Sixteen C57BL/6 mice (female, 6–8 weeks) of SPF grade were purchased from SiPeiFu Biotechnology Co. Ltd., Beijing, China. All mice were kept in separate cages in a temperature-controlled environment. The light–dark cycle was 12/12 h, and they had free access to food and water. After 1 week of adaptive feeding, mice were randomly divided into two groups, the control group (Control) and the simulated microgravity group (SMG), each group contained eight mice. Hindlimb suspension of mice by the tail model was followed to create a ground-based model of microgravity (Globus and Morey-Holton, 2016). The mice were suspended by tail with hindlimbs suspended at 15–30° horizontally, ensuring that the hindlimbs do not touch the cage floor. Mice in the control group were raised in identical cages without hindlimb unloading. Thirty-six fecal samples from 6 mice of each group were collected at the end of week 2, 4, and 8, and stored at −80°C until used for 16S rRNA gene sequencing. Mice were sacrificed at the end of week 8. Then, cecal contents from each mouse were obtained and saved at −80°C until used for metabolic analysis. All efforts were made to minimize the discomfort of the animals. The whole experimental design is shown in revised Supplementary Figure S1.

Total bacteria DNA was extracted from approximately 220 mg of stool samples using the CTAB method following the standard protocol and then sequencing libraries were created. Briefly, the V3-V4 region of the 16S rRNA genes was amplified using specific primer (341F, 5ˊ-CCTAYGGGRBGCASCAG-3ˊ, 806R, 5ˊ-GGACTACNNGGGTAT CTAAT-3ˊ) with the barcode. The sequencing libraries were generated using TruSeq® DNA PCR-Free Sample Preparation Kit (Illumina, USA) according to the manufacturer’s instructions, and index codes were added. The library quality was assessed on the Qubit@ 2.0 Fluorometer (Thermo Scientific). Finally, the library was sequenced on an Illumina NovaSeq platform and 250 bp paired-end reads were generated.

The analysis was conducted by following the “Atacama soil microbiome tutorial” of Qiime2 along with customized program scripts.¹ Briefly, raw data FASTQ files were imported into the format which could be operated by the QIIME2 system. Demultiplexed sequences were quality filtered, trimmed, de-noised, and merged. Then the chimeric sequences were identified and removed using the QIIME2 dada2 plugin to obtain the amplicon sequence variant feature table (ASV) (Callahan et al., 2016). To generate the taxonomy table, the QIIME2 feature-classifier plugin was then used to align ASV sequences to a pre-trained GREENGENES 13_8 99% database (Bokulich et al., 2018). Kruskal–Wallis and LEfSe were employed to identify the bacteria with different abundance among samples and groups. Alpha diversity indices were calculated using qiime2 diversity, with the observed OTU to calculate species richness and Shannon index to calculate species diversity. Beta diversity distance measurements were performed to investigate the structural variation of microbial communities across samples and then visualized via nonmetric multidimensional scaling (NMDS). Co-occurrence analysis was performed by calculating Spearman’s rank correlations between predominant taxa and the network plot was used to display the associations among taxa.

Fecal samples (100 mg) were individually grinded with liquid nitrogen and the homogenates were resuspended with prechilled 80% methanol by vigorous vortexing. The samples were incubated on ice for 5 min and then centrifuged at 15,000 g for 20 min at 4°C. Some of the supernatant was diluted to a final concentration of 53% methanol by LC–MS grade water. The samples were transferred to a new Eppendorf tubes and centrifuged at 15,000 g, 4°C for 20 min. Finally, the supernatant was injected into the LC–MS/MS system for analysis (Want et al., 2013).

