Repositioning of ezetimibe for the treatment of idiopathic pulmonary fibrosis

Abstract

Background: We previously identified ezetimibe, an inhibitor of Niemann-Pick C1-like intracellular cholesterol transporter 1 and European Medicines Agency-approved lipid-lowering agent, as a potent autophagy activator. However, its efficacy against pulmonary fibrosis has not yet been evaluated. This study aimed to determine whether ezetimibe has therapeutic potential against idiopathic pulmonary fibrosis.

Methods: Primary lung fibroblasts isolated from both humans and mice were employed for mechanistic in vitro experiments. mRNA sequencing of human lung fibroblasts and gene set enrichment analysis were performed to explore the therapeutic mechanism of ezetimibe. A bleomycin-induced pulmonary fibrosis mouse model was used to examine in vivo efficacy of the drug. Tandem fluorescent-tagged microtubule-associated protein 1 light chain 3 transgenic mice were used to measure autophagic flux. Finally, the medical records of patients with idiopathic pulmonary fibrosis from three different hospitals were reviewed retrospectively, and analyses on survival and lung function were conducted to determine the benefits of ezetimibe.

Results: Ezetimibe inhibited myofibroblast differentiation by restoring the mechanistic target of rapamycin complex 1-autophagy axis with fine control of intracellular cholesterol distribution. Serum response factor, a potential autophagic substrate, was identified as a primary downstream effector in this process. Similarly, ezetimibe ameliorated bleomycin-induced pulmonary fibrosis in mice by inhibiting mechanistic target of rapamycin complex 1 activity and increasing autophagic flux, as observed in mouse lung samples. Patients with idiopathic pulmonary fibrosis who regularly used ezetimibe showed decreased rates of all-cause mortality and lung function decline.

Conclusion: Our study presents ezetimibe as a potential novel therapeutic for idiopathic pulmonary fibrosis.

Illustration of the working mechanism of ezetimibe in pulmonary fibrosis. ECM: extracellular matrix; IPF: idiopathic pulmonary fibrosis; mTORC1: mechanistic target of rapamycin complex 1; SRF: serum response factor; TGF: transforming growth factor.

FIGURE 1

Transcriptomic analysis reveals that ezetimibe inhibits myofibroblast differentiation and activates autophagy. a, b) Ezetimibe treatment of transforming growth factor (TGF)β1-activated human primary lung fibroblasts (hLFs) isolated from normal lung tissue leads to a decrease in both α1 type 1 collagen (COL1A1) a) protein and b) mRNA levels proportional to treatment duration. hLFs were activated using 2 ng·mL⁻¹ of human recombinant (h)TGFβ1 and treated with 20 µM ezetimibe as indicated. c–l) hLFs were treated for 24 h with human TGFβ1 and/or ezetimibe, or vehicles of each drug. Total RNA samples were extracted from three biological replicates from each group. Transcriptomic data were analysed after mRNA library preparation and sequencing. c, d) Gene set enrichment analysis (GSEA) was performed on the MSigDB v7.5.1 C5 GO-CC gene sets. The top 10 enriched gene sets in the c) TGFβ1 and d) TGFβ1+ezetimibe groups are presented. A positive score indicates upregulation by ezetimibe, while a negative score indicates downregulation. Highlighted bars are c) fibrosis-related genes (red), and d) autophagy-related genes (blue). Asterisks indicate nominal p-values. e–g) Three of the highlighted GSEA results in (c) are shown. The presented gene sets are e) Gene Ontology Cellular Component (GO-CC): collagen containing extracellular matrix, f) GO-CC: external encapsulating structure and g) GO-CC: contractile fibre. h) Heatmap of the expression of genes from (e) and (g) in the three treatment groups shows the downregulation of fibrosis-related transcripts in the ezetimibe-treated group. i–k) The highlighted GSEA results in (d) are shown. The presented gene sets are i) GO-CC: autophagosome, j) GO-CC: autophagosome membrane and k) GO-CC: phagophore assembly site. l) Heatmap of the expression of genes from the GO-CC autophagosome gene set in (g) shows the upregulation of autophagy-related transcripts in the ezetimibe-treated group. ns: nonsignificant; ACTB: actin-β; ER: endoplasmic reticulum; ES: enrichment score; NES: normalised enrichment score; C: control group; T: TGFβ1 group; TE: TGFβ1+ezetimibe group. *: p<0.05; ***: p<0.001; ****: p<0.0001.

