A small-molecule TNIK inhibitor targets fibrosis in preclinical and clinical models

Abstract

Idiopathic pulmonary fibrosis (IPF) is an aggressive interstitial lung disease with a high mortality rate. Putative drug targets in IPF have failed to translate into effective therapies at the clinical level. We identify TRAF2- and NCK-interacting kinase (TNIK) as an anti-fibrotic target using a predictive artificial intelligence (AI) approach. Using AI-driven methodology, we generated INS018_055, a small-molecule TNIK inhibitor, which exhibits desirable drug-like properties and anti-fibrotic activity across different organs in vivo through oral, inhaled or topical administration. INS018_055 possesses anti-inflammatory effects in addition to its anti-fibrotic profile, validated in multiple in vivo studies. Its safety and tolerability as well as pharmacokinetics were validated in a randomized, double-blinded, placebo-controlled phase I clinical trial (NCT05154240) involving 78 healthy participants. A separate phase I trial in China, CTR20221542, also demonstrated comparable safety and pharmacokinetic profiles. This work was completed in roughly 18 months from target discovery to preclinical candidate nomination and demonstrates the capabilities of our generative AI-driven drug-discovery pipeline.

Fig. 1. AI-augmented pipeline for target discovery.

a, The PandaOmics target-discovery platform was applied to lung and kidney fibrosis datasets to generate target hypotheses, followed by the Chemistry42 platform application to generate small-molecule leads targeting TNIK. DE, differential gene; GWAS, genome-wide association study; hetero, heterogeneous; siRNA, small interfering RNA; IP, intellectual property. b, TNIK was scored the number 1 candidate using protein and receptor kinase PandaOmics settings based on relatively high values of network neighbors, mutated submodules, causal inference, pathways, interactome community, expression, heterogeneous graph walk and matrix factorization scores. GAK, cyclin G-associated kinase; MST1R, macrophage-stimulating 1 receptor; PKMYT1, protein kinase, membrane-associated tyrosine–threonine 1; STK26, serine–threonine kinase 26; Tchem, genes whose products can be targeted with small molecules better than the following bioactivity cutoff values: 30 nM for kinases, 100 nM for GPCRs and nuclear receptors, 10 μM for ion channels, and 1 μM for other target classes; Tbio, genes annotated with a Gene Ontology Molecular Function or Biological Process with an Experimental Evidence code, or targets with confirmed OMIM phenotype(s), or do not satisfy the Tdark criteria. c, TNIK is a member of the serine–threonine kinase STE20 family. This family does not contain any major targets of anti-fibrotic medications including nintedanib, the most prominent kinase inhibitor used for IPF treatment. This illustrates the relative novelty of the target. cAMP, cyclic AMP; FLT, FMS-related receptor tyrosine kinase 1; GSK, glycogen synthase kinase; KDR, kinase insert domain receptor; MAP, mitogen-activated protein; PKA, protein kinase A; ILD, interstial lung disease; K_d, dissociation constant.

Fig. 2. Architectural superposition of TNIK inhibitor structures with the predicted INS018_055-binding mode and effects on TGF-β-induced EMT and FMT cellular programs.

