples were loaded into a single flow cell for sequencing on the MiSeq sequencing platform (Illumina, San Diego) following the manufacturer's introduction. Clusters were auto-generated and paired-end sequenced with dual index reads in a single run with a read length of 2 × 300 bp. PANDAseq was used to collect paired end sequences, then Raw FASTQ files were obtained (27). Sequences were trimmed for primers and barcodes. The cleaned sequences were clustered at k = 10 (97% similarity) by deletion of chimeras and singleton reads (28, 29). Finally, OTUs were classified using BLASTn against a curated database derived from NCBI.

SOD, MDA, and glycogen were tested including the serum, liver, and muscle of the mice. The procedures were carried out according to the manufacturer's manuals (Nanjing Jiancheng Bioengineering Institute, China). SOD was assayed using a hydroxylamine reaction and detected at 550 nm. MDA was determined in a thiobarbituric acid reaction and measured at 532 nm. Glycogen was determined at 620 nm using microspectrometer.

A rat H9c2 cardiomyocyte cell line, obtained from American Type Cultural Collection (CRL1446, ATCC, USA), was cultured in Dulbecco's modified Eagle's medium (DMEM, Gibco, Oklahoma, USA) supplemented with 10% fetal bovine serum (FBS, Gibco, Oklahoma, USA) and 1% v/v penicillin/streptomycin (Gibco, Oklahoma, USA) at 37°C in a 5% CO₂ humidity environment. Desulfovibrio vulgaris was cultured in Brain Heart Infusion (OXOID, England) broth in the DWS A25 anaerobic chamber (Whitely, England) for 24 h. The metabolic products of D. vulgaris (MPDV) in the culture medium were centrifuged at 5,000 rpm to collect the supernatant for further experiments.

H9c2 cells were seeded into μ-Slide 8-well glass bottom plate (#80826, ibidi, Germany) at a total number of 10,000 per well. After treatments of MPDV (0, 10, 20, 40%), cells were incubated with 100 nM MitoTracker Red CMXRos (Beyotime, China) at 37°C for 20 min. Then cells were washed using Dulbecco's Modified Eagle Medium (DMEM) for five times. The images of mitochondrial morphology were captured by confocal microscopy (LSM 800, ZEISS, USA) equipped with a 63× oil immersion objective. Red fluorescence represented mitochondria stained by MitoTracker Red CMXRos.

H9c2 cells were seeded into μ-Slide 8-well glass bottom plate (#80826, ibidi, Germany) at a total number of 1 × 10⁴ per well. To measure the mitochondrial membrane potential (MMP), JC-1 mitochondrial membrane potential assay kit (Beyotime, China) was used according to the manufacturer's protocol. After treatments of MPDV (0, 10, 20, 40%), cells were incubated with 10 μg/ml JC-1 at 37°C for 20 min. The fluorescent images of JC-1 were captured by red and green fluorescence, respectively with Confocal Microscopy (LSM 800, ZEISS, USA) equipped with a 63× oil immersion objective. For the quantification of MMP, H9c2 cells were seeded in 96 well black/clear bottom plates. Ex 488/Em 535 nm and Ex 550/Em 600 nm were used and MMP was calculated by the ratio of red to green fluorescence.

H9c2 cells were seeded into μ-Slide 8-well glass bottom plate (#80826, ibidi, Germany) at a total number of 1 × 10⁴ per well. To measure the Reactive Oxygen Species (ROS) level, DCFH-DA ROS assay kit (Beyotime, China) was used according to the manufacturer's protocol. After treatments of MPDV (0, 10, 20, 40%), cells were incubated with 5 μM DCFH-DA at 37°C for 30 min. The fluorescent images of DCFH-DA were captured by green fluorescence with Confocal Microscopy (LSM 800, ZEISS, USA) equipped with a 63× oil immersion objective.

The ATP level of H9c2 cells, after treatments of MPDV (0, 10, 20, 40%), was measured by using a luminescent ATP assay kit (Beyotime, China) according to the manufacturer's protocol. The levels of ATP were detected by the Multi-Mode Detection Platform and calculated as the ratio to the control group (set as 1.0), respectively.

