Human SHP2 PTP domain (A237–I529) expression constructs were engineered by cloning PTPN11 (NP_002825.3) into the pET30 plasmid with a His-tag to support further protein purification. The K366A/Q510A double mutation was introduced into the wild-type template using a site-directed mutagenesis kit (E0554S; New England Biolabs). The expression constructs were sequenced and transformed into BL21 (DE3)-competent cells. E. coli cells were grown at 37°C in an LB culture medium in the presence of 100 μM mL⁻¹ kanamycin until the OD₆₀₀ reached 1.0. The medium was then cooled to 18°C and 1 mM isopropyl ß-D-1-thiogalactopyranoside (IPTG) was added to induce protein expression overnight.

Cell pellets were collected and suspended in a buffer of 50 mM Tris-HCl (Ph = 8.5), 150 mM NaCl supplemented with 10% glycerol, and lysed by homogeniser (JN-BIO, China). After centrifugation, the collected supernatants were diluted with a Tris-HCl buffer (pH = 8.5) containing 50 mM imidazole and 150 mM NaCl and then loaded onto a Ni-NTA column (#88221, Thermo). The bound SHP2 protein was eluted with a buffer containing 150 and 300 mM imidazole. The pooled protein was loaded onto a HiTrap Q column after dilution in 25 mM Tris-HCl buffer. After elution using a linear gradient of NaCl (0–0.5 M), peak fractions containing the SHP2 protein were concentrated and applied to a gel filtration Superdex 200 column (GE Healthcare). Purified SHP2-PTP proteins were flash-frozen and stored. Similar expression and purification protocols were used for the expression and purification of the full-length SHP2 and SHP2-PTP ^K366A/Q510A double mutants.

A surrogate phosphatase substrate, 6,8-difluoro-4-methylumbelliferyl phosphate (DiFMUP, Thermo Fisher, #D22065), was used to monitor the catalytic activity of SHP2 in a prompt fluorescence assay format. Specifically, the DiFMUP phosphatase assay was performed at 20°C in a 384-well Griener CELLSTAR black polystyrene plate in the buffer condition of 60 mM HEPES (pH7.5), 75 mM NaCl, 75 mM KCl, 1 mM EDTA, 0.05% Tween-20, 5 mM dithiothreitol (DTT). The final reaction volume was 25 μL. NP compound libraries were obtained from TargetMol at a stock concentration of 10 mM, dissolved in DMSO. The purity of the active compound LIN used in this study is 96.27% as tested by HPLC analysis. The HPLC traces is shown in Supporting Information. We co-incubated 0.5 nM of SHP2-PTP enzyme with tested NPs for 30 min at 20°C, followed by the addition of DiFMUP into the reaction. After 30 min, 5 μL of bpV(phen) solution was added to quench the dephosphorylation reaction. The signals were read on a plate reader (SpectraMax iD5, Molecular Devices). The excitation and emission wavelengths were 340 and 450 nm, respectively. The inhibitor dose-response curves were evaluated using a normalised IC₅₀ regression curve fitting with control-based normalisation. For the full-length SHP2 inhibition assay, 0.5 μM of a bisphosphorylated pIRS-1 peptide (ChinaPeptides Ltd. Shanghai) was added for partial SHP2 activation.

The BLI assay was performed using a FortéBio Octet 96 system. To label the biotin group on SHP2 proteins, NHS-PEG12-Biotin (Thermo Fisher, 21312) was added to the purified SHP2-PTP solution (Thermo Fisher, 21312) at 1.5:1 ratio for 2 h at 4°C. The reaction mixture was then loaded onto a column (GENEMORE, #G-MM-IGT) to remove excess free biotin. The biotin-labelled-SHP2-PTP protein was loaded onto super streptavidin (SSA) sensors (#18-5057, Octet) in a buffer containing 200 μL 25 mM HEPES, 150 mM NaCl, and 0.02% (v/v) Tween-20. After equilibrium, the kinetics of the LIN and SHP2-PTP association were analysed by soaking the SSA sensors in a compound solution with various LIN concentrations for 300 s (6.25, 12.5, 25, 50, 100, and 200 μM), followed by 300 s of dissociation in the same HEPES buffer. The equilibrium constant (KD) values were assessed using the FortéBio data analysis software by fitting the kinetic data using a 1:1 binding model.

The binding mode between SHP2 and LIN was analysed using the Glide module in the Schrödinger package. The SHP2 structure (PDB ID:4RDD) and LIN were processed using the Protein Preparation Wizard and LigPrep modules in the Schrödinger package (Fodor et al., 2018). Grids were generated for PTP binding sites using the Receptor Grid Generation module in the Schrödinger package. The Glide module of the Schrödinger package was used to generate the predicted binding positions between LIN and SHP2.