a, Crystal structure of the NCB-0846 (cyan)-bound TNIK kinase domain (PDB 5D7A) aligned with the predicted binding mode of INS018_055 (green). b, Crystal structure of the compound 9 (cyan)-bound TNIK kinase domain (PDB 5AX9) aligned with the predicted binding mode of INS018_055 (green). Differences in Met gatekeeper orientations between inhibitors bound to the TNIK kinase domain are depicted, with Met side chain color corresponding to ligand color (cyan, green). The hinge region is shaded orange. Key ligand interactions are marked with dashed lines in the ligand color (cyan, green). The conservative E91–K72 salt bridge is shown as a yellow dashed line. Ligands and key pocket residues are represented as sticks. c, Inhibitory effect of INS018_055 on TGF-β-induced α-SMA protein expression in MRC-5 cells. n = 3, mean ± s.d. GAPDH, glyceraldehyde 3-phosphate dehydrogenase. d, Top, representative INS018_055-inhibitory effect (green) on FMT in primary human lung fibroblasts by measuring α-SMA. Bottom, INS018_055-inhibitory effect (green) on EMT in primary human bronchial epithelial cells (HBECs) by measuring fibronectin. Percent remaining cells (black) reflects nuclear count, which is a measurement of cell percentage without nuclear loss (n = 2 experimental replicates). PIN, percentage inhibition. e, Representative western blots showing changes in E-cadherin, N-cadherin, SMAD2/SMAD3, phospho (p)-SMAD2/SMAD3 and β-catenin in different A549 cell fractions following TGF-β stimulation and treatment with INS018_055. HDAC2, histone deacetylase 2. f, Representative western blot showing the inhibitory effect of INS018_055 on phosphorylated and total NF-κB p65 in TGF-β- and TNF-α-stimulated A549 cells (n = 3 biological replicates). g, Representative western blots showing changes in E-cadherin, N-cadherin, fibronectin, phospho-FAK and phospho-SMAD2 induced by treatment of A549 cells with TNIK shRNA (shTNIK) (shTNIK-4). shCtrl, control shRNA. h, Significantly upregulated pathways induced by TGF-β treatment and restoration by INS018_055 (055) or shTNIK-1. Gene ontology (GO) enrichment (left, middle) and KEGG analysis (right). n = 3 in each condition. Enrichment analysis was performed using the gseapy.enrichr Python package. P values were computed using Fisher’s exact test (one-tailed hypergeometric test). Adjusted P values (q values) were calculated using the Benjamini–Hochberg method for correction for multiple-hypothesis testing. ECM, extracellular matrix; NS, not significant. i, Scheme showing TNIK function in regulating TGF-β, WNT, YAP–TAZ and TNF-α pathways as identified by in vitro perturbation experiments. Source data

Fig. 3. In vivo effects of INS018_055 treatment in mouse models of lung diseases.

a, Study design of the bleomycin-induced lung fibrosis model in C57BL/6 male mice (n = 10 per group). b, Lung function on day 21 measured by Penh (mean ± s.d.; group (G)1, n = 10; groups 2–5, n = 13) (group 1 (vehicle) compared to groups 2–5, P < 0.0001; group 5 (nintedanib) compared to groups 2, 3 and 4, with P = 0.9986, 0.9426 and 0.464, respectively). c, Representative measurements of mice treated with INS018_055 (3, 10 or 30 mg per kg, BID) and nintedanib (60 mg per kg, QD) from a showing Masson’s trichrome staining, modified Ashcroft scores and immunohistochemistry (IHC) of collagen 1 and ɑ-SMA. n = 10 per group (mean ± s.d.) (Masson’s trichrome staining: group 1 (vehicle) compared to groups 2, 3, 4 and 5, with P = 0.0233, 0.0004, 0.0004 and <0.0001, respectively; group 5 compared to groups 2, 3 and 4, with P = 0.0799, 0.8126 and 0.8289, respectively. Modified Ashcroft: group 1 compared to groups 2, 3, 4 and 5 with P = 0.1021, 0.0071, 0.004 and 0.0001, respectively; group 5 compared to groups 2, 3 and 4 with P = 0.1718, 0.791 and 0.8933, respectively. ɑ-SMA: group 1 compared to groups 2–5 with P ≤ 0.0001; group 5 compared to groups 2, 3 and 4 with P = 0.9689, 0.6125 and >0.9999, respectively. Collagen I: group 1 compared to groups 2–5 with P ≤ 0.0001; group 5 compared to groups 2, 3 and 4, with P = 0.9985, 0.9986 and >0.9999, respectively). Ordinary one-way ANOVA and post hoc Šídák’s multiple-comparison test were used to assess statistical significance. d, Study of INS018_055 in the LPS-induced acute lung injury model in C57BL/6 male mice. n = 8 per group. e, Lymphocyte cell counts from d. n = 8 per group (mean ± s.d.) (group 2 (vehicle) compared to groups 3, 4, 5, 6 and 7, P = 0.9898, 0.7465, 0.4720, 0.0035 and 0.0698, respectively). Dex, dexamethasone. f, Measurements of IL-6, IL-7, TNF-α, IL-1β and IL-4 in BALF by enzyme-linked immunosorbent assay (ELISA). n = 8 per group (mean ± s.d.) (group 2 (vehicle) compared for IL-6, IL-7 and IL-1β to groups 3–7 with P < 0.0001; for TNF-α, group 3, P = 0.0023 and groups 4–7, P < 0.0001; for IL-4, group 3, P = 0.004 and groups 4–7, P < 0.0001). Ordinary one-way ANOVA and post hoc Šídák’s multiple-comparison test was used to assess statistical significance (exact P values are provided except for ***P < 0.001). Source data

