Ligand-activated RXFP1 gene therapy ameliorates pressure overload-induced cardiac dysfunction

Abstract

Recurrent episodes of decompensated heart failure (HF) represent an emerging cause of hospitalizations in developed countries with an urgent need for effective therapies. Recently, the pregnancy-related hormone relaxin (RLN) was found to mediate cardio-protective effects and act as a positive inotrope in the cardiovascular system. RLN binds to the RLN family peptide receptor 1 (RXFP1), which is predominantly expressed in atrial cardiomyocytes. We therefore hypothesized that ventricular RXFP1 expression might exert potential therapeutic effects in an in vivo model of cardiac dysfunction. Thus, mice were exposed to pressure overload by transverse aortic constriction and treated with AAV9 to ectopically express RXFP1. To activate RXFP1 signaling, RLN was supplemented subcutaneously. Ventricular RXFP1 expression was well tolerated. Additional RLN administration not only abrogated HF progression but restored left ventricular systolic function. In accordance, upregulation of fetal genes and pathological remodeling markers were significantly reduced. In vitro, RLN stimulation of RXFP1-expressing cardiomyocytes induced downstream signaling, resulting in protein kinase A (PKA)-specific phosphorylation of phospholamban (PLB), which was distinguishable from β-adrenergic activation. PLB phosphorylation corresponded to increased calcium amplitude and contractility. In conclusion, our results demonstrate that ligand-activated cardiac RXFP1 gene therapy represents a therapeutic approach to attenuate HF with the potential to adjust therapy by exogenous RLN supplementation.

Graphical abstract

The hormone relaxin and its receptor RXFP1 mediate several beneficial cardiovascular effects. This report demonstrates that AAV-mediated expression of RXFP1 combined with relaxin administration rescues ventricular function in a pressure-overload model. The results suggest that ligand-activated cardiac RXFP1 gene therapy might represent a therapeutic approach to attenuate heart failure.

Introduction

Heart failure (HF) represents the final common endpoint of advanced cardiac disease and is a major cause of morbidity and mortality worldwide.¹ HF is characterized by insufficient cardiac output with a maladaptive neurohumoral response. Increased catecholamine and mineralocorticoid plasma concentrations, as well as activation of the renin-angiotensin-aldosterone system, all aim at temporarily increasing cardiac output. This compensatory mechanism allows for short-term stabilization of cardiac workload. However, with disease progression, fluctuations of cardiac function occur. These so-called decompensations are usually characterized by edema and pulmonary congestion, often resulting in hospitalizations for oxygen supplementation and intravenous (i.v.) diuretic therapy. With clinical improvements in HF therapy, these advanced disease stages become more and more frequent, rendering decompensated HF one of the most prevalent causes of hospital admissions in the western hemisphere.²

The class I recommended pharmaceutical HF therapy comprises β-adrenergic receptor blockers, angiotensin-converting enzyme inhibitors (ACEis)/angiotensin receptor blocker (ARB), and mineralocorticoid receptor antagonists (MRAs), all of which reduce overall HF progression and HF hospitalizations.³ However, in acute decompensation they might be detrimental due to their dampening effect on blood pressure and renal perfusion. The recently emerging angiotensin receptor-/neprilysin inhibitors (ARNIs) take advantage of endogenous compensatory mechanisms by increasing the natriuretic peptide (NP) plasma concentrations.⁴ Although ARNIs pave the way for preventing fluid retention and thus decompensations, there is still no suitable approach to increase cardiac workload when cardiac decompensation develops.^5,6

Relaxin (RLN) is a peptide hormone traditionally linked to maternal adaptation of the cardiovascular system during pregnancy and has been proposed as a pleiotropic and cardioprotective hormone.⁷ RLN exerts its physiological effects by binding to its cognate receptor—RLN family peptide receptor 1 (RXFP1)—a G protein coupled receptor (GPCR).⁸ As a GPCR, RXFP1 shares various downstream targets with “classical” GPCRs like the β-adrenergic receptors (β-ARs). However, there are certain critical differences.

At higher (i.e., nanomolar) RLN concentrations, RXFP1 couples to three distinct Gα proteins—Gα_s, Gα_oB, and Gα_i. Activation of Gα_s stimulates adenylyl cyclase (AC) and, similar to β-AR, increases intracellular cyclic AMP (cAMP).⁹ By activating protein kinase A (PKA), increased cAMP results in phospholamban (PLB) phosphorylation at serine 16, dissociation from SR Ca²⁺-ATPase (SERCA), and eventually positive inotropy.¹⁰ This effect is negatively modulated by Gα_Ob.^9,11,12 RXFP1 also uniquely couples to Gα_i, which stimulates a delayed second surge of cAMP in a Gβγ-phosphatidylinositol 3-kinase (PI3K)-protein kinase ζ (PKC-ζ)-AC5-dependent manner,^9,13, 14, 15, 16 ultimately resulting in a prolonged, bi-phasic cAMP response.¹⁴ In contrast, stimulation of β-AR leads to Gα_s-dependent cAMP accumulation, which has been shown to promote both PKA and CaMKII activation.17, 18, 19 CaMKII phosphorylates a number of cellular targets, including PLB at threonine 17 (PLB[T17]) and the Ryanodin receptor (RyR), and a growing body of evidence suggests that CaMKII activation is linked to maladaptive cardiac remodeling.²⁰

Similar to NPs, RLN is believed to be released from the failing myocardium into circulation.²¹ Efforts aimed at enhancing potential cardioprotective effects by administration of recombinant RLN in acute HF confirmed safety but failed in clinically relevant endpoints.²² RLN-mediated positive inotropic effects were first identified in isolated atrial cardiomyocytes from rats and later confirmed in atrial cardiomyocytes from explanted failing human hearts.^23,24 However, in contrast to atrial cells, ventricular cardiomyocytes from the same human hearts did not respond to RLN, which can be well attributed to the predominant atrial expression of RXFP1.^24,25

Considering these unique properties of RLN, we assumed a benefit from RXFP1 signaling in ventricular cardiomyocytes.^7,26 Therefore, we hypothesized that ectopic ventricular expression of RXFP1 using a gene therapeutic approach could enhance beneficial RLN effects.

By supplementation of exogenous RLN via subcutaneous administration, we sought to establish a non-β-AR approach for an adjustable and safe inotropic therapy in HF.