UHPLC–MS/MS analysis was performed using a Vanquish UHPLC system (Thermo Fisher, Germany) coupled with an Orbitrap Q Exactive^TM HF mass spectrometer (Thermo Fisher, Germany) in Novogene Co., Ltd. (Beijing, China). Samples were injected onto a Hypesil Gold column (100 × 2.1 mm, 1.9 μm) using a 17-min linear gradient at a flow rate of 0.2 ml/min. The eluents for the positive polarity mode were eluent A (0.1% FA in Water) and eluent B (Methanol).The eluents for the negative polarity mode were eluent A (5 mM ammonium acetate, pH 9.0) and eluent B (Methanol). Q Exactive^TM HF mass spectrometer was operated in positive/negative polarity mode with spray voltage of 3.5 kV.

The raw data files generated by UHPLC–MS/MS were processed using the Compound Discoverer 3.1 (CD3.1, Thermo Fisher) to perform peak alignment, peak picking, and quantitation for each metabolite. The normalized data were used to predict the molecular formula based on additive ions, molecular ion peaks, and fragment ions. And then peaks were matched with the mzCloud,² mzVaultand MassListdatabase to obtain the accurate qualitative and relative quantitative results. Statistical analyses were performed using the statistical software (Rohart et al., 2017). When data were not normally distributed, normal transformations were attempted by using area normalization method. Random Forest machine learning analysis was constructed through tools of bioincloud.tech and Mean Decrease Accuracy is used to measure the importance of metabolites in distinguishing groups in random forests. These metabolites were annotated using the KEGG database, HMDB database, and LIPIDMaps database. Furthermore, MetOrigin database³ was used to discriminate the origins of identified metabolites.

Statistical analysis of the results was performed by using the statistical software R (R version R-3.4.3) and GraphPad Prism 7.0 software. Statistical significance between SMG and control groups was evaluated by Mann–Whitney test and Kruskal–Wallis test. Differences between the groups were considered statistically significant at the 5% level (p < 0.05). All p -values for the statistical tests of metabolite variations in the two groups were corrected for multiple testing using Benjamini-Hochberg false discovery rates (FDR) method. Significant differences were indicated by *p < 0.05, **p < 0.01, ***p < 0.001. Correlations between the relative abundance of differential bacteria (p < 0.05, average relative adundance˃0.01%) and the concentration of major metabolites (FDR < 0.05) involving in major differential metabolic pathways were computed with Spearman’s test using Oebiotech tools⁴ and MetaboAnalyst 5.0.⁵

We performed the V3-V4 region of the 16S rRNA gene sequencing to assess whether the long-term SMG altered the intestinal flora at different time points after 8 weeks of SMG. The 36 fecal samples yielded a total of 3,070,422 raw reads corresponding to 7,719 OTUs. The data were normalized by established a depth of 63,210 reads per sample., Alpha diversity analysis showed there were no significant differences in the indices of observed OTU and Shannon index (p > 0.05, Figure 1A) compared to the control mice at weeks 2, 4 and 8. These results suggested that SMG did not affect the richness and diversity of gut microbiota during the 8 weeks of SMG. Beta diversity reveals that SMG had a statistically significant effect on community composition and structure at weeks 2, 4, and 8 (p < 0.01, Figures 1B,C). Non-metric multidimensional scaling (NMDS) based on the Bray–Curtis distance revealed a significant differences in microbial structure between the control group and the SMG group (p < 0.01, Figure 1B). Better separation of SMG and control mice was observed at the genus level throughout the 8 weeks of SMG (Figure 1C). In addition, the co-network analysis showed that the microbial associations differed between the SMG and the control groups at different time points (Supplementary Figure S3). For example, the relative abundance of Akkermansia is positively related to the beneficial bacteria Lactobacillus and negatively associated with Bacillus in the control mice at week 2. However, the relative abundance of Akkermansia is negatively relative with Helicobacter in SMG mice. Moreover, the difference between the two groups seemed more apparent in week 8.

