The Mineralocorticoid Receptor Promotes Fibrotic Remodeling in Atrial Fibrillation*

Background: We studied the role of the mineralocorticoid receptor (MR) for atrial fibrotic remodeling. Results: Increased 11β-hydroxysteroid dehydrogenase type 2 in atrial fibrillation enhances mineralocorticoid receptor pro-fibrotic signaling through connective tissue growth factor, lysyl oxidase, and microRNA-21. Conclusion: The MR regulates fibrogenesis in atrial fibrillation. Significance: The MR may represent a target for the prevention of atrial fibrosis. We studied the role of the mineralocorticoid receptor (MR) in the signaling that promotes atrial fibrosis. Left atrial myocardium of patients with atrial fibrillation (AF) exhibited 4-fold increased hydroxyproline content compared with patients in sinus rhythm. Expression of MR was similar, as was 11β-hydroxysteroid dehydrogenase type 2 (11β-HSD2), which also increased. 11β-HSD2 converts cortisol to receptor-inactive metabolites allowing aldosterone occupancy of MR. 11β-HSD2 was up-regulated by arrhythmic pacing in cultured cardiomyocytes and in a mouse model of spontaneous AF (RacET). In cardiomyocytes, aldosterone induced connective tissue growth factor (CTGF) in the absence but not in the presence of cortisol. Hydroxyproline expression was increased in cardiac fibroblasts exposed to conditioned medium from aldosterone-treated cardiomyocytes but not from cardiomyocytes treated with both cortisol and aldosterone. Aldosterone increased connective tissue growth factor and hydroxyproline expression in cardiac fibroblasts, which were prevented by BR-4628, a dihydropyridine-derived selective MR antagonist, and by spironolactone. Aldosterone activated RhoA GTPase. Rho kinase inhibition by Y-27632 prevented CTGF and hydroxyproline, whereas the RhoA activator CN03 increased CTGF expression. Aldosterone and CTGF increased lysyl oxidase, and aldosterone enhanced miR-21 expression. MR antagonists reduced the aldosterone but not the CTGF effect. In conclusion, MR signaling promoted fibrotic remodeling. Increased expression of 11β-HSD2 during AF leads to up-regulation of collagen and pro-fibrotic mediators by aldosterone, specifically RhoA activity as well as CTGF, lysyl oxidase, and microRNA-21 expression. The MR antagonists BR-4628 and spironolactone prevent these alterations. MR inhibition may, therefore, represent a potential pharmacologic target for the prevention of fibrotic remodeling of the atrial myocardium.

Atrial fibrillation (AF) 2 is the most common arrhythmia. The prevalence is increasing, especially in the elderly (1). AF is a major cause of thromboembolic stroke and cardiovascular morbidity (1). Atrial enlargement and fibrosis are predisposing factors for the development of AF. On the other hand, AF contributes to structural changes and fibrosis in the left atrium (2). Fibrosis is a hallmark of structural remodeling in left atria of AF patients. Interstitial fibrosis perturbs cardiac electrophysiological properties and promotes the formation micro-reentry circles leading to AF (2). Despite substantial progress in the field of structural and electrical remodeling, additional understanding of the underlying molecular pathways is needed to identify pharmacologic targets to prevent AF (3,4).
The activation of the mineralocorticoid aldosterone leads to left ventricular remodeling (5)(6)(7)(8)(9). Clinical trials show beneficial effects of mineralocorticoid receptor (MR) antagonist in patients with systolic heart failure (10 -12). There is evidence that MR antagonism may reduce the incidence of AF in heart failure patients (13). However it is currently unknown whether improvements in the hemodynamic situation or direct atrial effects of MR antagonist treatment are responsible for the beneficial effects. Recently, aldosterone was found to be able to induce AF and fibrosis (14). Aldosterone passes the cell membrane and binds at and activates the intracellular localized MR (6). Activation of the MR is regulated by the 11-␤-hydroxysteroid dehydrogenase 2 (11␤-HSD2), which oxidizes the glucocorticoid cortisol to the inactive metabolite cortisone, thereby preventing cortisol occupancy at the MR. The concentration of cortisol in the human serum is 100 -1000-fold higher compared with aldosterone, but its binding affinity to the mineralocorticoid receptor is similar. Therefore, cortisol mainly occupies the MR binding site in cells that do not express 11␤-HSD2, and 11␤-HSD2 expression prevents illicit MR binding by cortisol (15).