Results

Expression of RXFP1 is limited to atrial myocardium in human and mice

Expression of RXFP1 has been reported to be mainly restricted to atrial cardiomyocytes in humans. To verify this observation, we isolated total RNA from human and murine atrial and ventricular tissues. In mouse hearts, Rxfp1 was almost exclusively detectable in the atria with only minimal murine Rxfp1 mRNA in ventricular samples (Figure 1A). This pattern was similar in human specimens, where we found a solid human RXFP1 expression in atria, while RXFP1 mRNA was below the detection limit in ventricular samples (Figure 1B). In both species, RXFP1 expression in ventricles was not altered in failing left ventricle (LV) myocardium compared to non-failing myocardium (Figures 1A and 1B). In contrast to this, atrial natriuretic peptide (ANP) and brain natriuretic peptide (BNP), two typical markers associated with increased wall tension as observed in HF, were found to be significantly elevated in samples obtained from failing human and murine LV (Figures 1C–1F).

Figure 1.

Native RXFP1 expression in human and mouse heart

(A and B) Mouse Rxfp1 (A) and human RXFP1 (n = 8–10) (B) mRNA expression. (C–F) Mouse Anp (C), Bnp (n = 7–46) (D), and human ANP (E) and BNP (n = 8–10) (F) mRNA expression. RXFP1, ANP, and BNP mRNA expression quantification by qRT-PCR; HPRT1 was used as a reference gene. mRXFP1/huRXFP1 cDNA served as a positive control. Significant difference reported in (A)–(F) was determined by Kruskal-Wallis one-way ANOVA test with Dunn’s multiple comparison test. Values represent the mean ± SEM and ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001.

AAV-based overexpression of RXFP1 in isolated ventricular cardiomyocytes induces a dose-dependent cAMP accumulation upon RLN stimulation

With the aim to test the translational potential of a ligand-activated gene therapy, adeno-associated viral vectors (AAV) harboring the rat-derived Rxfp1 cDNA (AAV.rRXFP1) were generated. rRXFP1 was chosen to avoid adverse effects of introducing a foreign protein into neonatal rat ventricular cardiomyocytes (NRVCMs). An identical AAV carrying a luciferase (AAV.Luc) transgene served as control. In order to target NRVCMs, both transgenes were expressed under the control of a cytomegalovirus-enhanced ventricular myosin light chain promoter 2 (CMVenh-MLC2v). For in vitro transduction of NRVCMs, AAV serotype 6 (AAV6) was used. Increasing viral dosages produced a corresponding rise in transgene expression (Figure 2A) and moreover an expression-dependent cAMP response upon RLN stimulation (Figure 2B). At a multiplier of infection (MOI) of 10,000 viral genomes (vg) per cell, AAV6.rRXFP1 allowed for a 12-fold mRNA overexpression compared to endogenous expression of rRXFP1 mRNA in atrial cardiomyocytes from neonatal rats (NRACMs; Figure 2C). RLN stimulation resulted in a robust cAMP response (Figure 2D) and a significantly elevated phosphorylation of PLB at serine 16 (P-PLB[S16]), suggesting that the AAV6.rRXFP1 treatment results in the expression of functional rRXFP1 receptors. Isoproterenol (ISO), which was used as a positive control, induced a strong β-AR-mediated PLB(S16) phosphorylation (Figures 2E and 2F).

Figure 2.

Expression of rat-derived RXFP1 (rRXFP1) in combination with RLN stimulation induces cAMP-dependent signaling in isolated ventricular cardiomyocytes

(A) MOI-dependent rRXFP1 mRNA expression in transduced NRVCMs after 5 days (n = 3–6). (B) MOI-dependent cAMP response. NRVCMs were transduced for 5 days and stimulated with 100 nM of RLN (n = 3). (C) Transduced versus endogenous rRXFP1 mRNA expression. Transduced NRVCMs compared to native Rxfp1 mRNA expression in atrial cardiomyocytes (NRACMs) quantified by qRT-PCR with Hprt1 as a reference gene (n = 7–14). (D) cAMP production in response to increasing concentrations of RLN. NRVCMs were transduced with rRXFP1 virus for 5 days and stimulated with different dosages of RLN ranging from 1 pM to 100 nM and changes in cAMP production were measured (n = 6). (E and F) P-PLB(S16)/PLB from rRXFP1 overexpression and RLN stimulation. Transduced NRVCMs were stimulated with 100 nM RLN or 10 nM ISO and P-PLB(S16) was quantified by immunoblot; glyceraldehyde-3-phosphate dehydrogenase (GAPDH) immunodetection was used as an internal control (n = 3). Significant difference reported in (A) and (C) was determined by Kruskal-Wallis one-way ANOVA test with Dunn’s multiple comparison test, (B) was determined by ordinary one-way ANOVA test with Dunnett’s multiple comparison test, (D) was determined by two-way ANOVA with Sidak’s multiple comparisons test, and (E) was determined by two-way ANOVA with Tukey’s multiple comparisons test. (C–F) An MOI of 10,000 vg/cell was used. Values represent the mean ± SEM; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001.

RXFP1 improves calcium handling in cardiomyocytes

In order to evaluate potential positive inotropic effects of RXPF1-RLN gene therapy, calcium transients were measured. Isolated NRVCMs were transduced with AAV6.rRXFP1 or control virus (AAV6.Luc) and stimulated with RLN or saline. Compared to cells treated with the control virus, RLN stimulation of AAV6.rRXFP1-treated cells resulted in a significant increase (approximately 50%) in calcium transient amplitude (Figures 3A and 3B). Diastolic calcium, on the other hand, did not differ between the groups. mRNA extracted from the cells confirmed a robust rRXFP1 expression in AAV6.rRXFP1-treated cells (Figure 3C). ISO stimulation, which was used as a positive control, led to a strong increase in both calcium amplitude and diastolic calcium, independent of viral vector used (Figure 3D).

Figure 3.

rRXFP1-RLN treatment improves calcium handling in isolated cardiomyocytes

(A–C) Ca²⁺ transient amplitude (A), representative Ca²⁺ transients (B), and diastolic Ca²⁺ (C) after 5 min of saline, RLN, or ISO treatment (n = 3–8). NRVCMs were transduced with AAV6.rRXFP1 or AAV6.Luc (10,000 vg/cell) for 5 days. Ca²⁺ transients were measured with 1 Hz electrical stimulation. (D) rRXFP1 mRNA expression quantification by qRT-PCR, Hprt1 was used as a reference gene. Significant differences reported in (A), (C), and (D) was determined by ordinary one-way ANOVA test with Tukey’s multiple comparison test. Values represent the mean ± SEM; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001.

