Abstract Dysbiosis of gut microbiota significantly exacerbates the progression of rheumatoid arthritis (RA). Targeting gut microbiota may present a promising therapeutic strategy for RA. Gold nanospheres (GNS), known for excellent biocompatibility, stability and minimal toxicity, have emerged as precise modulators of gut microbiota, reshaping intestinal environments to treat various inflammatory diseases. Our study found that oral administration of 60-nm GNS effectively ameliorated collagen-induced arthritis (CIA) in mice, with a marked reduction in disease severity and synovial inflammation. Specifically, GNS notably enriched the probiotic genus Ligilactobacillus while restoring intestinal barrier function by upregulating tight junction proteins Claudin-1 and ZO-1. Targeted metabolomics analysis revealed GNS substantially increased the production of indole-3-propionic acid (IPA) and indole-3-acetic acid (IAA) in gut, which were shown to activate the aryl hydrocarbon receptor (AhR) pathway. Mechanistic studies demonstrated that the IPA/IAA mixture inhibited PTEN ubiquitination, stabilizing PTEN protein levels and suppressing NF-κB activation in synovial tissues. These changes correlated with reduced synovial hyperplasia and inflammatory infiltration. Our findings established GNS as an effective nanomodulator of the gut-joint axis, providing novel insights into microbiota-targeted therapies for RA and other inflammatory diseases. Graphical abstract [46]graphic file with name 12951_2025_3450_Figa_HTML.jpg Supplementary Information The online version contains supplementary material available at 10.1186/s12951-025-03450-7. Keywords: Rheumatoid arthritis (RA), Gold nanospheres (GNS), Gut microbiota, Tryptophan metabolism, Fibroblast-like synoviocytes (FLSs) Introduction Rheumatoid arthritis (RA) is a prevalent chronic autoimmune inflammatory joint disorder that affects tens of millions worldwide, leading to progressive joint damage, functional decline and irreversible joint deformity and disability [[47]1]. Persistent synovial inflammation and hyperplasia drive cartilage degradation and bone erosion in this autoimmune disorder [[48]2]. The etiology of RA is multifactorial, involving complex genetic, sex-related, and environmental factors [[49]3]. In recent years, the substantial role of gut microbiota in RA progression has attracted significant attention. Population-based studies have documented significant changes in the composition and diversity of the gut microbiota in RA patients compared to healthy individuals [[50]4, [51]5]. Notably, Prevotella copri is significantly increased in the gut microbiota of patients with untreated new-onset RA, correlating with a decrease in Bacteroides spp and the loss of beneficial microorganisms [[52]6]. Certain bacteria, such as Prevotella histicola, Bifidobacteria, and Bacteroides fragilis, can modulate the incidence and severity of RA [[53]7–[54]9]. For instance, Fusobacterium nucleatum is enriched in RA patients and aggravates RA progression by producing FadA-containing outer membrane vesicles, which activate the crucial modulator of inflammatory mediators YB-1 in synovial macrophages [[55]10]. In addition, Parabacteroides distasonis has been proved to ameliorate RA pathogenesis via the production of four bile acids, significantly inhibiting the differentiation of T-helper type-17 (Th17) cells and promoting the M2 polarization of macrophages [[56]11]. Interestingly, Lingjuan Jiang et al. found that Prevotella copri is abundant in RA patients, which cooperates with a high-fiber diet, promotes the digestion of complex fibers, leading to the excessive production of organic acids, triggering a pro-inflammatory response in macrophages and exacerbating arthritis [[57]12]. Collectively, RA results in imbalances among various gut microbiota and significant disturbances in metabolites related to energy production, tryptophan, fatty acids, and bile acids metabolism [[58]11, [59]13, [60]14]. Gut microbiota-derived metabolites performed a crucial role in the gut-joint axis, mediating the regulatory effects of gut microbiota on RA progression. These metabolites, particularly microbiota-derived tryptophan metabolites such as indole-3-propionic acid (IPA) and indole-3-acetic acid (IAA), participate in the progression of multiple diseases, including atherosclerotic cardiovascular disease, inflammatory bowel disease, obesity, and type 2 diabetes mellitus [[61]15–[62]18]. However, the specific mechanisms by which these tryptophan metabolites influence RA-associated inflammation and immune dysfunction remain poorly understood. These metabolites, which act as ligands for the aryl hydrocarbon receptor (AhR), have been implicated in regulating intestinal homeostasis and immune cell function. Still, their precise role in RA pathogenesis and therapeutic intervention has not been fully elucidated [[63]19]. Current treatments for RA rely on global immune suppression, which carries significant risks of adverse effects [[64]20]. The gut microbiota presents a promising therapeutic target, as its composition can be modulated by lifestyle, diet, and environmental factors [[65]12]. In this context, metallic nanoparticles (MNPs) offer unique advantages due to their biocompatibility, stability, and minimal toxicity [[66]21]. In particular, gold nanoparticles (GNPs) have demonstrated excellent anti-inflammatory and immunomodulatory properties. For instance, GNPs could affect macrophage and Th17 cell activity, promoting anti-inflammatory cytokine release and improving overall immune status [[67]22–[68]24]. Among various GNPs, gold nanospheres (GNS) are notable for modulating gut microbiota and are beneficial regulators in diseases like Alzheimer’s, osteoarthritis, and osteoporosis [[69]25–[70]27]. Research by Hua Kuang et al. showed that oral chiral GNS administration reshapes gut microbiota, increasing Lactobacillus and Clostridium levels while reducing Enterobacter and Desulfovibrio levels. Chiral GNS intervention notably alters tryptophan metabolism, boosting gut metabolites such as IAA, enhancing brain immune environments by increasing regulatory T cell proportions, reducing Th17 cells and IL-17γδ T cells, and lowering NLRP3 expression, thereby reducing neuroinflammation [[71]25]. However, in the context of RA, the potential of GNS to reshape gut microbial communities and influence tryptophan metabolism represents an understudied area with significant therapeutic implications. Our study aimed to investigate the role of GNS in modulating gut microbiota and tryptophan metabolism to alleviate RA symptoms. We found that GNS significantly improved joint damage and inflammation in CIA and reshaped intestinal flora microecology. GNS enriched microbial-derived tryptophan metabolites such as IPA and IAA, which reduced RA-related inflammation and immune responses via AhR activation. These results suggested that GNS-mediated modulation of gut microbiota and tryptophan metabolism represented a novel therapeutic strategy for RA, with the potential to restore immune balance and reduce inflammation. Our study bridged a critical gap in understanding the interplay between gut microbiota, tryptophan metabolism, and RA pathogenesis while providing mechanistic insights into the therapeutic potential of GNS in autoimmune diseases. Materials and methods Reagents and materials In this study, GNS were synthesized using the citrate reduction method and obtained from Wuhan MICE Biotechnology Co. Ltd. The synthesis of GNS was conducted in two main steps. Initially, trisodium citrate was employed as a reducing agent and a stabilizer, while chloroauric acid was the gold precursor. This involved the rapid addition of trisodium citrate to a boiling chloroauric acid solution in ultrapure water, followed by continued boiling for 20 min to produce gold seeds. In the subsequent step, the gold seeds acted as the core, with trisodium citrate as the stabilizer and ascorbic acid as the reducing agent. By varying the quantity of chloroauric acid, which acts as the gold precursor, gold shells of different thicknesses were formed, resulting in GNS with a range of particle sizes. The GNS samples underwent ultrasonic dispersion in water and were placed in a potential pool to measure zeta potential using a Zetasizer Nano ZS90. The morphological properties of the GNS, such as size distribution and homogeneity, were analyzed with a scanning electron microscope (Hitachi Regulus SU8100). The average nanoparticle size was used as the representative size for the samples in this study. CIA mice model establishment and relevant treatments Male DBA/1 J mice were procured from GemPharmatech (Nanjing, China). All experiments were conducted with gender- and age-matched DBA/1 J mice housed in a specific pathogen-free (SPF) environment with a 12-h light/dark cycle and ad libitum access to food and water. Experimental procedures were performed by the ethical regulations for animal care and use in China and approved by the Army Medical University (Approval No. AMUWEC20232385). The CIA model was established according to the protocol by Jiang et al. [[72]12]. Briefly, 