We also performed multivariate analysis to identify differential abundant taxa in SMG mice as compared to the control group at three-time points. At phylum level, fecal microbiota in both groups dominated by bacterial phyla Firmicutes, Bacteroidetes, and Proteobacteria (Figure 2A). The proportions of Bacteroidetes and Firmicutes significantly declined in the control group at week 8 (Supplementary Figure S2B). However, there were no significantly differences in abundance of Bacteroidetes and Firmicutes in the SMG groups (Supplementary Figure S2C). These results imply that long-term SMG may lead to spices belonging to Bacteroidetes prone to growth. SMG significantly decreased the relative abundance of Verrucomicrobia at a different time points, and Actinobacteria and Deferribacteres increased significantly at week 8 (Figure 2B). At the genus level, we observed that SMG showed a significant decrease in the relative abundance of Allobaculum and increased Bacteroides throughout the 8 weeks of SMG (Figures 2B,C, 3). At the species level, Desulfovibrio_c21_c20, Akkermansia_muciniphila, and Ruminococcus_flavefaciens were decreased considerably in SMG mice at week 2. At week 4, only Akkermansia_muciniphila were significantly decreased. At week 8, Bacteroides_uniformis and Faecalibacterium_prausnitzii were declined, while Ruminococcus_gnavus and Mucispirillum_schaedleri were significantly increased (Figures 2D). By performing LEFSe at the genus level, we identified two taxa (Bacteroides and Gracilibacter) characteristic of the SMG mice at week 2 and 8, while taxa such as Akkermansia and Allobaculum were enriched in the control group (Figure 3A). At week 4, we identified five taxa (Bacteroides, Odoribacter, Sporosarcina, Gracilibacter, and Clostridium) enriched in the SMG group (Figure 3B). Nine taxa were identified in SMG group at week 8, including Bacteroides, Prevotella, Odoribacter, Rothia, Mucispirillum, Anaerotruncus, Clostridium, Adlercreutzia, and Bilophila (Figure 3C). Together, these results suggested that Bacteroides, Allobaculum, and Akkermansia were the most affected genera throughout the simulated microgravity process. In line with previous studies (Wang et al., 2020; Sun et al., 2021), SMG could significantly reduce the relative abundance of beneficial bacteria, such as Akkermansia and Allobaculum. Our findings confirm that SMG induces a marked disturbance in the intestinal flora at different times, and longer SMG has a greater effect.

SMG treatment significantly altered the intestinal microbiota, raising our interest in investigating whether and how it could affect microbial metabolites, which could modulate circulating metabolites. We evaluated the intestinal metabolomics profiles from 15 samples of intestinal contents after 8 weeks of SMG. PCA and PLS-DA scores plots revealed clear significant differences in the composition of intestinal microbiota between the SMG and control groups (Figures 4A,B). A total of 1,193 metabolic features were identified across 15 samples. The volcano plots showed 275 features (23.05%) significantly changed in SMG (p<0.05) mice. Among them, 129 metabolites were upregulated and 146 downregulated (Figure 4C). On the super class level, the relative abundance of organic acids, peptide and carbohydrate significantly decreased after 8 weeks of SMG (Supplementary Table S1).Overall, these data indicate that long-term SMG significantly alter the intestinal metabolites, which may mainly contribute to disturbance in circulatory metabolism. We then used Random Forest machine learning analysis to determine the most discriminatory metabolites from SMG and control groups. The top 30 different metabolites including but not limited to 19(R)-Hydroxy-prostaglandin E2, DL-Tryptophan, 2,6-Dihydroxybenzoic acid, Tetrahydroaldosterone, Suberic acid, N-Carbamyl-L-glutamic acid, 5-Hydroxyindole, Prostaglandin D3, and 11-keto Testosterone (CRM), which were primarily from the amino acid, lipid, and organic acids (Figure 4D).

