# EVOLVING PICTURE OF CALCIUM HANDLING IN CARDIAC DISEASE

EDITED BY : Maria Fernandez-Velasco, Angélica Rueda, Gema Ruiz-Hurtado and Laetitia Pereira PUBLISHED IN : Frontiers in Physiology

#### Frontiers eBook Copyright Statement

The copyright in the text of individual articles in this eBook is the property of their respective authors or their respective institutions or funders. The copyright in graphics and images within each article may be subject to copyright of other parties. In both cases this is subject to a license granted to Frontiers. The compilation of articles constituting this eBook is the property of Frontiers.

Each article within this eBook, and the eBook itself, are published under the most recent version of the Creative Commons CC-BY licence. The version current at the date of publication of this eBook is CC-BY 4.0. If the CC-BY licence is updated, the licence granted by Frontiers is automatically updated to the new version.

When exercising any right under the CC-BY licence, Frontiers must be attributed as the original publisher of the article or eBook, as applicable.

Authors have the responsibility of ensuring that any graphics or other materials which are the property of others may be included in the CC-BY licence, but this should be checked before relying on the CC-BY licence to reproduce those materials. Any copyright notices relating to those materials must be complied with.

Copyright and source acknowledgement notices may not be removed and must be displayed in any copy, derivative work or partial copy which includes the elements in question.

All copyright, and all rights therein, are protected by national and international copyright laws. The above represents a summary only. For further information please read Frontiers' Conditions for Website Use and Copyright Statement, and the applicable CC-BY licence.

ISSN 1664-8714 ISBN 978-2-88966-123-7 DOI 10.3389/978-2-88966-123-7

#### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

#### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

#### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

#### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

# EVOLVING PICTURE OF CALCIUM HANDLING IN CARDIAC DISEASE

Topic Editors:

Maria Fernandez-Velasco, University Hospital La Paz, Spain Angélica Rueda, National Polytechnic Institute of Mexico (CINVESTAV), Mexico Gema Ruiz-Hurtado, Instituto de Investigación Hospital 12 de Octubre, Spain Laetitia Pereira, Institut National de la Santé et de la Recherche Médicale (INSERM), France

Citation: Fernandez-Velasco, M., Rueda, A., Ruiz-Hurtado, G., Pereira, L., eds. (2020). Evolving Picture of Calcium Handling in Cardiac Disease. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88966-123-7

# Table of Contents


*30 Long-Term Regulation of Excitation–Contraction Coupling and Oxidative Stress in Cardiac Myocytes by Pirfenidone* Adrián Monsalvo-Villegas, Diana Stephanie Osornio-Garduño and Guillermo Avila


Michal Cagalinec, Alexandra Zahradníková, Alexandra Zahradníková Jr., Dominika Kováčová, Ludovit Paulis, Simona Kureková, Matej Hot'ka, Jana Pavelková, Mario Plaas, Marta Novotová and Ivan Zahradník

*113 Impaired Activity of Ryanodine Receptors Contributes to Calcium Mishandling in Cardiomyocytes of Metabolic Syndrome Rats*

Gaudencio Fernández-Miranda, Tatiana Romero-Garcia, Tarín P. Barrera-Lechuga, Martha Mercado-Morales and Angélica Rueda

#### *128 Transmural Autonomic Regulation of Cardiac Contractility at the Intact Heart Level*

Yuriana Aguilar-Sanchez, Ainhoa Rodriguez de Yurre, Mariana Argenziano, Ariel L. Escobar and Josefina Ramos-Franco

*147 Unbalance Between Sarcoplasmic Reticulum Ca2 + Uptake and Release: A First Step Toward Ca2 + Triggered Arrhythmias and Cardiac Damage* Marilén Federico, Carlos A. Valverde, Alicia Mattiazzi and Julieta Palomeque

# Editorial: Evolving Picture of Calcium Handling in Cardiac Disease

Gema Ruiz-Hurtado1,2†, Angélica Rueda3†, Laetitia Pereira4† and María Fernández-Velasco5,6 \*

<sup>1</sup> Cardiorenal Translational Laboratory, Institute of Research i+12, University Hospital 12 de Octubre, Madrid, Spain, <sup>2</sup> CIBER-CV, University Hospital 12 de Octubre, Madrid, Spain, <sup>3</sup> Department of Biochemistry, Center for Research and Advanced Studies of the National Polytechnic Institute (CINVESTAV-IPN), Mexico City, Mexico, <sup>4</sup> UMR-S1180, Université Paris-Saclay, Châtenay-Malabry, France, <sup>5</sup> La Paz University Hospital Health Research Institute, IdiPAZ, Madrid, Spain, <sup>6</sup> CIBER-CV, Alberto Sols Biomedical Research Institute, Madrid, Spain

Keywords: calcium handling, ryanodine, heart disease, ion channels, EC coupling

**Editorial on the Research Topic**

#### **Evolving Picture of Calcium Handling in Cardiac Disease**

#### Edited by:

Christoph Fahlke, Julich Research Center, Helmholtz Association of German Research Centers (HZ), Germany

#### Reviewed by:

Gabriel Stölting, Charité – Universitätsmedizin Berlin, Germany

#### \*Correspondence:

María Fernández-Velasco maria.fernandez@idipaz.es; mvelasco@iib.uam.es

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology

> Received: 01 June 2020 Accepted: 24 July 2020 Published: 04 September 2020

#### Citation:

Ruiz-Hurtado G, Rueda A, Pereira L and Fernández-Velasco M (2020) Editorial: Evolving Picture of Calcium Handling in Cardiac Disease. Front. Physiol. 11:1013. doi: 10.3389/fphys.2020.01013

Cardiovascular diseases (CVDs) underlie a high rate of mortality worldwide. Both acquired and inherited CVDs are closely related to ionic channelopathies, compromising the expression and function of a wide number of ion channels. Ca2<sup>+</sup> mishandling has been associated with several CVDs, such as those involved in metabolic disorders, hypertrophy, heart failure, and several inherited heart diseases that implicate cardiac dysfunction and arrhythmias (Kushnir et al., 2018). Moreover, the archives of biomedical and life sciences are showing a multitude of contributions related to this topic, demonstrating the high interest of the scientific community in this specific research field. Indeed, Ca2<sup>+</sup> is the key regulator of the Ca2<sup>+</sup> induced Ca2<sup>+</sup> release (CICR) process, the main actor in cardiac excitation-contraction coupling (ECC) (Fabiato, 1983). Briefly, ECC begins with membrane depolarization of cardiomyocytes, activating sarcolemmal L-type voltagedependent Ca2<sup>+</sup> channels, thus promoting an inward Ca2<sup>+</sup> current (ICaL) that activates the intracellular Ca2<sup>+</sup> channel/ryanodine receptor type 2 (RyR2), located at the sarcoplasmic reticulum (SR), triggering a transient increase of cytosolic Ca2<sup>+</sup> levels, and eliciting cell contraction. To get new insights into these processes, Aguilar-Sanchez et al. evaluated in an elegant study whether the autonomous system regulates ECC in endocardial and epicardial layers from intact beating mouse hearts. The authors recorded action potentials (APs), Ca2<sup>+</sup> transients, and Ca2<sup>+</sup> currents in whole hearts, demonstrating that the Ca2<sup>+</sup> influx occurring in AP phase 1, and not in phase 2, triggers ECC (Aguilar-Sanchez et al.).

After contraction, relaxation takes place when cytosolic Ca2<sup>+</sup> levels return to physiological resting levels by two main mechanisms, the Sarco/Endoplasmic Reticulum Ca2<sup>+</sup> ATPase (SERCA2a pump) and the sarcolemma Na+- Ca2<sup>+</sup> exchanger (NCX) (Bers, 2002). More recently, other elements have been described as involved in the control of cardiac intracellular Ca2<sup>+</sup> handling, such as transient receptor potential (TRP) channels or organelles such as the nucleus or mitochondria that also rely on transporters or channels for Ca2<sup>+</sup> dynamics. The fine orchestration of all these elements results in adequate ECC and cell contraction.

Since its discovery (Inui et al., 1987), RyR2 has acquired special interest, and a large number of studies have analyzed whether changes in the structure, genetics, and function of this macro channel are involved in the pathogenesis of several CVDs. Post-translational modifications of RyR2, such as phosphorylation and oxidation, have gained major attention and have been related to channel dysfunction in some CVDs such as heart failure. Leading groups in the field have focused their interest in analyzing the physio-pathological regulation of RyR2 function by protein

**5**

kinase A or calmodulin kinase type II-dependent phosphorylation. In this line, Federico et al. have performed an interesting review article that addresses the most relevant findings related to changes in SR-Ca2<sup>+</sup> uptake and posttranslational modifications in RyR2, such as the phosphorylation and oxidation both linked to arrhythmias and other cardiac disturbances. In the same line, Fernández-Miranda et al. demonstrated in a prediabetic rat model of metabolic syndrome an "in situ" impairment of RyR2 function elicited by reduced levels of RyR2 phosphorylation, together with depressed SR-Ca2<sup>+</sup> uptake by SERCA2a, which are responsible for the poor cardiac outcome linked to this metabolic disorder.

In addition to phosphorylation, redox modifications of RyR2 such as S-nitrosylation and S-glutathionylation have emerged as new and potent modulators of channel function, demonstrating a crosslink between oxidation and phosphorylation changes. In many cases, changes in the redox state of RyR2 participate in the generation of abnormal Ca2<sup>+</sup> diastolic release, supporting an additional pro-arrhythmogenic mechanism in several CVDs. Nikolaienko et al. recapitulate in a remarkable review article the most significant findings related to whether redox modifications of RyR2 can induce channel dysfunction. In this line, Hamilton et al. tested the efficacy of mitochondria-targeted pharmacological interventions to attenuate cardiac arrhythmia at the cellular level and in "exvivo" hypertrophied thoracic aortic-banded mouse hearts. The authors demonstrated that the abnormal mitochondrial Ca2<sup>+</sup> accumulation induced an excessive production of reactive oxygen species that promoted the oxidation of RyR2 and enhanced its activity, inducing a pro-arrhythmic spontaneous Ca2<sup>+</sup> release (Hamilton et al.).

Besidesthis, increased reactive oxygen species can also activate the inflammatory response. In many cases, pro-inflammatory mediators result in intracellular Ca2<sup>+</sup> mishandling. In this regard, Val-Blasco et al. have determined whether the lack of the pro-inflammatory mediator NOD1 (a specific innate immune receptor) improves the β-adrenergic regulation of intracellular Ca2<sup>+</sup> dynamics in failing hearts. NOD1 is a well-known activator of the inflammatory response. An increasing amount of evidences suggest that inflammatory mediators participate in the progression of CVDs (Adamo et al., 2020). Inflammation and metabolic disorders are closely related, and in many cases involve changes in cardiac Ca2<sup>+</sup> homeostasis. A well-designed report, published by Delgado et al. has addressed the role of the inflammatory mediator tumor necrosis factor (TNF)-α in the modulation of intracellular Ca2<sup>+</sup> handling in a mouse model of diabetes type II (db/db). The authors demonstrated a different regulation of Ca2<sup>+</sup> handling mediated by TNF-α in male compared to female db/db mice, supporting the idea of a cardioprotective role of gender in the Ca2<sup>+</sup> mishandling linked to a pro-inflammatory environment (Delgado et al.).

Another interesting contribution has been supported by Cagalinec et al. using a transgenic rat model with a mutation in the gene encoding the wolframin protein (Wfs1−e5/−e5), linked to diabetic and neurological complications. These authors demonstrated that Wfs1−e5/−e5 rats are euglycemic at 4 months old and ventricular cardiomyocytes isolated from Wfs1−e5/−e5 animals showed a higher duration of Ca2<sup>+</sup> release; conjecturing a prolongation of RyR2 channel openings and modification in the RyR2 gating mediated by Wfs1 in cardiac cells (Cagalinec et al.).

All these functional studies and reviews have contributed to a better understanding of the basis of CVDs arising from ion channel dysfunction and intracellular Ca2<sup>+</sup> mishandling. These advances have also helped in the design of new drugs targeting ion channels, looking for more effective CVD treatments. Monsalvo-Villegas et al. performed an interesting pharmacological study describing a protective role of pirfenidone administration in isolated cardiomyocytes through modulation of CICR, ECC, and cell contractility. This study provides new explanations regarding the cardioprotective actions of pirfenidone (Monsalvo-Villegas et al.).

As previously mentioned, TRP channels have emerged as key partners in the pathogenesis of several CVDs, such as hypertrophy. Falcón et al. compiled in a commendable and wellorganized review article the involvement of various subtypes of TRPs in cardiac physiology and CVDs. The authors particularly described the involvement of these receptors in heart remodeling linked to cardiac hypertrophy, uncovering a new and interesting field of research (Falcón et al.).

In addition to cardiac function, Ca2<sup>+</sup> regulates vascular function. In this setting, Gutiérrez et al. have demonstrated in a well-conducted study whether different signaling pathways (PI3K, MAPK, and PKC) modulate Ca2<sup>+</sup> entry and intracellular Ca2<sup>+</sup> mobilization coupled to α1 adrenoceptor activation in the vascular smooth muscle of rat resistance arteries.

To conclude, the compendium of articles included in this Research Topic presents a new and interesting scenario for near future research related to novel actors in the picture of intracellular Ca2<sup>+</sup> mishandling associated with CVDs.

#### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

### FUNDING

GR-H and MF-V are supported by Instituto de Salud Carlos III, [MSII16/00047, CP15/00129, PI17/01093, PI17/01344], Sociedad Española de Cardiología (SEC 2017 and 2019) the European Regional Development Fund (Fondos FEDER and FSE). AR was supported by Fondo SEP-Cinvestav project #601410 FIDSC 2018/2; and Fondo SEP-Conacyt Ciencia Básica A1-S-9082. LP was supported by Institut de France Bourse Lefoulon Delaland and the ANR-11-1DEX-0003-02 as member of the Laboratory of Excellence LERMIT.

# REFERENCES


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Ruiz-Hurtado, Rueda, Pereira and Fernández-Velasco. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Deficiency of NOD1 Improves the β-Adrenergic Modulation of Ca2<sup>+</sup> Handling in a Mouse Model of Heart Failure

Almudena Val-Blasco<sup>1</sup>† , Jose A. Navarro-García<sup>2</sup>‡ , Maria Tamayo<sup>3</sup>†‡, Maria J. Piedras<sup>4</sup>‡ , Patricia Prieto<sup>3</sup>†‡, Carmen Delgado<sup>3</sup>† , Gema Ruiz-Hurtado<sup>2</sup> , Laura Rozas-Romero<sup>1</sup> , Marta Gil-Fernández<sup>1</sup> , Carlos Zaragoza<sup>5</sup>† , Lisardo Boscá<sup>3</sup>† and María Fernández-Velasco<sup>1</sup>† \*

#### Edited by:

Tarik Smani, Universidad de Sevilla, Spain

#### Reviewed by:

Marcelo Catalan, Arturo Prat University, Chile Jian Wu, University of Southern California, United States

#### \*Correspondence:

María Fernández-Velasco mvelasco@iib.uam.es; maria.fernandez@idipaz.es †CIBER-CV from ISCIII ‡These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology

Received: 01 March 2018 Accepted: 22 May 2018 Published: 14 June 2018

#### Citation:

Val-Blasco A, Navarro-García JA, Tamayo M, Piedras MJ, Prieto P, Delgado C, Ruiz-Hurtado G, Rozas-Romero L, Gil-Fernández M, Zaragoza C, Boscá L and Fernández-Velasco M (2018) Deficiency of NOD1 Improves the β-Adrenergic Modulation of Ca2<sup>+</sup> Handling in a Mouse Model of Heart Failure. Front. Physiol. 9:702. doi: 10.3389/fphys.2018.00702 1 Innate Immune Response Group, Instituto de Investigación Hospital Universitario La Paz, La Paz University Hospital, Madrid, Spain, <sup>2</sup> Cardiorenal Translational Laboratory and Hypertension Unit, Institute of Research i+12, Hospital Universitario 12 de Octubre, Madrid, Spain, <sup>3</sup> Departamento de Bioquímica, Facultad de Medicina, Instituto de Investigaciones Biomédicas Alberto Sols, Consejo Superior de Investigaciones Científicas, Madrid, Spain, <sup>4</sup> Department of Anatomy, Faculty of Health Sciences, Francisco de Vitoria University (UFV), Pozuelo de Alarcón, Spain, <sup>5</sup> Unidad de Investigación Cardiovascular, Universidad Francisco de Vitoria, Hospital Universitario Ramón y Cajal (IRYCIS), CIBERCV, Madrid, Spain

Heart failure (HF) is a complex syndrome characterized by cardiac dysfunction, Ca2<sup>+</sup> mishandling, and chronic activation of the innate immune system. Reduced cardiac output in HF leads to compensatory mechanisms via activation of the adrenergic nervous system. In turn, chronic adrenergic overstimulation induces pro-arrhythmic events, increasing the rate of sudden death in failing patients. Nucleotide-binding oligomerization domain-containing protein 1 (NOD1) is an innate immune modulator that plays a key role in HF progression. NOD1 deficiency in mice prevents Ca2<sup>+</sup> mishandling in HF under basal conditions, but its role during β-adrenergic stimulation remains unknown. Here, we evaluated whether NOD1 regulates the β-adrenergic modulation of Ca2<sup>+</sup> signaling in HF. Ca2<sup>+</sup> dynamics were examined before and after isoproterenol perfusion in cardiomyocytes isolated from healthy and from post-myocardial infarction (PMI) wild-type (WT) and Nod1−/<sup>−</sup> mice. Isoproterenol administration induced similar effects on intracellular [Ca2+]<sup>i</sup> transients, cell contraction, and sarcoplasmic reticulum (SR)-Ca2<sup>+</sup> load in healthy WT and Nod1−/<sup>−</sup> cells. However, compared with WT-PMI cells, isoproterenol exposure induced a significant increase in the [Ca2+]<sup>i</sup> transients and cell contraction parameters in Nod1−/−-PMI cells, which mainly due to an increase in SR-Ca2<sup>+</sup> load. NOD1 deficiency also prevented the increase in diastolic Ca2<sup>+</sup> leak (Ca2<sup>+</sup> waves) induced by isoproterenol in PMI cells. mRNA levels of β1 and β2 adrenergic receptors were significantly higher in Nod1−/−-PMI hearts vs WT-PMI hearts. Healthy cardiomyocytes pre-treated with the selective agonist of NOD1, iE-DAP, and perfused with isoproterenol showed diminished [Ca2+]<sup>i</sup> transients amplitude, cell contraction, and SR-Ca2<sup>+</sup> load compared with vehicle-treated cells. iE-DAP-treated

**8**

cells also presented increased diastolic Ca2<sup>+</sup> leak under β-adrenergic stimulation. The selectivity of iE-DAP on Ca2<sup>+</sup> handling was validated by pre-treatment with the inactive analog of NOD1, iE-Lys. Overall, our data establish that NOD1 deficiency improves the β-adrenergic modulation of Ca2<sup>+</sup> handling in failing hearts.

Keywords: β-adrenergic response, heart failure, NOD1, Ca2<sup>+</sup> handling, innate immune system

#### INTRODUCTION

Heart failure (HF) is a complex clinical disorder characterized by the inability of the heart to deliver blood and nutrients to metabolic tissues. Chronic HF is a progressive disease, with high morbidity and mortality, and poses a significant economic burden on the healthcare system. Given its poor prognosis, the identification of molecular pathways contributing to HF is a major research goal.

Compensatory activation of the adrenergic nervous system is a pathophysiological response to HF progression, which functions to maintain cardiac homeostasis through activation of neural hormones, mainly catecholamines (Gordan et al., 2015). Accordingly, sympathetic hyperactivity is a hallmark of HF, activating β-adrenergic receptors to increase heart rate and cardiac contractility via excitation–contraction (EC) coupling, in an attempt to counteract the decreased cardiac output. Chronic exposure of the heart to elevated levels of catecholamines may, however, ultimately lead to the desensitization of β-adrenergic receptors, exacerbating the loss of cardiac function and increasing the risk for triggered arrhythmias and sudden death (Feldman et al., 2005; Brum et al., 2006; Venetucci et al., 2008).

There is a growing body of evidence suggesting a link between the innate immune response and HF progression (Lin and Knowlton, 2014; Mann, 2015). Whether the myocardial inflammatory response impairs cardiac damage is a question of broad significance. Recent studies indicate that some receptors of the innate immune system, including nucleotide-binding oligomerization domain-like receptors (NLRs), play significant roles in the host response after cardiac damage (Bullón et al., 2016; Monnerat et al., 2016; Val-Blasco et al., 2017a). Nucleotidebinding oligomerization domain (NOD) receptors are NLR family members that recognize conserved motifs of bacterial peptidoglycans in many Gram-negative bacteria (Caruso et al., 2014). Beyond their role as microbial pattern recognition receptors, it has recently been described that NODs can also be activated by non-infectious factors such as endoplasmic reticulum stress (Keestra-Gounder et al., 2016). Ligand activation of NOD receptors induces a conformational change in the protein, leading to self-oligomerization and promoting the recruitment of its adaptor, receptor-interacting protein 2 (RIP2), which activates nuclear factor-κB and triggers the inflammatory response (Park et al., 2007). NOD-containing protein 1 (NOD1) is expressed in the heart, and is functional in resident fibroblasts and cardiomyocytes (Fernández-Velasco et al., 2012; Delgado et al., 2015). Several studies have demonstrated a role for NOD1 in the progression of cardiovascular diseases, including atherosclerosis and diabetic cardiomyopathy (Prieto et al., 2014; Kanno et al., 2015; Val-Blasco et al., 2017b). We have recently shown that NOD1 is upregulated in both mouse and human failing myocardium, and its genetic deletion or pharmacological blockade prevents cardiac dysfunction and deleterious remodeling in failing hearts principally by preventing the HF-related Ca2<sup>+</sup> mishandling (Val-Blasco et al., 2017a). Conversely, activation of NOD1 promotes a dysregulation of the intracellular Ca2<sup>+</sup> dynamics, similar to those observed in HF (Delgado et al., 2015).

Ca2<sup>+</sup> plays a key role in cardiac EC coupling. During EC coupling, an action potential leads to a small increase in intracellular Ca2<sup>+</sup> via activated sarcolemmal L-type Ca channels (ICaL), and this is amplified by a greater release of Ca2<sup>+</sup> from the sarcoplasmic reticulum (SR) by ryanodine receptors (RyR2), increasing the intracellular Ca2<sup>+</sup> concentration ([Ca2+]i) – a process known as Ca2+-induced Ca2<sup>+</sup> release. This elevation in [Ca2+]<sup>i</sup> results in myofilament activation and cell contraction. During relaxation, Ca2<sup>+</sup> is removed from the cytosol primarily by two mechanisms: Ca2<sup>+</sup> re-uptake by the SR-ATPase (SERCA2a) and Na+/Ca2<sup>+</sup> exchanger activation.

Excitation–contraction coupling is modulated by the sympathetic nervous system through activation of the β-adrenergic receptor coupled to Gs-type G-proteins, promoting an elevation in the intracellular concentration of cAMP (Bers, 2002). In turn, cAMP activates protein kinase A (PKA), resulting in the phosphorylation of several EC coupling-Ca2<sup>+</sup> transporters such as ICaL, RyR2, and phospholamban, among others (Mayourian et al., 2018).

Several lines of evidence have established a direct relationship between Ca2<sup>+</sup> dysregulation and HF progression (Gómez et al., 1997; Heinzel et al., 2011; Ruiz-Hurtado et al., 2015). Depressed systolic Ca2<sup>+</sup> release, the decline in the SR-Ca2<sup>+</sup> load, and the increased Ca2<sup>+</sup> diastolic leak seem to be key players in HF progression, and these changes worsen under chronic β-adrenergic activation. We recently showed that NOD1 modulates intracellular Ca2<sup>+</sup> handling in human and experimental failing hearts (Val-Blasco et al., 2017a); however, the role of NOD1 in the modulation of Ca2<sup>+</sup> handling under β-adrenergic stimulation has not been investigated. Using a mouse HF model, we determined whether the deficiency of NOD1 impairs the modulation of Ca2<sup>+</sup> dynamics under β-adrenergic stimulation.

#### MATERIALS AND METHODS

#### Myocardial Infarction Model

Male Nod1−/<sup>−</sup> mice on a C57BL/6J (6B; 129P2-Nod1tm1Nnz/J) background were used in this study (Chamaillard et al., 2003). As controls, we used wild-type (WT) C57BL/6J mice (The Jackson

Laboratory, Bar Harbor, ME, United States). Experiments involving mice were carried out in compliance with Spanish and European guidelines (2010/63/EU) regarding animal policy and welfare recommendations. The present study was performed in male mice because they have been reported to be more prone than female to develop cardiac injury and Ca2<sup>+</sup> mishandling after isoproterenol administration in experimental models of HF (Cross et al., 2002).

Mice were anesthetized by intraperitoneal (i.p.) injection of a mixture of ketamine (Imalgene <sup>R</sup> , 70 mg/kg) and xylazine (Rompun <sup>R</sup> , 10 mg/kg). Unconscious mice were shaved in the anterior region of the neck and the chest, intubated by tracheostomy, and connected to a small animal ventilator (MiniVent type 845, Harvard Apparatus) for artificial ventilation at 150 strokes/minute and 230 µL stroke volume. A 2–3 mm incision parallel to the lower costal edge was made and the left pectoralis major muscle was dissociated until the ribs were exposed. Left thoracotomy was performed between the third and fourth ribs to visualize the anterior surface of the heart and left lung. A 1-mm-thick piece of gelatin sponge (Spongostan <sup>R</sup> ), slightly moistened with saline, was inserted through the hollow thoracotomy to protect the lung. The ribs were separated with an eyelid-retractor and a branch of the left coronary artery was ligated with a blue polypropylene monofilament surgical non-absorbable suture 6/0. The Spongostan <sup>R</sup> sponge was then removed and the chest was closed with braided silk non-absorbable suture 4/0. Respiratory stimulation was performed to return the heart to spontaneous breathing. For postoperative analgesia, buprenorphine (Buprex <sup>R</sup> , 0.05 mg/kg) was applied subcutaneously. Mice were kept on a warm electric blanket until spontaneous recovery. Sham-operated mice were used as the control group and underwent the same procedure, but without coronary ligation.

#### Myocyte Isolation

Adult male mice (2 months of age) were anesthetized with sodium pentobarbital (100 mg/kg i.p.) and heparinized (4 UI/g i.p.). Hearts were rapidly removed and cannulated via the ascending aorta for Langendorff perfusion. Retrograde perfusion was initiated with a free calcium Tyrode solution containing 0.2 mM EGTA over 2–3 min at room temperature, and subsequently with Tyrode solution containing CaCl<sup>2</sup> (0.1 mM) and type II collagenase (1 mg/mL; Worthington Biochemical, Lakewood, NY, United States). Successful digestion was assumed when the flux rate increased and the heart color changed from red to pallor. The heart was then taken off the apparatus, and the ventricles were removed, minced into small pieces, and mechanically dissociated in a thermostatic bath at 37◦C in the enzymatic solution. The cardiomyocyte cell suspension was filtered through a nylon mesh (250 µm) and centrifuged at 300 rpm for 3 min. The cell pellet was suspended in Tyrode solution containing CaCl<sup>2</sup> (0.5 mM) and centrifuged as before. Finally, the cells were suspended in Tyrode solution containing CaCl<sup>2</sup> (1 mM). The Tyrode solution contained (in mM): 130 NaCl, 5.4 KCl, 0.5 MgCl2, 25 HEPES, 0.4 NaH2PO4, and 22 glucose, which was adjusted to pH 7.4 with NaOH.

#### Confocal Microscopy

Local increases in intracellular Ca2<sup>+</sup> concentration, [Ca2+]<sup>i</sup> transients, and Ca2<sup>+</sup> waves were analyzed in intact isolated cardiomyocytes loaded with the fluorescent Ca2<sup>+</sup> dye Fluo-3AM (5 µmol/L, Invitrogen). [Ca2+]<sup>i</sup> transients were recorded in cells electrically excited at 2 Hz by field stimulation using two parallel platinum electrodes. Ca2<sup>+</sup> waves were acquired in quiescent myocytes. SR Ca2<sup>+</sup> load was assessed by rapid caffeine application (10 mM) in cells previously excited at 2 Hz.

Images were acquired by confocal microscopy (Meta Zeiss LSM 710, 40× oil immersion objective with a 1.2 NA), by scanning the cell with an argon laser every 1.54 ms. Fluo-3AM was excited at 488 nm and emitted fluorescence was collected at >505 nm. Data analysis was performed with homemade routines using software written on the IDL platform (Research System Inc., Boulder, CO, United States), designed by Dr. AM Gómez (Inserm, UMR-S 1180). Images were corrected for background fluorescence. The fluorescence values (F) were normalized by the basal fluorescence (F0) to obtain the fluorescence ratio (F/F0).

## Drugs and Treatments

Isolated cardiomyocytes were perfused with 10−<sup>8</sup> M isoproterenol (Sigma-Aldrich, Madrid, Spain) for 1–5 min and intracellular Ca2<sup>+</sup> dynamics were recorded. In some experiments, cells were pretreated with the selective NOD1 agonist C12-iE-DAP (iE-DAP, InvivoGen, San Diego, CA, United States), which has significant cell membrane permeability and high potency (Fernández-Velasco et al., 2012; Delgado et al., 2015; Val-Blasco et al., 2017a), or iE-Lys (Invivogen), an inactive analog of NOD1. Nodinitib-1 (Cayman) was used as a selective NOD1 inhibitor and mice were injected IP with 5 µmol/L nodinitib-1 or vehicle (<0.01% DMSO) three times weekly for 6 weeks.

#### Real-Time PCR

Frozen heart tissue was pulverized to powder and then homogenized in TRIZOL Reagent <sup>R</sup> solution (Ambion) with a Polytron system (MWR). RNA (250 ng) was reverse transcribed to cDNA using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems #4368813). This template cDNA was used in the qPCR reaction with Power SYBR Green PCR Master mix (Applied Biosystems #4367659) and specific primers in a 7900HT Fast Real Time PCR system (Applied Biosystems). 18s-RNA was used as a housekeeping gene. The primer sequences (5<sup>0</sup> -30 ) for β1 and β2 adrenergic receptors and 18s-RNA were as follows:

β1-F CGCTGATCTGGTCATGGGAT β1-R GAAGAAGGAGCCGTACTCCC β2-F AATAGCAACGGCAGAACGGA β2-R TCACAAAGCCTTCCATGCCT 18s-F CCAGTAAGTGCGGGTCATAAGC 18s-R CCTCACTAAACCATCCAATCGG

#### Statistical Analysis

Results are reported as mean ± SEM. Statistical analysis was performed using two-way analysis of variance, paired (twosided) Student's t test or χ 2 test, as appropriate. All statistical

analyses were performed with Origin 8.0 software (OriginLab, Northampton, MA, United States). Significance was set at P < 0.05.

#### RESULTS

#### Hypertrophy Development in Mice Subjected to Myocardial Infarction

Compared with healthy WT mice, post-myocardial infarction (PMI)-mice 6 weeks after surgery showed significantly increased heart weight, heart weight/body weight ratio, and heart weight/tibia length ratio, consistent with the development of cardiac hypertrophy (**Table 1**). By contrast, Nod1−/−-PMI mice showed similar heart weight, heart weight/body weight, and heart weight/tibia length values to Nod1−/<sup>−</sup> mice (**Table 1**). These data indicate that deficiency of NOD1 attenuates cardiac hypertrophy in our PMI mouse model.

## Deficiency of NOD1 Improves the β-Adrenergic Modulation of Systolic Ca2<sup>+</sup> Release and Sarcoplasmic Reticulum Ca2<sup>+</sup> Load in PMI Cardiomyocytes

We first analyzed whether deficiency of NOD1 determined the β-adrenergic response to Ca2<sup>+</sup> handling in healthy myocytes. Isoproterenol perfusion induced a similar increase in [Ca2+]<sup>i</sup> transients amplitude (**Figure 1A**), cell contraction (**Figure 1B**), and SR-Ca2<sup>+</sup> load (**Figure 1C**) in WT and Nod1−/<sup>−</sup> cardiomyocytes.

Given that β-adrenergic regulation of Ca2<sup>+</sup> dynamics plays a key role in HF-related cardiac dysfunction, we next questioned whether NOD1 deficiency impacted isoproterenolinduced modulation of Ca2<sup>+</sup> handling in cardiomyocytes after PMI. **Figure 1D** shows representative line-scan Ca2<sup>+</sup> fluorescence images obtained after electric field stimulation at 2 Hz in all groups. Under β-adrenergic stimulation, both WT-PMI and Nod1−/−-PMI cardiomyocytes exhibited increased [Ca2+]<sup>i</sup> transients (**Figure 1E**), augmented cell contraction (**Figure 1F**), and faster time decay of [Ca2+]<sup>i</sup> transients (**Figure 1G**), but all these changes were significantly greater in Nod1−/−-PMI cells. These results reveal that, compared with WT-PMI cells, Nod1−/−-PMI myocytes show a significant improvement in Ca2<sup>+</sup> mishandling, which was mostly due to the increased amplitude of [Ca2+]<sup>i</sup> transients (**Figures 1D,E**) and also higher cell contraction parameters (**Figure 1F**). In this line, mean values of the amplitude of [Ca2+]<sup>i</sup> transients (F/F0) in cardiomyctes obtained under isoproterenol administration were 5.31 ± 0.52 (n/N = 13/5) in WT, 5.18 ± 0.32 (n/N = 14/4) in Nod1−/−, 4.18 ± 0.22 (n/N = 12/4) in WT-PMI, and 5.41 ± 0.21 (n/N = 13/4) in Nod1−/−-PMI (**Figures 1A,E**). Thus, [Ca2+]<sup>i</sup> transients obtained in the Nod1−/−-PMI group during isoproterenol perfusion were similar to those of equivalent healthy WT and Nod1−/<sup>−</sup> cardiomyocytes. Consistent with this finding, treatment of WT-PMI mice with the selective NOD1 inhibitor nodinitib-1 resulted in higher amplitude isoproterenolinduced [Ca2+]<sup>i</sup> transients in isolated cardiomyocytes compared with cells isolated from vehicle-treated mice (Supplementary Figure S1).

Next, to question whether changes in [Ca2+]<sup>i</sup> transients during β-adrenergic stimulation in PMI cardiomyocytes were related to modifications in SR-Ca2<sup>+</sup> load, we measured caffeine-evoked [Ca2+]<sup>i</sup> transients in all groups. **Figure 2A** shows examples of line-scan images of caffeine-evoked [Ca2+]<sup>i</sup> transients obtained in each group. Mean values of the amplitude of caffeine-evoked [Ca2+]<sup>i</sup> transients (F/F0) in cardiomyocytes during isoproterenol exposure was 7.07 ± 0.69 (n/N = 9/3) in WT, 7.34 ± 0.93 (n/N = 10/4) in Nod1−/<sup>−</sup> (**Figure 1C**), 5.77 ± 0.46 (n/N = 13/4) in WT-PMI, and 7.40 ± 0.50 (n/N = 12/4) in Nod1−/−-PMI (**Figure 2B**). Thus, the amplitude of [Ca2+]<sup>i</sup> transients triggered by caffeine was significantly higher in Nod1−/−-PMI mice than in WT-PMI mice, both in the absence or the presence of isoproterenol (**Figure 2B**). Additionally, the high values of SR-Ca2<sup>+</sup> load in the Nod1−/−-PMI group treated with isoproterenol were similar to those found in healthy WT and Nod1−/<sup>−</sup> cells treated with isoproterenol.

Taken together, β-adrenergic stimulation induced better systolic Ca2<sup>+</sup> release and cell contractility in Nod1−/−-PMI cells than in WT-PMI cells. These effects correlated with the significant increase in the SR-Ca2<sup>+</sup> load induced by isoproterenol in Nod1−/−-PMI cells.

#### Deficiency of NOD1 Prevents the Increased Diastolic Ca2<sup>+</sup> Release Induced by β-Adrenergic Stimulation in PMI Cardiomyocytes

Since SR-Ca2<sup>+</sup> load impairment is closely related to changes in diastolic Ca2<sup>+</sup> release, we analyzed spontaneous Ca2<sup>+</sup> waves in quiescent cells.


HW, heart weight; BW, body weight; TL, tibia length. ∗∗P < 0.01, ∗∗∗P < 0.001 vs WT and &&P < 0.01 vs WT-PMI.

fluorescence [Ca2+]<sup>i</sup> transients (F/F0, E), cell contraction (CC, %, F), and decay time of [Ca2+]<sup>i</sup> transients (tau, ms, G) in WT-PMI (n = 12 cells/4 mice) and NOD1−/−-PMI (n = 13 cells/4 mice) cardiomyocytes with and without Iso. Histograms represent the mean ± SEM: \$\$P < 0.01, \$\$\$P < 0.001 vs WT; <sup>+</sup>P < 0.05 vs NOD1−/−, ++P < 0.01 vs NOD1−/−, +++P < 0.001 vs NOD1−/−; <sup>∗</sup>P < 0.05 vs WT-PMI; &P < 0.05, &&P < 0.01, &&&P < 0.001 vs NOD1−/−-PMI; and #P < 0.05, ###P < 0.001 vs WT-PMI treated with Iso.

**Figure 3A** shows a representative line-scan image of a Ca2<sup>+</sup> wave recording from a WT-PMI myocyte. Ca2<sup>+</sup> wave occurrence analysis (**Figure 3B**) revealed no statistically significant changes between WT-PMI (14.60%) and Nod1−/−-PMI (10.50%) cells under basal conditions. By contrast, both groups showed a significant elevation in the Ca2<sup>+</sup> wave rate after isoproterenol perfusion, but this was significantly higher in WT-PMI cardiomyocytes perfused with isoproterenol (**Figure 3**). Accordingly, the percentage of cells with Ca2<sup>+</sup> waves during isoproterenol exposure was 34.07% in WT-PMI cells and 15.65% in Nod1−/−-PMI cells (P < 0.01).

These results indicate that deficiency of NOD1 prevents the increase of Ca2<sup>+</sup> diastolic leak induced by β-adrenergic stimulation in PMI cardiomyocytes, thus providing an explanation for the improvement in the SR-Ca2<sup>+</sup> load and the resulting better systolic Ca2<sup>+</sup> release in Nod1−/−-PMI cells under isoproterenol administration.

# mRNA Levels of the β1 and β2 Adrenergic Receptors Are Elevated in Nod1−/−-PMI Hearts

Analysis of the mRNA levels of β1 and β2 adrenergic receptors in hearts from WT-PMI and Nod1−/−-PMI mice revealed significantly increased expression of both receptors in Nod1−/−- PMI hearts (**Figures 4A,B**). These increased β1 and β2 adrenergic receptor levels can explain at least in part the improvement in Ca2<sup>+</sup> regulation observed in the Nod1−/−-PMI group under isoproterenol administration.

## Selective Activation of NOD1 Compromises the β-Adrenergic Regulation of Ca2<sup>+</sup> Handling in Wild-Type Cardiomyocytes

We next addressed whether the selective activation of NOD1 impairs the β-adrenergic regulation of Ca2<sup>+</sup> handling in

FIGURE 2 | Deficiency of NOD1 increases the β-adrenergic response on SR-Ca2<sup>+</sup> load in cardiomyocytes from failing mice. (A) Representative line-scan confocal images of caffeine-evoked [Ca2+]<sup>i</sup> transients obtained from WT-PMI and NOD1−/−-PMI cardiomyocytes with and without Iso perfusion. (B) Mean values of caffeine-evoked [Ca2+]<sup>i</sup> transients amplitude (SR-Ca2<sup>+</sup> load, F/F0) in WT-PMI (n = 23 cells/4 mice), NOD1−/−-PMI (n = 19 cells/4 mice), WT-PMI (n = 13 cells/4 mice), and NOD1−/−-PMI (n = 12 cells/4 mice) cardiomyocytes treated with Iso. Histograms represent the mean ± SEM: ∗∗P < 0.01, ∗∗∗P < 0.001 vs WT-PMI; &&&P < 0.001 vs NOD1−/−-PMI; and #P < 0.05 vs WT-PMI cardiomyocytes treated with Iso.

FIGURE 3 | Deficiency of NOD1 prevents the increase in diastolic Ca2<sup>+</sup> release induced by isoproterenol in cardiomyocytes from failing mice. (A) A representative line-scan image of a Ca2<sup>+</sup> wave recording from a WT-PMI myocyte. (B) The mean data for Ca2<sup>+</sup> wave occurrence from WT-PMI (n = 9 cells/3 mice) and NOD1−/−-PMI (n = 7 cells/3 mice) cardiomyocytes with and without Iso. Histograms represent the mean ± SEM: <sup>∗</sup>P < 0.05, ∗∗∗P < 0.001, vs WT-PMI; &P < 0.05, &&&P < 0.001 vs NOD1−/−-PMI; and ##P < 0.05 vs WT-PMI cardiomyocytes treated with Iso.

healthy cardiomyocytes. To do this, we pretreated WT cardiomyocytes for 1 h with the NOD1 ligand C12-iE-DAP (iEDAP; 20 µg/mL) and then we examined intracellular Ca2<sup>+</sup> dynamics in the absence or presence of isoproterenol perfusion. Under isoproterenol stimulation, iE-DAP-treated cells showed reduced amplitude of [Ca2+]<sup>i</sup> transients (**Figures 5A,B**), cell contraction (**Figure 5C**), and SR-Ca2<sup>+</sup> load (**Figure 5D**) compared with vehicle-treated cardiomyocytes. Treatment of cells with the inactive analog of IE-DAP, iE-Lys (20 µg/mL), resulted in a similar isoproterenol response for Ca2<sup>+</sup> handling to vehicle-treated cells, demonstrating the selective effect of iE-DAP on Ca2<sup>+</sup> handling under β-adrenergic stimulation (**Figures 5A–D**).

arbitrary units. ∗∗∗P < 0.001 vs WT-PMI.

Finally, we investigated whether the selective activation of NOD1 impairs diastolic Ca2<sup>+</sup> release under β-adrenergic stimulation. As shown in **Figure 5E**, isoproterenol perfusion promoted a significant increase in Ca2<sup>+</sup> waves in WT cells pretreated with iE-DAP as compared with vehicle- or iE-Lys-treated cells. Importantly, the percentage of Ca2<sup>+</sup> waves in cells that were pretreated with iE-DAP and under

(SR-Ca2<sup>+</sup> load), and (E) Ca2<sup>+</sup> wave occurrence in cardiomyocytes treated with vehicle (veh; n = 15 cells/7 mice), C12-iE-DAP (IEDAP n = 16 cells/6 mice), or with iE-Lys (IELYS, n = 14 cells/5 mice) in the absence or the presence of Iso. Histograms represent the mean ± SEM: <sup>∗</sup>P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001 vs veh; ##P < 0.01, ###P < 0.001 vs IEDAP; &P < 0.05, &&&P < 0.001 vs IELYS; \$P < 0.05, \$\$P < 0.01, \$\$\$P < 0.001 vs veh plus Iso; and ¬¬P < 0.01, ¬¬¬P < 0.001 vs IEDAP plus Iso.

isoproterenol perfusion was very similar to that found in WT-PMI cardiomyocytes under the same conditions (34.07% in WT-PMI cells and 29.20% in cells treated with iE-DAP; compare **Figures 3C**, **5E**). Supporting these data, iE-DAP administration failed to increase the percentage of Ca2+-wave-positive Nod1−/<sup>−</sup> cardiomyocytes (data not shown).

Overall, these results indicate that NOD1 activation limits the β-adrenergic regulation of Ca2<sup>+</sup> handling due to impairment of systolic Ca2<sup>+</sup> release ([Ca2+]<sup>i</sup> transients) and SR-Ca2<sup>+</sup> load, which are highly conditioned by the increased diastolic Ca2<sup>+</sup> leak.

#### DISCUSSION

Over the last decade, a large number of studies have shown that compensatory upregulation of the adrenergic nervous system response is a key factor in disease progression in chronic failing hearts. During the early stages of HF, there is a steady activation of the sympathetic nervous system to maintain cardiac output (Lymperopoulos, 2013), whereas in later stages, continuous adrenergic stimulation triggers negative feedback regulation of β-receptor activity, causing a downregulation in the number of receptors and also functional impairment, a process termed

receptor desensitization (Böhm et al., 1990; Madamanchi, 2007), ultimately contributing to toxicity and worsening of cardiac outcomes.

Some innate immune mediators, such as NLRs, play an important role in the host response to cardiac damage. Among them, NOD1 is expressed and functional in the heart (Fernández-Velasco et al., 2012; Delgado et al., 2015) and is upregulated in failing hearts both in a mouse HF model and in failing human myocardium (Val-Blasco et al., 2017a). Genetic deletion of NOD1 prevents HF-related cardiac dysfunction through modulation of EC coupling, chiefly by improving systolic Ca2<sup>+</sup> release and by maintaining Ca2<sup>+</sup> SR load, which are both compromised in HF (Val-Blasco et al., 2017a).

Our present results show that deficiency of NOD1 does not affect the regulation of Ca2<sup>+</sup> handling by β-adrenergic stimulation under physiological conditions, as systolic Ca2<sup>+</sup> release and SR-Ca2<sup>+</sup> load were similar between WT and Nod1−/<sup>−</sup> cardiomyocytes. In the failing heart, however, deficiency of NOD1 prevents the depression of [Ca2+]<sup>i</sup> transients and cell contractility due to maintained SR Ca2<sup>+</sup> load. Accordingly, during β-adrenergic stimulation, deficiency of NOD1 leads to a significant improvement in systolic Ca2<sup>+</sup> release, cell contraction parameters, and SR Ca2<sup>+</sup> load in PMI cardiomyocytes, with levels mirroring those found in isoproterenol-stimulated WT and healthy Nod1−/−cells. These results are interesting given that the decreased β-adrenergic function in HF plays a key role in the compromised cardiac contraction.

Several groups have shown that other pro-inflammatory mediators including TNFα, CXR4 or IL-1β impair the β-adrenergic response and reduce cardiac contractility (Gulick et al., 1989; Finkel et al., 1992; Schulz et al., 1995; Murray and Freeman, 1996; Pyo et al., 2006), and short-term treatment of rat cardiomyocytes with the pro-inflammatory cytokines IL-1α and TNFα blocks isoproterenol-induced increases in cell contractility (Bick et al., 1997). These findings are in agreement with an early report implicating IL-1β in uncoupling β-adrenergic responses in cardiomyocytes (Gulick et al., 1989). Consistent with these data, our study shows that a selective NOD1 agonist induces a minor but significant effect on cell contractility in isoproterenol-treated cardiomyocytes. NOD1 activation also reduced the β-adrenergic modulation of systolic Ca2<sup>+</sup> release due to reduced SR-Ca2<sup>+</sup> load, and by increasing the diastolic Ca2<sup>+</sup> leak. These results can be explained by the fact that NOD1 activation in cardiomyocytes induces the over-phosphorylation of RyR2, promoting the open state of the receptor and inducing Ca2<sup>+</sup> release from the SR to the cytoplasm, increasing the diastolic Ca2<sup>+</sup> leak (waves), and compromising the SR-Ca2<sup>+</sup> load (Delgado et al., 2015; Val-Blasco et al., 2017a).

In many cases, increased diastolic Ca2<sup>+</sup> release is the cause of depressed SR-Ca2<sup>+</sup> load in HF. Our data show that during β-adrenergic stimulation, diastolic Ca2<sup>+</sup> release is significantly increased in failing WT cells, whereas deficiency of NOD1 prevents the increase in the HF-related diastolic Ca2<sup>+</sup> waves. These data go some way to explain the improvement in the SR-Ca2<sup>+</sup> load and the resulting improved systolic Ca2<sup>+</sup> release and cell contractility in Nod1−/−-PMI cells under isoproterenol stimulation.

Nod1−/−-PMI hearts showed physiological levels of phospho-RyR2 (Val-Blasco et al., 2017a), thus contributing to block Ca2<sup>+</sup> diastolic leak in Nod1−/−-PMI cardiomyocytes. Aberrant diastolic Ca2<sup>+</sup> leak in many cases such as HF leads to arrhythmogenesis and cardiac dysfunction (Reiken et al., 2003; Ai et al., 2005; van Oort et al., 2010; Marks, 2013; Terentyev et al., 2014). RyR-derived Ca2<sup>+</sup> leak can cause arrhythmias by Ca2<sup>+</sup> waves that trigger delays after depolarizations (Lehnart et al., 2006). Pro-arrhythmogenic events are common in failing patients, particularly under β-adrenergic stimulation, and around 30–50% of these individuals die from sudden cardiac death, mostly associated with ventricular arrhythmias. Our data suggest that blocking NOD1 can be a new tool to prevent ventricular arrhythmias and cardiac dysfunction resulting from Ca2<sup>+</sup> mishandling in HF.

In line with our results, elevated levels of the inflammatory cytokines TNFα and IL-1β have been reported in both tissue and plasma from HF patients with arrhythmias (Levine et al., 1990; Testa et al., 1996; Aukrust et al., 2004). Moreover, other inflammatory mediators such as Toll-like receptors (TLRs) have been implicated in the generation of ventricular and atrial arrhythmias (Katoh et al., 2014; Monnerat-Cahli et al., 2014; Gurses et al., 2016). Importantly, TLRs seem to play a role in the regulation of the β-adrenergic response. Accordingly, Kizaki et al. (2008) demonstrated that TLR4 stimulation induces a downregulation of β2 receptors in macrophages (Wang et al., 2009). In this regard, our results demonstrate that hearts from Nod1−/−- PMI mice have higher mRNA levels of β1 and β2 adrenergic receptors than WT-PMI hearts, suggesting a possible mechanism for the larger systolic Ca2<sup>+</sup> release and SR Ca2<sup>+</sup> reuptake in by Nod1−/−-PMI mice. β-adrenergic receptors are activated through heterotrimeric G proteins and subsequent activation of the adenylyl cyclase, which modulates the activity of several proteins essential for EC Ca2<sup>+</sup> coupling, such as ICaL, RyR2, and phospholamban (Mayourian et al., 2018). Therefore, the elevated β-adrenergic receptor levels in Nod1−/−-PMI cells may increase the efficiency of cardiac EC coupling through the augmented function of ICaL, RyR, and SERCA, improving both systolic Ca2<sup>+</sup> release and SR Ca2<sup>+</sup> loading in failing hearts.

One of the most common treatments to manage β-adrenergic impairment in HF patients is the use of β-blockers. It has been demonstrated that β-blocker treatment reduces the detrimental effects of catecholamine stimulation, such as toxicity, pathological elevated heart rate, and adverse remodeling and arrhythmia development in failing patients. Improving β-adrenergic regulation of Ca2<sup>+</sup> handling by blocking NOD1 can be a new tool to improve the cardiac function in failing hearts.

#### CONCLUSION

Our study demonstrates that deficiency of NOD1 improves the β-adrenergic modulation of Ca2<sup>+</sup> handling in isolated cells from failing mice. Therefore, NOD1 emerges as a

new potential target in the treatment of cardiac dysfunction and ventricular arrhythmias associated with HF.

## AUTHOR CONTRIBUTIONS

MF-V conceived, designed, and discussed the experiments and wrote the paper. AV-B, JN-G, MT, PP, CZ, MP, LR-R, and MG-F performed the experiments. MF-V, AV-B, JN-G, CD, GR-H, and PP analyzed the data. MF-V, LB, GR-H, and CD contributed reagents, materials, and analysis tools.

#### FUNDING

MF-V is Miguel Servet II researcher of ISCIII (MSII16/00047 Carlos III Health Institute). This work was supported by grants PI14/01078, CP15/00129, PI17/01093, and PI17/01344 from ISCIII, Fondo Europeo de Desarrollo Regional (FEDER),

#### REFERENCES


Ministerio de Economía y Competitividad of Spain (SAF2014- 57190-R; SAF2017-84777-R), FSE, and CIBER-CV, a network funded by ISCIII.

#### ACKNOWLEDGMENTS

We acknowledge the technical assistance of L. Martin, V. Terrón, D. Navarro, and L. Sánchez. We would like to thank Dr. Gabriel Núñez (University of Michigan, United States) for kindly providing Nod1−/<sup>−</sup> mice and Dr. Kenneth McCreath for editorial support.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.00702/full#supplementary-material

transduction and pharmacogenomics in heart failure. Nat. Clin. Pract. Cardiovasc. Med. 2, 475–483. doi: 10.1038/ncpcardio0309


calcium leak and arrhythmias. Proc. Natl. Acad. Sci. U.S.A. 103, 7906–7910. doi: 10.1073/pnas.0602133103


RyR2 activity in failing mice cardiomyocytes. J. Gen. Physiol. 146, 295–306. doi: 10.1085/jgp.201511366


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Val-Blasco, Navarro-García, Tamayo, Piedras, Prieto, Delgado, Ruiz-Hurtado, Rozas-Romero, Gil-Fernández, Zaragoza, Boscá and Fernández-Velasco. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Redox Dependent Modifications of Ryanodine Receptor: Basic Mechanisms and Implications in Heart Diseases

#### Roman Nikolaienko, Elisa Bovo and Aleksey V. Zima\*

Department of Cell and Molecular Physiology, Loyola University Chicago, Maywood, IL, United States

Heart contraction vitally depends on tightly controlled intracellular Ca regulation. Because contraction is mainly driven by Ca released from the sarcoplasmic reticulum (SR), this organelle plays a particularly important role in Ca regulation. The type two ryanodine receptor (RyR2) is the major SR Ca release channel in ventricular myocytes. Several cardiac pathologies, including myocardial infarction and heart failure, are associated with increased RyR2 activity and diastolic SR Ca leak. It has been suggested that the increased RyR2 activity plays an important role in arrhythmias and contractile dysfunction. Several studies have linked increased SR Ca leak during myocardial infarction and heart failure to the activation of RyR2 in response to oxidative stress. This activation might include direct oxidation of RyR2 as well as indirect activation via phosphorylation or altered interactions with regulatory proteins. Out of ninety cysteine residues per RyR2 subunit, twenty one were reported to be in reduced state that could be potential targets for redox modifications that include S-nitrosylation, S-glutathionylation, and disulfide cross-linking. Despite its clinical significance, molecular mechanisms of RyR dysfunction during oxidative stress are not fully understood. Herein we review the most recent insights into redox-dependent modulation of RyR2 during oxidative stress and heart diseases.