6-week-old male DBA/1 J mice received a subcutaneous injection at the tail base on day 0 with mixed emulsion of bovine type II collagen (Cat#: 20,022, Chondrex Inc.) and complete Freund’s adjuvant (CFA, Cat#: 7001, Chondrex Inc.), reaching a final concentration of 1 mg/mL collagen and 0.5 mg/mL M. tuberculosis. Each mouse received 100 µL of this emulsion. A booster injection with an emulsion of collagen and incomplete Freund’s adjuvant (IFA, Cat#: 7002, Chondrex Inc.) was administered on day 21, maintaining the collagen concentration at 1 mg/ml, with each mouse receiving 100 µL. All mice were randomly assigned to five experimental groups: CIA mice treated with normal saline (RA + NS, n = 5), antibiotics (RA + ABX, n = 5), GNS (RA + GNS, n = 5), GNS along with AhR antagonist CH-223191 (RA + GNS + CH, n = 5), and CIA mice underwent fecal microbiota transplantation (FMT) from the RA + GNS group (FMT (RA + GNS), n = 5). GNS was administered orally at 0.01 mg/g body weight (bw) /day from day 21 to day 42 in the RA + GNS group. RA + NS mice received an equal volume of normal saline. The RA + GNS + CH group additionally received CH-223191 (Cat#: C303374, Aladdin) at 10 mg/kg bw/day via gavage. Procedures for FMT and ABX are detailed in the next section. A tryptophan metabolite mixture, comprising equal proportions of IAA (Cat#: I101072, Aladdin) and IPA (Cat#: I103959, Aladdin), dissolved in 0.5% CMC-Na, was administered to CIA mice (IPA + IAA group) via oral gavage daily from booster immunization until the end of this study (500 mg/kg). Control mice (CT group) received an equivalent volume of 0.5% CMC-Na. FMT and ABX treatment For FMT treatment, fresh feces from the RA + GNS donor mice were collected in sterile tubes, diluted with normal saline, and filtered using sterile gauze to prepare a fecal suspension, which was then administered to the recipient mice, FMT (RA + GNS) group via oral gavage (200 µl/mouse) until the end of this experiment. Notably, post booster immunization, recipient mice received a broad-spectrum ABX treatment for five days to deplete gut bacteria. This treatment included Vancomycin, 100 mg/kg bw/day (Cat#: MB1260, MeilunBio); Neomycin sulfate, 200 mg/kg bw/day (Cat#: N412785, Aladdin); Metronidazole, 200 mg/kg bw/day (Cat#: M109874, Aladdin); and Ampicillin Na, 200 mg/kg bw/day (Cat#: A105483, Aladdin). For the ABX treatment, a continuous ABX cocktail regimen described above was administered orally every other day from booster immunization to the end of this study. Assessment of arthritis severity Arthritis severity was evaluated as previously reported [[73]28]. Clinical scores, based on the severity of arthritic limbs, were assigned as follows: 0 = no symptoms, 1 = mild swelling and erythema limited to tarsals or ankle joint, 2 = mild swelling and erythema spreading from ankle to tarsals, 3 = moderate swelling and erythema spreading to metatarsal joints, and 4 = severe swelling encompassing the ankle and foot or ankylosing deformity. The clinical score for each mouse was the sum of the scores for each paw. Scoring was conducted every three days during the observation period. Histological evaluation The bilateral hind paws of mice were fixed in 4% paraformaldehyde, decalcified with 10% ethylenediaminetetraacetic acid (EDTA) for one week, and embedded in paraffin. These samples were then deparaffinized, rehydrated, and stained with hematoxylin–eosin (H&E) and tartrate-resistant acid phosphatase (TRAP) stains. To assess potential GNS side effects, H&E staining was also conducted on liver and kidney tissue samples from the mice. The paw-scoring criteria were conducted according to the method described by Jiang et al. [[74]12]. Immunohistochemical analysis Paw tissue sections were blocked with 5% BSA and incubated overnight at 4 °C with primary antibodies: anti-PTEN (Cat#: 60,300–1, Proteintech, 1:200), anti-p-p65 (Cat#: bs-3543R, Bioss, 1:200). Sections were subsequently incubated with secondary antibodies for one hour at room temperature. To evaluate intestinal permeability, colonic tissue sections were similarly blocked and incubated overnight with primary antibodies against ZO-1 (Cat#: PB9234, Boster, 1:200) and Claudin-1 (Cat#: A21770, ABclonal, 1:200), followed by secondary antibody incubation. Bone microstructure indexes assessment Mouse paws were harvested and fixed in 4% paraformaldehyde. Subsequently, ex vivo micro-computed tomography (micro-CT) was performed using a Bruker Micro-CT Skyscan 1272 system (Kontich, Belgium) at a resolution of 7 μm. The acquired images were reconstructed using Nrecon software (Ver. 1.6.10, Kontich, Belgium) and further processed with CT analyzer software (Kontich, Belgium). A spherical region of bone tissue, with a diameter of 1.2 mm centered on the metacarpophalangeal joint, was designated as the region of interest for quantitative micro-CT analysis. Additionally, 3D morphological analysis was conducted using CTvox 3D-model visualization software (Bruker Micro-CT, Ver. 3.3.1.0). Bone morphometric parameters, including bone mineral density (BMD), bone volume fraction (BV/TV), trabecular thickness (Tb.Th), trabecular separation (Tb.Sp), and trabecular number (Tb.N), were quantitatively assessed. Elisa assay The quantification of serum lipopolysaccharide (LPS) was investigated using Mouse LPS ELISA Kits (Cat#: rxj202425M, Quanzhou Ruixin Biotechnology, Co., Ltd.). All operations were performed according to the manufacturer’s instructions. Intestinal permeability assay We applied Fluorescein isothiocyanate-dextran 4 (FD4, Cat#: 60,842–46-8, Sigma-Aldrich) to perform an in vivo permeability assay, thereby evaluating the integrity of the intestinal barrier. After oral administration of FD4 (40 mg/100 g BW), blood samples were collected from mice 4 h later, centrifuged, and the resulting serum was transferred to a fresh centrifuge tube. The concentration of FD4 in the serum samples was quantified using a microplate system. The measurement results and the FD4 concentration standard curve were entered into data analysis software to calculate the FD4 content in the samples. Fecal metabolomics profiling Fecal metabolomics involved preparing 50 mg of each sample in Eppendorf tubes, treated with 500 μL of pre-cooled (−40 °C) extraction solution (methanol: acetonitrile: water, 2:2:1, including 0.1% formic acid and isotopically labeled internal standard). After 30 s of vortexing and 4 min of homogenization at 35 Hz, the mixture was sonicated in an ice-water bath for 5 min, twice. Samples were precipitated at −40 °C for an hour, centrifuged at 12,000 rpm for 15 min at 4 °C, and a 320 μL aliquot of the supernatant was dried under nitrogen and reconstituted in 80 μL of 0.1% formic acid solution. A final centrifugation at 12,000 rpm for 15 min at 4 °C produced a clear supernatant for UHPLC-MS/MS analysis. The ACQUITY Premier system (Waters) with a Waters ACQUITY UPLC HSS T3 column (100 × 2.1 mm, 1.8 μm) was used at 40 °C, with the auto-sampler at 10 °C and a 5 μL injection volume. Solvents were 0.1% formic acid in water and acetonitrile. Mass spectrometry was conducted on a SCIEX Triple Quad™ 6500 + spectrometer, equipped with an IonDrive Turbo V ESI interface. Data acquisition and MRM analysis were handled by SCIEX Analyst Work Station Software (v1.6.3) and Sciex MultiQuant Software (v3.0.3). Fecal 16S rDNA gene sequencing and analysis Fecal samples from experimental mice were gathered in sterile 2 mL cryovials and stored at −80 °C. DNA extraction from the RA + NS, RA + GNS, and RA + GNS + CH groups was performed using the CTAB method, as per the manufacturer’s guidelines, and resuspended in 50 μL of Elution Buffer. Prokaryotic 16S fragments were amplified under PCR conditions: initial denaturation at 98 °C for 30 s, 32 cycles of 98 °C for 10 s, 54 °C for 30 s, 72 °C for 45 s, and a final extension at 72 °C for 10 min. PCR products were confirmed by 2% agarose gel electrophoresis, purified with AMPure XT beads, and quantified using the Qubit system. Amplicon libraries were prepared and evaluated for size and quantity with the Agilent 2100 Bioanalyzer. Sequencing was conducted on the NovaSeq PE250 platform after library quantification. Sequences were assigned to samples via unique barcodes, with barcodes and primers removed. High-quality clean tags were generated using FLASH and fqtrim, and chimeric sequences were eliminated with Vsearch. DADA2 dereplication produced a feature table and sequence data. Alpha and beta diversities were calculated post-normalization using the SILVA classifier. Alpha diversity indices (Chao1, Observed species, Goods coverage, Shannon, Simpson) were derived using QIIME2. Beta diversity was assessed and visually represented using R packages. Sequence alignment and annotation were performed with Blast and the SILVA database, respectively, followed by additional analyses using the R package. Cell culture and cell viability assay MH7A cells were purchased from BeNa Culture Collection (BNCC, BNCC371792) and cultured in Dulbecco’s Modified Eagle Medium (DMEM; Cat#: G4524, Servicebio) supplemented with 10% fetal bovine serum (FBS; Cat#: C04001-050X10, VivaCell), and 100 U/mL penicillin–streptomycin solution (Cat#: C0222, Beyotime). Cells were maintained at 37 °C with 5% CO2. To measure the proliferation of MH7A cells, a Cell Counting Kit-8 (Cat#: C0038, Beyotime) was utilized according to the manufacturer’s instructions. 