Gut microbiota-induced perturbation in bile acids alter keratinocyte lipid metabolism via FXR-NQO1 signaling in psoriasis Authors: Panpan Lian, Renwei Lu, Chaode Gu, Junaid Wazir, Wentong Wang, Yangyongyi Zong, Yun Xia, Jie Ni, Huimei Chen, Enrico Petretto, Eckardt Treuter, Rongrong Fan, Airong Yang, Zhonglan Su, Hongwei Wang, Zhiqiang Huang Published: 2026 ABSTRACT Emerging evidence has shown that microbiota dysbiosis and aberrant bile acid (BA) profiles are correlated with disease progression. This study elucidates dysregulated BA metabolism in psoriasis patients and imiquimod-treated female mice, coupled with increased expression of the farnesoid X receptor (FXR) in keratinocytes. Activation of FXR by glycochenodeoxycholic acis (GCDCA) ameliorates psoriatic symptoms by enhancing lipid metabolism, particularly fatty acid oxidation. Mechanistically, the FXR-mediated enhancement of antioxidant capacity by upregulating NQO1 expression underlies its regulatory role in lipid metabolism, offering an insight into FXR’s role in oxidative stress and lipid metabolism. Conversely, keratinocyte-specific FXR ablation exacerbates psoriasis severity. Gut microbiota dysbiosis is further identified as a pivotal contributor to perturbations in BA profiles in psoriasis. These preclinical findings support a mechanistic model linking gut microbiota dysbiosis and BA alterations to FXR signaling in keratinocytes and psoriasis-associated metabolic dysregulation, suggesting therapeutic potential for microbiota-targeted interventions. Introduction Psoriasis is a multifactorial, chronic, and inflammatory skin disease with a global prevalence of 2–3%1. The clinical symptoms of psoriasis manifest as well-demarcated, erythematous, oval plaques with adherent silvery scales. Keratinocytes participate in both the initiation and maintenance phases of psoriasis. The pathological features are characterized by hyperproliferation and abnormal differentiation of keratinocytes, as well as infiltration of multiple inflammatory cells. This is accompanied by the increased expression of the cytoskeletal proteins keratin 16 (K16) and keratin 17 (K17)1,2. Our previous work revealed that sphingosine-1-phosphate receptor 3 (S1PR3)-driven STAT3 activation underlies pathological keratinocytes differentiation in psoriasis3. Recent studies have reported the gut microbiota dysbiosis in the feces of psoriatic patients. 16S rRNA sequencing and metagenomic analysis revealed that the α-diversity (number of bacterial species) of the gut microbiota in patients with psoriasis was decreased compared to healthy individuals. The β-diversity analysis showed distinct clustering between patients with psoriasis and healthy individuals, suggesting an altered composition of the gut microbiota in psoriasis4. Moreover, alleviation of psoriatic lesions after interleukin-17A (IL-17A) and interleukin-23 (IL-23) inhibitor treatment was accompanied by the restoration of gut microbiota homeostasis5,6, suggesting a close association between gut microbiota and psoriasis7. Bile acids (BAs) are synthesized from cholesterol in the liver, where they are converted into primary BAs, cholic acid (CA) and chenodeoxycholic acid (CDCA). These primary BAs are conjugated with glycine and taurine to increase their solubility in bile, and are then secreted into the gastrointestinal tract to facilitate the absorption of dietary nutrients8. Anaerobic gut bacteria in the colon produce bile salt hydrolase (BSH) enzymes to deconjugate conjugated BAs, and 7α- and 7β-dehydroxylases to convert primary BAs (CA, CDCA) into secondary BAs (DCA, LCA), respectively7. A recent study reported the gut microbiota dysbiosis and altered BAs composition in KRT14-Vegfa transgenic psoriatic mice9. Furthermore, treatment with secondary BAs, including LCA, DCA, and 3-oxo-lithocholic acid (3-oxoLCA) alleviated psoriasiform dermatitis in mice through inhibiting IL-17A expression and T cell trafficking10. However, the comprehensive BAs profiling and the identification of pivotal BAs involved in psoriasis development need further investigation. FXR (encoded by the NR1H4 gene) is a ligand-activated nuclear receptor that has also been identified as BAs receptor. The activation efficiency of FXR is dependent on the conjugation state and molecular structure of BAs. FXR is primarily expressed in the liver and intestine, where it plays a critical role in regulating the de novo synthesis of BAs and their enterohepatic circulation11. Accumulating evidence indicates that FXR is involved in various diseases, such as type 2 diabetes mellitus, obesity, nonalcoholic fatty liver disease (NAFLD), and colorectal cancer (CRC). FXR-null mice exhibit elevated serum triglyceride levels and reduced insulin tolerance, highlighting FXR’s central role in lipid and glucose homeostasis. Our results indicate that gut microbiota dysbiosis in psoriasis patients and imiquimod (IMQ)-induced psoriatic mice is associated with altered BA composition, together with impaired FXR activation, oxidative stress, and lipid accumulation in keratinocytes. Using transgenic mice with keratinocytes-specific FXR knockout, we found that FXR depletion in keratinocytes aggravates the IMQ-induced dermatitis. Conversely, GCDCA treatment attenuates the psoriatic phenotype and reduces lipid accumulation in skin, consistent with the FXR activation. Mechanistically, our data support a model in which FXR promotes lipid utilization in keratinocytes, at least in part, by upregulating NAD(P)H quinone dehydrogenase 1 (NQO1) and limiting reactive oxygen species (ROS) accumulation. Targeting the microbiota-FXR-NQO1 axis may address skin inflammation and systemic metabolic dysregulation, providing a potential therapeutic option for psoriasis. Results Disrupted serum bile acids profile in psoriasis patients and IMQ-induced psoriatic mice To investigate the profile of BAs, we performed ultra-performance liquid chromatography-mass spectrometry (LC-MS) to quantify the serum BAs metabolites in psoriasis patients and IMQ-induced mouse model (Fig. 1a). As shown in Fig. 1b, a total of 52 BAs were quantified in the serum of psoriasis patients (Fig. 1b). Analysis of 52 serum BAs revealed distinct compositional profile between psoriasis patients and healthy subjects (Fig. 1c). The orthogonal partial least squares discriminant analysis (OPLS-DA) plot revealed significant alterations in BAs composition between healthy subjects and psoriasis patients (Fig. 1d). Furthermore, differential analysis identified six BAs metabolites (four upregulated and two downregulated), as visualized on the volcano plot (Fig. 1e). This observation implied dysfunction in the gut microbiota-dependent BAs conversion in psoriasis, characterized by enhanced BAs transformation from primary to secondary BAs. We then examined the profile of serum BAs metabolites in IMQ-induced psoriatic mice and found significant alterations in BAs composition, as evidenced by distinct separation between control mice and IMQ-induced psoriatic mice in the OPLS-DA plot (Fig. 1f). A total of 26 BAs were quantified in the serum of IMQ-induced psoriatic mice, revealing a distinct metabolite profile in psoriatic mice compared to controls (Fig. 1g, h). Importantly, in IMQ-induced psoriatic mice, we found that the levels of primary BAs were significantly decreased, while the levels of secondary BAs were increased (Fig. 1i).Fig. 1Disrupted serum bile acids profile in psoriasis patients and IMQ-induced psoriatic mice.a Schematic diagram showing the collection of serum from psoriasis patients and psoriatic mice, followed by LC-MS analysis of BAs. The figure was created in BioRender. Zhao, C. (2026) https://BioRender.com/vutpv4i. b Heatmap of 52 BAs quantified by LC-MS analysis in the serum of psoriasis patients (n = 3, biological replicates). c The stacked histogram showing the compositional changes of 52 BAs in the serum of psoriasis patients (n = 3, biological replicates). d OPLS-DA for BAs profile between healthy subjects and psoriasis patients (n = 3, biological replicates). e Volcano-plot showing the significantly changed BAs in psoriasis. f OPLS-DA for BAs profile between control mice and IMQ-induced psoriatic mice (n = 5, biological replicates). g Heatmap of 23 serum BAs quantified by LC-MS in control and IMQ-induced psoriatic mice (n = 3, biological replicates). h The stacked histogram showing the relative aggregated BAs abundance in control and IMQ-induced psoriatic mice (n = 5, biological replicates). i Bar-chart showing the levels of primary and secondary BAs in conjugated and free forms in control and IMQ-induced psoriatic mice (n = 5, biological replicates). j Immunofluorescence (IF) staining of FXR in the lesions from psoriasis patients and healthy individuals (independently repeated three times). k IF staining of FXR in the lesions from IMQ-induced mice and control mice (independently repeated three times). l Representative macroscopic images (top) and H&E staining (bottom) of the dorsal skin in IMQ-induced and GW-treated mice. The data are presented as the mean ± SD. Statistical analyses are performed using two-tailed Student’s t test. All raw data are provided as a Source Data file. Disrupted serum bile acids profile in psoriasis patients and IMQ-induced psoriatic mice. a Schematic diagram showing the collection of serum from psoriasis patients and psoriatic mice, followed by LC-MS analysis of BAs. The figure was created in BioRender. Zhao, C. (2026) https://BioRender.com/vutpv4i. b Heatmap of 52 BAs quantified by LC-MS analysis in the serum of psoriasis patients (n = 3, biological replicates). c The stacked histogram showing the compositional changes of 52 BAs in the serum of psoriasis patients (n = 3, biological replicates). d OPLS-DA for BAs profile between healthy subjects and psoriasis patients (n = 3, biological replicates). e Volcano-plot showing the significantly changed BAs in psoriasis. f OPLS-DA for BAs profile between control mice and IMQ-induced psoriatic mice (n = 5, biological replicates). g Heatmap of 23 serum BAs quantified by LC-MS in control and IMQ-induced psoriatic mice (n = 3, biological replicates). h The stacked histogram showing the relative aggregated BAs abundance in control and IMQ-induced psoriatic mice (n = 5, biological replicates). i Bar-chart showing the levels of primary and secondary BAs in conjugated and free forms in control and IMQ-induced psoriatic mice (n = 5, biological replicates). j Immunofluorescence (IF) staining of FXR in the lesions from psoriasis patients and healthy individuals (independently repeated three times). k IF staining of FXR in the lesions from IMQ-induced mice and control mice (independently repeated three times). l Representative macroscopic images (top) and H&E staining (bottom) of the dorsal skin in IMQ-induced