#### Edited by:

Laetitia Pereira, Institut National de la Santé et de la Recherche Médicale (INSERM), France

#### Reviewed by:

Juan Antonio Rosado Dionisio, Universidad de Extremadura, Spain Geoffrey Woodard, Harvard University, United States

> \*Correspondence: Aleksey V. Zima azima@luc.edu

#### Specialty section:

This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology

Received: 27 September 2018 Accepted: 23 November 2018 Published: 06 December 2018

#### Citation:

Nikolaienko R, Bovo E and Zima AV (2018) Redox Dependent Modifications of Ryanodine Receptor: Basic Mechanisms and Implications in Heart Diseases. Front. Physiol. 9:1775. doi: 10.3389/fphys.2018.01775 Keywords: heart, Ca signaling, ryanodine receptor, sarcoplasmic reticulum, oxidative stress

# EXCITATION-CONTRACTION COUPLING

Regular heart contraction critically depends on precisely-controlled cytosolic Ca regulation during each cardiac cycle. The sarcoplasmic reticulum (SR) Ca release plays a particularly important role in activation of myocyte contraction (Zima et al., 2014). During systole, Ca release from the SR is a result of activation of specialized Ca channels – ryanodine receptors (RyR). These channels are activated by an inward Ca current via L-type Ca channels (LTCCs) during an action potential (AP) (**Figure 1A**). The mechanism of RyR activation by cytosolic Ca is known as Ca-induced Ca release (CICR) (Fabiato, 1983). In ventricular myocytes, CICR occurs at specialized microdomains where a T-tubule of the sarcolemma closely approaches a junction of the SR forming the dyad. The junctional SR membrane contains clusters of RyRs (Franzini-Armstrong et al., 1999; Hayashi et al., 2009). The activation of a single RyR cluster generates a local increase in cytosolic Ca ([Ca]i) called Ca spark (Cheng et al., 1993; **Figure 1B**). The spatio-temporal summation of thousands of

Ca sparks produces the global Ca transient that initiates contraction. During diastole, cytosolic Ca is pumped back into the SR by the Ca-ATPase (SERCA) and extruded from the cell by the Na-Ca exchanger (NCX) (Bers, 2002). The rate at which SERCA and NCX remove Ca from the cytosol determines how quickly cardiac muscle relaxes to allow the heart to fill with blood.

#### RYANODINE RECEPTOR COMPLEX

Three isoforms of RyR have been identified. The RyR1 isoform is dominant in skeletal muscles, whereas the RyR2 represents the cardiac RyRs isoform. The RyR3 isoform is found only at low expression levels in certain skeletal muscle types and brain (Lanner et al., 2010; Meissner, 2017). RyR forms a homo tetrameric assembly comprised of four 560 kDa subunits yielding a total molecular weight of 2,300 kDa (Van Petegem, 2015). The RyR2 is not only an ion channel but also a giant scaffolding protein on which several regulatory proteins and enzymes can be assembled (Fill and Copello, 2002; Bers, 2004; **Figure 1C**). RyR2 forms a multimolecular complex with FK506 binding protein 12.6 (FKBP12.6) and calmodulin (CaM) at the cytosolic side. FKBP12.6 binds to each subunit of RyR2 with high affinity and stabilizes the channel in closed conformation (Brillantes et al., 1994; Timerman et al., 1994; Wehrens et al., 2004). Another small regulatory protein CaM binds RyR with nanomolar affinity, at 1:4 ratio in the absence and presence of Ca (apoCaM and Ca-CaM) (Meissner, 2017). The effect of CaM on RyR activity is isoform specific. At free [Ca] > 1 µM CaM inhibits all three isoforms of RyR, whereas at submicromolar free [Ca] CaM activates RyR1 and RyR3, and inhibits RyR2 (Balshaw et al., 2001; Meissner, 2004). At the luminal side, RyR2 is associated with the complex consisting of triadin-1, junctin and calsequestrin that act together as a luminal Ca sensor (Györke et al., 2004). Other reports suggest RyR2 interactions with junctophilin, homer-1, sorcin and S100A1 (Song D.W. et al., 2011). Additionally, the RyR2 complex comprises enzymes that regulate the channel activity through the interplay of phosphorylation and dephosphorylation by protein kinases and protein phosphatases. Protein kinase A (PKA) phosphorylates RyR2 at serines 2030 and 2808, whereas Ca-CaM dependent protein kinase II (CaMKII) phosphorylates the channel at serine 2814 (Marx et al., 2000; Wehrens, 2004; Xiao et al., 2006). Phosphorylation can be reversed by associated with the RyR2 complex protein phosphatases 1 and 2A (PP1 and PP2A) (Marx et al., 2000).

# RYR2-MEDIATED SR Ca LEAK

The majority of SR Ca release occurs during systole via the mechanism of CICR. During diastole, however, spontaneous openings of RyR2s can produce SR Ca leak (Shannon et al., 2003; Zima et al., 2014). At normal physiological conditions, a large fraction of SR Ca leak occurs as uncoordinated openings of individual RyR2s or spark-independent Ca leak. This leak component can serve as an important protective mechanism against SR Ca overload. (Zima et al., 2010). However, in pathological conditions associated with increased RyR2 activity, the majority of SR Ca leak occur in a form of Ca sparks and Ca waves. By activating the electrogenic NCX, spontaneous Ca waves can generate delayed afterdepolarizations (DADs), an effective trigger of cardiac arrhythmias (Pogwizd and Bers, 2004). Increased diastolic SR Ca leak through RyR2s channel has been also implicated in the development of several cardiac pathologies, including heart failure (HF) (Shannon et al., 2003; Kubalova et al., 2005; Belevych et al., 2007; Bers, 2014). It has been suggested that SR Ca leak contributes to the depressed Ca transients and the reduced SR Ca load in HF. The increased RyR2 mediated Ca leak has been also implicated in the progression of arrhythmogenesis in failing hearts (Blayney and Lai, 2009). It has been suggested that oxidative post-translational modifications in RyR2 may play an important role in the abnormal channel activity and the increased SR Ca leak in many cardiac pathologies (Mochizuki et al., 2007; Terentyev et al., 2008; Belevych et al., 2009, 2011, 2012; Bovo et al., 2018a,b). The functional effect of these redox modifications, including disulfide oxidation, mixed disulfide formation (S-glutathionylation) and S-nitrosylation, will be further discussed in this review.

## ROS PRODUCTION IN CARDIAC MUSCLE

Cardiac muscle contraction strongly depends on ATP synthesis by the mitochondrial electron transport chain (ETC). In addition to energy production, the ETC activity can lead to generation of reactive oxygen species (ROS). During the reduction of molecular oxygen to water some redox centers, in particular complex I and III, in the ETC may leak electrons to oxygen producing superoxide anion (O<sup>2</sup> <sup>−</sup>•) (Turrens, 2003). O<sup>2</sup> <sup>−</sup>• is further converted into hydrogen peroxide (H2O2•) by superoxide dismutase (SOD) enzymes. Transitional metals can react with H2O<sup>2</sup> with a formation of even stronger oxidant – hydroxyl radical (OH•) (Madamanchi and Runge, 2013). ROS can be also generated by cytochrome P450-based enzymes, xanthine (XO) and NADPH oxidases (NOX) (Zima and Blatter, 2006; Santos et al., 2016). Additionally, cardiomyocytes express two isoforms of nitric oxide (NO•) synthase – endothelial and neuronal NOS (NOS1 and NOS3, respectively) that play an important role in the intracellular signaling (Santos et al., 2011). Besides generation of NO•, NOSs are also responsible for the production of reactive nitrogen species (RNS). Uncoupled NOSs can generate O<sup>2</sup> <sup>−</sup>•, which can be combined with NO• to produce the powerful oxidant peroxinitrite (ONOO−•).

Deleterious effects of increased ROS production include lipid peroxidation, protein oxidation, DNA mutagenesis and DNAprotein cross-linking. Therefore, amounts of ROS inside the cell must be tightly controlled by enzymatic and non-enzymatic antioxidant defense mechanisms (Madamanchi and Runge, 2013). These mechanisms comprise catalase, SOD, glutathione peroxidase and vitamins A, C and E (Zima and Blatter, 2006; Santos et al., 2016). O<sup>2</sup> <sup>−</sup>• produced in mitochondria is converted to H2O<sup>2</sup> by SOD, which is further reduced to water by glutathione peroxidase, catalase, peroxiredoxin

and thioredoxin systems (Foster et al., 2009). Non-enzymatic defense mechanisms rely on small antioxidant molecules, primarily on the intracellular pool of reduced glutathione (GSH). It has been shown that GSH can directly acts as ROS scavenger. Upon its reaction with ROS, GSH is oxidized into GSSG that can be reduced back to GSH by glutathione reductase (Schafer and Buettner, 2001). Glutathione reductase plays a key role in maintaining the intracellular GSH/GSSG ratio in a proper physiological range (30:1–100:1). GSH also contributes to the maintenance of the cellular redox state by acting as a substrate for glutathione peroxidase (Foster et al., 2009). When the amount of ROS production overwhelms the intracellular anti-oxidant defense, oxygen free-radicals cause damage of DNA, lipids and proteins. These uncontrolled conditions are also known as oxidative stress (Giordano, 2005).

At physiological conditions, however, low levels of ROS and RNS are involved in the cell signaling by inducing discrete, reversible and site-specific protein modification (Heymes et al., 2003; Giordano, 2005; Foster et al., 2009). The redox signaling is based on the ability of ROS/RNS to modulate protein cysteines, leading to S-nitrosylation, S-glutathionylation, and disulfide bond formation. Such redox modifications would affect activity of the proteins involved in different signaling cascades. For example, in cardiac muscle NOX activity and ROS production can be stimulated by growth factors and cytokines, such as angiotensin II, PDGF and TNF-α (Giordano, 2005). In these pathways, ROS act as important second messengers related to inflammatory and stress response. Redox modifications of signaling proteins can be reversed by specific enzymes. Disulfide bridges and mixed disulfides (S-glutathionylation) can be reduced by both thioredoxin and glutaredoxin systems. Additionally, thioredoxin system can reduce S-nitrosothiols. Both glutaredoxin and thioredoxin system utilize the reducing power of NADPH for reduction of their key enzymes – glutathione and thioredoxin reductases (Paulina Donoso and Gina Sánchez, 2013).

In the heart, increased ROS production during oxidative stress has been associated with different cardiac pathologies, including myocardial infarction (MI) and HF. During MI, the re-oxygenation of ischemic region is vital for heart survival. However, re-oxygenation also causes significant myocardium damage which has been linked to the toxic effects of ROS (Zweier and Talukder, 2006). Significant increase in ROS production during ischemia/reperfusion (I/R) arises from the uncoupled mitochondrial ETC as well us upregulation of NOX, NOS and XO resulting in increased generation of O•− 2 and decreased GSH/GSSH ratio (Zima and Mazurek, 2016). Oxidative stress during cardiac pathologies is commonly associated with increased SR Ca leak through the hyperactive RyR2. Moreover, ROS can activate hypertrophic and pro-apoptotic signaling pathways, leading to myocardial remodeling (Sawyer et al., 2002; Santos et al., 2016). HF is frequently viewed as a condition of chronic oxidative stress (Mak and Newton, 2001). The unbalanced metabolism has been suggested to play an important role in development of HF (Mak and Newton, 2001; Ventura-Clapier et al., 2004; Santos et al., 2011). As the disease progresses, oxidative stress worsens due to the increasing energy demand and workload of the failing heart, thus perpetuating a deleterious cycle (Seddon et al., 2007). Although HF is associated with a large number of complex intracellular changes, the focus of this review is directed at understanding the role of oxidative stress in SR Ca regulation and RyR2 function. To date, RyR2 dysfunction has been characterized by increased phosphorylation and oxidation levels. While functionally important phosphorylation sites on RyR2 have been characterized (Marx et al., 2000; Wehrens, 2004; Xiao et al., 2006), the specific mechanisms of oxidative modifications of RyR2 and their contribution to defective SR Ca cycling remain incomplete.

## REDOX REGULATION OF RYR2

fphys-09-01775 December 4, 2018 Time: 8:44 # 4

Historically, oxidative stress has been linked to an increase of SR Ca release (Trimm et al., 1986). It has been proposed that oxidation of cysteines in RyR2 causes significant changes in the channel gating (Abramson and Salama, 1989). Out of 90 cysteine residues in the single subunit of RyR2, about 21 are in the free thiol state and available for redox-modifications (Xu, 1998; Donoso et al., 2011). Therefore, it is no surprise that RyR2 has been characterized as the highly redox sensitive ion channel. During oxidative stress, sulfhydryl groups of cysteine residues on RyR2 can be oxidized by ROS producing sulfenic, sulfinic and sulfonic acids (Giles and Jacob, 2002). While there is no evidence for the functional significance of sulfinic and sulfonic acids, sulfenic acid can react with sulfhydryl groups, RNS and GSH, yielding disulfide bridges, S-nitrosylation, S-glutathionylation, respectively (Paulina Donoso and Gina Sánchez, 2013). In general, oxidation of cysteine residues has been suggested to cause RyR2 activation (Suzuki and Ford, 1999; Zima and Blatter, 2006). However, multiple studies have shown that effects of oxidative agents on RyR2 largely depend on experimental conditions (Mi et al., 2015). It has been demonstrated that low concentrations of oxidizing agents activate RyR2, whereas prolonged exposure or high concentrations of oxidants lead to irreversible RyR2 inhibition (Dulhunty et al., 2000). Different cysteine residues have been suggested to play a role in activation or inhibition of RyR2 by oxidative stress.

The role of RyR2 redox modifications in cardiac pathologies has been investigated in numerous studies. It has been shown that the non-selective beta-blocker with antioxidant properties carvedilol is more effective in a treatment of HF than other beta-blockers. In the pacing-induced HF model, carvedilol was able to preserve the cardiac function by stabilizing RyR2 structure. It has been suggested that some of the beneficial effects of carvedilol can be attributed to its ability to prevent oxidation of RyR2. In experiments with administration of the NO−•/ONOO−• donor SIN-1, carvedilol was able to prevent thiol oxidation in RyR2, presumably by acting as a scavenger of ONOO−• (Mochizuki et al., 2007). In the canine model of chronic HF, the increased SR Ca leak has been attributed to redox modification of RyR2 by ROS. Interestingly, reducing agents that target S-nitrosylation and S-glutathionylation failed to reverse SR Ca leak in HF. At the same time, the application of DTT (an agent that can reverse disulfide formation, S-nitrosylation and S-glutathionylation) decreased RyR2-mediated Ca leak, suggesting a central role of disulfide bond formation in this process (Terentyev et al., 2008). Studies of the canine post-MI model with ventricular fibrillation (VF) has also shown increased diastolic SR Ca leak and decreased SR Ca content. Similarly to HF studies, cardiomyocytes isolated from the canine post-MI model were characterized by increased levels of ROS production and RyR2 oxidation (Belevych et al., 2009). Another study of the rabbit I/R model has shown a critical role of RyR2 oxidation in the transition from the positive inotropic to arrhythmogenic effect during β-adrenergic receptor stimulation (Bovo et al., 2018a). Taken together, these data suggest that SR Ca leak due to RyR2 oxidation can be a common mechanism of SR Ca mishandling during different cardiac pathologies. Therefore, reactive cysteines on RyR2 may represent an important target in the development of new therapeutic strategies.

# IDENTIFICATION OF RYR2 REDOX-MODIFICATION SITES

Given the functional significance of redox modifications of RyR, several structural studies were performed in order to allocate and characterize sites of these modifications. In the skeletal isoform RyR1, cysteine 3635 located in the CaM-binding site has been shown to undergo functionally significant S-nitrosylation (Sun et al., 2001). Mass spectroscopy study has identified nine cysteine residues (including cysteine 3635) in RyR1 that were S-alkylated by low doses of the maleamide derivative CPM, suggesting their high-sensitivity to redox modifications (Voss et al., 2004). Other studies have proposed that cysteine 3635 can undergo S-glutathionylation and could be also involved in RyR disulfide cross-linking (Aracena-Parks et al., 2006). Twelve RyR1 cysteine residues, including cysteine 3635, have been proposed to be either S-nitrosylated, S-glutathionylated or to form disulfide bridges. Of those twelve cysteines, nine were shown to be endogenously redox-modified. It has been suggested that cysteine 3635 can form a disulfide bond with cysteine residues in the region 1–2401 of RyR1 (possibly with either cysteine 36, 2326, or 2363). However, the functional significance of these redox-modifications remains unclear yet.

While skeletal RyR1 hyper-reactive cysteines have been characterized, little is known about corresponding redox modifications in the cardiac RyR2. A functional role of RyR2 cysteine 3602 (which corresponds to the hyper-reactive cysteine 3635 in RyR1) in Ca overload-induced Ca release has been studied in HEK293 cells expressing recombinant RyR2. It has been shown that the mutation of RyR2 cysteine 3602 to alanine (C3602A) significantly increased SR Ca fractional release by decreasing termination and increasing activation threshold for CICR. Interestingly, the ability of N-ethylmaleamide (NEM) to increase the activation threshold for CICR for the RyR2 was lost for the C3602A mutant, suggesting functional significance of cysteine 3602 alkylation. However, no difference was observed in the effect of cysteine oxidants on WT and C3602A RyR2 (Mi et al., 2015). Despite ∼70% homology between RyR1 and RyR2, these channels appear to exhibit important structural and functional differences, particularly with respect to redox regulation.

#### S-GLUTATHIONYLATION OF RYR2

High pKa (>8.0) of a sulfhydryl group means that the majority of free thiols is protonated under physiological pH (7.0–7.4) and cannot be oxidized. As a result, protein redox modifications require an initial conversion of a sulfhydryl group to a thiolate anion. Proton dissociation from a thiol group is highly dependent on the local microenvironment and can be facilitated by closely located basic amino-acids (Grek et al., 2013). During oxidative

stress, spontaneous protein S-glutathionylation may occur via multiple pathways that include thiol-disulfide exchange between a protein thiol and GSSG, a reaction between sulfenic acid or thiyl radical (**-**S•) and GSH, as well as S-glutathionylation induced by RNS. In addition to spontaneous reactions, the rate and efficiency of S-glutathionylation can be enhanced by enzymatic activity of glutathione-S-transferases (GSTs) (Zhang et al., 2018). Because of the ability of glutaredoxin and thioredoxin systems to reverse S-glutathionylation, this redox modification can prevent irreversible oxidation of protein thiol groups into sulfonic acids (Townsend, 2007). However, the addition of large and negatively charged glutathione group may also significantly alter structural and functional properties of a protein. Moreover, two adjacent S-glutathionilated thiols can displace GSH groups with a formation of disulfide bond within the protein (Beer et al., 2004).

Several studies have provided evidence for the functional role of RyR2 S-glutathionylation during oxidative stress (**Figure 2**). It has been reported that preconditioning tachycardia increased NOX activity in the SR microsomes isolated from the canine ventricle. The increased NOX activity led to increased RyR S-glutathionylation and elevated systolic SR Ca release (Sánchez et al., 2005). The same effect was described for the SR microsomes isolated from animals preconditioned with exercises. It has been suggested that NOX activation and RyR2 S-glutathionylation during preconditioning may have a protective effect by increasing systolic SR Ca release while limiting diastolic SR Ca leak (Sánchez et al., 2008). In our recent study, we described the effects of oxidized glutathione (GSSG) on SR Ca leak in rabbit ventricular myocytes. We found that an application of GSSG to permeabilized myocytes increased diastolic SR Ca leak and induced RyR2 disulfide cross-linking (Mazurek et al., 2014). HF myocytes were also characterized by elevated GSSG level, RyR2 cross-linking and increased SR Ca leak. Blocking RyR2 crosslinking with the alkylating agent NEM decreased SR Ca leak and prevented depletion of SR Ca load. Based on these results, we suggested that oxidative stress in the failing heart promotes abnormal inter-subunit interactions within the RyR2 complex that increase channel opening leading to the increased SR Ca leak (Bovo et al., 2018a).

Interestingly, some studies suggest an interaction between RyR and GSTs, enzymes that catalyze S-glutathionylation. The members of GST family CLIC-2, GSTO1-1, and GSTM2-2 were identified as endogenous RyR modulators that exert a strong inhibitory effect on RyR2 and weak excitatory effect on RyR1. The isoform-specific effect of GSTM2-2 has been explained by its binding to the RyR2 divergent region 3, not present in RyR1 (Liu et al., 2012) The C-terminal non-catalytic region of the muscle specific GSTM2-2 has been shown to bind to RyR2 independently from N-terminal glutathione transferase activity (Dulhunty et al., 2011). However, it remains unclear whether GSTM2-2 may be involved in RyR2 S-glutathionylation in vivo.

#### S-NITROSYLATION OF RYR2

NO• plays an important role in cardiovascular signaling through its involvement in the cGMP-dependent vasodilation pathway. NO• binds to guanylate cyclase (GC) increasing the production of cGMP that activates several cGMP-dependent enzymes, including cAMP phosphodiesterase (PDE). The PDE activation would decrease cAMP level and PKA activity. This, in turn, would cause a decrease in PKA phosphorylation of several proteins, including RyR2, LTCC and phospholamban (PLB). cGMP can also activate cGMP-dependent kinase (PKG). It has been reported that PKG can phosphorylate RyR2 at the CaMKII site (Takasago et al., 1991). However, the functional significance of PKGmediated RyR2 phosphorylation remains elusive. NO•**-**based signaling reactions can also occur through the NO•**-**dependent redox-modification, known as S-nitrosylation (Lima et al., 2010; **Figure 2**). The reaction of S-nitrosylation involves a transfer of NO• group to a nucleophilic thiol group yielding S-nitrosothiol (Sun and Murphy, 2010). To initiate S-nitrosylation, thiol groups have to be oxidized (Donoso et al., 2011). It has been shown that intracellular proteins can be effectively nitrosylated by endogenously produced NO• as well by exogenous NO• donors (Hess et al., 2005).

The majority of functional studies of RyR S-nitrosylation have been conducted in lipid bilayer experiments with skeletal and cardiac SR vesicles. It has been shown that NO• donors S-nitrosoglutathione and S-nitrosocysteine can induce S-nitrosylation on each RyR2 subunit at three different sites. S-nitrosylation of RyR2 was associated with an increase of the channel open probability. Interestingly, a comparable RyR2 oxidation of 5-6 thiols per subunit showed no significant effect on the RyR2 function, suggesting a different role of thiol S-nitrosylation and oxidation in the channel activity (Xu, 1998). In another study, S-nitrosothiol reagents increased Ca release from skeletal and cardiac muscle SR vesicles, confirming that RyR S-nitrosylation activates RyR (Stoyanovsky et al., 1997). In both reports, the NO•/ONOO−• donor SIN-1 produced a significant RyR activation that could not be reversed by reducing agents, suggesting irreversible thiol oxidation. The effect of NO• donor SNAP on RyR1 activity in lipid bilayers was complex and dependent on several factors including, concentration, membrane potential and presence of agonists. Low concentrations of NO• donor SNAP increased RyR1 activity, whereas high concentrations (∼1 mM) caused channel inhibition (Hart and Dulhunty, 2000). In another study, a pretreatment of RyR1 with low concentrations of NO• donors prevented disulfide cross-linking induced by diamide. At the same time, NO• donors had no significant effect on the channel open probability. Higher doses of NO• donors, however, were able to increase RyR1 activity and induced disulfide cross-liking (Aghdasi et al., 1997). This type of complex regulation suggests the presence of several groups of functional thiol residues that differently modulate the RyR activity by RNS. Moreover, it has been shown that direct activation of RyR1 by submicromolar concentrations of NO• donors can only occur at physiological pO<sup>2</sup> ∼10 mmHg (in contrast to ambient pO<sup>2</sup> 150 mmHg). This suggests that RyR1 thiol residues have to be in the reduced state to be nitrosylated by NO• (Eu et al., 2000). However, these results could not be reproduced by another group. It was reported that NO• donors NOC-12 and GSNO activate RyR1 independently from pO<sup>2</sup> (Cheong et al., 2005). Furthermore, the effect of NO• donors

PKA-dependent phosphorylation may lead to increased FKBP12.6 dissociation from RyR2. The effects of oxidative stress on RyR2 mostly lead to the increased

on RyR1 involved the CaM dissociation and S-nitrosylation at cysteine 3635 (Sun et al., 2003). In contrast to RyR1, RyR2 can be activated only by GSNO, but not NOC-12. S-nitrosylation of cysteine 3602 (corresponding to cysteine 3635 in RyR1) was not required for RyR2 activation by GSNO (Sun et al., 2008).

diastolic SR Ca leak resulting in arrhythmias and contractile dysfunction in variety of cardiac pathologies.

In cardiomyocytes, NOS1 and NOS3 can induce S-nitrosylation of different substrates, depending on their subcellular localization (Lima et al., 2010). NOS3 is localized within the caveolae together with β-adrenergic receptors and LTCCs. NOS1 resides to the SR, where it modulates the activity of RyR2 (Barouch et al., 2002). Study of RyR2 S-nitrosylation in NOS1−/<sup>−</sup> mice has shown decreased S-nitrosylation, but increased oxidation of RyR2. Myocytes isolated from NOS1−/<sup>−</sup> mice were characterized by increased SR Ca leak, decreased SR Ca load and higher propensity of spontaneous Ca waves (Gonzalez et al., 2007). It has been also shown that NOS1 inhibition leads to the development of ventricular arrhythmias under condition of elevated cytosolic [Ca]. The molecular mechanism of arrhythmias has been attributed to the combination of decreased RyR2 S-nitrosylation and increased RyR2 oxidation. Under condition of NOS1 inhibition, the XO inhibitor allopurinol or the NO• donor GSNO could prevent ventricular arrhythmias. Interestingly, NOS1 inhibition was associated with decreased RyR2 phosphorylation at the CaMKII site. Moreover, the combination of NOS1 inhibition and oxidative stress had an additive effect, producing severe arrhythmical phenotype (Cutler et al., 2012). In another study of NOS1−/<sup>−</sup> myocytes, however, the decreased level of RyR2 S-nitrosylation was associated with decreased SR Ca release. NOS1 knockout or NOS inhibition reduced Ca spark frequency and SR Ca leak. Also, single channel recording revealed decreased open probability of RyR2 in the NOS1−/<sup>−</sup> mice. The NO• donor SNAP was capable to reverse the observed effects of NOS1 inhibition (Wang et al., 2010). The discrepancy between these two studies may be explained by the complex interplay between S-nitrosylation and oxidation in the regulation of RyR2 activity. It seems that physiological levels of RyR2 S-nitrosylation can be necessary for maintaining the proper channel activity in the healthy heart. Under conditions of oxidative stress, however, S-nitrosylation may function as a protective mechanism against irreversible thiol oxidation and RyR2 dysfunction. Thus, de-nitrosylation of RyR2 would inhibit the channel activity in control conditions and activate it during oxidative stress.

## RYR INTERSUBUNIT DISULFIDE CROSS-LINKING

The idea of intersubunit cross-linking has been first introduced for RyR1 (Abramson and Salama, 1989) and then further confirmed for RyR2 (Mazurek et al., 2014). It has been suggested that intersubunit disulfide bond formation leads to structural rearrangements of the channel that cause SR Ca leak (**Figure 2**). Structural aspects of disulfide cross-linking in RyR1 have been extensively characterized in several studies (Moore et al., 1999; Zhang et al., 2003; Aracena-Parks et al., 2006). A cryo-EM analysis of RyR1 oxidation by H2O<sup>2</sup> revealed a disulfide cross-linking together with significant changes in the channel morphology (Han et al., 2006). These data also revealed that

disulfide bonds can be formed between subunits within the RyR1 complex rather than between different RyR1 tetramers. Multiple studies of RyR1 have suggested that cysteine 3635 can be involved in the disulfide cross-linking and the CaM binding. However, recent Cryo-EM analysis of RyR1 and RyR2 did not provide a structural basis for the possible disulfide formation in this region (Yan et al., 2015; des Georges et al., 2016; Wei et al., 2016; Dhindwal et al., 2017). It has been suggested that disulfide bond formation may take place between N-terminal regions of RyR2 subunits (Zissimopoulos et al., 2013). Alternatively, RyR cross-linking may occur not only between subunits within one channel, but also between subunits of different RyR channels. It is well-known that skeletal and cardiac muscle RyRs are arranged as large clusters (Franzini-Armstrong et al., 1999; Cabra et al., 2016). Given the tight packing of RyRs in the cluster, there is a possibility of disulfide cross-linking between subunits of two neighboring channels (**Figure 2**). However, there are no direct data demonstrating this mechanism.

Because the intersubunit interactions within the RyR2 complex dictate the channel gating (Orlova et al., 1996; Serysheva et al., 2008; Tung et al., 2010), any post-translational modifications that affect an interaction between RyR2 subunits should have a significant impact on the RyR2 function. We found a positive correlation between RyR2 cross-linking and SR Ca leak, suggesting that the post-translational modification is a strong regulator of RyR2. We have also reported that RyR2 oxidation and the intersubunit cross-linking play an important role in activation of SR Ca leak during oxidative stress (Mazurek et al., 2014) and generation of pro-arrhythmogenic Ca waves during excessive adrenergic activation (Bovo et al., 2015). Furthermore, a significant level of RyR2 cross-linking was observed in ventricular myocytes isolated from the rabbit HF model (Bovo et al., 2018a). This RyR2 modification contributed to an increase of SR Ca leak and the blunted force**-**frequency response of the failing heart.

#### CALMODULIN DISSOCIATION DURING OXIDATIVE STRESS

CaM acts as an important regulatory protein of RyR2 that inhibits channel activity under both low and high [Ca] (Balshaw et al., 2001). During oxidative stress, increased levels of GSSG can decrease the binding affinity of apo-CaM and Ca-CaM to RyR2 (**Figure 2**). This would relieve the inhibitory effect of CaM on the RyR2 activity (Balshaw et al., 2001). Experiments in permeabilized ventricular myocytes have shown that RyR2 oxidation by H2O<sup>2</sup> caused an alteration of the channel structure toward its arrhythmogenic unzipped conformation. This abnormal RyR2 conformation decreased the binding affinity to CaM, which can be restored by the RyR antagonist dantrolene (Oda et al., 2015). At the same time, oxidation caused no effect on the FKBP12.6 binding to RyR2. Dissociation of CaM from RyR2 due to the defective inter-domain interaction (or unzipping) has been linked to the increased SR Ca leak in HF (Ono et al., 2010). It has been shown that CaM isoforms with increased binding affinity to RyR2 have a potential to rescue the aberrant SR Ca release in HF (Hino et al., 2012). Furthermore, a recombinant CaM with decreased rate of dissociation from its binding domain on RyR2 unveiled anti-arrhythmogenic properties. In vivo gene delivery of this recombinant CaM into the heart partially restored RyR2 refractoriness and decrease a chance of arrhythmias in the catecholaminergic polymorphic ventricular tachycardia (CPVT) model caused by calsequestrin mutation (Liu et al., 2018).

## RYR2 PHOSPHORYLATION AND OXIDATIVE STRESS

At least three functionally important phosphorylation sites have been identified in RyR2, including two PKA specific sites (serine 2030 and 2808) and one CaMKII site (serine 2814) (Marx et al., 2000; Xiao et al., 2006; Marx and Marks, 2013). RyR2 phosphorylation by PKA occurs in response to βadrenergic receptor activation. Epinephrine and norepinephrine binding to β-adrenergic receptors increases adenylyl cyclase activity and cAMP production, following by PKA activation (El-Armouche and Eschenhagen, 2009). It has been shown that RyR2 phosphorylation at serine 2808 increased RyR2 open probability and caused FKBP12.6 dissociation from the channel (Marx et al., 2000; **Figure 2**). However, studies from different laboratories have yielded conflicting results regarding the role of RyR2 phosphorylation by PKA in HF (Benkusky et al., 2007; Shan et al., 2010a,b; Zhang et al., 2012). Interestingly, increased levels of RyR2 oxidation and S-nitrosylation have been reported in transgenic mice with the RyR2 mutation mimicking the constitutively phosphorylated serine 2808 (S2808D). While RyR2 activity in the S2808D mutant was normal at young age, in older animals the RyR2 displayed increased SR Ca leak caused by its oxidation and S-nitrosylation together with depletion of FKBP12.6 and other regulatory proteins (Shan et al., 2010a). These results were in line with another study in which mutation of one of two RyR2 phosphorylation sites (serine 2808 or 2814) reduced both oxidative stress and SR Ca leak in the mouse model of Duchenne muscular dystrophy (DMD) (Wang et al., 2015). Earlier studies have shown that DMD patients can develop ventricular arrhythmias associated with leaky RyR2 due to S-nitrosylated and FKBP12.6 depletion (Fauconnier et al., 2010).

While the functional significance of RyR2 phosphorylation has been debated, it is important to highlight that both CaMKII and PKA are sensitive to the intracellular redox state (Johnston et al., 2015; **Figure 2**). In contrast to the cAMP-dependent PKA activation, PKA activation by oxidative stress is associated with the disulfide bond formation between two regulatory PKA RIα subunits. In the proposed model, the disulfide cross-linking of two RIα subunits leads to the increased affinity of PKA to AKAP, that promotes subcellular targeting of the kinase catalytic subunit to the corresponding substrate (Brennan et al., 2006). Similarly to PKA, CaMKII can be also activated by oxidative stress (Erickson

et al., 2008). The mechanism of ROS-dependent CaMKII activation required an initial activation by Ca-CaM followed by oxidation of methionine 281 and 282 yielding a persistent activation even after Ca-CaM dissociation. The mechanism of CaMKII activation by ROS was independent from the previously described activation by autophosphorylation. In in vivo experiments, NOX activation by angiotensin II (AngII) led to CaMKII oxidation. This activation was not observed in the p47−/<sup>−</sup> transgenic mice lacking NOX activity (Erickson et al., 2008).

Since HF is commonly associated with oxidative stress (Mak and Newton, 2001), CaMKII activity is expected to be upregulated during this pathology. Once activated by ROS, CaMKII can contribute to arrhythmogenesis in HF by activating late sodium current (Wagner et al., 2011). In another study, a short-termed exposure of myocytes to H2O<sup>2</sup> has resulted in the prolonged activation of CaMKII and long-termed facilitation of LTCC through mechanisms of CaMKII autophosphorylation and oxidation (Song Y.H. et al., 2011). Consistent with these findings, it has been shown that the pharmacological inhibition of CaMKII prevents arrhythmias induced by oxidative stress (Xie et al., 2009). The upregulation of CaMKII has been associated with the development of atrial fibrillation (AF). It has been shown that RyR2 phosphorylation at the CaMKII site was increased in AF patients, whereas RyR2 expression level was decreased. Cardiomyocytes from AF patients were characterized by increased SR Ca leak, which can be normalized by the CaMKII inhibitor KN93 (Neef et al., 2010). Moreover, it has been shown that NOX activation with AngII caused increased ROS production and CaMKII oxidation followed by an increase in Ca spark frequency. At the same time, the transgenic mice lacking the CaMKII phosphorylation site on RyR2 (S2814A) were protected from AF induced by AngII (Purohit et al., 2013). It has been shown that I/R-mediated arrhythmias were associated with CaMKII-dependent phosphorylation of RyR2 (Said et al., 2011). I/R also increased levels of RyR2 reversible redox modifications: S-glutathionylation and S-nitrosylation (Becerra et al., 2016). Arrhythmogenic effect of cardiac glucosides has been also attributed to the increased ROS production and RyR2 oxidation (Ho et al., 2011). It has been shown later that cardiac glucosides activate NOX2-mediated ROS production that causes CaMKII activation, RyR2 phosphorylation and Ca waves (Ho et al., 2014).

#### FKBP DISSOCIATION DURING OXIDATIVE STRESS

While there is no evidence that RyR2 oxidation directly causes FKBP12.6 dissociation from the channel, oxidative stress still may affect the RyR2-FKBP12.6 interaction through the mechanism of RyR2 phosphorylation (**Figure 2**). RyR2 hyperphosphorylation at the PKA site has been shown to promote FKBP12.6 dissociation from RyR2 resulting in the increased channel activity during HF. It was suggested that the RyR2 hyper-phosphorylation was a result of PP1 dissociation from the RyR2 macromolecular complex (Marx et al., 2000). The interplay between RyR2 oxidation and FKBP12.6 dissociation has been studied in WT and the transgenic mice lacking the PKA phosphorylation site in RyR2 (the S2808A mice). In the WT mice excessive β-adrenergic stimulation caused increased FKBP12.6 dissociation from RyR2 in WT mice, but not in the S2808A mice. At the same time, the levels of RyR2 oxidation were increased in both WT and S2808A mice. To study the effects of oxidative stress and phosphorylation on the RyR2- FKBP12.6 interaction, SR vesicles were treated either with H2O<sup>2</sup> or with H2O<sup>2</sup> combined with CaMKII or PKA. Both H2O<sup>2</sup> alone or in combination with CaMKII showed no significant effect on the FKBP12.6 dissociation from RyR2. However, RyR2 oxidation together with PKA caused almost complete FKBP12.6 dissociation. Chronic treatment of WT mice with isoproterenol increased levels of PKA-RyR2 phosphorylation, FKBP12.6 depletion and RyR2 oxidation, whereas in the S2808A mice these effects of adrenergic stimulation were partially abolished (Shan et al., 2010b).

# CONCLUSION

The vast body of evidence demonstrates a direct link between oxidative stress, RyR2 oxidation and increased SR Ca leak in several cardiac pathologies, including HF and MI. Molecular mechanisms of the impaired SR Ca handling can include structural and functional changes in the RyR2 complex due to thiol redox-modifications. However, effects of oxidative stress on RyR2 may go beyond the direct redox-modification of the channel and can involve dissociation of regulatory proteins and increased phosphorylation by PKA and CaMKII. It appears that phosphorylation and redox modifications may have an additive effect on RyR2 function. The experiments in the non-ischemic canine HF model have shown that RyR2 phosphorylation and thiol oxidation contribute to the different stages of HF: RyR2 phosphorylation by CaMKII manifests on the early stages of HF with following RyR2 oxidation during later stages (Belevych et al., 2011). In another HF study, the increased SR Ca leak was associated with RyR2 hyper-phosphorylation on both PKA and CaMKII sites together with thiol oxidation (Walweel et al., 2017), suggesting a combined effect of multiple factors on RyR2 dysfunction in HF. Despite abundant amount of experimental data pointing on the importance of oxidative stress, many of clinical trials that used antioxidant for treatment of HF did not yield promising results (Mak and Newton, 2001; Giordano, 2005; Thomson et al., 2007). However, trials with the beta-blocker with antioxidant properties carvedilol have shown some superiority over conventional beta-blockers, which can be attributed to its antioxidant effect (Packer et al., 2001). At the same time, it has been shown that carvedilol may also modulate the function of RyR2 independently from its beta-blocking function or antioxidant activity, but through the direct action on RyR2 activity (Zhou et al., 2011). Nevertheless, strategies focused on restoring the RyR2 structural integrity during oxidative stress may have a high therapeutic potential. Thus, further insights into molecular basis of RyR2 redox regulation are essential for the development of specific and effective therapeutic strategies.

# AUTHOR CONTRIBUTIONS

fphys-09-01775 December 4, 2018 Time: 8:44 # 9

All authors (RN, EB, and AZ) contributed to studying of literature and writing of the manuscript. All authors have approved the version to be published.

#### REFERENCES


## FUNDING

This work was supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under Award No. R01HL130231 (to AZ).


dystrophy. Proc. Natl. Acad. Sci. U.S.A. 107, 1559–1564. doi: 10.1073/pnas. 0908540107




rabbit ventricular myocytes. J. Physiol. 588, 4743–4757. doi: 10.1113/jphysiol. 2010.197913


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Nikolaienko, Bovo and Zima. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Long-Term Regulation of Excitation–Contraction Coupling and Oxidative Stress in Cardiac Myocytes by Pirfenidone

Adrián Monsalvo-Villegas, Diana Stephanie Osornio-Garduño and Guillermo Avila\*

Department of Biochemistry, Cinvestav-IPN, Mexico City, Mexico

#### Edited by:

Maria Fernandez-Velasco, Hospital Universitario La Paz, Spain

#### Reviewed by:

Dmitry Terentyev, Brown University, United States Julieta Palomeque, Universidad Nacional de La Plata, Argentina

> \*Correspondence: Guillermo Avila gavila@cinvestav.mx

#### Specialty section:

This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology

Received: 24 September 2018 Accepted: 29 November 2018 Published: 13 December 2018

#### Citation:

Monsalvo-Villegas A, Osornio-Garduño DS and Avila G (2018) Long-Term Regulation of Excitation–Contraction Coupling and Oxidative Stress in Cardiac Myocytes by Pirfenidone. Front. Physiol. 9:1801. doi: 10.3389/fphys.2018.01801 Pirfenidone (PFD) is used to treat human pulmonary fibrosis. Its administration to animals with distinct forms of cardiovascular disease results in striking improvement in cardiac performance. Here, its functional impact on cardiac myocytes was investigated. Cells were kept 1–2 days under either control culture conditions or the presence of PFD (1 mM). Subsequently, they were subjected to electrical stimulation to assess the levels of contractility and intracellular Ca2+. The PFD treatment promoted an increase in both peak contraction and kinetics of shortening and relaxation. Moreover, the amplitude and kinetics of Ca2<sup>+</sup> transients were enhanced as well. Excitation–contraction coupling (ECC) was also investigated, under whole-cell patch-clamp conditions. In keeping with a previous report, PFD increased twofold the density of Ca2<sup>+</sup> current (ICa). Notably, a similar increase in the magnitude of Ca2<sup>+</sup> transients was also observed. Thus, the gain of ECC was unaltered. Likewise, PFD did not alter the peak amplitude of caffeineinduced Ca2<sup>+</sup> release, indicating stimulation of Ca2+-induced–Ca2+-release (CICR) at constant sarcoplasmic reticulum Ca2<sup>+</sup> load. A phase-plane analysis indicated that PFD promotes myofilament Ca2<sup>+</sup> desensitization, which is being compensated by higher levels of Ca2<sup>+</sup> to promote contraction. Interestingly, although the expression of the Na+/Ca2<sup>+</sup> exchanger (NCX) was unaffected, the decay of Ca2<sup>+</sup> signal in the presence of caffeine was 50% slower in PFD-treated cells (compared with controls), suggesting that PFD downregulates the activity of the exchanger. PFD also inhibited the production of reactive oxygen species, under both, basal conditions and the presence of oxidative insults (acetaldehyde and peroxide hydrogen). Conversely, the production of nitric oxide was either increased (in atrial myocytes) or remained unchanged (in ventricular myocytes). Protein levels of endothelial and neuronal nitric oxide synthases (eNOS and nNOS) were also investigated. eNOS values did not exhibit significant changes. By contrast, a dual regulation was observed for nNOS, which consisted of inhibition and stimulation, in ventricular and atrial myocytes, respectively. In the latter cells, therefore, an up-regulation of nNOS was sufficient to stimulate the synthesis of NO. These findings improve our knowledge of molecular mechanisms of PFD action and may also help in explaining the corresponding cardioprotective effects.

Keywords: intracellular Ca2+, Ca2<sup>+</sup> channel, contractility, CICR, Ca2+-induced Ca2<sup>+</sup> release, cardiomyocyte

# INTRODUCTION

fphys-09-01801 December 11, 2018 Time: 17:41 # 2

Excitation–contraction coupling (ECC) is of pivotal relevance for cardiac contractility. ECC relies, in turn, on a critical phenomenon known as Ca2+-induced–Ca2+-release (CICR). An action potential (AP) activates voltage-gated Ca2<sup>+</sup> channels of the plasma membrane (CaV1.2), allowing entry of extracellular Ca2<sup>+</sup> (ICa) that activates ryanodine receptors (RyR2) and a consequent release of Ca2<sup>+</sup> from the sarcoplasmic reticulum (SR). The resulting increase in cytosolic Ca2<sup>+</sup> (Ca2<sup>+</sup> transient) activates the contractile machinery, and relaxation occurs as the levels of Ca2<sup>+</sup> decay thanks to the activity of both the sarco/endoplasmic reticulum Ca2<sup>+</sup> ATPase (SERCA) and the Na+/Ca2<sup>+</sup> exchanger (NCX; for reviews see Bers, 2002; Eisner et al., 2017; Landstrom et al., 2017).

A small synthetic compound termed pirfenidone (PFD, 185 g/mol) is used to treat human pulmonary fibrosis. Many studies have reported that PFD also improves cardiac performance in animal models of: atrial fibrillation (AF, Lee et al., 2006), myocardial infarction (Nguyen et al., 2010; Li et al., 2017; Adamo et al., 2018), Duchenne muscular dystrophy (Van Erp et al., 2006), pressure overload (Mirkovic et al., 2002; Wang et al., 2013; Yamagami et al., 2015), hypertrophy (Yamazaki et al., 2012), diabetic cardiomyopathy (Miric et al., 2001), and diphtheritic myocarditis (Adamo et al., 2018).

In marked contrast, little is known about the corresponding molecular mechanisms. A reduction in tissue fibrosis and stiffness, attributed to low expression levels of a profibrogenic cytokine (transforming growth factor-beta 1, or TGF-β1) is thought to be involved (Avila et al., 2014; Lopez-de la Mora et al., 2015; Graziani et al., 2018). Nevertheless, some effects do not require lowering levels of TGF-β1 or fibrosis. For example, although PFD improves cardiac contractility in dystrophindeficient (mdx) mice, this action occurs in the absence of changes in tissue fibrosis or stiffness (Van Erp et al., 2006). In addition, treating cardiac myocytes with PFD (1–2 days) induces a substantial increase in the activity of CaV1.2, by a TGF-β1 independent mechanism (Ramos-Mondragón et al., 2012).

Here, we have investigated whether PFD regulates the function of cardiac myocytes. Our present results show that, indeed, in both atrial and ventricular myocytes, PFD improves the magnitude and kinetics of contractility. This can be explained by a concomitant stimulus on both CaV1.2 and CICR. PFD also elicited a marked decrease in NCX activity and promoted changes in the synthesis rate of cytosolic free radicals [reactive oxygen species (ROS) and nitric oxide (NO)]. Interestingly, the compound also regulated the expression of neuronal NO synthase (nNOS), but not of that corresponding to NCX or eNOS. These findings may be essential to explaining how PFD exerts its cardioprotective effects.

#### MATERIALS AND METHODS

#### Isolation of Cardiac Myocytes

Atrial and ventricular myocytes were obtained as described elsewhere (Ramos-Mondragón et al., 2012; Santamaria-Herrera et al., 2016). The procedure was approved by the Institutional Animal Care and Use Committee (IACUC— CINVESTAV, 0199-16), complies with the Mexican Official Norm NOM-062-ZOO-1999 and is in accordance with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication, 8th Edition, 2011).

In brief, male Wistar rats of approximately 240 g were deeply anesthetized, using a mixture of Ketamine and Xylazine (100:10 mg/Kg, i.p.). The heart was then quickly removed by thoracotomy, cannulated (via aorta) and mounted into a homemade Langendorff system. A warm (37◦C) digestion buffer containing a mixture of collagenase type II and protease was retrogradely perfused, and the tissue (either ventricles or right atrium) was excised as soon as became softened. Subsequently, samples were carefully triturated, centrifuged and the resulting myocytes were plated into laminin-coated coverslips, which in turn were lying into 35 mm Petri dishes. The cells were transferred to a CO<sup>2</sup> incubator, kept at 37◦C and maintained in the presence of standard culture medium [control (CN)] or culture medium supplemented with PFD (1 mM). Unless otherwise specified, experiments were performed 1–2 days after of have initiated the PFD treatment. This protocol was chosen based on previous determinations of the maximal effect of PFD on CaV1.2 (Ramos-Mondragón et al., 2012).

#### Voltage-Clamp Experiments

Ca2<sup>+</sup> transients and ICa were investigated simultaneously using a combination of the voltage-clamp technique and epifluorescence (for a more comprehensive description, see Vega et al., 2011). Briefly, either atrial or ventricular myocytes were subjected to voltage-clamp, using the whole-cell patch-clamp technique. The patch-clamp electrodes were filled with an intracellular solution containing a Ca2+-sensitive dye (Fluo-4) and exhibited an electrical resistance of approximately 4 M. The cell capacitance (Cm) was estimated from linear capacitative currents, which were acquired under both cell-attached and whole-cell conditions. Leak currents were eliminated on-line, using a p/N leak subtraction procedure. For each cell, the peak value of ICa was normalized by C<sup>m</sup> and plotted as a function of Vm. The resulting I–V curves were fitted according to the following Boltzmann equation:

$$I\_{\rm Ca} = G\_{\rm max} (V\_{\rm m} - V\_{\rm rev}) / \{1 + \exp[(V\_{1/2} - V\_{\rm m})/k] \} (1)$$

where Gmax represents the maximal Ca2<sup>+</sup> conductance, V1/<sup>2</sup> is the potential for half-maximal activation of Gmax, k is a slope factor, and Vrev is the apparent reversal potential of ICa.

Levels of free cytosolic Ca2<sup>+</sup> were assessed using Fluo-4. The dye was excited using a 100 W mercury arc lamp (at 470–490 nm), and the emitted fluorescence was collected at 515–550 nm using a photomultiplier tube. The magnitude of Ca2<sup>+</sup> transients was estimated as the maximum fluorescence value during test pulses (F), divided by the basal fluorescence observed before the pulse (Fb), according to: 1F/F = (F – Fb)/Fb. The gain of EC coupling (ECC Gain) was calculated by normalizing maximal 1F/F values to its respective peak value of ICa. The increase in Ca2<sup>+</sup> signal

induced by caffeine (30 mM) was used to estimate the amount of releasable SR Ca2<sup>+</sup> content. Caffeine was applied via a local, rapid perfusion system (Warner Instruments Corporation), first in the absence and then in the presence of extracellular Na<sup>+</sup> (see Recording solutions). For each cell, the NCX function was examined at the holding potential (−50 mV, atrium; and −40 mV, ventricle) by analyzing Ca2<sup>+</sup> signals remaining at the end of caffeine applications (25 s).

#### Measurements of Contractility and Intracellular Ca2<sup>+</sup>

Ca2<sup>+</sup> transients and mechanical properties were recorded simultaneously in intact (non-patch clamped) myocytes, with the aid of an IonOptix system (as described elsewhere, Santamaria-Herrera et al., 2016). In short, the cells were loaded with Fura-2 AM (6 µM, 15 min) and then transferred to a stimulation chamber that contained Tyrode's solution. Subsequently, the chamber was mounted on an inverted microscope equipped with epifluorescence, and electrical stimuli (20 V, 20 ms) were delivery at 0.2 Hz. A particular myocyte was observed on a PC monitor, using a digital video camera. The fluorescence emission ratio of Fura-2 was recorded (F360/F380), and changes in myocyte length during shortening and re-lengthening were measured at both ends of the cell. In ventricular myocytes, the average sarcomere length was also investigated. Briefly, an area of welldefined striations was selected, and then the average periodicity of Z-line densities was determined (based on a fast Fourier transform algorithm). The sampling frequencies for recordings of contractions and intracellular Ca2<sup>+</sup> were 0.25 and 1.0 kHz, respectively.

#### Intracellular Production of ROS and NO

Production of ROS and NO was estimated using fluorescence sensors for these molecules (CM-H2DCF DA and DAF-FM DA; Ríos-Pérez et al., 2016). Either atrial or ventricular myocytes were incubated (15 min at room temperature) with Tyrode's solution supplemented with a 1:1 mixture of Fura-2 AM and the appropriated sensor (3 µM each). Fura-2 was excited at 360 nm (its isosbestic or Ca2+-insensitive point), and its fluorescence was used to normalize signals corresponding to CM-H2DCF DA and DAF-FM DA (to prevent for potential differences in loading capacity). Both sensors were excited at 470–490 nm and examined at 515–550 nm.

#### Western-Blot Analysis

Protein levels of NCX, nNOS and eNOS, were investigated by western-blot analysis (García-Castañeda et al., 2017). The cells were washed with ice-cold PBS and incubated 15 min in lysis buffer (containing inhibitors of proteases and phosphatases). The total protein concentration was determined by Bradford assay, and protein samples (25–50 µg) were separated by electrophoresis in Laemmli buffer before being electrotransferred to PVDF membranes. The membranes were then blocked with 3% BSA, washed and subjected to immunoblotting with primary antibodies for the protein of interest (at 4◦C, overnight). Subsequently, the membranes were incubated with a horseradish peroxidase-conjugated secondary antibody (at 37◦C for 60 min). Immunoreactivity was revealed by chemiluminescence, and bands were quantified using Image J (NIH, Bethesda, MD, United States). Each particular membrane was also probed for β-tubulin, to account for potential variability in the amount of protein loaded. The following antibodies were used. Primary: anti-NCX (R3F1, Swant, 1:1,000), anti-eNOS (610296, BD Bioscience, 1:800), anti-nNOS (610308, BD Bioscience, 1:800), and anti-β-tubulin (32-2600, Invitrogen, 1:4000). Secondary: goat anti-mouse IgG (G-21040, Invitrogen, 1:10,000).

#### Recording Solutions

The Tyrode's solution contained (in mM): NaCl 130, HEPES 25, Glucose 22, KCl 5.4, CaCl<sup>2</sup> 2, MgCl<sup>2</sup> 0.5, NaH2PO<sup>4</sup> 0.3, lactic acid 1.1, and pyruvic acid 3 (pH = 7.4).

In atrial myocytes, Ca2<sup>+</sup> transients and ICa were recorded in the presence of the following solutions. External (in mM): methanesulfonic acid 145, TEA-OH 115, HEPES 10, CaCl<sup>2</sup> 5, MgSO<sup>4</sup> 2, 4-aminopyridine 5, 9-anthracene carboxilic acid 1, glucose 10 and 2-3-butanedione monoxime 2. Internal (in mM): Cs-Asp 100, CsCl 10, HEPES 10, Cs-EGTA 1, CaCl<sup>2</sup> 0.2, glucose 10, TEA-HCl 20, ATP-Mg 5, GTP-tris 0.05 and Fluo-4 K<sup>5</sup> 0.15.

In ventricular myocytes, ICa and Ca2<sup>+</sup> transients were recorded in the presence of slightly different solutions (compared with those used for atrial myocytes). External (in mM): methanesulfonic acid 125, TEA-OH 115, HEPES 10, CaCl<sup>2</sup> 10, MgSO<sup>4</sup> 8, 4-aminopyridine 5, 9-anthracene carboxilic acid 1, 2 aminoethyl diphenylborinate 0.01, Creatine 25, glucose 10 and 2-3-butanedione monoxime 2. Internal (in mM): Cs-Asp 160, CsCl 10, HEPES 10, Cs-EGTA 6, CaCl<sup>2</sup> 1, glucose 10, ATP-Mg 5, TEA-HCl 25, MgCl<sup>2</sup> 0.5, GTP-tris 0.05 and Fluo-4 K<sup>5</sup> 0.05.

The NCX activity was investigated by comparing intracellular Ca2<sup>+</sup> responses triggered by caffeine (30 mM), which was applied using the above-cited external solutions (supplemented or not with 140 mM Na+. EGTA (0.5 mM) was added as well, to chelate any contaminating Ca2+. When required, TEA was omitted to balance the osmolarity.

In all patch-clamp solutions, the pH was adjusted to 7.3.

#### Data Analysis

All data were analyzed using the following software packages: SigmaPlot (Systat Software, Inc., San Jose, CA, United States), IonWizard (IonOptix, Westwood, MA, United States), and pCLAMP (Molecular Devices, Sunnyvale, CA, United States). Experimental results are shown as means ± SEM from the indicated number of experiments (n). Two groups were compared by Student's t-test, and ANOVA was used for multiple statistical comparisons.

All experiments were performed at room temperature (∼24◦C).

