A YAP/TAZ-miR-130/301 molecular circuit exerts systems-level control of fibrosis in a network of human diseases and physiologic conditions

Abstract

The molecular origins of fibrosis affecting multiple tissue beds remain incompletely defined. Previously, we delineated the critical role of the control of extracellular matrix (ECM) stiffening by the mechanosensitive microRNA-130/301 family, as activated by the YAP/TAZ co-transcription factors, in promoting pulmonary hypertension (PH). We hypothesized that similar mechanisms may dictate fibrosis in other tissue beds beyond the pulmonary vasculature. Employing an in silico combination of microRNA target prediction, transcriptomic analysis of 137 human diseases and physiologic states, and advanced gene network modeling, we predicted the microRNA-130/301 family as a master regulator of fibrotic pathways across a cohort of seemingly disparate diseases and conditions. In two such diseases (pulmonary fibrosis and liver fibrosis), inhibition of microRNA-130/301 prevented the induction of ECM modification, YAP/TAZ, and downstream tissue fibrosis. Thus, mechanical forces act through a central feedback circuit between microRNA-130/301 and YAP/TAZ to sustain a common fibrotic phenotype across a network of human physiologic and pathophysiologic states. Such re-conceptualization of interconnections based on shared systems of disease and non-disease gene networks may have broad implications for future convergent diagnostic and therapeutic strategies.

Figure 1. miR-130/301-specific fibrotic activity is active throughout a network of human diseases and physiologic conditions.

(A) Strategy used to identify miRNAs that exert systems-level control over fibrosis. (B) A fibrosis network, composed of known fibrotic genes (seed genes, circles) and their closest first-degree interactors (triangles, see Supplemental). Color-coding denotes inclusion in known annotated pathways relevant to fibrosis and the ECM (from the GO, Kegg, Reactome, NCBI PID, and Biocarta databases), and demonstrates the relevance of incorporated first-degree interactors to fibrotic processes. miR-130/301 was ranked among the top five miRNAs by spanning score in this network (targets encircled in black), reflecting its robust, systems-level control over fibrosis and matrix remodeling. (C) Based on transcriptomic profiling and network construction (see Methods), 137 networks representing various disease and physiologic states were grouped into cohorts, according to overlap with the fibrosis network. MiRNAs were then scored by one-way inverse correlation (one-way ANOVA) of their average assigned spanning score rank across cohorts. miR-130/301 was ranked among the top five miRNAs, underlining its importance to diseases with a strong fibrotic component (red box). (D) An abbreviated network was generated from 137 diseases and physiologic states (indexed in Table S1). Edges among conditions denote significant overlap (hypergeometric p-value < 0.1). Conditions in red were ranked highly based on their interconnectedness with the fibrosis network and the miR-130/301 family (top 25%, as ranked in Table S1), and all were found to share a distinct cohort of fibrotic genes embedded in the overlap with the PH network. The prevalence of fibrotic states predicted to involve miR-130/301 provides evidence for a shared fibrotic program controlled by this miRNA family in a wide range of human pathology and conditions beyond the pulmonary vasculature (see Table S3).

Figure 2. Positive correlation of ECM stiffening and YAP/TAZ-dependent expression of miR-130/301 in mice and humans suffering from lung fibrosis.

A mouse model of lung fibrosis (bleomycin-induction, n = 9–10/group) was analyzed. miR-130/301 was significantly increased by RT-qPCR (A), and serial sections of lung (B) displayed increased collagen (Picrosirius Red), miR-130a, and YAP by in situ hybridization. (C) In diseased lung, miR-130a and YAP nuclear localization was positively correlated with collagen crosslinking. (D) In situ staining for miR-130a and fibroblast markers (vimentin and α–SMA) was performed by fluorescent microscopy. Vimentin/miR-130a and α–SMA/miR-130a positive cells were increased in bleomycin-treated lung tissue (n = 8 per groups; 5 20X fields per slide were quantified). (E,F) In the lungs of patients with idiopathic pulmonary fibrosis, miR-130/301 and YAP1 were increased by RT-qPCR (E; [controls n = 5, fibrosis n = 15]) and in situ stain (F; [controls n = 8, fibrosis n = 10]). (See also Fig.S1). Data are expressed as mean ± SEM (*P < 0.05; ** P < 0.01).

Figure 3. Positive correlation of ECM stiffening and YAP/TAZ-dependent expression of miR-130/301 in mice and humans suffering from liver fibrosis.