and GW-treated mice. The data are presented as the mean ± SD. Statistical analyses are performed using two-tailed Student’s t test. All raw data are provided as a Source Data file. We then investigated the expression of the BAs receptor in both clinical psoriatic plaques and animal lesions. As shown in Fig. 1j, FXR was increased in the skin from psoriasis patients (Fig. 1j) and IMQ-induced psoriatic mice (Fig. 1k, Supplementary Fig. 1g). RNA sequencing (RNA-seq) and single-cell RNA sequencing (scRNA-seq) analysis revealed that FXR was upregulated and primarily expressed in keratinocytes, particularly in the spinous and granular layers of the epidermis in psoriasis (Supplementary Fig. 1a, b). Consistent with human data, qRT-PCR analysis revealed compensatory upregulation of Nr1h4 mRNA in the lesional skin of IMQ-induced psoriatic mice, suggesting the pivotal role of FXR signaling in psoriasis pathogenesis (Supplementary Fig. 1c). Additionally, we examined the expression of Takeda G protein–coupled receptor 5 (TGR5), another key BAs-sensing receptor12. In contrast to FXR, TGR5 expression was downregulated in both clinical patients and IMQ-induced psoriatic mice (Supplementary Fig. 1d–f)13. These findings suggest a potential role of keratinocytes-intrinsic FXR, and its dysregulation may contribute to psoriasis development. To investigate the functional role of FXR in psoriasis, we administered GW4064 (hereafter referred to as GW), an FXR agonist, to the IMQ-induced mouse model of psoriasis. Our results indicated that FXR activation ameliorated IMQ-induced psoriatic hyperplasia and reduced the psoriasis area and severity index (PASI) scores, including erythema, scaling, and infiltration (Fig. 1l, Supplementary Fig. 1h). Histopathological analysis revealed that GW treatment inhibited keratinocytes hyperproliferation and aberrant differentiation, as indicated by a reduction of Ki67 and K17 expression after GW-mediated FXR activation (Supplementary Fig. 1i, j). In summary, these findings indicate that serum BAs metabolites are disrupted in both psoriasis patients and mouse model, accompanied by dysregulated expression of FXR within psoriasis plaques. GW-mediated FXR activation alleviated the psoriatic phenotypes in mice, further supporting its functional involvement of BAs and FXR in psoriasis pathogenesis. Keratinocytes Nr1h4 knockout exacerbates psoriatic lesions in IMQ-induced psoriatic mice To precisely explore the pathological role of FXR in psoriasis, Nr1h4 knockout (KO) mice were subject to IMQ application to induce a psoriatic model (Supplementary Fig. 2a). The results showed that deletion of Nr1h4 exacerbated psoriasis, as evidenced by increased lesion severity and PASI scores (Supplementary Fig. 2b, c). Additionally, Nr1h4 KO mice exhibited a significant increase in spleen weight compared to IMQ-induced WT mice (Supplementary Fig. 2d), suggesting that the Nr1h4 deficiency-mediated exacerbation was involved in the systemic inflammatory response. Histopathological analysis revealed that Nr1h4 KO markedly promoted keratinocytes hyperproliferation as indicated by increased expression of K17 and P-STAT3 (Supplementary Fig. 2e, f). In addition, psoriasis-related inflammatory genes, such as Tnf, Il6, and Cxcl10, were significantly upregulated in IMQ-induced Nr1h4 KO mice (Supplementary Fig. 2g). We hypothesize that epidermal FXR plays a pivotal role in psoriasis pathogenesis. To investigate this, Nr1h4 was conditionally ablated in keratinocytes using a Cre-LoxP system (Fig. 2a, Supplementary Fig. 2h). Immunofluorescence (IF) staining confirmed the successful KO of FXR in keratinocytes (Fig. 2b). In the IMQ-induced psoriatic model, the keratinocytes FXR deficiency significantly exacerbated epidermal hyperplasia (Fig. 2c). Furthermore, the skin lesions were worsened by keratinocytes-specific Nr1h4 knockout, as evidenced by increased erythema, scaling, epidermal thickening, and PASI scores in IMQ-induced Nr1h4 knockout mice (Fig. 2c, d). Notably, the keratinocytes-specific deletion of FXR did not affect spleen weight, indicating that keratinocytes FXR is involved in psoriasis pathogenesis independent of systemic inflammation (Fig. 2e). Further histopathological analysis revealed that keratinocytes FXR KO led to epidermis hyperproliferation, as evidenced by increased expression of the hyperproliferation markers K17 and Ki67 after keratinocytes Nr1h4 knockout (Fig. 2f). Additionally, IF staining of CD3 and F4/80 revealed that keratinocytes-specific Nr1h4 KO increased lesional infiltration of T lymphocytes and macrophages into lesions (Fig. 2g). In summary, these results indicate that keratinocytes FXR plays a critical role in the pathogenesis of psoriasis, highlighting the extrahepatic and extraintestinal functions of FXR.Fig. 2Keratinocytes Nr1h4 knockout exacerbates psoriatic lesions in IMQ-induced psoriatic mice.a Schematic diagram of the generation of keratinocytes-specific Nr1h4 knockout (Nr1h4-cKO) mice. The figure was created in BioRender. Zhao, C. (2026) https://BioRender.com/2t9z9gf. b IF staining of FXR showing the knockout efficiency in Nr1h4-cKO mice following IMQ treatment (independently repeated three times). c Representative images of dorsal skin from WT and Nr1h4-cKO mice treated with vaseline or IMQ. d PASI scores assessing erythema, scaling and infiltration severity in dorsal lesions on the day of sacrifice (n = 5, biological replicates). e Macroscopic images of spleens and spleen indices (spleen weight/body weight ratios × 100%) in control and Nr1h4-cKO mice with or without IMQ treatment. Statistical quantification results are shown below (n = 5, biological replicates). f IF staining of Ki67 and K17 expression in dorsal lesions of mice (independently repeated three times). g IF staining of CD3 and F4/80 in dorsal lesions of mice (independently repeated three times). The data are presented as the mean ± SD. Statistical analyses are performed using two-tailed one-way ANOVA followed by the Dunnett multiple comparison. All raw data are provided as a Source Data file. Keratinocytes Nr1h4 knockout exacerbates psoriatic lesions in IMQ-induced psoriatic mice. a Schematic diagram of the generation of keratinocytes-specific Nr1h4 knockout (Nr1h4-cKO) mice. The figure was created in BioRender. Zhao, C. (2026) https://BioRender.com/2t9z9gf. b IF staining of FXR showing the knockout efficiency in Nr1h4-cKO mice following IMQ treatment (independently repeated three times). c Representative images of dorsal skin from WT and Nr1h4-cKO mice treated with vaseline or IMQ. d PASI scores assessing erythema, scaling and infiltration severity in dorsal lesions on the day of sacrifice (n = 5, biological replicates). e Macroscopic images of spleens and spleen indices (spleen weight/body weight ratios × 100%) in control and Nr1h4-cKO mice with or without IMQ treatment. Statistical quantification results are shown below (n = 5, biological replicates). f IF staining of Ki67 and K17 expression in dorsal lesions of mice (independently repeated three times). g IF staining of CD3 and F4/80 in dorsal lesions of mice (independently repeated three times). The data are presented as the mean ± SD. Statistical analyses are performed using two-tailed one-way ANOVA followed by the Dunnett multiple comparison. All raw data are provided as a Source Data file. The GCDCA-mediated FXR activation alleviates the psoriatic lesions in IMQ-induced psoriatic mice Based on the observations of altered serum BAs metabolites and FXR expression in both clinical and animal model (Fig. 1), we asked whether dysregulated BAs-mediated cutaneous FXR signaling is implicated in the pathogenesis of psoriasis. The BAs metabolites exhibiting consistent downregulation in both psoriasis patients and IMQ-induced psoriatic mouse model were investigated (Supplementary Fig. 3a). Among these downregulated BAs, GCDCA (hereafter referred to as GC) was identified as an endogenous FXR agonist capable of activating the luciferase reporter of NR0B2 (alias SHP), a canonical FXR downstream target gene14 (Fig. 3a, Supplementary Fig. 3b). Molecular docking analysis of GC and GW with FXR further revealed that GC could bind to FXR with a comparable affinity to GW (Fig. 3b, c). Given GC levels were reduced in the serum of both mice model and patients (Supplementary Fig. 3c), we subsequently investigated the levels of GC and the activation of FXR at lesional sites. The reduced expression of PPARG and FGF19, downstream genes of FXR, in lesional skin15,16, reflected the impaired activation of FXR signaling in psoriasis (Supplementary Fig. 3d). Critically, we quantified the lesional concentration of GC in IMQ-induced mice. Our data revealed significantly reduced GC concentrations in psoriatic skin compared to normal skin (Supplementary Fig. 3e), suggesting that impaired GC-mediated FXR signaling in psoriasis. Consistently, downregulating Pparg and unaffected Fgf15 expression in lesions of IMQ-induced psoriatic mice further supported the notion that GC failed to activate FXR in psoriasis (Supplementary Fig. 3f).Fig. 3The GCDCA-mediated FXR activation alleviates the psoriatic lesions in IMQ-induced psoriatic mice.a Schematic diagram of dual-luciferase reporter containing NR0B2 promoter to assess ligand-dependent FXR activation. The figure was created in BioRender. Zhao, C. (2026) https://BioRender.com/peitdqe. b Molecular docking analysis showing the predicted binding affinity of GW4064 ligand to FXR. c Molecular analysis showing the predicted binding affinity of GCDCA to FXR. d Representative images of dorsal skin (top), and H&E staining (bottom) of IMQ-induced psoriatic mice treated with GC (50 mg/kg) or GW (20 mg/kg). e Quantification of epidermal thickness in the dorsal skin using Image J (n = 4, biological replicates). f The PASI scores assessing the erythema, scaling and infiltration on dorsal lesional severity on the sacrifice day (n = 4, biological replicates). g Spleens were isolated from control, IMQ-induced, and GC‑ or GW‑treated IMQ-induced mice on the sacrifice day (n = 3, biological replicates). h Spleen index comparisons among control, IMQ-induced, and GC or GW-treated IMQ-induced psoriatic mice (n = 4, biological replicates). i IF staining of K17 in dorsal skin from control, IMQ-induced, and GC or GW-treated IMQ-induced psoriatic mice (independently repeated three times). j Western blot analysis of K16, Ki67, and FXR expression in dorsal lesions of mice. Representative blots are representative of three independent experiments. The data are presented as the mean ± SD. Statistical analyses are performed using two-tailed one-way ANOVA followed by the Dunnett multiple comparison. All raw data are provided as a Source Data file. The GCDCA-mediated FXR activation alleviates the psoriatic lesions in IMQ-induced