Fig. 4. In vivo effects of INS018_055 treatment by inhalation in rat models of lung fibrosis.

a, Study design of the bleomycin-induced lung fibrosis model in male Sprague Dawley rats (n = 12 for INS018_055 groups with three animals killed on day 28 for exposure analysis for plasma and lung, n = 12 for other groups). b, Lung function on day 29 measured by FVC, airway resistance (RL) and pulmonary compliance (Cdyn) (mean ± s.d.). Statistical analysis was performed using uncorrected two-sided Fisher’s least-significant difference test as post hoc analysis after ANOVA analysis. Compared with the model control group, P values of the sham group, INS018_055 groups (0.1 mg ml⁻¹, 0.3 mg ml⁻¹, 1 mg ml⁻¹, 6 mg ml⁻¹) and the pirfenidone group (350 mg per kg) are all <0.001 for FVC; <0.001, <0.001, 0.0062, <0.001, 0.0336 and <0.001, respectively, for airway resistance; <0.001, <0.001, 0.002, <0.001, <0.001 and <0.001, respectively, for pulmonary compliance; n = 9 for INS018_055 groups, n = 12 for other groups (exact P values are provided except for ***P < 0.001). c, Quantitation for INS018_055 (0.1, 0.3, 1 or 6 mg ml⁻¹, QD, inhalation) and pirfenidone (350 mg per kg, QD, oral) groups in the bleomycin-induced lung fibrosis model shown as modified Ashcroft score, Masson’s trichrome staining and hematoxylin and eosin (H&E) staining (mean ± s.d.). Statistical analysis was performed using Dunn’s multiple-comparison tests after the Kruskal–Wallis test (compared with the model control group, P values of the sham group, the INS018_055 (1 mg ml⁻¹) group, the INS018_055 (6 mg ml⁻¹) group and the pirfenidone (350 mg per kg) group are <0.0001, 0.0050, 0.0223 and 0.0066 for modified Ashcroft score, respectively; <0.0001, 0.0007, 0.0040 and 0.0028 for Masson’s trichrome staining, respectively; and <0.0001, 0.0008, 0.0250 and 0.0065 for H&E staining, respectively; n = 9 for INS018_055 groups, n = 12 for other groups (exact P values are provided except for ***P < 0.001). Source data

Fig. 5. In vitro and in vivo studies on the effect of INS018_055 on kidney cells and the mouse model of kidney fibrosis.

a, Left, representative western blot showing protein level changes of α-SMA with TGF-β and INS018_055 treatment. Right, graphical representation and IC₅₀ determination of the effect of α-SMA expression in HK-2 cells. This represents a single independent experiment: no statistical conclusions were made. b, Representative images of Sirius red staining and quantitation of positive area (bottom right). c, Quantification of hydroxyproline content in the kidney. d, Scoring of IHC staining of collagen type 1 (for b–d, n = 8, mean ± s.d.; statistical analysis was performed using two-sided Bonferroni multiple-comparison test (using Prism 6 software) to compare between the vehicle group and other treatment groups for b–d). P values < 0.05 were considered statistically significant. Compared with the vehicle group, P values of the sham control group, INS018_055 groups (3, 10 and 30 mg per kg, BID) and the SB525334 group (100 mg per kg, QD) are all <0.0001 for Sirius red-positive area; <0.0001, >0.9999, 0.0074, <0.0001 and 0.003, respectively, for kidney hydroxyproline content; and <0.0006, 0.0001, 0.0001, 0.0028 and 0.0001, respectively, for collagen type 1 score (exact P values are provided except for ***P < 0.001). Source data