To test the cardiac effects of ventricular rRXFP1 expression and RLN administration in vivo, we used AAV9 harboring the aforementioned constructs, which show superior cardiac tropism in rodents. To test for hemodynamic effects of RLN-RXFP1 treatment, mice received these viral vectors via tail-vein injection. 4 weeks later, the animals were injected with RLN 3 h before performing left-ventricular pressure measurements. Compared to animals treated with control vector, dP/dt maximum (dP/dt max) significantly increased in animals treated with both AAV9.rRXFP1 and RLN (n = 9–10; Figure S1A). Heart rate (HR) and dP/dt minimum (dP/dt min), on the other hand, were unaffected (Figures S1B and S1C). Successful viral transduction of cardiac tissue was confirmed by post-mortem analysis of rRXFP1 mRNA expression in AAV9.rRXFP1-treated animals (Figure S1D). In line with the increase in dP/dt max, P-PLB(S16)/PLB ratio was elevated in AAV9.rRXFP1-treated animals (Figure S1E). RLN plasma levels were comparable in all animals (Figure S1F).

Ligand-activated rRXFP1 gene therapy ameliorates cardiac dysfunction and attenuates remodeling in failing mouse myocardium

In order to test the therapeutic efficacy of AAV-mediated ventricular expression of rRXFP1 in vivo, the optimal dose required for significant expression of the receptor (AAV9.rRXFP1, AAV9.Luc) and stable RLN plasma concentration were determined by independent dose escalation studies. At a dosage of 1 × 10¹² vg/animal robust rRXFP1 expression was detected and thus chosen for subsequent in vivo studies (Figure S2A). RLN applied at a dosage of 0.675 μg/g/day using osmotic mini pumps resulted in a stable RLN plasma concentration of approximately 20 ng/mL (Figure S2B).

Effects of rRXFP1 gene therapy were evaluated in a model of pressure overload-induced HF. Transverse aortic constriction (TAC) was induced using a 26-gauge (G) needle resulting in a moderate aortic stenosis and thus gradually progressing HF phenotype. This enabled for post-TAC AAV transduction and allowed the transgene to reach stable expression before initiation of chronic RLN supplementation. 7 days after TAC, AAVs (AAV9.rRXFP1, n = 29; AAV9.Luc, n = 28) were administered via tail vein injection. To ensure continuous stimulation of RXFP1, we implanted mini osmotic pumps containing either RLN (AAV9.rRXFP1, n = 15; AAV9.Luc, n = 14) or saline (AAV9.rRXFP1, n = 14; AAV9.Luc, n = 14) 21 days after AAV delivery, when stable transgene expression could be expected, and continuously delivered RLN for 4 weeks (Figure 4A). Until day 28 post TAC, fractional shortening (FS) had decreased in all TAC-operated animals (Figures 4B and 4C). 4 weeks after pump implantation, FS increased significantly in the TAC/rRXFP1/RLN group, while the other TAC-operated groups presented a progressive decline in FS. Pronounced left ventricular dilation (left ventricular inner diameter in diastole; LVIDd) was detectable in TAC/Luc animals independent of RLN supplementation, as well as in TAC/rRXFP1/saline mice while this LV dilation was not observed in TAC/rRXFP1/RLN mice (Figure 4D). Again, all TAC-operated groups showed a significant hypertrophic response (LVAWd) without detectable impact of AAV9.rRXFP1 transduction or RLN administration (Figure 4E). In line with these changes, global longitudinal strain (GLS) at day 55 after surgery was impaired in TAC/Luc animals while TAC/rRXFP1/RLN animals presented nearly normal values (Figures S3A–S3F). HR was not significantly different between treatment groups (Figure 3F). Assessment of reverse longitudinal strain rate (rLSR), a parameter of diastolic function that correlates to the time constant of LV pressure decay (tau)^27,28 indicates maintained relaxation in TAC/rRXFP1/RLN mice (Figures S3G and S3H).

Figure 4.

rRXFP1 gene therapy with chronic administration of RLN improves cardiac function in TAC-induced HF

(A and B) Experimental design (A) and representative M-mode echocardiogram recordings (B) of experimental groups on day 55. (C–F) Fractional shortening (FS) (C), left ventricular internal diastolic diameter (LVIDd) (D), left ventricular diastolic anterior wall thickness (LVAWd) (E), and heart rate (HR) (F) of sham mice and TAC mice injected with 1 × 10¹² vg/animal of rRXFP1 virus 1 week after operation (n = 8–9 in sham and n = 14–15 in TAC groups). 4 weeks after TAC operation, TAC and sham animals were implanted with either saline pumps or RLN pumps. (G) rRXFP1 mRNA expression in the ventricle after 7 weeks of rRXFP1 gene therapy treatment compared to the native Rxfp1 expression in the atria of sham mice. (H) Bnp mRNA expression quantification by qRT-PCR; Hprt1 was used as a reference gene. (I) cardiac P-PLB(S16)/PLB quantification by immunoblot; GAPDH immunodetection was used as an internal control. (J) RLN plasma concentrations of study groups measured at day 55. Significant difference reported in (C) and (F) was determined by two-way ANOVA with Tukey’s multiple comparisons test, (G) and (H) was determined by Kruskal-Wallis one-way ANOVA test with Dunn’s multiple comparison test, and (I) and (J) was determined by ordinary one-way ANOVA test with Tukey’s multiple comparison test. A list with all group comparisons can be found in Data S1. For (C)–(F) ∗, TAC/Luc/saline versus TAC/rRXFP1/RLN; #, TAC/Luc/RLN versus TAC/rRXFP1/RLN; §, TAC/rRXFP1/saline versus TAC/rRXFP1/RLN; $, TAC/Luc/saline versus TAC/rRXFP1/saline; values represent the mean ± SEM; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001.

Analysis of ventricular mRNA expression confirmed significant levels of rRXFP1 mRNA in TAC animals treated with AAV9.rRXFP1 (Figure 4G). Bnp mRNA expression was highest in TAC mice receiving AAV9.Luc without RLN supplementation and moderately reduced upon AAV9.rRXFP1 transduction or RLN administration. Notably, the combination of both AAV9.rRXFP1 and RLN significantly reduced Bnp levels (Figure 4H). Conversely, the ratio of P-PLB(S16)/PLB was highest in animals that received both AAV9.rRXFP1 and RLN (Figure 4I; Figure S4). Plasma samples taken at study end verified comparable RLN plasma levels in all RLN-treated animals (Figure 4J). Also, in line with these changes, lower Anp and β myosin heavy chain (Myh7) was measured in TAC/rRXFP1/RLN mice (Figure S5). Moreover, fibrotic area and expression of fibrotic marker genes Collagen-1 (Col1a1), -3 (Col3a1), and Periostin (Postn) were reduced in TAC/rRXFP1/RLN mice (Figure S6), indicating that a combined treatment consisting of AAV-mediated expression of rRXFP1 and continuous RLN administration is able to attenuate cardiac dysfunction and cardiac remodeling.

