Vol 456 | 18/25 December 2008 | doi:10.1038/nature07511 LETTERS MicroRNA-21 contributes to myocardial disease by stimulating MAP kinase signalling in fibroblasts Thomas Thum1,2*, Carina Gross3*, Jan Fiedler1,2, Thomas Fischer3, Stephan Kissler3, Markus Bussen5, Paolo Galuppo1, Steffen Just6, Wolfgang Rottbauer6, Stefan Frantz1, Mirco Castoldi7,8,Ju¨rgen Soutschek9, Victor Koteliansky10, Andreas Rosenwald4, M. Albert Basson11, Jonathan D. Licht12, John T. R. Pena13, Sara H. Rouhanifard13, Martina U. Muckenthaler7,8, Thomas Tuschl13, Gail R. Martin5, Johann Bauersachs1 & Stefan Engelhardt3,14 MicroRNAs comprise a broad class of small non-coding RNAs that fibroblasts; expression was highest in fibroblasts from the failing control expression of complementary target messenger RNAs1,2. heart, but was low in cardiomyocytes (Fig. 2c and data not shown). Dysregulation of microRNAs by several mechanisms has been 3–5 6–10 a Early stage Intermediate stage Late stage described in various disease states including cardiac disease . 2 ) Whereas previous studies of cardiac disease have focused on 2 microRNAs that are primarily expressed in cardiomyocytes, the 1 role of microRNAs expressed in other cell types of the heart is 0 unclear. Here we show that microRNA-21 (miR-21, also known as Mirn21) regulates the ERK–MAP kinase signalling pathway in –1 control mice (log control failure model versus failure cardiac fibroblasts, which has impacts on global cardiac structure miRNA level in heart –2 and function. miR-21 levels are increased selectively in fibroblasts *** of the failing heart, augmenting ERK–MAP kinase activity through b Non-failing inhibition of sprouty homologue 1 (Spry1). This mechanism reg- 6 Failing ulates fibroblast survival and growth factor secretion, apparently Non-failing Failing *** controlling the extent of interstitial fibrosis and cardiac hyper- Early miR-21 4 trophy. In vivo silencing of miR-21 by a specific antagomir in a Intermediate miR-21 expression 2 mouse pressure-overload-induced disease model reduces cardiac Late miR-21 (fold control) miR-21 ERK–MAP kinase activity, inhibits interstitial fibrosis and attenu- 0 ates cardiac dysfunction. These findings reveal that microRNAs can contribute to myocardial disease by an effect in cardiac fibro- Early Late blasts. Our results validate miR-21 as a disease target in heart * Intermediate *** failure and establish the therapeutic efficacy of microRNA thera- c * d Non-failing Failing 5 peutic intervention in a cardiovascular disease setting. 3 pre- 100 4 In a transgenic mouse model of cardiac failure11, we found that the miR-21 2 3 cardiac microRNA expression signature was progressively deregu- 50 lated with increasing severity of the disease, with miR-21 expression 40 expression 2 12–14 1 30 being most strongly affected (Fig. 1a, b). Similarly, we observed control) (fold 1 miR-21/Rnu6-2 15,16 (fold non-failing) robust upregulation of miR-21 in other models of cardiac disease miR-21 miR-21 0 20 0 (Fig. 1c), as well as in human heart failure (Fig. 1d). The finding of Iso TAC Non- increased expression of the miR-21 precursor (pre-miR-21), as Failing failing assessed by northern blotting (Fig. 1d), suggested a transcriptional Control mechanism. Figure 1 | Deregulation of miR-21 expression in cardiac disease. Using in situ hybridization, we detected only weak miR-21 signals a, MicroRNA expression in left ventricular myocardium from control and in normal myocardium, whereas in failing myocardium the hybridi- b1-adrenergic receptor transgenic mice at various disease stages. miR-21 is marked in red. b, Northern blot analysis of miR-21 expression in mice from zation signal was greatly enhanced. At high magnification, the a. c, miR-21 expression in left ventricular myocardium from mice subjected hybridization signal was restricted primarily to small interstitial cells to TAC or isoproterenol (Iso) infusion. U6 small nuclear (Rnu6-2) was used (Fig. 2a), presumably cardiac fibroblasts. Consistent with our in situ as a control. d, Northern blot analysis of miR-21 in non-failing and failing hybridization data, when we isolated cell fractions from hearts human left ventricular myocardium. Data are mean and s.e.m.; *P , 0.05, (Fig. 2b), miR-21 expression was detected predominantly in cardiac ***P , 0.005. n 5 3–5 per experiment per group. 1Department of Medicine I, 2Junior Research Group, Interdisziplina¨res Zentrum fu¨r Klinische Forschung (IZKF), 3Rudolf Virchow Center, Deutsche Forschungsgemeinschaft (DFG) Research Center for Experimental Biomedicine, 4Institute of Pathology, University of Wuerzburg, 97080 Wuerzburg, Germany. 5Department of Anatomy, University of California, San Francisco, California 94158, USA. 6Department of Internal Medicine III, 7Department of Pediatric Hematology, Oncology and Immunology, 8Molecular Medicine Partnership Unit, University of Heidelberg, 69120 Heidelberg, Germany. 