We have recently observed that atrial fibrillation is associated with structural remodeling, including overexpression of connective tissue growth factor (CTGF), lysyl oxidase (LOX), and micro-RNA-21 (16 -18), but the role of the MR in this context is unknown. Therefore, the aim of this study was to investigate the role of the MR in the pathways that promote structural remodeling and fibrosis in the left atrial myocardium during atrial fibrillation.

EXPERIMENTAL PROCEDURES
Human Left Atrial Tissue-Tissue samples of the left atrial appendage of patients undergoing mitral valve surgery were analyzed in patients with sinus rhythm (SR) and with permanent AF (documented by ECG for Ͼ3 months). The samples were matched for atrial diameter (see Table 1). The patients did not receive drugs at least 12 h before surgery. The analysis was approved by the ethics committee of the Ärztekammer des Saarlandes (no. 131/00).
Cell Isolation and Culture-Neonatal rat cardiomyocytes and cardiac fibroblasts were isolated from 5-day-old neonatal Sprague-Dawley rat hearts (Charles River, Germany) of mixed sex as described (19,20). After 48 h in culture, cardiomyocytes exhibited regular spontaneous contractions and were used for experiments after 3-5 days of culture. Fibroblasts were separated by adhesion, and purity of fibroblasts was confirmed by vimentin staining. Neonatal cardiac fibroblasts were grown in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal calf serum, gentamicin (0.08 mg/ml), and penicillin (100 IU/ml). When subconfluent in passage three, cells were kept in serum-free medium for 24 h before treatment. Cellular protein content was quantified by a modified Lowry assay.
Electric Stimulation of Neonatal Rat Cardiomyocytes-Cellular viability of cardiomyocytes was determined by cell counting, morphology, and trypan blue exclusion. Cardiomyocytes were plated in 6-well culture trays (BD Falcon 6-well plate 353846, BD Biosciences) at 0.5-0.6 ϫ 10 6 per well (ϳ9 cm 2 ) containing F-10 medium (Invitrogen) supplemented with 10% horse serum, 5% fetal calf serum, 1% penicillin, and streptomycin and maintained in 95% air and 5% CO 2 at 37°C. After 48 h of culture, ϳ70% of the cells adhered to the culture dishes and began to oscillate (beat spontaneously). Thereafter, the culture medium was changed daily. On day 4, most cardiomyocytes beat synchronously and at a constant frequency. Thereafter cardiomyocytes were stimulated with an electronic multichannel cell stimulator (Cell Culture EP Stimulator, IonOptix Ltd) via carbon electrodes (C-Dish CLD6WBFC, IonOptix) at a frequency of 3 Hz, either regular or irregular. The irregular pacing was produced by a pseudo-random variation of the specified frequency within a 50% window. For example, if a frequency of 1 Hz with 50% variability is selected, 100 pulses will have occurred after 100 s, but the time period between any two pulses can be anything from 500 to 1500 ms. An equal number of electrical stimulations per min were applied either rhythmically or arrhythmically. Therefore, this model specifically addresses arrhythmia-rather than tachycardia-mediated cellular alterations. The duration of each pulse was 2-3 ms, and the strength of the electrical field was Ͻ7 V/cm, which is well below that reported to cause cell injury (21). Effective contraction by electric stimulation was checked by optical microscopy at 20ϫ magnification at the beginning and the end of each stimulation. To minimize electrolysis at the electrodes and a possible generation of oxidative species, the polarity of the electrodes was altered with each electrical pulse. After use, the electrodes were cleaned in distilled water with changing the water every 6 -12 h for at least 4 times and till the water was absolutely clear. Directly before use electrodes were soaked into 70% ethanol for 1 h for disinfection. After electric stimulation cells were washed with 2 ml of PBS, lysed with 200 l of ice-cold lysis buffer, scraped, and homogenized with 5 min of ultrasound. After homogenization, debris was removed by centrifugation at 14,000 ϫ g for 15 min at 4°C, and the supernatant was transferred to Eppendorf cups for storage at Ϫ20°C. Total protein concentration was quantified by a modified Lowry assay for Western blot analysis.
Animal Studies-Mice with cardiac overexpression of constitutively active (V12) Rac1 under the control of the ␣-myosin heavy chain promoter (RacET; provided by Prof. Mark A. Sussman, San Diego State University Heart Institute and Department of Biology, San Diego State University, San Diego, CA) and their wild-type (WT) controls (22) were fed with normal chow (ssniff, Germany). RacET mice show spontaneous or inducible atrial tachyarrhythmia in 45% as well as atrial conduction abnormalities at the age of 6 months (23). At a high age (16 months) they develop permanent AF spontaneously (24). The study was approved by the Animal Ethics Committee of the Universität des Saarlandes and is in accord with the "Guide for the Care and Use of Laboratory Animals" published by the National Institutes of Health (no. 85-23, revised 1996).
