Male C57BL/6J mice (8–10 weeks old) were purchased from Shanghai Laboratory Animal Center, Chinese Academy of Science (Shanghai, China). The animals were housed in a constant temperature room (22–24°C) and allowed to acclimatize to the animal facility environment for 2 weeks before the experimentation, with food and water ad libitum. In the normal rhythm group, mice were kept under strict 12 h light/12 h dark cycles with lights being turned on at 6 a.m. and turned off at 6 p.m. In the JL group, jet lag was induced according to the procedure described in earlier studies (Thaiss et al., 2014), in which mice were subjected to daily 8-h advance of the light/dark cycle for every 3 days. Briefly, mice were shifted between normal light conditions (lights on at 6 a.m. and off at 6 p.m.) and an 8-h time in advance (lights on at 10 p.m. and off at 10 a.m.) every 3 days. During the process of modelling, both control mice and jet-lagged mice were fed with a standard chow diet (Purina Laboratory Rodent Diet #5001). After 4 months, the stool samples and jejunal contents of mice were collected into sterile 1.5 mL Eppendorf microcentrifuge tubes (Corning, NY, USA). The collection of mouse jejunal contents was conducted in the clean bench. The jejunum was cut open with a sterile ophthalmic scissor and the contents were collected gently with sterile tweezers. The fresh pellets and jejunal contents were immediately frozen in liquid nitrogen and stored at -80°C. All experimental protocols conformed to the National Institutes of Health (NIH) Guidelines for the Care and Use of Laboratory Animals and were approved by the School of Medicine, Shanghai Jiao Tong University.

DNA was extracted from the samples using a QIAamp Fast DNA Stool Mini Kit (Qiagen, CA, USA) according to Manufacturer’s instructions. The concentration of bacterial DNA was measured using a NanoDrop 2000 spectrophotometer (Thermo Scientific, USA). Afterwards, the 16S rRNA genes were amplified with the bacterial primer pair 338F-806R flanking the V3–V4 region using FastPfu Polymerase. Amplicons were then purified by gel extraction (AxyPrep DNA GelExtraction Kit, Axygen Biosciences, CA, USA) and quantified using QuantiFluor-ST (Promega, Madison, WI, USA). Paired-end sequencing was performed using an Illumina MiSeq system (Illumina, CA, USA). High-throughput pyrosequencing of PCR products was then performed on the free online platform of Majorbio Cloud Platform (www.majorbio.com).

Sequencing reads were demultiplexed and filtered. Operational taxonomic units (OTUs) were picked at 97% similarity cutoff, and the identified taxonomy was then aligned using the Greengenes database (Version 13.8). The relative species abundance in each group was evaluated based on the rank-abundance curves, and alpha-diversity index (Shannon and Simpson) was analyzed. For beta-diversity analysis, principal component analysis (PCA), principal coordinate analysis (PCoA), and nonmetric multidimensional scaling (NMDS) were performed using Quantitative Insights into Microbial Ecology (QIIME) 1.9.1. Difference between groups was tested using analysis of similarities (ANOSIM). For the analysis of differences in bacterial composition, we used student t test with P < 0.05 as a threshold. The calculated P value underwent False Discovery Rate (FDR) correction and FDR P < 0.05 was considered statistically significant. From phylum to genus, the biomarkers in the two groups were quantitatively analyzed by the linear discriminant analysis (LDA) effect size (LEfSe) analysis. The LEfSe analysis, with the LDA threshold of >2, was performed using the non-parametric Kruskal-Wallis (KW) test to recognize the most differently abundant taxa.