Fig. 6. Pharmacokinetic analysis in the clinical phase I trial.

a,b, Plasma concentrations of INS018_055 versus time in SAD and MAD of the phase I study performed in New Zealand (a) (mean ± s.d., n = 6 per group) and in SAD and MAD of the phase I study performed in China (b) (mean ± s.d., n = 6 per group). Source data

Extended Data Fig. 1. Target ID Approach.

(a) Mutated sub-modules score scheme. The significance of gene’s involvement can be measured gene-wise. Colored nodes are genes associated with the disease of interest genes according to OMIM (green), ClinVar (orange), Open Targets (blue). (b) Causal inference score scheme. The significance of gene’s involvement can be measured TF-wise. (c) Expression score scheme. The Expression score is combination of each gene fold change amplitude, significance, and basal expression. (d) HeroWalk score scheme. HeroWalk is a guided random walk-based approach that is applied to a heterogeneous graph. The model learns node representations and then finds gene nodes close to the reference disease node. First, the ‘walks’ are sampled with a predefined meta-path, that is fixed sequence of node types in a walk, for example ‘gene’-‘disease’-‘gene.’ The node degree controls the probability of transition between the nodes while sampling. Following that, the SkipGram model learns the representation of each node based on the resulting corpus of walks. The cosine similarity between the specific disease and all genes produces a ranked list of genes. The top genes from this list are predicted to be promising target hypotheses. (e) Matrix factorization scheme. Matrix Factorization is a collaborative filtering algorithm widely used in recommender systems. The algorithm decomposes a sparse matrix derived from a gene-disease interaction graph into two lower-dimensionality matrices that consist of latent factors for genes and diseases. The algorithm uses graph regularization based on a fast kNN search to account for intraclass similarity between similar nodes. Recomputing the original interaction matrix from latent factors provides the scores for unobserved interactions; thus, gene ranking is obtained. (f) Performance of the models are evaluated using two metrics ELFC and HGPV. Time Machine approach was applied to demonstrate the ability of the models to predict truly novel target hypotheses.

Extended Data Fig. 2. PandaOmics scores transparency and in-silico validation of TNIK.

(a) High values of the graph-based scores can be explained by the significantly perturbed genes observed in the TNIK interactome community neighborhood formed by proteins implicated in IPF and other fibrotic diseases. (b) Causal inference score relies on transcriptional factor inference. Majority of the genes regulated by the transcriptional factors are associated with IPF according to OpenTargets and GWAS studies. (c) TNIK regulates major signaling pathways known to be causal for the development of fibrotic conditions. (d) Clustering of cells derived from single cell RNA-seq data for IPF and healthy lung tissue. (e) Visualization of TNIK gene-weighted density in myofibroblasts, cytotoxic T cells and club cells using Nebulosa package. (f) Single cell gene expression profiles of IPF lung tissue are presented in boxplots for various unique cell types: Non-classical Monocytes (n = 1058, 1931), ATII High-Surfactuns cells (n = 2655, 496), Low-info Multiplet Macrophages (n = 1000, 2765), Alveolar Macrophage (n = 23905, 27403), Macrophage (n = 40747, 60483), NK cells (n = 2744, 4007), Cytotoxic T cells (n = 4061, 6296), Myofibroblasts (n = 204, 2886), and Club cells (n = 226, 1855) for the control and IPF groups, respectively. These profiles show high expression of TNIK in the key cell types responsible for disease progression: myofibroblasts, cytotoxic T cells, and club cells. The differential expression between IPF and controls for each unique cell type was calculated using the tl.rank_genes_groups function of scanpy package (method = 'Wilcoxon', two-sided), and the resulting log-fold changes (with p-values adjusted using the Benjamini-Hochberg method and threshold = 0.05, FDR) between the IPF and control groups for each cell type of TNIK were plotted on a heatmap. The center lines in the boxplots indicate the medians, while the box limits represent the 25th and 75th percentiles. (g) Simulated knockout profile of TNIK in myofibroblasts derived from IPF patients by scTenifoldKnk virtual knockout tool confirms the importance of Hippo and YAP/TAZ signaling in the TNIK-mediated regulation of the major IPF-related pathways.

