Animals and husbandry details
Six-week-old healthy male SD rats were included and used in accordance with Guidelines for the Care and Use of Laboratory Animals of the Institute of Laboratory Animal Resources, Institutional Animal Care and Use Committee of Beijing University of Chinese Medicine. Animal weights were between 135 g and 165 g at the beginning of the first study. All rats (n=40) were housed with 5 animals per cage on a 12 h/12 h light/dark cycle in the animal facility of the Experimental Center of Beijing University of Chinese Medicine (room temperature: 25 ± 1 °C, humidity: 50±5%). All animals were allowed ad libitum access to food and water.
Treatment and sample collection
Bovine type II collagen (CII; 2 mg/ml) (Chondrex Inc., WA, United State) emulsified with complete Freund’s adjuvant (CFA; 2 mg/ml) (Sigma–Aldrich Co., St. Louis, United States) was obtained. All animals were divided into 4 groups (N=10) with the random number method after acclimation for 1 week: ① Control (CON) group, kept at 25 ± 1 °C and 50±5% humidity and injected with 0.9% NaCl solution (0.2 ml/rat) on Days 7 and 14; ② high humidity (HH) group, kept at 25 ± 1 °C and 80±5% humidity and injected with 0.9% NaCl solution (0.2 ml/rat) on Days 7 and 14; ③ CIA model (CIA) group, kept at 25 ± 1 °C and 50±5% humidity and injected with CII-CFA emulsion (1 mg/ml，0.2 ml/rat) on Days 7 and 14; and ④ CIA model with high humidity (H+CIA) group, kept at 25 ± 1 °C and 80±5% humidity and injected with CII-CFA emulsion (1 mg/ml，0.2 ml/rat) on Days 7 and 14. Arthritis scores were measured every 7 days after Day 21 [15,31]. Two fresh stool pellets were obtained from each rat after Day 59. Samples were placed in sterile conical tubes and immediately frozen at −80 °C. Rats were euthanized to obtain the hind limbs of the rats after Day 60. Blood samples were obtained from the inferior vena of rats and centrifuged at 3500 × g and 4 °C for 15 min to obtain the serum, which was immediately frozen at −80 °C. The group information and experimental process are shown in Fig. 1.
Assessment of arthritis variables
Arthritis scores were measured every 7 days after Day 21. The assessment criteria of arthritis scores were as follows: 0, no redness and swelling in the foot joints and claws; 1, mild swelling or redness of the foot joint and claws; 2, moderate swelling or mild redness of the foot joint and claws; 3, the claws are red and swollen below the ankle joint; and 4, severe redness, swelling and deformation of the ankle foot joint.
The articular cartilage was fixed in 4.0% paraformaldehyde for 3 days and then decalcified in 10% ethylenediamine tetraacetic acid (EDTA) decalcification solution. The tissues were dehydrated by gradient ethanol, embedded in paraffin, and sliced into 4 μm sections by pathological sectioning. Then, the slides were stained with safranin O to assess the articular cartilage damage.
Measurements of serum proinflammatory cytokines
The levels of TNF-α, IL-6 and IL-17 in serum were measured by enzyme-linked immunosorbent assay (ELISA). All reagents were purchased from CUSABIO (CUSABIO Technology LLC, Wuhan, China), and analyses were conducted according to the manufacturer’s instructions. The absorbance was measured at 450 nm.
DNA extraction and 16S rRNA sequencing
According to the manufacturer's instructions, a DNeasy PowerSoil kit (Cat. No. 12888; QIAGEN, Dusseldorf, Germany) was utilized to extract DNA from different samples. The purity and concentration of DNA were detected by agarose gel electrophoresis. Extracted DNA was diluted to a concentration of 1 ng/μl and stored at -20 °C until further processing. The DNA genome as a template was utilized to perform PCR amplification to ensure the efficiency and accuracy of amplification with barcode-specific primers and Takara Ex Taq hi-fi enzyme (Cat. No. RR001Q; Takara, Dalian, China). The corresponding areas of bacterial diversity identification were as follows: The V3-V4 region of the 16S rRNA genes was amplified with universally primed 343F (5' -TACgGRaggCAGCAGG-3') and 798R (5' -AgggTATCtaatCCT-3') using a commercial PCR kit (Cat. No. 51531; Qiagen, Dusseldorf, Germany).
PCR products were detected by gel electrophoresis and purified by AMPure XP beads after detection. The purified products were used as a second round of PCR template and amplified by a second round of PCR. After a second purification step with AMPure XP beads, PCR products were quantitatively analysed by a Qubit dsDNA detection kit (Cat. No. Q32854; Thermo Fisher Scientific, MA, United States). The samples were mixed in equal quantities according to the concentration of PCR products and then sequenced. An equal amount of purified amplicon was pooled for sequencing with a NovaSeq PE250 instrument.
Operational taxonomic unit (OTU) clustering and species annotation
Raw sequencing data were in FASTQ format. Paired-end reads were then preprocessed using Trimmomatic software (version 0.35) to detect and cut off ambiguous bases (N). The sliding window method was used to evaluate the average base quality. If the average mass value in the window was lower than 20, the back-end base was cut from the window. Paired-end reads were assembled using FLASH software (version 1.2.11). The stitching parameters were as follows: the smallest overlap was 10 bp, the largest overlap was 200 bp, and the maximum error matching rate was 20%. Sequences were further filtered as follows: the sequences containing ambiguity were removed, and reads with 75% of bases above Q20 were retained. Moreover, Quantitative Insights Into Microbial Ecology (QIIME, version 1.8) was used to detect and remove chimeric sequences.