The LX-2 cell line was purchased from the Cell Bank Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). Cells were cultured in DMEM supplemented with 10% [v/v] foetal bovine serum, 2 mM glutamine, 100 U/ml penicillin, and 100 mg/mL streptomycin. SHP-2 and control siRNAs were obtained from Suzhou GenePharma Co., Ltd. (sense 5'-3': GAG​AGA​GGA​AAG​AGU​AAA​UTT; antisense 5'-3': AUU​UAC​UCU​UUC​CUC​UCU​CTT). Transfection was performed according to the manufacturer’s instructions. Subsequent real-time PCR and western blotting were performed to analyse transfection efficiency.

C57BL/6 mice (20 ± 2 g) were obtained from Hangzhou Medical College Experimental Animal Center (Hangzhou, China). The animals were maintained under controlled temperature (22°C ± 1°C), humidity (50%), and light (12 h light/12 h dark). The animals were fed a standard laboratory diet and provided with free access to tap water. All animals received humane care according to institutional animal care guidelines approved by the Experimental Animal Ethical Committee of Hangzhou Medical College. Forty mice were randomly divided into six groups: (1) vehicle control (n = 7); (2) CCl₄ model (n = 7); (3) CCl₄ + LIN (20 mg/kg) (n = 7); (4) CCl₄ + LIN (40 mg/kg) (n = 7); (5) LIN (40 mg/kg) (n = 7); and (6) CCl₄ + silymarin (SIL) (200 mg/kg) (n = 5). LIN and SIL were dissolved in 0.5% CMC-Na solution. Mice were administered LIN or SIL (intragastric administration, i.g.) once per day and CCl₄ (intraperitoneal injection, i.p.; mixed 1:3 in olive oil, 2 mL/kg) twice per week for 4 weeks. Mice in the vehicle control group received olive oil (i.p. twice a week, 4 weeks) and 0.5% CMC-Na (i.g. every day, 4 weeks). After treatment, the mice were sacrificed, and plasma and liver tissues were collected and further analysed.

Serum was collected from blood samples after centrifugation at 860 × g for 15 min. Serum ALT (Nanjing Jiancheng Bioengineering Institute, C010-2-1) and AST (Nanjing Jiancheng Bioengineering Institute, C009-2-1) activities were measured using kits, according to the manufacturer’s instructions. Liver hydroxyproline content was determined using the alkaline hydrolysis method, as described in the kits. Serum levels of laminin (LN) were measured using ELISA kits, according to the manufacturer’s instructions (Shanghai YANJIN Biological Technology Co., LTD, F12019).

Total RNA was extracted from liver tissues and LX-2 cells using the TRIzol reagent. cDNA was synthesised using a PrimeScript RT Master Mix kit. Real-time PCR was performed using SYBR Green premix according to the manufacturer’s instructions. Relative expression of target genes was normalised to actin, analysed by the delta-delta-CT method and given as a ratio compared with the vehicle control. Primer sequences used in this study are listed in Supplementary Material.

A sample of the liver was fixed in 10% phosphate-buffered saline (PBS)-formalin solution and embedded in paraffin. Samples were sectioned (5 μm) and stained with haematoxylin-eosin (H&E) for histological observation of liver injury and stained with Masson’s trichrome and Sirius red to observe collagen deposition in the liver.

The formalin-fixed and de-paraffinized liver sections (5 μm) were incubated with 5% bovine serum albumin to minimise non-specific binding, and then incubated with α-SMA antibody (Santa Cruz, SC-53142) in a humidified chamber overnight at 4°C. After washing thrice with PBS, the sections were incubated with FITC-conjugated secondary antibody (SH0058; Skyhobio) for 1 h. After washing thrice, the liver sections were incubated with DAPI for 10 min. Images were captured using an inverted microscope (IX81; Olympus, Japan).

Paraffin-embedded liver sections were deparaffinized in xylene, rehydrated in a gradient of ethanol to distilled water, and then quenched by 3% hydrogen peroxide and then incubated with 5% bovine serum albumin, followed by incubation with vimentin antibody (ABclonal, A19607) at 4°C overnight, and further detected by using DAKO EnVision detection kits (Agilent Technologies Co., Ltd. K346811). The sections were counterstained with haematoxylin. Images were taken using an inverted microscope, and vimentin-positive cells were counted manually using Image-Pro Plus 6 in three random fields per sample (each group contained three samples).