Targeted metabolomics analysis of fecal samples was performed by Metabo-Profile Biotechnology (Shanghai, China), according to the methods described previously (Choi et al., 2020; Xie et al., 2021). Briefly, the fecal samples were thawed on ice. About 5 mg of each sample was added to 25 μL of ultrapure water and homogenized with zirconium oxide beads for 3 min before addition of methanol (120 μL) containing the internal standard to extract the metabolites. Then the samples were homogenized for another 3 min and centrifuged at 18,000g for 20 min. The supernatants (20 μL for each sample) were transferred to a 96-well plate. The following procedures were performed on an Eppendorf epMotion Workstation (Eppendorf Inc., Hamburg, Germany). The freshly prepared derivative reagents (20 μL) were added to each well. The plate was sealed, and the derivatization was carried out at 30°C for 60 min. After derivatization, 330 μL of ice-cold 50% methanol solution was added to dilute the samples. Then, the plate was stored at –20°C for 20 min, followed by centrifugation (at 4000g) at 4°C for 30 min. Thereafter, 135 μL of the supernatant for each sample was transferred to a new 96-well plate with 10 μL of internal standards in each well. Serial dilutions of derivatized stock standards were added to the wells on the left. Finally, the plate was sealed for analysis. All the standards were obtained from Sigma–Aldrich (MO, USA).

An ultra-performance liquid chromatography coupled to tandem mass spectrometry (UPLC-MS/MS) system (ACQUITY UPLC-Xevo TQ-S, Waters Corp., MA, USA) was used to analyze 11 targeted metabolites of interest. We used UPLC columns including an ACQUITY HPLC BEH C18 1.7 µm VanGuard pre-column (2.1 × 5 mm) and an ACQUITY HPLC BEH C18 1.7 µm analytical column (2.1 × 100 mm) to perform chromatographic separation of fecal samples at a constant temperature of 40°C. The injection volume of sample was 5 µL. The mobile phases: eluent A was 0.1% formic acid in water, and eluent B was acetonitrile/IPA (70:30). The gradient elution condition was: 0–1 min (5% B), 1–11 min (5–78% B), 11–13.5 min (78–95% B), 13.5–14 min (95–100% B), 14–16 min (100% B), 16–16.1 min (100-5% B), 16.1–18 min (5% B). The flow rate was set at 0.40 mL/min. For mass spectrometer, capillary: 1.5 (ESI+), 2.0 (ESI-) Kv; source temperature: 150°C; desolvation temperature: 550°C; desolvation gas flow: 1000 L/h. The quality control samples were prepared along with the test samples and run after each 14 samples to ensure reproducibility.

For data analysis, the raw data files generated by UPLC-MS/MS were processed with MassLynx software (Version 4.1, Waters, MA, USA) to perform peak integration, calibration, and quantitation for each metabolite. The powerful package R studio was used for statistical analysis. PCA, partial least square discriminant analysis (PLS-DA), and orthogonal partial least square discriminant analysis (OPLS-DA) were used for classification and identification of differently altered metabolites. Potential biomarkers of differential metabolites were characterized by variables with a variable influence on projection (VIP) > 1 and P < 0.05 in the Student t test or Wilcoxon test. Z-transform was conducted to observe the distribution of different metabolites between groups. The pathway enrichment analysis was performed using pathway impact and hypergeometric test. Also, Spearman’s rank correlation analyses were conducted to explore the correlation between metabolites and gut microbiota.

