Clinical sample collection
Samples of NAFLD subjects were derived from a clinical study, which was led by Shuguang Hospital affiliated to Shanghai University of Traditional Chinese Medicine. The protocol was approved by Institutional Review Board of Shuguang Hospital Affiliated to Shanghai University of Chinese Medicine (No. 2017-548-31) and registered at the Chinese Clinical Trial Registry (No. ChiCTR-IOR-17013491). The diagnostic criteria for NAFLD were referred to the Guidelines for the Diagnosis and Treatment of Nonalcoholic Fatty Liver Disease (2010) issued by Fatty liver and Alcoholic Liver Disease Group, Hepatology Society of Chinese Medical Association. Healthy subjects were recruited from the Phase I clinical program of Good Clinical Practice Center or the Physical Examination Center of Shanghai Shuguang Hospital, which was conducted in accordance with the Declaration of Helsinki. The protocol was approved by Institutional Review Board of Shuguang Hospital Affiliated to Shanghai University of Chinese Medicine (No. 2019-662-17-01). Written informed consent was obtained from all subjects. 34 NAFLD subjects (30 males and 4 females, ALT > 80, CAP score > 300) and 24 healthy subjects (23 males and 1 female, ALT < 40, CAP score < 235), aged 18-65 years old, were included in this study. The baseline characteristics are detailed in Supplementary Table 1. The inclusion and exclusion criteria for NAFLD patients and healthy subjects are supplied in the supplementary material. Sex was not considered in this study. The blood samples were drawn after fasting 12 h and serum samples were prepared and immediately frozen at −80 °C.
Animal studies
All animal experiments were conducted under the Guidelines for Animal Experiment of Shanghai University of Traditional Chinese Medicine and all animals received humane care according to the criteria outlined in the Guide for the Care and Use of Laboratory Animals. Animal protocols were approved by the Institutional Animal Ethics Committee of Shanghai University of Traditional Chinese Medicine (PZSHUTCM200417001, PZSHUTCM200430002, and PZSHUTCM2306200001). Male C57BL/6 wild type mice and Pparα−/− mice were purchased from Vital River Laboratory Animal Technology Co., Ltd. and Cyagen Biosciences (Suzhou) Inc., respectively. After 1 week acclimatization, all the mice were maintained in specific-pathogen-free (SPF) environment under a 12 h light/12 h dark cycle at 23°C–24°C with 60% ± 10% relative humidity. Each animal was checked for its suitability according to animal welfare authorities before individual experiment.
Animal experiment 1: NAFLD mice model after 24-week HFHS diet feeding
Based on our preliminary observational experiment and other report54, 24 weeks of HFHS feeding was sufficient for generating a typical NAFLD mouse model in C57BL/6 J mice. Thus, a total of 17 five-week-old male C57BL/6 J mice were randomly divided into two groups fed with a chow diet (C2018, SYSEBIO) and normal water or an HFD (60% fat, D12492, Research Diets) and supplemented with 30% sucrose in drinking water (HFHS group) for 24 weeks. The two groups were defined as Control (n = 8) and HFHS (n = 9) groups. Intraperitoneal glucose tolerance test (ipGTT) and intraperitoneal insulin tolerance test (ipITT) were administrated at the 22th and 23th week, respectively. The body weight and dietary intake were monitored during the feeding period.
Animal experiment 2: the effect of HDCA in HFHS-fed C57BL/6 J mice
The dose for HDCA treatment was referred to previous reports that the effective concentration range of HDCA for suppressing atherosclerosis formation, reducing plasma cholesterol levels, and improving diabetes in rodent animals is from 0.05 to 1.25% supplied in diet11,13,14,15,55,56. A preliminary experiment was conducted to determine the dose effect of HDCA on ameliorating NAFLD by using three doses of 1.25%, 0.625%, and 0.3125% (H106315, Aladdin, 98% purity) supplied in the diet. Finally, medium-dose HDCA (0.625%) was chosen in the formal experiment. Additionally, based on our preliminary observational experiment, the feeding duration of 12 weeks was chosen because 12-week HFHS feeding is sufficient for inducing successful hepatic steatosis based on the steatosis score and NAS, which is more applicable for evaluating the anti-NAFLD effect of HDCA in mice compared to the long term 24-week feeding. Then, a total of 18 five-week-old male mice were randomly divided into two groups (control group n = 6, HFHS group n = 12). Control group and HFHS group were fed with chow diet or HFHS diet for 12 weeks (n = 6 per group). HDCA group were fed with HFHS diet for 4 weeks and then supplemented with 0.625% HDCA in the diet for another 8 weeks (n = 6). The body weight and dietary intake were monitored during the feeding period. ipGTT and ipITT were administrated at the 11th and 12th week, respectively.