5 × 10^3 MH7A cells were seeded into 96-well plates and cultured overnight. Then they were exposed to different concentrations of IPA/IAA mixture (0, 1, 5, 10 μM) for 12, 24, 36, or 48 h. 10 μL CCK-8 buffer was added to each well and incubated for 0.5 h at 37 °C with 5% CO2 in a humidified incubator. Wells without cells but containing the CCK-8 reagent were designated as the blank. Ultimately, the OD at 450 nm was measured. Flow cytometric analysis To analyze the effect of IPA + IAA mixture on cell apoptosis, MH7A cells were treated with IPA + IAA mixture (10 μM). Then the cell apoptosis was detected by using the Annexin V-FITC/PI Apoptosis Detection Kit (Cat#: MA0220, Meilun Bio). Briefly, 1 × 10^6 cells were stained with 5 μL of annexin V-FITC and 10 μL PI. Apoptosis was analyzed using a FACS flow cytometer (CytoFLEX, USA) and analyzed with Flow Jo software (Flow Jo, USA). EdU staining To evaluate the proliferation of MH7A cells, the EdU solution (10 μM) was added to 10% FBS DMEM for two hours. Then, the cells were fixed with 4% paraformaldehyde (PFA) and permeabilized with 0.3% Triton X-100 for 15 min. The Elabscience® E-Click EdU Cell Proliferation Imaging Assay Kit (Cat#: E-CK-A377, Elabscience) was used to detect EdU-positive cells according to the manufacturer’s instructions. The IX81 inverted fluorescence microscope (Olympus, Japan) was used to visualize and analyze the EdU-positive cells. Transwell assay To assess cell migration, a Transwell insert with an 8 μm pore size polycarbonate filter membrane (Corning) was utilized. For invasion, additional Matrigel (Corning) was diluted (1:6) and added to the surface of the upper chambers followed by resting at 37˚C for 2 h. 5 × 10^3 MH7A cells treated with IPA + IAA mixture (10 μM) or not were seeded into the upper chambers in 2% FBS DMEM, while the lower chambers were loaded with 10% FBS DMEM. The following day, cells on the bottom surface of the filter membranes were stained with crystal violet staining solution (Cat#: C0121, Beyotime) for 10 min, washed 3 times, and imaged using the ECLIPSE Ni microscope (Nikon). The images were then analyzed using NIS-Elements D software (Nikon). RNA-sequencing assay RNA extraction was conducted using Trizol reagent (ThermoFisher, Cat#: 15,596,018) per the manufacturer’s instructions. RNA concentration and quality were assessed with the Bioanalyzer 2100 system and RNA 6000 Nano LabChip Kit (Agilent, USA, 5067–1511). Samples with an RNA Integrity Number (RIN) above 7.0 were selected for library preparation. From 5 μg of total RNA, mRNA was purified using Dynabeads Oligo (dT) (ThermoFisher, USA). The purified mRNA was fragmented at 94 °C for 5–7 min with the Magnesium RNA Fragmentation Module (NEB, USA). These fragments were converted into cDNA using SuperScript™ II Reverse Transcriptase (Invitrogen, USA). E. coli DNA polymerase I, RNase H, and dUTP Solution (NEB, ThermoFisher, USA) were used to synthesize U-labeled second-strand DNAs. The strands underwent end-repair, adenylation, and ligation to indexed adapters with a T-base overhang. Size selection of the ligation products was achieved using AMPure XP beads. Heat-labile Uracil-DNA Glycosylase (NEB, USA) was applied to U-labeled second-strand DNAs, followed by PCR amplification: initial denaturation at 95 °C for 3 min, 8 cycles of denaturation at 98 °C for 15 s, annealing at 60 °C for 15 s, extension at 72 °C for 30 s, and a final extension at 72 °C for 5 min. The cDNA library’s mean insert size was 300 ± 50 bp. Paired-end sequencing (2 × 150 bp, PE150) was performed on the Illumina NovaSeq™ 6000 system, adhering to the manufacturer’s protocol. Immunofluorescence staining MH7A cells were cultured with TNF-α (10 ng/mL) in the presence or absence of IPA + IAA mixture (10 μM) for 1 h before treatment of VO-Ohpic (5 μM; Cat#: 675,848–25-6, MedChemExpress). Then the cells were fixed with 4% PFA for 15 min and blocked in QuickBlock™ Blocking Buffer (Cat#: P0231, Beyotime) for 30 min. Cells were washed three times with PBS and then incubated with anti-p65 antibody (Cat#: bsm-33117 M, Bioss, 1:200) overnight at 4 °C. The following day, the cells were washed three times with PBS before being incubated with a CoraLite® Plus 594-conjugated secondary antibody (Cat#: RGAR004, proteintech, 1:200) and DAPI for 1 h at room temperature in the dark. The intensity value was analyzed by using the Image J. Western blot The total cellular lysates were obtained by lysing MH7A cells with RIPA buffer (25 mM Tris–HCl pH 7.6, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS) and a mixture of protease (Cat#: BL630B, Biosharp) and phosphatase (Cat#: P1081, Beyotime) inhibitors. Then, the cell lysates were sonicated (20% power, 5 s each time, 15 repetitions) and centrifuged at 12 000 rpm for 15 min at 4 °C. Equal amounts of total proteins (40 μg) were separated by SDS-PAGE (prepared with 5% acrylamide in the stacking gel and 10% acrylamide in the separating gel) and transferred onto PVDF membrane. The primary antibody anti- NF-κB p65 (Cat#: bsm-33117 M-1, bioss, 1:1000), anti-Phospho-NFKB p65 (Ser276; Cat#: bs-3543R, bioss; 1:1000), anti-IkB-α (Cat#: BS3601, bioworld; 1:1000), anti-p-IkB-α (phospho-S32/S36; Cat#: BS4105, bioworld; 1:1000), anti-PTEN (Cat#: sc-7974, santa cruz; 1:1000), anti-p-PTEN (Cat#: sc-377573, santa cruz; 1:500) was added to the PVDF membranes and incubated overnight at 4 °C. Next day, the PVDF membranes were washed 3 times with TBST and incubated with a secondary antibody for 1.5 h at room temperature. Chemiluminescence detection was performed with a BeyoECL Plus ECL kit (Cat#: P0018S, Beyotime) and imaged with a Bio-Rad ChemiDoc™ Touch Imaging System. The signal intensity of each protein band was measured with ImageJ. Immunoprecipitation and immunoblotting The MH7A cells were lysed using IP lysis buffer (50 mM Tris–HCl (pH 7.4), 150 mM NaCl, 1 mM EDTA, 1% Triton X-100) and sonicated as described above. The lysates were collected and incubated with an anti-PTEN antibody (Cat#: sc-7974, santa cruz; 1:50) for 2 h at 4 °C. Protein A/G beads (Cat#: KM0134, DIA-An Biotechnology) were then added to the cell lysates and incubated overnight at 4 °C after being washed 4 times with RIP binding buffer (1 × PBS, 1% TritonX-100, 0.01% NP-40, 5% Glycerol). To collect immune complexes, beads were centrifuged and heated at 95 °C for 5 min in 2 × SDS loading buffer. Then, the immunoprecipitated complexes were analyzed by Western blot using anti-Ubiquitin (P4D1) antibody (Cat#: sc-8017, santa cruz; 1:1000). Statistical analysis All quantitative data were analyzed using a specific software (GraphPad Prism 7.0), with results presented as mean with standard deviation. Unpaired Student’s t-tests were employed to assess the significance of differences between the two groups. Welch’s correction was applied when the F test was significant. The Kruskal–Wallis test was employed for bacterial taxonomic evaluations. Associations were determined through Spearman’s rank correlation analysis. Significant differences were indicated by *P < 0.05; **P < 0.01; ***P < 0.001; and ****P < 0.0001. No significant difference was indicated by NS. Results Synthesis and characterization of GNS Initially, we performed a comprehensive characterization of the synthesized GNS to determine its structure and physical properties. As shown in Fig. [75]1A, the scanning electron microscope (SEM) images exhibited particles with an average diameter of approximately 60 nm, indicating high uniformity and minimal aggregation of GNS. Additionally, zeta potential measurements reflected the surface charge properties of the synthesized GNS, approximately − 13.4 mV, demonstrating that the GNS is dispersively stable in solution, as the negative charge is sufficient to prevent particle aggregation (Fig. [76]1B). The experimental design and methodology for utilizing GNS in subsequent animal and cell experiments were shown in Fig. [77]1C. Fig. 1. [78]Fig. 1 [79]Open in a new tab The characterization of GNS and subsequent experimental design. A Representative SEM images of GNS showing 60 nm diameter particles with high uniformity. B Zeta potential measurement of GNS, showing − 13.4 mV surface charge for dispersive stability. C Experimental design for GNS application in animal and cell experiments GNS alleviated CIA in a gut microbiota-dependent manner and AhR antagonist CH-223191 reversed the protective effects of GNS To assess the effects of GNS on RA, CIA mice were randomly divided into two groups: one receiving GNS at a dose of 0.01 mg/g bw/day (RA + GNS group) and the other receiving an equivalent volume of saline (RA + NS). Previous study has demonstrated that GNS could maintain gut microbiota homeostasis and enrich tryptophan metabolites to activate AhR, thereby alleviating AD progression. Therefore, we have included antibiotic intervention group (RA + ABX group), fecal