Liver proteins were isolated by using a lysis buffer containing 50 mM Tris-HCl (pH7.5), 150 mM NaCl, 1 mM EDTA, 20 mM NaF, 0.5% NP-40, 10% glycerol, 1 mM phenylmethylsulphonyl fluoride, 10 μg/mL aprotinin, 10 μg/mL leupeptin, and 10 μg/mL pepstatin. The supernatants were collected after centrifugation at 3,000 g for 10 min at 4°C, and the protein concentration of each sample was determined and normalised to the same protein concentration. Protein samples were separated by sodium dodecyl sulphate-polyacrylamide gel electrophoresis and transferred onto polyvinylidene fluoride membranes. The membranes were incubated with the indicated primary and secondary antibodies, and the proteins in the membranes were visualised using a chemiluminescence kit. TGF-^® antibody (CST, 3711), Smad 2/3 antibody (CST, 8685), p-Smad 2/3 antibody (CST, 8828), anti-α-SMA antibody (Santa Cruz, SC-53142), vimentin antibody (ABclonal, A19607), desmin antibody (ABclonal, A3736), GAPDH antibody (Beyotime, AF0006), and SHP2 antibody (CST, 3397). The grey densities of the protein bands were normalised using β-actin as an internal control, and the results were further normalised to the control.

Total RNA was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA, United States), according to the manufacturer’s protocol. RNA purity and quantification were analysed using a NanoDrop ND-1000 instrument (Wilmington, United States). RNA integrity was assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, United States). The libraries were constructed using the TruSeq Stranded mRNA LT Sample Prep Kit (Illumina, San Diego, CA, United States) according to the manufacturer’s instructions. Transcriptome sequencing and analysis were conducted by LC-Bio Technology Co. Ltd. (Hangzhou, China). The libraries were sequenced on an Illumina HiSeq X Ten platform, and 150 bp paired-end reads were generated. After generating the final transcriptome, the expression levels of all the transcripts were estimated. Differentially expressed mRNAs were selected using the following criteria: fold change >2 or fold change <0.5 and p-value <0.05. Hierarchical cluster analysis of differentially expressed genes (DEGs) was performed to determine the expression patterns of genes in different groups and samples. Gene Ontology (GO) enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses of DEGs were performed using the R software based on the hypergeometric distribution.

Quantitative data are presented as mean ± standard deviation (s.d.), as specified in the figure legends. Statistical tests were performed using GraphPad Prism 7.0. Two-sided Student’s t-tests were used to compare the means of the data between the two groups.

The recombinant human catalytic domain of the SHP2 protein (rhSHP2-PTP) was expressed in E. coli and purified using an affinity column, ion exchange column, and gel filtration. SDS-PAGE results showed the purified SHP2-PTP protein had a molecular weight of ∼37 kDa and a purity >95% (Figure 1A). DiFMUP was used as a fluorogenic phosphatase substrate. Enzyme titration experiments showed that rhSHP2-PTP had robust dephosphorylation activity toward the substrate DiFMUP (Figure 1B). Additionally, DiFMUP was tested at different concentrations to evaluate the catalytic kinetics of the dephosphorylation activity of SHP2-PTP (Figure 1C). Based on these linear enzyme and substrate concentration assay results, the SHP2-PTP and substrate DiFMUP concentrations were set to 0.5 nM and 10 μM, respectively, in the subsequent high-throughput screening (HTS) study (Figure 1D).

With the assay conditions established, an in-house focused NP library with 800 small molecules was screened in 384-well format. A total of eight initial hits exhibited ≥80% inhibition at a concentration of 20 μM, which were chosen for dose-response confirmation (Figure 1E). In a verification experiment of initial hits, furanogermacrane sesquiterpene linderalactone (LIN, at point 12I) showed the best inhibitory activity among the screened hits (Figure 1E).

Lindera aggregata (Sims) Kosterm is a common traditional herb that has multiple bioactivities (Ohno et al., 2005; Wang et al., 2022). A previous phytochemical study indicated that sesquiterpenoids may be the active components of L. aggregate (Wang et al., 2015). However, the bioactive molecules that contribute to its pharmacological activity and the underlying molecular mechanisms remain unknown. Our HTS results and the promising in vitro data from LIN encouraged us to further explore the SHP2 inhibition potency of sesquiterpenoid derivatives.