Both rat and human RXFP1 in combination with RLN administration attenuate HF progression in mice

We established an AAV harboring the human RXFP1 cDNA sequence (AAV.huRXFP1) to test whether huRXFP1 exhibits similar therapeutic effects and further translational potential. To compare rRXFP1 and huRXFP1, we transduced NRVCMs with these AAVs and treated them with RLN in vitro, which confirmed comparable phosphorylation at PLB(S16) upon RLN stimulation (Figures 5A and 5B).

Figure 5.

huRXFP1 gene therapy with chronic administration of RLN improves cardiac function in TAC-induced HF

(A and B) In vitro, P-PLB(S16)/PLB upon stimulation of rRXFP1 and huRXFP1. NRVCMs were transduced with AAV6.rRXFP1 and AAV6.huRXFP1 virus using a MOI of 10,000 vg/cell, after 5 days the cells were stimulated with 100 nM of RLN for 5 h. GAPDH was used as an internal control. (C–D) In vivo huRXFP1 gene therapy experiment, (C) in vivo experimental design, and (D) representative M-Mode echocardiogram recordings of experimental groups on day 55. (E–H) FS (E), LVIDd (F), LVAWd (G), and HR (H) of sham and TAC mice injected with 1 × 10¹² vg/animal huRXFP1 virus 1 week after operation (n = 6–7 in sham and n = 12–21 in TAC groups). Saline or RLN pumps were implanted into the animals 4 weeks after operation. (I) huRXFP1 mRNA expression in the ventricle 7 weeks after huRXFP1 virus administration compared to native Rxfp1 expression in the atria of sham mice. (J) Bnp mRNA expression quantification by qRT-PCR; Hprt1 was used as a reference gene. (K) RLN plasma concentrations of study groups measured at day 55. Significant difference reported in (A) was determined by ordinary one-way ANOVA test with Tukey’s multiple comparison test, (E)–(H) was determined by two-way ANOVA with Tukey’s multiple comparisons test, and (I)–(K) was determined by Kruskal-Wallis one-way ANOVA test with Dunn’s multiple comparison test. A list with all group comparisons can be found in Data S1. For (E)–(H) ∗, TAC/Luc/saline versus TAC/huRXFP1/RLN; #, TAC/Luc/RLN versus TAC/huRXFP1/RLN; §, TAC/huRXFP1/saline versus TAC/huRXFP1/RLN; $, TAC/Luc/saline versus TAC/huRXFP1/saline; values represent the mean ± SEM; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001.

To evaluate huRXFP1 gene therapy (AAV9.huRXFP1, n = 39; AAV9.Luc, n = 29), we adopted both HF model and study protocol from the earlier rRXFP1 study (Figure 5C). Cardiac function was assessed by echocardiography at baseline and 7, 28, and 55 days post TAC operation (Figure 5D). Until day 28 post TAC, FS decreased in all TAC-operated animals. Upon implantation of osmotic pumps, a significant improvement of FS was detected in the TAC/huRXFP1/RLN animals (n = 22) compared to the constant decline in AAV9.Luc-treated mice, independent of RLN supplementation (Figure 5E). In contrast to the rat receptor isoform, LV dilation was not significantly different (Figure 5F). LVAWd presented a comparable increase in all TAC groups (Figure 5G). HR was similar between treatment groups (Figure 5H).

Analysis of ventricular mRNA confirmed a robust huRXFP1 mRNA expression in animals treated with AAV9.huRXFP1 (Figure 5I). A significantly lower Bnp mRNA expression was again observed in TAC/huRXFP1/RLN mice compared to TAC/Luc/RLN animals (Figure 5J). Treatment with RLN increased RLN plasma levels to a similar extent in all RLN groups (Figure 5K).

RXFP1 induces positive inotropy through a PKA-dominant pathway distinct from β-adrenergic signaling

PKA is the central hub for both RXFP1- and β-AR-mediated signaling. In order to evaluate differences in the downstream signaling pathways of these two GPCRs, NRVCMs were transduced with AAV6.rRXFP1 and treated with saline, RLN, or ISO. RXFP1 activation by RLN induced a gradual increase in P-PLB(S16) compared to saline and β-AR stimulation. A significant increase in P-PLB(S16) was detected after 10 min of stimulation with RLN and the signal was maintained throughout the experiment (Figure 6A). Additionally, a slow increase in CaMKII-dependent P-PLB(T17) was observed over time without becoming significant during the experiment (Figure 6B). On the other hand, β-AR activation induced a sudden and significant surge of both P-PLB(S16) and P-PLB(T17) compared to both saline and RLN stimulation (Figures 6A and 6B).

Figure 6.

RXFP1 stimulation induces PKA-dependent positive inotropic effects

Before conducting the experiments NRVCM were transduced with AAV6.rRXFP1. (A and B) P-PLB(S16)/PLB (A) and P-PLB(T17)/PLB (B) after prolonged activation of rRXFP1 and β-AR (n = 3–5). rRXFP1-expressing NRVCM were stimulated with RLN (100 nM) or ISO (10 nM) for 10 and 30 min, 1, 3, 6, and 12 h and harvested to quantify P-PLB(S16)/PLB by immunoblot. GAPDH was used as an internal control. (C and D) rRXFP1-expressing NRVCMs were pre-treated with KN93 (CaMKII-inhibitor) for 30 min and stimulated with RLN or ISO for 30 min. P-PLB(S16)/PLB (C) and P-PLB(T17)/PLB (D) were quantified by immunoblot; GAPDH was used as an internal control. (E and F) rRXFP1-expressing NRVCMs were pre-treated with (E) melittin (1 μM) for Gα_s and (F) H89 (5 μM) for PKA inhibition and stimulated with RLN for 30 min. P-PLB(S16) was quantified by immunoblot (n = 8–9). α-Actinin was used as an internal control. (G and H) rRXFP1-expressing NRVCMs were pre-treated with the β-blocker propranolol (1 μM) for 30 min and stimulated with RLN or ISO for 30 min. P-PLB(S16)/PLB was quantified by immunoblot. GAPDH was used as an internal control. Significant difference reported in (A)–(G) was determined by two-way ANOVA with Tukey’s multiple comparisons test. For (A) and (B) ∗, rRXFP1 RLN versus rRXFP1 saline; #, rRXFP1 ISO versus rRXFP1 saline; §, rRXFP1 RLN versus rRXFP1 ISO; values represent the mean ± SEM; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001.

To further investigate the influence of CaMKII activation in RXFP1 and β-AR signaling, we inhibited CaMKII activity with KN93 in AAV6.rRXFP1-transduced NRVCMs followed by either RLN or ISO stimulation. In the RLN-stimulated group, inhibition of CaMKII did not alter P-PLB(S16) or P-PLB(T17) levels (Figures 6C and 6D). In contrast, inhibition of CaMKII in the ISO-stimulated group produced a significant decrease in P-PLB(T17), but no effect on P-PLB(S16) was observed. These results suggest a PKA-dominant role in RXFP1 signaling.