psoriatic mice. a Schematic diagram of dual-luciferase reporter containing NR0B2 promoter to assess ligand-dependent FXR activation. The figure was created in BioRender. Zhao, C. (2026) https://BioRender.com/peitdqe. b Molecular docking analysis showing the predicted binding affinity of GW4064 ligand to FXR. c Molecular analysis showing the predicted binding affinity of GCDCA to FXR. d Representative images of dorsal skin (top), and H&E staining (bottom) of IMQ-induced psoriatic mice treated with GC (50 mg/kg) or GW (20 mg/kg). e Quantification of epidermal thickness in the dorsal skin using Image J (n = 4, biological replicates). f The PASI scores assessing the erythema, scaling and infiltration on dorsal lesional severity on the sacrifice day (n = 4, biological replicates). g Spleens were isolated from control, IMQ-induced, and GC‑ or GW‑treated IMQ-induced mice on the sacrifice day (n = 3, biological replicates). h Spleen index comparisons among control, IMQ-induced, and GC or GW-treated IMQ-induced psoriatic mice (n = 4, biological replicates). i IF staining of K17 in dorsal skin from control, IMQ-induced, and GC or GW-treated IMQ-induced psoriatic mice (independently repeated three times). j Western blot analysis of K16, Ki67, and FXR expression in dorsal lesions of mice. Representative blots are representative of three independent experiments. The data are presented as the mean ± SD. Statistical analyses are performed using two-tailed one-way ANOVA followed by the Dunnett multiple comparison. All raw data are provided as a Source Data file. In mouse experiments, GC supplementation markedly ameliorated IMQ-induced psoriatic skin lesions, with an effect comparable to that of GW, a synthetic FXR agonist (Fig. 3d). Histological analysis confirmed the beneficial effects of GC on lesional thickness (Fig. 3d, e). Furthermore, psoriasis-related features, including erythema, scaling, infiltration, and PASI scores, were significant reduced by GC treatment (Fig. 3f). In addition, GC and GW significantly reduced spleen weight (Fig. 3g, h). Pathological examination showed a significant reduction in the expression of Ki67 and K17 in the skin following GC treatment (Fig. 3i, j), indicating a marked improvement in epidermal hyperproliferation. We found that GC treatment did not alter FXR expression in HaCaTs and IMQ-induced mice (Fig. 3j, Supplementary Fig. 3g), suggesting that FXR remained constitutively expressed in keratinocytes to sense the GC levels. Critically, we investigated the role of increased secondary BAs in psoriasis pathogenesis. Taurodeoxycholic acid (TDCA) or nordeoxycholic acid (norDCA), the two most upregulated secondary BAs in IMQ-induced psoriatic mice, were administered to mice separately. TDCA or norDCA treatment slightly aggravated the psoriatic phenotype and pathological characteristic compared to IMQ-induced psoriatic mice (Supplementary Fig. 3h–k), indirectly supporting the protective role of primary BAs-mediated FXR activation in psoriasis. Notably, the proective role of GC was dependent on the epidermal FXR, as evidenced by high ROS levels and more severe psoriasis phenotype after Nr1h4 depletion in vivo and in vitro (Supplementary Fig. 3l–o). In summary, our results show that FXR activation by either GC or GW ameliorates IMQ-induced psoriatic lesions, suggesting the dysfunction of BAs-FXR axis is a pivotal cause of psoriatic dermatitis. FXR activation ameliorates psoriatic lesions through upregulating NQO1 expression Our results showed that the disruptions in BAs metabolism occurred during psoriasis pathogenesis and the restoration of BAs metabolite (e.g., GC) inhibits the psoriasis pathology via FXR in keratinocytes. Although abnormalities in lipid metabolism have been reported in psoriasis17, the underlying causes of abnormal lipid metabolism and the FXR-mediated mechanisms in keratinocytes are not fully understood. Using publicly available datasets, we found increased lipid uptake and decreased lipid utilization in psoriatic skin, involving genes such as CD36, SLC27A4 and CPT1A (Supplementary Fig. 4a), identifying the pattern of lipid metabolism in psoriasis. IF staining confirmed increased lipid accumulation in psoriatic skin, as supported by increased expression of PLIN2, a marker of lipid droplet (Supplementary Fig. 4b, d). In the IMQ-induced animal model, FXR activation significantly upregulated FABP5, an intracellular chaperone of fatty acid, suggesting the regulatory role of FXR in epidermal lipid accumulation (Supplementary Fig. 4c, e). Furthermore, we assessed the expression of CD36, SLC27A4 and CPT1A, crucial genes for fatty acid transport and utilization. Results showed that FXR activation reduced fatty acid transport and promoted fatty acid utilization (Supplementary Fig. 4f, g). Notably, GW/GC treatment increased fatty acid β-oxidation (FAO) and ATP production (Supplementary Fig. 4h, i). We chose 100 μM GC to treat HaCaTs for in vitro assay based on the CCK-8 results (Supplementary Fig. 4j, l). Likewise, GC/GW treatment significantly reduced lipid accumulation in keratinocytes and increased ATP production (Supplementary Fig. 4k, m), further validating that FXR-mediated amelioration of psoriasis was achieved by restoring keratinocytes lipid metabolism. Additionally, the GC-mediated FXR activation reduced ROS levels, lipid accumulation, and abnormal proliferation in normal human epidermal keratinocytes (NHEKs; primary human keratinocytes) (Supplementary Fig. 5a, b). To elucidate the molecular mechanism underlying FXR-associated lipid metabolism, we performed RNA-seq in keratinocytes. The in vitro experiments showed that upregulated genes by FXR activation were associated with the “oxidoreductase activity” pathway (Fig. 4a).Fig. 4FXR activation ameliorates psoriatic lesions through upregulating NQO1 expression.a KEGG enrichment analysis of upregulated genes after GW treatment in IL-17 + PA-induced HaCaTs. b H2DCFDA staining of ROS in IL-17 + PA-induced HaCaTs with/without GW and GC treatment (independently repeated three times). c–e The GSH levels, SOD activity, and T-AOC activity in IMQ-induced psoriatic mice skin after GW or GC treatment (n = 3, biological replicates). f Venn diagram of genes commonly upregulated in IL-6-induced HaCaTs and IL-17 + PA-induced HaCaTs after GW treatment. g qRT-PCR analysis of NQO1 expression in HaCaTs (n = 3, biological replicates). h Schematic representation of two predicted FXR binding sites in the NQO1 promoter. i Structural interaction between FXR protein and NQO1 promoter predicted by alphafold3. j ChIP‒PCR analysis of H3K27ac showing FXR activation with GW or GC regulating the transcription of NQO1 gene at site 2 of FXR motif in IL-17 + PA-induced HaCaTs (n = 3, biological replicates). k FXR ChIP-PCR analysis showing increased binding of FXR to the NQO1 promoter at site 2 of FXR motif in IL-17 + PA-induced HaCaTs (n = 3, biological replicates). l Luciferase reporter assays showing the regulatory effect of FXR on NQO1 promoter activity in IL-17 + PA-treated HaCaTs with GW/GC treatment under wild-type and site-specific mutation conditions (n = 3, biological replicates). m Western blot of NQO1 expression in IL-17 + PA-treated HaCaTs (top panel) and in skin from IMQ-induced psoriatic mice (bottom panel) after GW or GC treatment (independently repeated three times). n Schematic illustration showing that FXR alleviates lipid metabolism‑related oxidative stress by upregulating NQO1 expression. The figure was created in BioRender. Zhao, C. (2026) https://BioRender.com/ezu44mi. The data are presented as the mean ± SD. Statistical analyses are performed using two-tailed one-way ANOVA followed by the Dunnett multiple comparison. All raw data are provided as a Source Data file. FXR activation ameliorates psoriatic lesions through upregulating NQO1 expression. a KEGG enrichment analysis of upregulated genes after GW treatment in IL-17 + PA-induced HaCaTs. b H2DCFDA staining of ROS in IL-17 + PA-induced HaCaTs with/without GW and GC treatment (independently repeated three times). c–e The GSH levels, SOD activity, and T-AOC activity in IMQ-induced psoriatic mice skin after GW or GC treatment (n = 3, biological replicates). f Venn diagram of genes commonly upregulated in IL-6-induced HaCaTs and IL-17 + PA-induced HaCaTs after GW treatment. g qRT-PCR analysis of NQO1 expression in HaCaTs (n = 3, biological replicates). h Schematic representation of two predicted FXR binding sites in the NQO1 promoter. i Structural interaction between FXR protein and NQO1 promoter predicted by alphafold3. j ChIP‒PCR analysis of H3K27ac showing FXR activation with GW or GC regulating the transcription of NQO1 gene at site 2 of FXR motif in IL-17 + PA-induced HaCaTs (n = 3, biological replicates). k FXR ChIP-PCR analysis showing increased binding of FXR to the NQO1 promoter at site 2 of FXR motif in IL-17 + PA-induced HaCaTs (n = 3, biological replicates). l Luciferase reporter assays showing the regulatory effect of FXR on NQO1 promoter activity in IL-17 + PA-treated HaCaTs with GW/GC treatment under wild-type and site-specific mutation conditions (n = 3, biological replicates). m Western blot of NQO1 expression in IL-17 + PA-treated HaCaTs (top panel) and in skin from IMQ-induced psoriatic mice (bottom panel) after GW or GC treatment (independently repeated three times). n Schematic illustration showing that FXR alleviates lipid metabolism‑related oxidative stress by upregulating NQO1 expression. The figure was created in BioRender. Zhao, C. (2026) https://BioRender.com/ezu44mi. The data are presented as the mean ± SD. Statistical analyses are performed using two-tailed one-way ANOVA followed by the Dunnett multiple comparison. All raw data are provided as a Source Data file. Additionally, HaCaTs were treated with IL-6, which is closely associated with psoriasis via activation of the STAT3 pathway18. Upregulated genes in IL-6-induced HaCaTs exhibited the “ferroptosis” pathway (Supplementary Fig. 5c), which was characterized by oxidative stress-mediated lipid peroxidation. The H2DCFDA staining showed that GW or GC reduced ROS levels in IL-17/PA-stimulated HaCaTs (Fig. 4b). In animal model, we investigated the superoxide dismutase (SOD) and thioredoxin (TXN), which are responsible for defending against oxidative stress. Sod1, Sod2, Txn1, and Txn2 were significantly upregulated after FXR activation (Supplementary Fig. 5d). Importantly, glutathione (GSH), SOD activity and total antioxidant capacity (T-AOC) were enhanced by GC/GW treatment, highlighting the role of FXR in boosting oxidative stress defense (Fig. 4c–e). An integrative analysis identified potential target genes, including NQO1, SLC7A11, SCD, and ASNS (Fig. 4f). We investigated NQO1 and SLC7A11 for their antioxidant and anti-ferroptotic capability. FXR