Relaxin reverses maladaptive remodeling of the aged heart through Wnt-signaling

Healthy aging results in cardiac structural and electrical remodeling that increases susceptibility to cardiovascular diseases. Relaxin, an insulin-like hormone, suppresses atrial fibrillation, inflammation and fibrosis in aged rats but the mechanisms-of-action are unknown. Here we show that relaxin treatment of aged rats reverses pathological electrical remodeling (increasing Nav1.5 expression and localization of Connexin43 to intercalated disks) by activating canonical Wnt signaling. In isolated adult ventricular myocytes, relaxin upregulated Nav1.5 (EC50 = 1.3 nM) by a mechanism inhibited by the addition of Dickkopf-1. Furthermore, relaxin increased the levels of connexin43, Wnt1, and cytosolic and nuclear β-catenin. Treatment with Wnt1 or CHIR-99021 (a GSK3β inhibitor) mimicked the relaxin effects. In isolated fibroblasts, relaxin blocked TGFβ-induced collagen elevation in a Wnt dependent manner. These findings demonstrate a close interplay between relaxin and Wnt-signaling resulting in myocardial remodeling and reveals a fundamental mechanism of great therapeutic potential.

Here, we tested the hypothesis that the effects of RLX on Nav1.5 and fibrosis are mediated by canonical Wnt signaling. While RLX and Wnt signaling have been investigated in models of cancer [18][19][20] , there is no work on the interactions of the RLX and Wnt pathways in adult heart and 'healthy' aging as a precursor of cardiac diseases. We focused on hearts from 9-month (adult) and 24-month-old (aged) Fischer 344/Brown Norway F1 (F-344) male rats obtained from the National Institute of Aging. F-344 rats have been extensively studied and characterized as a model of aging because of their linear aging, and because, like humans, they exhibit an age-dependent excess of cardiac fibrosis and susceptibility to atrial fibrillation 5,21 . Rats were treated with RLX (400 μg/kg/day) or vehicle (sodium acetate) via subcutaneous osmotic mini-pumps for 14 days, resulting in an increase in the circulating plasma concentration of RLX from 0.017 ± 0.027 ng/mL to 23.76 ± 10.60 ng/mL, as determined by an ELISA assay.

Relaxin improves Ventricular function
Hearts were perfused in a Langendorff apparatus for optical mapping of atrial and ventricular action potentials to calculate conduction velocity (CV) 5,6 . Aged ventricles treated with RLX showed a marked increase of CV, an effect that was particularly pronounced at fast pacing rates compared to aged control ventricles (Fig. 1Aa). CV restitution kinetics measured by programmed stimulation revealed that premature stimuli elicited depolarizing waves in control aged rats with considerably slower CV than in RLX treated aged-rats (Fig. 1Ab). It is interesting to note that when paced at cycle lengths below 150 ms, control aged ventricles failed to capture and propagate or maintain a steady state CV whereas RLX-treated ventricles could be paced at those faster rates without significant decrement of CV (Fig. 1Aa). In ventricles of RLX treated rats, the slope of the restitution kinetics curve becomes less steep than for untreated animals which suggests a decrease in arrhythmia vulnerability (Fig. 1Ab).
A hallmark of advancing age is an escalation of excess fibrosis. Increased collagen deposits are associated with cardiac dysfunction due to disruption of cell-cell coupling and a depolarization of the resting membrane potential 22 . Collagen content was measured by staining left ventricular (LV) sections with Picro-Sirius Red and by calculating the collagen to tissue ratio in RLX treated and control animals. RLX treatment significantly reduced the collagen to tissue ratio (Fig. 1Ba-c). Most important, RLX reversed age-associated fibrosis to levels like those measured in 11-month-old rats 23 .
In addition, reductions in Nav1.5 expression contribute to arrhythmia and other heart pathologies by slowing CV 22 . RLX significantly increased Nav1.5 expression in LV sections compared to control (Fig. 1Ca-c).