# RESULTS

Our previous work has shown that the density of ICa is larger in PFD-treated myocytes than in controls (Ramos-Mondragón et al., 2012). Thus, we first reexamined this aspect, using

FIGURE 1 | Patch-clamp experiments reveal a pirfenidone (PFD)-induced increase in the magnitude of both ICa and Ca2<sup>+</sup> transients. Representative traces of ICa and Ca2<sup>+</sup> transients that were recorded simultaneously in either ventricular (A,B) or atrial myocytes (C). In (A,B), recordings obtained from control (CN) and PFD-treated (PFD) myocytes were superimposed, and the membrane was depolarized to +10 mV. In contrast, (C) illustrates traces obtained at different membrane potentials (Vm, in a control cell; the vertical calibration bar indicates 100 pA and 1.0 1F/F). (D,E) Average peak values of ICa (i) and Ca2<sup>+</sup> transients (ii) that were obtained as in (C), for both atrial (D), and ventricular (E) myocytes. Results are from a total of 28 atrial (11 CN, 17 PFD) and 32 ventricular (17 CN, 15 PFD) myocytes. (F,G) Voltage-dependence of ECC gain, estimated from data shown in (D,E). Closed and gray symbols denote control and PFD-treated cells, respectively. <sup>∗</sup>p < 0.05, ANOVA.

ventricular myocytes. At the same time, a possible effect on SR Ca2<sup>+</sup> release was investigated as well. Remarkably, PFD induced a twofold increase in the magnitude of both ICa and Ca2<sup>+</sup> transients. More specifically, the following values of ICa and Ca2<sup>+</sup> transients were obtained, for control (n = 14) and PFDtreated (n = 16) myocytes: −9.3 ± 2.0 and −20.9 ± 2.7 pA/pF (p < 0.003), and 0.34 ± 0.08 and 0.57 ± 0.06 1F/F (p < 0.05). Some examples of these determinations are show in **Figures 1A,B**.

In subsequent experiments, the voltage-dependence of ICa and Ca2<sup>+</sup> transients was investigated (see the representative traces of **Figure 1C**). Interestingly, the PFD effect on both ICa and Ca2<sup>+</sup> transients could be observed at different membrane potentials (**Figures 1D,E**). Fitting I–V curves to a Boltzmann function (Eq. 1) led to the conclusion that PFD causes a 65% increase in the maximal Ca2<sup>+</sup> conductance (Gmax), in accord with our previous work (Ramos-Mondragón et al., 2012). Specifically, the Gmax values that were estimated from control and PFD-treated myocytes were: 62 ± 8 and 105 ± 10 nS/pF (p < 0.01, atrium); and 184 ± 24 and 297 ± 25 nS/pF (p < 0.003, ventricle). No significant differences were found in the voltage-dependent parameters of GCa (mid-point and slope factor), between control and PFD-treated myocytes.

The gain of ECC was also investigated, and it was found identical in control and PFD-treated myocytes (**Figures 1F,G**). This result suggests that PFD induces an ICa that is not redundant for ECC. In other words, the extra ICa elicits the same degree of Ca2<sup>+</sup> release as its respective control. The activity of SERCA was examined next, as illustrated in **Figure 2A**. More specifically, the decaying phase of the Ca2<sup>+</sup> transient was fitted to a simple exponential equation. The resulting average values of τ are shown in **Figures 2B,C**. It can be appreciated that PDF did not alter this parameter in a wide range of membrane potentials, indicating that PFD does not alter the activity of SERCA.

To assess the amount of releasable SR Ca2+, caffeine was applied under both absence and presence of extracellular Na<sup>+</sup> (see the representative traces shown in **Figures 2D,E**). Interestingly no significant differences were found in the peak values of caffeine response (FPeak), between control and PFDtreated myocytes (in either the absence or presence of Na+, **Figures 2F,H**). These data indicate that the effect of PFD on Ca2<sup>+</sup> transients (**Figure 1**) is not due to changes in the load of SR Ca2+.

Two different concentrations of extracellular Na<sup>+</sup> were used in **Figures 2D–I**, because this approach allows investigating both the function of the NCX and Na+-independent transporters responsible for Ca2<sup>+</sup> extrusion. In particular, we examined 1F/F signals remaining at the end of 25 s of caffeine exposure, in the absence and presence of 140 mM Na (termed F<sup>0</sup> and FNa, see **Figures 2D,E**). In the absence of Na+, the rate of Ca2<sup>+</sup> signal decay seems to be slower in PFD-treated cells (**Figures 2D,E**, 0 Na+). To quantify this observation, for each cell, F<sup>0</sup> was normalized to its corresponding peak caffeine response (FPeak). The resulting average values were: 0.59 ± 0.04 and 0.78 ± 0.04 (p = 0.001), for control (n = 30) and PFD-treated cells (n = 31). This analysis suggests that PFD inhibits Na+-independent Ca2<sup>+</sup> extrusion systems.

The contribution of Na<sup>+</sup> to decreasing [Ca2+] was next estimated, from FNa/F<sup>0</sup> ratios (1.0 means no contribution). Surprisingly, in PFD-treated cells, the NCX activity was not detected (**Figures 2G,I**, gray bars), whereas in control myocytes the exchanger contributed by 45% (**Figures 2G,I**, closed bars). The NCX function was also more directly compared between control and PFD-treated cells (using the FNa/F<sup>0</sup> values of **Figures 2G,I**). On average, the FNa/F<sup>0</sup> values were: 0.60 ± 0.06 and 0.88 ± 0.06 (p = 0.012), for control (n = 21) and PFD-tread cells (n = 29). These data suggest that PFD inhibits the NCX activity by 30%.

Then we wonder if the effects observed in ECC (under nonphysiological conditions; i.e., patch-clamp, **Figure 1**), would have functional consequences. Thus, we investigated the contractility

of both sarcomere and cardiomyocytes. Remarkably, PFD increased not only the percent of cell and sarcomere contraction (by 50%) but also the corresponding kinetics of contraction and relaxation (by 30–70%; **Figure 3** and **Supplementary Figure 1**). These PFD effects should lead to improving cardiac blood flow, which is in keeping with results of Nguyen et al. (2010).

In **Figure 3**, contractility was recorded in parallel with Ca2<sup>+</sup> transients. An analysis of the corresponding Ca2<sup>+</sup> signal indicated that, in atrial myocytes, PFD promotes not only Ca2<sup>+</sup> transients of higher magnitude but also lower values of diastolic Ca2<sup>+</sup> (compared with controls, **Figure 4**, atrium). With regard to ventricular myocytes, PFD did not alter the amplitude of the Ca2<sup>+</sup> transient but increased the average values of systolic Ca2<sup>+</sup> (**Figure 4**, ventricle). Additionally, PFD increased both maximum rate of Ca2<sup>+</sup> output (in atrium, 21.1 ± 1.6 1Ratio/s versus 27.5 ± 1.5 1Ratio/s; p < 0.05) and lag time between electrical stimulus and initiation of Ca2<sup>+</sup> transient decay (in ventricle, 84 ± 4 ms versus 95 ± 4 ms; p < 0.05). As a consequence of these changes, the PFD treatment resulted in an increased integral of the initial phase of the Ca2<sup>+</sup> transient (**Supplementary Figure 2**). Thus, in intact cardiomyocytes, PFD stimulates CICR and thereby also enhances contractility (**Figure 3** and **Supplementary Figure 1**).

We also build phase-plane diagrams, to analyze myofilament Ca2<sup>+</sup> sensitivity (**Figure 5A**). The corresponding results indicate that, in the ventricle, PFD not only increases the amplitude of shortening (**Figure 3** and **Supplementary Figure 1**) but also shifts the phase-plane trajectories to the right (**Figures 5A,B**). Although in atrium a similar shift to the right was absent (**Figures 5A,B**), an increase in the slope of relaxation was found (**Figure 5C**). This increase was also present in the ventricle (**Figure 5C**) and suggests that PFD promotes myofilament Ca2<sup>+</sup> desensitization (Spurgeon et al., 1992), which is being compensated by higher levels of [Ca2+]<sup>i</sup> to promote contraction (see above).

FIGURE 3 | Pirfenidone improves the contractility of cardiac myocytes. (A,B) Electrically evoked contractions from atrial (A) and ventricular (B) myocytes previously kept in culture under either control (CN) or PFD conditions (PFD). Plotted are mean values that were estimated from a total of 56 atrial (30 CN, 26 PFD) and 110 ventricular (59 CN, 51 PFD) myocytes. The corresponding average values (mean ± SEM) of peak cell shortening (C,D) and the maximum speed of contraction (E,F) and relaxation (G,H) are also shown. Black and gray symbols/bars indicate control and PFD-treated cells. <sup>a</sup>p < 0.01, <sup>b</sup>p < 0.005, <sup>c</sup>p < 0.001.

stimulated atrial (A; 42 CN, 37 PFD) and ventricular (E; 81 CN, 80 PFD) myocytes. Average values (mean ± SEM) of diastolic (C,D) and systolic (E,F) Ca2<sup>+</sup> are also shown, along with the maximum amplitude of the Ca2<sup>+</sup> transient (G,H). Black and gray symbols/bars indicate control and PFD-treated cells. <sup>a</sup>p < 0.05, <sup>b</sup>p < 0.01.

The cAMP-dependent protein kinase (PKA) phosphorylates key elements of ECC, and thus we investigated whether this protein might be involved in the stimulus of contractility. Ventricular myocytes were exposed 30 min to PKI (a cellpermeable PKA inhibitor), which failed in reverting PFD effects (**Supplementary Figure 3**). Although in this experiment Ca2<sup>+</sup> transients were not investigated, the finding that PKI does not interfere with the magnitude or kinetics of cell shortening (**Supplementary Figure 3**) suggests that myofilament desensitization (**Figure 5**) is not due to PKA-dependent phosphorylation of cTnI (Cheng and Regnier, 2016).

Elevated levels of free radicals production are involved in several cardiovascular disorders (for reviews see Zhang et al., 2004; Korantzopoulos et al., 2007; Köhler et al., 2014). Thus, the possibility that PFD regulates the generation of ROS was also investigated. The corresponding results indicate that PFD causes a significant reduction in the signal of a ROS-sensitive fluorescent compound (CM-H2DCFDA), revealing a 15–30% inhibition of ROS production (**Figure 6A**). Conversely, a 1.3-fold upregulation of NO synthesis was also observed, particularly in atrial myocytes (**Figure 6B**). A minimum of 8 h of PFD-treatment was needed to observe a significant effect on NO and ROS (**Supplementary Figure 4**), suggesting that changes in protein expression might be involved.

An acute oxidative stress challenge was also investigated, in ventricular myocytes. Specifically, the production of ROS was assessed in cells exposed 30 min to either acetaldehyde (ACA) or hydrogen peroxide (H2O2). As predicted from results shown in **Figure 6A**, ROS production was again lower in PFDtreated cells, compared with CN (**Figure 6C**, PFD vs. CN). Additionally, in both CN (**Figure 6C**, white bars) and PFDtreated cells (**Figure 6C**, gray bars) the oxidative insults caused a significant increase in ROS production. The production of ROS, however, proved to be identical in the following three experimental conditions: CN, PFD + ACA, and PDF + H2O<sup>2</sup> (**Figure 6C**). Thus, the results of **Figures 6A,C** demonstrate that PFD attenuates oxidative stress, not only under basal conditions but also in the presence of pro-oxidants.

The regulatory actions on both NCX activity (**Figure 2**) and NO generation (**Figure 6B**) can potentially be explained by parallel changes in steady-state expression levels of related proteins. This point was next investigated, by western-blot analysis. PFD did not alter the expression of NCX (**Figure 7**), suggesting that in PFD-treated cells the disrupted ability of Na<sup>+</sup> to extrude Ca2<sup>+</sup> (**Figure 2**) reflects the lower activity of a fixed number of exchangers.

Pirfenidone also did not alter the expression of eNOS (p > 0.3, **Figure 7**). However, a dual regulation of nNOS was found.

The expression level of this enzyme was increased by 60% in atrial myocytes and decreased by 55% in ventricular myocytes (**Figure 7**). These data suggest that an elevation in expression levels of nNOS (Figure. 7, atrium) is sufficient to increase the synthesis of NO (**Figure 6B**, atrium). Conversely, a reduction in these levels (**Figure 7**, ventricle) is not sufficient for changing the production of NO (**Figure 6B**, ventricle).

# DISCUSSION

Cardiac myocytes are of paramount relevance for blood pumping, and its function is compromised in a number of cardiac conditions. On the other hand, many studies show that PFDtreated animals are less prone to develop cardiac dysfunction (Miric et al., 2001; Mirkovic et al., 2002; Lee et al., 2006; Van Erp et al., 2006; Nguyen et al., 2010; Yamazaki et al., 2012; Wang et al., 2013; Yamagami et al., 2015; Li et al., 2017; Adamo et al., 2018). However, no study has systematically investigated the impact of PFD on the function of cardiac myocytes. Consequently, here we have evaluated the influence of PFD on ICa, Ca2<sup>+</sup> transients, ECC gain, SR Ca2<sup>+</sup> content, contractility, ROS, NO, and expression levels of some related proteins (NCX, eNOS, and nNOS). Our results demonstrate, for the first time, that PFD upregulates not only CaV1.2 but also CICR and ECC. Remarkably, PFD also: enhances the degree and kinetics of contractility, inhibits Ca2<sup>+</sup> extrusion via NCX (and other secondary pathways), and lowers oxidative stress.

Below we discuss molecular mechanisms that may underlie our findings, along with the corresponding implications for the in vivo condition.

#### Molecular Mechanisms of PFD Action

Investigating contractility under physiological-like situations (i.e., in intact myocytes) is essential because the corresponding results are more likely to be significant for in vivo conditions. Unfortunately, however, deciphering the underlying mechanisms

is complicated because they may comprise several interrelated variables and processes —such as resting membrane potential, the fraction of inactivated channels, the affinity of myofilaments for Ca2+, and cytosolic ionic composition (Bers, 2001). On the contrary, interpreting results from voltage-clamp experiments (e.g., under whole-cell patch-clamp) is more straightforward, because many variables either remain constant or are measured (e.g., ICa, Cm, Vm, intracellular Ca2+, recording solutions). Here we have combined these approaches and obtained evidence suggesting that, by enhancing ICa and SR Ca2<sup>+</sup> release, PFD stimulates the contractility of both the sarcomere and cardiomyocytes. We cannot discard, however, a potential contribution of other mechanisms.

Our previous work demonstrated that PDF increases the number of functional CaV1.2 channels (Ramos-Mondragón et al., 2012). This finding, combined with the present observations that the magnitude of Ca2<sup>+</sup> transients is also enhanced at constant SR load (**Figures 1**, **2**), suggests that PFD prompts the recruitment of Ca2<sup>+</sup> release units. Atrial and ventricular myocytes are functional and structurally different (Bootman et al., 2011), and thus it will be of interest to investigate the molecular and ultrastructural basis for the recruitment of otherwise "silent" Ca2<sup>+</sup> release units. This potential mechanism relies on the assumption that, in control cells, Ca2<sup>+</sup> release units are not saturated by cytosolic Ca2<sup>+</sup> (even at voltages where ICa reaches its maximum). Evidence exists indicating that in rat atrial myocytes this happens, even under basal conditions. This is because Ca2<sup>+</sup> transients elicited by ICa cannot recruit RyRs located in central regions of the cell, due to the absence of t-tubules. Besides, the mitochondria act as a buffer of Ca2<sup>+</sup> that complicates free diffusion of the ion, thereby preventing the activation of this "reserve" of RyRs (Mackenzie et al., 2004). The reserve, however, can be recruited under exceptional conditions. For example, during β-adrenergic stimulation and probably also in response to PFD. The following

supports this view **(i)** PFD inhibits the Ca2<sup>+</sup> uptake by Na+ independent slow systems, which likely include mitochondrial transport; and **(ii)** PFD-treated cells must present higher [Ca2+]<sup>i</sup> in microdomains of the cell periphery, due to exacerbated ICa.

A similar model can be proposed for the ventricle. Gómez et al. (1997) reported that hypertrophied ventricular myocytes (derived from hypertensive rats) present a poor ability of ICa to trigger Ca2<sup>+</sup> release, in the face of an intact SR Ca2<sup>+</sup> load. β-adrenergic stimulation reverses this "defective EC coupling," which likely involves a change of the microarchitecture of the dyad (Gómez et al., 1997). In support of this view, it has been reported that the t-tubular system (and other structures) is altered in hypertrophied hearts (Page and McCallister, 1973). Moreover, biophysical evidence exists supporting the notion that in the hypertrophied myocytes the DHPR may be further from the RyR. More specifically, in these cells, the time course of ICa inactivation is slowed (Gómez et al., 1997). In our experiments, CN myocytes also likely develop a certain degree of defective EC coupling in culture, which PFD may have prevented. The following supports this hypothesis. Although rat ventricular myocytes retain ultrastructural properties like those of healthy, intact cardiac tissue, for at least during 3 days of culture, subtle structural changes began to appear earlier. For example, within 24 h, intercalated disks became internalized (Mitcheson et al., 1998). Moreover, we found evidence that PFD promotes accelerated kinetics of ICa inactivation, which represents biophysical evidence of enhanced DHPR/RyR functional interaction (**Supplementary Figure 5**).

Regarding ventricular myocytes, an apparent inconsistency exists between the results of **Figure 1** (whole-cell, patch-clamp) and **Figure 4** (intact cells). In particular, PFD increased the magnitude of the Ca2<sup>+</sup> transient in the former (**Figures 1B,E**) but not in the latter (**Figures 4B,H**). Conceivably, different rates of CaV1.2 activation and Ca2<sup>+</sup> diffusion could have led to obtaining distinct results. These processes are likely more rapid under patch-clamp, because: **(i)** in these conditions, membrane depolarization is more expedite than in an AP, and **(ii)** the recording solution does not contain Ca2+-binding biomolecules (as opposed to the cytosol). Thus, in patch-clamp experiments, PFD probably improved the DHPR/RyR functional interaction more efficiently (see above), because of both enhanced activation kinetics of ICa and optimized diffusion of the ion. These differences could also account for the observation that in intact ventricular myocytes PFD did not alter the rate of Ca2<sup>+</sup> output, but instead increased the latency for reaching the peak of the Ca2<sup>+</sup> transient as well as the corresponding integral (see **Figure 4**, **Supplementary Figure 2** and the corresponding description).

In response to PFD, the diastolic [Ca2+] was decreased in atrial myocytes (**Figures 4A,C**); however, it showed a tendency to be increased in ventricular myocytes (p = 0.08; **Figures 4B,D**). The reasons for these opposite effects are currently unknown. Additional data combined with a precise mathematical model may eventually help to explain how the fluxes of Ca2<sup>+</sup> are being balanced. In this respect, our results suggest that PFD-treated cells efficiently compensate for the higher flux of Ca2<sup>+</sup> into the cytoplasm: **(i)** PFD accelerates Ca2+-dependent inactivation of CaV1.2. As can be seen in **Supplementary Figure 5**, PFD

FIGURE 8 | The amount of Ca2<sup>+</sup> uptake by the SR is higher in PFD-treated cells, compared with controls. (A) This plot was built using average values of both τ (Figure 2) and Ca2<sup>+</sup> transients (Figure 1), according to: A<sup>100</sup> = A [1- (exp – (t100/τ)]; where A is the peak Ca2<sup>+</sup> transient amplitude, t<sup>100</sup> represents 100 ms, and A<sup>100</sup> means the estimated drop in A in 100 ms. (B) Same data as in a (A100) but plotted as a function of peak Ca2<sup>+</sup> transient amplitude (as opposed to Vm). Data from exclusively control cells (closed symbols) were fitted according to a linear equation, and the resulting parameters were used to create a theoretical function (solid line; the correlation coefficient, r, was 0.999). Of note, this function describes data from both control and PFD-treated cells (gray symbols), indicating that PFD does not alter SERCA activity but instead feeds the system with the substrate (Ca2+).

promotes a 1.3-fold increase in the percentage of ICa inactivation. Thus, while the influx of Ca2<sup>+</sup> via CaV1.2 is higher during the first ∼10 ms, a subsequent accelerated kinetics of inactivation imposes a drastic reduction of Ca2<sup>+</sup> influx (from at least 30 to 100 ms, **Supplementary Figure 5**). **(ii)** PFD increases the absolute level of SR Ca2<sup>+</sup> uptake, in proportion to its effects on Ca2<sup>+</sup> transients. The observation that large and small Ca2<sup>+</sup> transients return both to basal values with similar time constants (τ) means that the activity of SERCA is unchanged (**Figures 2A–C**). Nevertheless, it also implies that absolute levels of SR Ca2<sup>+</sup> uptake are necessarily higher in the former. In other words, our conclusion that PFD does not alter τ (**Figures 2A–C**) does not mean that SERCA is unable to detect the twofold change in intracellular Ca2<sup>+</sup> transient amplitude. On the contrary, it can be demonstrated that SERCA detects this change and thereby also accelerates the

SR Ca2<sup>+</sup> uptake. The demonstration consists of estimating the decay of Ca2<sup>+</sup> transients —in absolute 1F/F values—, for a fixed period. For example, the decay in 1F/F per 100 ms is larger in PFD-treated cells than controls (**Figure 8A**, A100). Accordingly, if one displays A<sup>100</sup> values as a function of the peak Ca2<sup>+</sup> transient amplitude, it becomes clear that PFD increases the velocity of Ca2<sup>+</sup> uptake as a consequence of enhancing the magnitude of Ca2<sup>+</sup> transients (**Figure 8B**).

Pirfenidone has been the subject of numerous studies regarding long-term in vivo effects. Its actions can be grouped as anti-inflammatory, antioxidant and antifibrotic. Changes in expression levels of proteins controlling the development of fibrosis (e.g., TGF-β1), inflammation (e.g., interleukins) and oxidative stress (e.g., catalase) are involved. To date, however, the molecular targets responsible for these changes remain unknown (Avila et al., 2014; Lopez-de la Mora et al., 2015; Graziani et al., 2018). Free radicals are the only PFD-binding molecules identified thus far. In particular, PFD scavenges hydroxyl and superoxide radicals (Giri et al., 1999; Misra and Rabideau, 2000; Mitani et al., 2008). Nevertheless, it is unknown whether this ROS scavenging activity mediates chronic effects.

The present study was designed to characterize long-lasting and persistent modifications of cardiomyocyte function (as opposed to rapid and readily reversible). Hence, cells were exposed to PFD for a minimum of 4 h (at 37◦C, in culture medium) and experiments were performed 15–50 min after finishing the treatment (i.e., using extracellular solutions devoid of PFD). Therefore, the changes we have observed persist long after PFD removal. Moreover, at least for the case of ROS and NO production, it took a minimum of 8 h of PFD treatment to detect a significant effect (**Supplementary Figure 4**).

Thus, most likely in our experiments PFD initially regulated protein expression (e.g., oxidative stress-related proteins and nNOS), and this, in turn, influenced free radicals production and cell function.

Nevertheless, the precise sequence of PFD effects on cardiomyocytes has yet to be firmly stablished, and it is likely that some of them be interrelated. For example, in atrial myocytes, the inhibition of ROS production (**Figure 6A** and **Supplementary Figure 4**) could have increased the expression of nNOS (**Figure 7**), and thereby PFD also induced higher levels of NO synthesis (**Figure 6B**). The time-course of PFD effects indirectly support these thoughts (**Supplementary Figure 4**). Specifically, ROS production starts to be regulated after 8 h of PFD-treatment (**Supplementary Figure 4B**; atrium), whereas the stimulation of NO generation requires 24 h to become evident (**Supplementary Figure 4D**; atrium). Moreover, these findings are also in line with results of Kar and collaborators, showing that prolonged exposure of HL-1 atrial myocytes to H2O<sup>2</sup> results in low expression levels of nNOS (Kar et al., 2015). Thus, in atrial myocytes, both the expression of nNOS and NO synthesis seem to be under control of chronic ROS production.

Additionally, it is well-known that ROS stimulates NCX activity (e.g., Goldhaber, 1996; Santacruz-Toloza et al., 2000; Kuster et al., 2010). Remarkably, this effect has been associated with redox modification of the exchanger, which likely interferes with Na+-dependent inactivation (Santacruz-Toloza et al., 2000). Thus, in our present control conditions, NCX activity is likely sustained by ROS, and possible removal of redox modification might represent the basis for PFD inhibiting the exchanger. On the other hand, ROS do not appear to be involved in the regulation by PFD of CaV1.2, because in atrial myocytes several classical antioxidants—such as DMSO, glutathione, and N-acetylcysteine (NAC)— do not reproduce the PFD effect on ICa (Ramos-Mondragón et al., 2012). Thus, in addition to ROS (Giri et al., 1999; Misra and Rabideau, 2000; Mitani et al., 2008), PFD also likely acts through other molecular targets (for review see Avila et al., 2014).

In our experiments, PFD inhibited the NCX function, which was estimated 25 s after caffeine stimulation. Thus, it remains unknown whether this effect has an impact on triggered electrical activity. Investigating the NCX by other methods would not only help to solve this question but also to reinforce our conclusions.

We previously speculated that PFD might decrease the production of NO in cardiomyocytes, based (at least partially) on evidence derived from hepatocytes (Nakanishi et al., 2004; Tsuchiya et al., 2004). However, the present study indicates that, in ventricular myocytes, PFD does not alter the NO production rate (despite causing a significant reduction in expression levels of nNOS). Furthermore, in atrial myocytes, the compound increases the production of NO (which is likely associated with the observed increase in nNOS amount). The NO signaling is both relevant and complex (for review, see Simon et al., 2014). This complexity originates, at least partially, in nNOS and eNOS presenting distinct subcellular distribution. Accordingly, it has been proposed that the former primarily modulate RyR and SERCA because of its closer proximity to the SR. Conversely, eNOS is thought to predominantly regulate Ca2<sup>+</sup> handling proteins near the sarcolemma, such as NCX, sarcolemma Ca2<sup>+</sup> pump and CaV1.2 (Jian et al., 2014). Additionally, a complex interplay exists between nNOS and ROS. As explained above, nNOS expression and NO levels, both depend on chronic ROS production, whereas nNOS activates, in turn, the ROS-NADPH oxidase 2 (Nox2) signaling (Girouard et al., 2009). Moreover, signaling downstream of NO is divided into two main pathways. One of them involves direct S-nitrosylation (of target proteins), whereas another is characterized by modulation of guanylate cyclase (GC) which in turn regulates cGMP- and cAMP-dependent protein kinases (PKG and PKA). Thus, a systematic study concerning nNOS regulation by PFD (**Figure 7A**) could begin by using selective modulators of NOS isoforms and protein kinases, combined with measurements of not only global levels of Ca2<sup>+</sup> and contractility, but also local production/changes of ROS, NO, and Ca2+. With regard to PKA, our preliminary results suggest that this kinase might not be involved in up-regulation of contractility (**Supplementary Figure 3**).

#### Potential Pathophysiological Relevance

Regardless of the precise molecular mechanisms involved, the present results are likely to have pathophysiological implications. In vivo, PFD decreases the susceptibility to develop arrhythmias

(AF and ventricular tachycardia), and these effects are thought to depend on a concomitant inhibition of fibrosis (Lee et al., 2006; Nguyen et al., 2010). Nevertheless, it is possible that functional changes in cardiomyocytes also contribute. The enhanced activity of CaV1.2 could increase the effective refractory period, and thereby decrease propensity toward electrical reentry (Ramos-Mondragón et al., 2012; Avila et al., 2014). Besides, the lower activity of the NCX may be decreasing the likelihood of delayed afterdepolarizations (DADs; Voigt et al., 2012), and hence PFD could also inhibit electrical instability and arrhythmia.

Stimulation of CICR, ECC, and contractility, on the other hand, may be of critical relevance to explaining why PFD improves: (**i)** ventricular developed pressure in mdx mice (both positive and negative dP/dt; Van Erp et al., 2006) and (**ii)** left ventricular ejection fraction in rats subjected to myocardial infarction (Nguyen et al., 2010).

The inhibition of ROS production by PFD could be relevant in more general terms, because these free radicals are involved in several cardiac conditions, including arrhythmias, myocardial infarct, hypertrophy, alcoholic cardiomyopathy, and heart failure (for reviews see Zhang et al., 2004; Korantzopoulos et al., 2007; Köhler et al., 2014). Thus, shielding myocytes from oxidant damage is likely essential for PFD in preventing cardiac dysfunction.

#### CONCLUSION

Pirfenidone has been long acknowledged to be of therapeutic potential for a number of heart symptoms. In vivo, the

#### REFERENCES


underlying mechanisms are likely numerous and interrelated. Here, its effects on isolated myocytes were investigated, limiting influences from additional tissues and cells. Our results show for the first time that PFD stimulates the function of cardiac myocytes. Other in vivo effects may be regarded as "auxiliary" because the compound acts on essential components of the cardiac muscle-pump. In particular, it stimulates CICR, ECC, and contractility. Moreover, it also decreases ROS production, suggesting that its effects on myocytes are cytoprotective.

# AUTHOR CONTRIBUTIONS

DO-G performed experiments and analyzed data. AM-V and GA performed experiments, analyzed data, and wrote the paper.

# FUNDING

This work was financed by Conacyt (Grant No. #253009).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2018.01801/full#supplementary-material


exchanger contributes to sarcoplasmic reticulum Ca2+ depletion in cardiac myocytes. Free Radic. Biol. Med. 48, 1182–1187. doi: 10.1016/j.freeradbiomed. 2010.01.038


Ca2+ channels and its arrhythmogenic significance. Pflugers Arch. 468, 1823– 1835. doi: 10.1007/s00424-016-1876-8


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Monsalvo-Villegas, Osornio-Garduño and Avila. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Pharmacological Modulation of Mitochondrial Ca2<sup>+</sup> Content Regulates Sarcoplasmic Reticulum Ca2<sup>+</sup> Release via Oxidation of the Ryanodine Receptor by Mitochondria-Derived Reactive Oxygen Species

#### Edited by:

Angélica Rueda, Centro de Investigación y de Estudios Avanzados (CINVESTAV), Mexico

#### Reviewed by:

Diego De Stefani, Università degli Studi di Padova, Italy José Javier López Barba, Universidad de Extremadura, Spain

#### \*Correspondence:

Dmitry Terentyev dmitry\_terentyev@brown.edu

#### Specialty section:

This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology

Received: 21 September 2018 Accepted: 06 December 2018 Published: 21 December 2018

#### Citation:

Hamilton S, Terentyeva R, Kim TY, Bronk P, Clements RT, O-Uchi J, Csordás G, Choi B-R and Terentyev D (2018) Pharmacological Modulation of Mitochondrial Ca2<sup>+</sup> Content Regulates Sarcoplasmic Reticulum Ca2<sup>+</sup> Release via Oxidation of the Ryanodine Receptor by Mitochondria-Derived Reactive Oxygen Species. Front. Physiol. 9:1831. doi: 10.3389/fphys.2018.01831 Shanna Hamilton<sup>1</sup> , Radmila Terentyeva<sup>1</sup> , Tae Yun Kim<sup>1</sup> , Peter Bronk<sup>1</sup> , Richard T. Clements2,3, Jin O-Uchi<sup>4</sup> , György Csordás<sup>5</sup> , Bum-Rak Choi<sup>1</sup> and Dmitry Terentyev<sup>1</sup> \*

<sup>1</sup> Department of Medicine, The Warren Alpert Medical School of Brown University, Rhode Island Hospital, Cardiovascular Research Center, Providence, RI, United States, <sup>2</sup> Department of Surgery, The Warren Alpert Medical School of Brown University, Rhode Island Hospital, Cardiovascular Research Center, Providence, RI, United States, <sup>3</sup> Vascular Research Laboratory, Providence Veterans Affairs Medical Center, Providence, RI, United States, <sup>4</sup> Lillehei Heart Institute University of Minnesota, Cancer and Cardiovascular Research Building, Minneapolis, MN, United States, <sup>5</sup> Department of Pathology, Anatomy and Cell Biology, Thomas Jefferson University, Philadelphia, PA, United States

In a physiological setting, mitochondria increase oxidative phosphorylation during periods of stress to meet increased metabolic demand. This in part is mediated via enhanced mitochondrial Ca2<sup>+</sup> uptake, an important regulator of cellular ATP homeostasis. In a pathophysiological setting pharmacological modulation of mitochondrial Ca2<sup>+</sup> uptake or retention has been suggested as a therapeutic strategy to improve metabolic homeostasis or attenuate Ca2+-dependent arrhythmias in cardiac disease states. To explore the consequences of mitochondrial Ca2<sup>+</sup> accumulation, we tested the effects of kaempferol, an activator of mitochondrial Ca2<sup>+</sup> uniporter (MCU), CGP-37157, an inhibitor of mitochondrial Na+/Ca2<sup>+</sup> exchanger, and MCU inhibitor Ru360 in rat ventricular myocytes (VMs) from control rats and rats with hypertrophy induced by thoracic aortic banding (TAB). In periodically paced VMs under β-adrenergic stimulation, treatment with kaempferol (10 µmol/L) or CGP-37157 (1 µmol/L) enhanced mitochondrial Ca2<sup>+</sup> accumulation monitored by mitochondrialtargeted Ca2<sup>+</sup> biosensor mtRCamp1h. Experiments with mitochondrial membrane potential-sensitive dye TMRM revealed this was accompanied by depolarization of the mitochondrial matrix. Using redox-sensitive OMM-HyPer and ERroGFP\_iE biosensors, we found treatment with kaempferol or CGP-37157 increased the levels of reactive oxygen species (ROS) in mitochondria and the sarcoplasmic reticulum (SR), respectively. Confocal Ca2<sup>+</sup> imaging showed that accelerated Ca2<sup>+</sup> accumulation reduced Ca2<sup>+</sup>

**43**

transient amplitude and promoted generation of spontaneous Ca2<sup>+</sup> waves in VMs paced under ISO, suggestive of abnormally high activity of the SR Ca2<sup>+</sup> release channel ryanodine receptor (RyR). Western blot analyses showed increased RyR oxidation after treatment with kaempferol or CGP-37157 vs. controls. Furthermore, in freshly isolated TAB VMs, confocal Ca2<sup>+</sup> imaging demonstrated that enhancement of mitochondrial Ca2<sup>+</sup> accumulation further perturbed global Ca2<sup>+</sup> handling, increasing the number of cells exhibiting spontaneous Ca2<sup>+</sup> waves, shortening RyR refractoriness and decreasing SR Ca2<sup>+</sup> content. In ex vivo optically mapped TAB hearts, kaempferol exacerbated proarrhythmic phenotype. On the contrary, incubation of cells with MCU inhibitor Ru360 (2 µmol/L, 30 min) normalized RyR oxidation state, improved intracellular Ca2<sup>+</sup> homeostasis and reduced triggered activity in ex vivo TAB hearts. These findings suggest facilitation of mitochondrial Ca2<sup>+</sup> uptake in cardiac disease can exacerbate proarrhythmic disturbances in Ca2<sup>+</sup> homeostasis via ROS and enhanced activity of oxidized RyRs, while strategies to reduce mitochondrial Ca2<sup>+</sup> accumulation can be protective.

Keywords: mitochondria, reactive oxygen species, ryanodine receptor, hypertrophy, ventricular arrhythmia, Ca2+-induced Ca2<sup>+</sup> release

# INTRODUCTION

Sudden cardiac death remains the leading global cause of mortality, and over half of patients with heart failure (HF) die suddenly due to the development of ventricular arrhythmia (Benjamin et al., 2018). Arrhythmogenesis in the failing heart is often linked to enhanced Ca2+-dependent triggered activity, in the form of early and delayed afterdepolarizations (Landstrom et al., 2017). These abnormal electrical activities arise in part as a consequence of untimely and dysregulated Ca2<sup>+</sup> release from the sarcoplasmic reticulum (SR), through SR Ca2<sup>+</sup> release channel, the ryanodine receptor (RyR). Abnormal activity of RyR leads to increased Ca2<sup>+</sup> leak and promotes the generation of spontaneous Ca2<sup>+</sup> waves (SCWs), that can subsequently propagate to trigger organ-wide arrhythmia (Bers, 2002).

Mitochondria play an essential role in cardiac Ca2<sup>+</sup> homeostasis in physiological conditions (Kwong et al., 2015; Luongo et al., 2017). Excitation-contraction coupling consumes large amounts of ATP and mitochondria increase oxidative phosphorylation to meet increased metabolic demand. Influx of Ca2<sup>+</sup> into the mitochondria is critical for the availability of ATP as major enzymes in the tricarboxylic acid cycle are activated by Ca2+. Mitochondria are in close spatial proximity to the SR (Dorn and Scorrano, 2010; Eisner et al., 2013; Lu et al., 2013; Seidlmayer et al., 2016; Lopez-Crisosto et al., 2017; Csordás et al., 2018), and it is well established that during higher workload, there is an elevation of cystolic Ca2<sup>+</sup> concentration in ventricular myocytes (VMs) that transpires to a small and slow rise in mitochondrial Ca2<sup>+</sup> concentration ([Ca2+]m), leading to enhanced energy production (Brandes and Bers, 1997; Luongo et al., 2015). Influx of Ca2<sup>+</sup> through the mitochondrial Ca2<sup>+</sup> uniporter (MCU) channel complex is driven largely by the negative membrane potential across the inner mitochondrial membrane (Kirichok et al., 2004; Baughman et al., 2011; De Stefani et al., 2011). Conversely, mitochondrial efflux mainly occurs via the mitochondrial Na/Ca2+/Li+-exchanger (NCLX) (Palty et al., 2010; Boyman et al., 2013; Luongo et al., 2017).

Mitochondria are a major source of reactive oxygen species (ROS) in the myocyte, and while an increase in oxidative stress is a prerequisite for many cellular stress responses, excessive ROS production in cardiac disease contributes to ventricular arrhythmogenesis by altering the function of multiple ion channels and transporters (Zima and Blatter, 2006; Niggli et al., 2013; Wagner et al., 2013). RyRs are highly sensitive to ROS, and contain multiple redox-sensitive cysteine residues (Zima and Blatter, 2006). Cysteine thiol oxidation of RyR increases channel activity, and many groups including ours have previously established that increased RyR oxidation in VMs from diseased hearts promotes proarrhythmic spontaneous SR Ca2<sup>+</sup> release in the form of propagating Ca2<sup>+</sup> waves that underlie increased triggered activity (Terentyev et al., 2008; Belevych et al., 2009; Cooper et al., 2013; Kyrychenko et al., 2013; Bovo et al., 2018). Scavenging of mitochondrial ROS was shown to improve Ca2<sup>+</sup> homeostasis and attenuate arrhythmic potential in multiple models of cardiac disease including HF, hypertrophy, diabetic cardiomyopathy or aging (Mochizuki et al., 2007; Terentyev et al., 2008; Belevych et al., 2012; Cooper et al., 2013; Luo et al., 2013; Joseph et al., 2016; Kim et al., 2017).

Given the contribution of mitochondrial dysfunction to multiple cardiac disease states, maintaining mitochondrial Ca2<sup>+</sup> homeostasis remains an attractive therapeutic target (Dietl and Maack, 2017). In conditions with defective intracellular Ca2<sup>+</sup> and Na2<sup>+</sup> homeostasis such as in models of HF, increasing [Ca2+]<sup>m</sup> above a specific threshold was suggested to improve metabolism and substrate utilization, as well as reduce oxidative stress and ROS overload in the myocyte (Liu and O'Rourke, 2008; Kolhaas et al., 2010; Liu et al., 2014). More recently, Schweitzer

et al. (2017) suggested that pharmacological enhancement of [Ca2+]<sup>m</sup> suppressed arrhythmia in a model of catecholaminergic polymorphic ventricular tachycardia (CPVT), a condition characterized by mutations in the RyR macromolecular complex that renders channels hyperactive. Given mitochondria are in close proximity to SR Ca2<sup>+</sup> release sites, it has been proposed that increasing mitochondrial Ca2<sup>+</sup> uptake may improve buffering capacity (Seguchi et al., 2005; Drago et al., 2012; Zhao et al., 2013), thereby limiting local Ca2<sup>+</sup> release events, Ca2<sup>+</sup> sparks, which would result in a decrease in generation and propagation velocity of proarrhythmic SCWs. Conversely, a reduction of mitochondrial Ca2<sup>+</sup> uptake may serve as an anti-arrhythmic strategy. In models of HF and ischemia-reperfusion, pathological mitochondrial dysfunction and mitochondria Ca2<sup>+</sup> overload contribute to oxidative stress and cell death (Santulli et al., 2015). Pharmacological inhibition (García-Rivas Gde et al., 2006; Xie et al., 2018) or genetic ablation (Kwong et al., 2015; Luongo et al., 2015) of MCU, as well as conditional NCLX overexpression (Luongo et al., 2017) has been shown to protect against ischemiainduced myocyte injury, the development Ca2+-dependent arrhythmia and the progression of HF.

In the present study, we aimed to determine the effects of pharmacological facilitation of [Ca2+]<sup>m</sup> accumulation and inhibition of mitochondrial Ca2<sup>+</sup> uptake on intracellular Ca2<sup>+</sup> homeostasis and arrhythmic potential using rat model of cardiac hypertrophy induced by thoracic aortic banding (TAB). To achieve this goal, we utilized whole heart optical mapping, genetically encoded ROS and mitochondrial Ca2<sup>+</sup> biosensors, confocal microscopy and biochemistry to dissect the influence of MCU enhancer kaempferol and NCLX inhibitor CGP-37157 on intracellular Ca2<sup>+</sup> cycling, in both healthy and hypertrophic VMs. Our results suggest that enhancement of mitochondrial Ca2<sup>+</sup> accumulation in either setting elevates mitochondrial ROS emission, increasing oxidation of RyR and aberrant spontaneous Ca2<sup>+</sup> release. Attenuating mitochondrial Ca2<sup>+</sup> uptake serves as an anti-arrhythmic treatment in hypertrophic hearts, whereby triggered activity was reduced by pharmacological inhibition of MCU with Ru360.

#### MATERIALS AND METHODS

#### Ethics Statement

Procedures involving animals were approved by The Rhode Island Hospital Institutional Animal Care and Use Committee and followed the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85-23, revised 2011).

#### Generation of Adenoviral Constructs

The mitochondrial targeting sequence cytochrome C oxidase subunit IV was fused as the N-terminal of the coding sequence of plasmid RCamp1h (Akerboom et al., 2013) to create a probe to monitor intra-mitochondrial Ca2+. pC1-HyPer-3 was a gift from Vsevolod Belousov (Addgene plasmid # 42131). The mitochondrial localization sequence of mAKAP1 followed by a linker was fused as the N-terminus of the coding sequence of pC1-HyPer-3 (Burns-Hamuro et al., 2003; DiPilato et al., 2004; Bilan et al., 2013). This enables anchoring of the probe to the outer mitochondrial membrane (OMM) to measure H2O<sup>2</sup> at the mitochondrial surface, and the subsequent viral construct is thus referred to as OMM-HyPer. ERroGFP\_iE\_pCDNA3 was a gift from David Ron (Addgene plasmid # 47954). The ERroGFP\_iE probe is targeted to the endoplasmic reticulum (ER) by the cleavable signal peptide and C-terminal KDEL ER retrieval signal (Avezov et al., 2013). Adenovirus carrying plasmid constructs were generated utilizing the ViraPower Gateway expression system (Thermo Fisher Scientific, Waltham, MA, United States). Briefly, coding regions were cloned into the pENTRTM 1A entry vector, and recombined into pAd/CMV/V5- DESTTM destination vector by LR recombinase reaction. Once sequence-verified, destination vector plasmids were digested with restriction enzyme PacI and transfected into HEK293A cells using LipofectamineTM 2000 (Thermo Fisher Scientific). Adenoviral stock titer was determined using the Adeno-X qPCR Titration Kit (Takara Bio USA, Inc., Mountain View, CA, United States).

#### Myocyte Isolation and Primary Culture

Myocytes were isolated from male 9- to 12-week-old Sprague-Dawley rats (controls) from Harlan Laboratories (Indianapolis, IN, United States). Male Sprague-Dawley rats with TAB surgery were purchased from Charles River Laboratories (Wilmington, MA, United States). Animals were shipped 5–7 days after surgery and acclimatized for 3–4 weeks in the Rhode Island Hospital animal facility. Experiments were performed 4–5 weeks after aortic banding procedure.

Bilateral thoracotomy was performed on euthanized rats and the heart plunged into ice cold Tyrode's solution. The hearts were mounted on a Langendorff apparatus and retrogradely perfused with Tyrode solution (Terentyev et al., 2009) containing collagenase II (Worthington Biochemical Corp., Lakewood, NJ, United States) at 37◦C for 16–17 min. Ventricles were minced and placed in a 37◦C water bath shaker in collagenase solution. Isolated VMs were plated onto laminin-coated glass coverslips in 24-well plates.

For experiments with cultured control and TAB rats VMs, myocytes were cultured in serum-free medium 199 (Thermo Fisher Scientific), supplemented with 25 mmol/L NaHCO3, 10 mmol/L HEPES, 5 mmol/L creatine, 5 mmol/L taurine, 10 µ/mL penicillin, 10 µg/mL streptomycin and 10 µg/mL gentamycin (pH 7.3). Unattached cells were removed after 1 h and remaining VMs were cultured for 48 h. Cultured VMs were infected with adenoviruses at multiplicity of infection (MOI) of 10 for all described constructs. Myocytes were cultured at 37◦C in 95% air, 5% CO<sup>2</sup> for 36–48 h before analysis.

# Pharmacological Modifiers of Mitochondrial Ca2<sup>+</sup> Uptake, Mitochondrial ROS, and RyR Activity

Kaempferol directly activates MCU (Montero et al., 2004; Vay et al., 2007), and was obtained from Millipore Sigma (Burlington, MA, United States), used at 10 µmol/L. CGP-37571 inhibits

NCLX (Liu and O'Rourke, 2008; Kolhaas et al., 2010; Liu et al., 2010), and was obtained from Millipore Sigma, used at 1 µmol/L. SB 202190 is an inhibitor of p38 mitogen-activated protein (MAP) kinase, has also been shown to activate MCU (Montero et al., 2004), and was obtained from Millipore Sigma, used at 30 µmol/L. Ru360 specifically inhibits mitochondrial Ca2<sup>+</sup> uptake through MCU (Matlib et al., 1998; García-Rivas Gde et al., 2006), and was obtained from Millipore Sigma, used at 2 µmol/L. MitoTEMPO, a specific scavenger of mitochondrial superoxide was obtained from Millipore Sigma, used at 20 µmol/L. Dantrolene, an RyR antagonist (Kobayashi et al., 2009; Maxwell et al., 2012) was obtained from Millipore Sigma and used at 2 µmol/L.

#### Confocal Imaging

Confocal imaging was performed using a Leica SP5 II confocal microscope equipped with 63 × 1.4 numerical aperture oil objective in linescan and x–y mode. All confocal imaging experiments were performed under β-adrenergic stimulation with 50 nmol/L isoproterenol (ISO, Millipore Sigma). Control VMs were paced via field stimulation at 2 Hz, while TAB VMs were paced at 0.5 Hz using extracellular platinum electrodes. Myocytes were studied in Tyrode's solution (Terentyev et al., 2009). Confocal imaging data were analyzed using Leica Software, Origin 8.0 (OriginLab, Northampton, MA, United States) and ImageJ (National Institutes of Health, Bethesda, MA, United States).

Intact VMs were loaded with Rhod-2 AM (Thermo Fisher Scientific) at room temperature for 12 min, followed by a 10 min wash. Rhod-2 was excited using 543 nm line of HeNe laser and fluorescence emission was collected at 560–660 nm wavelengths in linescan mode at 200 Hz sampling rate. Calcium transients were recorded at room temperature. To test for the propensity of triggered activity, VMs were paced for 20 s and latency between the last pacing stimulus and the subsequent SCW was calculated. To assess SR Ca2<sup>+</sup> load, 10 mmol/L caffeine was applied at the end of experiments. The data is presented as 1F/F0, where F<sup>0</sup> is basal fluorescence and 1F = F – F0.

Biosensor mtRCamp1h was excited using 543 nm line of HeNe laser and fluorescence emission was collected at 560–660 nm wavelengths. For permeabilized VM experiments, myocytes were saponin-permeabilized (0.001%) and equilibrated with a solution containing thapsigargin (10 µmol/L), cytochalasin D (10 µmol/L), FCCP (20 µmol/L), and ionomycin (5 µmol/L). Tyrode's solution containing Ca2<sup>+</sup> buffer EGTA (2 mmol/L) was applied to obtain minimum mtRCamp1h fluorescence. Maximum fluorescence was achieved by application of Ca2<sup>+</sup> (100 µmol/L). The data is presented as as 1F/F0, where F<sup>0</sup> is basal fluorescence and 1F = F − F0. Biosensors ERroGFP\_iE and OMM-HyPer were excited using 488 nm line of Argon laser and fluorescence emission was collected at 500–550 nm wavelengths, measured in x–y mode. Maximum fluorescence (Fmax) was obtained by application 200 µmol/L DTDP and minimum fluorescence was obtained by application ROS scavenger DTT (5 mmol/L). Data is presented as a percentage of 1F/1Fmax where 1F = F −Fmin, and 1Fmax = Fmax−Fmin. Mitochondrial membrane potential was monitored with the

voltage-sensitive fluorescent indicator, tetramethylrhodamine, methyl ester (TMRM; Thermo Fisher Scientific). Isolated VMs were loaded with 20 µmol/L for 1 min and washed thoroughly prior to imaging. TMRM was excited using 543 nm line of HeNe laser and fluorescence emission was collected at 560– 660 nm wavelengths, measured in x–y mode. Fluorescence of TMRM was normalized to the minimum fluorescence signal obtained by application of mitochondrial uncoupler, carbonyl p-(trifluoromethoxy) phenylhydrazone (FCCP, 50 µmol/L). For experiments with mitoTEMPO, isolated VMs were pretreated with mitoTEMPO (20 µmol/L, 30 min), before loading with TMRM as described.

The emission of ROS was measured in isolated VMs in Tyrode solution using MitoSOX Red mitochondrial superoxide indicator (Thermo Fisher Scientific; 20 µmol/L, 30 min loading). The indicator was excited with 514 nm line of an argon laser and emission was collected at 560–660 nm, measured in x–y mode. Fluorescence of MitoSOX was normalized to the maximum fluorescence signal obtained by application of peroxide H2O<sup>2</sup> (10 mmol/L).

# Oxidation of RyR and Western Blotting

Freshly isolated control or TAB rat VMs were treated with ISO (50 nmol/L, 5 min total), kaempferol (10 µmol/L, 5 min total) or CGP (1 µmol/L, 10 min total) and paced for 1 min at 2 Hz at room temperature before immediate lysis in lysis buffer from Cell Signaling (Danvers, MA, United States, Cat#9803S), supplemented with phosphatase (Calbiochem, San Diego, CA, United States, Cat#524625) and protease inhibitor cocktails (Millipore Sigma, Cat#P8340) as described previously (Terentyev et al., 2014). For co-immunoprecipitation, Pierce Coimmunoprecipitation Kit (Thermo Fisher Scientific, Cat#26149) was used. Lysate (500 µl) was pre-cleared with Control Agarose Resin for 30 min at 4◦C, centrifuged at 1,000 × g for 1 min. Flowthrough was incubated with antibody-coupled resin (anti-RyR2, Thermo Fisher Scientific, Cat#MA3-916 and negative control antibody, normal mouse IgG, Santa Cruz Biotechnology, Cat#sc-2025) for 2 h at 4◦C. Columns were washed three times. Protein complexes were eluted with elution buffer provided in the kit. To determine oxidation of RyR, the Oxidized Protein Western Blot Kit (Abcam, Cambridge, MA, United States, Cat#ab178020) was used, whereby carbonyl groups of immunoprecipitated RyR2 were derivatized to 2,4 dinitrophenylhydrazone (DNP) by reaction with 2,4 dinitrophenylhydrazine. For control we used Derivatization Control Solution, provided in the kit. The DNP-RyR2 protein samples were separated on 4–20% Mini-PROTEAN TGX gels (Bio-Rad Laboratories, Hercules, CA, United States, Cat#456-1094) and DNP-associated signal assessed by the kit-provided anti-DNP primary antibody and anti-RyR2 (Thermo Fisher Scientific, Cat#MA3-916), followed by HRP-conjugated secondary antibody and anti-mouse lgG(H+L) HRP secondary antibody (Promega, Madison, WI, United States, Cat#W4021). Abcam antibodies (Cat#ab57602 and Cat#ab101055) were used to assess expression levels of mitofusin 1 and mitofusin 2. Anti-glyceraldehyde 3-phosphate dehydrogenase (GAPDH) antibodies were used for loading control (Abcam, Cat#ab8245). Blots were developed with ECL (Bio-Rad Laboratories) and quantified using ImageJ and Origin 8 software.

# Ex vivo Optical Mapping

fphys-09-01831 December 19, 2018 Time: 18:27 # 5

Beating hearts were harvested from anesthetized TAB rats via thoracotomy and were retrogradely perfused through the aorta in a Langendorff perfusion system (Radnoti Glass Technology, Monrovia, CA, United States) with (in mmol/L): 130 NaCl, 24 NaHCO3, 1.0 MgCl2, 5.0 KCl, 1.2 NaH2PO4, 5 dextrose, and 1 CaCl2, at pH 7.4, gassed with 95% O<sup>2</sup> and 5% CO2. Constant flow perfusion was set to 10 mL/min with a peristaltic pump. Hearts were placed in a water-heated chamber to maintain temperature at 37 ± 0.2◦C, and 5 µmol/L blebbistatin was added to perfusate to reduce movement artifact. Hearts were loaded with Ca2<sup>+</sup> indicator Rhod-2 AM (Thermo Fisher Scientific), using 25 µL of stock solution (1 mg/mL of DMSO) delivered through a bubble trap, above the aortic cannula. The ECGs were continuously monitored with a Powerlab system (AD Instrument, Colorado Springs, CO, United States). The optical apparatus has been described previously (Kim et al., 2015). Fluorescence images of Rhod-2 signal were recorded from the anterior surface of the heart using a CMOS camera (100 × 100 pixels, 2000 frames/sec, 1.5 cm × 1.5 cm field of view, Ultima-L, SciMedia, Japan). Drugs (kaempferol and Ru360) were perfused for 20–30 min and ISO (50 nmol/L) was added to investigate the effect of drugs on VT/VF induction in TAB hearts. The fluorescence (F) from Rhod-2 was normalized with 1F/F. Hearts were stimulated with 150 ms cycle length followed by premature stimulation of 10 beats of S2 until refractoriness or VT induction. Propagation and duration of Ca2<sup>+</sup> transients were mapped using (dF/dt)max and 75% recovery, respectively, as previously described (Kim et al., 2015).

#### Statistics

Statistical analysis of Ca2<sup>+</sup> imaging and biochemical data was performed using Origin 8 (OriginLab). Data are presented as mean ± standard error (SEM) for single cell and ± standard deviation (SD) for intact heart optical mapping. Uppercase n (N) = number of animals, lowercase n = number of VMs. Statistical significance between groups were performed using Student's t-test (paired and unpaired), Fisher's exact test and oneway ANOVA with Bonferroni post hoctest where appropriate. For all analyses, a p-value of less than 0.05 was considered significant.