activation induced NQO1 expression in keratinocytes, while SLC7A11 expression remained unchanged (Fig. 4g, Supplementary Fig. 5e). This suggests the involvement of NQO1 in FXR-mediated antioxidant defense. FXR binding motifs were identified in the NQO1 promoter, supporting its role in NQO1 regulation (Fig. 4h). Furthermore, AlphaFold 3‑based structural predictions provide hypothesis‑supporting computational evidence for a potential interaction between FXR and the NQO1 promoter region19 (Fig. 4i). Chromatin immunoprecipitation (ChIP)-PCR showed that FXR activation increased H3K27ac activity at the NQO1 promoter (Fig. 4j) and promoted the binding of FXR on NQO1 promoter (Fig. 4k). A luciferase assay further confirmed that FXR upregulated the NQO1 promoter activity, and the mutation of FXR binding site reduced the GW‑induced luciferase activity (Fig. 4l). In cell and animal models, FXR activation also increased NQO1 at the protein level (Fig. 4m). In summary, these computational predictions, together with experimental validation, support the conclusion that FXR upregulates NQO1 expression in keratinocytes, thereby modulating oxidative stress and lipid metabolism (Fig. 4n). Blockage of NQO1 disrupts FXR-mediated fatty acid oxidation in IMQ-induced psoriatic mice While NQO1 has been implicated in oxidative stress, its relationship to FXR-mediated NQO1 expression in keratinocytes lipid metabolism remains unclear. We investigated NQO1 expression in psoriasis patients and found decreased NQO1 expression in psoriatic lesions (Fig. 5a). This finding supports our observation of decreased GC levels and reduced FXR activity in psoriasis. Moreover, inhibiting NQO1 with dicoumarol (Dico) significantly exacerbated skin lesions (Fig. 5b, c). Subsequently, to investigate the role of FXR-NQO1 axis in FXR-mediated lipid metabolism, we examined the effects of NQO1 inhibition on the expression of PLIN2, CPT1A and FABP5. NQO1 inhibition increased lipid accumulation and decreased lipid utilization (Fig. 5d). In vitro, NQO1 knockdown reduced CPT1A expression in IL-17- and palmitic acid (PA)-treated HaCaTs (Fig. 5e), suggesting that NQO1-mediated oxidative stress impairs lipid utilization in psoriasis. Furthermore, treatment of IL-17 and PA-induced HaCaTs with the ROS scavenger N-acetyl-L-cysteine (NAC) restored CPT1A and FABP5 expression. It reduced PLIN2 levels (Fig. 5f), confirming that NQO1-mediated oxidative stress disrupts lipid metabolism in psoriasis.Fig. 5Blockage of NQO1 disrupts FXR-mediated fatty acid oxidation in IMQ-induced psoriatic mice.a IF staining of NQO1 expression in psoriatic lesions and healthy skin (independently repeated three times). b Representative images of dorsal skin (top) and H&E staining (bottom) in IMQ-induced psoriatic mice with Dico treatment (20 mg/kg). c PASI scores assessing the erythema, scaling and infiltration on dorsal lesional severity on the sacrifice day (n = 5, biological replicates). d Western blot of CPT1A, PLIN2, FABP5, Ki67 and K16 in dorsal lesions. e Western blot of NQO1, CPT1A, and PLIN2 expression in HaCaTs transfected with NQO1 siRNA, IL-17 and PA treatment. f Western blot of CPT1A, PLIN2, and FABP5 in IL-17- with PA- induced HaCaTs with ROS scavenger NAC (10 μM) treatment. g H2DCFDA staining of ROS levels in HaCaTs with GC and Dico treatment. h Western blot of PLIN2 and FABP5 in HaCaTs with GC and Dico treatment. i Representative images of dorsal skin (top), H&E staining (middle), and spleen (bottom) in the IMQ-induced psoriatic mice with GC (50 mg/kg) and Dico treatment (20 mg/kg). j PASI scores assessing the erythema, scaling and infiltration on dorsal lesional severity on the sacrifice day (n = 5, biological replicates). k IF staining of Ki67 and K17 in dorsal skin of IMQ-induced psoriatic mice with GC and Dico treatment (independently repeated three times). Representative blots are representative of three independent experiments. The data are presented as the mean ± SD. Statistical analyses are performed using two-tailed one-way ANOVA followed by the Dunnett multiple comparison. All raw data are provided as a Source Data file. Blockage of NQO1 disrupts FXR-mediated fatty acid oxidation in IMQ-induced psoriatic mice. a IF staining of NQO1 expression in psoriatic lesions and healthy skin (independently repeated three times). b Representative images of dorsal skin (top) and H&E staining (bottom) in IMQ-induced psoriatic mice with Dico treatment (20 mg/kg). c PASI scores assessing the erythema, scaling and infiltration on dorsal lesional severity on the sacrifice day (n = 5, biological replicates). d Western blot of CPT1A, PLIN2, FABP5, Ki67 and K16 in dorsal lesions. e Western blot of NQO1, CPT1A, and PLIN2 expression in HaCaTs transfected with NQO1 siRNA, IL-17 and PA treatment. f Western blot of CPT1A, PLIN2, and FABP5 in IL-17- with PA- induced HaCaTs with ROS scavenger NAC (10 μM) treatment. g H2DCFDA staining of ROS levels in HaCaTs with GC and Dico treatment. h Western blot of PLIN2 and FABP5 in HaCaTs with GC and Dico treatment. i Representative images of dorsal skin (top), H&E staining (middle), and spleen (bottom) in the IMQ-induced psoriatic mice with GC (50 mg/kg) and Dico treatment (20 mg/kg). j PASI scores assessing the erythema, scaling and infiltration on dorsal lesional severity on the sacrifice day (n = 5, biological replicates). k IF staining of Ki67 and K17 in dorsal skin of IMQ-induced psoriatic mice with GC and Dico treatment (independently repeated three times). Representative blots are representative of three independent experiments. The data are presented as the mean ± SD. Statistical analyses are performed using two-tailed one-way ANOVA followed by the Dunnett multiple comparison. All raw data are provided as a Source Data file. The H2DCFDA staining confirmed that NQO1 inhibition disrupted GC- and GW-mediated antioxidant effects in HaCaTs (Fig. 5g, Supplementary Fig. 5f). Dico treatment reduced GC- and GW-mediated lipid utilization as indicated by increased PLIN2 levels (Fig. 5h, Supplementary Fig. 5g). In IMQ-induced mouse model, NQO1 inhibition combined with FXR activation reversed GC- and GW-induced improvements in psoriatic skin lesions (Fig. 5i, j, Supplementary Fig. 5h). IF staining showed that NQO1 inhibition increased K17 and Ki67 expression, consistent with a hyperproliferative phenotype (Fig. 5k, Supplementary Fig. 5i). Moreover, the protective role of GC was also examined in IL-23-induced mouse model of psoriasis. Consistent with the findings in IMQ-induced psoriatic mice, FXR expression was upregulated and FXR activation by GC alleviated psoriatic phenotypes in IL-23-induced mice (Supplementary Fig. 5j–l). We also investigated the effect of GC treatment on the STAT3 and NF-κB pathway, which are pivotal for skin inflammation and psoriasis pathogenesis. The results showed that GC treatment reduced the phosphorylation of STAT3 and P65 (Supplementary Fig. 5m). However, inhibition of STAT3 and NF-κB pathway did not influence FXR and GC-induced NQO1 (Supplementary Fig. 5n), suggesting that FXR-mediated anti-oxidative capacity was independent of the STAT3/NF-κB pathway. In sum, these above findings collectively indicate that NQO1 inhibition aggravates the IMQ-induced psoriatic lesions, and that FXR-mediated lipid metabolism is dependent on NQO1. Gut microbiota dysbiosis causes dysfunctional bile acids profile in psoriasis Our findings support a model in which BAs-FXR-NQO1 signaling is involoved in lipid metabolic homeostasis and keratinocytes growth in psoriasis. However, the underlying causes of abnormal BAs metabolism in this context warrant further investigation. Accumulating evidence shows that the gut microbiota serves as a key modulator of BAs metabolism. Consistently, we found that IMQ-induced psoriatic mice exhibited the altered composition of gut microbiota (Fig. 6a, b, Supplementary Fig. 6a–d). α-diversity analysis revealed decreased ace and shannon index in IMQ-induced psoriatic mice. (Supplementary Fig. 6e). Using germ-free (GF) mice, we indicated that the absence of the gut microbiota ameliorated the psoriatic phenotype, including pathological appearance and histological changes (Fig. 6c–e). Spleen weight associated with the psoriasis severity also showed significant improvement (Fig. 6f). Immunofluorescence staining revealed decreased expression of K17, Ki67, and CD31 in IMQ-induced GF mice, indicating the inhibited epidermal differentiation and angiogenesis by the gut microbiota ablation (Fig. 6g). Notably, analysis of serum BAs metabolism showed that absence of gut microbiota increased the primary BAs and decreased the secondary BAs, restoring the metabolic profile of BAs in psoriasis (Fig. 6h). Further differential BAs analysis identified 11 BAs were significantly changed (Fig. 6i, j). The marked restoration of primary BAs (e.g., GC) which activate FXR signaling was a key factor in improving the psoriasis lesion (Fig. 6k).Fig. 6Gut microbiota dysbiosis causes dysfunctional bile acids profile in psoriasis.a The stacked histogram showing the gut microbiota composition at the family levels between control and IMQ-induced psoriatic mice. b LEfSe analysis showing potential bacterial biomarkers in the microbial community of IMQ-induced psoriatic mice. c Diagram illustrating the experimental grouping and workflow of mice under specific germ free (SPF) and germ-free (GF) environment. The figure was created in BioRender. Zhao, C. (2026) https://BioRender.com/vo6wpa2. d Representative images and H&E staining of dorsal skin from mice. e PASI scores assessing the erythema, scaling, and infiltration in SPF and GF mice treated with IMQ (n = 4, biological replicates). f Macroscopic images of spleens and spleen indices in mice (n = 3, biological replicates). g IF staining of K17, Ki67, and CD31 in lesional skin from SPF and GF mice treated with IMQ. h Bar-chart showing the levels of primary and secondary BAs in conjugated and free forms in SPF and GF mice treated with IMQ (n = 5, biological replicates). i Heatmap of 23 BAs in the serum of IMQ-induced SPF mice and GF mice (n = 3, biological replicates). j Volcano-plot showing the significantly changed BAs in IMQ-induced SPF mice and GF mice. k Bar-plot showing the GCDCA levels in the serum of IMQ-induced SPF mice and GF mice (n = 5, biological replicates). l IF staining of PLIN2 in lesional skin from SPF and GF mice treated with IMQ (independently repeated three times). m Western blot analysis of the expression of NQO1, FXR, K16, and FABP5 in IMQ-induced SPF mice and GF mice. Representative blots are representative of three independent experiments. The data are presented as the mean ± SD. Statistical analyses are performed using two-tailed Student’s t test. All raw data are provided as a Source Data file. Gut microbiota