Further, β-catenin was reduced in LV sections of aged ( Fig. 2d) compared to adult (Fig. 2a) rat hearts and was diffusely distributed in the cytosol with relatively little localization at intercalated discs (IDs) while Cx43 was localized to the lateral membranes in aged vs adult LV (Fig. 2e vs. 2b). RLX shifted the localization of Cx43 from the lateral membranes to IDs, increased the levels of β-catenin, and β-catenin became co-localized with Cx43 at IDs (Fig. 2g-i). In general, LV of aged-rats with RLX-treatment resembled the LV of adult rats (Fig. 2a-c). This strongly suggested that RLX treatment reversed some of the aging-induced changes observed in 24-month-old rats. Similar findings were seen in aged atria (Fig. 3A). Control experiments showed that the same Relaxin treatment on young (9 months old) rats had no significant effects, see Supplemental Fig. S1.

Relaxin Acts Via Wnt-Signaling in Aged Rats
Based on previous work of Wnt signaling effects on Nav1.5 and ID proteins, we tested the role of Wnt-signaling in RLX-mediated upregulation of Nav1.5 in adult and aged animals. RNA from aged atria ± RLX-treatment followed by RT-PCR showed the expected reduction of fibrosis mRNA biomarkers (Metalloproteinases (MMPs: MMP-2 and 9, α-smooth muscle actin) but also showed a marked increase in Wnt1 expression (Fig. 3, Table 1). Likewise, RLX-treatment increased Wnt1 expression in LV sections (Fig. 3B). Next, tissue from aged LV ± RLX was tested for Dickkopf-1 (Dkk1) expression, an intrinsic inhibitor of Wnt signaling. Dkk1 has been shown to bind LRP6 and blocks canonical signaling of all Wnt ligands. RLX-treated LV showed a significant reduction of Dkk1 compared to control LV (Fig. 4). These data suggest that RLX acts to activate Wnt signaling by two complementary mechanisms: increasing Wnt1 expression and reducing Dkk1 and Wnt inhibition.

RLX-Wnt Interaction in Isolated
Cardiomyocytes. The effects of RLX were studied in isolated adult left ventricular myocytes by culturing the cells in 96 well plates with various doses (0-100 nM) of RLX for 2-days ( Fig. 5a,b). After treatment, the myocytes were fixed, labeled with Nav1.5 antibody and fluorescein-conjugated secondary antibody and fluorescent phalloidin (to label actin used as a counterstain for normalization purposes). Fluorescence intensities showed that RLX up-regulated Nav1.5 with an EC 50 of 1.3 nM (Fig. 5c-e), which is consistent with the reported affinity of RLX for its cognate receptor, RXFP1 24 .
As shown in Fig. 2, RLX increased β-catenin and Cx43 trafficking to the ID. To determine whether RLX also acted via a Wnt-related mechanism in isolated myocytes, we incubated freshly isolated ventricular myocytes with RLX (25 nM; 24-48 hours) or vehicle (Na + acetate) and then double stained with DAPI (nuclear stain) and a β-catenin antibody to measure the number of myocytes that expressed nuclear β-catenin. Control myocytes had a considerably smaller fraction of β-catenin-containing nuclei (Fig. 6Aa,c) compared to RLX treated myocytes (Fig. 6Ab,c), suggesting that RLX is acting in part by activating canonical Wnt signaling in isolated cardiomyocytes.
To determine if Wnt signaling is directly involved in Nav1.5 regulation, myocytes were treated with either CHIR-99021 (a GSK3β inhibitor, 3 µg/mL), RLX (25 nM) ± Dkk1 (0.1 µg/mL) or Wnt1 (0.1 µg/mL) ± Dkk1 for 24 hours. Activation of Wnt signaling by the GSK3β inhibitor, CHIR-99021 resulted in a significant increase in Nav1.5 expression to the same extent as RLX alone and compared to control myocytes. The simultaneous treatment of myocytes with CHIR-99021 and RLX did not result in a further increase of Nav1.5 expression (Fig. 6Ba-e). Inhibition of Wnt signaling by Dkk1 significantly blocked RLX (Fig. 6f-j) and Wnt1's ( Fig. 6Bk-o) effects on Nav1.5 expression. These results strongly suggest that, contrary to the effects reported in NRVM, canonical Wnt signaling in adult cardiomyocytes increases Nav1.5 expression.