A mouse model of liver fibrosis (CCl₄-induction, n = 9–10/group) was analyzed. miR-130/301 was significantly increased by RT-qPCR (A), and serial sections of lung (B) displayed increased collagen (Picrosirius Red), miR-130a, and YAP by in situ hybridization. (C) In diseased lung, miR-130a and YAP nuclear localization was positively correlated with collagen crosslinking. (D) In situ staining for miR-130a miR-130a and stellate cell markers (desmin and α–SMA) was performed by fluorescent microscopy. Desmin/miR-130a and a–SMA/miR-130a positive cells were increased in CCl₄-treated liver (n = 8 per groups; 5 20X fields per slide were quantified). (E) Similarly, in situ stain demonstrated an increase in miR-130a and YAP in fibrotic liver tissue of patients (n = 4 per group) with non-alcoholic steatohepatitis (NASH+fibrosis) (see also Table S4 and Fig.S2). Data are expressed as mean ± SEM (*P < 0.05; **P < 0.01).

Figure 4. miR-130/301 overexpression activates ECM remodeling and liver fibrosis progression in mouse.

(A) As assessed by RT-qPCR, serial intraperitoneal delivery of miR-130a mimic oligonucleotide (miR-130a) increased miR-130a in whole liver in mice as compared with control (miR-NC), either with or without weekly injection of suboptimal CCl₄ dose. (B) By Metavir score, either a suboptimal dose of CCl₄ or miR-130a independently increased liver fibrosis, while miR-130a+CCl₄ even more robustly increased such fibrosis. (C,D) In situ hybridization of miR-130a (top row) confirmed effective delivery in the liver of mice. Immunohistochemistry also revealed that miR-130a decreased Lrp8 and Pparγ as well as slightly increased YAP nuclear localization and collagen deposition and crosslinking. Moreover, miR-130a+CCl₄ more robustly decreased Lrp8 and Pparγ as well as more robustly increased YAP nuclear localization, collagen deposition, and crosslinking.

Figure 5. miR-130/301 inhibition ameliorates liver fibrosis.

In a model of liver fibrosis (CCl₄-induction), mice were treated with Short-NC or Short-130 (n = 8/10 per group). Short-130 delivery (n = 8 per group; A) and activity (n = 5 per group; (B) were confirmed. Short-130 decreased fibrosis, as assessed by α–SMA and collagen staining (C), as well as Metavir score (D). By in situ stain (C), Short-130 decreased vascular YAP nuclear localization and increased Pparγ and Lrp8. Short-130 also decreased transcript expression of Ctgf, collagen isoforms, and Lox (E) as well as decreased collagen deposition and collagen crosslinking (F). Data are expressed as mean ± SEM (*P < 0.05; **P < 0.01).

Figure 6. miR-130/301 inhibition ameliorates pulmonary fibrosis.

In a model of lung fibrosis (bleomycin-induction), mice were treated with control (Short-NC) or miR-130/301 inhibitor (Short-130) (n = 8/10 per group). Short-130 delivery (n = 8 per group; (A) and activity (n = 5 per group; (B) were confirmed. Short-130 decreased fibrosis, as assessed by α–SMA and collagen staining (C), as well as Ashcroft score (D). By in situ stain (C), Short-130 decreased vascular YAP nuclear localization and increased Pparγ and Lrp8. Short-130 also decreased transcript expression of Ctgf, collagen isoforms, and Lox (E) as well as decreased collagen deposition and collagen crosslinking (F). Data are expressed as mean ± SEM (*P < 0.05; **P < 0.01).

Figure 7. Pharmacological activation of APOE with LXR agonist GW3965 ameliorates pulmonary fibrosis.

In a model of lung fibrosis (bleomycin exposure), mice were treated with the LXR agonist GW3965 or control (n = 6 per group) via dietary intake (100 mg/kg). GW3965 ameliorated the degree of lung fibrosis as quantified by α-SMA labeling (A) and Aschcroft score (B). Picrosirius Red stain and in situ staining of mouse lung (A) demonstrated that GW3965 blunted bleomycin-mediated increases of collagen crosslinking and YAP, as well as decreased of Lrp8 and Ppar γ expression. C) By RT-qPCR, Ctgf, collagen isoforms, and Lox were increased in diseased lung (bleomycin), but GW3965 reduced such increased expression. (D) Unifying model of the central role of the YAP/TAZ-miR-130/301 circuit in control of ECM plasticity across related diseases. Data are expressed as mean ± SEM (*P < 0.05; **P < 0.01).