dysbiosis causes dysfunctional bile acids profile in psoriasis. a The stacked histogram showing the gut microbiota composition at the family levels between control and IMQ-induced psoriatic mice. b LEfSe analysis showing potential bacterial biomarkers in the microbial community of IMQ-induced psoriatic mice. c Diagram illustrating the experimental grouping and workflow of mice under specific germ free (SPF) and germ-free (GF) environment. The figure was created in BioRender. Zhao, C. (2026) https://BioRender.com/vo6wpa2. d Representative images and H&E staining of dorsal skin from mice. e PASI scores assessing the erythema, scaling, and infiltration in SPF and GF mice treated with IMQ (n = 4, biological replicates). f Macroscopic images of spleens and spleen indices in mice (n = 3, biological replicates). g IF staining of K17, Ki67, and CD31 in lesional skin from SPF and GF mice treated with IMQ. h Bar-chart showing the levels of primary and secondary BAs in conjugated and free forms in SPF and GF mice treated with IMQ (n = 5, biological replicates). i Heatmap of 23 BAs in the serum of IMQ-induced SPF mice and GF mice (n = 3, biological replicates). j Volcano-plot showing the significantly changed BAs in IMQ-induced SPF mice and GF mice. k Bar-plot showing the GCDCA levels in the serum of IMQ-induced SPF mice and GF mice (n = 5, biological replicates). l IF staining of PLIN2 in lesional skin from SPF and GF mice treated with IMQ (independently repeated three times). m Western blot analysis of the expression of NQO1, FXR, K16, and FABP5 in IMQ-induced SPF mice and GF mice. Representative blots are representative of three independent experiments. The data are presented as the mean ± SD. Statistical analyses are performed using two-tailed Student’s t test. All raw data are provided as a Source Data file. We then assessed BAs-mediated FXR activation and the downstream effects of lipid metabolism. Gut microbiota ablation significantly enhanced FXR’s activation, as evidenced by increased Nqo1 transcription and protein expression in IMQ-induced GF mice (Supplementary Fig. 6f, Fig. 6m). PLIN2 staining revealed that gut microbiota ablation reduced lesional lipid accumulation (Fig. 6l, m). To rule out possibility that the inherent immune deficiency in GF mice confounds the anti-psoriasis effects, GF mice were co-housing with SPF mice. The co-housed GF mice exhibited psoriatic phenotypic and histopathological features similar to those of SPF mice, suggesting that the alleviation in GF mice was attributed to gut microbiota ablation rather than to immune deficiency (Supplementary Fig. 6g–i). Additionally, we investigated the BAs profile and FXR expression in both BALB/c and C57BL/6 mice to assess the translatability of BALB/c-derived findings to the C57BL/6 background. Our cross-validation results showed that similar BAs profile and FXR expression between BALB/c and C57BL/6 mice, supporting the application of BALB/c-derived findings to C57BL/6 strain (Supplementary Fig. 6j, k). These findings reveal that during psoriasis pathogenesis, the gut microbiota dysbiosis leads to excessive transformation of primary BAs, which serves as a critical upstream signaling link to the skin FXR-NQO1 regulatory axis. Probiotics and antibiotics treatments alleviate psoriatic lesions in IMQ-induced psoriasis mice Gut microbiota dysbiosis was identified as an upstream trigger for the FXR-NQO1 regulatory axis in the pathogenesis of psoriasis. To develop a therapeutic strategy of modulating the gut microbiota in psoriasis, we implemented two distinct intervention strategies to increase the probiotic bacteria and decrease the pathogenic bacteria (Fig. 7a). Accordingly, we treated psoriatic mice with prebiotics and probiotics (Fig. 7b). Akkermansia (hereafter referred to as Akk) and inulin were used to enhance psoriasis associated probiotics, while antibiotics were used to reduce psoriasis-related pathogenic bacteria (Fig. 7b). Quantification of gut bacteria confirmed that antibiotics treatment was effective (Supplementary Fig. 7a, b). Inulin was selected as a prebiotic in this study because of its well-documented ability to promote the growth of Akk in vivo20,21. The co-administration of inulin with Akk was intended to synergistically enhance Akk colonization and amplify its therapeutic effects. Furthermore, 16S RNA sequencing and FISH analysis confirmed the increased abundance of Akk and its colonization in the colon (Fig. 7i, Supplementary Fig. 7c). Our results revealed that probiotics treatment alleviated psoriasis symptoms, with Akk and inulin combination yielding the most pronounced anti-psoriasis effects (Fig. 7c, d). Additionally, the antibiotics-induced clearance of the gut microbiota significantly alleviated the symptoms in the IMQ-induced psoriatic mice (Fig. 7c, d). Akk in combination with inulin and antibiotics treatment significantly decreased the spleen index compared to IMQ treatment (Fig. 7e, f). Histological examination revealed a reduction in epidermal thickness in IMQ-induced psoriatic mice following Akk, inulin, and antibiotics treatment (Fig. 7g). The decreased expression of K17 and Ki67 indicated that the epidermal abnormal differentiation and proliferation were inhibited by Akk and antibiotics (Supplementary Fig. 7d).Fig. 7Probiotics and antibiotics treatments alleviate psoriatic lesions in IMQ-induced psoriatic mice.a Schematic illustration showing the strategy of modulating gut microbiota to treat psoriasis. The figure was created in BioRender. Zhao, C. (2026) https://BioRender.com/ppvhfdr. b Diagram showing the treatment of probiotic (Akk), inulin (Inu), and four-combined antibiotics (Antibios) in IMQ-induced psoriatic mice. The figure was created in BioRender. Zhao, C. (2026) https://BioRender.com/zy3i89w. c Representative images of dorsal skin from mice. d PASI scores assessing the erythema, scaling, and infiltration in IMQ-induced psoriatic mice with Akk, inulin, and antibiotics treatment (n = 4, biological replicates). e, f Macroscopic images of spleens and spleen indices in IMQ-induced psoriatic mice with Akk, inulin, and antibiotics treatment (n = 4, biological replicates). g Histological H&E staining of dorsal skin from mice (independently repeated three times). h OPLS-DA plot of BAs profile in IMQ-induced psoriatic mice with Akk, inulin, and antibiotics treatment (n = 4, biological replicates). i The stacked histogram showing the composition of gut microbiota at the family levels in mice (n = 4, biological replicates). j Bar-plot showing the levels of serum BAs in IMQ-induced mice following the gut microbiota intervention (n = 4, biological replicates). k Bar-plot showing the primary to total BA ratio in IMQ-induced mice following the gut microbiota intervention (n = 4, biological replicates). l T-AOC in IMQ-induced psoriatic mice skin after gut microbiota intervention (n = 4, biological replicates). m Western blot analysis of the expression of NQO1, CPT1A, FABP5, and FXR in IMQ-induced psoriatic mice with Akk, inulin, and antibiotics treatment. n IF staining of PLIN2 in IMQ-induced psoriatic mice with Akk, inulin, and antibiotics treatment (independently repeated three times). Representative blots are representative of three independent experiments. The data are presented as the mean ± SD. Statistical analyses are performed using two-tailed one-way ANOVA followed by the Dunnett multiple comparison. All raw data are provided as a Source Data file. Probiotics and antibiotics treatments alleviate psoriatic lesions in IMQ-induced psoriatic mice. a Schematic illustration showing the strategy of modulating gut microbiota to treat psoriasis. The figure was created in BioRender. Zhao, C. (2026) https://BioRender.com/ppvhfdr. b Diagram showing the treatment of probiotic (Akk), inulin (Inu), and four-combined antibiotics (Antibios) in IMQ-induced psoriatic mice. The figure was created in BioRender. Zhao, C. (2026) https://BioRender.com/zy3i89w. c Representative images of dorsal skin from mice. d PASI scores assessing the erythema, scaling, and infiltration in IMQ-induced psoriatic mice with Akk, inulin, and antibiotics treatment (n = 4, biological replicates). e, f Macroscopic images of spleens and spleen indices in IMQ-induced psoriatic mice with Akk, inulin, and antibiotics treatment (n = 4, biological replicates). g Histological H&E staining of dorsal skin from mice (independently repeated three times). h OPLS-DA plot of BAs profile in IMQ-induced psoriatic mice with Akk, inulin, and antibiotics treatment (n = 4, biological replicates). i The stacked histogram showing the composition of gut microbiota at the family levels in mice (n = 4, biological replicates). j Bar-plot showing the levels of serum BAs in IMQ-induced mice following the gut microbiota intervention (n = 4, biological replicates). k Bar-plot showing the primary to total BA ratio in IMQ-induced mice following the gut microbiota intervention (n = 4, biological replicates). l T-AOC in IMQ-induced psoriatic mice skin after gut microbiota intervention (n = 4, biological replicates). m Western blot analysis of the expression of NQO1, CPT1A, FABP5, and FXR in IMQ-induced psoriatic mice with Akk, inulin, and antibiotics treatment. n IF staining of PLIN2 in IMQ-induced psoriatic mice with Akk, inulin, and antibiotics treatment (independently repeated three times). Representative blots are representative of three independent experiments. The data are presented as the mean ± SD. Statistical analyses are performed using two-tailed one-way ANOVA followed by the Dunnett multiple comparison. All raw data are provided as a Source Data file. To further investigate the role of BAs-FXR-NQO1 axis after probiotics and antibiotics interventions, we quantified serum BAs levels, assessed lesional FXR activation, and evaluated NQO1-mediated oxidative stress/lipid accumulation in Akk- and antibiotics-treated psoriatic mice. BAs profile analysis revealed that the gut microbiota intervention restored serum levels of primary BAs (Fig. 7h–k). Importantly, this shift in BAs metabolism coincided with FXR activation, as evidenced by upregulated NQO1 expression (Fig. 7m). To elucidate the functional consequences of the gut microbiota intervention in BAs/FXR/NQO1 axis, we further examined oxidative stress and cutaneous lipid accumulation following microbiota modulation. PLIN2 staining revealed that Akk and antibiotics treatment not only enhanced NQO1-mediated antioxidant responses, but also ameliorated pathological lipid accumulation in psoriatic lesions (Fig. 7m, n). Additionally, CD3 and F4/80 staining revealed that the gut microbiota intervention reduced the infiltration of T lymphocytes and macrophages into lesions (Supplementary Fig. 7d). Inulin administration alone failed to alleviate psoriatic symptoms (Supplementary