The effect of RLX on Dkk1 seen in rats treated with the hormone was also studied in isolated cardiomyocytes by treating the cultured myocytes with Na-acetate (control) or RLX for 2-days followed by labeling with an anti-Dkk1 antibody. In control myocytes, Dkk1 protein appeared as broadly distributed punctate spots inside the cells (Fig. 7 arrows) 25 . In contrast, RLX treatment significantly reduced the expression of Dkk1 in cardiomyocytes (Fig. 7). Thus, at least in part, RLX activates Wnt signaling in both cardiomyocytes and LV tissue through a downregulation of constitutively expressed DKK1.

Interplay Between RLX and Wnt in Fibroblasts
Fibrosis is a major contributor to impaired cardiovascular function and diastolic heart failure and has been shown to increase with age 23 . We have shown that RLX significantly reduces fibrosis in the atria 5 and ventricles of aged rats (Fig. 1). TGFβ significantly increased the differentiation of fibroblasts into myofibroblasts as determined by the increased levels of collagen I and F-actin, and RLX blocked these effects (Fig. 8). The suppression of the differentiation of fibroblasts to myofibroblasts by RLX has been reported in neonate or immortalized cell lines 8,26 Figure 1. Relaxin increases CV and reduces collagen in aged LV. (A) (a) RLX treatment resulted in a significant increase in CV during steady pacing (S1:S1). (b) CV restitution kinetics measured by programmed stimulation with premature pulses (S2 beats) was significantly increased by RLX, particularly during fast pacing times. *Indicates P < 0.05. 20x magnification (B) LV tissue sections were stained with Picro-sirius red and imaged on an Olympus Provis Light Microscope. The collagen to tissue ratio was significantly reduced in RLX (Bb, n = 6) treated animals compared to control (Ba, n = 5) by 23%. (C) LV tissue stained for Nav1.5 showed that RLX (Cb, n = 6) significantly increased Nav1.5 expression compared to control hearts (Ca, n = 5). 600x magnification, scale bars = 25 µm and apply to all panels. (2019) 9:18545 | https://doi.org/10.1038/s41598-019-53867-y www.nature.com/scientificreports www.nature.com/scientificreports/ but these cells do not necessarily respond to RLX in the same manner as adult primary cardiac fibroblasts and cannot be used as evidence of the anti-fibrotic actions of RLX in animals. To determine the role of Wnt signaling on the effects of RLX in cardiac fibroblasts, adult F-344 cardiac fibroblasts were isolated, cultured with RLX and/ or TGFβ and were treated with recombinant Dkk1 for 72 hours, after which, collagen levels were measured by immunofluorescence. TGFβ significantly increased collagen compared to control (Fig. 9a,b) and RLX blocked TGFβ's effects (Fig. 9b). The addition of Dkk1 resulted in a significant reversal of the effects of RLX on collagen expression (Fig. 9g). Dkk1 had no effects on RLX alone (Fig. 9f). These data are summarized in the bar graph Age and relaxin's effect on intercalated disk proteins in LV. In adult (9-months, n = 4) F-344 rats (Aa-c), β-catenin and Cx43 localized to the intercalated disks (arrows), with minimal movement of Cx43 to the lateral membranes (arrow heads). Untreated aged (24-months, n = 4) rats (Ad-f) exhibited a marked reduction in β-catenin and translocation of Cx43 to the lateral membranes. RLX treatment (n = 4) of aged animals (Ag-i) showed that RLX can reverse the effects of aging on β-catenin expression and results in trafficking of Cx43 to the intercalated disk, matching closely that seen in 9-month-old rats. (Aj-k). Quantification of β-catenin/Cx43 expression. 600x magnification, scale bars = 25 µm and apply to all panels. (2019) 9:18545 | https://doi.org/10.1038/s41598-019-53867-y www.nature.com/scientificreports www.nature.com/scientificreports/ shown in Fig. 9d. These findings are consistent with RLX's action in ventricular myocardium. Furthermore, RLX increased the expression of Wnt1 in cardiac fibroblasts (Fig. 10). These data strongly support our hypothesis: RLX increases activation of canonical Wnt signaling, and the effects of RLX on gene expression and TGFβ signaling are significantly inhibited by the canonical Wnt signaling inhibitor Dkk1.

