Clinical samples

Blood samples (n = 60) were collected from RA patients from the Department of Rheumatology, All India Institute of Medical Sciences (AIIMS), New Delhi, India [11]. Similarly, blood samples (n = 40) were collected from HC with no prior ailment and joint inflammation. The medical history of each patient was collected (Supplementary Table 1) [12]. See details in supplementary file).

Metabolomics analysis

To analyze the differential metabolites, HPLC-MS/MS was carried out using RA (n = 20) and HC (n = 20) plasma samples. Two complimentary LC-MS/MS metabolomics methods were applied: HILIC and C18 chromatography. The raw LC-MS (.wiff files) data file was analysed by Peak View (ABSciex). Fold change criteria was considered to categorize upregulated (fold change ≥ 1.5) and downregulated (fold change ≤ 0.5) metabolites, respectively, and p-value < 0.05 was considered [13, 14]. (See details in supplementary file).

Proteomics of plasma samples: SWATH-MS acquisition

Plasma samples of RA (n = 60) and HC (n = 40) were taken, and a total of 70 µg protein was estimated by BCA and digested overnight at 37 °C with 0.1 µg/µl trypsin (Promega, USA) [12, 15]. (See details in supplementary file).

Integration of metabolomics and proteomics

Association analysis between metabolomics and proteomics data was performed using significant differential metabolites and protein profiles between RA and HC groups. Joint Pathway Analysis was carried out using MetaboAnalyst 5.0 (https://www.metaboanalyst.ca/). It enabled us to visualize significant genes and metabolites that were enriched in a particular pathway and integrated the underlying relationships among differentially expressed metabolites and proteins [16].

Target Prediction of metabolites

The major concern in drug discovery is to validate the best-screened active compounds’ interaction with appropriate targets [17]. To identify the potential molecular targets of the screened metabolites, PharmMapper server database (http://lilab.ecust.edu.cn/PharmMapper) was used [18, 19]. The gene targets were matched with protein profile identified by SWATH analysis. (See details in supplementary file)

Western blot (WB) and enzyme-linked immunosorbent assay (ELISA)

For WB analysis, 4 pooled plasma proteins of RA and HC were used. Each pooled sample consists of RA (n = 10) and HC (n = 10); thus, 40 RA and 40 HC samples were used, and 20 µg of pooled plasma proteins were run on SDS-PAGE and anti-IGF1 (Santa Cruz, USA) (1:2000) as the primary antibody and anti-mouse (1:5000) as a secondary antibody were used. The indirect ELISA was performed using diluted (1 µl/200µl) RA plasma (n = 60) and HC (n = 40), coated into 96-well micro-titer plates (Thermo Scientific, Nunc, USA), followed by incubation with primary antibody (Anti-IGF1) and secondary antibody (anti-mouse). The absorbance was observed at 495 nm [12, 20]. (See details in supplementary file).

Peripheral blood mononuclear cell (PBMC) isolation, RNA isolation, WB, and qRT-PCR

PBMCs are the primary immune cells in the human body and offer discriminatory immune responses toward inflammation [5]. PBMCs were isolated by centrifuging RA blood (n = 6) and HC (n = 6) using histopaque reagent [20] and then used to perform WB with 3 pooled RA and HC samples, respectively (n = 2 in each pooled sample) [15]. Total RNA was extracted from PBMCs of HC (n = 6) and RA (n = 6) using Tri-Xtract Reagent (G-biosciences). GAPDH as an internal reference. Primers are shown in Supplementary Table 2 (See details in the supplementary file). Similarly, in our earlier study, protein and mRNA levels of TTR were checked in PBMCs of RA blood [20].

Correlation analysis

The association of GUDCA levels with RA disease activity, specifically with ACCPA and DAS28-ESR scores was investigated [20, 21]. Additionally, the relationship between IGF1 levels, (measured by ELISA), and RA disease activity (assessed by the DAS28-ESR score) was examined. (See details in the supplementary file)

In-Vitro analysisHuman synovial fibroblast SW982 cell culture and MTT test

SW982 cells were cultured in DMEM media and treated with GUDCA metabolite (1–70µM range) for 24 h in serum-free media [22]. Absorbance was measured at 540 nm. See details in the supplementary file)

Total protein extraction and western blotting

SW982 cells were cultured and pre-treated with GUDCA (50µM- 6.25 µM) for 24 h. Protein was extracted in RIPA buffer after 10 min induction with TNF-α (10ng/ml) [23]. The blot was incubated with anti-p65, anti-IGF1, and anti-TTR separately as primary antibodies [22]. (See details in supplementary file)