Fig. 7e, f). Additionally, treatment of sodium butyrate (NaB), bacteria-derived metabolite, did not influence the expression of FXR and NQO1 in keratinocytes (Supplementary Fig. 7g), suggesting the crucial role of bacteria-mediated bile acid metabolism in psoriasis. Collectively, these results show that probiotics and antibiotics treatments attenuated the psoriatic lesions in IMQ-induced psoriatic mice through BAs/FXR/NQO1 axis. These results suggest microbiota-targeted interventions, as well as BAs supplementation and pharmacologic FXR modulation, as hypothesis-generating therapeutic strategies for psoriasis that require further mechanistic investigation and clinical evaluation. Discussion Our study reveals a regulatory axis between gut microbiota and skin lipid metabolism involved in the pathogenesis of psoriasis, offering a viable strategy for clinical application. The key findings are as follows: (1) Dysbiosis of the gut microbiota and a reduction in probiotics (e.g., Akkermansia) lead to decreased levels of primary BAs (e.g., GCDCA). (2) Primary BAs directly interact with FXR in epidermal keratinocytes, promoting the expression of NQO1, which helps maintain cellular antioxidant defenses and facilitates fatty acid utilization. (3) Targeting FXR, supplementing primary BAs, and restoring gut microbiota composition can effectively alleviate psoriatic symptoms and reduce skin hyperproliferation (see Fig. 8). These findings suggest that the BAs–FXR signaling plays a crucial role in maintaining keratinocytes metabolic homeostasis by regulating oxidative stress responses.Fig. 8Proposed model of gut microbiota dysbiosis and psoriasis pathogenesis.We identified a gut-skin axis in keratinocytes lipid metabolism and the pathogenesis of psoriasis. Mechanistically, gut microbiota dysbiosis leads to the dysfunctional BAs metabolism in psoriasis, characterized by decreased serum levels of primary BAs and increased levels of secondary BAs. These altered BAs enter the skin via the circulation. The reduced primary BAs are insufficient to activate FXR receptors. Insufficient FXR signaling fails to directly bind to the NQO1 promoter, thereby reducing the skin’s antioxidant capacity. Dysregulated oxidative stress impairs lipid utilization and promotes lipid accumulation in the skin, ultimately contributing to the development of psoriasis. The illustration was created in BioRender. Zhao, C. (2026) https://BioRender.com/p1wlxvb. Proposed model of gut microbiota dysbiosis and psoriasis pathogenesis. We identified a gut-skin axis in keratinocytes lipid metabolism and the pathogenesis of psoriasis. Mechanistically, gut microbiota dysbiosis leads to the dysfunctional BAs metabolism in psoriasis, characterized by decreased serum levels of primary BAs and increased levels of secondary BAs. These altered BAs enter the skin via the circulation. The reduced primary BAs are insufficient to activate FXR receptors. Insufficient FXR signaling fails to directly bind to the NQO1 promoter, thereby reducing the skin’s antioxidant capacity. Dysregulated oxidative stress impairs lipid utilization and promotes lipid accumulation in the skin, ultimately contributing to the development of psoriasis. The illustration was created in BioRender. Zhao, C. (2026) https://BioRender.com/p1wlxvb. The nuclear receptor (NR) family of transcription factors are potent regulators in immune response and metabolic processes22. This is exemplified by PPAR, VDR, LXR, RAR, AHR, PXR and FXR, which not only regulate lipid homeostasis but also critically modulate cell proliferation and inflammation23–25. In dermatology, NR signaling impacts epidermal health-e.g., retinoic acid (RA) treats psoriasis and wrinkles. Activation of RAR is able to inhibit excessive keratinocytes differentiation, underlying the protective role of NR in skin health26. Xiao et al. reported that reduced LXRα and PPARγ correlates with increased keratinocytes proliferation and inflammation27. While PPARδ is markedly increased in psoriatic skin, which activating WNT signaling pathway to promote keratinocytes proliferation28. Additionally, NRs also regulate lymphocyte fate in skin. RAR-related orphan receptors (ROR), such as RORα and RORγ, are aberrantly overexpressed in psoriatic epidermis. Elevated RORγ in psoriatic skin promotes Th17 differentiation and IL-17-mediated inflammatory responses29,30. Critically, NRs govern sebaceous lipid metabolism and skin barrier integrity31. The aryl hydrocarbon receptor (AHR)/AHR-nuclear translocator (ARNT), sensor for dioxins, phytochemicals, microbial bioproducts, and tryptophan photoproduct, are abundantly expressed in the skin, strengthening skin barrier functions and accelerating epidermal terminal differentiation32, Notably, NRF2, via antioxidant responses, links oxidative stress to skin pathology33,34. Although FXR expression is increased in psoriatic lesions, we interpret this as inflammation-driven aberrant induction rather than negative feedback to altered BAs levels. Unlike PPARs and LXRs, which are downregulated, FXR resembles nuclear receptor related 1 (NURR1) and androstane receptor (CAR) in its expression pattern24,35–37. Our study indicates that FXR activation enhances lipid metabolism and reduces oxidative stress in keratinocytes, which is associated with the attenuation of psoriasis. Collectively, our data support a mechanistic link between BAs homeostasis and FXR activation in psoriasis and extend our understanding of NR functions in skin homeostasis. The FXR is a member of the nuclear receptor superfamily, binding to genomic FXR response element (FXRE) either in FXR monomer or FXR:RXR heterodimer. It plays a critical role in regulating the metabolic balance of BAs, lipids, and glucose38. FXR is highly expressed in the liver, intestine, kidney, and pancreas. Functionally, it modulates the development of intestinal infections, gallstones, liver regeneration, and tumor via regulating key metabolic hub genes, notably small heterodimer partner (SHP), fibroblast growth factor (FGF), aquaporin 2 (AQP2), and krueppel-like factor 11 (KLF11)39–43. These findings suggest that FXR function is highly tissue-dependent. Recently, its importance in tissues with lower expression levels, such as adipose tissue, muscle, nerve, and skin, has attracted attention. Notably, some studies have indicated that BAs and FXR are involved in ferroptosis44. Research has revealed that FXR activation suppressed ferroptosis by upregulating key proteins involved in lipid peroxidation, including GPX4, FSP1, PPARα, SCD1, and ACSL344,45, suggesting that the most fundamental function of FXR is antioxidation. Our study using a transgenic mouse model indicates that FXR upregulates NQO1 gene expression and enhances the antioxidant capacity in psoriasis, highlighting that the physiological role of FXR in maintaining skin homeostasis and the function of FXR in enhancing anti-oxidant capability. Collectively, our results identify FXR as a critical regulator of BAs–FXR signaling and redox homeostasis in psoriatic skin, supporting a broader role for FXR in peripheral tissue dysfunction associated with multifactorial diseases. Studies have reported some serum BAs in psoriasis patients46. A dysregulated BAs profile is reported to influence immune-mediated inflammatory diseases, including multiple sclerosis, inflammatory bowel disease, and Th2- and Th17-mediated dermatitis10. The composition and abundance of BAs were changed by gut microbiota through oxidation, deconjugation, and dihydroxylation. Research indicates that the gut microbiota dysbiosis is closely associated with various disorders, such as hepatitis, inflammatory bowel diseases, cholestatic jaundice, obesity, and psoriasis47–49. In patients with cholestatic jaundice, pathogenic bacteria such as Parabacteroides distasonis and Prevotella are increased, leading to alterations in BAs synthesis and conversion50. Many studies have reported that dysbiosis of the gut microbiota dysbiosis in psoriasis, and microbiota composition in psoriasis patients with severe lesions was different from that in psoriasis patients with more mild lesions or healthy controls51. Our study revealed a reduction in the abundance of Akkermansia, which was recognized as a psoriasis-associated gut bacterium52,53. Moreover, we found that supplementation with Akkermansia can improve psoriasis lesions through the gut-skin axis. Consistent with our results on Akkermansia-regulated BAs metabolism in psoriasis, it has been reported that Akkermansia increase the primary BAs levels in the cecum and enhance intestinal FXR signaling in liver diseases54. Additionally, emerging evidence reveals the potential mechanism: Akkermansia, via its metabolite harmaline (HAL), increases the levels of serum GCDCA and TCDCA by upregulating bile acyl-CoA:amino acid N-acyltransferase (BAAT) expression in hepatocytes, thereby catalyzing the conjugation of primary BAs with glycine or taurine55. Current evidence suggests that gut microbiota-derived metabolites, including short-chain fatty acids (SCFAs), tryptophan metabolites, and β-carboline alkaloids, primarily contribute to the regulation of immune homeostasis, mucosal barrier function, and inflammatory responses56,57. However, there is currently no evidence that these metabolites directly activate or inhibit FXR signaling, and any potential effects on FXR remain to be elucidated. Additionally, using GF mice, our finding reveals that the gut microbiota dysbiosis in psoriasis promotes BA transformation from primary BAs to secondary BAs. Dysregulated BAs profile is involved in psoriasis. These findings provide a perspective on the role of the gut microbiota-mediated BAs metabolism in skin health. We show that perturbation of the bile acid-FXR axis (exemplified by GCDCA) is a feature of psoriasis. This pathway serves as one significant regulatory element within the complex pathogenic network of psoriasis, potentially intersecting with mechanisms shared across other inflammatory conditions. However, several limitations must be considered when interpreting these findings. First, although the importance of GCDCA in psoriasis is highlighted, the other composition of BAs and their specific roles during psoriasis warrant further investigation9. Second, despite promising results in IMQ-induced mouse model, the applicability of these findings in the clinic requires further validation58. In addition, the precise functions of FXR in keratinocytes and its interactions with other signaling pathways remain poorly characterized due to limitations in skin tissue validation. Future research should address these issues to improve our understanding of the underlying pathological mechanisms. Methods Ethics statement This study complied with all relevant ethical guidelines. All human procedures