Extended Data Fig. 3. INS018_055 target affinity and cell viability analysis.

Surface plasmon resonance (SPR) assay to measure the binding kinetics of (a)INS018_055, (b) NCB-0846 and (c) KY-05009 to His tagged TNIK (9-315). (d) Cell viability analysis performed on MRC-5 cells (left), A549 cells (middle) and HK-2 cells (right). Cells were seeded at 8,000, 5,000 and 8,000 cells/well in 96-well plate, respectively and incubated with INS018_055 for 72 hours. Cell viability was measured by Cell Titer Glo. All experiments were performed in duplicate wells. CC₅₀ was calculated with the data of cell viability (%). To determine the CC₅₀ value, the data was fitted using an equation for a sigmoidal dose response (variable slope), as provided by GraphPad™ Prism software. The equation was shown as: Y = Bottom + (Top-Bottom)/(1+10^((LogIC₅₀-X)*HillSlope)). Source data

Extended Data Fig. 4. IPF donor fibroblast and epithelial cell imaging for FMT and EMT assays.

(a) Representative images of fibroblasts from a donor IPF patient stained with DAPI and alpha-smooth muscle actin (α-SMA) antibody. Cells were pre-treated with TGFβ to induce mesenchymal transition after which cells were treated with the indicated concentrations of INS018_055. Automated quantitation of α-SMA fluorescence intensity and cell density was made following INS018_055 treatment. (b) Representative images of fibroblasts from a donor IPF patient stained with DAPI and fibronectin (FN1) antibody. Cells were pre-treated with TGFβ to induce mesenchymal transition after which cells were treated with the indicated concentrations of INS018_055. Automated quantitation of FN1 fluorescence intensity and cell density was made following INS018_055 treatment. Fibroblasts and epithelial cells were isolated from 3 IPF donor tissue samples and 3 healthy donor samples. The experiment was carried out with all 6 independent samples in a single experiment by Charles River’s high throughput imaging pipeline. Source data is provided.

Extended Data Fig. 5. Effect of IN18_055 on TGF-β-induced, or combination of TGF-β and TNF-α induced EMT/FMT cellular programs.

(a) Representative pictures of morphology change of A549 cells. (b) Bar graphs (mean ± SD) of inhibition effect of INS018_055 on TGF-β induced changes of E-cadherin, N-cadherin, phospho-smad2, smad2/3 and beta-catenin levels in A549 cells; Bar graphs (mean ± SD) of inhibition effect of INS018_055 on TGF-β and TGF-α-induced phospho-NF-kb p65 and its total protein changes in A549 cells. Images of blots refers to Fig. 2e, f. Individual values are shown in dots. P values are analyzed by Welch’s t-test (two-sided) among 3 independent experiments. p values < 0.05 were considered statistically significant (*: significant compared to group treated with TGF-β but without INS018_055; #: significant compared to group treated with TGF-β and TGF-α but without INS018_055). Compared to TGF-β control group, group (no TGF-β), groups with INS018_055 (0.1, 0.3, 1 and 3 μM) showed p values (E-cadherin) of 0.0302, 0.024, 0.0288, 0.0127 and 0.031; p values (N-cadherin) of 0.0054, 0.0469, 0.0008, 0.0007 and 0.0012; p values (p-smad2/3) of 0.004, 0.0012, 0.0018, 0.001 and 0.0008; p values (chromatin β-catenin) of 0.5714, 0.0436, 0.0043 and 0.0028. Compared to TGF-β and TGF-α control group, groups with INS018_055 (0.1, 0.3, 1 and 3 μM) showed p values (p-p65) of 0.0383, 0.0675, 0.0341 and 0.0242; p values (p65) of 0.0503, 0.1243, 0.0492 and 0.0151. Source data

Extended Data Fig. 6. Effect of TNIK deletion on TGF-β signaling in A549 cells.