Animal experiment 3: the effect of HDCA in HFD-fed ob/ob mice
A total of 10 five-week-old ob/ob C57BL/6 J male mice were fed with an HFD diet with (n = 5) or without 1.25% HDCA (n = 5) for 8 weeks.
Animal experiment 4: the effect of HDCA in HFHS-fed Pparα
−/− mice
The control groups of wild-type mice (8-week-old, male, C57BL/6 N) and Pparα−/− mice (8-week-old, male, C57BL/6 N) were fed with HFHS for 12 weeks. HDCA groups of wild-type mice and Pparα−/− mice (8-week-old, male) were fed with HFHS for 4 weeks and then supplemented with 0.625% HDCA in HFHS diet for another 8 weeks. The body weight and dietary intake were monitored during the feeding period.
Animal experiment 5: the effect of HDCA in HFHS-fed Pparα
hep−/− mice
To obtain hepatocyte-specific Pparα-deficient or control mice, Pparαflox/flox mice (8-week-old, male, C57BL/6 J)57 were injected with AAV2/8-TBG-Cre or control virus AAV2/8-TBG-ZsGreen (Hanbio technology, Shanghai) via the tail vein, respectively. The mice were given HFHS diet the same day they were injected with the virus. The control groups of Pparαflox/flox and Pparαhep−/−mice were fed with HFHS for 12 weeks. HDCA groups of Pparαflox/flox and Pparαhep−/−mice were fed with HFHS for 4 weeks and then supplemented with 0.625% HDCA in the diet for another 8 weeks. The body weight and dietary intake were monitored during the feeding period.
At the end of all the experiments, overnight fasted mice were euthanized by deeply anesthetized with pentobarbital sodium (100 mg/kg, i.p.) followed by cervical dislocation. Tissues and intestinal content samples were harvested and immediately frozen at −80 °C for further analysis.
ipGTT and ipITT
For the GTT, glucose (1 g/kg body weight) was administered via intraperitoneal injection after overnight fasting. The glucose level of tail vein was measured at 0, 15, 30, 60, 90, 120 min after glucose load. For the ITT, insulin (0.75 U/kg body weight) was administered via intraperitoneal injection after 4 h fasting. The glucose level of the tail vein was measured at 0, 15, 30, 60, 90, 120 min after insulin load. Roche ACCU-CHEK Performa was applied for blood glucose data collection.
Serum biochemical assay, hepatic and fecal lipids, and inflammatory cytokines assay
The serum levels of triglycerides, total cholesterol, ALT, and AST were measured according to the instructions of specific kits (Nanjing Jiancheng Bioengineering Institute, China). The level of fasting insulin was measured by Insulin ELISA Assay (Merck Millipore, EZRMI-13K, USA). The serum level of β-hydroxybutyrate was determined using β-hydroxybutyrate test kit (MAK041, Sigma) according to the manufacturer’s protocol. Hepatic and fecal lipids were extracted according to the optimized Folch method58. TC, TG, NEFA level was measured with the instruction manual (Nanjing Jiancheng Bioengineering Institute, China). Hepatic TNF-α, IL-6, IL-1β were detected by Mlbio ELISA kits (M1002095, M1063132, M1002293) following the manufacture’s protocol.
Histopathological evaluation
Liver samples were fixed in 4% paraformaldehyde, embedded in paraffin and sectioned (3 μm thickness). The sections were subjected to hematoxylin-eosin staining (H&E) staining and Sirius Red using staining standard procedures. Images were acquired by ImageScope (Leica Biosystems Imaging, Inc., USA) or Digital slice scanner (3D Histech, Pannoramic MIDI, Hungary). Liver pathology was scored by pathologists according to NAFLD activity scoring system59. The scores comprised steatosis (0–3), lobular inflammation (0–3), and ballooning (0–2) scores. Steatosis were graded based on the percentage of involved hepatocytes, 0 (<5%), 1 (5–33%), 2 (34–66%), and 3 (>66%). Inflammation was graded based on the number of inflammatory foci per field (200×), 0 (no foci), 1 (<2 foci), 2 (2–4 foci), and 3 (>4 foci). Ballooning was scored as 0 (none), 1 (few balloon cells), 2 (many balloon cells).