#### RESULTS

# Mitochondrial Ca2<sup>+</sup> Accumulation Exacerbates the Proarrhythmic Phenotype of TAB Hearts ex vivo

We have previously reported that the rat model hypertrophy induced by TAB is highly arrhythmogenic, with incidences of non-sustained ventricular tachycardia and fibrillation (VT/VF) occurring in 100% of TAB hearts exposed to 50 nmol/L ISO (Kim et al., 2017). We investigated whether the MCU activator, kaempferol (Montero et al., 2004; Vay et al., 2007), increases mitochondrial Ca2<sup>+</sup> accumulation to reduce cytosolic Ca2<sup>+</sup> and suppress ventricular arrhythmias in this model. Hearts were isolated and perfused retrogradely as described in "Materials and Methods" and echocardiograms (ECGs) were monitored to investigate arrhythmogenesis under kaempferol and Ru360. ISO induced frequent premature ventricular contractions (PVCs) leading to VFs in TAB rat hearts (**Figure 1A** ECG traces under **TAB**+**ISO**). Pretreatment with MCU activator, kaempferol (10 µmol/L), did not prevent PVCs and VF induction. ECG traces in **Figure 1A** (**TAB**+**KAEM**+**ISO** panel) shows frequent PVCs (red stars) that led to VFs (blue bar) under kaempferol. However, MCU blocker Ru360 (2 µmol/L) suppressed number of PVCs and prevented VFs in TAB rat hearts (**Figure 1A**, **TAB**+**Ru360**+**ISO** panel).

We investigated the effect of kaempferol on Ca2<sup>+</sup> handling and arrhythmogenesis using optical mapping. Kaempferol reduces the Ca2<sup>+</sup> transient amplitude by 53% (**Figure 1B**, <sup>∗</sup>p = 0.002) but despite smaller Ca2<sup>+</sup> transients, kaempferol induced transient VTs in two of four TAB hearts, even without ISO. Addition of ISO caused frequent PVCs (**Figure 1D**) and long-lasting VFs (**Figure 1E**) in three of four hearts. Activation maps of PVCs (**Figure 1D**) and VFs (**Figure 1E**) suggest that focal activity play a major role in the initiation and maintenance of VFs. The frequencies of VF were significantly higher with kaempferol (17.6 ± 2.1 Hz, n = 3, in kaempferol vs. 14 ± 1.8 Hz, n = 7, control TAB with ISO, <sup>∗</sup>p = 0.042, **Figure 1C**). In contrast, MCU inhibitor, Ru360 (Matlib et al., 1998; García-Rivas Gde et al., 2006), suppressed spontaneous VT/VFs in the presence of ISO in four of four hearts. S1S2 pacing induced reentry and VFs in two of four hearts in the presence of Ru360 (**Figure 1F**), suggesting that conduction block, not focal activity, underlies VF induction during S1S2 pacing in Ru360 group. These intact heart optical mapping data suggest that enhancement of mitochondrial Ca2<sup>+</sup> accumulation may exacerbate ventricular arrhythmias in TAB rat hearts through increasing focal activity.

#### Pharmacological Enhancers of Mitochondrial Ca2<sup>+</sup> Accumulation Modulate Time Course, Not Amplitude During Periodic Pacing

To gain mechanistic insights as to how increasing mitochondrial Ca2<sup>+</sup> accumulation affects global Ca2<sup>+</sup> handling in VMs, we used a genetically encoded fluorescent Ca2<sup>+</sup> biosensor, mtRCamp1h. The RCamp1h indicator, with a Kd ∼ 1.3 µM (Akerboom et al., 2013), was fused with an N-terminal cytochrome C oxidase subunit IV tag for targeting to the mitochondrial matrix. We generated adenovirus encoding the sensor, and VMs isolated from control rat hearts were infected with adenovirus at a MOI of 10 and cultured for 48 h prior to experimentation. Rat myocytes are thought to preserve electrical properties and structure including T-tubule organization for the first 48 h of culture (Banyasz et al., 2008).

The correct cellular localization of mtRCamp1h was confirmed with co-expression of mitochondrial matrix-targeted GFP, as shown in **Figure 2A**. As shown in **Figure 2B**, basal mtRCamp1h fluorescence appears well within the dynamic range

PVCs before VF induction and focal activity in the presence of KAEM. Activation maps are shown in the right panel. (F) Conduction block and reentry formation by S1S2 pacing protocol in the presence of Ru360 (N = 4).

of the probe when adenovirally expressed in VMs. Furthermore, as seen in **Figure 2B**, we can indeed observe measurable changes in mitochondrial Ca2<sup>+</sup> concentration, as indicated by an increase in mtRCamp1h fluorescence, when cultured control VMs are paced at 2 Hz for 1 min under β-adrenergic stimulation with ISO (pacing indicated by red bar). We did not observe significant loading of mitochondria with Ca2<sup>+</sup> under baseline conditions with no ISO stimulation (data not

FIGURE 2 | Expression of biosensor mtRCamp1h in cultured control rat VMs reveals mitochondrial Ca2<sup>+</sup> accumulation is modified by kaempferol and CGP-37157. (A) Representative images of a cultured VM 48 h after adenoviral infection with mtRCamp1h (left, red, MOI of 10) and mitochondrial-targeted GFP (center, green, MOI of 10). A merged image is indicated on the right. (B) Recording of a permeabilized VM infected with mtRCamp1h. Fmax and Fmin were assessed by addition of EGTA (2 mmol/L) and Ca2<sup>+</sup> (100 µmol/L), respectively. (C) Representative trace from cultured VM expressing mtRCamp1h. VMs were treated with ISO (50 nmol/L) and paced at 2 Hz for 60 s intervals (red bars), then treated with kaempferol (KAEM, 10 µmol/L) and paced again peaks are overlaid in the inset for visualization, with ISO in black, ISO+KAEM in red. Graphs in (D) depict mean data ± SEM of mtRCamp1h transient amplitude, time to transient peak and tau of decay (n = 16 per group, N = 5, <sup>∗</sup>p < 0.05, paired Student's t-test). (E) Representative trace from VM expressing mtRCamp1h. VMs were treated with ISO and paced at 2 Hz for 60 s intervals (red bars), then treated with CGP-37157 (CGP, 1 µmol/L) and paced again. Peaks are overlaid in the inset for visualization, with ISO in black, ISO+CGP in blue. Graphs in (F) depict mean data ± SEM from mtRCamp1h transient amplitude, time to transient peak and tau of decay (n = 23, N = 4, <sup>∗</sup>p < 0.05, paired Student's t-test).

shown). To determine Fmin and Fmax for mtRCamp1h, VMs were saponin-permeabilized (0.001%) and equilibrated with a solution containing thapsigargin (10 µmol/L) to deplete SR Ca2+, cytochalasin D (10 µmol/L) to reduce cell contraction, as well as FCCP (20 µmol/L) and ionomycin (5 µmol/L) to dissipate mitochondrial membrane potential (1ψm). Solution containing Ca2<sup>+</sup> buffer EGTA at high concentration (2 mmol/L) was applied to obtain minimum mtRCamp1h fluorescence, while maximum fluorescence was achieved by application of Ca2<sup>+</sup> (100 µmol/L), as illustrated in **Figure 2B**.

Measurement of peak mtRCamp1h signal indicated accumulation of [Ca2+]<sup>m</sup> during workload, but after treatment with MCU enhancer kaempferol (10 µmol/L; **Figure 2C**) or NCLX blocker CGP-37157 (1 µmol/L; **Figure 2E**), no significant change in transient amplitude (an increase in [Ca2+]m) was observed (**Figures 2D,F**, respectively). However, accumulation of mitochondrial Ca2<sup>+</sup> during pacing was significantly faster after treatment with kaempferol (**Figure 2D**, time to peak 59.31 ± 1.01 s. ISO vs. 55.84 ± 1.24 s. ISO and kaempferol, <sup>∗</sup>p = 0.03), while the time constant of transient decay, τ, was significantly increased after application of CGP-37157 (**Figure 2F**, τ = 27.16 ± 3.10 ISO vs. 68.70 ± 8.81 ISO and CGP-37157, <sup>∗</sup>p < 0.001). The time to peak of transient after application of CGP-37157 was also increased (58.59 ± 1.23 s. ISO vs. 62.01 ± 0.98 s. ISO and CGP-37157, <sup>∗</sup>p = 0.01). This indicates that while pharmacological enhancers of [Ca2+]<sup>m</sup> do not increase overall mitochondrial Ca2<sup>+</sup> loading in cultured control VMs during workload of 1 min pacing, they modify the

time course for which those VMs accumulate or retain Ca2+. The finding that the rate of mitochondrial Ca2<sup>+</sup> uptake in intact myocytes is only modestly accelerated with kaempferol and even slowed down with CGP-37157 suggest the existence of overriding feedback mechanisms to prevent mitochondrial Ca2<sup>+</sup> overload that can cause cell death (Broekemeier et al., 1998; Hüser and Blatter, 1999; Elrod et al., 2010).

# Enhanced Mitochondrial Ca2<sup>+</sup> Accumulation Dissipates 1ψm

It was established that the excessive mitochondrial Ca2<sup>+</sup> uptake can be effectively limited by a reduction in 1ψm at least in part via activation of mitochondrial Ca2+-dependent K<sup>+</sup> channels (O'Rourke et al., 2005; Stowe et al., 2006). While irreversible mPTP opening collapses 1ψm, brief openings of the pore may also provide a protective efflux mechanism against mitochondrial Ca2<sup>+</sup> overload (Broekemeier et al., 1998; Elrod et al., 2010). We used isolated control rat VMs stained with voltage sensitive dye TMRM (20 µmol/L for 1 min) to determine if enhancement of mitochondrial Ca2<sup>+</sup> accumulation modifies 1ψm (**Figure 3**). As in **Figure 2**, VMs under β-adrenergic stimulation with ISO were paced at 2 Hz for 1 min, as indicated by red bars. Representative images **Figure 3A** shows VMs before and after the application of kaempferol. A representative trace is shown in **Figure 3B**, whereby signal was normalized to minimum fluorescence obtained by the application of FCCP (50 µmol/L). Application of kaempferol significantly reduced TMRM fluorescence (**Figure 3C**, 88.31 ± 1.73% ISO vs. 53.17 ± 4.69% ISO and kaempferol, <sup>∗</sup>p < 0.001). Application of CGP-37157 had similar effects (**Figures 3D–F**, 86.49 ± 1.95% ISO vs. 71.15 ± 3.38% ISO and CGP-37157, <sup>∗</sup>p < 0.001, respectively). The decrease in driving force due to the drop in 1ψm may explain why kaempferol or CGP-37157 are not able to increase the amplitude of [Ca2+]<sup>m</sup> effectively limiting mitochondrial Ca2<sup>+</sup> uptake, as measured with mtRCamp1h in **Figure 2**.

# Facilitation of Mitochondrial Ca2<sup>+</sup> Accumulation Promotes Proarrhythmic SCWs in VMs

Having demonstrated the effects of kaempferol and CGP-37157 on mitochondrial Ca2<sup>+</sup> and 1ψm, we next sought to establish the effects of modulating mitochondrial Ca2<sup>+</sup> on cytosolic Ca2<sup>+</sup> handling in VMs, as illustrated in **Figure 4**. Cultured control VMs were loaded with the fluorescent Ca2<sup>+</sup> indicator Rhod-2 and we

FIGURE 3 | Pharmacological enhancers of mitochondrial Ca2<sup>+</sup> accumulation modulate 1ψm in cultured control rat VMs. (A) Representative images of a VM treated with ISO (50 nmol/L) before and after application of kaempferol (KAEM, 10 µmol/L). 1ψm of VMs was monitored by TMRM fluorescence (20 µmol/L). (B) Representative recording of changes in 1ψm in response to application of ISO and pacing at 2 Hz (red bars) for 60 s, followed by application of KAEM. TMRM signal was normalized to minimum fluorescence obtained by application of FCCP (50 µmol/L) and represented as percentage of baseline. Graph in (C) depicts pooled data for minimum fluorescence after pacing and application of ISO or KAEM (n = 12, N = 4, <sup>∗</sup>p < 0.05, paired Student's t-test). (D) Representative images of a VM treated with ISO before and after application of CGP-37157 (CGP, 1 µmol/L). 1ψm of VMs was monitored by TMRM fluorescence. (E) Representative recordings of changes in 1ψm in response to application of ISO and pacing at 2 Hz for 60 s, followed by application of CGP. Graph in (F) depicts pooled mean data ± SEM for minimum fluorescence after pacing and application of ISO or KAEM (n = 14, N = 3, <sup>∗</sup>p < 0.05, paired Student's t-test).

induce SCW, before and after application of kaempferol (KAEM, 10 µmol/L). Graphs in (B) depict mean data ± SEM from Ca2<sup>+</sup> transient amplitude (n = 15, N = 5, <sup>∗</sup>p < 0.05, paired Student's t-test), SCW latency (n = 11, N = 5, paired Student's t-test) and the percentage of cells exhibiting waves (N = 5, <sup>∗</sup>p < 0.05, Fisher's exact test). (C) Representative confocal line scan images of Ca2<sup>+</sup> transients and Rhod-2 fluorescence (F/F0) profiles of ISO treated rat VM undergoing 2 Hz pace-pause protocol before and after application of CGP-37157 (CGP, 1 µmol/L). Graphs in (D) depict mean data ± SEM from Ca2<sup>+</sup> transient amplitude (n = 7, N = 4, ns is not significant, paired Student's t-test), SCW latency (n = 8, N = 4, <sup>∗</sup>p < 0.05, paired Student's t-test) and the percentage of cells exhibiting waves (N = 4, ns is not significant, Fisher's exact test).

recorded cytosolic Ca2<sup>+</sup> in the presence of β-adrenergic receptor agonist ISO (50 nmol/L), subjected to a burst-pace pause protocol (2 Hz, 20 s). We assessed Ca2<sup>+</sup> transient amplitude and SCW latency as an indication of the propensity for arrhythmogenic Ca2<sup>+</sup> release. As shown in **Figures 4A,B**, VMs treated with ISO and kaempferol have a significantly lower Ca2<sup>+</sup> transient amplitude in comparison to ISO alone (2.77 ± 0.44 1F/F<sup>0</sup> ISO vs. 1.57 ± 0.23 1F/F<sup>0</sup> ISO and kaempferol, <sup>∗</sup>p = 0.006). We observed a fivefold shortening in SCW latency after treatment with kaempferol (∗p < 0.001), as well a significant increase in the percentage of VMs exhibiting SCWs (63% ISO vs. 100% ISO and kaempferol, <sup>∗</sup>p < 0.001). A similar pattern was obtained in VMs treated with CGP-37157 (**Figures 4C,D**), whereby the SCW latency was shortened approximately twofold (∗p = 0.002). Changes in Ca2<sup>+</sup> transient amplitude and the percentage of cells exhibiting waves were not significant on treatment with CGP-37157.

To provide additional evidence that facilitation of mitochondrial Ca2<sup>+</sup> accumulation exerts strong deleterious effects on intracellular Ca2<sup>+</sup> handling we performed experiments using another MCU enhancer SB 202190 (Montero et al., 2004) (**Figure 5**). As seen from experiments in ISO-treated VMs

mtRCamp1h. VMs were treated with ISO (50 nmol/L) and paced at 2 Hz for 60 s intervals (red bars), then treated with SB 202190 (SB) and paced again. Peaks are overlaid in the inset for visualization, with ISO in black, ISO+SB in gray. Graphs in (B) depict mean data ± SEM of mtRCamp1h transient amplitude, time to transient peak and tau of decay (n = 12 per group, N = 3, <sup>∗</sup>p < 0.05, paired Student's t-test). (C) Representative confocal line scan images of Ca2<sup>+</sup> transients and Rhod-2 fluorescence (F/F0) profiles ISO treated (50 nmol/L) rat VM undergoing 2 Hz pace-pause protocol (red arrows) to induce SCW, before and after application of SB 202190 (SB, 30 µmol/L). Graphs in (D) depict mean data ± SEM from Ca2<sup>+</sup> transient amplitude (n = 16, N = 3, <sup>∗</sup>p < 0.05, paired Student's t-test) and SCW latency (n = 16, N = 3, <sup>∗</sup>p < 0.05, paired Student's t-test).

FIGURE 6 | Inhibition of RyR with dantrolene reduces proarrhythmic effects of kaempferol on Ca2<sup>+</sup> handling in cultured control rat VMs. (A) Representative confocal line scan images of Ca2<sup>+</sup> transients and Rhod-2 fluorescence (F/F0) profiles of TAB rat VMs undergoing 2 Hz pace-pause protocol (red arrows) to induce SCWs. VMs were treated with isoproterenol (ISO, 50 nmol/L) and kaempferol (KAEM, 10 µmol/L), or were pretreated for 5 min with dantrolene (DAN, 2 µmol/L) before ISO and KAEM treatment. Graphs in (B) depict mean data ± SEM from Ca2<sup>+</sup> transient amplitude (n = 13–15, N = 3–4, <sup>∗</sup>p < 0.05, Student's t-test) and SCW latency (n = 11–13, N = 3–4, <sup>∗</sup>p < 0.05, Student's t-test).

expressing mtRCamp1h, application of SB 202190 (30 µmol/L) moderately but significantly accelerated rate of mitochondrial Ca2<sup>+</sup> uptake while failed to increase overall amplitude of mitochondrial Ca2<sup>+</sup> transients (**Figures 5A,B**), in line with the effects of kaempferol (**Figure 2C**). Importantly, similar to kaempferol, incubation with SB reduced amplitude of Ca2<sup>+</sup> transients and shortened and latency of SCWs in periodically paced ISO-treated cells (**Figures 5C,D**).

Enhanced propensity to generate SCWs is often attributed to abnormally high activity of SR Ca2<sup>+</sup> release channels, RyRs (Bers, 2002). To test whether stabilization of RyRs can attenuate kaempferol–mediated effects on Ca2<sup>+</sup> handling, we treated VMs with dantrolene (2 µmol/L, 5 min), a specific inhibitor of RyR. **Figure 6** shows that dantrolene restores Ca2<sup>+</sup> transient amplitude and significantly reduces kaempferolmediated shortening of SCW latency in ISO-treated VMs (SCW latency 3.34 ± 0.73 s ISO and kaempferol vs. 2.64 ± 0.60 s ISO, kaempferol and dantrolene, <sup>∗</sup>p < 0.001).

Taken together, these data suggest that facilitation of mitochondrial Ca2<sup>+</sup> accumulation promotes generation of proarrhythmic SCWs by enhancement of activity of RyRs.

## Kaempferol and CGP-37157 Increase ROS and Oxidation of RyRs

RyR is well established as a ROS sensor, with increased oxidation and thus RyR activity associated with enhanced propensity for spontaneous SR Ca2<sup>+</sup> release and proarrhythmic SCWs.

Mitochondria is a major intracellular source of ROS and excessive RyR oxidation by mitochondria-derived ROS has been demonstrated in many models of HF and aging (Zima and Blatter, 2006). Facilitation of mitochondrial Ca2<sup>+</sup> uptake may accelerate the rate of emission of ROS by stimulating electron transport (Bertero and Maack, 2018). Therefore, to test the possible effects of kaempferol on mitochondrial redox state, control VMs were infected (MOI of 10) with adenovirus construct carrying sequence encoding the novel H2O<sup>2</sup> probe OMM-HyPer, and cultured for 48 h prior to imaging. As shown in **Figure 7**, application of kaempferol significantly increases the signal of mitochondrial-targeted peroxide-sensitive indicator (normalized fluorescence 16.29 ± 2.11% ISO vs. 54.23 ± 5.31% ISO and kaempferol, <sup>∗</sup>p < 0.001), confirming that facilitation of mitochondrial Ca2<sup>+</sup> accumulation induces mitochondrial ROS release.

Considering the close proximity of mitochondria and SR, we next sought to determine whether facilitation of mitochondrial Ca2<sup>+</sup> accumulation leads to an increase in local ROS levels in the vicinity of RyR in cultured control VMs, measured using the ER-tuned redox-sensitive biosensor ERroGFP\_iE (Avezov et al., 2013). The GFP sensor contains engineered cysteine residues that enable formation of di-thiol in response to oxidant stress (Cannon and Remington, 2009). After we generated adenovirus encoding the sensor, VMs were infected at a MOI of 10 and cultured for 48 h prior to imaging. As shown in **Figure 8A**, ERroGFP\_iE-infected VMs exhibit a striated pattern indicative of SR targeting of the probe.

**Figures 8B,D** show representative recordings where signal of ERroGFP\_iE was normalized to minimal fluorescence obtained by application of reducing agent dithiothreitol (DTT, 5 mmol/L) and maximal fluorescence obtained by application of oxidizing agent 2,2'-dithiodipyridine (DTDP, 200 µmol/L). At baseline under β-adrenergic stimulation with ISO, little change in oxidation in the form of increased fluorescence and di-thiol formation is observed. However, treatment with kaempferol or CGP-37157 still leads to a measurable and significant increase in ERroGFP\_iE signal (8.19 ± 0.83% ISO vs. 22.20 ± 2.49% ISO and kaempferol, <sup>∗</sup>p < 0.001 and 11.21 ± 2.03% ISO vs. 16.26 ± 2.32 ISO and CGP-37157, <sup>∗</sup>p = 0.006, respectively), indicative of increased oxidation and ROS emission in close proximity of RyR (**Figures 8C,E**, respectively).

To directly test the hypothesis that pharmacological enhancement of [Ca2+]<sup>m</sup> accumulation results in increased RyR oxidation, the free thiol content of immunoprecipitated RyRs was measured using DNP-antibody. **Figure 8F** demonstrates that

treatment with kaempferol or CGP-37157 (for 5 and 10 min, respectively) significantly increases oxidation of RyR fivefold and sixfold, respectively (**Figure 8G**, <sup>∗</sup>p = 0.047 and #p = 0.015), shown by increased DNP signal. These data demonstrate that in control VMs, enhanced mitochondrial Ca2<sup>+</sup> accumulation leads to increased ROS in the closely situated SR. Subsequent increased RyR oxidation is likely the responsible mechanism for the shortened latency for proarrhythmic RyR-mediated SCWs in VMs with pharmacologically enhanced mitochondrial Ca2<sup>+</sup> accumulation.

# The Effects of Pharmacological Modulators on Mitochondrial Ca2<sup>+</sup> Accumulation in TAB VMs

Kaempferol and CGP-37157 modulate mitochondrial Ca2<sup>+</sup> accumulation, membrane potential and ROS emission in cultured VMs from control rat hearts, as well as intracellular Ca2<sup>+</sup> cycling and SCW generation. However, in the diseased heart, redox balance is altered and both intracellular and mitochondrial Ca2<sup>+</sup> handling can be compromised already (Kim et al., 2017).

To test whether tethering of mitochondria to the SR is altered in TABs rat VMs, we performed western blot analysis of expression levels of mitofusin 1 and mitofusin 2; proteins that scaffold these two organelles (de Brito and Scorrano, 2008; Chen et al., 2012; Filadi et al., 2015). We did not find changes in expression levels of both these proteins in VMs of TABs vs. those of controls (**Figure 9**). Next, isolated TAB VMs were infected with adenoviruses to express mitochondrial Ca2<sup>+</sup> indicator mtRCamp1h. **Figures 10A–D** demonstrate that the effects of kaempferol and CGP-37157 on mitochondrial Ca2<sup>+</sup> accumulation are qualitatively similar to those in control VMs. As in controls, kaempferol accelerates Ca2<sup>+</sup> accumulation (time to peak 72.43 ± 3.75 s ISO vs. 56.14 ± 4.47 s ISO and kaempferol, <sup>∗</sup>p < 0.001) but the amplitude does not change significantly, while CGP-37157 impedes mitochondrial Ca2<sup>+</sup> transient decay (tau of decay 32.08 ± 7.49 s ISO vs. 72.38 ± 14.99 s ISO and CGP-37157, <sup>∗</sup>p = 0.006). In addition, we tested the effects of MCU inhibitor Ru360. **Figures 10E,F** show representative traces and respective pooled data for mitochondrial Ca2<sup>+</sup> transients in ISO-treated control VMs under basal conditions and after 30 min incubation with 2 µmol/L Ru360. **Figures 10G,H** demonstrates that Ru360 reduces mitochondrial Ca2<sup>+</sup> accumulation in TABs (0.29 ± 0.07 1F/F<sup>0</sup> ISO vs. 0.14 ± 0.03 1F/F<sup>0</sup> ISO and Ru360, <sup>∗</sup>p < 0.001). Notably, the amplitude of pacing-induced mitochondrial Ca2<sup>+</sup> transients in ISO-treated TAB VMs is significantly smaller that in controls (amplitude 1.15 ± 0.09 1F/F<sup>0</sup> ISO Control vs. 0.30 ± 0.07 1F/F<sup>0</sup> ISO TAB, <sup>∗</sup>p < 0.001, Student's t-test). Given this difference, we next performed experiments using freshly isolated VMs from hypertrophic TAB rat hearts, to test whether kaempferol and CGP-37157 have deleterious effects in a disease setting, or on the contrary could be protective.

FIGURE 9 | Expression of mitochondrial associated membrane proteins Mfn1 and Mfn2 is not altered in hypertrophic rat VMs. (A) Representative images of control and TAB VMs immunoblotted for expression of mitofusin 1 (Mfn1) and mitofusin 2 (Mfn2). GAPDH was used as a loading control. (B) Quantification of normalized optical densities (N = 4 per group, Student's t-test).

# Enhanced Mitochondrial Ca2<sup>+</sup> Accumulation Further Perturbs Intracellular Ca2<sup>+</sup> Cycling in TAB VMs

We studied the effects of enhancing mitochondrial Ca2<sup>+</sup> accumulation in TAB VMs on cytosolic Ca2<sup>+</sup> cycling using Rhod-2, as illustrated in **Figure 11**. As with control VMs, the burst pacing-pause protocol was used to assess the propensity of arrhythmogenic Ca2<sup>+</sup> release (**Figure 11A**). After treatment with kaempferol, TAB VMs has a significantly reduced Ca2<sup>+</sup> transient amplitude (5.59 ± 0.71 1F/F<sup>0</sup> ISO vs. 3.08 ± 0.42 1F/F<sup>0</sup> ISO and kaempferol, <sup>∗</sup>p = 0.002), as well as a approximately twofold in decrease in SCW latency (**Figure 11C**, <sup>∗</sup>p = 0.001). The percentage of cells exhibiting SCWs was also significantly increased after kaempferol application (71% ISO vs. 100% ISO and kaempferol, <sup>∗</sup>p = 0.01). These changes were accompanied by a significant decrease in SR Ca2<sup>+</sup> content assessed by application of 10 mmol/L caffeine (**Figure 11B**, 5.46 ± 0.77 1FcafF/F<sup>0</sup> ISO vs. 3.19 ± 0.30 1FcafF/F<sup>0</sup> ISO and kaempferol, <sup>∗</sup>p = 0.01). Decreased Ca2<sup>+</sup> transient amplitude and reduced SR Ca2<sup>+</sup> load is indicative of increased Ca2<sup>+</sup> leak via hyperactive RyRs (Belevych et al., 2007; Terentyev et al., 2008).

Excessive ROS production is a hallmark of hypertrophy and we have previously established that in TAB VMs, there is increased mitochondrial ROS production in comparison to healthy controls which results in oxidation and thereby abnormally high activity of RyRs (Kim et al., 2017). In the present study, parallel experiments using the mitochondriaspecific ROS indicator MitoSOX demonstrated that enhancement of mitochondrial Ca2<sup>+</sup> accumulation with kaempferol further increased ROS emission in diseased VMs (**Figures 11D,E**). Signal was normalized to maximal fluorescence obtained upon application of 10 mmol/L H2O2.

Similar effects on cytosolic Ca2<sup>+</sup> transients were obtained when enhancing mitochondrial Ca2<sup>+</sup> accumulation in TAB VMs with block of NCLX via CGP-37157 (**Figures 12A,B**). There was a significant reduction in SCW latency (**Figure 12C**, 4.79 ± 0.38 s. ISO vs. 1.99 ± 0.45 s. ISO and CGP-37157, <sup>∗</sup>p < 0.001), although no significant change in the SR Ca2<sup>+</sup> content assessed by caffeine application was observed (**Figures 12B,C**).

## Inhibition of Mitochondrial Ca2<sup>+</sup> Uptake With Ru360 Reduces ROS Emission and Increases Latency for Proarrhythmic SCWs

Given that application of MCU-inhibitor Ru360 attenuated triggered activity and arrhythmogenesis in TAB hearts ex vivo (**Figure 1**), we next assessed intracellular Ca2<sup>+</sup> handling of VMs preincubated with 2 µmol/L Ru360 for 30 min. There were no significant differences in the Ca2<sup>+</sup> transient amplitude and the percentage of cells exhibiting waves or the caffeine-sensitive SR Ca2<sup>+</sup> content (**Figures 13A–C**). However, there was a significant increase in SCW latency, (**Figures 13A,C**, 1.42 ± 0.19 s. ISO vs. 1.96 ± 0.19 s. ISO and Ru360, <sup>∗</sup>p < 0.001), indicative of stabilization of RyR-mediated Ca2<sup>+</sup> release. Myocytes from TAB hearts displayed a decrease in MitoSOX fluorescence after

FIGURE 10 | Pharmacological modulators of mitochondrial Ca2<sup>+</sup> in TAB VMs exert similar effects to those on controls, despite hypertrophy-related impairment in mitochondrial Ca2<sup>+</sup> homeostasis. (A) Representative trace from cultured TAB VM expressing mtRCamp1h. TAB VMs were treated with ISO (50 nmol/L) and paced at 0.5 Hz for 60 s intervals (red bars), before treatment with kaempferol (KAEM, 10 µmol/L) and further pacing. Peaks are overlaid in the inset for visualization, with ISO in black, ISO+KAEM in red. Graphs in (B) depict mean data ± SEM of mtRCamp1h transient amplitude, time to transient peak and tau of decay (n = 22 per group, N = 4, <sup>∗</sup>p < 0.05, paired Student's t-test). (C) Representative trace from cultured TAB VM expressing mtRCamp1h. VMs were treated with ISO and paced at 0.5 Hz for 60 s intervals (red bars), before treatment with CGP-37157 (CGP, 1 µmol/L) and further pacing. Peaks are overlaid in the inset for visualization, with ISO in black, ISO+CGP in blue. Graphs in (D) depict mean data ± SEM from mtRCamp1h transient amplitude, time to transient peak and tau of decay (n = 11, N = 3, <sup>∗</sup>p < 0.05, paired Student's t-test). (E) Representative trace from a cultured control VM expressing mtRCamp1h. Myocytes were treated with ISO (black line) and paced at 2 Hz for 60 s (red bar), or pretreated with Ru360 (2 µmol/L) for 30 min before treatment with ISO and pacing at 2 Hz (pink line). Graph in (F) depicts mean data ± SEM from mtRCamp1h transient amplitude (n = 17–18, N = 3, <sup>∗</sup>p < 0.05, Student's t-test). (G) Representative trace from a cultured TAB VM expressing mtRCamp1h. Myocytes were treated with ISO (black line) and paced at 2 Hz for 60 s (red bar), or pretreated with Ru360 (2 µmol/L) for 30 min before treatment with ISO and pacing at 0.5 Hz (pink line). Graph in (H) depicts mean data ± SEM from mtRCamp1h transient amplitude (n = 17, N = 4, <sup>∗</sup>p < 0.05, Student's t-test).

preincubation with Ru360 (**Figures 13D,E** 15.46 ± 1.91% ISO vs. 14.93 ± 2.15% ISO and Ru360, <sup>∗</sup>p < 0.001). There was also a significant reduction in oxidation of immunoprecipitated RyRs after treatment with Ru360 assessed using anti-DNP antibodies (**Figures 13F,G**). These data suggest that stabilization of SR Ca2<sup>+</sup> release stems from attenuation of ROS emission by mitochondria and normalization of RyR redox state.

# Modifiers of Mitochondrial Ca2<sup>+</sup> Uptake and Retention Do Not Alter the Velocity of SCWs

Regenerative SCWs propagate via the 'fire-diffuse-fire' mechanism (Keizer and Smith, 1998; Maxwell and Blatter, 2012), whereby Ca2<sup>+</sup> released from one cluster of RyR channels activates Ca2<sup>+</sup> release from another. Increasing Ca2<sup>+</sup> buffering can intercept Ca2<sup>+</sup> diffusing from cluster to cluster and modulate SCW velocity (Ramay et al., 2010; Eisner et al., 2017), as was shown with SR Ca2+-ATPase (SERCa) enhancers (Fernandez-Tenorio and Niggli, 2018). It could be suggested that slower SCW wave propagation after enhancement of mitochondrial Ca2<sup>+</sup> accumulation indicates a Ca2<sup>+</sup> buffering capacity of mitochondria, serving as a sink for cytosolic Ca2+. However, neither the enhancement (with kaempferol or CGP-37157) nor attenuation (with Ru360) of mitochondrial Ca2<sup>+</sup> accumulation altered the velocity of SCWs in either cultured control VMs (**Figure 14A**) or hypertrophic TAB VMs (**Figure 14B**). These data suggest that buffering capacity of mitochondria is insufficient to interrupt or slow SCWs, possibly due to concomitant changes in 1ψm and local ROS.

<sup>∗</sup>p < 0.05, Student's t-test). (D) Representative recording of ROS production measured with MitoSOX in VMs. Signal was normalized to maximum fluorescence obtained on application of 10 mmol/L H2O2. VMs were treated with ISO and kaempferol and paced at 2 Hz (red bars) for 60 s. Graph in (E) depicts mean

# DISCUSSION

The contribution of mitochondrial Ca2<sup>+</sup> flux to myocyte excitation-contraction remains the subject of intense research, with both enhancement or reduction of [Ca2+]<sup>m</sup> posited as therapeutic strategies to improve cardiac function and prevent arrhythmia in cardiac disease (Liu and O'Rourke, 2008; Kolhaas et al., 2010; Liu et al., 2014; Dietl and Maack, 2017; Schweitzer et al., 2017; Xie et al., 2018). Our present study provides evidence that enhanced mitochondrial Ca2<sup>+</sup> accumulation dissipates 1ψm and drives increased ROS in the mitochondria-SR microdomain. Subsequently, increased modification of RyR by ROS enhances channel activity and increases the propensity for proarrhythmic spontaneous SR Ca2<sup>+</sup> release in the form of SCWs. This mechanism further exacerbates proarrhythmic triggered activity in hypertrophic hearts. Conversely, inhibition

data ± SEM of normalized MitoSOX fluorescence (n = 7, N = 3, <sup>∗</sup>p < 0.05, paired Student's t-test).

of MCU is protective against arrhythmogenesis, attenuating oxidative stress and reducing aberrant activity of RyR.

# Modulation of Mitochondrial Ca2<sup>+</sup> and the Effects on Mitochondrial Function

The physical and functional coupling of the SR and mitochondria is critical for matching myocyte workload to mitochondrial ATP generation (Dorn and Maack, 2013; Lopez-Crisosto et al., 2017). The close association facilitates mitochondrial Ca2<sup>+</sup> influx upon SR Ca2<sup>+</sup> release (Sharma et al., 2000; Szalai et al., 2000; Csordás et al., 2001). To maintain Ca2<sup>+</sup> flux balance in the steady state, uptake of mitochondrial Ca2<sup>+</sup> is well matched to extrusion, so net [Ca2+]<sup>m</sup> does not significantly change. We assessed mitochondrial Ca2<sup>+</sup> uptake in cultured control and TAB VMs with the genetically encoded Ca2<sup>+</sup> probe mtRCamp1h.

In comparison to baseline, under β-adrenergic stimulation with ISO we observed an increase in [Ca2+]<sup>m</sup> during periodic pacing (**Figures 2B,E**, **10A,C**), likely to match the increased demand for ATP.

To modulate mitochondrial Ca2<sup>+</sup> influx in cultured control VMs, we applied kaempferol or SB 202190 to directly enhance uptake through MCU, or CGP-37157 to inhibit NCLX thus attenuating Ca2<sup>+</sup> efflux. Interestingly, application of these pharmacological agents did not increase [Ca2+]m, but instead altered the dynamics of Ca2<sup>+</sup> accumulation or retention (**Figures 2D,F**). Similar effects were observed in TAB VMs treated with kaempferol and CGP-37157 (**Figures 10B,D**). The inability to significantly increase total [Ca2+]<sup>m</sup> under these conditions points to the existence of limiting factors, preventing excessive mitochondrial Ca2<sup>+</sup> loading in VMs.

There are three established mechanisms to prevent mitochondrial Ca2<sup>+</sup> overload. Activation of mitochondrial Ca2+-activated K<sup>+</sup> channels limits depolarization of the mitochondria, reducing the electrochemical inward driving force for Ca2<sup>+</sup> (O'Rourke et al., 2005; Clements et al., 2015). Acidification of the mitochondrial matrix by an increased H<sup>+</sup> flow through ATP-synthase has been shown to inhibit MCU (Moreau and Parekh, 2009). Additionally, increased matrix Ca2<sup>+</sup> also promotes the opening of mPTP, dissipating 1ψm and limiting [Ca2+]<sup>m</sup> accumulation (Broekemeier et al., 1998; Hüser and Blatter, 1999; Elrod et al., 2010). This is evident in **Figure 3**, whereby application of kaempferol or CGP-37157 to cultured control VMs significantly depolarized 1ψm. Interestingly, preincubation of VMs with mitoTEMPO, a specific mitochondria ROS scavenger (20 µmol/L, 30 min), did not prevent dissipation of 1ψm by kaempferol (normalized TMRM fluorescence 89.7 ± 2.16% mitoTEMPO and ISO vs. 39.17 ± 3.59% mitoTEMPO, ISO and kaempferol, <sup>∗</sup>p < 0.001, paired Student's t-test, n = 19, N = 3), suggesting that this process is not ROS-dependent.

# Modulation of SR Ca2<sup>+</sup> Release by [Ca2+]<sup>m</sup> Is Mediated by Mitochondrial ROS

The RyR-mediated release of Ca2<sup>+</sup> from the SR is critical to contractile activation. Termination of SR Ca2<sup>+</sup> release allows for Ca2<sup>+</sup> released into the cytosol to be resequestered and maintains the refractoriness of Ca2<sup>+</sup> signaling during diastole (Terentyev et al., 2002; Szentesi et al., 2004; Sobie et al., 2005). Shortened Ca2<sup>+</sup> signaling refractoriness due to hyperactive RyR increases the rate of SCWs in diseased myocytes (Belevych et al., 2012; Brunello et al., 2013), contributing to the pathogenesis of triggered arrhythmias (Pogwizd and Bers, 2004).

We assessed the effects of [Ca2+]<sup>m</sup> modulation on Ca2<sup>+</sup> handling and initiation of triggered activity at the whole organ level, using hearts from rats with TAB-induced hypertrophy, as illustrated in **Figure 1**. Application of MCU activator kaempferol reduced Ca2<sup>+</sup> transient amplitude and exacerbated the proarrhythmic phenotype, with incidences of PVCs and spontaneous VT/VFs with higher VF frequencies. The focal activity and/or transmural outbreak-like activation patterns were frequently seen during PVCs and VFs, suggesting enhanced triggered activity may underlie exacerbated ventricular arrhythmias by kaempferol. In contrast the MCU blocker, Ru360, suppressed spontaneous VT/VFs. The activation maps of pacing induced VFs showed that rapid pacing caused conduction

block and reentry formation, suggesting that the protective effect of Ru360 is most likely through suppressing triggered activity. In recordings of intracellular Ca2<sup>+</sup> transients, it was evident that kaempferol and NCLX inhibitor CGP-37157 had detrimental effects on both cultured control VMs (**Figure 4**) and TAB VMs (**Figures 11**, **12**), shortening SCW latency and increasing the propensity for spontaneous Ca2<sup>+</sup> release, indicative of increased activity of RyR in both cultured control and hypertrophic TAB VMs. The experiments with RyR inhibitor dantrolene demonstrating attenuation of kaempferol-induced disturbances in Ca2<sup>+</sup> cycling (**Figure 6**) further confirm the central role of dysregulated RyR-mediated Ca2<sup>+</sup> release in this process.

Hyperactivity of RyR in cardiac disease is often attributed to posttranslational modifications, including phosphorylation of PKA- and CaMKII-specific sites, and oxidation of many reactive cysteines within the protein (Györke and Carnes, 2008; Niggli et al., 2013; Zima et al., 2014). Mitochondria are a major source of ROS in the myocyte and excessive ROS production is a hallmark of HF, hypertrophy and aging (Zima and Blatter, 2006), in parallel with perturbed Ca2<sup>+</sup> homeostasis (Terentyev et al., 2008; Cooper et al., 2013; Kim et al., 2017). Mitochondrial-ROS signaling has also been suggested to directly modulate Ca2<sup>+</sup> spark activity (Yan et al., 2008; Zhou et al., 2011). Our experiments using mitochondria-targeted ROS biosensor OMM-HyPer3 show that kaempferol induces surge in mito-ROS production (**Figure 7**). Furthermore, in the present study, **Figure 8** clearly demonstrates increased local ROS in the vicinity of RyR and RyR oxidation in myocytes treated with kaempferol and CGP-37157. Importantly, while oxidative stress is already significant in hypertrophic myocytes, application of kaempferol further increased mitochondrial ROS emission in TAB VMs (**Figure 11D**), exacerbating Ca2<sup>+</sup> mishandling.

Graphs depict mean ± SEM (n = 6–9, N = 4–5, ns is non-significant, paired Student's t-test). (B) Velocity of spontaneous Ca2<sup>+</sup> waves (SCWs) in TAB VMs. Graphs depict mean ± SEM (n = 10–20, N = 4–5, ns is non-significant, Student's t-test and paired Student's t-test where appropriate).

Scavenging of mitochondrial ROS has been sufficient to alleviate the arrhythmogenic phenotype in multiple disease states (Mochizuki et al., 2007; Belevych et al., 2012; Cooper et al., 2013; Luo et al., 2013; Joseph et al., 2016) and normalize the redox state of RyR in TAB-induced hypertrophy (Kim et al., 2017). Importantly, inhibition of mitochondrial Ca2<sup>+</sup> uptake with Ru360 was sufficient to attenuate Ca2+-dependent arrhythmia in ex vivo TAB hearts (**Figure 1**) and normalize Ca2<sup>+</sup> homeostasis in isolated TAB VMs, stabilizing RyRmediated Ca2<sup>+</sup> release and attenuating proarrhythmic SCWs (**Figures 13A–C**). Furthermore, Ru360 reduced mitochondrial ROS emission assessed using mitochondria-specific ROS indicator mitoSOX (**Figures 13D,E**). This indicates that block of mitochondrial Ca2<sup>+</sup> influx reduces mitochondrial ROS signaling in TAB VMs resulting in reduction of oxidation levels of RyR (**Figures 13F,G**). Our data suggest that increased mitochondrial Ca2<sup>+</sup> accumulation facilitates increased mitochondrial ROS emission and the oxidation of RyR. This underlies enhanced RyR activity, increased spontaneous Ca2<sup>+</sup> release in the form of arrhythmogenic SCWs, and a vicious cycle of Ca2+/ROS-induced myocyte dysfunction.

# Inhibition of NCLX-Mediated Mitochondrial Ca2<sup>+</sup> Efflux Exacerbates Ca2<sup>+</sup> Mishandling

During pathological mitochondrial Ca2<sup>+</sup> overload, opening of mPTP offers an additional Ca2<sup>+</sup> efflux pathway (Broekemeier et al., 1998; Hüser and Blatter, 1999; Elrod et al., 2010). Pharmacological inhibition of mPTP with cyclosporine A or genetic ablation of mPTP component cyclophilin D (that reduces opening) has shown to be protective against HF or ischemia-reperfusion injury (Griffiths and Halestrap, 1993; Hausenloy et al., 2010; Yarana et al., 2012; Gordan et al., 2016). However, there is ongoing controversy as to whether inhibition of NCLX can be protective in cardiac disease. In a guinea pig HF model, chronic inhibition of NCLX with CGP-37157 restored diminished [Ca2+]m, thereby improving redox homeostasis and protecting against arrhythmogenesis (Liu et al., 2014). In agreement with these findings, we recorded diminished mitochondrial Ca2<sup>+</sup> transients in VMs from diseased hearts in comparison to controls (**Figure 2** vs. **Figure 10**). However, incubation with CGP-37157 did not

change significantly mitochondrial Ca2<sup>+</sup> transient amplitude despite slowing down transient decay in TABs. It also did not improve but even worsened aberrant intracellular Ca2<sup>+</sup> handling (**Figure 12**). Xie et al. (2018) reported an increased mitochondrial Ca2<sup>+</sup> influx during diastolic period in mice with non-ischemic HF and posited that increased mitochondrial Ca2<sup>+</sup> efflux drives activation of the sarcolemmal NCX and initiates EADs. In this work, inhibition of both influx and efflux were reported to have anti-arrhythmic effects. In direct contrast, the present study clearly demonstrates that inhibition of NCLX in both control and TAB VMs promotes proarrhythmic spontaneous SR Ca2<sup>+</sup> release (**Figures 4**, **12**). Importantly, knockdown of NCLX in a conditional loss-of-function mouse model caused severe myocardial dysfunction, HF and sudden cardiac death (Luongo et al., 2017). This was attributed to substantially increased ROS generation due to mitochondrial Ca2<sup>+</sup> overload. Our results are in line with the latter and inhibition of NCLX with CGP-37157 significantly increased ROS emission in cultured control VMs (**Figures 7**, **8**), resulting in defective SR Ca2<sup>+</sup> handling in both cultured control (**Figure 4**) and TAB (**Figure 12**) VMs.

# Modification of Mitochondrial Ca2<sup>+</sup> Influx and Efflux Does Not Affect SCW Propagation

While mitochondrial Ca2<sup>+</sup> flux is closely associated with ATP generation and ROS emission, it has also been suggested that mitochondria may act as buffers that can shape global Ca2<sup>+</sup> transients during EC coupling (Maack et al., 2006; Yan et al., 2008; Walsh et al., 2009; Drago et al., 2012; Zhao et al., 2013). Although the low affinity of MCU for Ca2<sup>+</sup> (Kd ∼ 10–20 µmol/L Ca2+; Bernardi, 1999) would limit uptake during diastole and normal Ca2<sup>+</sup> transients, mitochondrial Ca2<sup>+</sup> uptake could occur at high local [Ca2+] near SR Ca2<sup>+</sup> release sites (Andrienko et al., 2009). One end of the mitochondria is in close proximity to these sites (∼37–270 nm; Sharma et al., 2000), and is physically tethered to the SR (García-Pérez et al., 2011; Chen et al., 2012) with strategic positioning of MCU near RyR (De La Fuente et al., 2016). This facilitates crosstalk between the organelles and tunneling of Ca2<sup>+</sup> between the two has been reported in striated muscle (Shkryl and Shirokova, 2006). In cardiac disease, mitochondria-SR interfaces and therefore Ca2<sup>+</sup> transport may be altered due to changes in expression levels of scaffolding proteins including Mfn1 and 2 (Dorn et al., 2015). We tested whether this occurs in our model of hypertrophy and did not find changes in expression levels of these proteins (**Figure 9**). Mitochondria may act as a buffer, serving as a sink of local Ca2+. Indeed, in a mouse model of CPVT, enhancement of mitochondrial Ca2<sup>+</sup> influx reduced frequency of arrhythmogenic Ca2<sup>+</sup> waves and incidences of VT/VF (Schweitzer et al., 2017). Also, Zhao et al. (2013) saw a reduction in the frequency of SCWs in myocytes treated with kaempferol after an FCCP-induced reduction in 1ψm, while Ru360 increased SCW frequency.

Although SR Ca2<sup>+</sup> release events are an important driver of [Ca2+]m, as the total mitochondria Ca2<sup>+</sup> flux is small, the overall ability of mitochondria to shape intracellular Ca2<sup>+</sup> dynamics remains debated (Dedkova and Blatter, 2008, 2013; O'Rourke and Blatter, 2009; Hohendanner et al., 2013; Williams et al., 2013; Eisner et al., 2017). Troponin I and SERCa are significant Ca2<sup>+</sup> buffers in the myocyte, and enhancement or increased expression of SERCa has been shown to significantly improve Ca2<sup>+</sup> buffering capacity and attenuate arrhythmogenic spontaneous Ca2<sup>+</sup> release (Lyon et al., 2011; Briston et al., 2014). In recent work of Fernandez-Tenorio and Niggli (2018), specific enhancers of SERCa activity reduced the frequency and velocity of SCWs in mouse VMs. In the present study, enhancement of mitochondrial Ca2<sup>+</sup> influx significantly increases the propensity for SCW in both cultured control and TAB VMs (**Figures 4**, **5**, **11**, **12**), while neither enhancement or inhibition of [Ca2+]<sup>m</sup> modulated the velocity and propagation of SCWs (**Figure 14**). Importantly, we do not find any differences in the rate of propagation of SCWs between control and TAB VMs where impairment of mitochondria to sequester Ca2<sup>+</sup> leads to profound approximately fourfold decrease in the amplitude of pacing-induced mitochondrial Ca2<sup>+</sup> transient (**Figures 2**, **10**). These data are in line with previous studies that suggest while Ca2<sup>+</sup> modulates mitochondrial function, mitochondria do not serve as a significant buffer of intracellular cytosolic Ca2<sup>+</sup> (Bers et al., 1993; Negretti et al., 1993; Lu et al., 2013; Williams et al., 2013). The effects of increased ROS emission on intracellular Ca2<sup>+</sup> homeostasis are likely overwhelming and cannot be compensated for by an increase in local Ca2<sup>+</sup> buffering capacity by mitochondria.

# CONCLUSION

In conclusion, our data suggest that pharmacological enhancement of mitochondrial Ca2<sup>+</sup> accumulation produces deleterious effects on Ca2<sup>+</sup> homeostasis under β-adrenergic stimulation. It promotes excessive ROS that enhances RyR activity thereby proarrhythmic spontaneous Ca2<sup>+</sup> release. In cardiac hypertrophy, where ROS defenses are weakened, it exacerbates the proarrhythmic alterations in Ca2<sup>+</sup> handling. Inhibition of mitochondria Ca2<sup>+</sup> uptake is protective because it reduces emission of ROS by mitochondria.

# AUTHOR CONTRIBUTIONS

SH, B-RC, JO-U, GC, and DT participated in the study design. SH wrote first draft of manuscript. SH, RT, TYK, PB, RC, JO-U performed the experiments. SH, RT, TYK, B-RC, DT conducted data interpretation and analyses. SH, RT, TYK, PB, RC, JO-U, GC, B-RC and DT reviewed the manuscript submitted for publication. All authors revised and approved the final version of the manuscript.

# FUNDING

This work was supported by American Heart Association Grant #18POST33960456 to SH, National Heart, Lung, and Blood Institute at the National Institutes of Health (NIH) RO1HL135236 to RC, NIH 7R01HL136757 to JO-U, NIH 1R01HL142864 to GC, NIH R01HL096669 to B-RC, American Heart Association Grant in Aid 15GRNT25650002 and NIH R01HL121796 to DT.

#### REFERENCES

fphys-09-01831 December 19, 2018 Time: 18:27 # 20


redox state of the electron transport chain. Biochemistry 22, 13059–13065. doi: 10.1021/bi980820c


pivotal roles in Ca2+ and reactive oxygen species signaling. J. Cell Sci. 126(Pt 14), 2965–2978. doi: 10.1242/jcs.093609



B56alpha and causing CaMKII- dependent hyperphosphorylation of RyR2. Circ. Res. 104, 514–521. doi: 10.1161/CIRCRESAHA.108.181651


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Hamilton, Terentyeva, Kim, Bronk, Clements, O-Uchi, Csordás, Choi and Terentyev. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Role of Phosphatidylinositol 3-Kinase (PI3K), Mitogen-Activated Protein Kinase (MAPK), and Protein Kinase C (PKC) in Calcium Signaling Pathways Linked to the α1-Adrenoceptor in Resistance Arteries

#### Alejandro Gutiérrez, Cristina Contreras, Ana Sánchez and Dolores Prieto\*

Departamento de Fisiología, Facultad de Farmacia, Universidad Complutense de Madrid, Madrid, Spain

#### Edited by:

Maria Fernandez-Velasco, Hospital Universitario La Paz, Spain

#### Reviewed by:

Peter S. Reinach, Wenzhou Medical University, China Marta Gil-Ortega, CEU San Pablo University, Spain

> \*Correspondence: Dolores Prieto dprieto@ucm.es

#### Specialty section:

This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology

Received: 17 September 2018 Accepted: 17 January 2019 Published: 06 February 2019

#### Citation:

Gutiérrez A, Contreras C, Sánchez A and Prieto D (2019) Role of Phosphatidylinositol 3-Kinase (PI3K), Mitogen-Activated Protein Kinase (MAPK), and Protein Kinase C (PKC) in Calcium Signaling Pathways Linked to the α1-Adrenoceptor in Resistance Arteries. Front. Physiol. 10:55. doi: 10.3389/fphys.2019.00055 Insulin resistance plays a key role in the pathogenesis of type 2 diabetes and is also related to other health problems like obesity, hypertension, and metabolic syndrome. Imbalance between insulin vascular actions via the phosphatidylinositol 3-Kinase (PI3K) and the mitogen activated protein kinase (MAPK) signaling pathways during insulin resistant states results in impaired endothelial PI3K/eNOS- and augmented MAPK/endothelin 1 pathways leading to endothelial dysfunction and abnormal vasoconstriction. The role of PI3K, MAPK, and protein kinase C (PKC) in Ca2<sup>+</sup> handling of resistance arteries involved in blood pressure regulation is poorly understood. Therefore, we assessed here whether PI3K, MAPK, and PKC play a role in the Ca2<sup>+</sup> signaling pathways linked to adrenergic vasoconstriction in resistance arteries. Simultaneous measurements of intracellular calcium concentration ([Ca2+]<sup>i</sup> ) in vascular smooth muscle (VSM) and tension were performed in endothelium-denuded branches of mesenteric arteries from Wistar rats mounted in a microvascular myographs. Responses to CaCl<sup>2</sup> were assessed in arteries activated with phenylephrine (PE) and kept in Ca2+-free solution, in the absence and presence of the selective antagonist of L-type Ca2<sup>+</sup> channels nifedipine, cyclopiazonic acid (CPA) to block sarcoplasmic reticulum (SR) intracellular Ca2<sup>+</sup> release or specific inhibitors of PI3K, ERK-MAPK, or PKC. Activation of α1-adrenoceptors with PE stimulated both intracellular Ca2<sup>+</sup> mobilization and Ca2<sup>+</sup> entry along with contraction in resistance arteries. Both [Ca2+]<sup>i</sup> and contractile responses were inhibited by nifedipine while CPA abolished intracellular Ca2<sup>+</sup> mobilization and modestly reduced Ca2<sup>+</sup> entry suggesting that α1-adrenergic vasoconstriction is largely dependent Ca2<sup>+</sup> influx through L-type Ca2<sup>+</sup> channel and to a lesser extent through store-operated Ca2<sup>+</sup> channels. Inhibition of ERK-MAPK did not alter intracellular Ca2<sup>+</sup> mobilization but largely reduced L-type Ca2<sup>+</sup> entry elicited by PE without altering vasoconstriction. The PI3K blocker LY-294002 moderately reduced intracellular Ca2<sup>+</sup> release, Ca2<sup>+</sup> entry and contraction induced by the α1 adrenoceptor agonist, while PKC inhibition decreased PE-elicited Ca2<sup>+</sup> entry and to a lesser extent contraction without affecting intracellular Ca2<sup>+</sup> mobilization. Under

**65**

conditions of ryanodine receptor (RyR) blockade to inhibit Ca2+-induced Ca2+-release (CICR), inhibitors of PI3K, ERK-MAPK, or PKC significantly reduced [Ca2+]<sup>i</sup> increases but not contraction elicited by high K<sup>+</sup> depolarization suggesting an activation of L-type Ca2<sup>+</sup> entry in VSM independent of RyR. In summary, our results demonstrate that PI3K, ERK-MAPK, and PKC regulate Ca2<sup>+</sup> handling coupled to the α1-adrenoceptor in VSM of resistance arteries and related to both contractile and non-contractile functions. These kinases represent potential pharmacological targets in pathologies associated to vascular dysfunction and abnormal Ca2<sup>+</sup> handling such as obesity, hypertension and diabetes mellitus, in which these signaling pathways are profoundly impaired.