(a) Representative pictures of morphology change of A549 cells. Single independent experiment- no statistical comparisons were made. (b) Western Blot images of inhibition of TGF-β induced changes of E-cadherin, N-cadherin, fibronectin, and phospho-smad2 by shTNIK (shTNIK-1). Single independent experiment- no statistical comparisons were made. (c) Bar graphs of Fig. 2g, the inhibition of TGF-β induced changes of E-cadherin, N-cadherin, fibronectin, phospho-FAK and phospho-smad2 by shTNIK (shTNIK-4). (d) Heatmap of log2(fold changes) of gene expression change of Hippo signaling related proteins. Source data

Extended Data Fig. 7. INS018_055 inhibits inflammation in a bleomycin-induced lung (BLM) fibrosis mouse model.

(a) Images and quantitation of inflammatory area under the treatment of INS018_055 (3, 10, or 30 mg/kg, BID) or nintedanib (60 mg/kg, BID) in bleomycin-induced lung fibrosis model by H&E staining. n = 10 per group. (mean ± SD) Individual values are shown in dots. Comparing G1 (model vehicle group) to G2, 3, 4, 5, yielded p = 0.0092, 0.0003, <0.0001, <0.0001, respectively; Comparing G5 (nintendanib) to G2, 3, 4: p = 0.0789, 0.6504, 0.9063, respectively) (b) cell counts of neutrophils, monocytes, and levels of IL-6, IL-1β in BALF of the animals in different treatment groups. n = 10 per group. (mean ± SD) Neutrophils: Comparing G1 (model vehicle) with G2, 3, 4, 5, yielded p = 0.1391, 0.0094, 0.0105, 0.0082, respectively; Comparing G5 (nintedanib) with G2, 3, 4 yielded p = 0.9073, >0.999, >0.999, respectively. Monocytes: Comparing G1 (model vehicle) with G2, 3, 4, 5, resulted in p = 0.2069, 0.01, 0.0031, and 0.0825, respectively; Comparing G5 (nintedanib) with G2, 3, 4, resulted in p = 0.9997, 0.9819, 0.8605, respectively; IL-6: Comparing G1 (model vehicle) with G2, 3, 4, 5, yielded p = <0.0001, <0.0001, <0.0001, <0.0001, respectively; Comparing G5 (nintedanib) with G2, 3, 4 resulted in p = 0.4435, <0.0001, 0.0403, respectively; IL-1β: Comparing G1 (model vehicle) with G2, 3, 4, 5, resulted in p = <0.0001, <0.0001, <0.0001, <0.0001, respectively; compared with G5: G2, 3, 4: p = 0.5115, <0.0001, <0.0001, respectively. All statistical analyses reported here are Ordinary One-way-anova with post-hoc Šídák’s multiple comparisons testing. (asterisks were used instead of an exact value only for *** when p < 0.0001). Source data

Extended Data Fig. 8. In vivo study on the effect of combination therapy of INS018_055 and sub-optimal dose Pirfenidone on mouse model of bleomycin induced lung fibrosis.