Oil Red O staining
The liver and ileum tissue fixed in 4% paraformaldehyde for 4 h were dehydrated for another 24 h and embedded in a frozen medium. The slices (8 μm) were stained with freshly prepared Oil Red O dye (D027, Nanjing Jiancheng Bioengineering) following the manufacturer’s protocol. The red lipid droplets were imaged using Digital slice scanner (3D Histech, Pannoramic MIDI, Hungary).
BODIPY gavage
The overnight fasted mice were gavaged with an olive-oil-infused bolus of fluorescent-labeled fatty acid BODIPY dye (2 μg/g BODIPY™ 500/510 C1, C12, Thermo Fisher, D3828) (10 μl/g mouse) and harvested 2 hours later. Jejunum sections were flash-frozen with OCT and then stained with DAPI. Slides were imaged using Digital slice scanner (3D Histech, Pannoramic MIDI, Hungary).
TUNEL staining
Samples of liver or ileum were fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned (3-μm thickness). The sections were subjected to TUNEL staining using staining standard procedures (Roche, 1168475910). Images were acquired by Digital slice scanner (3D Histech, Pannoramic MIDI, Hungary).
Quantitative analysis of BAs
BAs quantification was performed by Metabo-Profile Biotechnology (Shanghai) Co., LTD. All of the bile acids standards were synthesized by Metabo-Profile lab or obtained from Steraloids Inc. (Newport, RI, USA) and TRC Chemicals (Toronto, ON, Canada). For serum pretreatment, 20 μL sample serum was added to the 180 μL acetonitrile/methanol (v/v = 80:20) mixed solvent containing 10 μL internal standard into the 96-well plate. After shaking for 20 min at 200 × g, centrifugation was performed, and the supernatant was transferred to a microcentrifuge tube for freeze-drying. The dried samples were redissolved with 1:1 acetonitrile/methanol (v/v = 80:20) and ultrapure water, centrifuged at 4 °C for 20 min. The supernatant was transferred to a 96-well plate, and the injection volume was 5 μL. For tissue and intestinal contents: 10 mg of samples were weighted and homogenized with 180 μL acetonitrile/methanol (v/v = 80:20) mixed solvent containing 10 μL internal standard. After centrifugation for 20 min at 15,000 × g at 4 °C, the supernatant was transferred to 96-well plate and freeze-dried. After drying, the powders of actual samples, standard samples, and quality control samples were redissolved with 1:1 acetonitrile/methanol (v/v = 80:20) and ultrapure water, centrifuged at 15,000 × g at 4 °C for 20 min. The supernatant was transferred to a 96-well plate for UPLC/TQ-MS analysis with a volume of 5 μL. All the samples were run in a randomized order to minimize systematic analytical errors and pooled with quality control samples. The peak annotation and quantification were performed by MassLynx v4.1 and TargetLynx V4.1 (Waters Corp., Milford, MA, USA).
RNA sequencing analysis
RNA sequencing analysis was performed according to our previous published methods60. Briefly, total RNA was extracted by TRIzol method from frozen liver tissue. The library was constructed and sequenced by Shanghai Majorbio Bio-pharm Technology Co., Ltd. TruSeqTM RNA Sample Preparation Kit (Illmina, San Diego, CA) was used to construct RNA libraries. The high-throughput sequencing was performed via Illumina HiSeq XTEN/NovaSeq 6000 squencer (2 × 150 bp read length). SeqPrep (https://github.com/jstjohn/SeqPrep) and Sickle (https://github .com/najoshi/Sickle) were applied to trim and quality. Then clean reads were separately aligned to reference genome with orientation mode using HISAT2 (http://ccb.jhu.edu/software/hisat2/index. shtml)61 software. The mapped reads of each sample were assembled by StringTie (https://ccb.jhu.edu/software/stringtie/index.shtml?t=example) in a reference-based approach62. The expression of each gene was calculated according to Fragments Per Kilobases per Million reads63. RSEM (http://deweylab.biostat.wisc.edu/rsem/) was used to quantify gene abundances. The differential expression analysis was performed using the DESeq264. DEGs with |log2FC| > 1 and P adjust value ≤0.05 were considered to be significantly different expressed genes. The data were analyzed on the online platform of Majorbio Cloud Platform (www.majorbio.com).