Keywords: ERK-MAPK, PI3K, PKC, L-type Ca2<sup>+</sup> channel, RyR, intracellular Ca2<sup>+</sup> mobilization, α1-adrenergic vasoconstriction, resistance arteries

#### INTRODUCTION

Insulin resistance plays a key role in the pathogenesis of type 2 diabetes and is also associated to other metabolic and cardiovascular abnormalities such as obesity, dyslipidemia and hypertension, jointly referred as to metabolic syndrome (Ford, 2005). Imbalance between insulin vascular actions via the phosphatidylinositol 3-Kinase (PI3K) and the mitogen activated protein kinase (MAPK) signaling pathways in insulin resistant states results in impaired endothelial vasodilator PI3K/eNOS/NO and augmented vasoconstrictor MAPK/endothelin 1 (ET1) pathways leading to endothelial dysfunction and exacerbated vasoconstriction (Kim et al., 2006; Prieto et al., 2013). However, altered Ca2<sup>+</sup> homeostasis in the arterial wall usually underlies abnormal vasoconstriction and the vascular complications associated to metabolic disease, such as hypertension and coronary artery disease (Okon et al., 2005; Ford et al., 2010; Villalba et al., 2011).

Vascular smooth muscle (VSM) contraction is triggered by the elevation of free intracellular Ca2<sup>+</sup> concentration [Ca2+]<sup>i</sup> due to extracellular Ca2<sup>+</sup> influx and/or Ca2<sup>+</sup> release from intracellular stores in the sarcoplasmic reticulum (SR), followed by Ca2+-calmodulin-dependent activation of myosin light chain (MLC) kinase (MLCK), MLC phosphorylation and actin/myosin crossbridges formation (Liu and Khalil, 2018). However, increases in force development at a given cytosolic Ca2<sup>+</sup> concentration can also occur and hence a dissociation between [Ca2+]<sup>i</sup> , MLC phosphorylation and vasoconstriction mediated by Ca2<sup>+</sup> sensitization mechanisms (Somlyo and Somlyo, 2003; Villalba et al., 2007). Ca2<sup>+</sup> release from SR intracellular stores, Ca2<sup>+</sup> entry through plasma membrane channels and Ca2<sup>+</sup> sensitization mechanisms can differentially contribute to VSM contraction depending on the vasoconstrictor agonist and/or vessel size (Nobe and Paul, 2001; Villalba et al., 2007; Kitazawa and Kitazawa, 2012). Furthermore, relative contribution of protein kinases to Ca2<sup>+</sup> handling coupled to receptor-mediated arterial vasoconstriction has also been reported to be size-dependent and thus, involvement of PKC and Rho kinase (RhoK) increase and decrease, respectively, with decreasing arterial size (Kitazawa and Kitazawa, 2012; Martinsen et al., 2012). Although kinases such as PKC and RhoK have traditionally been associated to Ca2<sup>+</sup> sensitization mechanisms involved in smooth muscle contraction (Nobe and Paul, 2001; Somlyo and Somlyo, 2003; Villalba et al., 2007), increasing experimental evidence supports a role for protein kinase-mediated regulation of intracellular Ca2<sup>+</sup> mobilization and Ca2<sup>+</sup> entry mechanisms in VSM and cardiac myocytes (Ghisdal et al., 2003; Villalba et al., 2008; Smani et al., 2010), and differences between large/conductance and small/resistance arteries concerning the role of various kinases in Ca2<sup>+</sup> handling have also been demonstrated (Kitazawa and Kitazawa, 2012; Martinsen et al., 2012).

Peripheral small arteries or resistance arteries, whose vasoconstrictor activity is under the sympathetic nervous control, play a key role in blood pressure regulation, hypertension being a common vascular complication in metabolic syndrome and insulin resistant states (Ford, 2005). Both impairment of the signaling pathways including MAPK, PI3K, and PKC in endothelial cells and altered Ca2<sup>+</sup> handling in VSM have been reported to underlie abnormal vasoconstriction in metabolic disease (Okon et al., 2005; Kim et al., 2006; Villalba et al., 2011; Prieto et al., 2013). Since the role of MAPK, PI3K, and PKC in Ca2<sup>+</sup> handling of resistance arteries is poorly understood, we assessed here whether these kinases are involved in Ca2<sup>+</sup> signaling pathways linked to adrenergic vasoconstriction in resistance arteries.

#### MATERIALS AND METHODS

#### Animal Model

Animal care and experimental protocols conformed to the European Directive for the Protection of Animals Used for Scientific Purposes (European Union Directive 2010/63/EU) and were also supervised by the Animal Care and Use Committee Complutense University of Madrid. Male Wistar rats were housed at the Pharmacy School animal care facility under controlled suitable environmental conditions of temperature (24◦C), lighting (12 h light/12 h dark cycle) and humidity (50– 60%), and maintained on standard chow and water ad libitum. They were anesthetized with sodium pentobarbital (50 mg/kg, i.p.) and euthanized by decapitation and exsanguination at 12 weeks age.

# Dissection and Mounting of Mesenteric Resistance Arteries

After animals were euthanized, the mesentery was quickly removed and placed on cold physiological saline solution (PSS) of the following composition (mM): NaCl 119, NaHCO<sup>3</sup> 25, KCl 4.7, KH2PO<sup>4</sup> 1.17, MgSO<sup>4</sup> 1.18, CaCl<sup>2</sup> 1.5, ethylenediaminetetraacetic acid 0.027 and glucose 11, continuously gassed with a mixture of 5% CO<sup>2</sup> /95% O<sup>2</sup> to maintain pH at 7.4. Mesenteric resistance arteries, third order branches of the superior mesenteric artery, were carefully dissected by removing the surrounding connective and fat tissue. Arterial segments were mounted in parallel in double microvascular myographs (Danish Myo Technology, DMT-Denmark) by inserting two 40 µm tungsten wires and equilibrated for 30 min in PSS at 37◦C. The relationship between passive wall tension and internal circumference was determined for each individual artery. The arteries were set to an internal circumference (L1) equal to 90% of that given by a transmural pressure of 100 mmHg for a relaxed vessel in situ, L<sup>100</sup> (L<sup>1</sup> = 0.9 × L100) at which tension development is maximal (Mulvany and Halpern, 1977). At the beginning of each experiment, arteries were stimulated twice with (KPSS), similar to PSS except that NaCl was substituted for KCl on an equimolar basis, in order to test vessel viability. The endothelium was mechanically removed by inserting a human hair in the vessel lumen and guiding it back and forwards several times. The absence of functional endothelium was confirmed by lack of the relaxation to acetylcholine (10 µM). Arteries were chemically denervated by incubation with guanethidine (10 µM) for 45 min to inhibit adrenergic nerve endings.

#### Simultaneous Measurements of Intracellular Ca2<sup>+</sup> ([Ca2+]<sup>i</sup> ) and Tension

Simultaneous measurements of the intracellular calcium concentration ([Ca2+]i) and tension were performed by FURA-2 AM fluorescence in mesenteric resistance arteries as previously reported (Villalba et al., 2007). Arteries were incubated in the dark with 8 µM Fura-2 AM in PSS for 2 h at 37◦C. The myograph chamber was mounted on a Zeiss inverted microscope equipped for dual excitation wavelength fluorimetry (Deltascan, Photon Technology). After loading, arteries were illuminated with alternating 340 and 380 nm light using a monochromator-based system (Deltascan, Photon Technology). Fluorescence emission was detected at 510 nm wavelength. The Ratio (R) F340/F380 was taken as a measure of [Ca2+]<sup>i</sup> . At the end of each experiment fluorescence not related to Ca2<sup>+</sup> was measured by bathing the artery in PSS containing 25 mM MnCl<sup>2</sup> plus ionomycin (10 µM) to quench Ca2+-insensitive signals and the values obtained were subtracted from those obtained throughout the experiment.

### Experimental Procedures for the Functional Experiments

The role of ERK-MAPK, PI3K, and PKC kinases in Ca2<sup>+</sup> handling of resistance arteries was assessed in endotheliumdenuded arteries kept in Ca2+-free medium. Arteries were exposed for 5 min to Ca2+-free PSS (0 mM Ca2+, 0.1 mM EGTA) to remove all extracellular Ca2<sup>+</sup> available for contraction. The myograph solution was then replaced by "nominally Ca2+-free PSS" (0 mM Ca2+, 0 mM EGTA) and concentration-response curves (CRCs) for CaCl<sup>2</sup> (10 µM-3 mM) were performed in arteries activated with phenylephrine (PE, 10 µM), in the absence (controls) and presence of the selective blocker of L-type Ca2<sup>+</sup> channels nifedipine (0.3 µM), the inhibitor of the Orai1−mediated Ca2<sup>+</sup> entry Pyr6 (3 µM) or the specific inhibitors of ERK-MAPK (PD-98059, 3 µM), p38MAPK (SB-203580, 0.3 µM), PI3K (LY-294002, 3 µM) or PKC (GF-109203X, 0.1 µM). The effect of SR Ca2<sup>+</sup> store depletion on Ca2<sup>+</sup> entry and contraction stimulated by PE was assessed in arteries kept in nominally Ca2+-free medium stimulated with 10 µM PE and then activated with a single Ca2<sup>+</sup> concentration (1 mM), before and after SR Ca2+ATPase (SERCA) inhibition with cyclopiazonic acid (CPA, 10 µM), and then treatment with CPA plus nifedipine. The combined effect of SR Ca2<sup>+</sup> store depletion with CPA (10 µM) and inhibition of Orai1-mediated Ca2<sup>+</sup> entry channels with Pyr6 (3 µM) was also examined on Ca2<sup>+</sup> entry and vasoconstriction of mesenteric arteries stimulated with PE (10 µM).

The inhibitory effect of PD-98059 (3 µM), LY-294002 (3 µM), or GF-109203X (0.1 µM) on Ca2<sup>+</sup> entry was also assessed in KPSS-depolarized arteries. After a first stimulation with KPSS, arteries were incubated with the inhibitors for at least 30 min before a second stimulation with KPSS was repeated. To evaluate the potential relationship between the PI3K, MAPK, and PKC pathways and the ryanodine receptor (RyR)-mediated Ca2+ induced Ca2+-release (CICR) mechanism in resistance arteries (Sánchez et al., 2018), RyR was blocked by incubation with 10 µM ryanodine for 25 min and then 1.5 mM CaCl<sup>2</sup> was added to arteries depolarized with Ca2+-free high K<sup>+</sup> solution (KPSS<sup>0</sup> 0). The effect of the selective inhibitors of ERK-MAPK, PI3K, or PKC was further assessed in arteries under conditions of RyR blockade.

#### Solutions and Drugs

Ca2+-free PSS and Ca2+-free KPSS solutions were similar to PSS and KPSS, respectively, except that CaCl<sup>2</sup> was replaced by 100 µM of EGTA, which was omitted when CaCl<sup>2</sup> was administered ("nominally Ca2+-free solution," 0 mM Ca2+, 0 mM EGTA). Acetylcholine, guanethidine, and phenylephrine were obtained from Sigma-Aldrich (Spain). All of them were dissolved in distilled water. Nifedipine, CPA, Pyr6 and kinase inhibitors (PD-98059, LY-294002, GF-109203X, and SB-203580) were obtained from Tocris Cookson (Bristol, United Kingdom). Stock solutions of Pyr6, PD-98059 and LY-294002 were made in distilled water, and those of CPA Pyr6, ryanodine, SB-203580, and GF-109203X in DMSO and further diluted in water. Nifedipine was initially dissolved in ethanol and further dilutions were made in distilled water.

#### Statistical Analysis

Results are expressed as either absolute values (units of R F340/F380 or Nm−<sup>1</sup> of active tension) or as a percentage of the response to KPSS in each artery, as means ± SEM of 6– 10 arteries (one artery from each animal). Arterial sensitivity to

agonists was expressed in terms of pEC50, that was the negative value of log EC50, EC<sup>50</sup> being the concentration of agonist giving 50% of the maximal response or effect (Emax). Statistically significant differences between means were analyzed by using paired or unpaired Student's t-test where appropriate, or one-way ANOVA followed by Bonferroni's post hoc test for comparisons involving more than two groups. Probability levels lower than 5% (P < 0.05) were considered statistically significant. Calculations were made using a standard software package (GraphPad Prism 5.0, GraphPad Software, Inc., San Diego, CA, United States).

#### RESULTS

# Ca2<sup>+</sup> Signaling Mechanisms Coupled to the α1-Adrenoceptor in Resistance Arteries

In order to assess the involvement of intracellular Ca2<sup>+</sup> mobilization and Ca2<sup>+</sup> entry mechanisms coupled to the α1-adrenoceptor in resistance arteries, endothelium-denuded mesenteric arteries were kept in a nominally Ca2+-free medium, stimulated with PE and further activated with increasing Ca2<sup>+</sup> concentrations (**Figure 1A**). PE induced an initial rapid increase in VSM [Ca2+]<sup>i</sup> and a simultaneous phasic contraction showing intracellular Ca2<sup>+</sup> mobilization (**Figures 1A,C**), and a further sustained elevation of [Ca2+]<sup>i</sup> along with vasoconstriction upon Ca2<sup>+</sup> re-addition, indicative of VSM Ca2<sup>+</sup> entry (**Figures 1A,B**). While there were no significant differences in the initial PE-induced [Ca2+]<sup>i</sup> increases and contraction corresponding to intracellular Ca2<sup>+</sup> mobilization (**Figure 1C**), PE-induced vasoconstriction upon Ca2<sup>+</sup> re-addition was larger than the simultaneous sustained [Ca2+]<sup>i</sup> increases (**Figure 1C**). Involvement of Ca2<sup>+</sup> sensitization in the α1-adrenoceptormediated vasoconstriction was further depicted by the steep slope of the [Ca2+]<sup>i</sup> -tension relationship for PE, showing that large contractions are developed without parallel increases in [Ca2+]<sup>i</sup> levels (**Figure 1D**).

Treatment with the blocker of L-type voltage-dependent Ca2<sup>+</sup> channels nifedipine largely inhibited the CaCl<sup>2</sup> CCR in arteries stimulated with PE (**Figure 2A**), while the inhibitor of the Orai1−mediated Ca2<sup>+</sup> entry Pyr6 only induced a moderate decrease of these responses (**Figure 2B**). Combined treatment with the SERCA inhibitor CPA to deplete SR Ca2<sup>+</sup> stores plus the inhibitor of store-operated Ca2<sup>+</sup> channels Pyr6 caused a larger inhibition of PE-induced vasoconstriction (**Figure 2B**). The effect of SR store depletion by treatment with CPA was further assessed on changes in [Ca2+]<sup>i</sup> and contraction elicited by PE, in arteries kept in nominally Ca2+-free medium and further stimulated with 1 mM Ca2<sup>+</sup> (**Figures 2C,D**). CPA inhibited PEinduced intracellular Ca2<sup>+</sup> mobilization and phasic contraction and reduced the sustained Ca2<sup>+</sup> entry and vasoconstriction elicited by the α1-adrenoceptor PE. The latter were abolished by combined treatment with CPA plus the blocker of the L-type voltage-dependent channels nifedipine (**Figures 2C,D**). These data demonstrate that α1-adrenergic vasoconstriction is largely due to Ca2<sup>+</sup> influx through L-type voltage-dependent Ca2<sup>+</sup> channels and to lesser extent to Ca2+release from the SR and store-operated Ca2<sup>+</sup> entry.

## PI3K Inhibition Decreased Intracellular Ca2<sup>+</sup> Mobilization and Ca2<sup>+</sup> Entry Induced by PE

Treatment with the PI3K inhibitor LY-294002 was used to evaluate whether PI3K is involved in Ca2<sup>+</sup> entry, Ca2<sup>+</sup> mobilization and/or Ca2<sup>+</sup> sensitization coupled to the α1 adrenoceptor in resistance arteries. This blocker moderately reduced both Ca2<sup>+</sup> entry and vasoconstriction induced by PE (**Figures 3A,B** and **Table 1**), without affecting [Ca2+]i-contraction relationships for this agonist which suggests no changes in Ca2<sup>+</sup> sensitization (**Figure 3C**). Interestingly, PI3K inhibition also reduced PE-induced intracellular Ca2<sup>+</sup> mobilization and the associated phasic contraction (**Figures 3D,E**).

# ERK-MAPK Inhibition Reduced Ca2<sup>+</sup> Entry but Not Vasoconstriction Coupled to the α1-Adrenoceptor

The effects of the inhibitor of ERK-MAPK PD-98059 on changes in VSM [Ca2+]<sup>i</sup> and contraction in mesenteric resistance arteries kept in nominally Ca2+-free medium and activated by PE (10 µM) before increasing CaCl<sup>2</sup> concentrations were added are shown in **Figure 4**. Treatment with PD-98059 largely reduced increases in [Ca2+]<sup>i</sup> elicited by PE upon Ca2<sup>+</sup> re-addition (**Figure 4A** and **Table 1**) without altering vasoconstriction (**Figure 4B** and **Table 1**). PE-induced contractions were not altered either by treatment with the p38MAPK inhibitor SB-203580 (0.3 µM) (**Supplementary Figure S1**). The relationship [Ca2+]i-contraction for PE was left-shifted upon ERK-MAPK kinase blockade in resistance arteries (**Figure 4C**), indicating decreased Ca2<sup>+</sup> sensitization under conditions of ERK-MAPK blockade and suggesting Ca2<sup>+</sup> entry through L-type channels not coupled to vasoconstriction and linked to ERK-MAPK kinase cascade in mesenteric resistance arteries. However, PD-98059 treatment did not affect PE-induced intracellular Ca2<sup>+</sup> mobilization and contraction associated to the α1-adrenergic stimulation (**Figures 4D,E**).

# PKC Inhibition Reduced Ca2<sup>+</sup> Entry and Contraction Elicited by PE

The PKC inhibitor GF-109203X was used to assess the involvement of PKC in Ca2<sup>+</sup> handling coupled to the α1 adrenoceptor in mesenteric resistance arteries. Treatment with GF-109203X reduced the increases in [Ca2+]<sup>i</sup> and to a minor extent vasoconstriction induced by Ca2<sup>+</sup> re-addition in arteries stimulated by PE kept in a nominally Ca2+-free medium (**Figures 5A,B** and **Table 1**). Both Ca2<sup>+</sup> sensitization (**Figure 5C**) and PE-induced intracellular Ca2+mobilization and phasic contraction (**Figures 5D,E**) remained unaffected by PKC inhibition.

FIGURE 1 | α1-Adrenoceptor activation involves intracelular Ca2<sup>+</sup> mobilization, Ca2<sup>+</sup> entry and Ca2<sup>+</sup> sensitization associated to contraction (A) Representative traces illustrating the changes in both [Ca2+]<sup>i</sup> mobilization (top) and vasoconstriction (bottom) induced by activation of α1-adrenoceptors by a single dose of Phenylephrine (PE) (10 µM) and further cumulative addition of CaCl<sup>2</sup> (closed bar) in endothelium-denuded mesenteric arteries kept in nominally Ca2+-free medium (0 mM Ca2+, open bar). (B,C) Summarized data showing changes in [Ca2+]<sup>i</sup> and vasoconstrictor responses stimulated by 10 µM PE addition in arteries kept in nominally Ca2+-free medium (B) and further cumulative re-addition of CaCl<sup>2</sup> (C). (D) Relationship between [Ca2+]i-contraction in response to cumulative addition of Ca2<sup>+</sup> in arteries kept in Ca2<sup>+</sup> free medium and stimulated with 10 µM PE. Responses are expressed as absolute values of either [Ca2+]<sup>i</sup> (1F340/F380) or tension (Nm−<sup>1</sup> ) (A) or relative to those elicited by KPSS (B–D). Values are means ± SEM of n = 7 arteries (one from each animal).

#### Inhibition of ERK-MAPK, PI3K, and PKC Pathways Reduced L-Type Ca2<sup>+</sup> Channel-Mediated [Ca2+]<sup>i</sup> Increases Independently of RyR

Since [Ca2+]<sup>i</sup> changes coupled to the α1-adrenoceptor in resistance arteries are largely due to Ca2<sup>+</sup> entry through L-type Ca2<sup>+</sup> channels, the effects of PD-98059, LY-294002, and GF-109203X were tested on the increase in [Ca2+]<sup>i</sup> elicited by high K <sup>+</sup> depolarization in order to assess whether the modulatory effect of ERK-MAPK, PI3K, and PKC on Ca2<sup>+</sup> entry elicited by PE is exerted through L-type Ca2<sup>+</sup> channels. Treatment with PD-98059, LY-294002, or GF-109203X reduced Ca2<sup>+</sup> entry stimulated by KPSS (**Supplementary Figures S2A–C**) suggesting that these kinases regulate L-type Ca2<sup>+</sup> channel entry in resistance arteries.

RyR-mediated Ca2+-induced Ca2+-release (CICR) upon L-type channel activation has recently been shown in VSM of mesenteric resistance arteries (Sánchez et al., 2018), and therefore we further assessed the potential relationship

between PI3K, PKC, and MAPK pathways and RyRmediated CICR mechanism. Treatment with ryanodine (10 µM) to selectively block the RyR (Meissner, 2017) and the subsequent SR Ca2<sup>+</sup> mobilization and amplification of Ca2<sup>+</sup> entry through L-type channels, reduced nifedipinesensitive increases in [Ca2+]<sup>i</sup> and contraction elicited by Ca2<sup>+</sup> readdition in high K+-depolarized endotheliumdenuded arteries (**Figures 6A,B**), thus confirming CICR upon L-type channel activation. On the other hand, in ryanodine-treated arteries to block CICR, selective inhibition of MAPK with PD-98059 (**Figures 6C,D**), PI3K with LY-294002 (**Figure 6E**) or PKC with GF-109203X (**Figure 6F**) resulted in a further significant reduction of the increases in [Ca2+]<sup>i</sup> but not contraction elicited by Ca2<sup>+</sup> re-addition in high K+-depolarized arteries, thus demonstrating a direct stimulatory effect of MAPK, PI3K, and PKC on L-type Ca2<sup>+</sup> entry independent of RyR mechanisms.

# DISCUSSION

An increasing body of experimental evidence during the last decade gives support to a key role for protein kinase-mediated regulation of Ca2<sup>+</sup> handling in arterial and cardiac myocytes, through phosphorylation of channels involved in either Ca2<sup>+</sup> entry through plasma membrane or Ca2<sup>+</sup> release from SR intracellular stores. The present study provides new insights into the modulation of Ca2<sup>+</sup> handling by PI3K, MAPK, and PKC signaling pathways coupled to the α1−adrenoceptor in resistance arteries and linked to both contractile and noncontractile functions. A graphical summary of our findings is depicted in **Figure 7**.

Hypertension and augmented vasoconstriction associated to metabolic disease have traditionally been ascribed to protein kinase-mediated Ca2<sup>+</sup> sensitization of the VSM contractile machinery leading to increased vascular tone and systemic vascular resistance (Martínez et al., 2000; Naik et al., 2006; Villalba et al., 2007, 2011; Crestani et al., 2017). However, Ca2<sup>+</sup> signaling mechanisms and the relative contribution of Ca2<sup>+</sup> sensitization and Ca2<sup>+</sup> mobilization mechanisms to arterial contraction are size-dependent. Thus, sensitivity of vasoconstriction and [Ca2+]<sup>i</sup> changes induced by α1 adrenoceptor activation to pharmacological inhibitors of voltage-dependent L-type Ca2<sup>+</sup> entry is higher in resistance arteries compared to large conductance arteries (Prieto et al., 1991; Kitazawa and Kitazawa, 2012; Martinsen et al., 2012). Moreover, contractions of large arteries have been reported to involve kinases such as RhoK and PKC to varying degrees, while vasoconstriction of resistance arteries seem to be mediated exclusively by PKC (Budzyn et al., 2006; Kitazawa and Kitazawa, 2012). The present results confirm early studies demonstrating the involvement of Ca2<sup>+</sup> sensitization mechanisms in the noradrenergic vasoconstriction of small mesenteric arteries (Buus et al., 1998; Kitazawa and Kitazawa, 2012), but further demonstrate that PI3K, MAPK, and PKC are involved in the regulation of Ca2<sup>+</sup> entry and intracellular Ca2<sup>+</sup> mobilization coupled to α1-adrenoceptor activation in resistance arteries.

The PI3K/Akt signaling pathway is stimulated upon activation of the insulin receptor in the vascular endothelium and coupled to eNOS phosphorylation, NO production and vasodilatation. This pathway is impaired and associated to endothelial dysfunction in insulin resistant states (Kim et al., 2006; Contreras et al., 2010). However, PI3Ks play also a main role in the Ca2<sup>+</sup> signaling of neurons and cardiac and vascular myocytes (Viard et al., 2004; Ghigo et al., 2017). In VSM, PI3Ks have been involved in the regulation of L-type Ca2<sup>+</sup> entry in response to vasoconstrictors and growth factors (Macrez et al., 2001; Le Blanc et al., 2004) but also in the Ca2+-dependent Rho-mediated negative control of MLCP linked to Ca2<sup>+</sup> sensitization and contraction (Wang et al., 2006). In the present study, pharmacological inhibition of PI3K moderately reduced both Ca2<sup>+</sup> entry and vasoconstriction elicited by PE in resistance arteries, without affecting the relationship Ca2+-tension for the α1-adrenergic agonist indicative of changes in Ca2<sup>+</sup> sensitization. These findings are consistent with earlier studies

involving PI3K activation in the transduction pathways for the angiotensin II (AII)-induced Ca2<sup>+</sup> responses and contraction in VSM (Le Blanc et al., 2004; Rakotoarisoa et al., 2006). In vascular myocytes, PI3Kγ isoform was initially shown to mediate AII-induced activation of L-type voltage-dependent Ca2<sup>+</sup> currents, to increase [Ca2+]<sup>i</sup> and elicit contraction (Quignard et al., 2001; Le Blanc et al., 2004; Rakotoarisoa et al., 2006). This is confirmed in the endothelium-denuded intact resistance arteries of the present study by the inhibitory effect of the PI3K blocker LY294002 on L-type Ca2<sup>+</sup> entry elicited by high K<sup>+</sup> depolarization. PI3Kγ/Akt phosphorylates Cavβ<sup>2</sup> and induces Cavα1C translocation thus increasing L-type Ca2<sup>+</sup> currents in arterial myocytes, and pharmacological inhibition PI3Kγ has vasodilator effects and reduces arterial blood pressure (Carnevale and Lembo, 2012). However, our results further demonstrate that PI3K inhibition markedly reduced intracellular Ca2<sup>+</sup> release and


TABLE 1 | Effects of the inhibitors of PI3K LY-294002 (3 µM), ERK-MAPK kinase PD-98059 (3 µM) and PKC GF-109203X (0.1 µM) on the sensitivity and maximal responses of the CaCl<sup>2</sup> concentration-response curves in mesenteric arteries stimulated by PE (10 µM) in a Ca2+-free medium.

Data are means ± SEM; "n" number of arteries (one per animal) from Wistar rats. Results are expressed as absolute values, as the increases in the intracellular Ca2<sup>+</sup> concentration [Ca2+]<sup>i</sup> (ratio F340/F380) and tension (Nm−<sup>1</sup> ) elicited by PE or CaCl2. pEC<sup>50</sup> is – log(EC50), EC<sup>50</sup> being the concentration of agonist giving 50% of the maximum effect (Emax). Significant differences were analyzed by paired Student's t-test. <sup>∗</sup>P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001 vs. control.

percentage of control maximal responses. Values are means ± SEM of n = 7 arteries (one from each animal). <sup>∗</sup>P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001 vs. control.

phasic contraction induced by PE in resistance arteries, which suggests that PI3K might also activate voltage-independent store-operated Ca2<sup>+</sup> entry through stimulation of intracellular Ca2<sup>+</sup> release. This pathway is linked to the α1-adrenoceptor in resistance arteries, as depicted by the inhibitory effect on Ca2<sup>+</sup> entry and contraction induced by SR depletion by blockade of SERCA with CPA, or by treatment with the store-operated Ca2<sup>+</sup> entry blocker Pyr6 (Santiago et al., 2015).

α1-Adrenoceptor stimulation causes intracellular Ca2<sup>+</sup> mobilization via IP<sup>3</sup> receptor-mediated Ca2<sup>+</sup> release from SR stores (Liu and Khalil, 2018). Therefore, it seems likely that the reduction of PE-induced intracellular Ca2<sup>+</sup> mobilization by the PI3K blocker LY294002 to be due to the inhibition of PI3K-stimulated IP3-dependent intracellular Ca2<sup>+</sup> release, as earlier reported for the IGF-1 receptor linked to a G protein-PI3K-phospholipase C signaling pathway in cardiac myocytes

vasoconstriction (bottom) induced by CaCl<sup>2</sup> (1.5 mM) re-addition in endothelium-denuded mesenteric arteries depolarized with high K<sup>+</sup> in Ca2<sup>+</sup> free medium (KPSS<sup>0</sup> <sup>0</sup>). (B) Summarized data showing the average effects of ryanodine (10 µM) on changes in [Ca2+]<sup>i</sup> and contraction elicited by CaCl<sup>2</sup> (1.5 mM) re-addition in arteries depolarized with a high K<sup>+</sup> solution. (C) Representative traces showing the effects of inhibition of MAPK with PD-98059 (3 µM) on changes in both [Ca2+]<sup>i</sup> (top) and vasoconstriction (bottom) induced by CaCl<sup>2</sup> (1.5 mM) re-addition in endothelium-denuded mesenteric arteries depolarized with high K<sup>+</sup> (KPSS<sup>0</sup> <sup>0</sup>) and treated with ryanodine (30 µM) to block RyR. (D) Summarized data showing the average effects of PD-98059 (3 µM) on changes in [Ca2+]<sup>i</sup> and vasoconstriction elicited by CaCl<sup>2</sup> (1.5 mM) re-addition in arteries depolarized with a high K<sup>+</sup> solution under conditions of RyR blockade with ryanodine. Average effects of the PI3K inhibitor LY-294002 (3 µM) (E) or the PKC inhibitor GF-109203X (0.1 µM) (F) on changes in [Ca2+]<sup>i</sup> and vasoconstriction elicited by CaCl<sup>2</sup> (1.5 mM) re-addition in arteries depolarized with a high K<sup>+</sup> solution under conditions of RyR blockade. Responses are expressed as absolute values of either [Ca2+]<sup>i</sup> (1F340/F380) or tension (Nm−<sup>1</sup> ) (A,C) or relative to those elicited by KPSS (B,D–F). Values are means ± SEM of n = 5 arteries (one from each animal). Significant differences were analyzed by paired Student's t-test or one-way ANOVA. <sup>∗</sup>P < 0.05 vs. control. ∗∗∗P < 0.001 vs. control. †††P < 0.001 vs. ryanodine-treated.

(Ibarra et al., 2004). However, we have recently demonstrated RyR-mediated CICR upon activation of L-type channels as a potent amplifying mechanism of Ca2<sup>+</sup> entry and contraction in mesenteric resistance arteries (Sánchez et al., 2018). Since RyR1 and RyR2 can be phosphorylated by kinases such as PKA leading to SR Ca2<sup>+</sup> leak (Bovo et al., 2017; Meissner, 2017), we further investigated whether PI3K may regulate RyR-mediated CICR stimulated by L-type Ca2<sup>+</sup> channel activation. Under conditions of RyR blockade with ryanodine, selective PI3K inhibition with LY294002 significantly reduced increases [Ca2+]<sup>i</sup> but not contraction elicited by high K<sup>+</sup> depolarization. Although RyR phosphorylation cannot be ruled out, the present finding suggest that PI3K activation mainly stimulates Ca2<sup>+</sup> entry through direct regulation of L-type Ca2<sup>+</sup> channels in resistance arteries, as discussed above and reported for isolated vascular myocytes (Le Blanc et al., 2004).

Enhanced activity of PI3K and up-regulation of PI3Kδ have been found to be associated to augmented L-type Ca2<sup>+</sup> entry in arterial myocytes from rat models of type I diabetes (Pinho et al., 2010) and insulin resistance (Sánchez et al., 2018). While in the former PI3K contributed to enhanced vasoconstriction, in the latter PI3K-mediated increased Ca2<sup>+</sup> entry compensated for SR Ca2<sup>+</sup> store dysfunction. In the heart, PI3Kα/Akt signaling is involved in insulin inotropic actions and activates Ca2<sup>+</sup> currents in microdomains containing L-type Ca2<sup>+</sup> channels (Lu et al., 2009). This pathway and the corresponding L-type Ca2<sup>+</sup> entry is defective in insulindeficient and resistant states which might contribute to the cardiac contractile dysfunction in diabetic cardiomyopathy (Lu et al., 2007; Ghigo et al., 2017).

On the other hand, ERK1/2 MAPK pathway is involved in insulin mitogenic actions but has also been associated to insulin-mediated enhanced vasoconstriction in insulin resistant states (Kim et al., 2006; Contreras et al., 2010; Prieto et al., 2013). In the present study, the selective inhibitor of ERK-MAPK PD98059 did not alter the initial rapid Ca2<sup>+</sup> increase in response to PE corresponding to intracellular Ca2<sup>+</sup> mobilization, but caused a profound inhibition of the sustained Ca2<sup>+</sup> entry stimulated by PE without altering the associated contraction in resistance arteries. These findings differ from earlier studies involving ERK-MAPK in both Ca2+-dependent vasoconstriction coupled to the AII receptor in human resistance arteries (Touyz et al., 1999), and in Ca2<sup>+</sup> sensitization-mediated contraction of coronary artery myocytes linked to the ET1 receptor (Cain et al., 2002). Moreover, the present results differ from reports involving the p38 and JNK limbs of the MAPK pathway in nifedipinesensitive vasoconstriction linked to α-adrenoceptors in VSM of large arteries (Ok et al., 2011). In contrast, our results suggest a major role of MAPK in the regulation of L-type Ca2<sup>+</sup> entry coupled to the α1-adrenergic adrenoceptor in resistance arteries, supported by the marked reduction elicited by PD98059 on the high K<sup>+</sup> depolarization-induced Ca2<sup>+</sup> entry under conditions of RyR receptor blockade. These findings are in agreement with reports showing that ERK MAPK phosphorylates Cav1.2 channels and enhances L-type currents in cardiac myocytes in response to growth factors involved in cardiac hypertrophy (Takahashi et al., 2004). Differences in the involvement of the ERK-MAPK in Ca2<sup>+</sup> handling and contraction in vascular myocytes may be ascribed to the differential activation of this pathway by various receptors. Our results suggest that ERK-MAPK-mediated modulation of Ca2<sup>+</sup> handling might be related to VSM non-contractile proliferative pathways coupled to the α1 adrenoceptor in resistance arteries (Kudryavtseva et al., 2013). Further studies are needed to elucidate the role of receptorcoupled MAPK pathways as modulators of both L-type and non-L-type Ca2<sup>+</sup> influx involved in cell proliferation.

In cardiac myocytes, ERK MAPK phosphorylation of Cav1.2 channels and increased L-type currents are activated upstream by PKC linked to activation of G protein-coupled receptors (Smani et al., 2010). The role of PKC in arterial Ca2<sup>+</sup> handling and vasoconstriction is well-documented, although this kinase has mostly been involved in Ca2<sup>+</sup> sensitization mechanisms coupled to the α1-adrenoceptor in resistance arteries (Buus et al., 1998; Villalba et al., 2007; Kitazawa and Kitazawa, 2012). PKC phosphorylates the phosphoprotein CPI-17, a potent inhibitor of MLCP, rendering it inactive and therefore promoting increased vascular tone (Somlyo and Somlyo, 2003). However, in our study, the PKC inhibitor GF-109203X did not alter the initial Ca2<sup>+</sup> rise but markedly reduced the sustained [Ca2+]<sup>i</sup> increase in response to PE, and to a lesser extent the simultaneous vasoconstriction of mesenteric resistance arteries, which suggests a major role for PKC in the regulation of Ca2<sup>+</sup> entry rather than in the enhancement of myofilament Ca2<sup>+</sup> sensitivity of arterial myocytes, as also supported by the unchanged Ca2+-tension relationship for the α1-adrenergic agonist in presence of the

PKC inhibitor. Involvement of PKC in Ca2<sup>+</sup> entry in resistance arteries partially agrees with that recently reported by Kitazawa and Kitazawa (2012). However, these authors showed that GF-109203X induced a small reduction of PE-induced Ca2<sup>+</sup> entry, while it abolished vasoconstriction and reduced phosphorylation of MLC, CPI-17 and MYPT1, supporting a major involvement of PKC in Ca2<sup>+</sup> sensitization. The discrepancies between our data showing a minor inhibitory effect of GF-109203X on Ca2<sup>+</sup> sensitization coupled to PE vasoconstriction and those in the study by Kitazawa and Kitazawa (2012) might be ascribed to the fact that different PKC isoforms mediate Ca2<sup>+</sup> sensitization and Ca2<sup>+</sup> entry pathways in arterial smooth muscle (Liu and Khalil, 2018). Thus, the conventional Ca2+-dependent PKC isoforms (α, -β1, -β2, and -γ) are activated by cytosolic Ca2<sup>+</sup> and diacylglycerol, and in turn phosphorylate a wide array of substrates implicated in the regulation of Ca2<sup>+</sup> fluxes; specifically, PKCα has been involved in the regulation of L-type Ca2<sup>+</sup> entry in both vascular and cardiac myocytes (Yang et al., 2009; Gulia et al., 2013). In contrast, the novel PKC isoforms appear to mediate vasoconstriction coupled to Ca2<sup>+</sup> sensitizing pathways (Liu and Khalil, 2018). GF-109203X is selective for the conventional PKCα and β1 isoforms but it may also inhibit the novel δ and ε PKC isoforms at the higher micromolar range used in the study of Kitazawa and Kitazawa (2012) and Liu and Khalil (2018).

In the present study, involvement of PKC in the sustained Ca2<sup>+</sup> entry induced by PE and sensitive to L-type Ca2<sup>+</sup> channel blockade is further supported by the marked inhibitory effect found for GF-109203X on the L-type channel-mediated Ca2<sup>+</sup> entry elicited by high K<sup>+</sup> depolarization, thus suggesting that PKC modulates voltage-dependent L-type Ca2<sup>+</sup> entry coupled to the α1-adrenergic vasoconstriction in resistance arteries, and supporting that PKC-mediated modulation of L-type channels in arterial myocytes (Gulia et al., 2013) contributes not only to myogenic tone but also to agonist-induced vasoconstriction (Cobine et al., 2007; Potts et al., 2012). Activity of both conventional and novel PKC isoforms is chronically enhanced by hyperglycemia, lipotoxicity and oxidative stress, which has been associated to the cardiovascular complications in the insulin resistant states (Geraldes and King, 2010; Turban and Hajduch, 2011). Abnormal PKC activity contributes to the augmented arterial L-type Ca2<sup>+</sup> entry and enhanced vasoconstriction in diabetic arteries (Ungvari et al., 1999) and arteries from genetically obese rats (Sánchez et al., 2018), and in the latter compensates for SR Ca2<sup>+</sup> store dysfunction.

#### CONCLUSION

The present findings demonstrate that PI3K, MAPK, and PKC are involved in the regulation of Ca2<sup>+</sup> entry and intracellular Ca2<sup>+</sup> mobilization coupled to α1-adrenoceptor activation in resistance

#### REFERENCES

Bovo, E., Huke, S., Blatter, L. A., and Zima, A. V. (2017). The effect of PKA-mediated phosphorylation of ryanodine receptor on SR Ca2+ leak in arteries and mostly associated to non-contractile functions of VSM. Under conditions of vascular disease, vascular myocytes change from a contractile to a synthetic phenotype that is able to proliferate and migrate. The present results showing a protein kinase-mediated regulation of Ca2<sup>+</sup> handling linked to the α1 adrenoceptor in resistance arteries suggest that changes in PI3K, MAPK, and PKC signaling pathways involving enhanced Ca2<sup>+</sup> mobilization not coupled to contraction, might participate in the changes toward a VSM proliferative phenotype and be involved in vascular remodeling in hypertension and other insulin resistant states. Further studies are needed to elucidate this issue.

#### AUTHOR CONTRIBUTIONS

DP conceived, designed, and discussed the experiments. AG and AS performed the experiments and analyzed the data. AG, AS, and DP wrote the manuscript. AG, CC, and AS contributed to discussion, and edited the manuscript.

#### FUNDING

This work was supported by grants SAF2012-31631 and SAF2016-77526R from MINECO (Spain) cofounded by the FEDER Program of EU.

#### ACKNOWLEDGMENTS

Manuel Perales and Francisco Puente are thanked for expert technical assistance.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2019.00055/full#supplementary-material

FIGURE S1 | p38MAPK inhibition does not alter PE-induced vasososntriction. Average effects of the p38MAPK inhibitor SB-203580 (0.3 µM) on the contractions elicited by cumulative addition of CaCl<sup>2</sup> in endothelium-denuded arteries kept in a nominally Ca2+-free medium and stimulated by 10 µM PE. Results are expressed as a percentage of control maximal responses. Values means ± SEM of n = 6 arteries (two from each animal).

FIGURE S2 | ERK-MAPK, PI3K, or PKC kinase inhibitors reduce Ca2<sup>+</sup> entry through voltage-dependent L-type channels. Average inhibitory effects ERK-MAPK inhibitor PD-98059 (3 µM) (A), the PI3K inhibitor LY-294002 (3 µM) (B) or the PKC inhibitor GF-109203X (0.1 µM) (C), on the increases in [Ca2+]<sup>i</sup> elicited by depolarization high K<sup>+</sup> (KPSS). Results are expressed as absolute values of [Ca2+]<sup>i</sup> (1F340/F380). Values are means ± SEM of five arteries (one from each animal). Significant differences were analyzed by paired Student's t-test; <sup>∗</sup>P < 0.05 vs. control.

ventricular myocytes. J. Mol. Cell. Cardiol. 104, 9–16. doi: 10.1016/j.yjmcc.2017. 01.015

Budzyn, K., Paull, M., Marley, P. D., and Sobey, C. G. (2006). Segmental differences in the roles of rho-kinase and protein kinase C in mediating

vasoconstriction. J. Pharmacol. Exp. Ther. 317, 791–796. doi: 10.1124/jpet.105. 100040


dysfunction. Circulation 120, 318–325. doi: 10.1161/CIRCULATIONAHA.109. 873380



rats: involvement of rho kinase. Am. J. Physiol. Heart Circ. Physiol. 300, H2044–H2053. doi: 10.1152/ajpheart.01204.2010


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Gutiérrez, Contreras, Sánchez and Prieto. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Gender-Dependent Alteration of Ca2+ and TNF**α** Signaling in *db*/*db* Mice, an Obesity-Linked Type 2 Diabetic Model

#### *Edited by:*

*José Antonio Pariente, Universidad de Extremadura, Spain*

#### *Reviewed by:*

*Juan Martinez-Pinna, University of Alicante, Spain Gines M. Salido, Universidad de Extremadura, Spain Belma Turan, Ankara University, Turkey*

#### *\*Correspondence:*

*Carmen Delgado cdelgado@iib.uam.es Ana-Maria Gomez ana-maria.gomez@inserm.fr*

*† These authors have contributed equally to this work*

#### *Specialty section:*

*This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology*

*Received: 04 November 2018 Accepted: 14 January 2019 Published: 07 February 2019*

#### *Citation:*

*Delgado C, Gomez A-M, Samia El Hayek M, Ruiz-Hurtado G and Pereira L (2019) Gender-Dependent Alteration of Ca2+ and TNFα Signaling in db/db Mice, an Obesity-Linked Type 2 Diabetic Model. Front. Physiol. 10:40. doi: 10.3389/fphys.2019.00040*

*Carmen Delgado1 \*, Ana-Maria Gomez2 \*, Magali Samia El Hayek2 , Gema Ruiz-Hurtado3† and Laetitia Pereira2†*

*1Instituto de Investigaciones Biomédicas "Alberto Sols" (CSIC-UAM)/CIBER-CV, Madrid, Spain, 2INSERM UMR-S 1180, University of Paris-Sud, University of Paris-Saclay, Châtenay-Malabry, France, 3Cardiorenal Translational Laboratory, Institute of Research i+12, Hospital Universitario 12 de Octubre/CIBER-CV, Madrid, Spain*

Cardiovascular complications are the primary death cause in type 2 diabetes, where inflammation can play a role. We, and others, have previously shown that, in diabetic cardiomyopathy, cardiac dysfunction is associated with Ca2+ mishandling. It is possible that diabetic cardiomyopathy differently affects men and women, as the latter present higher risk to develop heart failure and a higher plasmatic level of the pro-inflammatory cytokine, tumor necrosis factor alpha (TNFα), than men. However, the gender-dependent regulation of Ca2+ signaling in diabetes and its relationship with TNFα signaling are still unclear. Here, we analyzed TNFα signaling pathway and its role in Ca2+ signaling dysfunction in male and female rodent models of type 2 diabetes linked to obesity (*db*/*db* mice) using confocal microscopy in freshly isolated cardiomyocytes. TNFα increased [Ca2+]i transient amplitude and accelerated its decay without affecting SR Ca2+ load or Ca2+ spark frequency in cells from control mice. All TNFα effects on Ca2+ handling were prevented by the inhibition of the ceramidase and the phospholipase A2 (PLA2). While the plasmatic level of TNFα was similar in male and female *db*/*db* mice, only male *db*/*db* hearts over-expressed both TNFα converting enzyme (TACE) and the protective TNFα receptors 2 (TNF-R2). TNFα receptor 1 (TNF-R1) expression, involved in negative inotropic response of TNFα, was unchanged in both male and female *db*/*db* mice compared to controls. We found that male *db*/*db* mice cardiomyocytes presented a decrease in [Ca2+]i transient amplitude associated to a drop of sarcoplasmic reticulum Ca2+ load, not seen in female *db*/*db* mice. Interestingly, sustained incubation with TNFα did not restored Ca2+ signaling alteration observed in male *db*/*db* mice but still induces an increase in Ca2+ spark frequency as seen in control littermates. In cardiomyocytes from female *db*/*db* mice, TNFα had no visible effects on Ca2+ handling. In conclusion, our study shows that the alteration of Ca2+ signaling and TNFα, seen in *db*/*db* mice, is gender specific presenting an increase in TNFα cardio-protective pathway in male mice.

Keywords: diabetic cardiomyopathy, TNF**α**, calcium, gender difference, *db*/*db* mice

# INTRODUCTION

Cardiovascular complications, such as coronary artery diseases, hypertension, and heart failure, are a leading cause of death in type 2 diabetes (Laakso, 1999; Bauters et al., 2003; Bell, 2007). Preclinical studies have shown that diabetic cardiac dysfunction, with depressed contraction and relaxation, results from dysregulation of metabolism, mitochondrial function, oxidative stress, and Ca2+ handling (Bugger and Abel, 2014). These knowledge result almost exclusively from male animal studies. However, in the clinical setting, the risk for developing cardiac diseases in diabetes is known to be gender specific (Galderisi et al., 1991; Rutter et al., 2003; Toedebusch et al., 2018). Indeed, the Framingham Heart Study showed that diabetic women present a 5.1-fold increased risk to develop heart failure than non-diabetic patients, whereas in diabetic men, this risk is only multiplied by 2.4 (Galderisi et al., 1991; Rutter et al., 2003). In addition, the hospital admission rate for cardiovascular diseases is higher in diabetic women compared to diabetic men. Yet, the gender differences in the alterations of cardiac cellular function in diabetes are unclear, notably regarding Ca2+ mishandling.

Ca2+ regulates contraction through the excitation-contraction coupling in cardiomyocytes. For each heartbeat, sarcolemmal L type Ca2+ channels open during the action potential, leading to Ca2+ influx that activates Ca2+ release from the ryanodine receptors (RyR) located at the sarcoplasmic reticulum (SR). This release of Ca2+ by the RyR (visualized as a [Ca2+]i transient) activates contractile myofibrils to generate cardiomyocyte contraction. After the contraction, the Ca2+ is re-uptaken into the SR by the SERCA pump and extruded outside the cardiomyocytes mainly by the Na+ /Ca2+ exchanger, resulting in cardiomyocyte relaxation. We and others have shown that, in animal models of type 2 diabetes linked to obesity, contractile dysfunction is associated with a decrease in the Ca2+ transient amplitude. This lower Ca2+ transient amplitude is associated to reduced L-type Ca2+ current density combined with downregulation of RyR expression (Belke et al., 2004; Pereira et al., 2006b, 2014). We found that these alterations may be different in male and female *db*/*db* mice (Pereira et al., 2014); however, the mechanisms remain unclear.

Clinical and preclinical studies pointed out an increase in plasmatic level of TNFα, in type 2 diabetes, notably in women (Yamakawa et al., 1995; Pereira et al., 2006a; Preciado-Puga et al., 2014). TNFα is an inflammatory cytokine commonly associated to infectious and non-infectious cardiomyopathy, such as viral myocarditis, congestive heart failure, and myocardial infarction. The level of TNFα seems correlated to the development of cardiac dysfunction (Feldman et al., 2000; Blum and Miller, 2001), and its over-expression leads to cardiac hypertrophy, fibrosis, arrhythmia, and dysfunction (Kubota et al., 1997; Kadokami et al., 2000; London et al., 2003). Yet, whether TNFα is a cause or a consequence of cardiac dysfunction is still under debate. The biological response of TNFα is mediated through two receptors, the TNFα receptor 1 (TNF-R1) and TNFα receptor 2 (TNF-R2). TNF-R1 activation is responsible for a cardiac negative inotropic response, whereas TNF-R2 mediates cardiac positive inotropic response (Meldrum, 1998). At the cellular level, TNFα regulates contraction either by direct regulation of Ca2+ signaling in acute condition or *via* iNOS activation in sustained conditions (Fernandez-Velasco et al., 2007). Still, whether TNFα activation positively or negatively alters the Ca2+ transient is quite controversial, and studies found either a decrease, an increase, or no effect on Ca2+ transient. Those discrepancies seem to depend on the animal model, the concentration of TNFα used, and the incubation time (Yokoyama et al., 1993; Goldhaber et al., 1996; Bick et al., 1997; Sugishita et al., 1999; Li et al., 2003; Zhang et al., 2005; Duncan et al., 2010; Greensmith and Nirmalan, 2013). In addition, whether the regulation of TNFα signaling in type 2 diabetic cardiomyopathy linked to obesity is gender specific remains unknown.

Considering all these controversial findings surrounding TNFα regulation of Ca2+ handling, we first studied the effect of TNFα on Ca2+ signaling in WT mice. Then, using the *db*/*db* mice, an animal model of type 2 diabetes with insulin resistance linked to obesity, we found that both Ca2+ and TNFα signaling underwent distinct alterations in male compared to female. Here, we found that male *db*/*db* mice presented a depressed Ca2+ transient associated with a lower SR Ca2+ load, not seen in female *db*/*db* mice. More interestingly, in male *db*/*db*, cardiomyocytes seem to put in place a protective mechanism to counteract those alterations by increasing the expression of cardio-protective TNF-R2 signaling pathway.

# MATERIALS AND METHODS

#### Cell Isolation

Experiments were carried out according to the ethical principles of the French Ministry of Agriculture and the European Parliament on the protection of animals. Ventricular adult cardiomyocytes were isolated from 8 weeks old male C56Bl6 mice, male and female 15 weeks old *db*/*db* (Janvier), and their control littermates (*db*/*+*). Mice were euthanized by intraperitoneal injection of sodium pentobarbital (100 mg/kg). Cardiac ventricular myocyte isolation was performed by standard enzymatic methods (collagenase type II, Worthington) using the Langendorff perfusion as previously described (Pereira et al., 2006b, 2007, 2012; Leroy et al., 2011; Ruiz-Hurtado et al., 2015). After isolation, cells were kept in 1 mM [Ca2+] for an hour prior experiments. Only rod-shaped cells and quiescent cells when unstimulated and excitable were used for the Ca2+ experiments.

**Abbreviations:** ATK, arachidonyl trifluoromethyl ketone; TNFα, tumor necrosis factor alpha; TNF-R1, TNFα receptor 1; TNF-R2, TNFα receptor 2; KO, knock-out; NO, nitric oxide; NOE, n-oleoylethanolamine; o.i., oil immersion; PKA, protein kinase A; PLA2, phospholipase A2; RyR, cardiac ryanodine receptor; SR, sarcoplasmic reticulum; SERCA, sarco/endoplasmic reticulum Ca2+-ATPase; TACE, TNFα converting enzyme.

# Measurements of Plasmatic TNF**α**

TNFα determination by ELISA Soluble TNFα concentration was determined in plasma samples from mice using commercial ELISA test (BIOTRAK, Amersham Life Science, Sweden).

## Confocal Microscopy

Ca2+ handling was recorded in freshly isolated ventricular adult cardiomyocytes loaded with the fluorescent Ca2+ dye, the Fluo-3 acetoxymethyl ester (Fluo-3 AM, Molecular Probes) at 5 μM diluted in a mixture of DMSO-pluronic acid 20%. A line scan across the longitudinal axis of the myocyte was performed to measure cardiomyocyte shortening. Cardiomyocyte shortening corresponds to the difference between cardiomyocyte length at rest and cardiomyocyte length during contraction (during electrical stimulation), as previously described (Fernandez-Velasco et al., 2009). Ca2+ transient, Ca2+ sparks, and SR Ca2+ load were recorded using confocal microscopy (Meta Zeiss LSM 510, objective w.i. 63×, n.a. 1.2) in line scan mode (1.54 ms) along the longitudinal axis of the cell. Ca2+ transients were evoked by field stimulation (1 Hz) applied through two parallel platinum electrodes. Spontaneous Ca2+ sparks were recorded in quiescent cells after Ca2+ transient recording. Ca2+ transient decay time corresponds to the kinetic of the relaxation phase due to the re-uptake of Ca2+ into the SR by the SERCA pump as well as the extrusion of Ca2+ by the Na2+/Ca2+ exchanger. Ca2+ transient decay time is calculated using a monoexponential function to fit the Ca2+ transient decline phase. SR Ca2+ load was assessed by rapid caffeine application (10 mM) after 1 min pacing to reach the steady state. Parameters were studied with or without TNFα (1 h to 1 h 30 min) supplemented or not with a ceramidase inhibitor n-oleoylethanolamine (NOE, 5 μM) and a phospholipase A2 (PLA2) inhibitor (ATK, 10 μM) (Sigma-Aldrich). Fluo-3 AM was excited with an Argon laser (λex = 488 nm), and emission was collected at wavelengths >505 nm. Image analysis was performed using homemade routines in interactive data language (IDL).