Specification of Drugs-Aldosterone and TGF-␤ 1 (both Sigma) were used as endogenous mediators of fibrosis. Aldosterone is a physiological mineralocorticoid that mediates its cellular effects via binding at the MR but also possesses non-MR effects. Cortisol (synonym, hydrocortisone; Rotexmedica GmbH) is an endogenous glucocorticoid that binds at the glucocorticoid and the mineralocorticoid receptor. To investigate the competitive binding between aldosterone and cortisol at the MR, we used either a physiological concentration of cortisol (127 ng/ml) (25) or a 100-fold higher concentration (12.7 g/ml) for mechanistic studies, as indicated in the figure legends. TGF-␤ was used at the concentration of 5 ng/ml that was shown to stimulate the expression of CTGF and cardiac fibrosis (26). The MR antagonist spironolactone (Sigma) is a widely used therapeutic drug with well characterized clinical and biochemical properties. BR-4628 (Bayer Pharma AG) is a new MR antagonist derived from the chemical structure of dihydropyridines with a advanced specificity for the MR (27). BR-4628 and spironolactone exhibit a similar IC 50 at the MR and were, therefore, used in the same concentrations (500 nM). This concentration is based on pharmacological studies to inhibit the MR activation by aldosterone binding (27). CN03 (Cytoskeleton) is a RhoA activator with an active site based on the catalytic domain of the bacterial cytotoxic necrotizing factor toxins. CN03 activates Rho GTPase isoforms by deamidating glutamine-63. This modification converts glutamine 63 to glutamate that blocks GTPase activity, resulting in constitutively active Rho (28). We used concentrations following the manufacturer's datasheet (1 g/ml) to achieve a constitutive active RhoA. The Rho kinase inhibitor Y-27632 was used at a concentration of 10 M that has been shown to down-regulate the Rho-dependent protein kinase (29). The concentration of CTGF was established in previous studies to induce pro-fibrotic changes in cardiac fibroblasts (16,18).
Micro-RNA Isolation and TaqMan-PCR-MicroRNAs were isolated by a miRNA isolation kit (mirVana, Ambion). For TaqMan-PCR, a target-specific stem loop structure and reverse transcription primers were used. After reverse transcription miR-21 expression was quantified with specific TaqMan hybridization probes (TaqMan miR-21 microRNA assay, Applied Biosystems). The small RNA molecule U6 small nuclear (Rnu6-2) was amplified as a control. The 11␤-HSD2 mRNA expression was measured using the TaqMan probe Rn-00492539-m1 from Applied Biosystems (30). The 18 S rRNA was amplified as the control.
Measurement of Hydroxyproline-Hydroxyproline content in left atrial myocardium (LA) and supernatants of cultured cardiac fibroblasts were determined by enzyme-linked immunosorbent assay using a hydroxyproline assay kit (Quickzyme, The Netherlands) according to the manufacturer's instruction.
Statistical Analysis-Band intensities were analyzed by densitometry. Data were calculated from at least three independent experiments and are presented as the mean S.E. Statistical anal-

RESULTS
Mineralocorticoid Receptor and 11␤-Hydroxysteroid Dehydrogenase Type 2 in Human LA-We investigated samples of LA from patients undergoing mitral valve cardiac surgery who had either permanent atrial fibrillation or sinus rhythm. The characteristics of the patients are displayed in Table 1. Groups did not differ significantly with regard to cardiovascular medications. The samples were matched for atrial diameter. Patients with AF are characterized by increased left atrial fibrosis compared with patients with SR (Sirius red staining; Fig. 1, A and B), and total collagen content was quantitated by a hydroxyproline assay (425 Ϯ 103%, p Ͻ 0.05; Fig. 1C).
The expression of MR mRNA and protein was similar in AF and SR patients (107 Ϯ 13% (p ϭ ns) and 87 Ϯ 9% (p ϭ ns), respectively; Fig. 1, D and E). MR expression did not differ between very large LA (63 Ϯ 2 mm) and large LA (48 Ϯ 3 mm) (p ϭ 0.42, data not shown). Because the MR has similar affinities to bind aldosterone and glucocorticoids, the expression of the cortisol-inactivating enzyme 11␤-HSD2 was investigated. LA of AF patients exhibited increased expression of 11␤-HSD2 compared with SR patients (416 Ϯ 124%, p Ͻ 0.05; Fig. 1F).