microbiota transplantation group (FMT(RA + GNS) group), and an AhR antagonist CH-223191 intervention group (RA + GNS + CH group) (Fig. [80]1C) [[81]25]. GNS exhibited great biocompatibility without liver and kidney toxicity (Supplementary Fig. [82]1A and B). Compared with the RA + NS group, GNS administration significantly alleviated the severity of RA, decreasing the arthritis score, which demonstrated effective protective effects on RA (Fig. [83]2A). H&E staining showed that GNS treatment markedly reduced synovial hyperplasia and inflammatory infiltration, and maintained joint space and cartilage integrity, significantly alleviating RA-induced pathological changes (Fig. [84]2B and C). Additionally, TRAP staining exhibited significantly reduced activity and number of TRAP-positive osteoclasts in RA + GNS mice compared to untreated RA control groups. The areas of bone erosion and resorption were notably diminished, indicating a reduction in bone destruction and an attenuation of the osteolytic activity typically associated with RA (Fig. [85]2D and E). To further verify the therapeutic effects of GNS, we assessed the bone erosion in CIA mice through three-dimensional micro-CT analysis. The three-dimensional reconstruction of bone structure in the metacarpophalangeal joints exhibited a more intact trabecular structure, fewer cracks and gaps, more complete cortical bone, and reduced areas of erosion in the RA + GNS group (Fig. [86]2F). Quantitative analyses of bone histomorphometry index showed higher BMD, BV/TV, and Tb.N, as well as lower Tb.Sp, further demonstrating that GNS significantly improved joint bone destruction and mitigated RA progression (Fig. [87]2G). Fig. 2. [88]Fig. 2 [89]Open in a new tab GNS protected against RA-induced joint damage and bone erosion in CIA mice. A Arthritis scores in CIA mice. * indicated significant differences between RA + GNS and RA + NS, # indicated significant differences between RA + GNS + CH and RA + GNS, + indicated differences between RA + ABX and RA + NS, and ▲indicated differences between FMT(RA + GNS) and RA + NS. B Representative histology images of joints. C Histological scores of joints. D Representative TRAP staining images of joints. E Quantitative analyses of TRAP staining. F Representative photographs and micro-CT images of paws. The bright areas in the micro-CT images of the metacarpophalangeal joints on the left were the delineated region of interest (ROI) used for subsequent bone histomorphometry index analysis. G Quantitative analyses of bone histomorphometry indexes including bone mineral density (BMD), bone volume/total volume (BV/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th), and trabecular separation (Tb.Sp) in CIA mice. H Representative immunohistochemical images showing in situ expression of Claudin-1 and ZO-1. I and J The Average Optical Density (AOD) of Claudin-1 and ZO-1 To investigate whether the protective effects of GNS on the CIA mice are mediated by the gut microbiota, we performed ABX intervention and gut microbiota transplantation from the RA + GNS group to the FMT(RA + GNS) group. Compared with RA + NS mice, RA + ABX and FMT(RA + GNS) groups exhibited marked attenuated CIA-associated symptoms (Fig. [90]2A). These changes were accompanied by improved metacarpophalangeal joints, synovial hyperplasia and bone erosion (Fig. [91]2B–G), indicating that gut microbiota performed an important role in RA progression. Microbial communities in the faeces of GNS-treated CIA mice showed comparable ameliorative effects against CIA as observed in the GNS-treated group. Interestingly, the intervention of AhR antagonist CH-223191 reversed the protective effects of GNS on RA progression, indicating that the AhR pathway is integral to the therapeutic benefits of GNS in ameliorating RA (Fig. [92]2A–G). Additionally, intestinal permeability analysis showed that GNS treatment produced significant improvements. we observed a marked reduction in both FD4 fluorescence intensity and serum LPS levels. These changes indicate that GNS effectively alleviated the abnormal increase in intestinal permeability during RA progression (Supplementary Fig. [93]1C and D). Consistently, GNS significantly upregulated the expression of the critical tight junction proteins including Claudin-1 and ZO-1, indicating a significant enhancement in gut barrier function. Notably, CH-223191 similarly reversed this upregulation (Fig. [94]2H–J). These findings supported our hypothesis that GNS may significantly alleviated the progression and severity of CIA by promoting gut microbiota homeostasis and enriching tryptophan metabolites, which in turn activate the AhR pathway and improve gut permeability. GNS reversed gut dysbiosis in CIA mice As mentioned above, GNS contributed to the improvement of RA in a gut microbiota-dependent manner, therefore, we examined the alteration in the composition and structure of gut microbiota mediated by GNS treatment via 16S rDNA gene sequencing. The Venn diagram illustrating the distribution of ASVs (Amplicon Sequence Variants) revealed that there were 761 ASVs shared among all three groups, while the unique ASVs were 1245 for the RA + NS group, 863 for the RA + GNS group, and 1975 for the RA + GNS + CH group (Fig. [95]3A). This result indicated a significant alteration in the gut microbiota composition across the different treatment groups. Subsequently, we performed alpha diversity analyses. The flat region observed on the rarefaction curves indicated that the sampling efforts have effectively captured the majority of the microbial diversity, supporting the thoroughness of the analysis (Supplementary Fig. [96]2A). In specific, GNS decreased the Shannon and Chao1 indexes, a change that was effectively reversed by the AhR antagonist CH-223191. This reversal demonstrated that CH-223191 can counteract the GNS-mediated reduction in alpha diversity, which reflected community diversity through species richness and species evenness in CIA mice (Fig. [97]3B–C and Supplementary Fig. [98]2B). Jaccard and Bray curtis-based principal coordinates analysis (PCoA) revealed a distinct clustering of microbiota composition between the three groups (Fig. [99]3D and E). PCoA based on the Unweighted UniFrac and Weighted UniFrac analyses exhibited similar clustering (Supplementary Fig. [100]2C). We employed Phylogenetic Investigation of Communities by Reconstruction of unobserved States 2 (PICRUSt2) to obtain functional annotation results for the gut microbiota in the three groups using the KEGG Orthology (KO) database. Several metabolic pathways, including Tyrosine metabolism (involved in neurotransmitter production), Pyruvate metabolism (essential for energy production), Propanoate metabolism (critical for bacterial growth and energy supply), and Glutathione metabolism (providing antioxidant protection), were enhanced under GNS treatment but were reversed by CH-223191 (Fig. [101]3F). Additionally, the phenotypic functions of microbial communities measured by BugBase was changed significantly, further demonstrating that GNS reshaped the profile of gut microbiota (Supplementary Fig. [102]3A). Fig. 3. [103]Fig. 3 [104]Open in a new tab Alteration in gut microbiota composition and structure mediated by GNS. A Venn diagram showing ASVs distribution among the three groups. B Shannon index in gut microbiota across the three groups. C Chao1 index reflecting species richness in the gut microbiota. D Principal coordinates analysis (PCoA) plot based on Jaccard distance. E PCoA plot based on Bray–Curtis distance F PICRUSt2 functional annotation using the KEGG Orthology database. G Bar graph about gut microbiota at phylum taxonomic level among RA + NS, RA + GNS and RA + GNS + CH groups. H Heatmap depicting different gut microbiota at the phylum taxonomic level. The color scheme within the heatmap was employed to signify specific overall abundance, with blue indicating lower abundance and red indicating higher abundance. I Circos plot illustrating the relative abundance of bacterial phyla. The various colored ribbons standed for specific phyla, and the width of the ribbon was directly proportional to the abundance of the phylum. The ribbons linked bacterial taxa to their corresponding sample. J The ratio of F/B. K The relative abundance of Verrucomicrobiota among RA + NS, RA + GNS, and RA + GNS + CH groups As is shown in Fig. [105]3G–I, the microbial composition at the phylum level revealed the dominance of five principal bacteria, comprising Firmicutes, Bacteroidota, Campylobacterota, Desulfobacterota, and Actinobacteriota. The F/B ratio, which measures the proportion of Firmicutes to Bacteroidota in the gut microbiota, plays a pivotal role in maintaining gut homeostasis and optimizing the structure and function of the intestinal microbiome. An imbalance, characterized by an overgrowth of Bacteroidota or a depletion of Firmicutes, can disrupt this equilibrium, fostering a pro-inflammatory environment that exacerbates chronic conditions such as inflammatory bowel disease and breast cancer [[106]29]. Previous study has proved that the ratio of F/B was decreased in RA patients and correlated