#### Western-Blot Analysis

Adult ventricular homogenates were quickly frozen in liquid nitrogen and then placed in Tris solution (50 mmol/L, pH = 7.4) containing proteases and phosphatase inhibitors (10 μg/ml leupeptin, 10 μg/ml trypsin inhibitor, 2 μg/ml aprotinin, and 5 μM okadaic acid). Homogenization was performed on ice using a Politron. Homogenate was centrifuged at 18,925 g for 10 min at 4°C. Proteins were resuspended in Laemmli (5%) sample buffer, boiled (90°C for 5–10 min), and separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) using 10% polyacrylamide gels. After separation, proteins were transferred to polyvinylidene fluoride membranes (Amersham Biosciences), and non-specific binding sites were blocked overnight at 4°C in 5% dried milk and Tris-buffer saline (TBS, pH = 7.4) and 0.01% Tween 20. Membranes were incubated overnight (at 4°C) for the rabbit polyclonal anti-TACE (1:300; Proscience) and the rabbit polyclonal anti-TNFR2 (H-202) (1:250; Santa Cruz), at room temperature for 1 h 30 min for the rabbit polyclonal anti-TNFR1 (H-271) (1:500; Santa Cruz). A secondary horseradish peroxidaseconjugated goat anti-rabbit IgG (Amersham Biosciences) was used in combination with an enhanced chemiluminescence detection system (SuperSignal West Pico Chemiluminescent Substrate, Pierce) to visualize the primary antibodies. Band densities were determined with a laser-scanning densitometer (HP-3970) and Quantity One software (BioRad SA). Protein loading was controlled by probing all Western blots with anti-GADPH antibody (1:4,000) (Ambion).

#### Statistical Analysis

Results were expressed as mean ± SEM. Significance between two groups was determined using unpaired Student's *t* test or non-parametric Mann-Whitney test. Data involving more than two groups were analyzed using either one-way ANOVA or two-way ANOVA as appropriate. We used GraphPad Prism 7 (GraphPad) for statistical comparison. Differences with values of *p*<0.05 were considered significant.

# RESULTS

#### Sustained TNF**α** Exposure Increases Ca2+-Induced Ca2+ Release

TNFα-mediated Ca2+ signaling regulation is quite controversial, which is probably due to protocol differences. Therefore, we first studied, in our experimental settings, the effect of sustained activation (1–1 h 30 min) of TNFα on Ca2+ handling parameters such as Ca2+ transient, Ca2+ spark frequency, and SR Ca2+ load (**Figure 1**). In our hands, 10 and 50 ng/ml TNFα treatment significantly increased Ca2+ transient amplitude (F/F0 of 3.1 ± 0.3 for 10 ng/ml, 3.5 ± 0.3 for 50 ng/ml vs. 2.5 ± 0.14 for baseline, *p* < 0.05). Moreover, TNFα significantly accelerated the Ca2+ re-uptake into the SR as shown by the faster SR Ca2+ transient decay time (**Figures 1A,B**) (~29% faster for 10 ng/ml and ~25% for 50 ng/ml, *p* < 0.01). This acceleration of Ca2+ re-uptake did not modified SR Ca2+ load (**Figure 1D**) and did not affect Ca2+ spark frequency (**Figures 1E,F**) at any concentration studied. However, 100 ng/ml of TNFα had no effects on either Ca2+ transient amplitude, Ca2+ spark frequency, or SR Ca2+ load. However, 100 ng/ml of TNFα still accelerated the Ca2+ transient decay (**Figure 1C**). These results clearly show that sustained TNFα activation mediates an increase in systolic Ca2+ release. Altogether, our results lean toward the idea of a positive inotropic effect.

#### PLA2 and Ceramidase Mediate TNF**α** Regulation of Ca2+ Signaling

Previous work has suggested that TNFα response is mediated by the sphingosine signaling pathway (Hofmann et al., 2003). To investigate the signaling pathway involved in TNFα regulation of Ca2+ signaling, we used a ceramidase inhibitor

10 ng/ml, *n* = 15 for 50 ng/ml, *n* = 9 for 100 ng/ml). (D) Mean of sarcoplasmic reticulum (SR) Ca2+ load obtained by caffeine application after 1 min field stimulation in same conditions (respectively, *n* = 10, *n* = 11, *n* = 11, and *n* = 7). (E) Examples Ca2+ spark frequency (CaSpF) recording in freshly isolated cardiomyocytes at baseline, at 10 ng/ml TNFα and 100 ng/ml TNFα. (F) Mean of CaSpF (number of sparks (#) per second per 100 μm) obtained in same groups as in (A) (respectively, *n* = 19, *n* = 20, *n* = 15, and *n* = 7). \**p* < 0.05, \*\**p* < 0.01.

(5 μM NOE) and a PLA2 inhibitor (10 μM ATK). NOE fully prevented the increase of Ca2+ transient amplitude (**Figures 2A,B**) and the faster Ca2+ transient decay time induced by 10 ng/ml of TNFα (**Figure 2C**). NOE had no significant effects on neither the Ca2+ spark frequency nor the SR Ca2+ load (**Figures 2D**–**F**). Similarly, the phospholipase A2 inhibitor blunted all TNFα-mediated effects on the Ca2+ transient and the Ca2+ transient decay time (**Figures 2B,C**). As for NOE, ATK had no effect on SR Ca2+ load (**Figure 2D**). However, ATK, contrarily to NOE, did significantly reduce basal Ca2+ spark frequency. Altogether, those results suggest that TNFα alters Ca2+ signaling *via* the activation of the ceramidase and phospholipase A2 signaling pathway.

#### Gender Differences in Upstream TNF**α** Signaling Pathway in Obesity-Linked Type 2 Diabetic Mice (*db*/*db*)

Since plasmatic TNFα level is significantly elevated in type 2 diabetic patients, we first measured the plasmatic level of TNFα in male and female *db*/*db* mice. At 15 weeks old, *db*/*db* mice develop a type 2 diabetes linked to obesity with associated cardiomyopathy (Pereira et al., 2006b). Surprisingly, neither male nor female *db*/*db* mice presented an increase in their plasmatic level of TNFα compared to control (**Figure 3A**). Then, we measured the expression of key proteins involved in the TNFα signaling pathway, such as type 1 and type 2 TNFα receptors and the TNFα conversion enzyme TACE in both male and female *db*/*db*

mice. Interestingly, TACE expression was significantly higher in male *db*/*db* mice compared to controls, whereas no change was detectable in the female group (**Figure 3B**). Moreover, while TNF-R1 receptor expression was unchanged in both *db*/*db* groups (**Figure 3C**), TNF-R2 in the *db*/*db* male group was significantly increased (**Figure 3D**). These results clearly suggest that in male *db*/*db* mice hearts, the TNF-R2, known to mediate a cardioprotective pathway, is over-expressed, probably to protect the heart from diabetic-induced stress.

## Gender Differences in Obesity-Linked Type 2 Diabetic (*db*/*db*) Ca2+ Mishandling

In *db*/*db* mice, cardiac dysfunction has been associated with a decrease in SR Ca2+ transient amplitude and SR Ca2+ load (Belke et al., 2004; Pereira et al., 2006b, 2014). Here, we confirmed, in isolated cardiac myocytes from male *db*/*db* mice, that Ca2+ transient amplitude is significantly decreased (**Figures 4A,B**). This drop in Ca2+ transient amplitude (~51% lower than control, *p* < 0.01) is correlated with a drop in SR Ca2+ load (**Figure 4D**) (~51% lower than control, *p* < 0.01), which could explain the smaller (although not significant) cardiac cell shortening (**Figure 4C**). In our experimental conditions, Ca2+ spark frequency does not seem to be altered in *db*/*db* compared to control (*db*/*+*) (*p* = N.S.) (**Figures 4E,F**). In female *db*/*db* mice, the Ca2+ handling was similar in *db*/*db* compared to their control littermates (**Figure 5**). Indeed, all parameters such as Ca2+ transient amplitude (**Figure 5A**), Ca2+ spark frequency (**Figure 5C**), SR Ca2+ load (**Figure 5D**), and cell shortening (**Figure 5B**) were not significantly modified in freshly isolated cardiomyocytes in female *db*/*db* compared to control. In

percentage of relative density normalized by GAPDH signal in male and female control (*n* = 13, *n* = 13) and *db*/*db* (*n* = 13, *n* = 15) heart tissue homogenates. \**p* < 0.05.

conclusion, we found a gender-specific alteration of Ca2+ handling in *db*/*db* mice, with lower SR Ca2+ release associated to a drop in SR Ca2+ load in male, not seen in female.

#### Gender Differences of TNF**α**-Mediated Effect in Type 2 Diabetic (*db*/*db*)

Next, we compared TNFα regulation of Ca2+ signaling between male and female *db*/*db* mice. In male *db*/*db* mice, 10 ng/ml TNFα did not alter Ca2+ transient amplitude, cell shortening, nor SR Ca2+ load (**Figures 4A–C,F**). However, 10 ng/ml of TNFα similarly increased Ca2+ spark frequency in both control (~3.29 fold, *p* < 0.05) and *db*/*db* (1.5 fold, *p* = 0.06) (**Figure 4D**). In female control, the higher Ca2+ transient amplitude and cell shortening did not reach significance. Both female *db*/*db* and control had unchanged Ca2+ spark frequency. Those results suggest that, in 15 weeks old female *db*/*db*, the excitation-contraction coupling is unchanged compared to female control. Moreover, TNFα fails to show the effects found in male *db*/*db* (**Figure 4D**). Therefore, there are gender differences in Ca2+ mishandling and the underlying mechanisms in type 2 diabetes.

# DISCUSSION

We have previously shown that cardiac dysfunction in type 2 diabetes is associated with cardiomyocyte Ca2+ mishandling, resulting from a decrease in the Ca2+ channels involved in the Ca2+-induced Ca2+ release process (RyR and L-Type Ca2+ channels) (Belke et al., 2004; Pereira et al., 2006b). Although TNFα is elevated in diabetic patient and animal model of diabetes (Yamakawa et al., 1995; Pereira et al., 2006a; Preciado-Puga et al., 2014), little was known about its role in cellular alteration, notably regarding the Ca2+ signaling pathway and gender specificity in animal model of diabetes linked to obesity. Here, we found a gender-specific alteration of Ca2+ and TNFα signaling in *db*/*db* mice, a common model of type 2 diabetes linked to obesity. Indeed, we found that male *db*/*db* mice, not

cardiomyocytes at in *db*/+ and *db*/*db* with or without TNFα (F) Ca2+ spark frequency in the same groups (for *db*/*db*: *n* = 12 and 17, for *db*/+: *n* = 12 and *n* = 18)

female, presented the previously described Ca2+ mishandling with lower systolic Ca2+ release and SR Ca2+ load. More interestingly, we found that male and female *db*/*db* mice expressed differently TNF-R2, with an increased expression in male *db*/*db* mice that might reflect the activation of the TNFα cardio-protective TNF-R2-dependent pathway, not seen in female *db*/*db*.

\*\**p* < 0.01 compared to *db*/+ and \$\$*p* < 0.05 compared to *db*/*db* without TNFα treatment.

# Cardiac Positive Inotropic Effect of TNF**α**

Discrepancies regarding the TNFα regulation of Ca2+ signaling are quite important in the literature with reported positive or negative ionotropic effect. For instance, in cat cardiomyocytes, short time exposure of TNFα reduced Ca2+ transient amplitude in response to a disruption of Ca2+ influx *via* L type Ca2+ channels leading to cellular shortening, supporting, then, a negative ionotropic effect of TNFα (Yokoyama et al., 1993). This negative inotropic effect of TNFα has been also described, in rabbit and guinea pigs, with TNFα-induced impaired cellular shortening cardiomyocytes mediated by NO dependent but Ca2+ independent (Goldhaber et al., 1996; Sugishita et al., 1999). However, various studies performed in rodents have shown that TNFα can lead to inotropic positive effects (Bick et al., 1997; Greensmith and Nirmalan, 2013). Here, we found that TNFα treatments (10 and 50 ng/ml) induced a time and concentration-dependent effect leading to a significant increase in Ca2+ transient amplitude between 1 h and 1 h 30 min suggesting a positive inotropic effect. Our results are in concordance with Bick et al. study (Bick et al., 1997), who have found that TNFα incubation increases Ca2+ transient and cellular contraction in neo-natal cardiomyocytes. In adult rat cardiomyocytes treated with 50 ng/ml of TNFα (Greensmith and Nirmalan, 2013), Ca2+ transient amplitude and cellular shortening were also increased (Greensmith and Nirmalan, 2013). The absence of effect observed under 100 ng/ml of TNFα might be explained by its bimodal effect, as previously described in cardiomyocytes, depending on exposure time or dose (Amadou et al., 2002; Shanmugam et al., 2016). Then, 100 ng/ml TNFα

or higher doses, and with prolonged exposure, is expected to induce negative inotropic effects on Ca2+ handling.

#### In Mice Cardiomyocytes, TNF**α** Regulates Ca2+ Signaling *via* the Sphingosine and PLA2 Pathways

Previous studies have shown that TNFα produces myocardial effects (negative or positive inotropic effect) through different mechanisms such as PLA2 or sphingosine signaling pathway (Murray and Freeman, 1996; Oral et al., 1997; Liu and McHowat, 1998). Here, we found that exposure of TNFα (1 h to 1 h 30 min) mediates Ca2+ transient increase *via* the activation of both ceramidase (sphingosine precursor) and PLA2 (for arachidonic acid production). Sphingosine is commonly associated to short-term (within minutes) negative inotropic effect of TNFα (Oral et al., 1997). However, other studies have shown that ceramide enhanced SR Ca2+ release and SR Ca2+ re-uptake in adult ventricular myocytes (Liu and Kennedy, 2003). Those results are in line with our prevention of TNFαmediated elevation of systolic Ca2+ release and Ca2+ transient decay time in cardiomyocytes treated with the ceramidase inhibitor NOE (**Figures 2B,C**). Moreover, inhibition of the PLA2 prevented TNFα-mediated increase in Ca2+ transient amplitude and SR Ca2+ transient decay time, suggesting that TNFα induces Ca2+ mishandling *via* PLA2-mediated phosphorylation of RyR. Indeed, 10 ng/ml of TNFα has been shown to increase Ca2+ transient amplitude as a result of PLA-2 mediated RyR PKA phosphorylation at serine 2,808 in wildtype mice of RASSF1A knock out (Mohamed et al., 2014). This PKA-dependent mediated effect of PLA-2/arachidonic acid on the RyR phosphorylation state perfectly explains why we observed a dramatic drop of Ca2+ spark frequency under the inhibition of the PLA-2 (**Figure 2F**). In addition, TNFα also accelerates SR Ca2+ re-uptake reflecting an increase in SERCA pump activity as seen under PKA phosphorylation of phospholamban supporting the TNFα/PLA-2/PKA pathway. This mechanism is confirmed by the restoration of the TNFαmediated acceleration Ca2+ transient decay time under ATK, the PLA-2 inhibitor (**Figures 2B,C**).

#### Gender-Dependent Ca2+ Mishandling in *db*/*db* Mice, an Obesity-Linked Type 2 Diabetic Model

Type 2 diabetes is the most common form of diabetes. In western countries, 80% of type 2 diabetic patients have developed a diabetes linked to obesity resulting in severe glucose intolerance Delgado et al. Gender and TNFα in *db/db* Mice

compared to lean type 2 diabetic patients (Schaffer and Mozaffari, 1996). Our study was performed in *db*/*db* mice, a model that recapitulates, in that sense, the human pathology. Indeed, the leptin receptor mutation of *db*/*db* mice impairs the satiety feeling and leads to obesity around 4–5 weeks of age, which is followed by diabetic state with hyperglycemia and insulin resistance (Coleman, 1978). In type 2 diabetes linked to obesity, cardiac dysfunction has been associated to Ca2+ mishandling and structural remodeling (Belke et al., 2004; Pereira et al., 2006b; Falcao-Pires and Leite-Moreira, 2012). Indeed, overall, animal models of type 2 diabetes present a reduced Na+ /Ca2+ exchanger activity, and depressed Ca2+ transient linked to downregulation of Ca2+ channels, RyRs, and reduced SERCA activity (Netticadan et al., 2001; Zhong et al., 2001; Abe et al., 2002; Belke et al., 2004; Pereira et al., 2006b; Boudina and Abel, 2010). Here, our results show that those effects are recapitulated in male *db*/*db* mice (**Figure 4**), but not in female *db*/*db* mice. However, the gender-specific regulation in Ca2+ handling and/or β-adrenergic response has been previously described (Parks et al., 2014). Supporting this idea, we found that basal Ca2+ transient amplitude is lower in female control compared to male control cardiomyocytes. Although Parks et al. (2014) have shown that Ca2+ current, diastolic Ca2+, and SR Ca2+ load were similar between control male and female, basal cAMP level was lower in control female compared to control male due to higher PDE4B expression in female. These results are in line with our previous work showing that *db*/*db* female mice have reduced phosphorylation of the RyR, which reduce Ca2+ spark frequency and could explain the preserve SR Ca2+ load and Ca2+ transient seen in female *db*/*db* compared to *db*/*db* male. Our results are paradoxical compared to the higher risk to develop heart failure for type 2 diabetic women compared to diabetic men. This discrepancy could be explain as follow: the decrease in [Ca2+]i transient in male *db*/*db* mice could be protective at long term, maybe by preventing Ca2+ toxic effects such as apoptosis or preserve ATP content by limiting the ATP expense in pumping Ca2+ (Javorkova et al., 2010; Parks et al., 2014). Future studies will be needed to confirm this hypothesis.

#### Gender Dependent Alteration of Molecular TNF**α** Signaling Pathway in *db*/*db*

To our knowledge, plasmatic TNFα level parallels the degree of cardiac dysfunction in diabetic patients. In the *db*/*db* mice, we did not observe any changes in the plasmatic level of TNFα compared to control. Even though circulating TNFα is unchanged, male *db*/*db* mice present an increase in TACE expression suggesting a paracrine elevation of TNFα in the heart. Surprisingly, despite cardiomyocyte treatment with 10 ng/ml of TNFα, a concentration within the *in vivo* range measured under stress and injury (Bitterman et al., 1991), TNFα did not induce an increase in Ca2+ transient amplitude or decay time in *db*/*db*, as seen in C57Bl6 mice (**Figures 4B,C**). One explanation could be that in *db*/*db* control littermate strain background (C57BKS/J strain), TNFα is not as effective as in C57Bl6 strain. Indeed, genetic background, such as between C57BL6/J and C57BL6/N, has been shown to influence cardiac phenotype and propensity to develop cardiomyopathies (Tian et al., 2011; Simon et al., 2013). This could also explain the ineffective response of TNFα in female control and *db*/*db* mice (**Figure 5**). Although TNFα activation has been linked with oxidative stress, no genderspecific difference in cardiomyocytes redox state at baseline or during pathology has been observed (Ren, 2007; Bell et al., 2015). Another possibility could be that in male *db*/*db*, the dramatically reduced SR Ca2+ load would prevent the high Ca2+ systolic release induced by TNFα probably due to the phosphorylation of the RyR *via* the activation of PLA2. Indeed, we found in the presence of TNFα an increase in Ca2+ spark frequency in both *db*/*+* and *db*/*db* mice reflecting an elevated diastolic RyR opening resulting from RyR phosphorylation by PKA previously described in male *db*/*db* (Pereira et al., 2014). Interestingly, in male *db*/*db* mice, the TNF-R2 was overexpressed, which is known to exert cardio-protective effects *via* the activation of NF-κB (Burchfield et al., 2010). Indeed, in liver, TNFα inhibits PDE3 expression elevating cAMP level and PKA activation (Ke et al., 2015). This activation of PKA could explain, in cardiomyocytes, the elevation of Ca2+ spark frequency in male *db*/+ cardiomyocytes treated with TNFα (**Figure 4F**). Moreover, TNF-R2 is known to be involved in positive cardiac inotropic effect (Defer et al., 2007). As a result, [Ca2+] overload was prevented and Ca2+ transient increased leading to an increase in inotropic response. The over-expressed TNF-R2 in a male *db*/*db* appears as an attempt to counteract the already present Ca2+ mishandling to protect from cardiac dysfunction. Indeed, prolonged activation of the TNF-R2 pathway in the *db*/*db* male cardiomyocytes could then activate phosphorylation of excitationcontraction coupling key proteins, such as phospholamban, to restore Ca2+ transient and cardiomyocytes contraction.

In conclusion, we found for the first time that both Ca2+ and TNFα signaling are altered only in male type 2 diabetic mice, whereas female does not seem to be affected. Although this study has several limitations in the interpretation such as non-comparable hormonal state between female *db*/*db* mice and diabetic women, lower effect of TNFα in *db*/*+* than C57BL6 control, we still clearly show that male *db*/*db* mice develop Ca2+ mishandling leading to impaired contraction already at a young age, while woman seemed to be protected. Moreover, we found that male *db*/*db* mice put into place a protective mechanism to counteract those negative effects by over-expressing TNF-R2 cardio-protective signaling pathway.

# DATA AVAILABILITY

The datasets generated for this study are available on request to the corresponding author.

# ETHICS STATEMENT

The study was carried out in accordance to the ethical principles of the French Ministry of Agriculture and the European Parliament on the protection of animals. The protocol was approved by the French Ministry of Agriculture and Bioethical Committee of the CSIC following recommendation of the Spanish Animal Care and the European Parliament on the protection of animals.

#### AUTHOR CONTRIBUTIONS

CD and AG conceived and designed the project, supervised the data acquisition and participated in analysis. LP and GR performed most of the experiments and analyses. LP interpreted the data and wrote the first draft of the manuscript. MS

#### REFERENCES


participated in the figure preparation. All authors have edited the manuscript.

#### FUNDING

This work has been funded by Acuerdos Bilaterales España-Francia (CSIC-INSERM) grant no. 2005FR0020 and partially by SAF2017-84777R from the Spanish Ministerio de Industria, Economia y Competitividad (to CD) and partially by CP15/00129 and PI17/01093 (to GR), and the ANR-11-1DEX-0003-02 to LP as members of the Laboratory of Excellence LERMIT.


hypertensive rats. *Clin. Exp. Pharmacol. Physiol.* 34, 432–438. doi: 10.1111/j. 1440-1681.2007.04591.x


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Delgado, Gomez, Samia El Hayek, Ruiz-Hurtado and Pereira. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# TRP Channels: Current Perspectives in the Adverse Cardiac Remodeling

*Debora Falcón1† , Isabel Galeano-Otero1† , Eva Calderón-Sánchez1,2† , Raquel Del Toro1,2 , Marta Mart´ın-Bórnez1 , Juan A. Rosado3 , Abdelkrim Hmadcha4,5 and Tarik Smani1,2 \**

*1Department of Medical Physiology and Biophysics, Institute of Biomedicine of Seville, University of Seville, Sevilla, Spain, 2CIBERCV, Madrid, Spain, 3Department of Physiology (Cell Physiology Research Group), University of Extremadura, Cáceres, Spain, 4Department of Generation and Cell Therapy, Andalusian Center for Molecular Biology and Regenerative Medicine (CABIMER), University of Pablo de Olavide-University of Seville-CSIC, Sevilla, Spain, 5CIBERDEM, Madrid, Spain*

#### *Edited by:*

*Maria Fernandez-Velasco, University Hospital La Paz, Spain*

#### *Reviewed by:*

*Antonio Ferrer-Montiel, Universidad Miguel Hernández de Elche, Spain Karel Talavera, KU Leuven, Belgium*

*\*Correspondence:* 

*Tarik Smani tasmani@us.es*

*† These authors have contributed equally to this work*

#### *Specialty section:*

*This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology*

*Received: 18 November 2018 Accepted: 08 February 2019 Published: 01 March 2019*

#### *Citation:*

*Falcón D, Galeano-Otero I, Calderón-S´anchez E, Del Toro R, Mart´ın-Bórnez M, Rosado JA, Hmadcha A and Smani T (2019) TRP Channels: Current Perspectives in the Adverse Cardiac Remodeling. Front. Physiol. 10:159. doi: 10.3389/fphys.2019.00159*

Calcium is an important second messenger required not only for the excitation-contraction coupling of the heart but also critical for the activation of cell signaling pathways involved in the adverse cardiac remodeling and consequently for the heart failure. Sustained neurohumoral activation, pressure-overload, or myocardial injury can cause pathologic hypertrophic growth of the heart followed by interstitial fibrosis. The consequent heart's structural and molecular adaptation might elevate the risk of developing heart failure and malignant arrhythmia. Compelling evidences have demonstrated that Ca2+ entry through TRP channels might play pivotal roles in cardiac function and pathology. TRP proteins are classified into six subfamilies: TRPC (canonical), TRPV (vanilloid), TRPM (melastatin), TRPA (ankyrin), TRPML (mucolipin), and TRPP (polycystin), which are activated by numerous physical and/or chemical stimuli. TRP channels participate to the handling of the intracellular Ca2+ concentration in cardiac myocytes and are mediators of different cardiovascular alterations. This review provides an overview of the current knowledge of TRP proteins implication in the pathologic process of some frequent cardiac diseases associated with the adverse cardiac remodeling such as cardiac hypertrophy, fibrosis, and conduction alteration.

Keywords: calcium, TRP channels, cardiac remodeling, hypertrophy, fibrosis, conduction disorders

# INTRODUCTION

Calcium (Ca2+) is an important second messenger necessary for the excitation-contraction (EC) coupling process in cardiac myocytes (Berridge, 2002; Bers, 2002; Eisner et al., 2017), which occurs as a consequence of Ca2+ entry to the cytosol due to L-type Ca2+ channels that provoke Ca2+ release from sarcoplasmic reticulum causing cardiac myocyte contraction. Cardiac relaxation starts when intracellular Ca2+ concentration ([Ca2+]i) decreases as a result of the activity of sarco/endoplasmic reticulum Ca2+ ATPase (SERCA) responsible for the Ca2+ reuptake into the sarcoplasmic reticulum and the Na+ /Ca2+ exchanger responsible of the Ca2+ extrusion out of cardiomyocytes (Bers, 2002; Eder, 2017). Ca2+ is also required for the activation of signaling pathway that plays minor roles in the healthy heart, for example, those involved in the cardiac remodeling and heart failure (Yue et al., 2015; Eder, 2017; Freichel et al., 2017; Avila-Medina et al., 2018; Domínguez-Rodríguez et al., 2018).

Among the wide Ca2+-permeable channels known in the heart, transient receptor potential (TRP) channels contribute to the Ca2+ influx induced by a wide spectrum of physicochemical stimuli from cellular microenvironments, such as thermal, mechanical stresses, and neurohormonal (Freichel et al., 2017). Likewise, a wide variety of vasoactive agent, including endothelin-1, thrombin, ATP, angiotensin-II, or bradykinin, also stimulates TRP (Suzuki et al., 2011; Albarrán et al., 2013; Gerhold and Schwartz, 2016; Sawamura et al., 2017). TRP channels comprise a large Ca2+-permeable cation channel superfamily showing a common architecture. They contain six transmembrane domains (TM1–TM6) and the cation-permeable pore region formed by a loop between TM5 and TM6 (Gaudet, 2007). They are divided into six major subgroups based on their specific function and sequence analogies (Nilius and Droogmans, 2001; Owsianik et al., 2006; Peng et al., 2015): the canonical channel (TRPC), the vanilloidrelated channel (TRPV), and the melastatin-related channel (TRPM) formed by 7, 6, and 8 different channel proteins, respectively (Montell et al., 2002; Venkatachalam and Montell, 2007; Yu et al., 2010); the ankyrin-related channel (TRPA) subfamily containing only one protein; the mucolipin-related channel (TRPML) formed by three proteins; and polycystinrelated channel (TRPP) constituted by three members also known as polycystic kidney disease protein (PKD2) (Montell et al., 2002; Venkatachalam and Montell, 2007).

TRP channels display diverse cation selectivity and activation mechanisms (Venkatachalam and Montell, 2007). Most of them are nonselective for cation and are permeable to monovalent and divalent cations, more permeability for Ca2+ than Na+ (ratio PCa/PNa) that ranges from channels selective for monovalent cations, such as TRPM4 and TRPM5 with a ratio PCa/PNa < 0.05, to highly Ca2+ selective channels, as TRPV5 and TRPV6, which exhibit a ratio PCa/PNa close to 100 (Yue et al., 2015). TRPM6 and TRPM7 are also permeable to Mg2+, Ca2+, Na+ , Zn2+, and other metals [for an extensive review see Freichel et al. (2017)]. Most of TRP channels lack the typical voltage sensor; therefore, they are not gated by voltage (Ramsey et al., 2006; Alonso-Carbajo et al., 2017; Jardín et al., 2017) but they can be modulated by a different chemical and physical stimuli, including temperature fluctuations and mechanical stretch (Islas, 2017; Nazıroğlu and Braidy, 2017; Yamaguchi et al., 2017), extracellular and intracellular ions (including H+ , Ca2+, and Mg2+) (Launay et al., 2002; Jiang et al., 2005; Li et al., 2007), intracellular ligands [as diacylglycerol (Palazzo et al., 2013), phosphoinositide-4,5-bisphosphate (PIP2) (Ong and Ambudkar, 2017)], and various exogenous natural and synthetic ligands (Vetter and Lewis, 2011; Holzer and Izzo, 2014).

TRP channels have been suggested as regulators of local micro domains signaling pathway related with changes in the [Ca2+]i

Falc´on et al. TRPs in Cardiac Remodeling

or by the interplay with Ca2+-dependent regulatory proteins. Actually, they contribute to Ca2+ homeostasis by directly conducting Ca2+ or indirectly *via* membrane depolarization and modulation of voltage-gated Ca2+ channels (Nilius and Droogmans, 2001; Owsianik et al., 2006; Freichel et al., 2014; Pires et al., 2017). Hence, during the last two decades, TRPs have been suggested as intermediaries of diverse physiological and pathophysiological cardiovascular processes (Inoue et al., 2006, 2018; Egginton, 2009; Smani et al., 2015; Yue et al., 2015).

#### EXPRESSION OF TRP CHANNELS IN CARDIAC CELLS

RT-PCR, western blot, immunostaining, and functional current recordings demonstrated that TRPs are expressed ubiquitously in cardiac myocytes and fibroblasts of different species (Sabourin et al., 2011; Yue et al., 2015). In the case of TRPC channel, the seven members TRPC1–7 are expressed in the majority of the cell types in heart (Eder, 2017; Freichel et al., 2017). Consistently, all TRPCs, except TRPC5, were detected in the sinoatrial node (Ju et al., 2007). Interestingly, significant overexpression of TRPC1/C3/C4/C5 or TRPC6 was detected in patients with heart failure as compared to nonfailing heart (Bush et al., 2006; Morine et al., 2016). Interestingly, these TRPC channels show distinct profiles of expression in the ventricles of patients with heart failure as it happens in murine models of univentricular pressure overload (Morine et al., 2016). As in other cell types, TRPC channels are implicated in signal transduction in cardiac myocytes (Flockerzi and Nilius, 2014; Eder, 2017; Freichel et al., 2017). TRPC family requires the phospholipase C (PLC) pathway for activation. TRPC3, TRPC6, and TRPC7 interact directly with diacylglycerol (Yamaguchi et al., 2018), while TRPC1, TRPC4, and TRPC5 are activated indirectly through a still unidentified mechanism (Sabourin et al., 2011; Zhang and Trebak, 2014; He et al., 2017). Some TRPC channels are activated by intracellular Ca2+ store depletion, which stimulates the store-operated Ca2+ entry (SOCE) required for diverse cardiac physiopathological process (Ong et al., 2016; Eder, 2017). It has been proposed that TRPC1 associates with TRPC4 or TRPC5, thereby forming the store-operated Ca2+ channel, while TRPC3, TRPC6, and TRP7 are suggested to form the receptor-operated channel (Ju and Allen, 2007; Saleh et al., 2008; Sabourin et al., 2012). Others studies demonstrated that long-term stimulation of cardiac myocytes with angiotensin II, phenylephrine, endothelin-1, or aldosterone evoked an exacerbated SOCE elicited by thapsigargin, correlating with an increment in the expression or activation of TRPC1, TRPC4, and/or TRPC5 (Watanabe et al., 2008; Makarewich et al., 2014; Camacho Londoño et al., 2015; Sabourin et al., 2016). The use of dominant negative mutants confirmed that TRPC4 is sensitive to passive Ca2+ store depletion, while TRPC3 and TRPC6 respond to the diacylglycerol stimulus, regardless of store depletion (Makarewich et al., 2014). Furthermore, upregulation of TRPC3/C4 in adult ventricular cardiomyocytes correlated with the enhanced SOCE and pro-arrhythmic

**Abbreviations:** AF, Atrial Fibrillation; [Ca2+]i, concentration of intracellular Ca2+; ROS, reactive oxygen species; SOCE, store-operated Ca2+ entry; TRP, transient receptor potential; TRPC, transient receptor potential-canonical; TRPV, transient receptor potential-vanilloid; TRPM, transient receptor potential-melastatin; TRPA, transient receptor potential-ankyrin; TRPML, transient receptor potential-mucolipin; TRPP, transient receptor potential-polycystin.

spontaneous Ca2+ waves (Domínguez-Rodríguez et al., 2015). Importantly, transient occlusion of coronary artery in rats also enhanced the expression of TRPC1/C3/C4/C5 and TRPC6 either in risk or in remote zone of the infarcted heart (Domínguez-Rodríguez et al., 2018). Finally, TRPC7 activation was proposed to initiate angiotensin-II activation to myocardial apoptosis (Satoh et al., 2007).

TRPV channels were also detected in mammalian hearts, especially TRPV1, TRPV2, and TRPV4 (Yue et al., 2015). Most of TRPV channels are sensitive to temperature and ligands, and they participate in sensation of hot temperature and in chemoreception (Vriens et al., 2007; Islas, 2017). TRPV1 was identified principally in sensory nerves in the cardiovascular system but also in the myocardium (Zahner et al., 2003; Gao et al., 2015; Randhawa and Jaggi, 2017). Bradykinin evoked a TRPV1-dependent [Ca2+]i increase in cardiac neurons, indicating that TRPV1 activation was responsible for stimulation/ sensitization by bradykinin of cardiac nociceptors (Wu and Pan, 2007). An early study demonstrated that after *trpv1* gene deletion, an exacerbated inflammation and cardiac remodeling occurred due to impaired post-ischemic recovery in isolated perfused infarcted heart (Wang and Wang, 2005). More recently, the overexpression of TRPV2 after myocardial infarction was observed in cardiac tissue of rats (Entin-Meer et al., 2014), and TRPV2 downregulation in knockout mice was related to a better recovery after myocardial infarction (Entin-Meer et al., 2017), probably because of an attenuated pro-inflammatory response in these mice. Another study also suggested that TRPV2 may play a critical role in stretch-activated Ca2+ influx pathway in dystrophic cardiomyopathy, contributing to [Ca2+]i mishandling (Lorin et al., 2015). In the case of TRPV4, it is also highly expressed in the heart and is activated during myocardial ischemia and reperfusion, which induced Ca2+ influx with subsequent reactive oxygen species (ROS) release (Wu et al., 2017). Recently, TRPV4 upregulation in cardiomyocytes was also linked with aging in mice (Jones et al., 2018). Indeed, pharmacological inhibition of TRPV4 with HC067047 prevented stress-induced cardiomyocyte death and ischemia and reperfusion-induced cardiac damage in aged mice. These findings might have potential implications in the treatment of elderly populations at increased risk of myocardial infarction.

Regarding the expression of TRPM channels in hearts, it is known that eight isoforms are present in different parts of the heart, and TRPM2/4/7 is expressed specifically in cardiac myocytes (Yue et al., 2015). The analysis of mRNA and/or protein levels showed that levels of TRPM2, TRPM3 and TRPM8 were reduced in left and right ventricle of patients with failing heart (Morine et al., 2016), meanwhile increased expression of TRPM7 was observed in the left ventricle of patients with ventricular tachycardia (Parajuli et al., 2015). TRPM2 seems essential for cardiac myocyte bioenergetics maintenance (Hoffman et al., 2015). Its activation protected the heart from ischemia and reperfusion injury by improving mitochondrial dysfunction and reducing ROS levels (Miller et al., 2014). TRPM4 is thought to interfere in cardiac myocyte contraction by its activation of voltage-gated Ca2+ channel (Alonso-Carbajo et al., 2017). Actually, the exacerbated activity of TRPM4, which is activated by physiological range of Ca2+ concentration, was related with arrhythmic changes (Hu et al., 2017). Also, TRPM7 is believed indispensable during the myocardial proliferation in early stages of cardiogenesis, since the deletion of *trpm7* gene, before embryonic day 9 of mice, provoked heart failure and embryonic death (Sah et al., 2013). Nevertheless, the mechanism by which TRPM7 regulates cardiac cell proliferation remains unknown (Chubanov et al., 2017).

In relation to the expression and function of other TRP channels in heart, TRPA1 was detected in cardiac endothelial cells, vascular smooth muscle, and in cardiac myocytes (Yue et al., 2015), where its activation seems relevant to attenuate ischemia and reperfusion injury (Lu et al., 2016). Furthermore, the presence of TRPP2 has been proven in knockout mice who died before birth as a result of cardiac malformations (Pennekamp et al., 2002). Recently, a study confirmed that TRPP2 was able to regulate autophagy through Ca2+ homeostasis in cardiac myocytes (Criollo et al., 2018). However, there are only few studies which looked on the role of these channels in the function of cardiac cells.

#### ROLE OF TRP CHANNELS IN THE ADVERSE CARDIAC REMODELING

Ca2+ plays critical role in the adaptation of the heart to environmental demands, leading to cardiac remodeling (Eder, 2017; Avila-Medina et al., 2018). Physical exercise or pregnancy are reversible stimuli that induce physiologic cardiac hypertrophy to adapt the increase in consumption of nutrients and oxygen, whereas sustained neurohumoral activation, hypertension, or myocardial injury can lead to pathological heart hypertrophy followed by interstitial fibrosis (Pfeffer and Braunwald, 1990; Klug et al., 1993; Hill and Olson, 2008). These events might cause left ventricular dilation and dysfunction, what is known as the adverse cardiac remodeling, which increases the risk of heart failure and malignant arrhythmia (Hill and Olson, 2008; van der Bruggen et al., 2017).

Initial attentions have been given to describe the role of TRP channels in the appearance of cardiac remodeling using different experimental procedures *in vitro* and *in vivo* in animal models (Guinamard and Bois, 2007; Eder and Molkentin, 2011; Yue et al., 2015). Here, we will focus on the role of TRPCs, TRPVs, and TRPMs role in the adverse cardiac remodeling, since little is known regarding the other subfamilies and their participation in these processes.

#### TRPs and Cardiac Hypertrophy

Compelling evidence confirmed that the activity and expression of several TRP channels are upregulated in pathological hypertrophy and heart failure as reviewed elsewhere (Watanabe et al., 2008; Yue et al., 2015; Freichel et al., 2017), and as summarized in **Figure 1**. Special attention has been given to TRPC's role in cardiac hypertrophy, but the implication of TRPV1, TRPV2, and TRPM4 has been also demonstrated.

#### TRPCs and Hypertrophy

The role of TRPCs in the pathological cardiac hypertrophy has been extensively studied in isolated neonatal and adult cardiac myocytes and in different animal model of cardiac hypertrophy (Nishida and Kurose, 2008; Watanabe et al., 2008; Eder and Molkentin, 2011; Xiao et al., 2017). Ca2+ entry through TRPCs is considered essential to activate signaling pathways involving PKC, calmodulin-kinase, and calcineurin/ NFAT, which promotes cardiac hypertrophy by the re-expression of the fetal gene program (Heineke and Molkentin, 2006; Nakayama et al., 2006; Eder and Molkentin, 2011). Independent reports demonstrated that stimulation of neonatal rat cardiac myocytes with hypertrophic agonists such as ATP, endothelin-1, phenylephrine, or angiotensin-II, increased cells size, correlating with the upregulation and activation of TRPC1 (Ohba et al., 2007), TRPC3 (Brenner and Dolmetsch, 2007), TRPC5 (Sunggip et al., 2018), and TRPC7 (Satoh et al., 2007). Using animal models of pressure overload-induced heart hypertrophy, different studies demonstrated that TRPC1 (Ohba et al., 2007), TRPC3 (Bush et al., 2006; Brenner and Dolmetsch, 2007), and TRPC6 (Kuwahara et al., 2006) are upregulated in heart. Conversely, the overexpression of TRPC3 in transgenic mice increased cardiac hypertrophy through calcineurin/NFAT activation when mice were subjected to angiotensin-II and phenylephrine treatment or pressure overload (Nakayama et al., 2006). Similarly, the overexpression of TRPC6 in transgenic mice leaded to the development of cardiac hypertrophy and heart failure (Kuwahara et al., 2006). In contrast, the use of cardiomyocyte-specific dominant-negative mutants in transgenic mice for TRPC3 and TRPC6 or for TRPC4 reduced cardiac hypertrophic responses following either the infusion of neuroendocrine agonists or pressure overload stimulation through calcineurin/NFAT (Wu et al., 2010). Moreover, *trpc1* gene deletion in mice attenuated pressure overload-induced hypertrophy by the alteration of calcineurin/NFAT and Akt signaling pathway (Seth et al., 2009). Intriguingly and in contrast to other studies mentioned here, a previous study reported that TRPC3/TRPC6 double knockout mice did not develop pressure-overload induced hypertrophy; however, *trpc3* or *trpc6* gene's deletion did not protect the heart from hypertrophy or dysfunction due to pressure overload, suggesting the need of both channels to promote the deleterious hypertrophic response (Seo et al., 2014). This finding worth further investigation and confirmation in other model of pathological cardiac hypertrophy.

#### TRPVs and Hypertrophy

Most studies focused on TRPV1 and TRPV2 role in hypertrophy and the adverse cardiac remodeling. An early study revealed that TRPV1 knockout mice showed a reduced increase in heart's weight as compared to wild-type mice, and they were partially protected from pressure-overload induced cardiac hypertrophy (Buckley and Stokes, 2011). A recent study demonstrated that TRPV1's expression is increased in hypertrophied heart of mice evoked by transverse aorta constriction (TAC) (Chen et al., 2016). However, *Trpv1* gene's deletion was associated with excessive inflammation, exaggerated cardiac hypertrophy, and abnormal cardiac function after TAC, suggesting a protective role of TRPV1 (Zhong et al., 2018a). This study proposed that TRPV1, highly expressed in sensory nerves, might be involved in the regulation of cardiovascular function due to its anti-inflammatory effects *via* calcitonin gene-related peptide. Conversely, a previous study revealed that pharmacological activation of TRPV1 with SA13353 prevents the increase of cardiomyocyte size evoked by endothelin-1 (Zhang et al., 2012b). SA13353 also attenuated and reduced cold stress-induced hypertrophy and decreased the fractional shortening among others functional cardiac and cellular parameters (Zhang et al., 2012b). In contrast, another study showed that TRPV1 activation with capsaicin can antagonize high-salt dietmediated cardiac hypertrophy, by ameliorating the mitochondrial complex I oxidative phosphorylation and suggesting that TRPV1 mediated amendment of mitochondrial dysfunction may represent a novel target for the management of early cardiac dysfunction (Lang et al., 2015). Therefore, the role of TRPV1 in the molecular mechanism underlying pathological cardiac hypertrophy still remains unclear.

TRPV2 is also significantly upregulated in wild-type mice subjected to TAC, whereas the absence of functional TRPV2 reduces significantly the left ventricular hypertrophy after TAC, suggesting a role of TRPV2 in the development of cardiomyocyte hypertrophy because of an increased afterload (Koch et al., 2017). In this way, the overexpression of TRPV2 was also associated with enlarged hearts in patients with dilated cardiomyopathy (Iwata et al., 2013). TRPV3 is also overexpressed in angiotensin-II-induced cardiomyocyte hypertrophy, which aggravated cardiac hypertrophy through calcineurin/NFATc3 signaling pathway activation (Zhang et al., 2018).

#### TRPMs and Hypertrophy

In the case of TRPMs, different studies focused especially on TRPM4 role in hypertrophy, while little is known regarding other TRPM channels. TRPM4 is thought to fine-tune the amount of Ca2+ influx into cardiomyocytes via store-operated Ca2+ channels after chronic angiotensin-II stimulation, through a mechanism involving calcineurin–NFAT activation (Kecskés et al., 2015). TRPM4 upregulation was observed in hypertrophied ventricular cardiomyocytes freshly isolated from well-established genetic model of spontaneously hypertensive rats when compared to control, the Wistar Kyoto rats (Guinamard et al., 2006). Interestingly, selective removal of TRPM4 from the heart resulted in an increased hypertrophic growth after chronic mice treatment with angiotensin-II as compared to wild type mice (Kecskés et al., 2015), suggesting a protective role of TRPM4. Recently, a study confirmed the beneficial role of TRPM4 in hard training-induced physiological hypertrophy because TRPM4 knockout mice developed a pathological cardiac hypertrophy when they were subjected to endurance training (Demion et al., 2014). The deletion of the *Trpm4* gene in 12-week-old mice was linked with moderate cardiac hypertrophy as well as ventricular dilation, increased cellular density, and reduced left ventricular cardiomyocytes size, suggesting that TRPM4 might act as a negative regulator of cardiomyocytes proliferation during prenatal development (Demion et al., 2014).

#### Role of TRPs in Interstitial Fibrosis

Cardiac fibroblasts represent ∼75% of all cardiac cells, although they represent about ∼10–15% of total cardiac cell volume due to their small size. However, the activation of cardiac fibroblasts plays a crucial role in cardiac pathology. Cardiac fibrosis is caused by an excessive extracellular matrix proteins produced by myofibroblasts, the differentiated phenotype of fibroblasts under stress situations (Yue et al., 2013), where the Ca2+ signaling is essential for transcriptional control regulation and myocardial fibrosis (Ramires et al., 1998), as illustrated in **Figure 2**. The role of TRP channels in cardiac fibroblasts activation, proliferation, and differentiation has been extensively investigated (Du et al., 2010; Yue et al., 2013; Certal et al., 2017). However, unlike in cardiac myocytes, TRPs-activated Ca2+ signaling mechanisms are not fully understood in fibroblasts.

#### TRPCs and Fibrosis

TRPC1, TRPC3, and TRPC6 are considered the main TRPCs implicated in cardiac fibrosis, but most studies focused on TRPC3's role in cardiac fibrosis, as reviewed recently (Numaga-Tomita et al., 2017). A role of background Ca2+ entry through TRPC1 and TRPC4 was associated with fibrosis in double knockout mice subjected to pressure overload (Camacho Londoño et al., 2015). In mechanically stressed hearts, TRPC3 activation triggered an aberrant increase of ROS production, leading to RhoA activation in both cardiac myocytes and fibroblasts, resulting in interstitial fibrosis (Kitajima et al., 2016). Actually, *trpc*3 gene deletion suppressed cardiac fibrosis in response to pressure overload or to angiotensin-II infusion (Domes et al., 2015; Kitajima et al., 2016; Numaga-Tomita et al., 2016). TRPC3 blockade with Pyr3 also inhibited angiotensin-II-induced Ca2+ entry, proliferation, and differentiation of fibroblasts isolated from left atrial of electrically maintained atrial fibrillation in a dog model (Harada et al., 2012). Recently, chronic treatment of rat models of pressure overload with GSK503A, proposed to inhibit both TRPC3 and TRPC6, showed no interstitial cardiac fibrosis (Seo et al., 2014), suggesting that TRPC3 and TRPC6 are needed for the fibrosis appearance. TRPC6 has been supposed as a regulator of myofibroblast differentiation, since its silencing in human cardiac fibroblasts attenuated the upregulation of α-SMA, a marker of myofibroblast transformation, induced by TGF-β1, a pro-fibrotic agonist (Kapur et al., 2014). Conversely, the loss of TRPC6 in knockout mice prevented TGF-β1-mediated

myofibroblast transformation. In addition, TRPC6 overexpression using adenovirus promoted fibroblast conversion to myofibroblast, a hallmark of fibrosis (Davis et al., 2012). In contrast to this finding, an early report proposed that silencing of TRPC6 enhanced myofibroblast formation (Nishida et al., 2007), which questions whether TRPC6 is relevant for cardiac fibrosis.

#### TRPVs and Fibrosis

TRPV channels seem also critical for cardiac fibroblast differentiation to myofibroblasts. It was shown that the ablation of TRPV1 markedly enhanced the fibrosis due to myocardial infarction, by the stimulation of the TGF-β1 and Smad2 signaling pathway (Huang et al., 2010). In contrast, the administration of a TRPV1 antagonist in TAC mice protected the heart from the fibrotic process (Horton et al., 2013). Conversely, in a pressure-overload mice model, it demonstrated a reduction of hypertrophy and fibrosis-mediated TRPV1 channel activation by capsaicin (Wang et al., 2014). Capsaicin blunted pressure overload–induced upregulation of TGF-β1 and Smad2/3 phosphorylation. It also reduced angiotensin-II-induced proliferation of cardiac fibroblast isolated from wild-type but not from TRPV1-knockout mice (Wang et al., 2014). Recently, the overexpression of TRPV1 in transgenic mice-attenuated isoproterenol-induced myocardial fibrosis *in vitro* and *in vivo* in primary cultured cardiac fibroblasts (Wang et al., 2016).

Other reports suggested that TRPV2, TRPV3, and TRPV4 channels also participate in cardiac fibrosis. TRPV2 functional deletion was associated with a decreased development of fibrosis associated with aging (Jones et al., 2017), while TRPV2 upregulation was associated with enlarged hearts, increased fibrosis, and myocardial structural defects in patients with dilated cardiomyopathy (Iwata et al., 2013). Furthermore, TRPV3 activation intensified cardiac fibrosis, stimulating cardiac fibroblast proliferation in the pressure-overloaded rat hearts (Liu et al., 2018). TRPV4 seems also necessary for cardiac fibroblasts differentiation to myofibroblasts, since AB159908, a TRPV4 antagonist, as well as TRPV4 siRNA inhibited TGF-β1 induced fibroblasts differentiation (Adapala et al., 2013). Actually, TRPV4-knockout mice presented decreased fibrosis after myocardial infarction (Adapala et al., 2018).

#### TRPMs and Fibrosis

Several reports made special attention to the abnormal expression of TRPM7 and the development of cardiac fibrosis; however, little is known regarding other TRPM channels (Xu et al., 2015). Recently, higher amount of mRNA and protein levels of TRPM6 in addition to significant increase in markers of myocardial fibrosis as TGF-β1, collagen I and III, were detected in atrial fibrillation patients, suggesting possible contribution of TRPM6 to atrial fibrosis (Zhang et al., 2015).

Experimental strategies treating human or animal cardiac fibroblasts with pro-fibrotic agonists or with hormones as angiotensin-II increased significantly the expression of TRPM7. In fact, TGF-β1 addition to human atrial fibroblasts upregulated the expression of TRPM7; meanwhile, TRPM7 silencing inhibited fibroblasts proliferation, differentiation, and collagen production induced by TGF-β1 (Du et al., 2010). Angiotensin-II also stimulated TRPM7 resulting in an increased expression of α-SMA and fibronectin protein (Yu et al., 2014). Moreover, rat cardiac fibroblasts incubated with angiotensin-II increased levels of protein expression of TRPM7, collagens I and III, which promoted fibrosis (Zhou et al., 2015). In contrast, the downregulation of TRPM7 decreased its related current density and inhibited angiotensin-II mediated cardiac fibroblasts proliferation, differentiation, and collagen synthesis (Li et al., 2017). Moreover, in rats with sick sinus syndrome, TRPM7 regulated angiotensin II-induced cardiac fibroblasts proliferation and collagen synthesis of sinoatrial node, involving Smad signaling pathway (Zhong et al., 2018b). Recently, a role of TRPM7 was reported in miRNA-135a inhibition of isoproterenol-induced cardiac fibrosis (Wu et al., 2018) and in fibrosis evoked by H2O2 and hypoxia (Guo et al., 2014; Lu et al., 2017). Therefore, TRPM7 stand out as an interesting possible target to attenuate pathological cardiac fibrosis.

#### TRPs and Conduction Disorders

Atrial fibrillation (AF) is the most sustained clinical arrhythmia, which occurs due to structural remodeling, involving prominent fibrotic changes (Yue et al., 2011). TRP channels do not influence the excitability of heart's pacemaker but apparently their upregulation mediate the arrhythmogenesis and the progression of electrical remodeling of the diseased heart (Yue et al., 2015). Special advances have been made by studying the role of TRPM4, TRPC3, and TRPM7 in conduction disorders in the heart.

TRPM4 contribution to cardiac conduction as well as the development of arrhythmias has been demonstrated using different approaches, such as channel inhibition with 9-phenantrol, a blocker for TRPM4 (Grand et al., 2008), using TRPM4-deficient mouse models, and by the identification of TRPM4 mutants detected in a variety of inherited human cardiac arrhythmias (Freichel et al., 2017). A pro-arrhythmic role of TRPM4 and its participation in membrane potential depolarization likely explain the triggering of spontaneous beating and the increase of action potential duration described in the hypertrophied heart (Demion et al., 2014). *In vitro* experiments showed that 9-phenanthrol decreased Ca2+ oscillations in atrial HL-1 mouse cardiac myocytes, thought to play a critical role in arrhythmias (Burt et al., 2013). TRPM4 inhibition with 9-phenanthrol also mimicked the reduction of action potential duration evoked by TRPM4 deletion in atrial cells (Simard et al., 2013) and reverted the early after-depolarization involved in cardiac arrhythmias observed after a process of hypoxia and re-oxygenation (Simard et al., 2012). Recently, it was shown that physiological range of [Ca2+]i could activate TRPM4, and its upregulation altered action potential characteristics in HL-1 cells treated with angiotensin-II, which increased the arrhythmic propensity of cardiac tissue in pathological situation (Hu et al., 2017). Interestingly, *trpm4* gene mutations were linked to progressive familial heart block type 1 (Kruse et al., 2009; Daumy et al., 2016), isolated cardiac conduction disease (Liu et al., 2010), atrio-ventricular block (Stallmeyer et al., 2012; Syam et al., 2016), right bundle branch block (Stallmeyer et al., 2012), Brugada syndrome (Liu et al., 2013; Gualandi et al., 2017), and recently to either complete heart block or idiopathic ventricular fibrillation (Bianchi et al., 2018).