MR, Hydroxyproline and CTGF Expression in Cultured Cardiac Fibroblasts-MR expression and its role in fibrotic remodeling were studied in cultured neonatal rat cardiac fibroblasts. Cardiac fibroblasts express MR mRNA and protein. Neither aldosterone nor TGF-␤ treatment altered the basal MR expres- sion (Fig. 3A). We found no protein expression of 11␤-HSD2 in cultured cardiac fibroblasts (data not shown).
Regulation of 11␤-HSD2 in Cardiomyocytes and Transgenic RacET Mice-11␤-HSD2 expression was not found in cardiac fibroblasts (data not shown), which is in concordance with the literature (40). Cardiomyocytes expressed 11␤-HSD2, and neither aldosterone nor TGF-␤ treatment altered 11␤-HSD2 expression (data not shown). Aldosterone increased CTGF expression in cardiomyocytes (141 Ϯ 12%, p Ͻ 0.05; Fig. 6A), but a physiological concentration of cortisol in the cell culture medium completely abolished the aldosterone effect (72 Ϯ 8%, p Ͻ 0.01 versus aldosterone; Fig. 6A), indicating MR occupancy by cortisol. Cortisol itself did not alter CTGF expression (Fig.  6A). To investigate the functional interaction between cardiomyocytes and cardiac fibroblasts, the latter were incubated with cardiomyocyte-conditioned medium. We observed an increase of collagen production by cardiac fibroblasts exposed to cell culture medium from cardiomyocytes that were incubated with aldosterone (154 Ϯ 12%, p Ͻ 0.001; Fig. 6B). Co-incubation with cortisol prevented this aldosterone-induced increase of collagen production (114 Ϯ 12%, p Ͻ 0.05 versus aldosterone; Fig. 6B).
To characterize the expression of 11␤-HSD2 in vivo we studied transgenic mice with cardiac overexpression of constitutively active (V12) Rac1 under the control of the ␣-myosin heavy chain promoter (RacET). These mice show a marked 30-fold up-regulation of Rac1 activity compared with WT mice and developed AF spontaneously at old age (24). 12-Month-old RacET mice exhibited a Ͼ6-fold up-regulation of atrial 11␤-HSD2 expression compared with wild type control mice (FVBN) (665 Ϯ 189%, p Ͻ 0.05; Fig. 6H). MR expression was similar in RacET and wild type control mice (120 Ϯ 5%, p ϭ ns; Fig. 6G).

DISCUSSION
The data identify the mineralocorticoid receptor and its regulation by the 11␤-hydroxysteroid dehydrogenase 2 as important contributors for atrial fibrotic remodeling. The 11␤-HSD2 converts cortisol to receptor-inactive metabolites and thereby allows aldosterone to bind at the MR. Importantly, 11␤-HSD2 is up-regulated in human LA from patients with AF compared with SR, in a mouse model of spontaneous AF (RacET mice) and in irregularly compared with regularly paced cardiomyocytes. This finding explains the increased pro-fibrotic MR-dependent signaling despite equal expression of MR. The subsequent mechanistic studies identified the RhoA/Rho kinase pathway, CTGF, LOX, miR-21, and Sprouty-1 as components of the MR-controlled pro-fibrotic atrial remodeling.
Atrial fibrillation is a disease of the left atrial myocardium. Therefore, the use of left atrial samples for the analysis is appropriate. The expression of MR in moderately enlarged LA was similar in AF and SR patients, whereas other studies have reported increased expression of MR in AF patients (38,39). Potential explanations for these differences include the use of left atrial tissue compared with the right atrial appendage in other studies (38) and the study of more severely diseased hearts and atria (39).