were approved by the Clinical Research Ethics Committee of the First Affiliated Hospital of Nanjing Medical University, and written informed consent was obtained from all participants, in accordance with the principles of the Declaration of Helsinki. The animal experiment was approved by the Ethics Committee of Medical School of Nanjing University and were conducted in accordance with institutional guidelines and relevant regulations. Human specimens Skin biopsy samples were obtained from psoriasis patients and healthy subjects undergoing esthetic plastic surgery. The clinical characteristics of healthy subjects and psoriasis patients, including BMI, sex, and age, were presented (Supplementary Table 1). Mice GF BALB/c mice were purchased from GemPharmatech Company (strain no. N000296). Germ-free mice were bred and maintained in special plastic isolators. All mice were housed under a strict 12 h light cycle. Mice were supplied with a 50-kGy irradiated sterile pelleted normal chow diet and autoclaved tap water ab libitum (Synergistic Biology, XTI01WC-009). The bedding was replaced every 7 days in all the experiments. All GF mice were screened for bacteria, viral, and fungus contamination (Supplementary Table 2). For the co-housing experiments, 6-week-old female GF mice from the same breeder were divided into two groups: GF mice and GF mice co-housing with SPF mice. The co- housed GF mice were maintained with age-matched female SPF mice in the same cages. In parallel, SPF mice housed individually served as controls. All mice were subjected to IMQ cream application (Mingxinlidi, Sichuan) for 1 week. Nr1h4 knockout (KO) mice (strain no. T012640), and KRT14-iCre mice (strain no. T004833) were purchased from GemPharmatech Company. In Nr1h4 KO mice, exon2 of Nr1h4 gene was deleted by CRISPR/Cas9 technology. For keratinocytes-specific Nr1h4 KO, KRT14-iCre mice were co-housed with Nr1h4 loxp/loxp mice (strain no. T009517) to produce the KRT14-iCre-Nr1h4 loxp/loxp mice. Six- to 8-week-old female BALB/c and C57BL/6 mice were purchased from GemPharmatech Company. The psoriatic mouse model induced by imiquimod (IMQ) was induced on the shaved dorsal skin59. In brief, the dorsal skin of each mouse was shaved 2 cm × 3 cm and then depilated until it was fully exposed. A total of 62.5 mg of 5% IMQ cream (Mingxinlidi, Sichuan) was applied to the exposed skin for 6 consecutive days. In parallel, the control mice were treated with Vaseline cream. During induction of the IMQ-induced psoriasis model, mice were received once-daily intraperitoneally administered with GW4064 (20 mg/kg; MedChemExpress, #HY-50108), GCDCA (50 mg/kg; MedChemExpress, #HY-N2334), Dicumarol (20 mg/kg; MedChemExpress, HY-N0645), TDCA (50 mg/kg; MedChemExpress, #HY-B1899) and NorDCA (50 mg/kg; MedChemExpress, HY-N8268) for 6 consecutive days. Vehicle controls matched the solvent composition without active compounds. The vehicle was formulated by sequential addition of 10% dimethyl sulfoxide (DMSO), 40% polyethylene glycol 300 (PEG300), 5% Tween-80, and 45% physiological saline (0.9% NaCl). The IL-23-induced psoriatic model was induced by intradermal injection with 20 µL of PBS alone or PBS containing 0.5 μg recombinant mouse (rm) IL-23 (Sinobiological, #CT028-M08H) into the ear. Mice were subjected to IL-23 injection every other day for 18 days. Probiotics supplementation was conducted using a commercially available Akkermansia muciniphila (Akk) strain (1 × 10⁹ CFU/mL). Probiotics were freshly dissolved in sterile phosphate-buffered saline (PBS) and administered daily by oral gavage at a dosage of 200 µL/mouse for 14 days. Prebiotics were provided by supplementing the mouse diet with 5% (w/w) inulin (Synergistic Biology, Customized products). The inulin-supplemented diet was provided ad libitum. To deplete the gut microbiota, an antibiotic cocktail consisting of ampicillin (1 g/L), neomycin (1 g/L), metronidazole (1 g/L), and vancomycin (0.5 g/L) was administered in drinking water. The animal experiment was approved by the Medical School of Nanjing University Ethics Committee. All animal procedures, including welfare monitoring and euthanasia, were carried out in full compliance with the ARRIVE guidelines and all applicable regulations. Psoriasis area and severity index (PASI) score The PASI is a widely-recognized metric for evaluating the severity of psoriasis. It integrates the index of severity (erythema, scaling and infiltration) and the percentage of affected area. Erythema was quantified based on the degree of redness on psoriatic lesions. Scaling was defined as the visible flaking or peeling of psoriatic plaques. Skin infiltration assessed the thickness or elevation of the lesions. Each index was scored on the scale from 0 (none) to 4 (severe). Cell culture Human epidermal keratinocytes (HaCaTs) (XL biotec, #300493) and normal human epidermal keratinocytes (NHEKs) (Keycell, #QS-Q125) were cultured in Dulbecco’s modified Eagle’s medium (Gibco) supplemented with 10% fetal bovine serum (Gibco) and 100 U/mL penicillin‒streptomycin (Thermo Fisher Scientific) at 37 °C in 5% CO2. After reaching 70% confluence, keratinocytes were treated with IL-17 (50 ng/mL) and PA (100 μM) for 24 h. GW4064 (10 μM) and GCDCA (100 μM) were added to the culture medium to activate FXR signaling. HaCaTs were harvested for western blot, transcriptomic sequencing, H2DCFDA staining, and BODIPY staining. Hematoxylin and eosin (H&E) staining Dorsal skin samples were fixed in 4% paraformaldehyde for 48 h and embedded in paraffin. After deparaffinization, the skin samples were cut into 10-μm-thick slices. These slides were then subjected to H&E staining for morphological analysis. Hyperkeratosis, parakeratosis, and Munro’s microabscess were evaluated in H&E-stained sections. Images were acquired at 200× magnification using a Leica microscope system. Immunofluorescence (IF) and immunohistochemistry (IHC) analysis The skin sections were deparaffinized in dimethylbenzene and rehydrated in an ethanol solution. After antigen retrieval, the sections were incubated with the appropriate primary antibody at 4 °C overnight. For IHC, the sections were washed with PBS, followed by incubation with an HRP-conjugated secondary antibody. The sections were then developed with a DAB substrate and counterstained with hematoxylin. For IF, the sections were washed with PBS and then incubated with a fluorescence-conjugated secondary antibody. The sections were mounted with mounting medium containing DAPI and visualized using a fluorescence microscope. The primary antibodies used included anti-phospho-STAT3 (Tyr705) (Cell Signaling Technology, #9145), anti-STAT3 (Cell Signaling Technology, #12640), anti-Ki67 (Beyotime, #AF1738), anti-Keratin 16 (Santa Cruz Biotechnology, #sc-53255), anti-Keratin 17 (Cell Signaling Technology, #12509), anti-PLIN2 (Proteintech, #15294-1-AP), anti-FABP5 (Proteintech, #12348-1-AP), anti-FXR (Abclonal, #A24015), anti-NQO1 (Abclonal, #A19586), CD31 (Servicebio, #GB11063-3-100), anti-P65 (Beyotime, #AF1234), anti-p-P65 (Beyotime, #AF5875), anti-TGR5 (Abcam, #ab72608), anti-CD3 (Cell Signaling Technology, #78588), anti-F4/80 (Cell Signaling Technology, #CST70076). Fluorescence in situ hybridization (FISH) Colon tissue samples were fixed in 2.5% glutaraldehyde. Akkermansia-specific probe [Akk1437 (S-G-Akk-1437-a-A-20), 5′-CCTTGCGGTTGGCTTCAGAT-3′] was diluted in 1:100 to make the hybridization solution. The probe solution was hybridized with the tissue sections at 46 °C for 2 h according to a liquid-FISH protocol60. Superoxide dismutase (SOD), reduced glutathione (GSH), and total antioxidant capacity (T-AOC) analysis SOD activity was measured via a SOD assay kit (Jiancheng, #A001-3-2), which was based on the WST-1 method. The absorbance was measured at 450 nm. The SOD activity was calibrated by using a SOD standard curve. The results are presented as units per microgram of lysis in the skin (U/mg protein). One unit was defined as the amount of enzyme required to exhibit 50% superoxide radical dismutation. The amount of reduced GSH was measured via the enzymatic recycling of 5,5’-dithio-bis (2-nitrobenzoic acid) (DTNB). According to a GSH assay kit (Jiancheng, #A006-2-1), reduced glutathione (GSH) reacts with DTNB to form a yellow compound, which can be measured at 405 nm. The reduced GSH levels were quantified as micromoles per gram of lysis in the skin (μmol/g protein). The T-AOC was determined following the kit instructions (Jiancheng, #A015-2-1). T-AOC inhibited the formation of the green 2,2’-azino-bis-3-ethylbenzthiozoline-6-sulfonate (ABTS+) radical. The capacity of the skin antioxidant was compared to that of the water-soluble vitamin E analog (Trolox), which is a standard for the total antioxidant capacity assay. The absorbance was measured at 405 nm. The total antioxidation ability is presented as millimoles per Trolox equivalents. Reactive oxygen species (ROS) The ROS levels in HaCaTs were measured by H2DCFDA (MedChemExpress, #HY-D0940), an ROS-sensitive fluorescent probe. HaCaTs were seeded in 12-well plates and treated with IL-17 (Sino, #CT047-HNAE), PA (Kunchuang, #KC001), GW4064(MedChemExpress, #HY-50108), and GCDCA (MedChemExpress, #HY-N2334) for 24 h. The growth media was removed, and the samples were washed with PBS for 3 times. Freshly prepared working solutions of H2DCFDA (20 μM) were added to HaCaTs in the dark for 2 h of incubation. The intensity of fluorescence was monitored at 200× magnification via an FV3000 confocal microscope system. ATP measurement ATP levels in the lesional skin of mice and IL-17 + PA-induced HaCaTs were determined by chemiluminescence with an ATP Assay Kit (Beyotime, #S0026B). According to the manufacturer’s instructions, the lysis buffer was added to an equal amounts of skin tissue and HaCaTs. After preparing the different concentrations of ATP standard solution to generate the curve, the supernatants of lysis and standard samples were mixed with ATP detection solution to detect the luminescence. The ATP concentrations of the samples were calculated based on the regression equation derived from the standard curve, correlating luminescence intensity (RLU) with known ATP concentrations (nmol/L). Lipid droplets staining The lipid droplet contents were stained by the lipophilic dye BODIPY 493/503 (Beyotime, #C2053S). The working solution of BODIPY 493/503 was prepared following the manufacturer’s instructions. HaCaTs were cultured in six-well plates. The growth medium was removed, and the samples were washed with PBS for 3 times. HaCaTs were fixed with 4% paraformaldehyde, followed by incubation with BODIPY 493/503 staining solution for 30 min. The intensity of fluorescence was monitored at 40× magnification via an Megarobo (CellVue® T3000) microscope system. Cell transfection