Proteomics analysis
The quantitative proteomics analysis was performed at the Shanghai Institute of Materia Medica, Chinese Academy of Sciences65. Mouse liver tissues were lysed in SDT lysis buffer. The lysates were homogenized with sonication, denatured, and reduced at 95 °C for 5 min and then centrifuged at 12,000 × g for 25 min. The supernatants were collected and further quantified by a Bradford assay. Peptides were prepared following the Filter Assisted Sample Preparation procedure.
Peptide labeling was conducted with TMT 10-plex reagents according to the manufacturer’s protocol (Thermo Fisher Scientific). To increase the depth of protein identification, high-pH reverse-phase liquid chromatography was used for peptide fractionation. The peptides were separated using a Waters reversed phase XBridge BEH C18 column (150 × 2.1 mm, 3.5 μm) at a flow rate of 0.2 mL/min using Agilent 1200 HPLC systems.
The proteome analysis was performed on an Orbitrap Q-Exactive (Thermo Fisher Scientific) platform connected to an online nanoflow EASY nLC1200 HPLC system (Thermo Fisher Scientific). Peptides were loaded on a self-packed column (75 μm × 150 mm, 3 μm ReproSil-Pur C18 beads, Dr. Maisch GmbH, Ammerbuch, Germany) and separated with a 120 min gradient for each sample at a flow rate of 300 nL/min. A homemade column oven maintained the column temperature at 50 °C. A data-dependent acquisition MS method was used, in which one full scan (350–1700 m/z, R = 120,000 at 200 m/z) at a target of 3 × 106 ions was first performed, followed by top 20 data-dependent MS/MS scans with higher-energy collisional dissociation (HCD) at a resolution of 60,000 at 200 m/z. Other instrument parameters were set as follows: 32% normalized collision energy (NCE), 1 × 105 AGC target, 120 ms maximum injection time, 1.0 m/z isolation window.
Raw mass spectrometry data were processed using Maxquant 1.6.5.0 (Thermo Scientific) against the human Uniprot database, with a false-discovery rater (FDR) < 0.01 at the level of proteins and peptides. Further bioinformatics analysis was conducted in DAVID (https://david.ncifcrf.gov) for Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis. The fold changes >1.2 with p values < 0.05 were considered as a cutoff for differential proteins.
Cell culture and treatment
AML12 (alpha mouse liver-12) hepatocytes were purchased from National Collection of Authenticated Cell Cultures (SCSP-550) and cultured in DMEM/F-12 medium (Meilunbio, PWL107, China) supplemented with 10% fetal bovine serum, 1% ITS (10 µg/ml insulin, 5.5 µg/ml transferrin, 5 ng/ml selenium), 40 ng/ml dexamethasone and 1% penicillin-streptomycin, at 37 °C, 5% CO2. RAW264.7 (a mouse macrophage cell line) (ATCC, TIB-71) and HEK293T (ATCC, CRL3216) were cultured in DMEM (Meilunbio, PWL0212, China) containing 10% fetal bovine serum (FBS), 1% penicillin-streptomycin, at 37 °C, 5% CO2. No commonly misidentified cell line was used in this study. All the cell lines were routinely tested negative for mycoplasma contamination.