The MR has similar affinities for aldosterone and cortisol and, hence, MR is occupied by cortisol under physiological conditions because of its 1000 time higher serum concentrations. The enzyme 11␤-HSD2 inactivates cortisol to cortisone. High expression levels of 11␤-HSD2 protect MR from glucocorticoid binding and enables aldosterone to bind to the MR (15,40). Our data demonstrate that 11␤-HSD2 is up-regulated in LA of AF compared with SR patients despite similar MR expression and that 11␤-HSD2 correlates with fibrosis in human LA. We, therefore, studied the role of 11␤-HSD2 regarding MR-mediated pro-fibrotic signaling in cardiomyocytes. Aldosterone increased the expression of CTGF but in the presence of cortisol the aldosterone effect was abolished. Cardiac fibroblasts do not express 11␤-HSD2 (40). Conditioned medium from cardiomyocytes treated with aldosterone induced a pro-fibrotic response in fibroblasts; however, this effect was absent when the cardiomyocytes were treated with aldosterone in the presence of cortisol. These findings suggest that under physiological conditions aldosterone has little pro-fibrotic activity because the MR is occupied by cortisol. In paced cardiomyocytes we observed an arrhythmia-associated increase in the 11␤-HSD2 protein expression. Because 11␤-HSD2 mRNA expression was not increased significantly, the regulation of 11␤-HSD2 appears to be, at least in part, posttranscriptional, as described (42,43). Transgenic mice with cardiac overexpression of constitutive active Rac1 (RacET) that develop spontaneous AF (23,24) showed increased atrial 11␤-HSD2 expression compared with wild type mice. These data point to an arrhythmia and Rac1-mediated regulation of 11␤-HSD2. In summary, these results provide evidence that 11␤-HSD2 up-regulation in AF facilitates a pro-fibrotic signaling via the MR.
Aldosterone increased collagen production in cardiac fibroblasts, thus validating this cell culture model for investigating MR-dependent fibrotic remodeling in vitro. CTGF is a mediator of fibrosis in AF (16,18) and of aldosterone-induced cellular alterations (41,44). Our data demonstrate that aldosterone induces CTGF expression via the MR, whereas cortisol did not. Recent data identified RhoA/Rho kinase as a potential mediator of aldosterone-induced renal injury including up-regulation of CTGF (45). We recently showed that RacET mice have an increased RhoA activity in the myocardium (16). RhoA correlates with fibrosis in human left atrial myocardium. Our in vitro data reveal that aldosterone increases RhoA activity and that the RhoA/Rho kinase pathway regulates CTGF expression and collagen production downstream of the MR. RhoA/Rho kinase was confirmed as a common pathway where TGF-␤ and MR signaling converge.
Fibroblasts exposed to CTGF exhibited increased expression of LOX, the key enzyme for collagen cross-linking (18,46). Aldosterone increased LOX expression in a MR-and Rho kinase-dependent manner. These results confirm a dependence of LOX on CTGF as a subsequent pathway, and both CTGF and LOX are induced by aldosterone via the MR.
MicroRNAs are small non-coding RNAs that control expression of complementary target messenger RNAs (47). MicroRNA-21 increases fibroblast survival that leads to cardiac fibrosis and dysfunction (36). We and others have observed that miR-21 promotes atrial structural remodeling (17,48,49). LA of patients with AF exhibit increased miR-21 expression (17), and miR-21 as well as Sprouty-1 expression in human LA correlates positively with fibrosis. Here, these data are extended showing that the MR-dependent pro-fibrotic signaling involves miR-21 and Sprouty-1.
BR-4628 is a dihydropyridine-derived non-steroidal and selective MR antagonist (27). BR-4628 and spironolactone exhibit similar potency against MR but differ in their affinity to other steroid receptors; BR-4628 possesses the highest specificity for MR (27). The prevention of aldosterone induced alterations, e.g. collagen production, CTGF, LOX, miR-21, and Sprouty-1 expression, by the MR antagonists BR-4628 and spironolactone suggests that 1) aldosterone-induced fibrotic changes are strongly MR-dependent, and 2) MR blockade is a potential target to prevent structural remodeling and fibrosis. Important limitations of this study include that isolated human atrial myocytes and fibroblasts in culture were not available for the mechanistic studies. We are aware of the limitations of neonatal rat cardiac fibroblasts and myocytes prepared from left ventricles. To prevent differentiation, these cells were used until passage 3 or less only. As all animal models and cell culture have potential limitations with regard to the human disease, the human left atrial myocardium is the primary readout of the study, and our cell culture and animal studies are performed to dynamically study the signal transduction.

CONCLUSION
In summary, MR signaling promotes fibrotic atrial remodeling. Increased expression of 11␤-HSD2 leads to up-regulation of pro-fibrotic mediators by aldosterone, specifically RhoA activity as well as CTGF, LOX, miR-21 expression, and collagen synthesis. The MR antagonists BR-4628 and spironolactone prevented these alterations. MR inhibition may, therefore, represent a potential pharmacologic target for the prevention of fibrotic remodeling of the atrial myocardium.