Fig 2: Screening, prediction, and validation of caffeine directly targets Dusp9. (A) Workflow for identifying potential drug targets of caffeine. (B) Molecular docking of caffeine with the protein Dusp9. (C, D) The affinity of CAFF for Dusp9 was determined by using SPR. (E) Representative immunoblots from (Cellular Thermal Shift Assay) CETSA experiments in THLE-2 cells (42/48/53/58/63 °C). (F) Melt curves of Dusp9 in THLE-2 cells after caffeine or DMSO treatment by CETSA. (G) Drug affinity responsive target stability (DARTS) assay was performed to test the direct binding of Caffeine to Dusp9. (H) Fold change of Dusp9 in THLE-2 cells after caffeine (0.25/0.5/1 mM) or DMSO treatment by DARTS. (I) Expression levels of hepatic Dusp9 protein were examined with Western blot in HFHC MASH mice. (J) Western blot data of relative expression of Dusp9 protein in HFHC MASH mice. GAPDH served as a loading control. (n = 3 per group). (M) Immunohistochemical detection of Dusp9 (brown/tan) and nuclei (blue) in liver tissue from HFHC-fed MASH mice. (K) Western blot data of relative expression of Dusp9 protein in FFA-induced THLE-2 cells. (L) Relative levels of Dusp9 were analyzed in THLE-2 cells. Experiments were performed three times independently, and values represent means ± SD.∗, compared with Con p < 0.05; ∗∗, compared with Con p < 0.01; #, compared with HFHC p < 0.05; ##, compared with HFHC p < 0.01. The data were analyzed with a two-sided Student's t-test or one-way ANOVA.

Fig 3: Liver-specific Dusp9 knockdown exacerbates glycolipid metabolism and abolishes caffeine-mediated multiple amelioration in MASH mice. (A) The flow chart shows administration modeling, tail vein injection of AAV, administration, and sacrifice time point (n = 8, per group). (B–C) Body weight and liver weight. (D) Representative images showed Oil red O, H&E, and PSR staining from indicated groups; PSR quantification is analyzed in (M). (E) Hepatic TG contents (mg/g) in each group. (F–H) Fasting plasma glucose, fasting insulin, and insulin resistance index (HOMA-IR) in each group. (I–J) Serum ALT and AST activities. (K) Hepatic MASH score (MASH activity score) in each group. (L) Hydroxyproline content of liver tissue in each group (n = 8, per group). Values represent means ± SD.∗p < 0.05; ∗∗p < 0.01; n. s. = no significance. The data were analyzed with a two-sided Student's t-test or one-way ANOVA.

Fig 4: Dusp9 silencing impairs the effects of caffeine on lipid accumulation, inflammation, and MAPK signaling in FFA-induced THLE-2 cells. (A) Flow chart illustrating the experimental design for Dusp9 silencing by lentiviral transfection, followed by FFAs and CAFF incubation. The experiment was repeated three times. (B, C) Representative pictures showed Oil red O staining and semi-quantitative assessment of Oil red O content. Scale bar, 20 μm (n = 3 independent experiments per group). (D) Levels of TGs in THLE-2 cells in the indicated groups. (E) Dusp9 downstream signaling molecules including p-ASK1, ASK1, p-p38, p38, p-JNK, and JNK were tested with Western blot in FFA-induced THLE-2 cells. Relative levels of p-ASK1, p-p38, and p-JNK were analyzed in THLE-2 cells. (F) Quantitative PCR was performed to determine the hepatic mRNA levels of genes related to lipid metabolism (Srebp1, Fasn, Scd1, and Acc1) in the indicated groups. (G) Quantitative PCR was performed to determine the hepatic mRNA levels of genes related to inflammation (Tnf-α, Il-1β, Ccl2, and Cxcl2) in the indicated groups. Experiments were performed three times independently, and values represent means ± SD.∗, compared with Con p < 0.05; ∗∗, compared with Con p < 0.01; #, compared with HFHC p < 0.05; ##, compared with HFHC p < 0.01. The data were analyzed with a two-sided Student's t-test or one-way ANOVA.

Fig 5: Caffeine treatment inhibits MAPK signaling in vivo and in vitro MASH models. (A) Dusp9 downstream signaling kinases including hepatic p-ASK1, ASK1, p-p38, p38, p-JNK, and JNK, were detected with Western blot in MASH mice. (B) Schematic illustrating CAFF rescues the expression of Dusp9 attenuated by the HFHC diet and further blocks the phosphorylation of MAPK signaling. (C) Relative levels of p-ASK1, p-p38, and p-JNK were analyzed. (D) MAPK signaling molecules including p-ASK1, ASK1, p-p38, p38, p-JNK, and JNK were tested with Western blot in FFA-induced THLE-2 cells. (E) Relative levels of p-ASK1, p-p38, and p-JNK were analyzed in THLE-2 cells. Experiments were performed three times independently, and values represent means ± SD.∗, compared with Con p < 0.05; ∗∗, compared with Con p < 0.01; #, compared with HFHC p < 0.05; ##, compared with HFHC p < 0.01. The data were analyzed with a two-sided Student's t-test or one-way ANOVA.