Protein isolation and western bolting
Liver tissues and cell sample lysates were prepared by RIPA buffer supplemented with 1 mM PMSF and phosphatase inhibitors. The protein concentration was measured by BCA Protein Quantification Kit (20201ES90, Yeasen Biotechnology, Shanghai, China). Specifically, the nuclear protein was acquired by a Nuclear and Cytoplasmic Protein Extraction kit (Beyotime Institute of Biotechnology, Shanghai, China). Proteins were separated by SDS-PAGE electrophoresis and transferred to PVDF membranes. Membranes were then probed with anti-PPAR alpha (Abcam, ab126285, ab215270, 1:1000 for WB), anti-CD36 (Abcam, ab133625, 1:1000 for WB), anti-FABP1 (Proteintech, 13626-1-AP, 1:1000 for WB), anti-CPT1A (D3B3) (Cell singaling, 12252 S, 1:1000 for WB), anti-CPT2 (Proteintech, 26555-1-AP, 1:1000 for WB), anti-HMGCS2 (Cell singaling, 20940 S, 1:1000 for WB), anti-NF-κB p65 (Cell singaling, 8242 S, 1:1000 for WB), anti-Phospho-NF-κB p65 (Ser536) (Cell singaling, 3033 S, 1:1000 for WB), anti-CRM1 (Santa Cruz Biotechnology, sc-74454, 1:500 for WB), anti-RAN (Santa Cruz Biotechnology, sc-271376, 1:500 for WB), anti-Lamin B1 (Abcam, ab229025, 1:1000 for WB), anti-GAPDH (Proteintech, 60004-1-Ig, 1:5000 for WB), anti-β-Actin (Cell singaling, 4970, 1:1000 for WB), anti-FLAG (Sigma, F1804, 1:1000 for WB), anti-HA (Proteintech, 51064-2-AP, 1:1000 for WB) antibodies followed by incubation with HRP-conjugated secondary antibodies (CST, 7076, 1:5000 for WB) (ABclonal, AS014, 1:5000 for WB). The signals were detected by chemiluminescence using the Amersham Imager 600 (GE, USA). All the unprocessed scans of bands were supplied in Source Data file or Supplementary information.
Real-time quantitative PCR
Total RNA was extracted with Trizol reagent (15596018, Thermo Fisher Scientific) from cells or tissues. cDNA was synthesized by the High Capacity cDNA Reverse Transcription Kit (KR118-02, TIANGEN). Then, quantitative PCR (qPCR) was carried out using the SYBR Green Master Mix (11201ES03, Yeasen Biotechnology, Shanghai, China) according to the manufacturer’s instructions. The primers were synthesized by BioSune Biotechnology (Shanghai) Co., LTD, and primer sequences of the genes are listed in Supplementary Table 2.
Immunofluorescence
The cells were fixed with 4% paraformaldehyde solution for 15 min and rinsed twice by PBS, then permeabilized with 0.5% Triton X-100 (ST797, Beyotime) at room temperature followed by PBS washing for twice. After washing, the cells were further incubated with 10% donkey serum (Jackson Immuno Research, USA) for 1 h to block non-specific antibody binding, then incubated with primary antibodies to PPARα (Abcam, 126285, 1:100) and subsequently incubated with ABflo® 594-conjugated Goat Anti-Rabbit IgG (H + L) (ABclonal, AS039, 1:100) for 1 h. After incubating with DAPI (P0131, Beyotime) for 10 min, images were directly captured using confocal microscopy (Leica, TCS SP8).
Samples of the liver were fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned (3 μm thickness). The sections were subjected to immunohistochemistry staining using standard procedures. The samples were incubated with primary antibodies to PPARα (Abcam, 215270, 1:100) overnight and subsequently incubated with ABflo® 488-conjugated Goat Anti-Rabbit IgG (H + L) (ABclonal, AS053, 1:100) for 50 min. After incubating with DAPI for 10 min, images were acquired by fluorescent microscope (NIKON ECLIPSE C1, Japan).
Luciferase reporter gene assays
After 24 h co-transfection with pCMV-Script-hFXR and PGL4-Shp-TK firefly luciferase plasmids66, followed by continuous 6 h 1% FBS DMEM medium, the HEK293T cells were exposed to DMSO or different concentration of HDCA (20 μM and 100 μM) for 24 h to test FXR activity. Cell activity assays and luciferase assays were then performed by CellTiter-Blue Reagent (G8080, Promega) and Firefly-Glo Luciferase Reaction (MA0519, Meilunbio), respectively. Data were collected by SpectraMax M5e (Molecular Devices, USA).