To evaluate the contribution of Gα_s and PKA to the RXFP1 signaling cascade, we inhibited the Gα_s subunit of the heterotrimeric G-protein by melittin before stimulation of RXFP1 with RLN. This inhibition resulted in a significant decrease in P-PLB(S16) signal (Figure 6E). In line with Gα_s inhibition, PKA inhibition with H89 significantly decreased P-PLB(S16) (Figure 6F). Therefore, Gα_s and PKA are crucial downstream signal mediators in the RXFP1 signaling cascade. Lastly, blockage of β-AR with the unspecific β-blocker propranolol did not influence RXFP1 activation. Similar levels of P-PLB(S16) were detected with or without the presence of propranolol in AAV6.rRXFP1-transduced NRVCMs treated with RLN (Figures 6G and 6H).

Discussion

The pregnancy hormone RLN and its cognate receptor, RXFP1, exert distinct effects in the cardiovascular system by increasing cAMP production, which results in anti-fibrotic, anti-apoptotic, and inotropic responses. We found RXFP1 to be predominantly expressed in the atrial myocardium of rodents and humans. These findings are in accordance with previous studies, where RLN-dependent positive inotropy was exclusively detected in atrial cardiomyocytes.^12,13,16 We therefore hypothesized that ventricular expression of RXFP1 could mediate beneficial effects in the failing ventricular myocardium as well.

A cardio-specific, AAV-based expression system was generated and successfully tested on NRVCMs where AAV transduction achieved a robust rRXFP1 expression. Additional RLN stimulation led to a dose-dependent increase in the central downstream messenger cAMP, thereby confirming an intact coupling to the RXFP1 signaling pathway. On the molecular level, we found a rise in PLB(S16). This observation is consistent with a previous publication that found a similar effect in RLN-treated atrial cells with endogenous RXFP1 expression.²⁴ The potential impact on Ca²⁺ handling²⁹ led us to conduct Ca²⁺ transient measurements, which revealed a moderate but significant increase in Ca²⁺ amplitude in AAV6.rRXFP1-transduced and RLN-stimulated cells, which is a well-established link to positive inotropy.^30,31 In vivo, hemodynamic measurements following intraperitoneal RLN injection revealed significantly increased contractility in AAV9.rRXFP1-transduced mice compared to control animals (Figure S1). This further supports the notion that RXFP1 signaling in ventricular cardiomyocytes induces positive inotropy. Yet, the magnitude of contractility rise we observed in vivo could still be hampered from interference with β-AR stimulation in stressed animals or limited transgene expression within the cardiomyocytes.

As the primary focus of this project was to evaluate translational potential, this ligand-activated cardiac gene therapy was assessed in vivo in a murine HF model. We chose a moderate TAC model to ensure gradual HF development and the AAV9 as vector for transduction. The study was designed to approximate the later clinical situation where a patient is admitted to the clinic due to a preceding disease. Thus, we chose to deliver AAV after exposure to pressure overload. RLN supplementation in turn was initiated 3 weeks after AAV transduction when stable transgene expression can be expected.^32,33 Both rat (rRXFP1) and human RXFP1 (huRXFP1) isoforms in combination with exogenous human RLN supplementation were examined with similar results. In the TAC model, animals expressing these receptor isoforms and supplemented with RLN presented not only attenuated HF but even recovery of cardiac function with nearly normalized systolic function compared to sham-operated mice. Moreover, progression to overt HF with progressive LV-dilation as observed in control animals was absent in RXFP1/RLN mice, emphasizing the potential of RXFP1/RLN. Of interest is the observation that RXFP1 expression per se showed a trend toward attenuated HF development, but only supplementation of RLN could fully abrogate HF. This suggests an endogenous cardio protection mediated by RXFP1 expression, which was further supported by fetal gene regulation. The correlation of myocardial BNP mRNA expression to the severity of HF and BNP plasma levels is an established observation in HF.³⁴ The ectopic ventricular expression of either rRXFP1 or huRXFP1 in combination with RLN abrogated myocardial Bnp mRNA upregulation. We found a trend toward attenuated cardiac fibrosis in RLN receiving mice (Figure S6), which is well in accordance with current knowledge on RLN signaling.⁷ Yet, RXFP1/RLN could reduce fibrotic area and pro fibrotic marker genes beyond RLN supplementation alone. The maintained rLSR in RXFP1/RLN mice points toward a preserved diastolic function, which further supports this notion. However, the impact of RLN/RXPF on fibrotic remodeling and diastolic function will need further investigation to distinguish between secondary effects due to improved cardiac performance and direct effects on myocytes and myocyte/fibroblast crosstalk.

Of note, the RLN plasma levels detected in both in vivo experiments were equivalent to the mean plasma levels in several human clinical trials.³⁵ Interestingly, we could not detect positive chronotropic effects upon RLN stimulation. This is to some degree controversial to previous publications,36, 37, 38 which suggested an increase in HR in perfused rat hearts exposed to RLN. However, as these data originated from ex vivo analysis, the previously described positive chronotropy could well be a phenomenon only detectable in denervated heart tissue without sympathetic and parasympathetic interference.

To gain insight into the mechanisms underlying the beneficial effects of RXFP expression, we assessed the downstream signaling cascade of the RXFP1-RLN signaling pathway in vitro. In cardiomyocytes, cAMP accumulation has been shown to modulate gene expression, mitochondrial function, and excitation-contraction coupling, depending on its micro-domain.³⁹ cAMP is known to activate PKA, which in turn phosphorylates specific downstream targets including PLB. In vitro, we could confirm RLN-dependent rise in cAMP and subsequent PLB phosphorylation at the PKA-specific S16 phospho-site, which is known to influence calcium handling in cardiomyocytes. This observation is consistent with our in vivo findings, suggesting RLN-RXFP1-PKA-PLB as the responsible pathway for RXFP1-mediated positive inotropy. Apart from P-PLB(S16) and calcium,^40,41 the increase in cAMP production had no effect on other classical downstream signaling proteins in the RXFP1-RLN signaling pathway including MAPKs,⁴² ERK1/2,^43,44 and AKT.44, 45, 46 RLN treatment of highly pure ventricular cardiomyocytes transduced with RXFP1 did not result in significant phosphorylation of these proteins. This suggests that activation of these pathways might be highly cell-type-specific as data for these pathways were generated in different in vitro models (Figure S7).