We used 16S rRNA high-throughput sequencing to determine whether circadian disruption influenced colonic microbiota. Sequencing analysis of fecal microbe revealed 551,482 raw reads. Based on the 97% similarity level, all the effective reads were clustered into operational taxonomic units (OTUs). According to the histogram of species analysis on OTU level ( Figure 1A ), the Con and JL groups had 345 identical OTUs. In addition, 128 OTUs exclusively belonged to the Con group, while only 82 OTUs belonged to the JL group. Rank-Abundance curves ( Figure 1B ) showed that the relative abundance of colonic microbiota was decreased in the JL group. Moreover, as shown in Figures 1C, D on α-diversity analysis, the community diversity of the colonic microbiota was decreased in JL group. Quantitative analysis showed that Shannon index was decreased by 11.9% (P = 0.040). Simpson index showed a tendency to increase (by 65.1%, P = 0.078). As shown in Figure 1E , the results of ANOSIM indicated that the colonic microbial structure was significantly different between the two groups (R = 0.813 and P = 0.003). β-diversity analysis (PCA, PCoA, NMDS) revealed a distinct clustering of colonic microbiota in the two groups (ANOSIM analysis; P = 0.003; Figures 1F–H ). Next, the phylum-level analysis indicated that the relative abundance of Bacteroidetes phylum was significantly decreased (P = 0.002, FDR P = 0.016). The abundance of Actinobacteria was increased in the JL group (P = 0.00002, FDR P = 0.00032) and that of Firmicutes showed a tendency to increase (P = 0.014, FDR P = 0.070) for FDR P were greater than 0.05 ( Figure 2A ). As shown in Figure 2B , the ratio of Firmicutes to Bacteroidetes (F:B) was increased in the JL group (P = 0.031). On genus level, we chose the top 18 abundance genus for analysis. The relative abundance of Bifidobacterium (P = 0.000078, FDR P = 0.0054) was significantly increased in the JL group. The abundance of Turicibacter (P = 0.003, FDR P = 0.086), Ileibacterium (P = 0.0067, FDR P = 0.15) and Eubacterium_fissicatena_group (P = 0.018, FDR P = 0.18) showed a tendency to increase. The abundance of norank_f_Muribaculaceae (P = 0.011, FDR P = 0.16) and Prevotellaceae_Ga6A1_group (P = 0.011, FDR P = 0.16) showed a tendency to decrease in the JL group ( Figure 2C ). These results suggested that JL suppressed the abundance, richness, and diversity of colonic microbiota but increased the F: B ratio. Finally, LEfSe analysis was performed to identify specific bacteria associated with JL. The results showed that Erysipelotrichales, Erysipelotrichaceae, Bacilli, Ileibacterium, Turicibacter, Actinobacteriota, and Bifidobacteriaceae dominated in the JL group ( Figures 3A, B ).

The results of 16S rRNA high-throughput sequencing showed that jejunal microbiota might also be affected by JL. Sequencing analysis revealed 569,943 raw reads. According to a 97% similarity level, all effective reads were clustered into OTUs. Figure 4A shows species distribution in the Con and JL groups on OTU level. The two groups have 913 identical OTUs. Further, 216 OTUs exclusively belonged to the Con group, while 77 OTUs belonged to JL group. Rank-abundance curves revealed that the richness of the jejunal microbiota was decreased by JL ( Figure 4B ). In addition, α-diversity analysis showed an overt difference in jejunal microbiota between the Con and JL groups. Shannon index was significantly decreased by 29.6% (P = 0.029) ( Figure 4C ), while Simpson index was markedly increased by 167.7% (P = 0.036) in the JL group ( Figure 4D ). As is shown in Figure 4E , the results of ANOSIM analysis revealed a significant difference between two groups (R = 0.2773, P = 0.033). β-diversity analysis showed an obvious clustering of jejunal microbiota composition between Con and JL groups using PCoA and NMDS analyses ( Figures 4G, H ), while no obvious separation of the two groups was observed using PCA analysis ( Figure 4F ). These results suggested that JL could affect the abundance, richness, and diversity of jejunal microbiome. According to the community analysis, JL tended to decrease the relative abundance of Bacteroidota phylum (P = 0.046, FDR P = 0.18) in the jejunal microbiota. The abundance of Firmicutes showed a tendency to increase in the JL group, with no statistically significant difference ( Figure 5A ). Chronic JL caused an increase in the ratio F: B (P = 0.032) ( Figure 5B ). Further, on the genus level, Figure 5C showed that the abundance of Ileibacterium (P = 0.046, FDR P = 0.34) showed a tendency to increase in the JL group, while the abundance of Streptococcus (P = 0.028, FDR P = 0.34), Bacteroides (P = 0.010, FDR P = 0.34), Ruminococcus_torques_group (P = 0.027, FDR P = 0.34), and Faecalibacterium (P = 0.018, FDR P = 0.34) showed a tendency to decrease after JL. All the P values were greater than 0.05 after FDR correction, indicating the changes of jejunal microbiota caused by chronic JL may be relative slight compared with colonic microbiota. Furthermore, LEfSe analysis was performed to recognize the specific microbiota related to JL ( Figures 6A, B ). Five types of bacteria dominated in JL group, including Erysipelotrichaceae, Erysipelotrichales, Ileibacterium, Turicibacter, and Faecalibaculum. These results indicated that, besides colonic microbiota, the bacterial community of jejunum might also be affected by chronic JL treatment.