Human proteome microarray
The HuProtTM 20 K Human Proteome Microarrays (CDI Laboratories, Baltimore, MD) were used in this study. The experiments were performed by Wayen Biotechnologies (Shanghai, China) according to the following procedure. Briefly, proteome microarrays were blocked with blocking buffer (5% BSA in 0.1% Tween 20, PBST) for 1 h at room temperature with gentle agitation. Biotin-HDCA and biotin were diluted to 100 μM in PBST and incubated on proteome microarray at room temperature for 1 h (Biotin-HDCA was synthesized by Xi’an Ruixi Biological Technology Co., Ltd). After washing with PBST three times, microarray was incubated with 0.1% Cy5-Streptavidn solution for 20 min at room temperature in the dark, followed by three 5-min washes in PBST. The microarray was spun dry at 100 × g for 2 min and scanned with GenePixTM 4000B (Axon Instruments, CA). GenePixTM Pro v6.0 was used for data analysis. The median fluorescence signal at each site (F635_Median) was divided by the background (B635_Median) for data analysis, that is, original signal strength (I) = F635_Median/B635_Median. The median and SD of I were calculated to obtain the corrected data Z Score (corrected signal strength) for each site. Proteins with a Z Score greater than 2.8 were those that bind to HDCA. The mean value (I mean) represents the mean value of the original signal strength of each protein. I Mean Ratio is the ratio between the Biotin-HDCA group and Biotin group. To call the candidates, the cutoff was set as a P value ≤ 0.05 and I mean ratio ≥ 1.4.
Cellular thermal shift assay (CETSA)
AML12 cells were treated with DMSO or 100 μM HDCA for 1 h and washed with PBS. The cell pellets were resuspended in 1 mL PBS with protease inhibitor. Cells were divided into 100 μl aliquots and heated with a thermal gradient from 40 °C to 65 °C for 3 min. After freeze-thawing three times with liquid nitrogen, the supernatant was acquired by centrifugation at 2000 × g for 10 min at 4 °C, followed by Western blotting.
Co-immunoprecipitation (Co-IP)
All the GFP/Flag/HA-tagged plasmids (HA_PPARα_pcDNA3.1(+)-C-eGFP, FLAG_RAN_pcDNA3.1(+)-N-eGFP, CRM1_pcDNA3.1(+)-N-eGFP, G5A/P49A/ V51A-FLAG_RAN_pcDNA3.1(+)-N-eGFP) were produced and sequencing confirmed by GenScript (Nanjing, China). Co-IP was performed following the protocol of Co-immunoprecipitation Kit (10007D, invitrogen). The equilibrated beads were incubated with the Flag antibody (Sigma-Aldrich, F1804, 1:50 for IP), HA antibody (Proteintech, 51064-2-AP, 1:50 for IP), CRM1 antibody (Santa Cruz Biotechnology, sc-74454, 1:25 for IP) or Control IgG antibody (Rabbit Control IgG ABclonal, AC005, 1:100 for IP, Mouse Control IgG ABclonal, AC011, 1:100 for IP) at room temperature for 30 min and then incubated with the protein extracts at 4 °C overnight. The magnetic beads were collected by magnetic separator. All the non-specifically bounded proteins were removed by wash buffer. The bounded proteins were eluted from the beads with elution buffer for 10 min at 70 °C, then followed by Western blotting analysis.
Duolink PLA
The AML12 cells were treated with 100 μM HDCA or DMSO for 24 h. Duolink PLA was performed according to the protocol of Duolink® PLA (DUO092101, Sigma-Aldrich). The primary antibodies are anti-PPARα (Proteintech, 15540-1-AP, 1:25), anti-CRM1 (Santa Cruz Biotechnolog, sc-74454, 1:10), anti-CRM1 (ABclonal, A19625, 1:10), anti-RAN (Santa Cruz Biotechnolog, sc-271376, 1:20).
Molecular docking
Docking analysis was carried out by Swiss-Dock software from the Swiss Institute of Bioinformatics (http://www.swissdock.ch/)67, with default parameters. The Protein structure of GTP-binding nuclear protein RAN (UniProt: P62827) was chosen as a target using the target selection tab in Swiss-Dock, and the HDCA (PubChem CID: 5283820) structure was uploaded using the ligand molecular selection tab in Swiss-Dock. PyMOL software (version: 2.5.2) was applied for analysis of the interaction of residues between RAN and HDCA.
Statistical analysis
Except for RNA sequencing and proteomics, most of the plots were generated by GraphPad Prism 9.0 (GraphPad Software, San Diego, USA). The differential analysis was performed using the two-tailed Student’s t test, Kruskal–Wallis test, Mann–Whitney U, two-sided Fisher’s exact test, or one-way ANOVA followed by Tukey’s multiple comparison in the Graphpad Prism 9.0. Correlations between BA and phenotype were performed using SPSS 21.0 or Graphpad Prism 9.0. All data are expressed as mean ± standard error of mean (SEM) unless otherwise noted. Significance was set at p < 0.05.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.