with the severity of RA [[107]30]. Although the precise functional implications of F/B ratio changes in RA remain to be fully elucidated, the observed increase in F/B ratio in RA + GNS mice (compared to RA + NS)—characterized by enrichment of beneficial Firmicutes species—aligned with improved clinical outcomes. This suggested that F/B ratio elevation may serve as a potential indicator of disease amelioration in RA. CH-223191-mediated reversal of this ratio was associated with aggravated inflammation (Fig. [108]3J), further supporting the potential beneficial role of F/B ratio modulation in RA management. Notably, Verrucomicrobiota was significantly enriched in the GNS administration group and decreased with CH-223191 intervention (Fig. [109]3K). Subsequently, we observed the specific alteration of key bacterial strains among the three groups at the class, order and family levels. At the class level, the RA + GNS group showed a significant increase in Bacilli and Verrucomicrobiae, as well as a decrease in Erysipelotrichia (Supplementary Fig. [110]3B–F). At the order level, we found that the relative abundances of Verrucomicrobiales and Lactobacillales were higher, while Oscillospirales, Peptostreptococcales-Tissierellales and Erysipelotrichales were lower in the RA + GNS group (Supplementary Fig. [111]4A–G). At the family level, Lactobacillaceae, Akkermansiaceae, and Pasteurellaceae exhibited a significant upward trend in RA + GNS mice, and the level of Oscillospiraceae and Erysipelotrichaceae observably decreased after GNS administration (Supplementary Fig. [112]5A–C). Additionally, we compared the alterations in the relative abundance of key genera between the three groups and specifically identified the bacterial strains that are closely related to GNS-mediated protective effects on RA. As shown in Fig. [113]4A and B, Ligilactobacillus, Muribaculaceae_unclassified, Lactobacillus, Lachnospiraceae_NK4A136_group, and HT002 occupied the predominant positions in the microbial community, with Ligilactobacillus exhibiting the highest relative abundance among the three groups. The bubble plot demonstrated that Ligilactobacillus, Lactobacillus, Lachnospiraceae_NK4A136_group, and HT002 all belonged to the phylum Firmicutes, while Muribaculaceae_unclassified is classified under the phylum Bacteroidota (Fig. [114]4C). Further evolutionary branching tree of the gut microbiota revealed that Ligilactobacillus and Lactobacillus belong to Lactobacillaceae family, Lachnospiraceae_NK4A136_group and HT002 belong to Lachnospiraceae family (Fig. [115]4D). The bacterial genera with higher relative abundance (Ligilactobacillus > Muribaculum > Desulfovibrio > Akkermansia > Anaero tignum > Clostridium > Intestinimonas) were all significantly modulated by GNS and CH-223191 administration compared with the RA + NS group (Fig. [116]4E). We noticed that GNS treatment could enrich Ligilactobacillus (23.34% abundance in RA + NS, 37.70% in RA + GNS, 13.41% in RA + GNS + CH, P = 0.0221), which was greatly reduced in the RA + GNS + CH group, exhibiting the most striking change. These changes were accompanied by an increased abundance of Akkermansia and a decreased abundance of Muribaculum, Desulfovibrio, and Anaerotignum in the RA + GNS group. The Sankey plot further indicated the dominant position of Ligilactobacillus in the gut microbiota of three groups of mice, suggesting that its abundance increased in the RA + GNS group, while it decreased in the RA + NS and RA + GNS + CH groups (Fig. [117]4F). Fig. 4. [118]Fig. 4 [119]Open in a new tab Microbial composition and key genera modulated by GNS and CH-223191. A Bar graph about gut microbiota at genus taxonomic level among RA + NS, RA + GNS and RA + GNS + CH groups. B Heatmap depicting different gut microbiota at the genus taxonomic level. C Bubble plot showing key genera’ relative abundance. D The evolutionary branching tree of the gut microbiota. E Bar charts of gut microbiota at the genus taxonomic level with statistically significant differences. F Sankey diagram illustrating taxonomic changes at the genus (right) and phylum (middle) levels across RA + NS, RA + GNS, and RA + GNS + CH groups (left). Branch color and width depict the flow of specific genera within various phyla. G A phylogenetic tree with a cladogram generated by the LEfSe algorithm illustrates taxonomic associations among microbiome communities across RA + NS, RA + GNS, and RA + GNS + CH groups. H LEfSe was used to identify the most differentially abundant taxa among RA + NS, RA + GNS and RA + GNS + CH groups. I Correlation analyses between key genera and arthritis score. Colorings represented the median Spearman correlation coefficient To elucidate substantial differences in gut microbiota composition across the three groups and identify specific bacteria that played a crucial role in the anti-arthritic effects induced by GNS, we carried out linear discriminant analysis (LDA) Effect Size (LEfSe) and Cladogram (based on the highest relative abundance difference at each taxonomic level) analyses (Fig. [120]4G and H). LEfSe analysis (LDA cutoff = 3) identified an overrepresentation of taxa such as the orders of Clostridia_UCG_014 and Monoglobales (including the family of Monoglobaceae and the genus of Monoglobus), the genera of Desulfovibrio and Eubacterium__xylanophilum_group in RA + NS group, which may be associated with RA progression. Meanwhile, the RA + GNS group was characterized by an abundance of taxa such as the orders of Lactobacillales (including the family of Lactobacillaceae and the genus of Ligilactobacillus) and Verrucomicrobiales (including the family of Akkermansiaceae and the genus of Akkermansia), as well as the genus of Lachnospira, which may be linked to GNS-mediated protective effects on RA. Additionally, we noticed that the genera of Duncaniella, Muribaculum, Anaerotignum and Alloprevotella, the family of Tannerellaceae (including the genus of Parabacteroides) and Clostridiaceae (including the genus of Clostridium), the orders of Erysipelotrichales and Oscillospirales (including the family of Oscillospiraceae and the genus of Intestinimonas) were enriched in RA + GNS + CH, which may suggest an association with the reversal of GNS-mediated protective effects on RA by CH-223191. Among these specific bacterial taxa, Desulfovibrio, as an opportunistic pathobionts, has been proved to participate in the progression of multiple diseases such as IBD, liver cirrhosis, and autoimmune diseases such as systemic sclerosis and multiple sclerosis [[121]31]. The overgrowth of Desulfovibrio in CIA mice may exacerbate the autoimmune process by promoting local inflammation and altering the gut microbiota composition, leading to a more pro-inflammatory environment. Additionally, Ligilactobacillus and Akkermansia, as two promising probiotics, play an important role in balancing the gut microbiota, maintaining the integrity of the intestinal barrier and modulating inflammation [[122]32, [123]33]. Lachnospira, known for its ability to produce short-chain fatty acids (SCFAs), particularly butyrate, has protective effects against various diseases, particularly metabolic disorders, inflammatory conditions, and possibly even autoimmune disease such as systemic lupus erythematosus, Graves’ disease, and behcet disease [[124]34–[125]36]. Interestingly, these bacterial taxa including Ligilactobacillus, Akkermansia and Lachnospira, exhibit a positive correlation with tryptophan, indole, and their derivates in the gut environment, which further validated that GNS might alleviate the progression of RA by modulating the tryptophan metabolism of gut microbiota to activate the AhR [[126]37–[127]39]. Ultimately, we conducted correlation analyses between specific microbial components and the arthritis score. The genus of Ligilactobacillus was negatively associated with the severity of CIA, further demonstrating that enrichment of Ligilactobacillus may be involved in the protective effects of GNS against RA (F[128]ig. [129]4I). GNS markedly enriched tryptophan metabolites and was inverted by CH-223191 Given the significant alterations in tryptophan metabolism during the progression of RA, as well as the regulatory effect of GNS on tryptophan metabolism, and the reversal effects of the AhR antagonist CH-223191 on GNS, we conducted targeted quantitative tryptophan metabolome sequencing of the feces from three groups of mice including RA + NS, RA + GNS and RA + GNS + CH. The fecal metabolomes exhibited distinct differences among the three groups as demonstrated by principal components analysis. (PCA) (Fig. [130]5A). In targeted metabolomics analysis, we identified 27 tryptophan metabolites, with 11 of them showing significant changes in abundance following intervention with either GNS or CH-223191 (Fig. [131]5B). This indicated that these metabolites may be responsive to these treatments and could play a role in their effects on RA. Further analysis revealed a notable negative correlation between the arthritis scores and three metabolites: IAA, IPA, and nicotinic acid, which might be associated with reduced severity of arthritis (Fig. [132]5C). Interestingly, these metabolites were significantly enriched with GNS treatment but decreased with CH-223191 intervention, implying a possible therapeutic effect of GNS through the modulation of these specific metabolites (Fig. [133]5D–F). To identify which tryptophan metabolites mediated the potential protective effects of GNS on CIA, we analyzed the correlation between metabolites and micro-CT results to investigate their relationship with joint erosion. IPA and IAA were significantly positively correlated with joint BMD, indicating they might contribute to maintaining bone density and integrity in the context of RA, whereas nicotinic acid showed a weaker correlation (Fig. [134]5G–I). Additionally, we explored the interactions among differential metabolites and revealed a strong positive correlation between IPA and IAA levels, suggesting a coordinated regulation or shared pathway between these two metabolites during the treatments (Fig. [135]5J). These findings collectively highlighted the potential of targeting specific tryptophan metabolites for RA treatment strategies. Following this, we carried out a holistic microbiome-metabolome analysis to uncover the complex relationships between different microbial genera and a range of tryptophan metabolites (Fig. [136]5K). The abundance of Ligilactobacillus exhibited a conspicuously positive correlation with IAA and IPA, which was consistent with GNS-mediated alterations in gut microbiota. Collectively, GNS may alleviate the severity of CIA by altering the gut microbiota, especially increasing the genus of Ligilactobacillus, which may lead to the enrichment of tryptophan metabolites like IPA and IAA, collectively mitigating the progression of RA. Fig. 5. [137]Fig. 5 [138]Open in a new tab GNPs treatment altered the profile of tryptophan metabolism. A principal component analysis (PCA) of fecal metabolomes from RA + NS, RA + GNS, and RA + GNS + CH mice reveals distinct tryptophan metabolism profiles. B Heatmap of tryptophan metabolites. The blue indicated lower abundance, and red indicated higher abundance. C Correlation analysis between tryptophan metabolites and arthritis score. D–F The concentration of IAA, IPA and nicotinic acid from feces among RA + NS, RA + GNS, and RA + GNS + CH mice. G–I Correlation analyses examining the relationship between fecal concentrations of IAA, IPA, and nicotinic acid and the BMD within joint structures. J Heatmap of correlation analysis for group RA + GNS + CH vs. RA + GNS vs. RA + NS. The horizontal and vertical axes represented the differential metabolites of the groups being compared. The color blocks at different positions indicated the correlation coefficients between the corresponding metabolites at those positions. Red represented a positive correlation, blue represented a negative correlation, and the darker the color, the stronger the correlation. Additionally, significant correlations were marked with asterisks (*). K Correlation analysis between tryptophan metabolites and gut microbiota at genus taxonomic level IPA/IAA mixture suppressed PTEN ubiquitination and degradation to attenuate RA in MH7A cells RA is a chronic immune-mediated disease that primarily affects the synovial tissues. In the course of RA, the abnormal proliferation of fibroblast-like synoviocytes (FLSs) transforms the synovium into a hyperplastic, invasive tissue, producing pro-inflammatory cytokines and matrix-degrading enzymes, leading to the destruction of cartilage and bone [[139]40]. MH7A cells, a human fibroblast-like synoviocyte cell line, exhibit characteristics similar to synovial cells from RA patients. TNF-α-treated MH7A cells are commonly used to simulate the inflammatory state of FLSs (serving as the CT group in subsequent experiments). CCK8 test revealed that IPA/IAA mixture treatment significantly inhibited the proliferation of MH7A cells in a concentration-dependent manner (Fig. [140]6A). We chose IPA/IAA mixture at a concentration of 10 µM for the following experiments. Flow cytometry analysis demonstrated that the IPA/IAA mixture treatment led to an increase in the percentage of apoptotic cells compared to the CT group, with the majority of the increased apoptosis occurring in the late stage (Fig. [141]6B and C). EdU staining indicated a significant decrease in proliferation in the IPA + IAA group compared to the CT group. Quantitative analysis showed that the percentage of EdU-positive cells was lower in the IPA + IAA group, suggesting that IPA/IAA treatment inhibited MH7A cell proliferation (Fig. [142]6D and E). Furthermore, the transwell assay demonstrated that the IPA/IAA mixture markedly inhibited the migration and invasion of MH7A cells (Fig. [143]6F-6H). Fig. 6. [144]Fig. 6 [145]Open in a new tab IPA/IAA mixture inhibited proliferation, migration/invasion and promotes apoptosis of MH7A cells. A The proliferation of MH7A cells was assessed using the CCK8 assay. IPA/IAA mixture treatment significantly inhibited the proliferation of MH7A cells in a concentration-dependent manner (0, 1, 5, 10 μM). B Flow cytometry analysis of apoptosis. MH7A cells were treated with IPA/IAA mixture (10 µM) and analyzed for apoptosis by flow cytometry. C Quantification of apoptosis. D Representative images of EdU staining for the proliferation of MH7A cells. E Quantification of EdU-positive cells. F Representative images of transwell assay for the migration and invasion of MH7A cells. G and H Relative ratio of cell migration and invasion To elucidate the detailed mechanisms by which IPA/IAA inhibited the proliferation, migration, and invasion of MH7A cells, we employed RNA sequencing to investigate the potential genes regulated by the IPA/IAA mixture. The PCA analysis revealed distinct clustering between IPA/IAA-treated and control groups, indicating a clear separation in gene expression profiles (Fig. [146]7A). Compared to the CT group, the IPA + IAA group showed 8209 differentially expressed genes (|fold change|> 1.35, p < 0.05), with 5046 genes upregulated and 3163 genes downregulated (Fig. [147]7B and C). The GO enrichment analyses demonstrated that IPA/IAA treatment markedly many biological processes including cell cycle, apoptotic process, cell migration, and cell adhesion (Supplementary Fig. [148]6A). Consistent with previous results obtained from IPA/IAA mixture treatment of MH7A cells, IPA/IAA inhibited the expression of genes associated with MH7A cell migration and adhesion (Fig. [149]7D and E). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis determined that the PI3K-AKT signaling pathway constituted one of the most prominently enriched pathways, which was downregulated in the IPA + IAA group (Fig. [150]7F, G and (Supplementary Fig. [151]6B). The PI3K/AKT signaling pathway is a critical regulator of cell survival, growth, and metabolism, and its dysregulation has been implicated in various inflammatory and autoimmune diseases. In the context of RA, the activation of PI3K/AKT signaling promotes the survival and proliferation of synovial fibroblasts, leading to synovial hyperplasia and joint inflammation. Additionally, this pathway mediates the production of inflammatory cytokines (including IL-1, IL-6, IL-17, IL-21, and IL-22) and proteases, which contribute to cartilage and bone destruction, hallmark features of RA [[152]41]. Within the PI3K/AKT signaling cascade, PTEN (phosphatase and tensin homolog), as a key negative regulator, serves as a crucial brake on synovial fibroblast activation and inflammation, thereby influencing the course of RA [[153]42]. Notably, RNA-seq revealed that PTEN was significantly upregulated in the IPA/IAA mixture-treated group compared to the control group, which is consistent with the observed inhibition of MH7A cell proliferation, migration, and invasion, suggesting that PTEN may play a pivotal role in mediating the anti-RA effects of IPA/IAA (Fig. [154]7H and I). Fig. 7. [155]Fig. 7 [156]Open in a new tab IPA/IAA mixture significantly inhibited NF-κB signaling pathway via the upregulation of PTEN. A PCA analysis showed distinct clustering between IPA/IAA-treated and control groups. B Heatmap of Differentially expressed genes (DEGs) in the IPA + IAA group compared to the control group. Different colors represent different levels of gene relative expression, ranging from blue through white to red, indicating expression levels that range from low to high. Red indicates highly expressed genes, and blue shows lowly expressed genes. C Bar plot of DEGs between IPA + IAA and CT groups (|fold change|> 1.35, p < 0.05). D and E Positive regulation of cell migration (enrichment score =  − 0.48) and focal adhesion (enrichment score =  − 0.47) pathways were down-regulated in the IPA/IAA group compared to the CT group as validated by GSEA. F KEGG enrichment bar plot. The y-axis represents the names of the top 20 KEGG pathways with the smallest p-values, and the x-axis represents the -log10 value of the p-value from the KEGG pathway enrichment analysis. G PI3K/AKT pathway (enrichment score =  − 0.38) was down-regulated in IPA/IAA as validated by GSEA. H Volcano plot of DEGs in the IPA + IAA group compared to the CT