The fecal metabolome represents a functional readout of the activity of gut microbiome, and thus can serve as an intermediary phenotype reflecting the host-microbiome interactions (Liu et al., 2020). We subsequently analyzed the fecal metabolites of Con and JL mice using a targeted UPLC-MS/MS based metabolomics approach. As shown in Figure 7A , the relative abundance of fecal metabolites in JL mice was different from that in the Con group. The abundance of amino acids (14.11% vs 30.04%, P = 0.006), indoles (0.11% vs 0.20%, P = 0.008), peptides (0.06% vs 0.11%, P = 0.006) and phenylpropanoids (0.03% vs 0.05%, P = 0.013) in fecal metabolites were decreased in the JL group compared with the Con group. The abundance of carbohydrates and SCFAs was similar between two groups.

In addition, the results of PCA and PLS-DA analysis showed apparently distinct clustering of fecal metabolites ( Figures 7B, C ). Based on the results of both univariate and multivariate statistical analyses for the differential metabolites, 52 potential biomarkers which have biological significance were obtained ( Figure 7D ). A scatter plot and a heatmap of Z scores were obtained after the Z-transform of each biomarker ( Figures 8A, B ). Based on the above-mentioned results and pathway clustering analysis shown in Figure 8C , we subsequently explored Try and its derivatives, bile acids, and itaconic acids. Among them, Try had the largest impact value. Compared with the control group, the levels of Try (1001.46 ± 343.55 vs 2185.52 ± 502.33 nmol/g, P = 0.0014), indole-3-carboxaldehyde (IAId; 802.04 ± 509.31 vs 1937.13 ± 553.35 nmol/g, P = 0.0065), and indole-3-methyl acetate (2.86 ± 2.49 vs 7.98 ± 4.59 nmol/g, P = 0.047) were all significantly decreased in the JL group ( Figures 8D–F ). Bile acids are also important bacterial metabolites in the intestine. Via gut microbiome–mediated metabolism, the primary bile acids are transformed into the secondary bile acids, namely deoxycholic acid (DCA), lithocholic acid (LCA) and ursodeoxycholic acid (UDCA) (Bajor et al., 2010; Jia et al., 2018). The results showed that compared with Con group, the levels of the secondary bile acids and their hepatic microsomal metabolites, such as isoalloLCA (3.58 ± 1.54 vs 22.19 ± 14.66 nmol/g, P = 0.030), β-UDCA (23.34 ± 13.22 vs 59.05 ± 24.44 nmol/g, P = 0.045), isoLCA (10.57 ± 8.09 vs 23.88 ± 11.03 nmol/g, P = 0.047), 6-ketolithocholic acid (59.40 ± 47.09 vs 316.31 ± 142.46 nmol/g, P = 0.0043), hyodeoxycholic acid (HDCA; 224.18 ± 98.07 vs 1594.63 ± 870.24 nmol/g, P = 0.0043), β-HDCA (38.45 ± 15.97 vs 276.91 ± 145.94 nmol/g, P = 0.0043), and taurine-hyodeoxycholic acid (THDCA; 2.27 ± 1.78 vs 33.83 ± 37.83 nmol/g, P = 0.0043) were decreased in the jet-lagged mice ( Figures S1A–G ). Whereas, the levels of other secondary bile acids, such as α-muricholic acids (αMCA; 953.74 ± 298.75 vs 494.93 ± 340.21 nmol/g, P = 0.041) and βMCA (828.47 ± 336.09 vs 382.51 ± 269.34 nmol/g, P = 0.045), as well as the levels of the primary bile acids such as cholic acid (CA; 585.34 ± 317.84 vs 181.05 ± 132.39 nmol/g, P = 0.044) and norcholic acid (NorCA; 39.52 ± 16.41 vs 9.32 ± 12.48 nmol/g, P = 0.017), were increased in the jet-lagged mice ( Figures S1H–K ). The level of itaconic acid, which is a major physiological regulator of the global metabolic rewiring, was decreased in JL group (2.79 ± 0.63 vs 8.28 ± 5.37 nmol/g, P = 0.0043) ( Figure S2 ). Therefore, chronic JL could not only change the microbiome of both jejunum and colon but also disturb the concentration of fecal metabolites.