(a) Study design of bleomycin-induced lung fibrosis model in C57BL/6 male mice (n = 10 per group). (b) lung function on day 21 measured by Penh (mean ± SD) (n = 10 for G1, and n = 18 for G2-4, with data of animals for PK sampling included). Comparing G1 (model vehicle) with G2, G3, G4, yielded p value = 0.0984, 0.9896, and 0.0061, respectively; Comparing G4 (combo group) with G2, G3 yielded p values = 0.7383 and 0.0062, respectively. (c) Quantification of INS018_055 in bleomycin-induced lung fibrosis model Masson’s trichrome staining for fibrosis area, Modified Ashcroft Score, and IHC staining of Collagen 1 and ɑ-SMA (mean ± SD, n = 10). Individual values are shown in dots. MT staining: Comparing G1 (model vehicle) with G2, G3, G4, resulted in p value = 0.0078, 0.149, and 0.0002, respectively; Comparing G4 (combo group) with G2, G3 yielded p values = 0.6478 and 0.0774, respectively. Modified Ashcroft Score: Comparing G1 (model vehicle) with G2, G3, G yielded p value = 0.5428, 0.998, and 0.0112, respectively; Comparing G4 (combo group) with G2, G3 resulted in p values = 0.3414 and 0.0298, respectively. α-SMA: Comparing G1 (model vehicle) with G2, G3, G4 yielded p value = <0.0001, <0.0001, and <0.0001, respectively; Comparing G4 (combo group) with G2, G3 resulted in p values = 0.3935 and 0.7877, respectively. Collagen I: Comparing G1 (model vehicle) with G2, G3, G4 yielded p value = 0.8867, 0.9992, and 0.2988, respectively; Comparing G4 (combo group) with G2, G3 resulted in p values = 0.8884 and 0.4865, respectively. (d) Representative histopathology images including M&T Staining, α-SMA IHC Staining and Collagen I IHC Staining. All statistical analyses reported here are Ordinary One-way-anova with post-hoc Šídák’s multiple comparisons testing. (asterisks were used instead of an exact value only for *** when p < 0.0001). Source data

Extended Data Fig. 9. In vivo effects of INS018_055 treatment in combination with maximum therapeutic dose of Pirfenidone in mouse models of lung fibrosis.

(a) study design of bleomycin-induced lung fibrosis model in C57BL/6 male mice. (b) Left, body weight change during study process; right: Disease-free curve based on clinical observations (Group,1,2, n = 10, Group 3-7, n = 16) (mean ± SD). (c) Lung function measured by Penh (Group,1,2, n = 10, Group 3-7, n = 16) (mean ± SD). Comparing G2 (model vehicle) with G1, p = 0.0005; G3-G7, p = 0.4629, 0.5745, 0.9888, 0.8805 and 0.1635, respectively. All statistical analyses reported here are Ordinary One-way-anova with post-hoc Šídák’s multiple comparisons testing. Source data

Extended Data Fig. 10. In vitro and in vivo study of INS018_055 on skin fibrosis.

(a) IC₅₀ values of inhibition effect of INS018_055 on levels of α-SMA expression, and levels of fibronectin and procollagen I in supernatant in cultured NHDF cells. Left: Cells were treated with INS018_055 and TGF-β (0.1 ng/ml) for 72 hours. The fluorescence intensity of α-SMA normalized to the total number of nuclei identified by staining with Hoechst 33258. Right and below: Cells were treated with INS018_055 and TGF-β (10 ng/ml) for 72 hours. Fibronectin and procollagen I contents were measured in the culture supernatants. Relative inhibition (%) = (Mean Stimulated control - Value)/ (Mean Stimulated control –Mean Non-stimulated control) X 100. All experimental conditions were performed in n = 5. (mean ± SEM). (b) INS018_055 inhibited collagen level in bleomycin induced skin-thickening rat model. Sprague Dawley rats were injected with 0.1 mL 1 mg/mL bleomycin subcutaneously into two sites on the shaved regions, once daily for 4 weeks. Treatments were administered with vehicle, Rapamycin (1.5 mg/kg, intraperitoneal) or INS018_055 (0.05%, 0.15% or 0.45%, topical), 30 minutes before administration of bleomycin, everyday for four weeks. Hydroxyproline and collagen content was measured (mean ± SD), (n = 5). The concentrations of hydroxyproline and collagen were normalized to total protein content using the Bradford method. Statistical analysis was performed using one-way ANOVA, then Dunnett’s multiple comparison test was performed. For hydroxyproline, compared with vehicle control group (Group 2), p values of Group naive control, Group of rapamycin, Groups of INS018_055 [0.05%, 0.15% and 0.45%], are 0.0165, 0.0085, 0.0015, 0.0261, and 0.0033, respectively. For total collagen, p values of Group naive control, Group of rapamycin, Groups of INS018_055 [0.05%, 0.15% and 0.45%], are 0.0025, 0.0003, 0.0001, 0.0004, and 0.0012, respectively. p values < 0.05 were considered statistically significant. Source data