Discussion
Based on intriguing clinical and pre-clinical data, RLX engendered significant enthusiasm as a potential therapy for cardiopulmonary diseases. In the acute heart failure trial RELAX-AHF, RLX treatment (30 µg/kg/day for 2-days, i.v.) improved patient survival by a remarkable 37% in 6 months 27 . These exciting results led the FDA to declare RLX as a "break-through" therapy made all the more significant because the trial included patients with systolic and diastolic HF. Unfortunately, the reduced mortality benefits were not duplicated in a subsequent larger clinical trial (www.novartis.com). Detailed analysis of the larger trial has not been reported and the failure of RLX to significantly reduce mortality is not fully understood. A possible explanation is that the control group of patients receiving standard of care for HF fared considerably better than in earlier trials but another problem has been the design methodology of a 2-days treatment which has justifiably received substantial criticism 28 . Our previous studies on the effects of RLX in experimental animals provide compelling evidence of significant    www.nature.com/scientificreports www.nature.com/scientificreports/ beneficial effects of the hormone in cardiac physiology. We reported that RLX suppressed AF in aged (24-months old) rats by increasing conduction velocity (CV) of atrial action potentials 5 . These effects were linked to increased expression of the voltage-gated sodium channel (Nav1.5) and current (I Na ), and a marked decrease in fibrosis 5 , both effects confirmed here in ventricles. At the cellular level, Relaxin was shown to upregulate I Na in 24-hours     but the reversal of fibrosis required more than a week due to the slow turn-over of collagen in the extracellular matrix 6 . Besides electrical and ECM remodeling, we reported that Relaxin acted as a potent anti-immune and anti-inflammatory agent in the ventricles of aged animals 2 . Here we examine the mechanisms responsible for RLX's functions in the heart. Our results show that RLX's effects are largely mediated by the modulation of canonical Wnt signaling which can act as a master controller of gene expression in heart and other organs.
This conclusion is based on several major lines of evidence: a) GSK3β inhibitors and exogenous Wnt1 protein mimic the effects of RLX in cardiomyocytes; b) recombinant Dkk1, a specific natural inhibitor of canonical Wnt signaling, blocks the effects of RLX on Nav1.5 and collagen I expression; c) Dkk1 expression is reduced by RLX in LV tissue and isolated cardiomyocytes; and d) RLX increases the expression of the Wnt1 ligand in cardiac fibroblasts. The integration of RLX and Wnt signaling was previously suggested by studies in prostate cancer 18 , but the mechanisms described here are very different. Whereas in prostate cancer RLX promotes tumor progression via its effects on the protocadherin PCDHY 18 , the effects we described are mediated, at least in part, by increased expression of Wnt1 and reduced expression of Dkk1.
Wnt signaling in the heart is complex, and different Wnt ligands have distinct effects. Previous studies of the actions of Wnt-signaling focused on isolated neonate rat ventricular myocytes (NRVM) and non-cardiac fibroblasts. In these earlier studies, treatment with Wnt1 and inhibition of GSK3β increased Cx43 expression and trafficking to the ID 15,16 , results that are consistent with those reported here. The role of Wnt signaling in the regulation of Nav1.5 is more controversial. Liang et al. reported that incubation of NRVM with Wnt3a or CHIR-99021 down-regulated Nav1.5 11 , results that were confirmed using the immortalized cardiomyocyte cell line HL-1 14 . In contrast, findings by Asimaki et al. 17 , Chelko et al. 16 and data presented here, show that canonical Wnt signaling and GSK3β inhibition increase Nav1.5 expression in adult, aged, and diseased cardiomyocytes, suggesting that, like Wnt-signaling, RLX signaling is highly dependent on cell type and age.