After the sequencing data were preprocessed to generate high-quality sequences, Vsearch software (Version 2.4.2) was used to classify the sequences into multiple operational taxonomic units (OTUs) according to their similarity. A parameter sequence similarity greater than or equal to 97% is classified as an OTU unit. The QIIME software package was used to select representative sequences of each OTU, and all representative sequences were annotated and blasted against the Unite database (ITS rDNA) using pynast (v0.1).
Sample preparation and GC–MS analysis
The samples stored at -80 °C were thawed on ice, 60 mg of stool sample was accurately weighed and placed into a 1.5-ml centrifugation tube, 40 μl of internal standard (l-2-chloro-phenylalanine, 0.3 mg/ml, methanol configuration) was added to each sample, and 2 small steel balls and 360 μl of cold methanol were successively added. Samples were stored at -20 °C for 2 min and ground in a grinding machine (60 Hz, 2 min). Samples were sonicated in an ice water bath for 30 min, and then 200 μl of chloroform was added and the mixture vortexed (60 Hz, 2 min). Then, 400 μl of water was added and the mixture vortexed (60 Hz, 2 min). Ultrasonic extraction was performed in an ice water bath for 30 min, and the samples were allowed to stand at -20 °C for 30 min. Then, the extract was centrifuged for 10 min (13000 × g, at 4 °C), and 300 μl of the supernatant was put into a glass-derived bottle and dried in a centrifugal freeze dryer. Next, 80 μl of methoxamine hydrochloride pyridine solution (15 mg/ml) was added to each sample, followed by vortexing for 2 min and ice water ultrasonic treatment for 3 min. The oxime reaction was carried out for 90 min in an incubator shaker at 37 °C. A total of 80 μl of trifluoroacetamide (containing 1% chlorotrimethylsilane) derivatizing reagent and 20 μl n-hexane were added, and 11 internal standards (C8/C9/C10/C12/C14/C16, 0.8 mg/mL; C18/C20/C22/C24/C26, 0.4 mg/ml, all prepared in chloroform) were added at a volume of 10 μl, followed by vortexing for 2 min and reaction at 70 °C for 60 min. After the samples were removed, they were placed at room temperature for 30 min for GC–MS metabolomics analyses. All extraction reagents were precooled at -20 °C before use.
Meteorological chromatography-mass spectrometric conditions
Metabolite analysis was carried out by a gas chromatograph-mass spectrometer (7890B-5977A; Agilent J&W Scientific, Folsom, CA, United States). Separation was performed by loading a 30-m × 0.25-mm × 0.25-μm DB-5MS fused silica capillary column (Agilent J&W Scientific, Folsom, CA, United States). The flow rate of the carrier gas, high-purity helium (purity not less than 99.999%), was 1.0 ml/min, and the inlet temperature was 260 °C. The injection volume was 1 μl, and the solvent delay was 5 min. The temperature programme was achieved using the following gradient: The initial temperature of the GC oven was 60 °C, and then the temperature was ramped to 125 °C at 8 °C/min, to 210 °C at 5 °C/min, to 270 °C at 10 °C/min, and to 305 °C at 20 °C/min and held steady for 5 min. Mass spectrometric conditions were as follows: ionization source, electron impact ionization; ion source temperature, 230 °C; quadrupole temperature, 150 °C; collision energy, 70 eV; solvent delay, 3 min; scan mode, full scan (scan mode); and mass scan range, m/z 50-500.
Data preprocessing and statistical analysis
Raw GC–MS mass spectra were converted to ABF format files by converting AnalysisBaseFileConverter software (version 4.0). Then, the data were imported to MS-DIAL software (version 3.9) for preprocessing. Finally, the raw data matrix included sample information, and the name of each peak, retention time, mass-to-charge ratio, and mass spectral response intensity (peak area) were derived. The NIST database (https://webbook.nist.gov/chemistry/) was used for material qualitative analysis. Principal component analysis (PCA) and Partial least-squares-discriminant analyses (PLS-DA) were performed to visualize the changes in metabolites between the experimental groups after mean centering (Ctr) and Pareto variance (Par) scaling, respectively. Variable importance in the projection (VIP) values were obtained according to the PLS-DA model. VIP >1 was used to identify potential biomarkers. The PLS-DA model was tested for 200 response sequencing tests; that is, the x-matrix was fixed, the variables of the previously defined classification Y matrix (such as 0 or 1) were randomly arranged n times (n=200), and the corresponding PLS-DA model was established to obtain R2 and Q2 values of the random model. Linear regression was performed with R2Y and Q2Y of the original model, and the intercept values of the regression line and Y-axis were R2 and Q2, respectively, which were used to measure whether the model was overfitting. The VIP values were obtained based on the PLS-DA model, and p values were derived from a two-tailed Student’s t test using the normalized peak areas. The criteria for screening differential metabolites were VIP >1 and p <0.05.
Student’s t test for unpaired data (95% confidence interval) was used for comparisons between each group using GraphPad Prism (Version 9.3; GraphPad Software, San Diego, CA, USA). The data are expressed as the mean ± standard deviation of the mean (S.D.). One-way analysis of variance (ANOVA) was used for measurement data of multiple groups, and least-significant difference (LSD) was used for pial comparison between groups. If homogeneity of variance was not satisfied, the rank sum test was used for comparison of multiple independent samples. Error bars represent the standard deviation. The degree of significance is indicated as *p < 0.05, **p < 0.01, #p < 0.05 and ##p < 0.01.