Linderane, isolinderalactone, and lindenenol, which have similar core scaffolds, were selected to test their SHP2-PTP inhibitory activity compared to that of LIN. Dose-response experiments of LIN showed an IC₅₀ value of 1.87 ± 0.85 μM against the SHP2 PTP domain (Figures 2A, B). Linderane, isolinderalactone, and lindenenol, which were all incorporated with methylfuran group, displayed dose-dependent suppression of SHP2-PTP activity with IC₅₀ values ranging from 2.1 to 12.6 μM, respectively (Figure 2B). Interestingly, compounds with an oxacycloundecin core (LIN and isolinderalactone) were more favourable for SHP2 inhibitory activity than linderane and lindenenol. Based on the in vitro enzyme assay results and its interesting novel structural skeleton, furanogermacrane sesquiterpene LIN was selected for further target and mechanistic studies.

As shown in Figure 2A, LIN showed a similar inhibitory effect on the enzyme activity of full-length SHP2, further demonstrating that LIN interacts with the PTP domain of SHP2. To generate more biochemical evidence to verify the direct binding with SHP2, the most active component, LIN, was selected to analyse the binding affinity with the PTP domain of SHP2 in vitro using a bio-layer interferometry (BLI) experiment. As shown in Figure 2C, LIN directly interacted with SHP2-PTP in a dose-dependent manner, with a K_D value of 665.9 μM.

To understand the underlying molecular basis of the LIN-SHP2 interaction, a molecular docking study was conducted using the PTP domain crystal structure of SHP2. The results revealed that LIN blocked the catalytic site of the PTP domain, thereby inhibiting the phosphorylated substrate from binding to SHP2. Predominantly, docking analysis indicated that residues Lys-366 and Gln-510 of the Q loop contributed to the LIN-SHP2 interaction by forming key hydrogen bonds with the two LIN furan fragments (Figure 2D). Additionally, the 3-methylfuran moiety of LIN forms hydrophobic interactions with the WPD loop of the SHP2 pocket, thus stabilising its binding to SHP2 and preventing the substrate from entering the active pocket.

To further validate the interacting residues and binding mode of LIN, K366A/Q510A double mutations were introduced into SHP2-PTP. As shown in Figure 2E, the K366A/Q510A mutation significantly eliminated the binding affinity between LIN and SHP2-PTP in the BLI experiment with a response value lower than 0.2 at the concentration of 200 μM. These experimental data facilitated the elucidation of the molecular mechanism by which LIN targets the key catalytic site of SHP2.

With promising in vitro results, the CCl₄-induced liver fibrosis animal model was used to evaluate the therapeutic effect of LIN in vivo. Silymarin (SIL), a well-known hepatoprotective drug, was used as a positive control at a dose of 200 mg/kg. As shown in Figures 3A, B, serum AST/ALT levels were significantly increased in the CCl₄ group, which is predictive of liver cell damage. Both ALT and AST levels were significantly reduced in the 20 and 40 mg/kg LIN-treated groups (p < 0.05), showing a protective effect similar to that of the 200 mg/kg SIL group. Histological analysis of the liver sections was performed to further evaluate the therapeutic effect of LIN on liver fibrosis. As shown in Figure 3C, CCl₄-induced liver damage, including hepatic infiltration of immune cells, swelling, and necrosis of hepatocytes, was ameliorated by LIN treatment. In contrast, mice that received LIN alone at 40 mg/kg had no obvious alteration in ALT/AST and pathological changes in mouse liver tissue, indicating low LIN toxicity (Figures 3A–C). These findings indicated that LIN effectively attenuated CCl₄-induced liver injury in mice.

Next, we investigated how LIN alleviated pathological changes in a mouse model of hepatic fibrosis. During liver fibrosis, the dynamic balance between synthesis and degradation of the extracellular matrix is disrupted, leading to ECM accumulation (including laminin, fibronectin, collagen I, III, and IV) and fibrosis formation (Biagini and Ballardini, 1989). Hydroxyproline is a characteristic fibrillar collagen component (Bradshaw et al., 2009). The antifibrotic effect of LIN was further evaluated by biochemical analysis of liver hydroxyproline and serum laminin levels. As shown in Figures 3D, E, hydroxyproline and laminin levels induced by CCl₄ were decreased by LIN in a dose-dependent manner. Additionally, the increase in mRNA levels of Col1a1 and Col3a1 induced by CCl₄ was significantly reversed after LIN treatment at a low dose (20 mg/kg) (Figure 3F).