FIGURE  2

Ezetimibe stimulates autophagic flux to suppress collagen accumulation in lung fibroblasts. a) The increase in the microtubule-associated protein light chain (LC)3B-II to LC3B-I ratio and LC3B-II protein level is proportional to the duration and dose of ezetimibe treatment. The representative immunoblots and quantitated results from four independent experiments are shown. b) Autophagic cargo flux was probed using a green fluorescent protein (GFP)-LC3 cleavage assay, measured by immunoblotting (IB) cell lysates with anti-GFP antibody. Mouse primary lung fibroblasts (mLFs) isolated from GFP-LC3 transgenic or nontransgenic mice were treated with the indicated drug for 24 h. The amount of free GFP reflects the degree of autophagic degradation. Representative immunoblots and quantified results from three independent experiments are shown. c) Autophagic carrier flux, probed with red fluorescent protein-GFP-LC3 puncta, is increased under ezetimibe treatment. mLFs isolated from RFP-GFP-LC3 transgenic mice were cultured on a coverslip and treated with the indicated drug and protein for 24 h. The cells were fixed and stained with Hoechst 33342, which was contained in the mountant. Fluorescent proteins were imaged using a confocal microscope. The representative images with quantified results are shown. Scale bars=20 μm. d) Prolonged chemical autophagic inhibition by chloroquine leads to α1 type 1 collagen (COL1A1) accumulation in lung fibroblasts and averts the activity of ezetimibe. Representative immunoblots with quantified results from three independent experiments are shown. Data are shown as mean±sem, statistically analysed with one-way ANOVA and Tukey's multiple comparisons adjustment. TGF: transforming growth factor; ACTB: actin-β; ns: nonsignificant. *: p<0.05; **: p<0.01; ****: p<0.0001; ^#: unadjusted p<0.05; ^##: unadjusted p<0.01.

FIGURE 3

Autophagic degradation of serum response factor (SRF) prevents myofibroblast differentiation. a) Transcriptomic analysis unveils SRF as the main transcription factor controlled by ezetimibe. Gene set enrichment analysis on MSigDB v7.5.1 C3 ontology gene sets reveals SRF-targeted gene sets (red) as the most downregulated by ezetimibe. The hierarchical clustering heatmap depicts the differentially expressed genes from the six SRF-targeted gene sets and their main Gene Ontology clusters. b) The decrease in SRF levels is consistent with the duration- and dose-related action of ezetimibe. Representative immunoblot images are shown. c) Prolonged chemical autophagic inhibition by chloroquine leads to SRF accumulation in lung fibroblasts and averts the activity of ezetimibe. Representative immunoblots and quantified results from three or four independent experiments are shown. d) Immunofluorescent analysis reveals differences in SRF localisation in autophagosomes and nuclei. Green fluorescent protein (GFP)-microtubule-associated protein light chain 3 (LC3) transgenic primary mouse lung fibroblasts (mLFs) treated with indicated drugs were immunostained with SRF. Representative images with quantified results are shown. Nuclear staining with Hoechst 33342 is blinded for better visualisation of GFP and SRF. Top: merged images (scale bars=10 μm); lanes 2–4: magnified optical images of SRF, GFP-LC3 and merged images, respectively. Arrows indicate GFP puncta with SRF colocalisation (scale bars=5 μm). Data are presented as mean±sem, statistically analysed with one-way ANOVA and Tukey's multiple comparisons adjustment. FDR: false discovery rate; ACTB: actin-β; ns: nonsignificant; MFI: mean fluorescence intensity. *: p<0.05; **: p<0.01; ***: p<0.001.