Real-time quantitative reverse transcription PCR (qRT-PCR)

SW982 cells were cultured and incubated with GUDCA (50µM) for 24 h. The effect was investigated by TNF-α treatment (10 ng/ml) on GUDCA pretreated cells for 1 h. Total RNA was isolated and subjected to cDNA preparation, and mRNA expression was evaluated and quantitated using 2−ΔΔCT formula [22]. Human-specific primer sequences are shown in Supplementary Table 2. (See details in supplementary file)

Scratch assay analysis

SW982 cells were grown in a culture plate, the vertical scratch was drawn, and each scratch area was measured before and after the treatment with GUDCA (50µM). Bright-field images were taken at 0 h and 48 h and analyzed using a Nikon Eclipse 650 (NIKON, Tokyo, Japan) at ×10 magnification. The images were analyzed using ImageJ software [24]. (See details in supplementary file)

Total reactive oxygen species (ROS) estimation

SW982 cells were pretreated with GUDCA (50µM) with and without TNF-α (24 h), followed by adding 10µM working solution of DCFH-DA into each well for 30 min. Fluorescence images were taken by ZOE Fluorescent Cell Imager and analyzed by ImageJ software [22]. (See details in supplementary file)

In vivo studiesDevelopment of collagen-induced arthritis (CIA) rat model

Female Wistar rats (60–80 g) were procured from the ICMR -National Institute of Nutrition in Hyderabad, India. The work design was approved by the Institute’s Animal Ethical Committee (IGIB/IAEC/3/3/Mar 2023). The animals were randomly divided into four groups (n = 4). The untreated group/ healthy control (HC) (Group 1), Collagen-Induced Arthritis (CIA) (Group 2), vehicle control (VC + CIA) (Group 3), and GUDCA treated (CIA + GUDCA) (Group 4). CIA rats (Except the HC group) were then induced with 2 mg/ml collagen (Type II) from chicken (Sigma, USA). GUDCA was administered at 800 µg/Kg of rat body weight mixed with corn oil/ benzyl alcohol (95:5 v/v) and was injected subcutaneously [25, 26]. (See details in supplementary file)

Measurement of CIA induction in experimental groups and detection of RA

Throughout the study, paw volume and arthritis index (AI) were assessed in individual animals to monitor disease progression from day 0th to day 28th [27]. Splenic index and liver index were calculated for each rat as the ratio of the spleen/liver: body weight [28]. (See details in supplementary file)

Enzyme-linked immunosorbent assay (ELISA) of cytokines in plasma

Rat plasma was separated and added (100 µl) to the pre-coated ELISA plate, followed by the manufacturer’s guidelines. TNFα, IL(Interleukin)1β, and IL-6 cytokines were quantified using ELISA kits (ELK Biotechnology, China) [12].

Hematoxylin and eosin staining (H & E)

Rat synovium was sliced and fixed in 10% formalin, fixed in the paraffin block, sliced (5 μm thick) using a microtome, and slides were prepared. Slides were viewed under a Nikon microscope. Images of the slides at 10X magnification were taken, and Image-J software was used to analyze the images [25] (See details in supplementary file).

Western blot analysis of the CIA model rat plasma

The blood samples were drawn through direct heart puncture and collected in EDTA coated vacutainer tubes (BD, Franklin Lakes, NJ, USA), and plasma was separated. Similarly, synovial tissue was collected and crushed in liquid nitrogen from each rat, and the lysate was prepared in RIPA buffer and centrifuge; further supernatant was stored at -80 °C for further analysis. For WB analysis, the total protein (10 µg) concentration was run on the gel as mentioned above [29] (See details in supplementary file).

Enzyme-linked immunosorbent assay (ELISA) of synovium

The indirect ELISA was performed using synovium lysate of all groups diluted (10 µl/90µl) with coating buffer into 96-well micro-titer plates (Thermo Scientific, Nunc, USA) and incubated overnight at 4ºC and proceeded as mentioned above [29] (See details in supplementary file).

Statistical analysis

All non-parametric Mann–Whitney tests were performed using Graph pad Prism 9.0. The complete data set was analyzed using MetaboAnalyst 3.0. Pathway enrichment analysis was performed to find the related pathway with the altered metabolites. Statistical analysis was performed with the paired student’s t-test to compare the data between two groups, and ANOVA was used to compare data among multiple groups. The obtained p-values were represented by asterisks on the graph (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001). Each experiment was repeated at least three times.