Excessive accumulation of ECM is the major hepatic fibrosis pathogenesis, and collagen is considered to be the main ECM component. In the following study, two staining methods, Masson’s trichrome and Sirius red, were used to investigate the therapeutic effect of LIN on collagen deposition. Based on the results in Figures 3G, H, the CCl₄-treated group showed disrupted hepatic architecture, increased collagen content, and bridging fibrosis. In contrast, LIN treatment ameliorated the extent of collagen deposition in a dose-dependent manner, especially in the 40 mg/kg LIN treatment group, which displayed mild collagen deposition in the liver without bridging fibrosis.

To uncover the underlying mechanism of the antifibrotic effect of LIN, RNA sequencing analysis was performed using mouse liver tissue samples from the control, CCl₄ group, and CCl₄ + LIN groups. Among the 2163 DEGs, 716 genes were found to be upregulated, and 1447 genes were downregulated between the CCl₄ and CCl₄ + LIN groups. Additionally, 6480 differentially expressed transcripts with 3450 transcripts were upregulated and 3030 transcripts were downregulated. We used GO annotation analysis of DEGs to characterise their respective biological functions. Most biological-process-related genes between CCl₄ and CCl₄ + LIN groups were annotated with GO terms associated with “collagen−containing extracellular matrix”, “collagen fibril organization”, “extracellular space” and “extracellular matrix” (Figure 4A). Interestingly, based on KEGG annotation analysis, transforming growth factor beta (TGF-β) signalling emerged as the top 20 signalling pathways with statistical significance (Figure 4B), which plays a vital role in HSCs activation and ECM deposition that promote liver fibrosis.

We next investigated whether LIN prevented CCl₄-induced liver fibrosis by intervening in the canonical fibrogenic TGF-β signalling pathway. Western blot and real-time-PCR results showed that the amplified mRNA and protein expression of TGF-β was reduced in mice treated with LIN (40 mg/kg) (Figures 4C–E). As shown in Figure 4F, Western blot analysis confirmed that the phosphorylation level of Smad 2/3 increased in the livers of CCl₄-challenged mice, whereas LIN treatment significantly inhibited Smad2/3 phosphorylation in the liver. In addition, we checked the expression of SHP2 after LIN treatment (Figure 4D and Supplementary Figure S2). The result indicated that compound LIN do not affect the expression level of SHP2 protein, which demonstrated that SHP2 inhibitor LIN only inhibits the catalytic activity of SHP2. These results demonstrate that the SHP2 inhibitor, LIN, may alleviate liver fibrosis by interfering with the TGF-Smad3 pathway and HSCs activation.

Activated HSCs are recognised as the major matrix-producing cells during liver fibrosis progression. Several studies have indicated that SHP2 in HSCs contributes to process activation and fibrosis (Gao et al., 2020). Additionally, LIN may interfere with the TGF-Smad3 pathway based on RNA-sequencing analysis data; thus, we next investigated the effect of the identified SHP2 inhibitor LIN on HSC activation in vivo. Three well-known biomarkers, α-SMA, vimentin, and desmin, were used to investigate HSCs activation. data in Figures 5A–C showed that LIN (20 and 40 mg/kg) significantly reduced hepatic mRNA expression of acta2, Vim, and Des (p < 0.05, p < 0.01) induced by CCl₄ in vivo. Treatment with LIN at a 20 mg/kg dose also caused a significant reduction in the protein expression of hepatic α-SMA, Vim, and Des (p < 0.05) (Figures 5D, E). Immunofluorescence staining (red) further confirmed that α-SMA accumulation was largely inhibited by LIN (Figure 5F). Additionally, a similar antifibrotic trend was observed in vimentin-positive cells after LIN treatment (Figure 5G).

To confirm that the effect of LIN on HSC activation resulted from SHP2 inhibition, LX-2 cells were treated with 10 μM LIN for 24 h. As shown in Figure 6A, fibrogenic gene expression in LX-2 cells (e.g., ACTA2, VIM, TGF-β, FN1, DES, and COL1A1) was significantly decreased after LIN treatment, demonstrating the therapeutic potential of LIN to inhibit HSC activation.

To validate that the effect of LIN on LX-2 cell activation was mediated by inhibiting SHP2 activity, we further tested LIN in SHP2 knockdown cells. LX-2 cells were transfected with SHP2 siRNA and validated using western blotting and PCR (Figure 6B). As shown in Figure 6A, the mRNA expression levels of liver fibrosis markers were significantly suppressed compared to those in the control group, which is consistent with the positive role of SHP2 during fibrosis. Interestingly, SHP2-depleted LX-2 cells were much less sensitive to LIN treatment, suggesting that SHP2 may be the major target of LIN. Consistent with the in vivo result, LIN treatment led to significant reduced phosphorylation level of smad2/3 in LX-2 cells (Figure 6C).