We subsequently analyzed the correlation between colonic microbiota and its metabolites ( Figures 9A, B ). On phylum level, the abundance of Try was negatively correlated with that of Actinobacteriota (r = -0.76, P = 0.016). The abundance of IAId, a downstream product of Try, was positively correlated with that of Bacteroidetes (r = 0.82, P = 0.0068). On the genus level, Try level was positively correlated with abundance of Prevotellacea UCG-003 (r = 0.68, P = 0.030), Negativibacillus (r = 0.78, P = 0.0084), Dubosiella (r = 0.81, P = 0.0041), Prevotellaceae_Ga6A1_group (r = 0.89, P = 0.00065), Anaerotruncus (r = 0.87, P = 0.00095), Ruminococcus (r = 0.76, P = 0.016), Eubacterium_ventriosum_group (r = 0.77, P = 0.0096), Streptococcus (r = 0.77, P = 0.0099), and Fournierella (r = 0.75, P = 0.013). On the contrary, it was negatively correlated with the abundance of bacteria including Enterorhabdus (r = -0.75, P = 0.018), Erysipelatoclostridium (r = -0.85, P = 0.0035), Eubacterium_siraeum_group (r = -0.81, P = 0.0041), Adlercreutzia (r = -0.68, P = 0.030), Faecalibaculum (r = -0.70, P = 0.026), Pseudogracilibacillus (r = -0.71, P = 0.022), Faecalitalea (r = -0.69, P = 0.028), Bifidobacterium (r = -0.85, P = 0.0035), Ileibacterium (r = -0.81, P = 0.0082), and Turicibacter (r = -0.75, P = 0.012). The level of IAId was positively correlated with the abundance of Peptococcus (r = 0.75, P = 0.012), Acetatifactor (r = 0.66, P = 0.036), Prevotellacea UCG-003 (r = 0.74, P = 0.014), Negativibacillus (r = 0.70, P = 0.026), Dubosiella (r = 0.86, P = 0.0015), Prevotellaceae_Ga6A1_group (r = 0.78, P = 0.0075), Anaerotruncus (r = 0.83, P = 0.0028), Ruminococcus (r = 0.82, P = 0.0068), Eubacterium_ventriosum_group (r = 0.88, P = 0.00085), Streptococcus (r = 0.72, P = 0.020), and Fournierella (r = 0.76, P = 0.01). However, it was negatively correlated with the abundance of Eubacterium_siraeum_group (r = -0.87, P = 0.0011), Pseudogracilibacillus (r = -0.71, P = 0.022), Bifidobacterium (r = -0.72, P = 0.024), Ileibacterium (r = -0.68, P = 0.035), and Turicibacter (r = -0.64, P = 0.047).