Additional indications suggested that TRPM7 and TRPC3 might also mediate the pathogenesis of AF. In atrial fibroblasts from AF patients, TRPM7 is notably upregulated (Du et al., 2010; Zhang et al., 2012a) and was suggested to play a pivotal role in AF due to fibrogenesis in the atrium since fibrosis is the main factor for AF. TRPM7 knockdown suppressed endogenous TRPM7 currents, decreased Ca2+ influx in atrial fibroblasts, and inhibited TGF-β1-induced fibroblast proliferation, differentiation, and collagen production (Du et al., 2010). TRPC3 is also significantly upregulated in the atria of AF patients (Zhao et al., 2013). In fibroblasts freshly isolated from left atrial of dogs undergoing AF, by sustained atrial tachypacing, it was observed a significantly increase in TRPC3 protein expression, currents, ERK phosphorylation, and extracellular matrix gene expression (Harada et al., 2012). Further evidence for a role of TRPC3 has been demonstrated in experiments using TRPC3 knockout mice, in which the effect of angiotensin-II addition to pacing-induced AF mice was significantly reduced compared to wild-type mice (Ju et al., 2015). Interestingly, a recent study examined the expression of different TRPs in leukocytes of patients with nonvalvular AF (Düzen et al., 2017). They observed marked increase in gene expression of TRPC1- C7, TRPV1-V6, TRPM1-M8, TRPML1-ML3, TRPA1, and TRPP2. A possible correlation between these leukocytes genes' expression and those examined from the atrium will be of major interest. Therefore, further investigations are undoubtedly needed for understanding the role of all these TRP channels in AF.

# CONCLUSIONS AND PERSPECTIVES

A growing body of evidence has demonstrated that, by controlling Ca2+ homeostasis, different TRP isoforms are critically involved in pathological cardiac remodeling and heart failure. However, molecular mechanisms which trigger the transition of the heart from adaptation to maladaptation by these channels are still unknown. In the last two decades, the use of genetically modified animal and mice models of cardiac disease provided valuable information about TRPs implication in cardiac remodeling. Nevertheless, substantial work is still required to understand why many TRPs from different subfamilies are activated by similar pro-hypertrophic or pro-fibrotic stimuli, and whether they associate or interact between them to activate signaling pathway involved in hypertrophy, fibrosis, or conduction disorders. Unfortunately, the wide range of agonists and antagonists used to modulate TRPs failed to determine which TRPs might be the right target(s) to characterize and consider as therapeutic tools. More specific inhibitors/activators of TRPs are eagerly awaited to shed a light on the complex mechanism of cardiac diseases associated with remodeling. Of hope, the increasing amount of available information related to TRP-drug interaction sites and gating processes of TRP channels is expected to facilitate the development of novel therapeutic concepts by pharmaceutical companies. Overall, and in the light of the reported studies, TRP channels are still considered promising therapeutic targets to regulate pathological cardiac plasticity.

#### AUTHOR CONTRIBUTIONS

TS conceived the concept of the review. TS, IG-O, DF, RT, MM-B, and EC-S wrote the review. IG-O, DF, AH and JR

#### REFERENCES


designed and formatted the figures. TS, IG-O, RT, DF, AH, and JR read and edited the review manuscript.

#### FUNDING

This work was supported by Spanish Ministry of Economy and Competitiveness [BFU2016–74932-C2]; Institute of Carlos III [PI15/00203; PI16/00259; CB16/11/00431]; the Andalusia Government [PI-0313-2016]. This study was co-financed by FEDER Funds.


canine atrial fibrillation. *Int. J. Cardiol.* 168, 4417–4418. doi: 10.1016/j. ijcard.2013.05.029


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Falcón, Galeano-Otero, Calderón-Sánchez, Del Toro, Martín-Bórnez, Rosado, Hmadcha and Smani. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Calcium Signaling and Contractility in Cardiac Myocyte of Wolframin Deficient Rats

Michal Cagalinec1,2,3 \*, Alexandra Zahradníková1,2, Alexandra Zahradníková Jr.1,2 , Dominika Kovácová ˇ 4 , Ludovit Paulis4,5, Simona Kureková<sup>2</sup> , Matej Hot'ka2,6 , Jana Pavelková1,2, Mario Plaas<sup>3</sup> , Marta Novotová1,2 and Ivan Zahradník1,2

<sup>1</sup> Department of Cellular Cardiology, Institute of Experimental Endocrinology, Biomedical Research Center, University Science Park for Biomedicine, Slovak Academy of Sciences, Bratislava, Slovakia, <sup>2</sup> Institute of Molecular Physiology and Genetics, Centre of Biosciences, Slovak Academy of Sciences, Bratislava, Slovakia, <sup>3</sup> Institute of Biomedicine and Translational Medicine, Faculty of Medicine, University of Tartu, Tartu, Estonia, <sup>4</sup> Institute of Pathophysiology, Faculty of Medicine, Comenius University, Bratislava, Slovakia, <sup>5</sup> Institute of Normal and Pathological Physiology, Centre of Experimental Medicine, Slovak Academy of Sciences, Bratislava, Slovakia, <sup>6</sup> Department of Neurophysiology and Neuropharmacology, Center of Physiology and Pharmacology, Medical University of Vienna, Vienna, Austria

#### Edited by:

Gema Ruiz-Hurtado, Instituto de Investigación Hospital 12 de Octubre, Spain

#### Reviewed by:

Marcela Sorelli Carneiro-Ramos, Federal University of ABC, Brazil Senka Holzer, Medical University of Graz, Austria

> \*Correspondence: Michal Cagalinec michal.cagalinec@savba.sk

#### Specialty section:

This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology

Received: 05 October 2018 Accepted: 12 February 2019 Published: 13 March 2019

#### Citation:

Cagalinec M, Zahradníková A, Zahradníková A Jr, Kovácová D, ˇ Paulis L, Kureková S, Hot'ka M, Pavelková J, Plaas M, Novotová M and Zahradník I (2019) Calcium Signaling and Contractility in Cardiac Myocyte of Wolframin Deficient Rats. Front. Physiol. 10:172. doi: 10.3389/fphys.2019.00172 Wolframin (Wfs1) is a membrane protein of the sarco/endoplasmic reticulum. Wfs1 mutations are responsible for the Wolfram syndrome, characterized by diabetic and neurological symptoms. Although Wfs1 is expressed in cardiac muscle, its role in this tissue is not clear. We have characterized the effect of invalidation of Wfs1 on calcium signaling-related processes in isolated ventricular myocytes of exon5-Wfs1 deficient rats (Wfs1−e5/−e5) before the onset of overt disease. Calcium transients and contraction were measured in field-stimulated isolated myocytes using confocal microscopy with calcium indicator fluo-3 AM and sarcomere length detection. Calcium currents and their calcium release-dependent inactivation were characterized in wholecell patch-clamp experiments. At 4 months, Wfs1−e5/−e5 animals were euglycemic, and echocardiographic examination revealed fully compensated cardiac function. In field-stimulated isolated ventricular myocytes, both the amplitude and the duration of contraction of Wfs1−e5/−e5 animals were elevated relative to control Wfs1+/<sup>+</sup> littermates. Increased contractility of myocytes resulted largely from prolonged cytosolic calcium transients. Neither the amplitude of calcium currents nor their voltage dependence of activation differed between the two groups. Calcium currents in Wfs1−e5/−e5 myocytes showed a larger extent of inactivation by short voltage prepulses applied to selectively induce calcium release-dependent inactivation of calcium current. Neither the calcium content of the sarcoplasmic reticulum, measured by application of 20 mmol/l caffeine, nor the expression of SERCA2, determined from Western blots, differed significantly in myocytes of Wfs1−e5/−e5 animals compared to control ones. These experiments point to increased duration of calcium release in ventricular myocytes of Wfs1−e5/−e5 animals. We speculate that the lack of functional wolframin might cause changes leading to upregulation of RyR2 channels resulting in prolongation of channel openings and/or a delay in termination of calcium release.

Keywords: calcium current, calcium transient, contractility, cardiac myocyte, Wolfram syndrome, wolframin

# INTRODUCTION

fphys-10-00172 March 12, 2019 Time: 16:54 # 2

Wolfram syndrome (WS; OMIM 222300) is a rare hereditary disorder, first identified as a clinical entity separate from the juvenile type of diabetes mellitus by Wolfram and Wagener (1938). WS is characterized by symptoms including diabetes insipidus (DI), diabetes mellitus (DM), optical atrophy (OA), and deafness (D), and therefore it is termed also DIDMOAD (Barrett et al., 1995). Most WS cases are caused by recessive mutations of the Wfs1 gene, located on chromosome 4p16.1. The Wfs1 protein is highly expressed in the brain, heart and pancreatic β-cells (Inoue et al., 1998; Yamada et al., 2006); pancreas and brain represent the crucial organs responsible for most of clinical symptoms in WS. Cardiac symptoms of WS were not originally recognized; however, emerging clinical findings include heart malformations as well as sinus tachycardia, atrial or ventricular arrhythmias (Medlej et al., 2004; Fabbri et al., 2005; Ganie et al., 2011; Korkmaz et al., 2016). The high expression of Wfs1 in the heart tissue, and the cardiac symptoms identified until now suggest functional importance of Wfs1 in the heart.

Although the Wfs1 gene was identified 20 years ago (Inoue et al., 1998), the function of Wfs1 has not been resolved yet, and little is known about its 3D-structure. Wfs1 is composed of 890 amino acids (MW of ≈100 kDa), and was shown to reside on the membrane of endoplasmic reticulum (ER; Takeda et al., 2001). It was proposed to contain at least nine transmembrane helices (Hofmann et al., 2003), and the amino- and carboxy- terminals were shown to be located in the cytoplasm and in the lumen of ER, respectively. Wfs1 seems to exist predominantly as a tetramer (Hofmann et al., 2003), and ion channel activity was observed after reconstitution of microsomes of Wfs1-expressing Xenopus oocytes in lipid bilayers (Osman et al., 2003). However, homology modeling studies (Qian et al., 2015; Safarpour Lima et al., 2016) produced structures that lack clear channel-forming helices.

Wfs1 was suggested to participate in the response of cells to ER stress: In Wfs1-transfected COS7 cells, Wfs1 negatively regulated the activating transcription factor 6α (ATF6α), a key transcription factor involved in ER stress signaling, and stabilized the E3 ubiquitin ligase HRD1 (Fonseca et al., 2010). In line with this, Cagalinec et al. (2016) have shown that overexpression of Wfs1 leads to massive activation of the key factors of ER stress, namely, ATF6, ATF4, and XBP1 in primary cultured rat cortical neurons. Silencing of Wfs1 by specific shRNA in neurons also induced increased expression of these factors but only to a moderate extent. ER stress caused by Wfs1 deficiency was implicated also in the disruption of β-cell function (Morikawa et al., 2017; Ohta et al., 2017). Wfs1 was also suggested to participate in calcium handling: silencing of Wfs1 by specific shRNA in neurons resulted in depression of calcium transients and of Ca2<sup>+</sup> release from the ER (Cagalinec et al., 2016). Moreover, expression of Wfs1 in HEK293 cells has been shown to positively modulate Ca2<sup>+</sup> levels in the ER by increasing the rate of Ca2<sup>+</sup> uptake (Takei et al., 2006). In addition, Wfs1 co-immunoprecipitates with SERCA, the pump transporting Ca2<sup>+</sup> from cytosol to lumen of the reticular membrane system (Zatyka et al., 2015). Depletion of Wfs1 led to decreased and delayed cytosolic Ca2<sup>+</sup> elevations in response to glucose stimuli (Ishihara et al., 2004) and to increased expression of SERCA in β-cells and β-cell lines (Zatyka et al., 2015). Wfs1 has been shown as a molecular partner of calmodulin (Yurimoto et al., 2009) and affected the function of the calcium-dependent protease calpain2 (Lu et al., 2014). In addition, it has been demonstrated recently that Wfs1 forms a complex with neuronal calcium sensor 1 (NCS1) and inositol 1,4,5-trisphosphate receptor (IP3R) to promote Ca2<sup>+</sup> transfer between the ER and mitochondria in WS patient fibroblasts (Angebault et al., 2018). All these facts demonstrate strong involvement of Wfs1 in calcium signaling and ER-stress mediated pathways.

To understand the role of Wfs1 on the cellular, organ and body level, a Wfs1 deficient mouse model (Luuk et al., 2009) and a Wfs1 loss-of-function rat model (exon5-Wfs1 deficient; Wfs1−e5/−e5; Plaas et al., 2017) were developed. In the latter, DM develops at the age of 13 months (Plaas et al., 2017).

Since calcium ions are the sole trigger for myocyte contraction and Wfs1 is strongly involved in calcium metabolism, in this work we evaluated calcium transients and contractility of left ventricular myocytes freshly isolated from the exon5-Wfs1 deficient (Wfs1−e5/−e5) rats. To assess whether the cardiac complications in this model develop independently from insulin deficiency, in this study we have studied animals at the age of 4 months, i.e., before the development of hyperglycemia.

# MATERIALS AND METHODS

#### Experimental Model

The impact of Wfs1 malfunction on calcium signaling in cardiac myocytes was studied using the Wfs1 loss-of-function (exon5 deficiency, Wfs1−e5/−e5) rat model (Plaas et al., 2017). We used animals at the age of 4 months to avoid complications due to involvement of diabetes.

Rats were kept in groups of two to three in polypropylene cages under a standard 12:12 h light/dark regime at a temperature of 22 ± 2 ◦C and 60–70% humidity. A standard balanced pellet diet and tap water were provided ad libitum. The genotype of rats in the control group (Wfs1+/+; n = 17) and the experimental group (Wfs1−e5/−e5; n = 19) was verified by PCR (see below). The number of animals used in individual experiments is given together with the data. All experiments conformed to the European directive 2010/63/EU and to Act No. 377/2012 of the Government of the Slovak Republic, were carried out in compliance with the guidelines for the care and use of laboratory animals, and were approved by the Ethical committee of the Centre of Biosciences, Slovak Academy of Sciences and by the State Veterinary and Food Administration of the Slovak Republic (approval No. Ro-1007/16-221 and Ro-2498/18-221a). Rats were sacrificed under full anesthesia by exsanguination. Before chest opening, rats were heparinized (5000 U/kg i.p.) and deeply anesthetized with sodium pentobarbital (100 mg/kg i.p.) until cessation of the paw withdrawal reflex and corneal reflex. Excised hearts attached to the cannula were mounted on the Langendorff set-up for isolation of myocytes, or for whole heart fixation.

The genotype of rats was verified using primers from Plaas et al. (2017), namely, rwfs\_zf\_genoR1 (5<sup>0</sup> - AAGAGTGGGTATGGTGCTGG-3<sup>0</sup> ) and rwfs\_zf\_genoF1 (50 -AGAAGTGGCTACCCAGGGAT-3<sup>0</sup> ). The PCR product was separated in 1.5% agarose gel. A single, 333 (Wfs1+/+) or 149 bp (Wfs1−e5/−e5) long band was detected in every animal.

## Echocardiography

fphys-10-00172 March 12, 2019 Time: 16:54 # 3

Transthoracic echocardiography (Liu and Rigel, 2009) was carried out before cardiotomy and isolation of cardiac myocytes. The examination was performed using a GE Medical Vivid E9 (GE Healthcare, Horten, Norway) using a 10-MHz matrix probe (ML6-15). The rat under sodium pentobarbital anesthesia was placed in supine position on a warming pad (37◦C) and the anterior chest was shaved. Left ventricular function and structure were assessed from parasternal short axis views and from the four-chamber apical view. LV end-systolic and end-diastolic diameters and posterior wall thickness during systole and diastole were measured from two-dimensional anatomical M-mode in parasternal short axis view by the leading edge method. LV ejection fraction was calculated by using the Teichholz formula. Aortic outflow was acquired using pulsed wave Doppler. Left ventricular diastolic function was assessed using transmitral flow parameters—the ratio of peak velocity of early (E) and late (A) diastolic filling—from the four-chamber apical view with conventional pulsed wave Doppler. Each measurement was obtained from three consecutive cardiac cycles.

#### Blood Glucose Measurements

Blood samples were taken from the tail vein on the day of the experiment after induction of anesthesia and before echocardiographic examination. Glycemia was measured by the glucose oxidase method (Beckman, United States) as described previously (Kurdiova et al., 2014).

#### Solutions

Ripa buffer (in mmol/l): 150 NaCl, 25 Trizma base (pH 7.6), supplemented with 1% deoxycholate, 0.1% SDS, 1% Triton X-100, and 50 µl/ml protease inhibitor cocktail (Roche Diagnostics, Indianapolis).

PBST solution (in mmol/l): 150 NaCl, 3 KCl, 2 KH2PO4, 8 Na2HPO4, supplemented with 0.1% Tween-20.

1Ca Tyrode solution (in mmol/l): 135 NaCl, 5.4 KCl, 5 MgCl2, 1 CaCl2, 0.33 NaH2PO4, 10 HEPES; pH adjusted to 7.25 with 1 mol/l NaOH, osmolarity 300 mosm/l.

0Ca Tyrode solution (in mmol/l): 135 NaCl, 5.4 KCl, 5 MgCl2, 0.003 CaCl2, 0.33 NaH2PO4, 10.0 HEPES; pH adjusted to 7.25 with 1 mol/l NaOH, osmolarity 300 mosm/l.

Supplemented 0Ca Tyrode solution was prepared from 0Ca Tyrode solution by adding (in mmol/l): 10 glucose, 10 creatine, 10 taurine; pH adjusted to 7.25 with 1 mol/l NaOH, osmolarity 330 mosm/l.

Low-calcium Tyrode solution was prepared from the Supplemented 0Ca Tyrode solution by adding 50 µmol/l CaCl2.

Enzymatic solution was prepared from the low-calcium Tyrode solution by adding 0.05–0.2 mg/ml Liberase TM (Roche, Cat. No. 05 401 127 001).

External bath solution (in mmol/l): 135 NaCl, 5.4 CsCl, 10 HEPES, 5 MgCl2, 0.33 NaH2PO4, 1 CaCl2, 0.01 IBMX, 0.02 TTX.

Internal solution (in mmol/l): 135 CsCH3SO3, 10 CsCl, 10 HEPES, 3 MgSO4, 3 ATPNa2, 0.05 cAMP, 1 EGTA, and 0.1 Fluo-3.

Perfusion solution was prepared from the Supplemented 0Ca Tyrode by adding an aliquot of 1 mol/l CaCl<sup>2</sup> stock solution (Merck) to a final concentration of 1.2 mmol/l CaCl2.

Caffeine solution was prepared from Perfusion solution by adding caffeine at a final concentration of 20 mmol/l.

#### Western Blotting

Ventricular cardiomyocytes were lyzed in Ripa buffer. The samples were incubated for 45 min on ice. Total extracts were cleared by centrifugation for 10 min at 4◦C at 14,000 × g and assayed for protein content by Lowry's method. Aliquots from each cell lysate containing 25 µg of protein were separated by SDS-PAGE on an 8% gel and transferred to nitrocellulose membrane. Membranes were blocked for 2 h with 5% skim milk in PBST solution, and then incubated with the primary antibody (suspended in 5% skim milk-PBST) overnight or 4 h at room temperature. Primary antibodies used were anti-SERCA2 (1:1000, SC-376235, Santa Cruz) and anti-GAPDH (1:1000, MAB374, Merck-Millipore). Membranes were washed with PBST (three times, 10 min each) and incubated with HRP anti-mouse antibody (1:2000, W4021, Promega) for 1.5 h at room temperature. The bands were visualized with a chemiluminescence immunodetection system (Amersham Biosciences, Piscataway, NJ, United States). Images were analyzed with the Origin software (OriginLab, Ver. 9) by subtracting the background, integrating the signal in each band, and normalizing the SERCA signal to the GAPDH signal in each lane.

#### Electron Microscopy

Samples of left ventricular muscle were prepared for morphological analysis as described previously (Mikusová et al., 2009). In brief, the hearts of experimental animals in deep anesthesia were rapidly excised, mounted on a Langendorff setup, and perfused with 1Ca Tyrode solution for 3 min, with 0Ca Tyrode solution for 3 min, and finally with 2% glutaraldehyde in cacodylate buffer (in mmol/l: 150 Nacacodylate, 2 CaCl2, pH 7.3) for 5 min. Three samples of left ventricular muscle per animal (five animals per group) were dissected, exposed to 2% glutaraldehyde in cacodylate buffer for 45 min, postfixed by 1% osmium tetroxide in cacodylate buffer for 30 min, and contrasted with 1% aqueous solution of uranyl acetate. After dehydration in graded ethanol series and propylene oxide, the samples were embedded in Durcupan blocks (ACM Fluka, Sweden). Ultrathin (58–60 nm) longitudinal sections were cut from each block at three distant levels using an ultramicrotome (Power-Tome MT-XL, RMC/Sorvall, Tucson, United States) and mounted on formvar-coated copper grids. The sections were contrasted with lead citrate and examined with a JEM 1200 electron microscope (Jeol, Tokyo, Japan) at 80 kV. The images were recorded with a CCD camera (Gatan DualVision 300W). The endomembrane distribution and ultrastructure were studied at magnification of 2,000–100,000×.

#### Isolation of Cardiac Myocytes

fphys-10-00172 March 12, 2019 Time: 16:54 # 4

Left ventricular cardiac myocytes were isolated from 6 Wfs1+/<sup>+</sup> and 8 Wfs1−e5/−e5 male rats at the age of 4 months. The hearts of experimental animals in deep anesthesia were rapidly excised, cannulated, and mounted on a Langendorff perfusion system. The heart was retrogradely perfused first with 1Ca Tyrode solution until the blood was completely washed out. This was then replaced with Supplemented 0Ca Tyrode solution and perfused for further 5 min. After this, the heart was perfused with Enzymatic solution until the tissue became marble redwhite (usually 6–8 min). All solutions were oxygenated and equilibrated at 37◦C. During perfusion, the heart was positioned inside a heating chamber maintaining the temperature at 37◦C. The heart was then taken off from the perfusion system, the atria and the right ventricle were discarded and the left ventricle was cut gently into several small pieces. The tissue was filtered through a nylon mesh (100 µm). The filtrate containing the dissociated myocytes was gently centrifuged at 50 × g for 1 min, the supernatant was discarded and replaced with 1.5 ml lowcalcium Tyrode solution. The tissue that remained on the filter was further enzymatically digested for further 5 min at 37◦C with Enzyme solution, in which collagenase activity was reduced using BSA (1% w./v.). This procedure was repeated four to five times. The batch with the highest yield of viable myocytes was then processed further by increasing the concentration of CaCl<sup>2</sup> in low-calcium Tyrode solution to 100, 500, and 1000 µmol/l in three 15-min steps to maintain the myocytes calcium tolerant. All experiments were performed within 8 h after isolation.

#### Patch Clamp

Myocytes were whole-cell patch clamped using patch-pipettes of 1.6–1.8 M filled with Internal solution and a VE-2 amplifier (Alembic Instruments, Canada), a Digidata 1320A A/D converter, and pClamp software (Ver. 10, all from Axon Instruments, United States). Cell capacitance and series resistance were compensated to 50–85%. Myocytes were kept at a −50 mV holding potential. The myocytes were kept in a phosphorylated state by the use of ATP and cAMP in the Internal solution and of the membrane-permeable phosphodiesterase inhibitor IBMX in the External solution (Zahradnik and Palade, 1993). Calcium currents elicited by voltage pulses applied by specific protocols were filtered at 10 kHz and sampled at 20 kHz. Current–voltage curves were measured by applying a series of 70 ms test pulses to −40 to +50 mV in 10-mV increments once in 30 s. Inactivation of ICa was assessed by two types of twopulse protocols (Zahradníková et al., 2004). Protocol 1 consisted of a series of twin pulses, applied at 30-s intervals, where the prepulses (duration 5 ms, prepulse potential increasing from −50 to +40 mV by 10 mV increments) were followed by a constant 70-ms test pulse to 0 mV delivered 30 ms after the prepulse. This protocol was used to determine the voltage dependence of the voltage-, current-, and calcium release-dependent components of ICa inactivation. Protocol 2 consisted of a series of twin pulses, applied at 30-s intervals, where the prepulses (prepulse potential +60 mV, duration increasing from 0 to 4.5 ms by 0.5 ms) were followed by a constant 70-ms test pulse to 0 mV delivered 30 ms after the prepulse. This protocol served for determination of the calcium release-dependent component of ICa inactivation. In both inactivation protocols, the fraction of non-inactivated calcium current was calculated for each prepulse as the ratio of the peak amplitude of test ICa to the peak amplitude of the test ICa in the absence of the prepulse.

#### Confocal Imaging

Myocytes were loaded with the calcium indicator fluo-3 AM (1 µmol/l) for 15 min at room temperature. Fluo-3-labeled myocytes suspended in the experimental chamber were left to settle on the coverglass for 1 min and then perfused at a rate of 1 ml/min with Perfusion solution and electrically paced to contract at a frequency of 1 Hz using field stimulation (32 mA, 40 ms, DS3 Isolated Current Stimulator, Digitimer Ltd., United Kingdom). The TCS SP-8 STED confocal microscope (Leica Microsystems) was used in the line scan mode (2.5 ms per line, 8000 lines, excitation 500 nm, emission 510–570 nm) to record fluo-3 fluorescence images in parallel with the transmission images. Thus, calcium transients and the corresponding myocyte shortenings could be assessed simultaneously (**Figure 1**). In a subset of experiments, rapid perfusion with Caffeine solution was applied after the series of electrical stimuli to evoke maximal calcium release and to assess the sarcoplasmic reticulum calcium content.

The transmission images were used to determine changes in sarcomere length using the SarConfoCal plugin (Pasqualin et al., 2017) in the Fiji environment (Schindelin et al., 2012) of ImageJ (Schneider et al., 2012). The adjacent averaging filter was set to 17. Only the intracellular regions of the images were analyzed.

The time courses of fluorescence and sarcomere length were normalized to the diastolic values (1F/F0, 1L/L0), and imported into the Clampfit software (pClamp 10, Molecular Devices, United States). Here, the response to each stimulus, i.e., the calcium transient and its corresponding sarcomere length transient was analyzed. The following parameters were determined for calcium transient and sarcomere length transient: baseline, amplitude, half-width, rise time (20–80%), and decay time (80–20%). Smoothing window was set to five samples.

#### Data Processing

Current–voltage data were fitted by the equation

$$I\_{\rm Ca}(V) = \rm g\_{\rm max}(V - V\_{\rm r})/(1 + \exp(-(V - V\_{1/2a})/V\_{\rm Sa}))^4 \tag{1}$$

where ICa is the peak calcium current amplitude, V is the test pulse voltage, gmax is the maximal conductance, V<sup>r</sup> is the reversal potential, V1/2<sup>a</sup> is the potential of half-activation, and VSa is the slope factor of activation.

Inactivation curves of the two-pulse inactivation protocols were fitted by the respective Boltzmann functions:

$$I\_{\rm Ca}(V\_{\rm p}) = -\rm F\_{\rm iV}/(1 + \exp((V\_{\rm p} - V\_{1/2i})/V\_{\rm SiV})) + \rm F\_{\rm iV} \tag{2}$$

where V<sup>p</sup> is the prepulse voltage, FiV is the inactivated fraction of the peak ICa amplitude at the test pulse, V1/2i is the halfinactivation potential, and VSiV is the slope factor, and

$$I\_{\rm Ca}(t\_{\rm p}) = -\rm F\_{\rm it}/(1 + \exp((t\_{\rm p} - t\_{1/2i})/V\_{\rm sit})) + \rm F\_{\rm it} \tag{3}$$

where t<sup>p</sup> is the prepulse duration, Fit is the inactivated fraction of the peak ICa amplitude at the test pulse, t1/2<sup>i</sup> is the prepulse duration for half-inactivation, and VSit is the slope factor.

Statistical analysis was performed in Origin (OriginLab, Ver. 9) using Student's t-test. The P-values were calculated for twotailed data distributions.

Data fitting was performed in Origin using the Levenberg– Marquardt method. The fitted parameters of two datasets were considered significantly different if their 95% confidence intervals did not overlap.

#### RESULTS

#### Invalidation of Wolframin Does Not Lead to Heart Disease or Diabetes at the Age of 4 Months

The genotype of all experimental animals was verified using PCR. Overall examination of animals and their hearts excised for experiments did not show malformations or differences in comparison with standard rats. The weight of Wfs1−e5/−e<sup>5</sup> rats taken into experiments was significantly smaller (524 ± 11 g, n = 19, P < 0.05) than that of their Wfs1+/<sup>+</sup> littermates (570 ± 19 g, n = 17). Similar values and differences in the body weight at the age of 4 months were reported by Plaas et al. (2017). Tests for postprandial blood glucose showed slightly but significantly higher glycemia in Wfs1−e5/−e5 (7.79 ± 0.47 mmol/l, n = 8) than in Wfs1+/<sup>+</sup> animals (6.10 ± 0.16, n = 6, P = 0.009) but no hyperglycemia in either group, in agreement with the original study for this age group (Plaas et al., 2017). This confirms that the cohort of experimental animals used in this study was not biased by specific or unspecific side effects related to breeding.

Echocardiographic examination did not reveal myocardial hypertrophy, infarction, or other major changes in the studied hearts. Quantitative evaluation provided similar parameters in both groups (n = 6 for Wfs1+/<sup>+</sup> and n = 8 for Wfs1−e5/−e5 , **Table 1**). The posterior wall thicknesses, either diastolic or systolic, were unchanged. Functional parameters showed some small differences between groups. The heart rate was somewhat higher, while the enddiastolic and stroke volumes were somewhat smaller in Wfs1−e5/−e5 than in Wfs1+/<sup>+</sup> animals; these differences did not reach statistical significance. The aortic valve maximal flow velocity was significantly higher in Wfs1−e5/−e5 than in Wfs1+/<sup>+</sup> animals, although its time integral was not different in the two groups. Based on measurements of the ejection fraction, which did not show significant differences between the two groups of animals, systolic function in

#### TABLE 1 | Echocardiographic parameters.

fphys-10-00172 March 12, 2019 Time: 16:54 # 6


PWd – posterior wall thickness (diastolic), PWs – posterior wall thickness (systolic), EDV – end diastolic volume, SV – stroke volume, EF – ejection fraction, AVmax – aortic valve maximal flow velocity, AV VTI – aortic valve velocity time integral, MV E/A – mitral valve early (E)/late (A) peak velocity ratio, n – the number of animals, P – P-value of the t-test for two-tailed distribution.

#### TABLE 2 | Parameters of cell shortening.


n – The number of analyzed cells. Data are given as mean ± SEM. P – P-value of the t-test for two-tailed distribution.

#### TABLE 3 | Parameters of calcium transients.


n – The number of analyzed cells. Data are given as mean ± SEM. P – P-value of the t-test for two-tailed distribution.

Wfs1−e5/−e5 and Wfs1+/<sup>+</sup> animals was fully preserved. At the organ level, therefore, the heart function of Wfs1−e5/−e5 animals can be considered compensated at this stage of disease development.

#### Invalidation of Wolframin Leads to Elevated Contraction and Prolonged Calcium Transients

Contractile and calcium transient responses of isolated cardiac myocytes to field or caffeine stimulation (six Wfs1+/<sup>+</sup> and eight Wfs1−e5/−e5 rats) were evaluated by optical methods of transmission and fluorescence microscopy, respectively (**Figure 1**). The resting sarcomere length was the same in both studied groups (**Table 2**); however, the contractile responses were different. The amplitude of peak contraction as well as the duration of contraction at the half-peak amplitude were significantly higher in Wfs1−e5/−e5 than in Wfs1+/<sup>+</sup> myocytes. Since the rise time and the decay time of contraction did not change, prolongation of contraction resulted from prolonged duration of the plateau. Since contractions were recorded in the unloaded isotonic mode, they do not reflect metabolic or energy supply aspects of myocyte function.

In contrast to contractions, the respective peak amplitudes of stimulated calcium transients in Wfs1−e5/−e5 myocytes did not differ from those in Wfs1+/<sup>+</sup> myocytes; however, their duration in Wfs1−e5/−e5 myocytes was significantly longer than in Wfs1+/<sup>+</sup> myocytes (**Table 3**). The prolongation of contractions correlated well with prolongation of calcium transients, resulting in similar increases in FDHM (10 vs. 12%, respectively, **Tables 2**, **3**).

Stimulation of isolated myocytes by caffeine is a useful tool for estimation of calcium content of sarcoplasmic reticulum. Calcium transients in response to caffeine application were not different between groups (**Table 3**), indicating similar calcium content of the sarcoplasmic reticulum. Caffeine-induced

contractures were always of supramaximal amplitude (unloaded cells) and therefore their kinetics were not evaluated.

# Invalidation of Wolframin Affects Release-Dependent Inactivation of Calcium Channels

The change in contractility could result from a change of the calcium current (the trigger for calcium release) and/or from a change of calcium release itself. We estimated both calcium signals in isolated cardiac myocytes from 6 Wfs1+/<sup>+</sup> and 8 Wfs1−e5/−e5 rats by means of a voltage-clamp experiment (Zahradníková et al., 2004). The voltage dependence of calcium currents is shown in **Figure 2A** together with the fitted curves according to Eq. 1. No statistically significant differences between parameters of the current–voltage curves of the two myocyte groups were observed. The parameters of the fit are summarized in **Table 4**.

The efficiency of calcium current to induce calcium release was analyzed according to Zahradníková et al. (2004). The extent of calcium release was estimated from the extent of calcium releasedependent inactivation of calcium current. The fraction of inactivated current (Fi) was assessed by two two-pulse protocols. First, inactivation of ICa was assessed by Protocol 1, in which the potential of the 5-ms prepulse was varied. The dependence of FiV on prepulse potential is shown in **Figure 2B**, and parameters of the curves fitted using Eq. 2 are given in **Table 4**. The voltage parameters of inactivation were not significantly different but the fraction FiV of inactivated calcium current was significantly larger in Wfs1−e5/−e5 than in Wfs1+/<sup>+</sup> myocytes. In the second

FIGURE 2 | Analysis of calcium currents. The parameters of the fitted curves are in Table 4. (A) Voltage dependence of the peak calcium current amplitudes. The curves represent the respective fits by Eq. 1. (B) Fractional inactivation of the test pulse calcium current amplitude by variable prepulse voltages of 5-ms duration. The curves represent the respective fits by Eq. 2. (C) Fractional inactivation of the test pulse calcium current amplitude by prepulses to +60 mV of variable duration. The curves represent the respective fits by Eq. 3. In all panels, black – Wfs1+/+, gray – Wfs1-e5/-e<sup>5</sup> . Data are given as mean ± SEM (the number of cells n = 4, 4, and 2 for A, B, and C, respectively).

TABLE 4 | Parameters of calcium currents and the release dependent inactivation depicted in Figure 2.


For Figure 2A: the best fit parameters using Eq. 1: gmax – conductance, V<sup>r</sup> – reversal potential, V1/2a – potential of half-activation, VSa – slope factor of the voltagedependent activation.

For Figure 2B: the best fit parameters using Eq. 2: FiV – maximum fraction of ICa inactivated by 5-ms prepulses, V1/2i – prepulse potential for half-inactivation, VSiV – slope factor of the inactivation curve.

For Figure 2C: the best fit parameters using Eq. 3: Fit – maximum fraction of ICa inactivated by +60-mV prepulses, t1/2i – prepulse duration for half-inactivation, VSit – slope factor of the inactivation curve.

n – The number of fitted data points. <sup>∗</sup>Significantly different from Wfs1+/+.

two-pulse protocol experiment, calcium release was triggered by prepulses to +60 mV with variable duration that activated brief tail calcium currents of variable amplitude, known to induce only the calcium release-dependent component of inactivation (Zahradníková et al., 2004). The subsequent test pulse to 0 mV was used to assess the amplitude of calcium current that was not inactivated by calcium release (Protocol 2). The dependence of the fraction of inactivated current, Fit, on the prepulse duration is shown in **Figure 2C** and the parameters of the curves fitted using Eq. 3 are shown in **Table 4**. The parameters of inactivation were not significantly different between the two myocyte groups except for the fraction Fit, which was again significantly elevated in Wfs1−e5/−e5 myocytes.

These data show that the fraction of calcium current inactivated by short prepulses is increased in Wfs1−e5/−e5 animals, despite no observed changes in the amplitude or the voltage-dependence of the calcium current itself. Since neither the half-inactivation nor slope factors of prepulseinduced ICa inactivation were changed, these data suggest that the voltage dependence of ICa inactivation was not affected by wolframin invalidation. To summarize, these two-pulse experiments revealed an increased effect of calcium release on calcium current inactivation in Wfs1−e5/−e5 myocytes relative to Wfs1+/<sup>+</sup> myocytes. This could be caused only by increased calcium release at individual dyads.

## Invalidation of Wolframin Does Not Affect Ultrastructure of Cardiomyocytes

Electron microscopic examination of ultrastructure of cardiomyocytes (five Wfs1+/<sup>+</sup> and five Wfs1−e5/−e5 rats) revealed very similar morphology in both groups of experimental animals (**Figure 3**). The myocytes displayed standard shape, contractile myofibrils were well organized, and mitochondria formed columns between myofibrils. The ribosomes were present in the cytosol between mitochondria and myofibrils. Both the rough and smooth reticular membrane systems occurred in the intermyofibrillar space. Cardiomyocytes of both groups contained dyads near Z-lines. According to visual examination, the diameter of t-tubule profiles in dyads seemed somewhat different between groups, albeit not substantially (**Figures 3A,B**). The two groups seem to differ by higher occurrence of small non-dyadic t-tubules near Z-lines in the Wfs1−e5/−e5 rats in comparison to Wfs1+/<sup>+</sup> (**Figures 3C,D**).

#### Invalidation of Wolframin Does Not Affect Protein Expression of SERCA

To determine whether invalidation of wolframin affects expression of SERCA, its protein levels in myocytes (six Wfs1+/<sup>+</sup> and eight Wfs1−e5/−e5 rats) were determined by western blots using the housekeeper protein GAPDH as a loading control (**Figure 4**). The analysis yielded values of 0.44 ± 0.07 vs. 0.45 ± 0.06 a.u. for Wfs1+/<sup>+</sup> and Wfs1−e5/−e5, respectively (mean ± SEM, n = 4), indicating no difference between protein expression of SERCA of either group.

#### DISCUSSION

In this work, we studied the effect of invalidation of wolframin on cardiac function using exon5-Wfs1 deficient rats (Wfs1−e5/−e5),

with emphasis on calcium signaling related to contractility of cardiac myocytes. At the level of whole heart, we used echocardiography to assess cardiac functional parameters. At the level of isolated ventricular myocytes, we characterized cell contraction and intracellular calcium transients together with calcium currents. In addition to genotyping and blood glucose level verification, we examined changes in the expression of SERCA2 and in the ultrastructure of the left ventricular myocytes.

This is to our knowledge the first study of the effect of wolframin invalidation on calcium handling in cardiac myocytes. Previous works on wolframin were performed mostly in pancreatic β-cells and neurons or in cell lines. In those studies, the changes in cell function induced by wolframin invalidation or silencing went hand-in-hand with induction of ER stress (Takei et al., 2006; Hara et al., 2014; Zatyka et al., 2015; Cagalinec et al., 2016; Morikawa et al., 2017), with suppression of SERCA activity (Takei et al., 2006; Hara et al., 2014; Morikawa et al., 2017) and mRNA expression (Morikawa et al., 2017). Paradoxically, increase of SERCA protein expression was also observed (Zatyka et al., 2015). It has been proposed that the role of wolframin in pancreatic β-cells is to target SERCA to proteasome-mediated degradation (Zatyka et al., 2015). On the other hand, induction of ER stress by wolframin deficiency was not observed in either cardiac or skeletal muscle cells under conditions, when ER stress was clearly present in pancreatic β-cells (Yamada et al., 2006). It should be noted that in β-cells the role of SERCA is to damp the amplitude of cytosolic calcium oscillations during cell depolarization by sequestration to the ER (Félix-Martínez and Godínez-Fernández, 2014). In cardiac myocytes, SERCA primes calcium stores for stimulated calcium release and it returns calcium back to the store after the stimulus. It is therefore plausible that the role of wolframin in cardiac myocytes differs from that in pancreatic β-cells and neurons.

Ultrasonographic examination showed that the function of Wfs1−e5/−e5 hearts was fully compensated at this age of WS development, despite the lower body weight of Wfs1−e5/−e5 rats relative to control littermates. Additionally, no overt difference in the ultrastructure of ventricular myocytes or SERCA2 expression was observed between the two groups in our study. These findings comply with previous findings that symptoms of WS develop in Wfs1−e5/−e5 rats at the age of 13 months (Plaas et al., 2017). However, we have observed several differences in the function of cardiomyocytes between the two groups. Invalidation of wolframin led to increased release of calcium measured either as fractional inactivation of calcium current, or as prolongation of cytosolic calcium transients and myocyte contraction, or as increased amplitude of cell contraction in response to field stimulation. Interestingly, the amplitudes of calcium transients induced by caffeine were not different, suggesting that the SR content was similar in both myocyte groups. Together with no difference in expression of SERCA2, this indicates no effect of wolframin invalidation on calcium accumulation in the SR of Wfs1−e5/−e5 myocytes at the studied age of rats.

Prolonged cytosolic calcium transients at the same maximal intensity and prolonged contraction observed in Wfs1−e5/−e5 myocytes can be explained by the local control mechanism of calcium release (Cannell et al., 1994; Zahradníková et al., 2010; Janícek et al., 2012 ˇ ). According to this, the amplitude and the rate of cytosolic calcium increase result from temporal summation of individual release events arising at dyads and the number of dyads recruited from their cellular pool. The resulting contraction convolves with a certain delay the cytosolic calcium increase, the supply of ATP to drive the myosin motors, and sequestration of calcium to SR by SERCA.

It could be that in the field-stimulated isolated myocytes the observed differences occurred in part due to longer action potential duration in Wfs1−e5/−e5 myocytes. In this respect, patch-clamp experiments showed similar voltage dependent characteristics of calcium currents in both groups, i.e., no change in function of L-type calcium channels that could cause action potential prolongation. To the contrary, we observed increased calcium current inactivation due to calcium induced calcium release mechanism (Sham, 1997; Zahradníková et al., 2004) that should contribute to faster termination of the action potential. However, we could not exclude changes in potassium currents controlling termination of action potentials.

Our results show that calcium release in Wfs1−e5/−e5 is prolonged relative to that of Wfs1+/<sup>+</sup> myocytes under conditions of identical activating stimuli. We speculate that the longer duration but unchanged intensity of calcium release might arise by a yet unknown mechanism from the regulatory effect of wolframin on RyR2 channel open time and/or termination of calcium release, or on modulation of the calcium signal by mitochondria. All of these pathways are plausible, since wolframin was shown to interact with calmodulin (Yurimoto et al., 2009), a regulator of RyR2 activity with a potential role in RyR refractoriness (Liu et al., 2018), and since wolframin deficiency affected mitochondrial function in cortical neurons (Cagalinec et al., 2016).

#### CONCLUSION

We have shown that invalidation of Wfs1 results in subtle changes in calcium signaling, present before overt onset of disease. These subtle changes result in significant augmentation of both amplitude and duration of contraction. However, the sensitivity of our methods did not allow

us to find the molecular basis of the observed changes. To resolve this question, further studies are necessary, such as direct measurements of calcium release flux from individual dyads, and/or analysis of expression and function of all proteins involved in calcium cycling. It is tempting to speculate on wolframin regulation of RyR2 gating, e.g., by prolongation of channel openings or by delaying calcium release termination. Moreover, since the structure, density, size, and distribution of dyads was recognized as an important player in proper function of cardiac calcium release, the role of wolframin in their formation is worth further study.

#### DATA AVAILABILITY

The datasets generated for this study are available on request to the corresponding author.

#### AUTHOR CONTRIBUTIONS

MC, AZ, and IZ conceived of or designed the study. MC and AZ contributed to measurements of calcium transients and myocyte shortening. AZjr and MH contributed to electrophysiology. MN and IZ contributed to electron microscopy. DK and LP contributed to echocardiography. SK and AZjr contributed to western blotting. MC, JP, AZ, AZjr, and IZ analyzed the data. AZ, IZ, and MC wrote the manuscript. MP provided materials.

# REFERENCES


MC, AZ, AZjr, and IZ edited the manuscript for important intellectual content.

# FUNDING

The work was supported by the SASPRO grant no. 0063/01/02, financed by the Slovak Academy of Sciences and by a "Cofinancing of regional, national, and international programs (COFUND)," which is part of the Marie Curie Action of the EU 7th Framework Program, under Grant Agreement No. 609427; Slovak Research and Development Agency grant APVV 15-0302; Science Grant Agency grants VEGA 2/0169/16, VEGA 2/0143/17; and by the project "Completing the infrastructure for modern research of lifestyle diseases" (ITMS 26230120006), funded by ERDF.

#### ACKNOWLEDGMENTS

We thank the Centre of Experimental Medicine SAS – Institute for Heart Research for providing the GE Medical Vivid E9 Echocardiograph. We thank M. Nemec, Laboratory of Diabetes and Metabolic Derangements, IEE BRC SAS, for determination of blood glucose levels. We thank G. Gajdošíková for animal care and assistance with isolation of cardiac myocytes. We also thank L. Novota for technical assistance with electron microscopy.



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Cagalinec, Zahradníková, Zahradníková, Kováˇcová, Paulis, Kureková, Hot'ka, Pavelková, Plaas, Novotová and Zahradník. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Impaired Activity of Ryanodine Receptors Contributes to Calcium Mishandling in Cardiomyocytes of Metabolic Syndrome Rats

Gaudencio Fernández-Miranda† , Tatiana Romero-Garcia† , Tarín P. Barrera-Lechuga† , Martha Mercado-Morales and Angélica Rueda\*

Departamento de Bioquímica, Centro de Investigación y de Estudios Avanzados del IPN (CINVESTAV), Mexico City, Mexico

#### Edited by:

Tarik Smani, Universidad de Sevilla, Spain

#### Reviewed by:

Aleksey V. Zima, Loyola University Chicago, United States Alicia Mattiazzi, Centro de Investigaciones Cardiovasculares, Argentina

#### \*Correspondence:

Angélica Rueda arueda@cinvestav.mx

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology

Received: 14 December 2018 Accepted: 11 April 2019 Published: 30 April 2019

#### Citation:

Fernández-Miranda G, Romero-Garcia T, Barrera-Lechuga TP, Mercado-Morales M and Rueda A (2019) Impaired Activity of Ryanodine Receptors Contributes to Calcium Mishandling in Cardiomyocytes of Metabolic Syndrome Rats. Front. Physiol. 10:520. doi: 10.3389/fphys.2019.00520 Metabolic syndrome (MetS) has become a global epidemic. MetS is a serious health problem because of its related cardiovascular complications, which include hypertension and delayed heart rate recovery after exercise. The molecular bases of cardiac dysfunction in MetS are still under scrutiny and may be related to anomalies in the activity and expression of key proteins involved in the cardiac excitation–contraction coupling (ECC). The cardiac Ca2<sup>+</sup> channel/ryanodine receptor (RyR2) participates in releasing Ca2<sup>+</sup> from internal stores and plays a key role in the modulation of ECC. We examined alterations in expression, phosphorylation status, Ca2<sup>+</sup> sensitivity, and in situ function (by measuring Ca2<sup>+</sup> sparks and Ca2<sup>+</sup> transients) of RyR2; alterations in these characteristics could help to explain the Ca2<sup>+</sup> handling disturbances in MetS cardiomyocytes. MetS was induced in rats by adding commercially refined sugar (30% sucrose) to their drinking water for 24 weeks. Cardiomyocytes of MetS rats displayed decreased Ca2<sup>+</sup> transient amplitude and cell contractility at all stimulation frequencies. Quiescent MetS cardiomyocytes showed a decrease in Ca2<sup>+</sup> spark frequency, amplitude, and spark-mediated Ca2<sup>+</sup> leak. The [3H]-ryanodine binding data showed that functionally active RyRs are significantly diminished in MetS heart microsomes; and exhibited rapid Ca2+-induced inactivation. The phosphorylation of corresponding Ser2814 (a preferential target for CaMKII) of the hRyR2 was significantly diminished. RyR2 protein expression and Ser2808 phosphorylation level were both unchanged. Further, we demonstrated that cardiomyocyte Ca2<sup>+</sup> mishandling was associated with reduced SERCA pump activity due to decreased Thr17-PLN phosphorylation, suggesting a downregulation of CaMKII in MetS hearts, though the SR Ca2<sup>+</sup> load remained unchanged. The reduction in the phosphorylation level of RyR2 at Ser2814 decreases RyR2 availability for activation during ECC. In conclusion, the impaired in situ activity of RyR2 may also account for the poor overall cardiac outcome reported in MetS patients; hence, the SERCA pump and RyR2 are both attractive potential targets for future therapies.

Keywords: ryanodine receptor, calcium sparks, ryanodine binding, SERCA pump, cardiomyocytes, calcium mishandling, metabolic syndrome

# INTRODUCTION

fphys-10-00520 April 30, 2019 Time: 14:59 # 2

Metabolic syndrome (MetS) is a cluster of biochemical and physiological risk factors for cardiovascular disease and diabetes mellitus type 2 (DM2); it represents a severe public health problem around the world (Alberti et al., 2009). Risk factors for MetS include obesity (particularly central obesity), elevated triglyceride (TG) levels, low high-density lipoprotein cholesterol (HDL-C) levels, high blood pressure, and dysglycemia. Insulin resistance is considered to be the critical factor underlying MetS, though the pathogenesis remains unclear (Alberti et al., 2009). The definition of what clinically constitutes MetS has generated considerable debate; however, it is generally accepted that a combination of at least three or more of the risk factors must exist to diagnose MetS (Nolan et al., 2017; Tune et al., 2017). People diagnosed with MetS exhibit delayed heart rate recovery after exercise and 50–60% higher cardiovascular risk than those without it (Sung et al., 2006; Qiao et al., 2007). Specifically, MetS is associated with a twofold increase in cardiovascular mortality, myocardial infarction, and stroke (Mottillo et al., 2010). MetS-associated alterations in heart function include impaired myocardial contractility and diastolic dysfunction that predisposes to congestive heart failure (HF) (Tune et al., 2017). At the molecular level in cardiac cells, excitation– contraction coupling (ECC) initiates with T-tubule membrane depolarization, which induces L-type Ca2<sup>+</sup> channel (LTCC) activation. Ca2<sup>+</sup> entry is not of sufficient magnitude to activate the contractile machinery but triggers a larger Ca2<sup>+</sup> release from the stores in the sarcoplasmic reticulum (SR) via the activation of the intracellular Ca2<sup>+</sup> channel/ryanodine receptor (RyR2), thus promoting cell contraction. This process is known as Ca2+-induced Ca2<sup>+</sup> release (CICR). Relaxation takes place after Ca2<sup>+</sup> removal, primarily through its recapture into the intracellular Ca2<sup>+</sup> stores by the sarco/endoplasmic reticulum Ca2<sup>+</sup> ATPase (SERCA pump) and its extrusion by the action of both the Na+/Ca2<sup>+</sup> exchanger (NCX1) and the plasma membrane calcium ATPase (PMCA) (Fabiato, 1983; Bers, 2014). ECC defects in MetS cardiomyocytes are chiefly characterized by a slow rate of shortening and relengthening. Moreover, depressed cell shortening and slower cytosolic Ca2<sup>+</sup> clearing have been documented in cardiomyocytes of the prediabetic sucrose-fed rat model at early stages of MetS development (6–18 weeks of sucrose treatment) (Dutta et al., 2001; Hintz and Ren, 2002; Davidoff et al., 2004; Wold et al., 2005; Vasanji et al., 2006; Balderas-Villalobos et al., 2013; Okatan et al., 2016), but the participation of the RyR2 has not been fully elucidated.

The study of RyR2 dysregulation in cardiovascular diseases—including those of inherited and non-inherited etiology—has gained considerable interest in recent years due to its crucial role in the development of HF and cardiac arrhythmias. In the context of type 2 diabetic cardiomyopathy, depressed RyR2 activity is reportedly linked to a reduction in its protein expression; however, other studies did not observe changes in RyR2 at the protein level (Pereira et al., 2006, 2014). In the context of MetS, the function, expression, and phosphorylation status of cardiac RyR are still under analysis. Moreover, altered RyR phosphorylation status has been linked to the impairment of RyR activity, which leads to diabetic cardiomyopathy (Belke et al., 2004; Pereira et al., 2014). Additionally, intricate effects of both protein kinase A- (PKA) and Ca2+/Calmodulin-dependent protein kinase II- (CaMKII) mediated RyR phosphorylation have been reported in cardiomyocytes from prediabetic rats, resulting in defective RyR2 regulation (Okatan et al., 2016; Sommese et al., 2016).

Alterations in the in situ activity, expression, and regulation by phosphorylation of RyR2 have not been thoroughly evaluated in rat experimental models of MetS, particularly after 24 week treatment with 30% sucrose in drinking water, which elicits a condition resembling a chronic state of MetS in humans. Thus, we sought to examine RyR2 performance in the sucrose-induced MetS rat model. We evaluated the in situ activity of RyRs in cardiomyocytes by characterizing electrically-stimulated Ca2<sup>+</sup> transients, Ca2<sup>+</sup> spark properties, and spark-mediated Ca2<sup>+</sup> leak. We also used biochemical approaches ([3H]-Ryanodine binding and Western Blots assays) to quantify functional RyR2 and to measure the Ca2<sup>+</sup> sensitivity, protein expression, and phosphorylation status at Ser2808 and Ser2814 of RyR2 in heart homogenates and SRenriched membranes.

## MATERIALS AND METHODS

#### Development of the Sucrose-Induced Metabolic Syndrome Model

Male Wistar rats, aged 25 days, were divided into two groups and maintained under a dark–light cycle of 12 h and controlled temperature of 22 ± 2 ◦C. The first experimental group (MetS) received 30% sucrose (refined commercial sugar) in their drinking water and commercial rat chow (PicoLab Rodent Diet 20, LabDiet, St. Louis, MO, United States) ad libitum for 24 weeks. The second group (control, C) received water and commercial rat chow ad libitum during the same 24 weeks. Measurement of systolic blood pressure (SBP) was carried out by the tail-cuff method with a sensor connected to a pressure transducer and a PC equipped with special software (Grass PolyView) for data capture and processing. Recordings were taken in quadruplicate and were compared with those from invasive techniques: no difference has been found, in agreement with previous reports (Perez-Torres et al., 2009; de Alba-Aguayo et al., 2017).

#### Measurement of Serum Glucose, Triglycerides, Total Cholesterol, and HDL-Cholesterol

After fasting overnight, rats were treated with heparin (1000 units/kg) and anesthetized with sodium pentobarbital (50 mg/kg, i.p.). Anesthetized animals were subjected to a thoracotomy, during which the heart was exposed and excised

rapidly, and blood samples were collected immediately from the sectioned aorta. The serum levels of glucose, triglycerides, total cholesterol, and HDL-cholesterol were measured using glucose and lipid panel strips with the CardioCheck PA analyzer (PTS Diagnostics, Indianapolis, IN, United States).