Furthermore, our experiments show that both the Gα_s subunit of the heterotrimeric G-protein and PKA play a crucial role in the signaling cascade of RXFP1 in NRVCMs. In addition, the signal from this branch seems to exceed the signal from the Gα_i branch of the cascade, since P-PLB(S16) signal was significantly diminished by Gα_s and PKA inhibition, but not by inhibiting either Gα_i or PI3K (Figure S8). This contradicts data from previous studies, which suggested RXFP1 to primarily signal through Gα_i and PI3K.^13,47 However, this discrepancy could be explained by the different cell types used in those studies and may not be comparable to our ventricular cardiomyocyte model. Finally, our data highlight that the Gα_i branch plays only a subsidiary role in AAV6.rRXFP1-transduced NRVCMs.

Although RXFP1 roughly resembles the classical β-adrenergic signaling, we found that temporal activation, amplitude, and specificity of downstream targets differed considerably among the two pathways. RLN and ISO both act as positive inotropic agents when coupled to their respective receptors. They signal through Gα_s and utilize cAMP as a second messenger to activate PKA, which leads to an increase in phosphorylation of PLB. β-adrenergic activation by ISO resulted in an excessive peak of PLB phosphorylation at both PLB(S16) and PLB(T17) in NRVCM, which can be attributed to PKA and CamKII activation. This strong activation of downstream targets is well known and explains the robust effects of catecholamine therapy and its narrow therapeutic window. In contrast, RXFP1 stimulation led to a predominant PKA activation, resulting in a significant rise in phosphorylation of PLB(S16). However, the slope of the rise and the peak-plateau levels were less pronounced compared to ISO. Interestingly, RXFP1 activation did not induce a significant increase in phosphorylation of PLB(T17), supporting the notion that a PKA-dominant signaling is driving RXFP1-induced inotropy. This is substantiated by data based on chemical inhibition of CaMKII, which could significantly reduce PLB(T17) phosphorylation in ISO-treated, but not in RLN-treated RXFP1-expressing NRVCM.

On the level of Ca²⁺ we found no change in diastolic Ca²⁺ in RLN-stimulated NRVCM, which indicates that the RXFP1-induced peak in Ca²⁺ amplitude does not exceed the cellular capacities for calcium handling. Thus, the positive inotropy of RXFP1 pathway might be less disruptive compared to β-AR. In accordance with the observation of attenuated HF in RXFP1 mice without RLN supplementation, RXFP1-transduced NRVCMs showed a trend toward increased Ca²⁺ amplitude even without RLN. This possible endogenous receptor activity will need further analysis but goes beyond the scope of this project.

Lastly, this study also confirms that RXFP1 activation is independent of β-adrenergic activation, since blocking β-AR had no effect on RXFP1 signaling. Hence, we demonstrated how RXFP1 represents a signaling pathway independent from classical β-AR with a distinct activation pattern of the shared downstream targets. Due to its more gradual and more PKA-specific signaling profile, chronic activation results in sustained effects, which attenuated HF in a murine model.

We envision this therapeutic approach for patients with chronic systolic HF and recurrent decompensations. It is well established how each episode promotes further deterioration of cardiac performance.⁴⁸ In these patients, a moderate increase in systolic function can be sufficient to prevent decompensation and hospital submission. Clinically, AAV administration could be performed upon hospitalization for HF if other reversible reasons have been excluded. With RLN having structural similarities to insulin, various forms of application are conceivable. Besides i.v. application, which has already been evaluated in humans,²² subcutaneous administration proved feasible in our study. Thus, subcutaneous injection in an outpatient environment or even self-administration of RLN for a defined time frame could be possible. This might prevent hospitalization and thus disease progression for the chronic HF patient.

In conclusion, ectopic RXFP1 expression is able to mediate positive inotropy via PKA-specific PLB phosphorylation. Activation of this pathway could be successfully harnessed as a HF therapy using vector-based gene transfer of RXFP1, which allows for a stable and safe expression of the receptor. Recombinant RLN is well tolerated and can be easily administered to make use of the beneficial effects of RXFP1-RLN interaction, as demonstrated in this study. Moreover, the dosing strategy of RLN can be adjusted to meet the clinical need and can correct for varying gene transfer efficacies that are inevitably associated with current gene transfer methods. In this regard, this study describes a ligand-activated cardiac gene therapy using RXFP1 and RLN and presents a novel and unique therapeutic approach for HF treatment.

Materials and methods

Ethics statement

For the collection of human blood and tissue samples, the study protocol was approved by the ethics committees of the University of Heidelberg (Germany; Medical Faculty Heidelberg, S-017/2013). Written informed consent was obtained from all patients, and the study was conducted in accordance with the Declaration of Helsinki. All animal procedures and experiments were performed in accordance with the Guide for the Care and Use of Laboratory Animals (National Institutes of Health [NIH]) and approved by the Regierungspräsidium Karlsruhe, Baden-Württemberg, Germany (reference numbers G/52-16, G/155-16, G/261-17).

Chemicals

If not further specified, chemicals were purchased from Sigma-Aldrich, Munich, Germany. The chemicals melittin (1 μM) for Gα_s inhibition, pertussis toxin (PTX; 500 ng/mL) for Gα_i inhibition, H89 for PKA inhibition (5 μM), wortmannin (100 nM) for PI3K inhibition, and KN93 and its control KN92 (each 10 μM) for CAMKII inhibition were purchased from Tocris Bioscience (Bristol, UK). All chemicals were diluted and stored according to the manufacturer’s protocols. Propranolol (1 μM) was obtained from the University of Heidelberg pharmacy. Recombinant RLN H2 (R&D Systems, Minneapolis, MN, USA) was diluted according to the manufacturer’s protocol.

Study design

The study results are reported in accordance with the ARRIVE (Animal Research: Reporting In Vivo Experiments) guidelines. Animals were randomized into the respective groups, and all measurements and analysis were performed under blinded conditions. An identical protocol was followed for the rRXFP1 and huRXFP1 gene therapy study. Cardiac function measurements were performed before TAC operation (day −1), before virus infusion (day 7), before pump implantation (day 28), and at a 2-month follow-up time point (day 55). At day 7, animals were randomized to systemic injection of AAV9.Luc or RXFP1 vector (AAV9.rRXFP1 or AAV9.huRXFP1). Osmotic pumps were implanted on day 28 with animals randomized to receive either saline or RLN pumps. Follow-up assessments were performed on day 55 and the animals were sacrificed the following day. In addition, mice were sham operated to serve as control animals and randomized to be implanted with either saline- or RLN-containing pumps Data S2.

Experimental animals

Male wild-type C57BL/6N mice used for this study were obtained from Janvier Laboratories, Saint Berthevin Cedex, France. If not otherwise stated, male, 8-week-old mice were used.