Further, to investigate the potential role of SMG-related metabolites, we annotated metabolic features by using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. The SMG mice exhibited distinct patterns of metabolites involved in 36 metabolic pathways, including tryptophan metabolism, Tyrosine metabolism, alanine, aspartate and glutamate metabolism, and caffeine metabolism (Supplementary Table S2). Metabolites that were increased in abundance in SMG reflected pathways mainly involved in tryptophan metabolism (L-tryptophan, 2-ketoadipic acid, and 5-hydroxytryptophan), histidine metabolism (carnosine, histamine and 1-methylhistidine), cysteine and methionine metabolism (S-adenosylmethionine, L-cystathionine and 5′-deoxy-5′(methylthio) adenosine), and arginine and proline metabolism (S-adenosylmethionine, oxaceprol, and 4-acetamidobutyric acid) (p < 0.05) (Figures 5A,C). Metabolites whose abundance was decreased in SMG reflected pathways involved in steroid hormone biosynthesis (etiocholanolone, hydrocortisone, and cortisone) (p < 0.05) (Figures 5B,C). Overall, these results revealed that long-term SMG significantly affects the intestinal metabolites and metabolic pathways.

We further sought to determine the potential origin of the metabolites that were highly correlated with gut bacteria. Total of 106 co-metabolism metabolites, 36 microbiota metabolites, and 21 host metabolites were identified (Supplementary Figure S4A,B). We performed functional enrichment analysis of these metabolites and observed that these metabolites were mainly enriched in steroid hormone biosynthesis, histidine metabolism, tryptophan metabolism, tyrosine metabolism, Galactose metabolism, arginine biosynthesis, arginine and proline metabolism, beta-Alanine metabolism, and glutathione metabolism (Supplementary Figure S4C). We found that metabolites related to 9 pathways were originated from microbiota, including tryptophan metabolism, tyrosine metabolism, Galactose metabolism, arginine and proline metabolism, pyrimidine metabolism, folate biosynthesis, aminobenzoate degradation, and carotenoid biosynthesis. Interestingly, only the pathway of steroid hormone biosynthesis was identified to be related to host metabolism and 27 pathways involved in co-metabolism, such as histidine metabolism, tryptophan metabolism, tyrosine metabolism, arginine biosynthesis, arginine and proline metabolism, beta-alanine metabolism, and glutathione metabolism. Collectively, the differential metabolites we identified here were mostly produced or influenced by the gut microbiota. Our results suggested that the altered intestinal metabolism may have a strong correlation with the changed gut microbiota induced by SMG.

To further investigate the potential dependencies between the perturbed intestinal metabolites and gut microbiota, the Spearman’s correlation coefficient between differential metabolites and the abundance of differential gut bacteria was determined. Strong connections were identified between the perturbed intestinal metabolites and gut microbiota (Table 1). Our results showed that the level of several differential metabolites were significantly correlated with the relative abundance of biomarker microbial genera (ρ > 0.6, p < 0.05). Notably, Allobaculum, Akkermansia, Phascolarctobacterium, and Faecalibacterium who have been reported to produce SCFAs, exhibited a negative association with metabolites involved in tryptophan metabolism (5-hydroxyindole-3-acetic acid, 2-ketoadipic acid, Dl-3-hydroxy-kynurenine, L-tryptophan, and oxoadipic Acid), histidine metabolism (1-methylhistidine, carnosine, anserine, and histamine), arginine and proline metabolism (S-adenosylmethionine, oxaceprol, and 4-acetamidobutyric Acid), but positive with hydroxyprogesterone caproate and etiocholanolone involving in steroid hormone biosynthesis. Moreover, genera of Bacteroides, Lachnospiraceae_Clostridium, Rothia, Bilophila, and Coprococcus were positively correlated with arginine and proline metabolism (4-acetamidobutyric acid and oxaceprol), tryptophan metabolism (5-hydroxyindole-3-acetic acid), Dl-3-hydroxy-kynurenine, L-tryptophan), and histidine metabolism (anserine, carnosine, and histamine), but negatively associated with steroid hormone biosynthesis (etiocholanolone, hydroxyprogesterone caproate, and cortisol). The relative abundance of Alistipes showed significant positive correlation with histidine metabolism (urocanic acid) and steroid hormone biosynthesis (etiocholanolone and cortisol). Our results suggested that the alteration of intestinal metabolites and pathways under long-term of SMG have a strong association with the imbalance of gut microbiota.