9Regulus Therapeutics, Carlsbad, California 92008, USA. 10Alnylam Pharmaceuticals, Cambridge, Massachusetts 02142, USA. 11Department of Craniofacial Development, King’s College, London SE1 9RT, UK. 12Northwestern University Feinberg School of Medicine, Chicago, Illinois 60611, USA. 13Laboratory of RNA Molecular Biology, Rockefeller University, New York, New York 10065, USA. 14Institute of Pharmacology and Toxicology, Technische Universitaet Muenchen (TUM), 80802 Muenchen, Germany. *These authors contributed equally to this work. 980 ©2008 Macmillan Publishers Limited. All rights reserved NATURE | Vol 456 | 18/25 December 2008 LETTERS ad* synthetic miR-21 precursor led to a significant increase in fibroblast miR-21 RNA growth factor 2 (FGF2) secretion (Fig. 2f). Non-failing 30 Our expression data showing that miR-21 expression in cardiomyo- 20 cytes is low and does not increase in failing hearts (Fig. 2c) indicate that miR-21 does not have a significant function in cardiomyocytes. 10 Neither enhancement nor suppression of miR-21 levels affected the 0 morphology, size or number of primary rat cardiomyocytes under Non-failing Failing resting or hypertrophy-inducing conditions (Supplementary Fig. 1). Failing *** We also generated transgenic mice overexpressing miR-21 in cardio- e *** myocytes (Supplementary Fig. 2a), but detected no notable differ- 15 *** ences in myocardial structure or function, either with age or when * scr-miR subjected to cardiac pressure overload (Supplementary Fig. 2a, b). 10 pre- Moreover, very few differences were detected by transcriptional miR-21 profiling or by determination of protein expression of predicted Apoptosis (% annexin-V-positive cells) Apoptosis (% annexin-V-positive 5 miR-21 targets in left ventricular myocardium from such miR-21 anti- b Cardiac Cardio- 0 miR-21 transgenic mice and their wild-type littermates (Supplementary Fig. fibroblasts myocytes Control PD98059 2c and data not shown). Likewise, we found few transcriptome P4HB c * f changes in primary cardiomyocytes transfected with miR-21 precur- 12 scr-miR pre-miR-21 sors and controls (Supplementary Fig. 2d). Although it has recently p-ERK1/2 been suggested that overexpression of miR-21 alters cardiomyocyte 8 14 ACTN2 morphology , our data provide strong evidence that miR-21 exerts its ERK1/2 4 effects on the heart primarily in cardiac fibroblasts. CM * Using bioinformatic algorithms to predict miR-21 targets, we iden- CF 0 tified SPRY1, a potent inhibitor of the Ras/MEK/ERK pathway19,20,asa Rnu6-2 ** 4 PECAM1 / Neonatal * 3 candidate messenger RNA target (Supplementary Table 1). The 8 3 39 untranslated region (UTR) of Spry1 mRNA contains several pre- ** miR-21 6 2 dicted microRNA-binding sites, of which only one corresponds to a 2 ACTA2 4 microRNA highly upregulated during cardiac disease (that is, miR-21). 1 1 We also identified a few other potential miR-21 targets, but these were (fold control) 2 p-ERK/ERK unaffected or only moderately downregulated on miR-21 expression in 0 0 FGF2 in supernatant 0 Adult fibroblasts (Supplementary Fig. 3). Transfection of primary cardiac fibroblasts with miR-21 precursors had no detectable effect on the Non- Spry1 mRNA level (data not shown), but resulted in strong repression failing Failing scr-miR scr-miR pre-miR-21 pre-miR-21 of SPRY1 protein expression (Fig. 3a), indicating that miR-21 affects Figure 2 | miR-21 promotes ERK–MAP-kinase-mediated cell survival in Spry1 expression by repressing translation. When we fused the Spry1 cardiac fibroblasts. a, In situ hybridization of miR-21 in non-failing and 39 UTR to a luciferase reporter gene and determined luciferase activity failing myocardium. Scale bar, 100 mm. b, Cell fractions from rat hearts in cells transfected with synthetic miR-21 precursors, miR-21 signifi- stained with DAPI and antibodies against prolyl 4-hydroxylase (P4HB), a2- cantly repressed luciferase activity, whereas neither miR-133a nor actinin (ACTN2), CD31 (PECAM1) and smooth muscle a2-actin (ACTA2). miR-206 had any effect (Fig. 3a). These experiments identify Spry1 Scale bar, 100 mm. c, miR-21 expression in cardiomyocytes (CM) and cardiac as a direct target of miR-21. fibroblasts (CF) isolated from neonatal rat or adult mouse hearts. We next determined the Spry1 expression pattern in the heart. In d, Percentage apoptotic cells in the cardiac fibroblast fraction isolated from situ hybridization on sections of neonatal wild-type heart showed b individual non-failing (wild-type) and failing (intermediate stage 1- that within the myocardium Spry1 mRNA localized to interstitial adrenergic receptor transgenic) mouse hearts. Horizontal bars indicate the mean. e, Percentage apoptotic primary rat cardiac fibroblasts after treatment cells and the smooth muscle of the coronary vessels, but was not with scrambled-miR (scr-miR), synthetic miR-21 (pre-miR-21), miR-21 detected in cardiomyocytes (data not shown). We generated an allele lacZ antagonists (anti-miR-21) and the ERK–MAP kinase inhibitor PD98059. of mouse Spry1 (Spry1 ) in which the lacZ gene replaces part of the f, Western blot assay for ERK1/2 and phospho-ERK1/2 (p-ERK1/2), and Spry1 coding sequence (Supplementary Fig. 4). Assays for lacZ ELISA of secreted FGF2. Data are mean and s.e.m.; *P , 0.05, **P , 0.01, expression in 3-week-old Spry1lacZ/1 animals indicated that Spry1 ***P , 0.005.