Construction and production of recombinant viral vectors

Pseudotyped, recombinant rAAV6 or rAAV9 were used for all in vitro or in vivo experiments, respectively. The AAV viral vectors AAV6 and AAV9.CMV.MLC260.FLAG.Rxfp1 (designated AAV6.rRXFP1 or AAV9.rRXFP1) and AAV6 and AAV9.CMV.MLC260.FLAG.RXFP1 (designated AAV6.huRXFP1 or AAV9.huRXFP1) were generated as follows. Both optimized Rattus Norvegicus and Homo sapiens RXFP1 cDNA (GenBank: NM_201417.1 and GenBank: NM_021634.3) were synthesized and cloned into a pCDNA3.1 plasmid carrying an N-terminal FLAG-Tag using PstI and XhoI restriction sites. After extensive in vitro testing, the transgene cassette was sub-cloned into an AAV transfer plasmid downstream of a cardiomyocyte-specific CMV-enhanced short 260 bp myosin light chain promoter (MLC260) using BamHI and XhoI restriction sites and the whole construct was flanked by two inverted terminal repeats (ITRs). Recombinant vectors were generated by packaging of AAV-inverted terminal repeat recombinant genomes into AAV6 or AAV9 capsids using two plasmid transfection protocol as described before.^49,50 High titer vectors were produced in cell stacks with polyethylenimine transfection. After 72 h, the vectors were harvested and purified by density filtration in an iodixanol gradient. Using the same method, the control vectors AAV6 and AAV9.CMV.MLC260.FLAG.Luc^Firefly (designated AAV6.Luc or AAV9.Luc) were generated. Viral titers from all viral vectors were quantified at the same time using a SYBR-green real-time polymerase chain reaction (PCR) assay (Bio-Rad) and expressed as vg/mL.

TAC

A pressure overload-induced HF model was generated as described before⁵¹ with minor modifications. Anesthesia was induced with intraperitoneal injection of Xylazin (16 mg/kg body weight) and Ketamine (120 mg/kg body weight). The mice were then intubated, and anesthesia was maintained with 0.5% isoflurane mixed with 0.5−1.0 L/min 100% O₂. A partial thoracotomy in the first intercostal space was performed under a surgical microscope and the ribs retracted using 4 chest retractors. Blunt angled forceps were used to gently separate the thymus and fat tissue from the aortic arch. After the transverse aorta was identified, a small piece of an 8.0 silk suture was placed between the right carotid and innominate arteries. Two loose knots were tied around the transverse aorta and a small piece of a blunt needle (26G needle) was placed parallel to the transverse aorta. The two knots were placed against the needle before carefully removing it. Chest and skin were closed with sutures and the animals were allowed to recover. For postoperative pain control carprofen (5 mg/kg body weight) was injected subcutaneously. In the sham control mice, the entire procedure was identical except for the ligation of the aorta.

Systemic administration of viral vectors in mice

A dosage of 1 × 10¹² vg in a volume of 100 μL was injected into the tail vein of C57BL/6 mice 7 days after TAC operation. An equivalent amount and volume of AAV9.Luc was injected into TAC operated mice as controls.

Assessment of cardiac function

Echocardiography was performed in wake animals using Vevo 2100, VisualSonics system. The mice were shaved on the chest and held in prone position. Warm ultrasound coupling gel (37°C) was placed on the shaved chest area, and the MS400 transducer was positioned to obtain 2D B-mode parasternal long and short axis views and a 1D M-mode short axis view. The ejection fraction (EF), fractional shortening (FS), left ventricular internal diameter diastolic (LVIDd), left ventricular internal diameter systolic (LVIDs), and left ventricular diastolic anterior wall thickness (LVAWd) were calculated from 6 consecutive heartbeats in the M-mode using LV trace function. Global longitudinal strain (GLS), longitudinal or radial strain, and strain rate, as well as reverse longitudinal strain rate (rLSR), were calculated from 3 consecutive heartbeats in the long axis B mode using speckle tracking echocardiography as previously described.^28,52

Pressure loop measurement

Pressure loop measurement was performed according to a protocol modified from Abraham and Mao.⁵³ In brief, the animal was anesthetized with a weight-adapted intraperitoneal injection of 500 μg/kg medetomidine, 5 mg/kg midazolam, and 50 μg/kg fentanyl. After the righting reflex was lost, the mouse was secured to the operating table with surgical tape. Prior to the measurement, the catheter tip was equilibrated in warm saline for 30 min and calibrated to 0 mmHg. The neck and chest area were cleaned with ethanol and the fur was removed. An incision was made over the right carotid from mandible to sternum. The surrounding tissue was bluntly dissected to expose the right carotid avoiding damage to the adjacently localized vagal nerve. A distal sterile 6.0 silk suture was placed around the carotid artery, tied, and secured. 2 proximal sutures were loosely tightened, and the proximal carotid artery occluded by a hemostat. A small incision was made between the upper sutures. The catheter tip was inserted into the vessel through the incision and the catheter was secured using middle suture. Releasing the hemostat, the catheter was gently advanced into the ascending aorta and into the LV through the carotid guided by pressure tracing to ensure correct placement.

Calcium transients

Intracellular Ca²⁺ transients were measured according to Yu et al.⁵⁴ with minor modifications.⁵⁵ NRVCMs were transduced with AAV6.rRXFP1 or AAV6.Luc vectors for 5 days. The medium was changed every second day. The transfected NRVCMs were loaded with 1 μM Fura 2 AM in M199 at 37°C for 15 min. After loading, the cells were washed with warm M199 medium for 15 min at 37°C and eventually connected to an electrode system on the table of an inverted fluorescence microscope (IX70 Olympus). Electrically stimulation was applied at a frequency of 1 Hz and the Fura 2 AM loaded cells were excited at 380 nm using a monochromator. The alternative measurement of the bispectral emission (340 nm/380 nm) was performed with the fluorescence microscope for a period of 5 min and the emission ratio was calculated with the software TILL vision. The data were exported to an online script. The following parameters were analyzed: (1) the amplitude of the transient (Δ[340 nm/ 380 nm]) and (2) the diastolic Ca²⁺ ([340 nm/ 380 nm]).

Tissue sample harvesting

At the end of the study, all mice were sacrificed. The chest was opened, and the heart and lungs removed. The heart was washed in cold PBS, weighed, and cut into 7 pieces. The lungs were cleaned and weighed. Other organs such as liver, kidney, adrenals, spleen, and skeletal muscle were collected. All tissues pieces were frozen on dry ice and stored at −80°C until further processing.

Histological staining

Representative formalin-fixed paraffin-embedded heart specimens were cut in 4–6 μm sections using a HM 340E Rotary Microtome (Thermo Fisher Scientific, Waltham, MA, USA) and stained with hematoxylin and eosin (H&E) and acid fuchsin orange G (AFOG) according to standard protocols. Additionally, a periodic-acid Schiff reaction (PAS) was performed. Images were acquired at 20-fold magnification using a Hamamatsu NanoZoomer Digital Pathology system (Hamamatsu Photonics, Hamamatsu, Japan).