group. PTEN was significantly upregulated in the IPA + IAA group. I The relative expression of PTEN between CT and IPA + IAA groups. J Relative expression of p-p65, p65, p-IκBα, IκBα, p-PTEN, and PTEN in the presence or absence of IPA/IAA mixture during the stimulation for 0–120 min at the protein level. β-actin was used as an internal control. K–P Quantitative analyses of p-p65, p65, p-IκBα, IκBα, p-PTEN, and PTEN expression. Q Representative images of the intracellular location of the NF-κB p65. R MH7A cells were cultured with or without the IPA/IAA mixture for 12 h and treated with cycloheximide (CHX) (50 μg/mL). Total cell lysates were subjected to immunoblotting analysis. S Quantitative analyses of PTEN expression treated with CHX in the presence or absence of IPA/IAA mixture (0, 4, 8, 12 h). T MH7A cells were pretreated with IPA/IAA mixture for 8 h, with or without MG132 (50 μM) for 2 h. The cell lysates were immunoprecipitated with PTEN antibody and immunoblotted with ubiquitin and PTEN antibodies. U–V Relative expression levels of ubiquitin in PTEN Furthermore, the NF-κB signaling pathway is pivotal in the progression of RA, as it regulates a wide array of inflammatory processes and immune responses. Activation of this pathway amplifies the release of pro-inflammatory cytokines and promotes tissue damage in RA synovial tissues [[157]43]. PTEN, known for its role in antagonizing the PI3K/AKT pathway, also exerts an inhibitory effect on NF-κB signaling [[158]44]. To explore the interaction between PTEN and NF-κB signaling in response to IPA/IAA mixture treatment, we assessed the levels of PTEN and phosphorylated PTEN (p-PTEN), as well as NF-κB pathway components, including p65, phosphorylated p65 (p-p65), NF-κB inhibitor IκB-α, and phosphorylated IκBα (p-IκBα) (Fig. [159]7J). As expected, the IPA/IAA mixture treatment significantly elevated the expression of PTEN, indicating its potential role in mediating the observed anti-RA effects. Interestingly, the treatment did not alter the levels of p-PTEN, suggesting that the modulation of PTEN expression by IPA/IAA was not achieved through changes in its phosphorylation status (Fig. [160]7O and P). This implied that the mechanisms underlying the effects of IPA/IAA on PTEN may involve other pathways or interactions that influence its protein stability, activity, and overall function without directly affecting its phosphorylation. Similarly, IPA/IAA mixture effectively suppressed the phosphorylation of p65 at 120 min and significantly inhibited the induced phosphorylation of IκBα at 60 min. These results suggested that IPA/IAA mixture exerts its anti-inflammatory effects, at least in part, by dampening the NF-κB signaling pathway (Fig. [161]7K–N). The nuclear translocation of p65 is a critical step in the activation of the NF-κB signaling pathway, as it facilitates the binding of p65 to target DNA sequences, thereby initiating the transcription of pro-inflammatory genes [[162]45]. To investigate the impact of IPA/IAA mixture on the nuclear localization of p65, we performed immunofluorescence assays, which revealed that treatment with IPA/IAA mixture significantly inhibited the nuclear translocation of p65. This observation aligned with the suppression of phosphorylated p65 and IκBα levels, indicating a coherent mechanism by which the IPA/IAA mixture exerts its anti-inflammatory effects by disrupting the NF-κB signaling cascade. Interestingly, we also observed that the addition of VO-Ohpic, a specific inhibitor of the PTEN, partially reversed the inhibitory effect of the IPA/IAA mixture on p65 nuclear translocation, which showed that the mixture inhibited NF-κB signaling components in a PTEN-dependent manner (Fig. [163]7Q). Subsequently, we investigated the specific mechanisms by which IPA/IAA mixture modulated the expression of PTEN in MH7A cells. When treated with 50 μg/mL of the protein synthesis inhibitor cycloheximide (CHX), PTEN protein levels exhibited a slight reduction at 4 h and a marked decrease at 8 h. In contrast, treatment with the IPA/IAA mixture prevented PTEN degradation and sustained PTEN protein levels, which indicated that the IPA/IAA mixture likely functioned to stabilize PTEN protein, thereby extending its presence and activity within the cells (Fig. [164]7R and S). PTEN is known to be regulated by various post-translational modifications, with ubiquitination being particularly crucial [[165]46]. Ubiquitination typically marks proteins for proteasomal degradation, and since the IPA/IAA mixture appeared to prevent PTEN decline, we hypothesized that PTEN might be regulated through ubiquitination. Immunoprecipitation assays demonstrated that IPA/IAA treatment resulted in a decrease in PTEN ubiquitination, which could account for the observed stabilization of PTEN protein levels upon IPA/IAA treatment (Fig. [166]7T–V). Collectively, these findings indicated that the anti-inflammatory effects of the IPA/IAA mixture in MH7A cells are, at least partially, mediated through the modulation of the PTEN/NF-κB signaling pathway. The mixture enhanced PTEN expression by reducing PTEN ubiquitination, which suppressed NF-κB signaling and in turn diminished the inflammatory response characteristic of MH7A cells. Microbial typtophan metabolites exhibited anti-CIA effects in mice To further examine the anti-CIA effects of IPA/IAA mixture in vitro, we randomly divided the CIA mice into two groups: the IPA + IAA group (500 mg/kg bw/day) and the control group (equal volume of saline). The results indicated that the IPA/IAA mixture significantly alleviated the severity of RA, reduced hind paw swelling, and decreased arthritis scores in the CIA mice (Fig. [167]8A). Histological examination using H&E and TRAP staining revealed a decrease in synovial hyperplasia and inflammatory cell infiltration, along with a marked reduction in bone erosion and resorption areas (Fig. [168]8B, C, G and H). Micro-CT analysis demonstrated that administration of the IPA/IAA mixture significantly improved joint bone destruction and slowed the progression of RA (Fig. [169]8D and I). Additionally, we assessed the expression of PTEN/NF-κB signaling pathway in the FLSs. The IHC results demonstrated that IPA/IAA treatment observably decreased the expression of p-p65 and increased PTEN expression, which were consistent with the in vitro experiments (Fig. [170]8E, F, J and K). In summary, these in vivo findings substantiated the anti-inflammatory and protective effects of the IPA/IAA mixture in RA. Fig. 8. [171]Fig. 8 [172]Open in a new tab The IPA/IAA mixture exerted significant anti-inflammatory and protective effects in CIA mice. A Arthritis severity scores in CIA mice treated with the IPA/IAA mixture or normal saline. B Representative histological images of joint sections stained with H&E. C Representative TRAP-stained images of joint sections in CIA mice. D Representative photographs and micro-CT images of paws. E and F Representative immunohistochemical images showing expression of p-p65 and PTEN in fibroblast-like synoviocytes. G Histological scores of joints. H Quantitative analyses of TRAP staining. I Quantitative analyses of bone histomorphometry indexes. J and K The Average Optical Density (AOD) of p-p65 and PTEN Discussion Dysregulation of the gut microbiota can disrupt the intestinal barrier function, leading to increased bacterial translocation and activation of immune cells, which in turn triggers system inflammation and contributes to the development of RA. Building on our group’s prior discovery that GNPs mitigated osteoarthritis progression via the “microbiota-gut-joint” axis by restoring microbial diversity and elevating anti-inflammatory SCFAs, this study explored the potential of GNS as a nanotherapeutic agent targeting RA by modulating gut microbiota and tryptophan metabolism [[173]26]. Notably, the gut microbiota remodeling effects observed here aligned mechanistically with our earlier findings in osteoarthritis models, but further extended the scope to RA-specific pathways. Specifically, GNS reversed the dysbiosis of gut microbiota, increased the levels of multiple probiotic strains including Ligilactobacillus, Akkermansia and Lachnospira, enriched tryptophan-related metabolites such as IPA and IAA, and alleviated synovial hyperplasia and inflammatory infiltration by inducing PTEN activity, thus inhibiting the NF-κB pathway (Fig. [174]9). Fig. 9. [175]Fig. 9 [176]Open in a new tab Schematic for GNS-mediated protective effects on RA progression. GNS treatment reshaped the gut microbiota profile, increased IPA and IAA production, improved synovial inflammation and restored gut barrier function First, GNS reshaped the composition and structure of the microbial community to favor beneficial taxa. In specific, at the phylum level, GNS mediated the increase in the F/B ratio, which may be crucial for its anti-inflammatory effects. The enrichment of Verrucomicrobia, particularly Akkermansia, demonstrated its role as a beneficial microbe that supports gut barrier integrity and immune modulation, contributing to the overall therapeutic effect of GNS. At the genus level, the marked