Plasmid DNA was purified via Endo-Free Plasmid Midi Kits (Omega Biotec, #D6950). Oligonucleotides for RNAi were synthesized and provided by Sangon Biotech Co., Ltd. For siRNA transfection, NQO1 siRNA, NR1H4 siRNA, and negative control siRNA were transfected into HaCaTs by using Lipofectamine™ RNAiMAX (Invitrogen, #13778030) according to the manufacturer’s guidelines. The sequences of siRNA used are listed in Supplementary Table 3. Western blot Protein from HaCaTs and skin tissue was extracted using RIPA lysis buffer (Beyotime, #P0013B) supplemented with protease and phosphatase inhibitors. The protein concentration was measured with a BCA assay kit (Beyotime, #P0009). Proteins (10–20 μg) were separated by 10% SDS‒PAGE and transferred to a PVDF membrane, which was subsequently blocked in 5% bovine serum albumin (BSA) for 1 h at room temperature. The PVDF membrane was incubated with the primary antibody at 4 °C overnight. The next day, the membranes were incubated with a secondary antibody conjugated with HRP (Bioworld, #BS13278) for 1 h. Finally, enhanced chemiluminescence was used to visualize protein levels using digital imaging systems. The following antibodies were used: anti-GAPDH (Bioworld, #BS60630). Additionally, anti-phospho-STAT3 (Tyr705) (Cell Signaling Technology, #9145), anti-STAT3 (Cell Signaling Technology, #12640), anti-Ki67 (Beyotime, #AF1738), anti-Keratin 16 (Santa Cruz Biotechnology, #sc-53255), anti-Keratin 17 (Cell Signaling Technology, #12509), anti-PLIN2 (Proteintech, #15294-1-AP), anti-FABP5 (Proteintech, #12348-1-AP), anti-CD36 (Proteintech, #18836-1-AP), and anti-CPT1A (Proteintech, #66039-1-Ig), anti-FXR (Abclonal, #A24015), anti-NQO1 (Abclonal, #A19586), and anti-CD31 (Servicebio, #GB11063-3-100). The uncropped and unprocessed scans of blots are provided in the Source Data file. RNA extraction and qRT-PCR Total RNA was isolated from HaCaTs and skin tissue using TRIzol reagent (Vazyme, #R401-01). The isolated RNA was reverse transcribed into single-stranded cDNA with a reverse transcription kit (Vazyme, #R302-01). All the quantitative PCR assays were performed with PowerUp SYBR Master Mix (Applied Biosystems, #A25742). The relative quantity of mRNA was assessed by the 2−ΔΔCT method. The PCR primers were synthesized and provided by Sangon Biotech Co., Ltd. The sequences of the primers used are listed in Supplementary Table 3. Microbiota quantification Total gut microbial load was assessed through the relative quantification of the bacterial 16S rRNA gene via qRT-PCR, with detailed procedures outlined below. Microbial genomic DNA was isolated from fecal samples of mice using the kit (Yeasen, #18820ES50). DNA purity and concentration were measured prior to analysis. qRT-PCR amplification was performed using universal primers 515-F and 806-R targeting the V3-V4 regions (Supplementary Table 3). The PCR primers were synthesized and provided by Sangon Biotech Co., Ltd. The relative abundance of the total gut microbiota was quantified by comparing the 16S rRNA gene copy number to the baseline (day 0). Results are expressed as the fold change in bacterial load relative to day 061. Online data analysis Single-cell RNA sequencing (scRNA-seq) data for psoriasis lesions and normal skin were obtained from an online repository (https://yz-studio.shinyapps.io/psoriaticskincellatlas2/)62. Additionally, RNA-seq data containing 92 psoriatic and 82 normal skin samples were obtained from the GEO database (GSE54456)63. The molecular docking between FXR and NQO1 promoter was predicted by online alphafold3 (https://alphafoldserver.com/). 16S rRNA sequencing Fecal samples were fast frozen with liquid nitrogen, and stored at −80 °C. After fecal genomic DNA extraction, the purity and concentration of DNA were examined. The diluted genomic DNA was used as a template for amplification using primers (515F and 806R) in the 16S V4 region, and the PCR products were separated on a 2% agarose gel. The PCR product was subjected to magnetic-bead purification and library preparation, which were then sequenced by NovaSeq6000. Raw sequencing data were demultiplexed and adapters (barcodes/primers) trimmed. Quality filtering was performed using fastp to obtain high-quality reads. Paired-end reads were assembled into clean tags with FLASH. All high-quality tag counts were clustered into operational taxonomic units (OTUs) at 97% identity threshold using the UPARSE algorithm. SSU rRNA sequences were taxonomically annotated against the SILVA database (http://www.arb-silva.de/) using 0.8–1.0 similarity thresholds. Taxonomic profiling was then performed across all hierarchical levels (phylum to species). Bacterial composition analysis (α-diversity and β-diversity) was performed with the phyloseq (v1.40.0) and vegan (v2.6.2) packages in the R software to plot PCoA. Gut microbiota composition was analyzed using Aitchison distance (v 2.0.8), with between-group differences tested by PERMANOVA (v 0.2.0). Differential bacteria between groups were identified using linear discriminant analysis effect size (LEfSe, v1.1.2) with a LDA score setting of 4, and ANCOM analysis (v 4.0.2) in the R software. The 16S rRNA sequencing of feces was performed by the Metware company. Targeted bile acids measurement Serum samples were collected from human (n = 6, three psoriasis patients and three healthy controls) and mice (n = 10, five IMQ-induced GF mice and five IMQ-induced SPF mice) for bile acids analysis; the ethics statement and participant age information are provided in relevant sections. Quality control was ensured by pooled QC samples. An aliquot of serum was mixed with ice-cold methanol containing internal standards. After vortexing and centrifugation, the supernatant was carefully transferred to a new tube and dried under a gentle stream of nitrogen gas. The dried extract was reconstituted in initial mobile phase for LC-MS analysis. All samples were filtered through a 0.22 μm membrane before LC-MS analysis. LC–MS analysis was performed using an ultrahigh-performance liquid chromatography (UHPLC) system (Waters, USA) coupled to an AB SCIEX 5000 tandem mass spectrometer (AB SCIEX, USA) to profile serum bile acids. Bile acids were annotated by matching their accurate mass (<5 ppm error) and retention time to commercial standards. For others, putative identification was based on matching accurate mass and MS/MS spectra with entries in the HMDB database. Univariate statistical tests (two-tailed Student’s t-test, were performed using GraphPad Prism 8). The profile of serum BAs in IMQ-induced psoriatic mice and psoriasis patients was provided by the Metware company. Quantitative lipidomics The quantitative lipidomics analysis was performed using an LC-ESI-MS/MS system consisting of an ExionLC AD UPLC and a QTRAP® 6500+ mass spectrometer. For sample preparation, 20 mg of ice-thawed −80 °C samples were homogenized. Lipids were extracted with 1 mL MTBE:MeOH (3:1, v/v) containing internal standards, then 200 μL water was added. The sample extracts were analyzed using an LC-ESI-MS/MS system (UPLC, ExionLC AD; MS, QTRAP® 6500+ System). Mass spectrometric detection was conducted in both positive and negative ESI modes with the following parameters: turbo spray ion source at 500 °C, ion spray voltages of 5500 V (positive) and -4500 V (negative), GS1 = 45 psi, GS2 = 55 psi, CUR = 35 psi. Multiple reaction monitoring (MRM) transitions were optimized for each metabolite, with nitrogen as collision gas (5 psi). Data acquisition and processing were controlled using Analyst 1.6.3 software. RNA sequencing (RNA-seq) HaCaTs were cultured in six-well plates and the total RNA was then isolated by using TRIZOL reagent. RNA-seq was performed by the Beijing Genomics Institute using standard protocols. Briefly, RNA samples were denatured and mRNA were enriched using oligo(dT)-attached magnetic beads. The RNA was then fragmented. First-strand cDNA was synthesized, followed by synthesis of the second strand using dUTP. The adapters were ligated, and quality control was performed. This cDNA is then subjected to high-throughput sequencing technologies to generate millions of short-read sequences. These reads were then aligned to a GRCh38/hg38 genome using bioinformatics tools HISAT2. Differential expression analysis was conducted to identify significant genes by using DESeq2. Chromatin immunoprecipitation (ChIP) assays HaCaTs were cultured in 150-mm culture dishes and fixed with 1% formaldehyde for 15 min. Chromatin was fragmented and treated with protein A/G magnetic beads (MedChemExpress, #HY-K0202). Anti-H3K27ac (Abcam, #4729), anti-FXR antibody (Cell Signaling Technology, #72105) and mouse anti-IgG were added to equal aliquots of chromatin and incubated at 4 °C overnight. The purified DNA from the input and immunoprecipitated samples were then subjected to PCR with NQO1 promoter-specific primers. The sequences of the NQO1 promoter-specific primers used are listed in Supplementary Table 3. Luciferase reporter assay Luciferase reporter assays were performed using a dual-luciferase assay, normalizing firefly to Renilla activity, as described previously3. The proximal promoter regions of human NQO1 and corresponding mutants were cloned into the firefly luciferase reporter construct pGL3-basic. HaCaTs were transfected with pRL-TK and pGL3-WT (or pGL3-MUT) vector at a ratio of 1:20, followed by IL-17 and PA treatment. After 48 h, luciferase activity was measured according to the manufacturer’s instructions (Vazyme, #DL101-01). Molecular docking AutoDock-GPU1 was used for molecular docking simulation of molecules with human FXR to obtain possible conformations and orientations of the ligand at the binding sites. The crystal structure of human FXR (PDB code: 3DCT) was downloaded from the RCSB Protein Data Bank (https://www.rcsb.org). The 3D structures of GW4064 and GCDCA were downloaded from PubChem (https://pubchem.ncbi.nlm.nih.gov/) and used as ligands. The Discovery Studio Visualizer (Dassault Systèmes Biovia Corp, Version 18.1.100.18065) was employed to visualize the interactions between molecules and proteins. The three-dimensional binding models of molecules and proteins were mapped by PyMOL (Schrödinger, LLC. Version 2.5.0). Statistical analysis The data are shown as the means ± standard deviations (S.D.). The statistical significance of the data was calculated by unpaired two-tailed Student’s t test for two groups, and one-way ANOVA followed by the Dunnett multiple comparison test for more than two groups in GraphPad Prism 8 software (*p < 0.05, **p < 0.01, or ***p < 0.001). Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplemental materials Reporting Summary Transparent Peer Review file Supplemental materials Reporting Summary Transparent Peer Review file Source data Source Data Source Data