FIGURE 4

Ezetimibe suppresses activation of the intracellular cholesterol sensor, mechanistic target of rapamycin complex (mTORC)1, via inhibition of lysosomal cholesterol accumulation in lung fibroblasts. a) Transforming growth factor (TGF)β1-induced phosphorylation of the well-known mTORC1 substrates, ribosomal protein S6 kinase (p70S6K) and ribosomal protein S6 (RPS6), is suppressed proportional to ezetimibe dose. Representative immunoblots with quantified results from three to four independent biological replicates are shown. b) Immunofluorescence staining of mTOR and lysosome-associated membrane glycoprotein 1 (LAMP1) reveals TGFβ1-induced mTOR localisation to the lysosome in primary human lung fibroblasts. Representative images with from each stain and merged images including nuclear stain (Hoechst 33342) are shown. Scale bars=10 μm. The graph below images represents colocalisation coefficients of mTOR and LAMP1 (n=5–6). c) Ezetimibe decreases free cholesterol (FC) accumulation in lysosomes. Mouse primary lung fibroblasts (mLFs) were stained with Filipin III and LysoTracker Red to visualise free cholesterol and lysosomes, respectively. The mountant did not stain the nucleus. Representative images from each stain and merged images are shown. Scale bars=10 μm. The graph below shows colocalisation coefficients of Filipin and LysoTracker Red (n=11–16). d) Chronic cholesterol depletion abolishes TGFβ1-induced mTORC1 activation, α1 type 1 collagen (COL1A1) expression, and the effects of ezetimibe; whereas cholesterol supplementation increases p70S6K phosphorylation (mTORC activation) and COL1A1 expression and restores the effects of ezetimibe. Lung fibroblasts were incubated in Opti-MEM media containing vehicle, 100 μM methyl-β-cyclodextrin (MβCD), or 50 μM MβCD+free cholesterol and treated with the indicated drugs. Representative immunoblots for annotated antibodies are shown with quantified results from four independent experiments. The results were analysed with two-way ANOVA. Data are shown as mean±sem, statistically analysed with one-way ANOVA and Tukey's multiple comparisons adjustment, unless otherwise indicated. ns: nonsignificant. ACTB: actin beta; *: p<0.05; **: p<0.01; ***: p<0.001; ****: p<0.0001.

FIGURE 5

Ezetimibe prevents fibrosis progression in bleomycin-induced pulmonary fibrosis mouse model via regulation of the mechanistic target of rapamycin complex (mTORC)1–autophagy axis. Pulmonary fibrosis was induced by oropharyngeal aspiration of bleomycin. 7 days after exposure, ezetimibe or its vehicle was orally gavaged to the mice three times a week. The mice were sacrificed 21 days after the initial bleomycin aspiration. a) An illustration of the basic experimental design is shown with changes in bodyweight after bleomycin-induced lung injury. The bodyweight of ezetimibe-treated mice recovered faster than that of vehicle-treated mice (n=32–38 per group). The data were analysed using two-way ANOVA with adjustment for Tukey's multiple comparisons adjustment. Asterisks indicate the difference between the bleomycin group and the ezetimibe group. b) Ezetimibe alleviates collagen accumulation in the murine lung (n=13–15 per group). Hydroxyproline assay and Sircol Soluble Collagen assay were performed using the whole lysates of right upper lobes. c) Respiratory mechanics including static compliance and inspiratory capacity were recovered from bleomycin-induced pulmonary fibrosis. On the day of sacrifice, the mice were fully anaesthetised and tracheostomised using 18g catheter, and their respiratory functions were measured using flexiVent. d) Areas of fibrotic lesions are smaller in the lungs of ezetimibe-treated mice. Representative scanned images of Masson trichrome stain from paraffin-embedded slide sections of the left lungs are presented. Graphs below images are quantified measures of blue-stained collagen area and Ashcroft scores (n=19–20). Scale bars=2 mm. e) mRNA levels of fibrosis-related genes Col1a1 (α1 type 1 collagen), Col1a2 (α2 type 1 collagen), Eda-Fn (extra domain A containing fibronectin) and Col3a1 (α1 type 3 collagen) are downregulated in the ezetimibe-treated group compared with those in the bleomycin group (n=4–5 per group). f) Oral ezetimibe treatment results in an increase in autophagosome conversion and a decrease in mTORC1 activity and fibrosis. Representative immunoblots of single lysates from the right lungs of each group of mice are shown (n=4–5 per group). Data are presented as mean±sem. Each dot on a dot plot represents a result from an individual mouse. *: p<0.05; **: p<0.01; ***: p<0.001; ****: p<0.0001.