increase in Ligilactobacillus in the RA + GNS group, coupled with its strong negative correlation with arthritis severity, suggested it may play a protective role in RA management. This genus, along with other enriched taxa such as Lachnospira, is known for producing SCFA like butyrate, which may modulate immune responses by influencing gut barrier function and systemic inflammation levels. Conversely, the increased abundance of potentially pathogenic taxa such as Desulfovibrio in the RA + NS group demonstrated a detrimental role in exacerbating inflammatory conditions. Desulfovibrio, known to promote inflammatory responses, might contribute to the autoimmune processes seen in RA, further suggesting the importance of maintaining a balanced microbial community. Notably, the AhR antagonist CH-223191 was found to reverse the protective effects of GNS on RA. This reversal appeared to stem from its ability to block the activation of the AhR by tryptophan metabolites such as IPA and IAA produced by the altered gut microbiota under the influence of GNS. These metabolites are crucial for mediating the anti-inflammatory effects of GNS. Additionally, CH-223191 also reversed the changes induced by GNS in the gut microbiota, including adjustments to the F/B ratio, decreased alpha diversity, and the proliferation of beneficial bacteria along with a decrease in pathogenic ones. These observations suggest that CH-223191 not only acted as an AhR pathway inhibitor but also significantly affected microbial composition and community structure, displaying its potential as a gut microbiome modulator. The ability of CH-223191 to impact both microbial and host pathways presented a unique opportunity for further investigation. Second, by conducting targeted tryptophan metabolomics, we identified notable changes in metabolite abundance influenced by both GNS and the AhR antagonist CH-223191. Among these metabolites, IPA and IAA, in conjunction with nicotinic acid, demonstrated a negative correlation with arthritis scores, suggesting their potential role in mitigating RA severity. This was further substantiated by the positive correlation between these metabolites and bone mineral density (BMD), indicating a protective effect on joint integrity. Importantly, the increase in Ligilactobacillus abundance corresponded with elevated levels of IPA and IAA, signifying a microbial influence on metabolite-mediated anti-inflammatory effects. Furthermore, IPA/IAA mixture suppressed PTEN ubiquitination and degradation, thereby inhibiting the NF-κB signaling pathways, indicating that IPA/IAA can modulate key molecular pathways involved in inflammation. In vivo experiments corroborated these findings, demonstrating significant reductions in RA symptoms following IPA/IAA treatment in CIA mice. Collectively, these results elucidated a novel mechanism through which microbiota-derived tryptophan metabolites can exert protective effects in RA. Third, there are several limitations in our study that need to be addressed in future research: (i) Our study was conducted on a limited number of mice, which may not fully represent the broader population. Additionally, results obtained from animal models may not directly translate to human clinical outcomes. (ii) Although we have elucidated the role of IPA/IAA in GNS-mediated anti-inflammatory effects, several key questions remain unresolved. The precise mechanisms by which GNS enhances microbial production of IPA/IAA require further investigation, particularly regarding whether this involves direct modulation of bacterial enzymatic pathways or indirect shifts in microbial community structure. Notably, IPA and IAA may serve as key signaling molecules within the microbiome-host crosstalk, potentially influencing bacterial quorum sensing or interspecies metabolic cooperation to sustain a balanced gut ecosystem. The specific relationship between these metabolites and microbiome-mediated anti-inflammatory effects warrants deeper exploration, including their potential synergetic or antagonistic interactions with other microbial metabolites. Future studies should focus on personalized microbiome responses to GNS intervention, employ single-bacterium experiments to clarify causal metabolite-microbiome correlations, and validate these mechanisms in clinically relevant models. (iii) The concentrations of metabolites used in both in vitro and in vivo experiments may not reflect physiologically relevant levels found in humans, and dose–response relationships were not comprehensively explored. (iv) The absence of a healthy control group limits the ability to establish baseline comparisons and may confound the interpretation of treatment effects. A healthy control group would have provided a reference for normal biological processes, enabling a more accurate assessment of whether observed effects are due to the treatment or the disease state itself. Therefore, future studies should include a healthy control group to enhance the validity and generalizability of the findings. (v) Although our antibiotic treatment protocol was carefully optimized to minimize systemic effects, this broad-spectrum antibiotic may still influence host physiology beyond their intended effects on gut microbiota. Future investigations employing germ-free animals or more targeted microbiota manipulation approaches would provide valuable complementary data to our current findings. In conclusion, our findings on GNS-mediated Ligilactobacillus enrichment and intestinal barrier restoration are now contextualized with prior reports demonstrating the beneficial effects of probiotics (e.g., Lactobacillus strains) in CIA models [[177]14]. Importantly, our study advanced this field by establishing a novel link between GNS-induced microbial metabolites (IPA/IAA) and the PTEN/NF-κB pathway in synovial tissues—a mechanistic insight not yet thoroughly explored. Furthermore, compared with other nanoformulations (e.g., silica or polymeric nanoparticles) used for microbiota modulation, we proved that GNS possessed unique, such as oral bioavailability and precise gut-joint axis targeting [[178]25–[179]27]. Additionally, we discussed how our results on AhR activation by IPA/IAA complement and extended previous findings on microbial tryptophan metabolites in RA pathogenesis. Future research should aim to identify specific microbial metabolites and their interactions with host pathways to fully elucidate the mechanistic underpinnings of GNS-mediated protective effects and explore the potential for targeted microbiota-based therapies in RA. Supplementary Information [180]12951_2025_3450_MOESM1_ESM.tif^ (6.1MB, tif) Additional file 1: Fig. 1 GNS exhibits good biocompatibility without liver and kidney toxicity. (A) Representative images of liver with H&E staining. (B) Representative images of kidney with H&E staining. (C) Lipopolysaccharide (LPS) level in mouse serum. (D)Intestinal permeability measurement by Fluorescein isothiocyanate-dextran 4 (FD4) [181]12951_2025_3450_MOESM2_ESM.tif^ (2.2MB, tif) Additional file 2: Fig. 2 Alpha and beta diversity analyses of gut microbiota. (A) Rarefaction curves derived from alpha diversity. (B) Boxplots of related alpha diversity consisting of Simpson and Pielou_e. (C and D) PCoA based on the Unweighted UniFrac and Weighted UniFrac analyses [182]12951_2025_3450_MOESM3_ESM.tif^ (2.1MB, tif) Additional file 3: Fig. 3 The dramatic alterations of gut microbiota at the class level. (A) Bar plot about the abundance of microbiota with statistically significant differences. (B) Bar graphs of the bacteria at the class taxonomic level. (C) Heatmap depicting different gut microbiota at the class taxonomic level. (D-F) The relative abundance of Bacilli, Verrucomicrobiae and Erysipelotrichia among RA+NS, RA+GNS, and RA+GNS+CH groups [183]12951_2025_3450_MOESM4_ESM.tif^ (2MB, tif) Additional file 4: Fig. 4 The alterations of gut microbiota at the order level. (A) Bar graphs of the bacteria at order taxonomic level. (B) Heatmap depicting different gut microbiota at the order taxonomic level. (C-G) The relative abundance of Oscillospirales, Verrucomicrobiales, Peptostreptococcales-Tissierellales, Lactobacillales, and Erysipelotrichales among RA+NS, RA+GNS, and RA+GNS+CH groups [184]12951_2025_3450_MOESM5_ESM.tif^ (1.7MB, tif) Additional file 5: Fig. 5 The alterations of gut microbiota at the family level. (A) Bar graphs of the bacteria at the family taxonomic level. (B) Heatmap depicting different gut microbiota at the family taxonomic level. (C) Bar charts of gut microbiota at the family taxonomic level with statistically significant differences [185]12951_2025_3450_MOESM6_ESM.tif^ (3.9MB, tif) Additional file 6: Fig. 6 Functional and pathway annotation of differentially expressed genes. (A) GO enrichment bar plot. Using the top 20 Biological Process terms with the smallest Q-values for plotting, the vertical axis represented the GO terms, and the horizontal axis represented the -log10 value of the Q-value for the enrichment analysis of each GO term. (B) The KEGG enrichment bubble plot displayed the top 20 pathways with the smallest P-values Acknowledgements