H9c2 cells were seeded into μ-Slide 8-well glass bottom plate (#80826, ibidi, Germany) at a total number of 1 × 10⁴ per well. To measure the calcium (Ca²⁺) ion level, Fluo-4 AM calcium assay kit (Beyotime, China) was used according to the manufacturer's protocol. After treatments of MPDV (0, 10, 20, 40%), cells were incubated with 2 μM Fluo-4 AM probe at 37°C for 30 min. The fluorescent images of Fluo-4 AM were captured by green fluorescence with confocal microscopy (LSM 800, ZEISS, USA) equipped with a 63× oil immersion objective.

Cytotoxicity was assessed by cell viability assay. Cell viability was measured by using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay (MTT assay kit, Beyotime, China). The absorbance of 570 nm was evaluated by a Multi-Mode Detection Platform (SpectraMax Paradigm, Molecular Devices, USA). Cell viability was calculated as the ratio to the control group (set as 1.0), respectively.

The functional prediction of GM was performed using FAPROTAX on the Cloud-platform of Biomicroclass (http://www.cloud.biomicroclass.com/CloudPlatform). SPSS version 22.0 and GraphPad prism version 9.0 were used for statistical analysis and graphical presentation. Kolmogorov–Smirnov test was used to determine the data normality. Student t-test, Kruskal-Wallis (non-parametric data) tests, and Two-way ANOVA (for parametric data) were employed to determine the significant changes. Dunn-Bonferroni test and Bonferroni test were performed for parametric and non-parametric multiple comparison p-values correction, respectively.

To investigate the effect of HA on the mice's GM profile, 6–8 weeks old C57BL/6 mice were orally fed daily with 50 mg/kg HA for 15 days. Fecal samples were collected on Days 0 and 15 from individual mice for gut microbial DNA extraction (Figure 1A). After 15 days of treatment with HA, a slight decrease in food consumption and body weight were detected in the HA group compared with the control group, but the differences are not statistically significant. Comparing the body weight of the Ctrl vs. HA groups, the P-values for 0-, 5-, 10-, and 15-day were >0.9999, 0.2284, 0.6583, and 0.2386, respectively) (Figure 1B). ERIC-PCR and gel electrophoresis were performed as previously described (25). The resulting data were plotted with PLS-DA to display the similarity of GM profiles between the HA-treated and the control mice. Results showed a similar GM composition between the control and HA groups on day 0. However, by the end of the experiment, HA groups displayed an obvious different GM profile compared with the control group (Figure 1C; Supplementary Figures S1, S2).

To further analyze the change of GM induced by treatment with HA, 16S rRNA gene sequencing was used to profile the GM composition from fecal genomic DNA. The alpha diversity analysis showed that HA treatment decreased the alpha diversity (observed and Chao1 indexes) and richness (Shannon index) compared with the control group (Figure 2A). Compared with the control group, HA treatment significantly altered the relative abundance of the phyla Bacteroidetes, Cyanobacteria, and Spirochaetes (Figure 2B). Meanwhile, in family taxa, HA mainly increased Porphyromonadaceae, Eubacteriaceae, and decreased Verrucomicrobioacease, Rikenellaceae, Prevotellaceae, Gloeobacteraceae (Figure 2C). Moreover, in genus taxa, HA significantly increased the relative abundance of Barnesiella, Tannerella, and Eubacterium, while reduced Rikenella and Desulfovibrio (Figure 2D). The change of GM genus taxa was associated with the increase of species Barnesiella spp., Tannerella spp., and Eubacterium spp., whereas the decrease of Rikenella spp. and Desulfovibrio sp. (Figure 2E).

Based on the published literature from the NCBI database (PubMed), we grouped the altered species into potential pathogenic and beneficial bacteria (Supplementary Tables S1, S2). The potential pathogens suppressed by HA treatment were mainly the species from the genus Alistipes associated with inflammation, the sulfate-reducing bacteria (SRB) Desulfovibria, and the Helicobacter associated with gastrointestinal inflammation and cancer (Figure 3A).