FIGURE 6

Ezetimibe mediates autophagy–serum response factor (SRF) pathway activity in myofibroblasts of mice with bleomycin-induced pulmonary fibrosis. Pulmonary fibrosis was induced in green fluorescent protein (GFP)-microtubule-associated protein light chain 3 (LC3) transgenic mice, followed by treatment with ezetimibe or vehicle from day 7 after bleomycin aspiration. Mice were euthanised on day 21. a) A GFP-LC3 cleavage assay performed on whole-lung lysates reveals increased GFP-LC3 cleavage in ezetimibe-treated pulmonary fibrosis mice. Representative immunoblots and quantified results are shown. b–d) Frozen lung sections from each GFP-LC3 transgenic mouse were immunostained with smooth muscle α2 actin (ACTA2) and SRF. b) Representative images of immunofluorescent staining are shown. ACTA2 and SRF were stained with Alexa Fluor 555 and Alexa Fluor 647, respectively. GFP-LC3 puncta represent autophagosomes. Scale bars=10 μm. c) Magnified images from the boxed area in (b) are shown. ACTA2 was blinded to enhance for better visualisation of GFP-LC3 and SRF. Scale bars=5 μm. d) ACTA2-expressing myofibroblasts in ezetimibe-treated mice showed increased number of autophagosomes and heightened SRF intensity within autophagosomes, along with reduced SRF intensity in the nuclei. The nuclei and GFP-positive vesicles in ACTA2-positive cells were identified, and SRF intensities were quantified using Imaris software. Each small dot represents one image. The means of each biological replicate are shown in larger coloured blocks. Two-tailed t-tests were performed to compare results from each group (n=3 per group). Data are presented as mean±sem. COL1A1: α1 type 1 collagen; ACTB: actin-β; ns: nonsignificant. *: p<0.05; **: p<0.01.

FIGURE 7

Ezetimibe reduces the rate of lung function decline and death in patients with idiopathic pulmonary fibrosis (IPF). a) Kaplan–Meier survival curves of patients with IPF categorised into four groups are shown: pirfenidone and ezetimibe users (PFD+ezetimibe), pirfenidone-only users (PFD), ezetimibe-only users and nonusers (none). The solid lines represent the survival estimate, and the dotted lines with shaded area represent the 95% confidence interval. Differences between the groups, stratified by the year of diagnosis, were analysed using log-rank tests and adjusted using the Tukey–Kramer test for multiple comparisons. Overall survival estimates were derived from all-cause mortality, and transplant-free survival estimates were derived from all-cause mortality and lung transplantation data. b) Cox multivariate regression analyses reveal decreased risk of death for users of ezetimibe and pirfenidone. All analyses were adjusted for several variables, including sex, age at diagnosis, comorbidities and gender-age-physiology stage for IPF. c–d) Progressive changes in c) forced vital capacity (FVC) and d) diffusing capacity of the lung for carbon monoxide (D_LCO) were analysed using a mixed model for repeated measures. Patients with IPF who used pirfenidone were categorised into two groups according to their ezetimibe prescription records, as indicated. The model included the following fixed effects: age at diagnosis, sex, lung cancer history, ezetimibe prescribed group, follow-up month and the interaction effect of ezetimibe and follow-up month. The random effects included the hospital factor and year of diagnosis. The within-individual variation from repeated measures was also adjusted according to the patient identifier. Data are presented graphically as the estimated mean±sem. Asterisks above each red dot reflect the p-value for difference on each time period. The lines represent lung function decline trend from a simple linear regression. HR: hazard ratio; ns: nonsignificant. *: p<0.05; **: p<0.01; ^#: unadjusted p<0.05.