Neonatal ventricular cardiomyocytes isolation

2–3 days old neonatal Wistar rats were decapitated and hearts were collected, washed, and then minced in cold ADS buffer (120 mmol/L NaCl, 20 mmol/L HEPES, 8 mmol/L NaH2PO4, 6 mmol/L glucose, 5 mmol/L KCl, 0.8 mmol/L MgSO₄, pH 7.4). Up to six digestion steps with pancreatin (Sigma-Aldrich, Munich, Germany; 0.6 mg/mL) and collagenase type II (Worthington, Lakewood, USA; 0.5 mg/mL) in sterile ADS buffer were performed. Single cell containing supernatant was collected after each step and the pelleted cells stored in fetal calf serum (FCS; Biochrom/Merck, Berlin, Germany) at 37°C. The combined pellets were resuspended in cold, sterile ADS buffer and cardiomyocytes were separated from fibroblasts using a Percoll gradient (GE Healthcare Sciences, Freiburg, Germany) in sterile ADS buffer. The cardiomyocyte phase was collected and washed 2 times in cold, sterile ADS. Cardiomyocytes were resuspended in Medium 199 supplemented with 10% v/v fetal calf serum, 100 U/mL penicillin, 100 μg/ streptomycin, 2 mM L-glutamine, and 1 mM calcium chloride and plated in cell culture plates. Cells were seeded at densities between 0.1 and 0.2 million cells/cm². The NRVCMs were cultured at 37°C and 5% CO₂ humidified atmosphere. After 48 h FCS was reduced to 0.5%, followed by medium renewal every other day.

RNA isolation and quantitative reverse-transcriptase PCR (qRT-PCR)

RNA from adherent cells (2 × 10⁶ cells) or 10–15 mg tissue was collected using 1 mL TriZol reagent (Life Technologies GmbH, Darmstadt, Germany). Tissue was homogenized using ceramic beads in a bead mill for 20 s at 5,500 rpm. RNA isolation was done according to the manufacturer’s recommendation. For precipitation of RNA from adherent cells, the water/2-propanol mixture was supplemented with 1.5 μL Polyacryl carrier. Precipitation was done by incubating the water/2-propanol mixture at −20°C overnight. Purified RNA was transcribed using iScript cDNA Synthesis Kit (BioRad Laboratories GmbH, Munich, Germany) according to the manufacturer’s recommendation. 1 μg of RNA was used in 20 μL reaction volume. qRT-PCR was performed in duplicates using diluted cDNA (1:100) on a Bio-Rad CFX96 real-time PCR detection system (Bio-Rad) with SYBR Green PCR master mix (Bio-Rad). The 2^–ΔΔCT method was used to calculate relative gene expression levels between samples. Specificity of PCR products was confirmed by melting point determination and gel electrophoresis. All qRT-PCR primers used are listed in Table S1.

Quantitative immunoblot analysis

For protein isolation, tissue samples were homogenized in 100 μL/10 mg SDS Lysis Buffer (1% [v/v] SDS, 1 mM EDTA, 1 mM EGTA) supplemented with 1 × cOmplete Protease inhibitor (Sigma-Aldrich) and phosphatase inhibitor cocktails 2 and 3 (Sigma-Aldrich) using ceramic beads in a bead mill for 2 × 30 s at 6,000 rpm. Homogenate was incubated on ice for 30 min. For protein isolation from adherent cells, cells were lysed in 50 μL SDS Lysis Buffer/well and incubated on ice for 30 min. Cell and tissue lysates were centrifuged for 15 min, 18,000 × g, 4°C to remove debris. Protein concentration of supernatant was measured using DC Protein Assay Kit (BioRad). Typically, 25–50 μg denatured protein were segregated on a 4%–20% gradient Tris-Gylcine gel (Thermo Fisher Scientific, Waltham, MA, USA) and transferred to a polyvinylidene fluoride (PVDF) membrane (Millipore, Dundee, UK) in a Trans-Blot Semidry blotting chamber (BioRad). After blocking in i-Block blocking reagent (Thermo Fisher Scientific) for 1 h, membranes were probed with the following antibodies diluted in iBlock solution: PLB (mouse monoclonal; Thermo Fisher; Cat. MA3-922; dilution 1:5,000), P-PLB(S16) (rabbit polyclonal; Merck Millipore; Cat. 07-052; dilution 1:7,500), P-PLB(T17) (rabbit polyclonal; Badrilla, Leeds, UK; Cat. A010-13AP; dilution 1:5,000). Mouse monoclonal antibodies against GAPDH (Merck-Millipore, Burlington, MA; Cat. MAB374; dilution 1:10,000) and Sarcomeric α-Actinin (Sigma Aldrich; Cat. A7811; 1:2,500) were used as internal controls. Anti-mouse antibody coupled to Alexa Fluor 680 (Thermo Fisher Scientific; Cat. A-21057; dilution 1:10,000) and anti-rabbit secondary antibody coupled to Dylight 800 (Cell Signaling Technology; Cat. 5151; 1:10,000) were used to detect the signals from the immunoblots utilizing the Odyssey CLx infrared imager (LI-COR) detection system. Images were processed using Odyssey imaging software.

cAMP Assay

cAMP measurements were performed using the cAMP Glo assay kit from Promega (Mannheim, Germany) according to the manufacturer’s instructions. 500 μM IBMX and 100 μM Ro 20-1724 were used to keep cAMP from degradation. The luminescence was read using a luminometer.

RLN H2 measurement

Circulating levels of RLN H2 were measured using a ready-to-use RLN H2 Quantikine ELISA kit from R&D Systems (DLR200). Plasma samples were diluted 1:200 to stay within the detection range of the assay. 50 μL of negative control, positive control, standard, and diluted samples were used for the measurement.

Statistics

Data are presented as means ± SEM. GraphPad Prism software version 6.0c was used for statistical analysis. Normal distribution was examined using D’Agostino-Pearson omnibus normality test. Unpaired Student’s t test was used to test for differences between the two groups if normal distribution was assumed. Otherwise, Mann-Whitney U test was used. To compare means among three or more independent groups with normal distribution, we performed one-way analysis of variance (ANOVA). A Tukey post hoc test was applied when multiple comparisons were conducted. For non-normal data distributions, a Kruskal-Wallis test was performed, and a Dunn post hoc test was applied when multiple comparisons were conducted. Two-way ANOVA was used when appropriate. For all tests, a p value of ≤ 0.05 was accepted as statistically significant. To reduce complexity in some graphs, comparisons of two specific groups are indicated with individual symbols (∗, §. $, +), whereas a duplication of a given symbol indicates the significance level reached: ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. The raw data can be found in the supplemental information (Data S2).