Finally, many details of the mechanism by which RLX modulates canonical Wnt signaling remain to be explored. RXFP1 has been reported to signal via cAMP-dependent mechanisms downstream of the activation of heterotrimeric proteins of the G s , G i and G o families 29 . Our research, however, suggests that the effects of RLX on Nav1.5 expression are independent of cAMP signaling, since adenylate cyclase and PKA inhibitors did not alter the effects of RLX on the expression of Nav1.5 (data not shown). Furthermore, attempts to show RLX-dependent regulation of cAMP levels in cardiomyocytes using a sensitive colorimetric immunoassay were unsuccessful. Likewise, we failed to see any significant effects of RLX treatment on the levels of phosphorylation and/or nuclear translocation of CREB in isolated cardiomyocytes (not shown). Thus, we concluded that the substantial effects of RLX on Nav1.5 expression are not a consequence of the effects of RLX on cAMP signaling. RLX has also been reported to signal through a variety of pathways, including nitric oxide, PI3 Kinase, or via a relatively uncharacterized pathway that involves a direct interaction of RLX with the glucocorticoid receptor and the latter's subsequent activation 30 . The involvement of GR-dependent pathways is unlikely given that GR signaling increases Dkk1 expression and blocks canonical Wnt signaling 31 . However, we cannot rule out at this point a mechanism mediated by the production of nitric oxide.
Note that there is no simple correlation between rats treated with Relaxin for 2-weeks with isolated cells incubated with Relaxin for 24-48 hours. Delivery of exogenous Relaxin systemically to a rat raises questions of dosing, duration, circulating levels of RLX and its actions at many targets that can alter hemodynamics, immune and inflammatory responses; not just the heart. Relaxin added to cells dissociated with collagenase raises questions about cellular damage by the isolation procedure and the brief survival of primary cell cultures, particularly aged cardiac myocytes which in our hands lasted less than 24 hours and could not be used to test a genomic mechanism. Nevertheless, the cellular experiments show that Relaxin acts at the cellular level to explain most of its actions on the heart and that the genomic changes on Nav1.5, connexin 43 and β-catenin occur within 24 hours. Hence, cell studies offer a reasonable option to elucidate the modes of action of Relaxin.

Methods
Animal study design. Twenty-four-month-old Fischer 344/Brown Norway F1 male rats from the National Institute of Health were separated into RLX (400 µg/ml/day, n = 6) and control (Na + acetate, n = 5) groups and Plasma relaxin concentration. Circulating levels of RLX in the aged rats were determined using a pre-packaged Quantikine ELISA kit (R&D Systems, DRL200) according to manufacturer instructions.
Immunohistochemistry. Tissue sections (7 μm) were treated with 0.1% Triton X-100 followed by block with 2% BSA. Primary antibody was added for 1 hour at room temperature. Guinea Pig and Rabbit anti-Nav1.5 (AGP-008, Alomone Labs, 1:200 in tissue and ab56240, Abcam, 1:200 in cells), rabbit anti-β-catenin (ab32572, Abcam, 1:250), mouse anti-Cx43 (sc-13558, Santa Cruz Biotechnology, Inc, 1:100), rabbit anti-collagen I (ab34710, Abcam, 1:200), mouse anti-Wnt1 (10C8, Thermo Fisher, 1:200) antibodies were used to measure protein expression. Secondary antibodies were applied for 1 hour at room temperature. Immunofluorescence imaging was performed using an Olympus Fluoview 1000 confocal microscope or an Olympus Provis light microscope. Isolated cardiomyocytes and fibroblasts were stained in a similar manner after fixation with 4% paraformaldehyde. In order to ensure quantitative reliability of the results, all samples being compared were immunostained simultaneously with the same antibody preparation, and identical confocal microscope settings were utilized for the recording of all data. NIH ImageJ was used for image analysis. Quantitation of fluorescence labeling was done after subtracting backgrounds using the built-in background subtraction function of ImageJ. To measure the fluorescence intensity of proteins associated to the intercalated disk, the cell-cell contact regions of adjacent cardiomyocytes were identified in the microscopic image, selected, and the intensities measured and recorded using ImageJ tools.  (10), Glucose (5.5), pH 7.0. Heart was perfused for 11-14 minutes with digestion buffer containing: 50 mL Perfusion Buffer with 2 mg/mL Collagenase Type II. Heart was placed in small glass beaker with 3 mL digestion buffer and minced into small pieces with surgical scissors. Additional mincing was conducted by cutting plastic transfer pipettes at 45 degrees at largest diameter, slowing reducing pipette diameter size as tissue becomes solubilized. Stopping buffer (10 mL) was added containing: 45 mL Perfusion Buffer, 5 mL of 10% FBS and 12.5 µM CaCl 2 . Myocytes were transferred to a 50-mL tube through a cell strainer primed with stopping buffer and allowed to pellet (~20 minutes). Myocytes were washed with Calcium re-introduction solutions (100 µM, 400 µM and 900 µM: diluted in stopping buffer) with pelleting allowed between washes. Cells were mixed with plating medium containing: 10% FBS, Blebbistatin (25 µM), HEPEPS (10 mM), ATP (2 mM) and Primocin TM (diluted 50 mg to 500 mL MEM, Invivogen, San Diego, CA. USA). Cells were placed on laminin coated coverslips, in 24 well plates and incubated at 37 °C for 2 hours. Plating medium was replaced by culture medium containing: 0.1% BSA, Blebbistatin (25 uM), HEPES (10 mM), ITS (250 uL of 100x stock) and Primocin TM .
Picro-sirius red stain. Left ventricular sections (7 μm) were washed in Xylene followed by washes in 100% and 95% EtOH. Tissue was placed in Hematoxylin, followed by tap water wash and Picro-Sirius Red application. This was followed by washes with acidic water, 100% EtOH and Xylene. Finally, coverslips were mounted onto slides with Permount. Imaging was performed using an Olympus Provis Light Microscope at 10x magnification. Data is reported as collagen: tissue ratio.
RT-PCR analysis. RNA was isolated (RNAEasy, Qiagen) and copied to cDNA (High Capacity Reverse Transcription kit, Applied Biosystems) according to manufacturer protocols. A Syber-green-based formulation (Absolute Sybr-Green, Thermo Fischer Scientific, Waltham, MA) was utilized for fluorescence-based kinetic real-time PCR using an Applied Biosystems model 7000 detection system (Applied Biosystems Inc., Foster City, CA). Expression levels of RNAs of interest were normalized to that of GAPDH using the ΔΔCt method, and reported relative to the mean of the WTV group. Primer pair sequences (forward and reverse for each target, listed 5′ to 3′) used for RT-PCR are as follows: MMP-2: gcaccaccgaggattatgac, cacccacagtggacatagca; MMP-9: cctctgcatgaagacgacataa, ggtcaggtttagagccacga; αSMA: tggctgatggagtact-tc, gatagagaagccaggatg; Wnt1: cctgcacctgcgactacag, ggttcatgaggaagcgtagg GAPDH: agctggtcatcaatgggaa, atttgatgttagcgggatc.

Statistical analysis.
Comparisons between two groups were done using an unpaired, 2-tailed t-test, or the Mann-Whitney U test. Comparisons of three or more groups were done using ANOVA or Kruskal-Wallis tests. Nuclear β-catenin was compared using Chi-squared. All data is presented as mean ± SEM, unless otherwise noted. Statistical comparisons were performed using Graphpad Prism software. A value of P < 0.05 was considered to be statistically significant.
Ethical approval and informed consent. 1. An approval of experimental protocols is stated in the Methods section from the University of Pittsburgh. 2. All methods were carried out in accordance with NIH guidelines and the animal usage committee of the University of Pittsburgh. 3. There was no informed consent as this did not involve patients or human tissues.

Data availability
All data relevant to the manuscript are available for the reviewers to examine and should be manuscript be accepted for publication in Scientific Reports the authors assert that they submit these data in a supplemental data manuscript, or it will be added to the supplement.