On the other hand, HA could promote the growth of certain beneficial bacteria, especially the short-chain fatty acids (SCFAs)-producing bacteria and xylan/cellulose-degrading bacteria (Figure 3B). Among the SCFAs-producers are Bacteroides uniformis, Bifidobacterium cheorinum, Eubacterium rectale, and Eubacterium spp. The SCFAs-producers play an important role to break down indigestible foods and convert to acetate, propionate, butyrate, and valerate for gut-healthy maintenance (30). Meanwhile, HA treatment also stimulated the growth of xylan/cellulose degrading bacteria, including Bacteroides xylanolyticus, Blautia producta, and Clostridium sulfatireducens.

Furthermore, FAPROTAX was used to evaluate the impact of HA on GM metabolic pathways. The findings showed that HA increased the GM xylanolysis function and nitrate reduction. Simultaneously, HA reduced the accumulation of nitrite respiration and nitrate ammonification. More importantly, results also showed that HA could reduce the accumulation of the pathogens-associated functional pathways, including gastroenteritis and diarrhea (Figure 3C).

The above positive modulating effects on GM triggered our investigation of the health status of the HA-treated mice. We first took on the rotarod test to evaluate the potential anti-fatigue property of HA. C57BL/6 mice at 6–8 weeks of age were first trained on the rotarod for 5 min. After 30 min rest, mice were tested on the rotarod for 15 min each time for three consecutive trials for each mouse. The retention time on the rotarod from all three attempts was recorded and analyzed. The results showed that HA-treated mice significantly yielded higher scores on the retention time than the untreated control (Figure 4A). Moreover, the results also showed that HA-treated mice fell from the rotarod less than the control mice (Figure 4B). The fatigue-associated biomarkers, SOD, MDA, and glycogen were measured in the serum, liver, and muscle samples. The results showed that HA significantly increased the level of glycogen in the liver. Meanwhile, the levels of SOD in the serum and muscle were effectively increased in the HA group. While the level of MDA in the serum and liver was markedly decreased after 15 days of HA treatment compared to the control group (Figure 4C).

As mentioned above, HA reduced the relative abundance of Desulfoviibrio spp. in HA-treated mice. We wonder whether the metabolic products of this group of bacteria might have harmful effects on mitochondrial functions. We then chose D. vulgaris, the representative Desulfovibrio bacteria for mitochondria functional tests in cardiomyocytes. We firstly used mitotracker staining dye to evaluate the mitochondrial morphology exposed to MPDV at the dosages of 0, 10, 20, and 40% in H9c2 cardiomyocytes. Under normal conditions, mitochondria displayed thread-like or punctate morphology due to their dynamic fusion-fission balance. However, MPDV treatment significantly changed mitochondrial morphology, contributing to punctate morphology to a large extent. The presence of MPDV caused mitochondrial fragmentation in a dose-dependent manner (Figure 5A). Mitochondrial membrane potential (ΔΨm) is a sensitive and important index to evaluate mitochondrial health. MPDV (40%) treatment significantly caused the decrease of ΔΨm when compared with the control group by using JC-1 staining dye, suggesting MPDV induced mitochondrial depolarization (Figures 5B,C). As mitochondrial membrane potential was an important factor to generate ATP along the electron transport chain (ETC), decreasing ΔΨm could lead to excessive ROS generation. Application of DCFH-DA, a ROS-specific staining dye, showed that MPDV treatment enhanced ROS burden in a dose-dependent manner (Figures 5D,E). Furthermore, the decrease of ΔΨm probably caused the reduction of ATP. Therefore, the assessment of ATP content was used to evaluate MPDV in mitochondrial biological functions. The luminescent assay showed that MPDV decreased the ATP content in a dosage-dependent manner (Figure 5F). we also evaluated the Ca²⁺ level in treated cells since defects of mitochondrial functions might trigger mitochondrial and cytosolic Ca²⁺ overloading. Therefore, we applied Fluo-4 AM staining dye to measure Ca²⁺ level. The result showed that strong signals of Ca²⁺ were detected after MPDV treatment in the whole cell (Figures 5G,H). Finally, MPDV also decreased cell viability, suggesting its cytotoxic effects (Figure 5I). Taken together, the presence of MPDV could damage the mitochondrial functions and exert cytotoxic effects on H9c2 cardiomyocytes.