# Cardiomyocyte Isolation

Left ventricle myocytes were enzymatically isolated from MetS and control rat hearts following a previously reported protocol (de Alba-Aguayo et al., 2017). Animals were treated with heparin (1000 units/kg) and anesthetized with sodium pentobarbital (50 mg/kg, i.p.). The heart was excised rapidly via a thoracotomy and placed in ice-cold (0◦C) oxygenated Tyrode solution containing (in mM): NaCl 130, KCl 5.4, NaH2PO<sup>4</sup> 0.4, MgCl<sup>2</sup> 0.5, glucose 22, Hepes 25 and insulin 10−<sup>3</sup> (pH 7.4 with NaOH). The aorta was cannulated above the aortic valve and heart was perfused by gravity with warm (37◦C) Tyrode solution supplemented with 0.1 mM EGTA for 5 min. Enzyme solution containing 0.8 g/L collagenase Type II (Worthington Biochemical, Corp., Lakewood, NJ, United States) in Tyrode solution supplemented with 0.1 mM CaCl<sup>2</sup> was then perfused until the aortic valve was digested (confirmed by the increased outflow of perfusate). The heart was transferred to a Petri dish containing enzyme solution supplemented with 1 mg/mL bovine serum albumin (BSA) and gently shaken for 2–3 min at 37◦C to disperse individual myocytes. The resulting cell suspension was centrifuged for 3 min at 170 × g. The cell pellet was suspended in Tyrode solution supplemented with 0.5 mM CaCl<sup>2</sup> and was centrifuged again at the same speed. Finally, the cell pellet was suspended in storage solution containing Tyrode solution supplemented with 1 mM CaCl2.

# Recording of Ca2<sup>+</sup> Sparks, Ca2<sup>+</sup> Transients, and Assessment of SR Ca2<sup>+</sup> Load

Spontaneous Ca2<sup>+</sup> sparks and field-stimulated Ca2<sup>+</sup> transients were imaged in Fluo-3 loaded cardiomyocytes as previously reported (de Alba-Aguayo et al., 2017), with some modifications. Cells were superfused with recording solution (in mM: NaCl 130, CaCl<sup>2</sup> 1.8, MgCl<sup>2</sup> 2.0, KCl 5.4, glucose 10, HEPES 10, insulin 0.01, pH 7.4) and 2D-images were obtained with a confocal microscope (Leica TCS SP5, Leica Microsystems, Wetzlar, Germany) in the line-scan mode (1.9 ms/line). To obtain electrically-stimulated Ca2<sup>+</sup> transients, cells were paced at different frequencies using a Grass stimulator set at 70 V and pulse length of 20 ms; 5 images of the cell were taken for each stimulation frequency. The Y-axis of the image indicates the length of the cell (in µm) and the X-axis indicates the scan time (in s). Fluo-3 was excited at 488 nm using an argon laser at 2% intensity. The emitted fluorescence of the dye was measured at 510 nm. To determined SR Ca2<sup>+</sup> load, myocytes were paced at 1 Hz for at least 1 min to restore steady state SR Ca2<sup>+</sup> load, then confocal laser scanning was initiated and four electrically-evoked Ca2<sup>+</sup> transients were followed by rapid application of caffeine (10 mM) (Fernández-Velasco et al., 2009; de Alba-Aguayo et al., 2017).

# Analysis of Confocal Images

Image analysis was performed using IDL 5.5 software (Research Systems, Inc.) running a custom protocol written by Ana Maria Gómez (Inserm UMR-S 1180, Châtenay-Malabry, France), including baseline fluorescence correction and normalizing fluorescence levels (F) with respect to basal fluorescence (F0) (de Alba-Aguayo et al., 2017). The fluorescence transient was obtained by averaging the fluorescence values in a 1.4-µm frame over time. Amplitude was measured as the maximum value of F/F0, where F is the fluorescence signal, and F<sup>0</sup> is the basal fluorescence (measured as the average of the 50 lowest values on the fluorescence transient). Cell shortening was reported in % respect to total cell length. Cells with cell shortening values below 4.5% were not considered in the analysis. Decay Time (in ms) was determined at 50% of recovery of the Ca2<sup>+</sup> transient peak. Parameters such as amplitude (F/F0), frequency (events/s<sup>∗</sup> 100 µm), full duration at half maximum (FDHM in ms), full width at half maximum (FWHM in µm), time-to-peak (in ms) and decay time constant (in ms), were analyzed for Ca2<sup>+</sup> sparks (de Alba-Aguayo et al., 2017). The spark-mediated Ca2<sup>+</sup> leak was calculated according to the method reported by Biesmans and collaborators, in which the spark-mediated Ca2<sup>+</sup> leak is defined as spark frequency<sup>∗</sup> spark mass; and spark mass is calculated as spark amplitude∗duration∗width (Biesmans et al., 2011).

# Preparation of Heart Homogenates and SR-Enriched Fractions

Heart homogenates and SR-enriched fractions were prepared from left-ventricle tissue pulverized to a fine powder with a mortar and pestle over liquid N2. Tissue powder was suspended in homogenization buffer (in mM: sucrose 300, NaF 20, HEPES 20, aprotinin 5.2 × 10<sup>4</sup> , benzamidine 0.5, leupeptin, 0.012, PMSF 0.1; pH 7.2 with KOH) and homogenized with a Potter-Elvehjem homogenizer (Cole-Parmer, Vernon Hills, IL, United States) and spun at 2000 × g for 10 min; the resulting supernatant (homogenate) was spun at 8000 × g for 10 min; SR-enriched fractions (or microsomes) were isolated using the second supernatant by ultracentrifugation at 40000 × g for 30 min at 4◦C. The resulting pellets were suspended in homogenization buffer and protein concentration was determined using the Lowry method as previously reported (de Alba-Aguayo et al., 2017).

# Measurement of SR Ca2<sup>+</sup> Uptake

Fura-4F acid form (pentapotassium salt, Molecular Probes/Thermo Fisher Scientific, Inc.) was added outside SR-membrane vesicles and used to monitor the SERCAdependent Ca2<sup>+</sup> uptake accordingly to a previously described method (de Alba-Aguayo et al., 2017). SR Ca2<sup>+</sup> uptake was measured in a buffer containing (in mM): KCl 100, MgCl<sup>2</sup> 4, MOPS 20, potassium oxalate 10, ATP-Mg 1.25, creatine phosphate 1.5, pH 7.4 with KOH, and 0.3 U/ml creatine phosphokinase. 10 µM Ruthenium Red was added to inhibit Ca2<sup>+</sup> release through RyRs. Fura-4F was added

to a final concentration of 0.5 µM. 600 µg of protein from SR-enriched vesicles were added to a cuvette with uptake buffer and equilibrated at 37◦C during 5 min with stirring. Fluorescence was measured in a QM-8 spectrofluorometer (PTI, excitation wavelengths at 340, 360, and 380 nm; emission was collected at 510 nm). The addition of 5 µM CaCl<sup>2</sup> initiated Ca2<sup>+</sup> uptake reactions; the fluorescence was recorded up to 900 s. The experiment was finished by adding Thapsigargin (1 µM). To determine Rmax, 200 µM CaCl<sup>2</sup> was added; to measure Rmin, 250 µM EGTA was added. Ratio values of fluorescence (F340/F380) were subsequently transformed into Ca2<sup>+</sup> concentrations using the Grynkiewicz equation (Grynkiewicz et al., 1985). Ca2<sup>+</sup> uptake speed was calculated as the first derivative of the monoexponential decay equation used to fit the Ca2<sup>+</sup> uptake curve.

#### [ <sup>3</sup>H]-Ryanodine Binding Assays

[ <sup>3</sup>H]-ryanodine binding experiments were performed as previously described using SR-enriched fractions (Fernández-Velasco et al., 2009; de Alba-Aguayo et al., 2017). [3H]-ryanodine saturation curves were carried out in medium containing 1 M KCl, 0.1 mM CaCl2, 20 mM HEPES (pH 7.2 with KOH); and [3H]-ryanodine in the range of 0.625 to 20 nM. Equilibrium binding data were fitted to the one-site model (Hill equation), apparent dissociation constant (Kd) and total functional RyRs (Bmax) were calculated from the fitted curve. The Ca2<sup>+</sup> dependence of [3H]-ryanodine binding was assessed in incubation medium containing 0.2 mM KCl, 20 HEPES (pH 7.2 with KOH), 50 µg of SR-enriched fractions, 10 nM [3H]-ryanodine, 1 mM EGTA, and CaCl<sup>2</sup> necessary to set free Ca2<sup>+</sup> in the range of 10 nM to 10 mM (total volume 100 µl). Ca2+-EGTA ratios were calculated using the MaxChelator program<sup>1</sup> (Bers et al., 2010). All incubations lasted 90 min at 36◦C. Samples were run in duplicate, filtered onto glass fiber filters (Whatman GF/B), and washed three times with 5 ml of cold water using a Brandel M-24R cell harvester. The filters were placed in scintillation vials, 8 ml of liquid scintillation mixture was added, and the retained radioactivity was measured in a Beckman LS-6500 β-counter. The specific binding was defined as the difference between the binding in the absence (total binding) and presence (non-specific binding) of 20 µM unlabeled ryanodine. Data represent the normalized mean ± SEM of the indicated n in the respective figure. Fitting of data was accomplished with the computer program OriginPro 8 (OriginLab Corporation, Northampton, MA, United States) to the following equation: B = Bmax([Ca2+] na/([Ca2+] na+K<sup>a</sup> na))(1– [Ca2+] ni/([Ca2+] ni+K<sup>i</sup> ni))+C, modified from Meissner's group (Liu et al., 1998), where B is the [3H]-ryanodine binding value at a given [Ca2+], Bmax is the binding maximum, K<sup>a</sup> and K<sup>i</sup> are Hill activation and inactivation constants, respectively, n<sup>a</sup> and n<sup>i</sup> are the respective Hill coefficients, and C is an initial [ <sup>3</sup>H]-ryanodine binding value at very low [Ca2+] (pCa8) (Fernández-Velasco et al., 2009; de Alba-Aguayo et al., 2017).

#### SDS-PAGE and Western Blot Analysis

SDS-PAGE and Western Blots were performed as a modification of previously reported protocols (Fernández-Velasco et al., 2009; de Alba-Aguayo et al., 2017). Gradient polyacrylamide gels (4–16%). Gels were loaded with the indicated µg of protein from heart homogenates in Laemmli buffer. Each gel was transferred onto a PVDF membrane for 2 h, 100 V at 4◦C in a humid chamber. PVDF membrane was blocked from non-specific binding with 5% non-fat dry milk in PBS-T buffer (composition in mM: KH2PO<sup>4</sup> 2.9, Na2HPO<sup>4</sup> 10.07, NaCl<sup>2</sup> 150.58 and Tween 20, 0.1%) for 1 h, before incubation with primary antibody against RyR2 (C3-33 dil: 1:5000, Cat# MA3-916, Thermo Fisher, Inc., Waltham, MA United States), pSer2808RyR2 (dil: 1:5000, Cat# A010-30, Badrilla, Leeds, United Kingdom), pSer2814RyR2 (dil: 1:5,000 Cat# A010- 31, Badrilla, Leeds, United Kingdom), SERCA2a (dil: 1,5000 Cat# A010-20, Badrilla, Leeds, United Kingdom), PLN (dil: 1:25,000 Cat# A010-14, Badrilla, Leeds, United Kingdom), pSer16PLN (dil: 1:3,000 Cat# A010-12, Badrilla, Leeds, United Kingdom), pThr17PLN (dil: 1:5,000 Cat# A010-13, Badrilla, Leeds, United Kingdom), CaMKII (dil: 1:2,500 Cat# 4436S, Cell Signaling, Danvers, MA, United States), and GAPDH (dil: 1:100,000 Cat# AM4300, Ambion <sup>R</sup> , Thermo Fisher Scientific, Waltham, MA, United States) for 2 h at room temperature. After washing, the membranes were incubated with corresponding secondary peroxidaseconjugated antibodies (dil 1:5,000 of goat anti-mouse IgG peroxidase conjugated, Cat. No. 401215, or goat anti-rabbit IgG peroxidase conjugated, Cat. No. 401315; both from Calbiochem <sup>R</sup> Merck, Co., Kenilworth, NJ, United States) for 1 h. The membranes were washed (3X, 10 min each with PBS-T buffer). Proteins were visualized by chemiluminescent reaction (SuperSignal <sup>R</sup> West Pico Chemiluminescent Substrate, Thermo Fisher Scientific, Waltham, MA, United States) and the relative amount of protein was determined by densitometric analysis using Kodak MI SE Software (v.5.01.30. Molecular Imaging Software).

#### Statistical Data Analysis

Data are presented as the mean ± SEM of indicated independent determinations, where "N" was used to designate the number of animals or hearts; and "n" to specify the number of cells, Ca2<sup>+</sup> release events, or experiments. Statistical significance was evaluated by Student's t-test or one-way ANOVA followed by Tukey post hoc test, when appropriated (OriginPro 8 software from Origin Lab Corporation, Northampton, MA, United States). Significance was defined at P < 0.05.

# RESULTS

#### Characteristics of the Chronic Sucrose-Induced MetS Model

Male rats that received 30% sucrose in the drinking water for 24 weeks developed several of the hallmark characteristics of MetS: for instance, a significant increase in body weight

<sup>1</sup>http://maxchelator.stanford.edu/index.html

(32%) and visceral fat accumulation (2.4-fold) compared to the control group (**Table 1**). Additionally, average SBP was significantly higher in MetS rats than controls (**Table 1**). Although the heart weight was slightly augmented in the MetS condition, features such as heart weight-to-body weight ratio, left ventricle weight, and left ventricle-to-heart weight ratio showed no alterations, indicating the lack of severe structural remodeling in the heart (**Table 1**). Serum parameters evidenced the development of hypertriglyceridemia (2.6-fold increase in serum triglycerides levels); and a significant increase in the TG-to-HDL-C ratio, which is indicative of a higher degree of cardio-metabolic risk in the insulin-resistant condition (Salazar et al., 2014). These alterations in both body and serum parameters are strong evidence of the development of MetS in rats as a consequence of the sucrose-rich diet (Wong et al., 2016).

# Diminished Ca2<sup>+</sup> Transient Amplitude in Cardiomyocytes of Metabolic Syndrome Rats

Previous studies have demonstrated depressed cardiac contractile function in sucrose-fed animals, evidenced by the reduction of ejection fraction, left ventricle systolic function, and heart fractional shortening (Vasanji et al., 2006). Moreover, reduced peak tension in papillary muscle strips has been reported in MetS condition (Okatan et al., 2016). Because the intensity of cardiomyocyte contractile response relies on the magnitude and duration of the intracellular Ca2<sup>+</sup> transient (Spurgeon et al., 1992), we investigated whether MetS affected electrically-evoked Ca2<sup>+</sup> transients and associated cell shortening in single cardiomyocytes at three different stimulation frequencies. **Figure 1A** shows representative confocal images (acquired in line scan mode) of Fluo-3 loaded cardiomyocytes from control (top) and MetS (bottom)

TABLE 1 | Body characteristics and biochemical parameters of metabolic syndrome and control animals.


Data are means ± SEM of N = 14 control rats and N = 12 MetS rats. Bw, body weight; HDL-C, high-density lipoprotein cholesterol; Hw, heart weight; LV, left ventricle; TGs, triglycerides. <sup>∗</sup>P < 0.05 and ∗∗P < 0.001 vs. Control group.

rats at 0.5 Hz, with their corresponding Ca2<sup>+</sup> transient profiles (F/F0), cellular shortening (% of resting cell size), and Ca2<sup>+</sup> transient decay time (ms). Cardiomyocytes of both experimental groups exhibited a rate-dependent fall in the Ca2<sup>+</sup> transient amplitude; however, in the MetS condition, the magnitude of the Ca2<sup>+</sup> transient was consistently lower than in control cells (**Figure 1B**). In MetS cells, the amplitude of the Ca2<sup>+</sup> transient represented 76.6, 62.8, and 61.5%, of corresponding Ca2<sup>+</sup> transient amplitude in control cells, at 0.5, 1, and 2 Hz, respectively. Interestingly, a rate-dependent decline in cell contraction was found in the MetS group, with a significant reduction of 32.8, 48, and 65.9% in cell shortening of MetS cardiomyocytes at 0.5, 1, and 2 Hz, respectively (**Figure 1C**). We investigated whether alterations in the distance between T-tubules in cardiomyocytes to estimate sarcomere length (SL), could account for these results (see **Supplementary Materials and Methods**), since SL above or below the optimal range is known to contribute to a decrease in cardiomyocyte contractility. However, we found similar T-tubule distances in RyR-immunostained cardiomyocytes from both experimental groups (average T-tubule distance, in µm: 1.67 ± 0.03 in n = 5 control cells, vs. 1.64 ± 0.019 in n = 5 MetS cells, P = 0.535. **Supplementary Figure 1**), ruling out the presence of structural T-tubule remodeling in MetS cardiomyocytes. Analysis of Ca2<sup>+</sup> transient kinetics revealed that the decay time (at 50% of the peak) was slightly increased in MetS cells at all frequencies (**Figure 1D**), suggesting that the Ca2<sup>+</sup> clearing mechanisms are slowed. The Ca2<sup>+</sup> handling abnormalities could be related to a decreased rate of SR Ca2<sup>+</sup> uptake and/or an increased SR Ca2<sup>+</sup> leak; we therefore evaluated the Ca2<sup>+</sup> spark-mediated Ca2<sup>+</sup> leak.

# Reduced Diastolic Ca2<sup>+</sup> Leak in Cardiomyocytes of Metabolic Syndrome Rats

The extent of SR Ca2<sup>+</sup> leak impacts the SR Ca2<sup>+</sup> available for release, causing systolic dysfunction. Increased SR Ca2<sup>+</sup> leak elevates diastolic Ca2<sup>+</sup> level, contributing to diastolic dysfunction; causes triggered arrhythmias and it is energetically costly (Bers, 2014). Thus, in a first approach, we analyzed the Ca2<sup>+</sup> -spark mediated Ca2<sup>+</sup> leak in intact Fluo 3-loaded cardiomyocytes of both experimental groups. **Figure 2A** shows representative confocal images of spontaneous Ca2<sup>+</sup> sparks taken from control (top) and MetS (bottom) cardiomyocytes in line scan mode. Analyses of Ca2<sup>+</sup> spark characteristics are summarized in **Figures 2B** through 2H. Ca2<sup>+</sup> sparks recorded in MetS cells showed significantly reduced frequency (by 34.7%, **Figure 2B**) and amplitude (by 7%, **Figure 2C**) compared to control cells. Ca2<sup>+</sup> spark mass calculated as the product of Ca2<sup>+</sup> spark amplitude∗duration∗width was lower in MetS cell than in controls, resulting in a significant reduction of the Ca2<sup>+</sup> spark-mediated Ca2<sup>+</sup> leak (**Figure 2F**). Other Ca2<sup>+</sup> spark parameters such as duration (**Figure 2D**), size (**Figure 2E**) time-to-peak (**Figure 2G**), and decay time (**Figure 2H**) remained unchanged.

images (A) recorded during electric stimulation (0.5 Hz) of Fluo 3-loaded ventricular myocytes isolated from control (C) and MetS rats. Green lines indicate the start of electric stimulation. The corresponding normalized fluorescence traces (F/F0) and contraction profiles (in %) are shown below. (B) Average of Ca2<sup>+</sup> transient amplitude (expressed as F/F0, where F is the peak fluorescence signal and F<sup>0</sup> the diastolic fluorescence) of control (open circles) and MetS (orange circles) cardiomyocytes obtained by field stimulation at 0.5, 1, and 2 Hz. Average cell shortening (C), and decay time at 50% of the normalized Ca2<sup>+</sup> transient (D) at 0.5, 1, and 2 Hz in n = 17 control cells vs. n = 32 MetS cardiomyocytes. <sup>∗</sup>P < 0.05, ∗∗P < 0.001 respect to control cells. #P < 0.05, ##P < 0.001 respect to 0.5 Hz in the same condition.

# SR Ca2<sup>+</sup> Load Is Unaffected in MetS Cardiomyocytes

The amount of Ca2<sup>+</sup> stored inside the SR that is available for release during the cardiomyocyte contraction is primarily determined by the delicate balance between diastolic Ca2<sup>+</sup> leak and SERCA pump-mediated Ca2<sup>+</sup> uptake (Zima et al., 2010). Thus, we hypothesized that the decrease in the Ca2<sup>+</sup> sparkmediated Ca2<sup>+</sup> leak in MetS cells could be attributable to a significant reduction of the SR Ca2<sup>+</sup> load. Accordingly, we determined the SR Ca2<sup>+</sup> load in steady state conditions by measuring the amplitude of the caffeine-induced Ca2<sup>+</sup> transient in cardiomyocytes previously paced at 1 Hz. **Figure 3A** shows representative line-scan images of caffeine-induced Ca2<sup>+</sup> transients recorded in Fluo-3 loaded control and MetS cells. Our results show that the development of MetS did not modify the SR Ca2<sup>+</sup> loading in steady-state conditions because the amplitude of the caffeine-induced Ca2<sup>+</sup> transient was similar in both groups (**Figure 3B**), in agreement with previous reports in sucrose-fed rats at early stages of MetS development (Wold

et al., 2005; Balderas-Villalobos et al., 2013). Unchanged SR Ca2<sup>+</sup> content is incompatible with reduced diastolic Ca2<sup>+</sup> leak in the MetS cardiomyocytes—unless SERCA pump activity is compromised. Although the decay time of Ca2<sup>+</sup> transients was found to be slightly increased in MetS cells, this parameter is a combination of several mechanisms, including SR Ca2<sup>+</sup> uptake, cytoplasmic Ca2<sup>+</sup> buffering, and plasma membrane Ca2<sup>+</sup> extrusion, among others (Bode et al., 2011). Therefore, by further analyzing the caffeine-induced Ca2<sup>+</sup> recordings, we specifically examined SERCA function using a previously reported method (Bode et al., 2011; Delgado et al., 2015). We calculated the SERCA-dependent portion of the rate constant of decay of the Ca2<sup>+</sup> transient (k SERCA) by subtracting the rate constant of decay of the caffeine-evoked Ca2<sup>+</sup> transient from that of the systolic Ca2<sup>+</sup> transients. These calculations demonstrated that k SERCA was significantly reduced in MetS cardiomyocytes (4.23 ± 0.38 s−<sup>1</sup> , n = 6 in control vs. 3.37 ± 0.12 s−<sup>1</sup> , n = 8, in MetS group; P < 0.05). These data

FIGURE 3 | SERCA pump activity is impaired, but this condition does not alter the SR Ca2<sup>+</sup> load in MetS cardiomyocytes. (A) Representative line-scan images of caffeine-induced Ca2<sup>+</sup> transients recorded in isolated Fluo 3-loaded cardiomyocytes from control (C, top) and MetS (bottom) rats to determine SR Ca2<sup>+</sup> loading. Cells were superfused with recording solution and paced at a frequency of 1 Hz (red lines) before caffeine challenge (indicated by the blue arrow). (B) The bar graph represents average values of the caffeine-induced Ca2<sup>+</sup> transient amplitude (F/F0) in control (white bar, n = 14 cells) and MetS (orange bar, n = 22 cells) cardiomyocytes. (C) Representative traces of Ca2<sup>+</sup> uptake assays in SR-enriched microsomes obtained from control (C, black trace) and MetS (orange trace) hearts. (D) The bar graph compares the average Ca2<sup>+</sup> uptake speed (in µM/s) in control (white bar, n = 10), and MetS (orange bars, n = 8) microsomal preparations. (E) Representative Western Blot images of phosphorylated PLN at Ser16, Thr17, total PLN, and SERCA pump in whole heart homogenates of control (C) and MetS rats. Bar graphs of normalized phosphorylation levels of pS16 PLN (F) and pT17 PLN (G) in whole heart homogenates of control (C, white bars) and MetS (orange bars) rats. N = 4–7 heart preparations for each experimental group. All values in graphs are presented as Mean ± SEM. <sup>∗</sup>P < 0.05 vs. control values.

enabled us to hypothesize impaired SERCA pump function in MetS cardiomyocytes.

We therefore used a more direct approach to investigate SERCA pump enzymatic activity by performing a SERCA pumpmediated Ca2<sup>+</sup> uptake assay in cardiac SR-enriched membranes (**Figure 3C**). These experiments revealed a 33% decrease in the rate of Ca2<sup>+</sup> uptake speed in MetS microsomes compared to the control microsomes (**Figure 3D**). SERCA-pump mediated Ca2<sup>+</sup> uptake was slowed in MetS microsomes, though SERCA pump protein expression was not affected (normalized SERCA pump expression: 1.0 ± 0.11 in n = 11 control samples vs. 1.05 ± 0.14 in n = 11 MetS samples; P > 0.05, N.S.). This result agrees with our finding of unchanged SR Ca2<sup>+</sup> load despite the reduced diastolic Ca2<sup>+</sup> leak observed in MetS cardiomyocytes.

Because PLN regulates SERCA-mediated Ca2<sup>+</sup> uptake, changes in PLN expression level and/or phosphorylation status would be expected modify SERCA activity (Mattiazzi and Kranias, 2014). To gain insight into the molecular mechanisms involved in the observed SERCA pump dysfunction, we analyzed PLN expression and phosphorylation status of phosphosites Ser16 (a PKA site) and Thr17 (a CaMKII site; **Figure 3E**). Total PLN expression and PLN/SERCA pump ratio were similar in both experimental conditions (PLN/SERCA pump, normalized data: 1.0 ± 0.16 in n = 7 control samples vs. 1.31 ± 0.18 in n = 7 MetS samples, P < 0.2307). Western Blot data showed a slight but not significant decrease in the phosphorylation level of Ser16- PLN (**Figure 3F**). However, Thr17-PLN phosphorylation was significantly reduced in MetS heart homogenates (**Figure 3G**), which correlates with the reduction in SERCA activity observed in SR Ca2<sup>+</sup> uptake assays (**Figures 3C,D**).

#### Dysfunctional Cardiac RyRs in the Metabolic Syndrome Condition

In our experimental model of MetS, the decrease in both the Ca2<sup>+</sup> transient amplitude and the Ca2<sup>+</sup> spark-mediated Ca2<sup>+</sup> leak in isolated cardiomyocytes cannot be explained by a reduction in the SR Ca2<sup>+</sup> load; therefore, we explored additional alterations in the RyR that might enable us to understand the changes in its activity. The fact that ryanodine binds only to open RyRs enables estimation of functional RyRs under in vitro conditions. **Figure 4A** shows saturation binding curves for [3H]-ryanodine in SR-enriched fractions from MetS and control hearts. SR membranes from MetS hearts bound significantly less [3H]-ryanodine (38.2%) than those from control hearts. Importantly, the Kd for [3H] ryanodine was found in the nanomolar range and remained unchanged (**Table 2**), indicating that the apparent ryanodine affinity of RyRs remains similar in both experimental groups. We also examined whether the Ca2+-dependent activation of [3H]-ryanodine binding might differ between MetS and control microsomes. **Figure 4B** shows bell-shape-like curves for both control and MetS heart microsomes. Both data sets were fitted with the same equation, indicating that RyRs followed a bimodal Ca2<sup>+</sup> activation/inactivation response as previously reported (Liu et al., 1998; de Alba-Aguayo et al., 2017). The fitted parameters indicate that Ca2<sup>+</sup> was equally

[ <sup>3</sup>H]-ryanodine. (A) Aliquots of 50 µg of protein from microsomes (SR-enriched fraction) from control (N = 6) and MetS (N = 5) hearts were incubated with [3H]-ryanodine (concentration range of 0.625–20 nM) in incubation medium (containing in mM: 1000 KCl, 0.1 CaCl<sup>2</sup> 20 HEPES, pH 7.2) for 90 min at 37◦C. (B) Specific Ca2+-dependent [3H]-ryanodine binding curves of SR-enriched fractions from control (open symbols; N = 9) and MetS (orange symbols; N = 6) hearts. Microsomes were incubated with 7 nM [ <sup>3</sup>H]-ryanodine and the indicated [Ca2+]<sup>i</sup> at 37◦C for 90 min. Values have been normalized and fitted to a modified version of the equation B = Bmax([Ca2+] na/([Ca2+] na+K<sup>a</sup> na))(1–[Ca2+] ni/([Ca2+] ni+K<sup>i</sup> ni))+C. Non-specific [3H]-ryanodine binding was determined in the presence of 20 µM ryanodine and subtracted from all reported values. Open symbols, control; orange symbols, MetS; dashed line indicates the fitting curve for control and dark solid line the fitting curve for MetS microsomes. <sup>∗</sup>P < 0.05 with respect to corresponding control values.

effective in activating RyRs from control and MetS heart microsomes, since the range for Ka value remained similar (**Table 3**) and in agreement with previous reports (Liu et al., 1998). However, RyRs from MetS hearts exhibited rapid Ca2+ induced inactivation (**Figure 4B**), and showed a significant decrease (26.3%) in the total amount of RyRs capable of binding [3H]-ryanodine at optimal [Ca2+] (indicated by the Bmax value in **Table 3**), with a slight, but not significant, decrease in the initial quantity of active RyR at very low [Ca2+] (indicated by the C value in **Table 3**). Thus, our results




Values are Mean ± SEM of indicated independent determinations for each experimental group. Bmax indicates the maximum number of active RyRs, K<sup>a</sup> and K<sup>i</sup> are Hill activation and inactivation constants, respectively; and C is an initial [3H]-ryanodine binding value at very low [Ca2+] (pCa8). N indicates the number of hearts. <sup>∗</sup>P < 0.05 vs. control group.

FIGURE 5 | Reduced phosphorylation levels of Cardiac Ryanodine Receptor at Ser-2814 underlie its dysfunctional activity. (A) Representative Western Blot images of phosphorylated RyR2 at Ser2808 (pS2808-RyR2), Ser2814 (pS2814-RyR2), and total ryanodine receptor (RyR2) in whole heart homogenates of control (C) and MetS rats. GAPDH signals were used as loading controls. 50 µg of protein from whole heart homogenates of each experimental group (C and MetS) were subjected to 4–16% gradient SDS-PAGE, blotted onto PVDF membranes during 2 h, and probed with the respective primary and secondary antibodies. Bar graphs indicate normalized expression levels of total RyR2 (B), pS2808RyR2/RyR2 total (C), and pS2814RyR2/RyR2 total (D) in whole heart homogenates of control (C, white bars) and MetS (orange bars) rats. N = 6–7 heart preparations for each experimental group. All values in graphs are presented as Mean ± SEM. <sup>∗</sup>P < 0.05 vs. control values.

show an important reduction of active RyR2 that inactivate faster in MetS microsomes, which could help explain the decrease in both Ca2<sup>+</sup> transient amplitude and the diastolic Ca2<sup>+</sup> leak in MetS cardiomyocytes despite the unchanged SR Ca2<sup>+</sup> load.

Western blot analysis was used to determine the total amount of RyR2 protein expression in control and MetS heart preparations. We also analyzed the phosphorylation levels of the corresponding Ser2808 (a substrate for several kinases, PKA and CaMKII among them) and Ser2814 (a preferential target for CaMKII) in RyR2, in light of the finding that modifications of RyR phosphorylation level account for the dysfunctional activity of cardiac RyRs in insulin-resistance and prediabetes (Dincer, 2012; Okatan et al., 2016; Sommese et al., 2016). **Figure 5A** shows representative immunoblots of pSer-2808-RyR2, pSer-2814-RyR2, and total RyR. The bar graph in **Figure 5B** shows that the total amount of RyR2 was unmodified in MetS heart preparations in comparison with controls (normalized total RyR2 expression: 1.0 ± 0.15 in N = 6 control hearts vs. 0.91 ± 0.10 in N = 7 MetS hearts; P > 0.05, N.S.). Nevertheless, we found a significant reduction (46%) in the phosphorylation status of Ser2814 (Normalized pSer2814RyR2: 1.0 ± 0.13 in N = 6 control vs. 0.54 ± 0.14 in N = 6 MetS hearts; <sup>∗</sup>P < 0.05. **Figure 5D**), while the phosphorylation level of Ser2808 remained unchanged (normalized pSer2808RyR2: 1.0 ± 0.37 in N = 6 control vs. 0.7 ± 0.32 in N = 6 MetS hearts; P > 0.05, N.S.; **Figure 5C**). The decreased levels of pSer2814-RyR2 and pThr17- PLN, both well-known CaMKII target sites, suggests that CaMKII activity might be compromised in our MetS model. Although CaMKII expression was similar in both experimental groups (total CaMKII expression: 1.0 ± 0.07 in N = 7 control hearts vs. 1.01 ± 0.07 in N = 8 MetS hearts; P > 0.05, N.S.; **Supplementary Figure 2**), we cannot rule out that changes in its activity could account for the Ca2<sup>+</sup> handling alterations reported in this work. Since CaMKII activity is modulated by a wide range of posttranslational modifications, further analysis regarding this issue will be required.

#### DISCUSSION

In diabetic cardiomyopathy, the mechanisms that regulate intracellular Ca2<sup>+</sup> handling are altered and the function of

RyR2 is clearly compromised; the latter has been associated with a decrease in RyR2 protein expression and reduced SR Ca2<sup>+</sup> load (Pereira et al., 2014); however, the data concerning the contribution of RyRs to the Ca2<sup>+</sup> mishandling in MetS cardiomyocytes has, until now, been lacking.

Thus, the main aim of this study was to more closely examine the in situ activity, protein expression, Ca2<sup>+</sup> sensitivity, and phosphorylation status of the RyR2 in an animal model of MetS. Sucrose administration (30% in drinking water) for 24 weeks successfully generated key features of MetS in rats, characterized by the presence of central obesity, hypertension, hypertriglyceridemia, and increased TG/HDL-C ratio (**Table 1**). Previous reports in experimental models of MetS at early stages of development induced by administration of either fructose or sucrose in the diet have shown differential development of hypertriglyceridemia, hypertension, and increased abdominal fat accumulation, though not necessarily an increase in body weight (Hintz and Ren, 2002; Davidoff et al., 2004; Schwanke et al., 2006; Balderas-Villalobos et al., 2013; Sommese et al., 2016). Under our experimental conditions, the sucrose-fed rats showed a phenotype closely related with MetS patients, in whom a significant increase in blood presure, TG/HDL-C ratio, body weight, and visceral fat accumulation are considered key cardiovascular risk factors (Wilson et al., 2005). In our MetS model, several hallmark metabolic disturbances are strongly established, leading to a condition that more closely resembles chronic prediabetic state in obese humans. We believe that the differences in our model respect to others are mainly attributable to the prolonged administration of sucrose (e.g., 6 months rather than only 4).

Our results demonstrate the detrimental effect of a high carbohydrate diet on the metabolic condition—and specifically in cardiomyocyte performance, which agrees well with previous publications (Cárdenas et al., 2006; Balderas-Villalobos et al., 2013; Okatan et al., 2016).

Alterations in heart contractile activity have been previously reported in sucrose-fed rats; intracellular Ca2<sup>+</sup> mishandling was proposed as the underlying mechanism (Carvajal and Baños, 2002; Vasanji et al., 2006; Okatan et al., 2016). Indeed, we consistently found a significant reduction in the Ca2<sup>+</sup> transient amplitude of MetS cardiomyocytes (**Figure 1B**); our data agreed with previous reports (Hintz et al., 2003; Balderas-Villalobos et al., 2013; Okatan et al., 2016). Contrary to what might have been expected, we found no rate-dependent increase in the Ca2<sup>+</sup> transient peak as previously reported (Bassani et al., 1995). The reason for this discrepancy is not apparent to us. Possible factors influencing this discrepancy are the different age and strain of our rats vs., the ones used by Bassani and cols. With regard to contraction, in the case of rodents (like rats and mice) the force-frequency relationship (FFR) is even or negative at very low ex vivo frequencies (Janssen and Periasamy, 2007). In our hands, the contraction-frequency relationship showed a pronounced drop in MetS cells while remaining unchanged in controls (**Figure 1C**). Of note, there is a dissociation between the important decrease in the amplitude of Ca2<sup>+</sup> transients with increasing frequency and the less important or even lack of decrease in the amount of shortening. This dissociation may reveal an increase in Ca2<sup>+</sup> myofilament sensitivity with increasing stimulation frequency, which appears as more important in control than in MetS cells. A frequencydependent 'sensitization' of the myofilaments has been previously suggested (see for instance Gao et al., 1998). In addition, a possible explanation for the impairment of cell contractility in MetS cells might be related to the finding that in male Wistar rats under high-sucrose diet, both myosin and actomyosin Ca2+- ATPase are downregulated in the heart (Pierce et al., 1989); this should be addressed in the future.

In light of the fact that k SERCA was reduced in MetS cardiomyocytes, the observation that the Ca2<sup>+</sup> transient decay time was not considerably augmented might seem surprising. However, it should be pointed out that additional Ca2<sup>+</sup> clearing mechanisms are involved in the Ca2<sup>+</sup> transient decline (i.e., Ca2<sup>+</sup> extrusion by the NCX, cytoplasmic Ca2+-buffering proteins, dynamic RyR open time, etc.). Some of these are excluded when k SERCA is calculated as reported by Bode et al. (2011), unveiling a depressed SERCA function in MetS cells.

Because local, spontaneous Ca2<sup>+</sup> release events (Ca2<sup>+</sup> sparks) constitute the building blocks of Ca2<sup>+</sup> transients (Cheng and Lederer, 2008), we studied the Ca2<sup>+</sup> spark properties in MetS and control cells. Our results revealed a significant decrease in Ca2<sup>+</sup> spark frequency and amplitude, reducing the diastolic Ca2<sup>+</sup> leak in MetS cells (**Figure 2**); this suggests reduced RyR recruitment or activation (Hoang-Trong et al., 2015). Sucrose-fed rat hearts have a marked diminution of ATP and phosphocreatine content and creatine kinase activity (Carvajal et al., 2003). Because diastolic Ca2<sup>+</sup> leak is energetically costly, the finding of a significant reduction of this parameter in MetS cardiomyocytes is not surprising, though it is in disagreement with other reports in insulin-resistant rats (Okatan et al., 2016; Sommese et al., 2016). However, when the heart is energetically compromised as in MetS condition, increasing the diastolic Ca2<sup>+</sup> leak may add additional complications to already established energetic issues; thus, we propose that reducing the amount of Ca2<sup>+</sup> leakage during diastole could be rather beneficial at rest, but may compromise heart function under high demand (i.e., β-adrenergic stimulus, stress, or exhausting exercise). Although an augmentation in diastolic [Ca2+]<sup>i</sup> might participate in the impairment of the relaxation speed in MetS cells, previous reports in related insulin resistant models showed no changes in basal Ca2<sup>+</sup> in cardiac cells (Dutta et al., 2001; Hintz et al., 2003), which is consistent with the decreased spark-mediated Ca2<sup>+</sup> leak observed in our MetS model. However, it should be mentioned that increased diastolic Ca2<sup>+</sup> levels have been also reported in sucrose- and fructose-fed rats (Okatan et al., 2016; Sommese et al., 2016). Moreover, at this moment we cannot rule out alterations in other forms of diastolic Ca2<sup>+</sup> leak in MetS cardiomyocytes (Bovo et al., 2011).

In physiological conditions, free [Ca2+]SR influences RyR2 activity by establishing the driving force for SR Ca2<sup>+</sup> release (Györke and Terentyev, 2008). A reduction in the SR Ca2<sup>+</sup> content induces a decrease in Ca2<sup>+</sup> spark frequency (Satoh et al., 1997; Bers et al., 2003; Zima et al., 2010). In light of these findings, we analyzed caffeine-induced Ca2<sup>+</sup> transients to assess whether the diminished diastolic Ca2<sup>+</sup> leak observed

in MetS cardiomyocytes was due to changes in SR Ca2<sup>+</sup> load. Our results showed that MetS development does not modify the SR Ca2<sup>+</sup> loading in cardiomyocytes, which is in agreement with some reports (Wold et al., 2005; Balderas-Villalobos et al., 2013) but not with others (Okatan et al., 2016; Sommese et al., 2016). The discrepancies are related to differences in the diastolic Ca2<sup>+</sup> leak, which, in two of the abovementioned reports, was increased and explained the reduction in SR Ca2<sup>+</sup> load (Okatan et al., 2016; Sommese et al., 2016). SERCA pump is one of the most studied proteins in the context of cardiomyopathies associated with diabetes; in the case of prediabetes, it is generally demonstrated that its activity decreases (Wold et al., 2005; Vasanji et al., 2006; Balderas-Villalobos et al., 2013; Okatan et al., 2016). Independently of the unchanged SR Ca2<sup>+</sup> load, we corroborated the existence of a dysfunctional SERCA pump in MetS hearts, by the use of direct (SR Ca2<sup>+</sup> uptake assays) and indirect (by calculating k SERCA) approaches (**Figure 3**). One conceivable explanation for the decrease in SERCA pump activity despite unchanged protein expression is that the oxidative stress level could be exacerbated in insulin resistant cardiomyocytes (Dutta et al., 2001; Hintz et al., 2003), this would increase the number of oxidized thiols in the SERCA pump, thereby depressing its activity (Balderas-Villalobos et al., 2013). Another explanation relies on the fact of reported reduced phosphorylation levels of PLN. In this work we found that pThr-17 PLN was significantly reduced, in agreement with one study in sucrose-fed rats reporting decreased phosphorylation of PLN at both Ser-16 and Thr-17, which could account for its reduced activity (Vasanji et al., 2006). On the other hand, SERCA pump activity is highly dependent on energy supply; thus, it is not surprising to find its activity depressed in MetS cardiomyocytes (**Figure 3**).

Neither SR Ca2<sup>+</sup> load nor RyR2 protein expression was modified in MetS condition; therefore, this parameter cannot explain the depressed in situ activity of RyR2 in MetS cardiomyocytes. We performed [3H]-ryanodine binding assays to gain more insight into the molecular modifications of RyR2. Our results revealed that in MetS heart, the maximum number of functional RyRs was significantly reduced, though no significant changes (besides Bmax value) in Ca2<sup>+</sup> dependent RyR activation/inactivation curves were revealed. We therefore examined RyR2 phosphorylation status at Serines 2808 and 2814 (in human and mouse nomenclature). These post-translational modifications, induced by kinases such as PKA and CaMKII, have been defined as a critical regulatory mechanism for this intracellular Ca2<sup>+</sup> channel (Takasago et al., 1989, 1991; Huke and Bers, 2008; Camors and Valdivia, 2014). In our experimental model of MetS, we found a significant reduction in the phosphorylation level of Ser2814, which no doubt contributed to the impaired functional activity of RyRs in MetS hearts.

A seminal study in a dog model of MetS reported dysfunctional cardiac RyRs, characterized by a decreased capacity for [3H]-Ryanodine binding despite higher levels of Ser2808 phosphorylation compared with control hearts (Dincer et al., 2006). In our MetS model, the phosphorylation of Ser2808 remained unchanged. Perplexing a different study reported diminished phosphorylation of Ser2808 in MetS heart preparations, but did not further explore the impact of this variation in RyR functional activity (Paulino et al., 2010). Finally, additional studies have proposed that the "leaky" RyR2 phenotype in MetS cardiomyocytes is associated with augmented RyR phosphorylation at either Ser-2808 (Okatan et al., 2016) or Ser-2814 (Sommese et al., 2016). The only point where most publications—including ours— agree is about the fact that RyR2 protein expression remains unmodified in MetS models (Dincer et al., 2006; Paulino et al., 2010; Okatan et al., 2016; Sommese et al., 2016).

We are aware of the controversial role of Ser2808 phosphorylation in the abnormal activity of RyR2 in the setting of HF. The unsolved controversy is supported by a considerable amount of data that is primarily in favor of phasing out the importance of Ser2808 as the key player in HF (Valdivia, 2012); our data from a completely distinct cardiac disease support this idea. A definitive role for RyR2 phosphorylation has not yet been reached (Camors and Valdivia, 2014). Thus it is plausible that, in the case of MetS cardiomyopathy, a RyR2 that is dephosphorylated at a crucial residue (e.g., Ser2814) could be more difficult to activate. Indeed, the work of Dhindwal et al. (2017) supports the latter. This group proposed that phosphorylation causes RyR2 to adopt a protein conformation that requires less energy for the transition to the open state, using a structural basis to explain activating effects of phosphorylation in RyRs. Thus, we propose that in a metabolic condition such as MetS, the RyR is primarily dephosphorylated at Ser2814, thus impairing its physiological activation.

The downregulation of the CaMKII kinase could account for the reduced RyR phosphorylation, particularly at the Ser2814 site (**Figure 5C**). To test the involvement of this kinase, we evaluated the phosphorylation of PLN at Thr-17, which was significantly reduced in MetS heart preparations. However, CaMKII expression was not significantly changed (**Supplementary Figure 2**). It is important to note that CaMKII activity is regulated through several mechanisms. For example, CaMKII is susceptible to several post-translational modifications, including autophosphorylation, oxidation, S-nitrosylation, and O-GlcNAcylation, all of which are capable of sustaining the kinase's activation (Mollova et al., 2015). Thus, we intend to address this issue in future work. Another mechanism that could be responsible for the apparent CaMKII downregulation could relate to disturbances of the CaM-CaMKII interaction due to the altered CaM oxidation state. However, this has not been demonstrated in the cardiac isoform (Robison et al., 2007).

#### CONCLUSION

Principal findings of this work are that abnormal Ca2<sup>+</sup> transient amplitude, contractile dysfunction; and impaired relaxation of MetS cardiomyocytes underlies intrinsic dysfunctional RyR2 and SERCA pump. Abnormal activity of RyRs was evidenced by its decreased ability to bind [3H]-ryanodine. Although the MetS condition does not modify RyR2 protein expression, its phosphorylation at Ser2814 is decreased, which impairs its

capacity for activation during ECC. The dysfunctional RyRs, together with a decreased activity of SERCA pump due to decreased Thr17-PLN phosphorylation suggest a downregulation of CaMKII in MetS hearts, though its expression remained unchanged. Dysfunctional RyR2 and SERCA pump participate in the Ca2<sup>+</sup> mishandling of MetS cardiomyocytes (i.e., reduced Ca2<sup>+</sup> transient amplitude, and decreased Ca2<sup>+</sup> sparkmediated Ca2<sup>+</sup> leak), may account for the poor overall cardiac outcome reported in MetS patients and could be targeted for future therapies.

#### ETHICS STATEMENT

This study was carried out in accordance with the ethical guidelines of the Mexican Official Norm (NOM-062-ZOO-1999) and the National Institutes of Health Guide for the Care and Use of Laboratory Animals (NIH publication updated in 2011); and was approved by the Institutional Bioethical Committee for Care and Handling of Laboratory Animals at the CINVESTAV-IPN (approved CICUAL Protocol No. 0105-14).

#### AUTHOR CONTRIBUTIONS

GF-M, TR-G, and TB-L contributed to the generation and characterization of the experimental model of MetS in rats, isolation of cardiomyocytes, confocal imaging of Ca2<sup>+</sup> sparks, electrically stimulated Ca2<sup>+</sup> transients, determination of SR Ca2<sup>+</sup> load; [3H]-ryanodine bindings, Ca2<sup>+</sup> uptake assays and data analysis. M-MM contributed towards SDS-PAGE, Western Blots, and densitometric analysis. AR contributed towards the experimental design, data analysis, data interpretation, and manuscript preparation.

#### REFERENCES


#### FUNDING

This work was supported by the Secretaría de Ciencia y Tecnología of México City (SECITI-CDMX, Grant No. 331/2010) by Conacyt CB Grant 80960 to AR; and by a PRODEP-SEP grant to the Academic Group CINVESTAV-CA-10, ID 28915/2018.

#### ACKNOWLEDGMENTS

We deeply thank Dr. Agustín Guerrero-Hernández (Department of Biochemistry, CINVESTAV-IPN, Mexico) and Dr. José Antonio Arias-Montaño (Department of Physiology, Biophysics and Neurosciences, CINVESTAV-IPN, Mexico) for providing us with the facilities to perform the microsome Ca2<sup>+</sup> uptake and [3H]-ryanodine binding experiments, respectively. Anti pSer16-PLN, pSer17-PLN, PLN and CaMKII antibodies were a generous gift from Dr. Héctor H. Valdivia (University of Wisconsin-Madison, Madison, WI, United States). We thank Dr. E. Michelle Frank (The Well-Tempered Word LLC) for critical reading of the manuscript. We also wish to acknowledge the technical assistance of Juan Carlos García Torres; and M.Sc. Iván Galván (Microscopy Unit, LANSE-CINVESTAV-IPN). We also thank Roche S. A. de C. V., for the award of Medical Research Dr. Jorge Rosenkranz in Diabetes field, Young category to AR.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2019.00520/full#supplementary-material




**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Fernández-Miranda, Romero-Garcia, Barrera-Lechuga, Mercado-Morales and Rueda. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Transmural Autonomic Regulation of Cardiac Contractility at the Intact Heart Level

Yuriana Aguilar-Sanchez<sup>1</sup> , Ainhoa Rodriguez de Yurre<sup>2</sup> , Mariana Argenziano<sup>3</sup> , Ariel L. Escobar<sup>4</sup> \* and Josefina Ramos-Franco<sup>1</sup> \*

<sup>1</sup> Department of Physiology and Biophysics, School of Medicine, Rush University Medical Center, Chicago, IL, United States, <sup>2</sup> Laboratorio de Cardio Inmunologia, Instituto de Biofisica Carlos Chagas Filho, Rio de Janeiro, Brazil, <sup>3</sup> Division of Human Genetics, Children's Hospital of Philadelphia, Philadelphia, PA, United States, <sup>4</sup> Department of Bioengineering, School of Engineering, University of California, Merced, Merced, CA, United States

#### Edited by:

Angélica Rueda, Centro de Investigación y de Estudios Avanzados (CINVESTAV), Mexico

#### Reviewed by:

Nieves Gómez-Hurtado, Autonomous University of Madrid, Spain Roberto R. M., University of Michigan, United States

> \*Correspondence: Ariel L. Escobar

aescobar4@ucmerced.edu Josefina Ramos-Franco jrfranco@rush.edu

#### Specialty section:

This article was submitted to Integrative Physiology, a section of the journal Frontiers in Physiology

Received: 10 January 2019 Accepted: 03 June 2019 Published: 03 July 2019

#### Citation:

Aguilar-Sanchez Y, Rodriguez de Yurre A, Argenziano M, Escobar AL and Ramos-Franco J (2019) Transmural Autonomic Regulation of Cardiac Contractility at the Intact Heart Level. Front. Physiol. 10:773. doi: 10.3389/fphys.2019.00773 The relationship between cardiac excitability and contractility depends on when Ca2<sup>+</sup> influx occurs during the ventricular action potential (AP). In mammals, it is accepted that Ca2<sup>+</sup> influx through the L-type Ca2<sup>+</sup> channels occurs during AP phase 2. However, in murine models, experimental evidence shows Ca2<sup>+</sup> influx takes place during phase 1. Interestingly, Ca2<sup>+</sup> influx that activates contraction is highly regulated by the autonomic nervous system. Indeed, autonomic regulation exerts multiple effects on Ca2<sup>+</sup> handling and cardiac electrophysiology. In this paper, we explore autonomic regulation in endocardial and epicardial layers of intact beating mice hearts to evaluate their role on cardiac excitability and contractility. We hypothesize that in mouse cardiac ventricles the influx of Ca2<sup>+</sup> that triggers excitation–contraction coupling (ECC) does not occur during phase 2. Using pulsed local field fluorescence microscopy and loose patch photolysis, we show sympathetic stimulation by isoproterenol increased the amplitude of Ca2<sup>+</sup> transients in both layers. This increase in contractility was driven by an increase in amplitude and duration of the L-type Ca2<sup>+</sup> current during phase 1. Interestingly, the β-adrenergic increase of Ca2<sup>+</sup> influx slowed the repolarization of phase 1, suggesting a competition between Ca2<sup>+</sup> and K<sup>+</sup> currents during this phase. In addition, cAMP activated L-type Ca2<sup>+</sup> currents before SR Ca2<sup>+</sup> release activated the Na+-Ca2<sup>+</sup> exchanger currents, indicating Cav1.2 channels are the initial target of PKA phosphorylation. In contrast, parasympathetic stimulation by carbachol did not have a substantial effect on amplitude and kinetics of endocardial and epicardial Ca2<sup>+</sup> transients. However, carbachol transiently decreased the duration of the AP late phase 2 repolarization. The carbachol-induced shortening of phase 2 did not have a considerable effect on ventricular pressure and systolic Ca2<sup>+</sup> dynamics. Interestingly, blockade of muscarinic receptors by atropine prolonged the duration of phase 2 indicating that, in isolated hearts, there is an intrinsic release of acetylcholine. In addition, the acceleration of repolarization induced by carbachol was blocked by the acetylcholinemediated K<sup>+</sup> current inhibition. Our results reveal the transmural ramifications of autonomic regulation in intact mice hearts and support our hypothesis that Ca2<sup>+</sup> influx that triggers ECC occurs in AP phase 1 and not in phase 2.

Keywords: mouse ventricle, endocardium, epicardium, adrenergic, cholinergic, whole heart

# INTRODUCTION

fphys-10-00773 July 2, 2019 Time: 17:44 # 2

Cardiac excitability and contractility are tightly controlled by the autonomic nervous system (ANS) (Lee and Shideman, 1959; Katz, 1967; Lindemann and Watanabe, 1985; Cohn, 1989; Henning, 1992). The ANS comprises two antagonistic branches, the sympathetic and parasympathetic nervous systems. The sympathetic nervous system modulates the cardiac function through the local release of norepinephrine and systemic action of epinephrine that alter both, Ca2+dynamics (Lee and Shideman, 1959; Evans, 1986; Marks, 2013) and electrical activity of myocardial cells. At the cellular level, these catecholamines bind to β-adrenergic receptors (β-ARs) leading to the activation of adenylyl cyclase (AC) (Hildebrandt et al., 1983; Brum et al., 1984), through a G<sup>s</sup> protein mechanism, to increase cAMP levels (Osterrieder et al., 1982). The cAMP activates PKA (Krebs, 1972; Hayes and Mayer, 1981), inducing the dissociation of the catalytic subunit to subsequently phosphorylate several key Ca2<sup>+</sup> handling proteins. Some of these key proteins include the L-type Ca2<sup>+</sup> channel (Collins et al., 1981; Osterrieder et al., 1982), ryanodine receptor (RyR2) (Suko et al., 1993; Valdivia et al., 1995) and phospholamban (PLN) (Mundiña de Weilenmann et al., 1987), a protein that inhibits the sarcoplasmic reticulum Ca2<sup>+</sup> ATPase (SERCA).

The other branch, the parasympathetic nervous system, acts through the release of acetylcholine (ACh) from postganglionic cholinergic intracardiac neurons. The locally released ACh binds to muscarinic (M2) receptors which in turn inhibit AC through a G<sup>i</sup> protein (Krebs, 1972). The AC inhibition favors low levels of cAMP, reducing the fraction of the activated PKA. The reduction of activated PKA will decrease the degree of phosphorylation of Ca2<sup>+</sup> handling proteins. In atria, SA and AV nodes, the G<sup>i</sup> protein-βγ subunits interact with an inward rectifying K<sup>+</sup> channel increasing its open probability to produce an outward hyperpolarizing K<sup>+</sup> current (IKACh) (Breitwieser and Szabo, 1988; Clapham and Kim, 1989; Kurachi et al., 1989; Wickman et al., 1999). Although the role of IKACh in regulating the electrical properties of atria, SA and AV nodes have been extensively studied, the role of IKACh during ventricular excitability is still obscure (Hiltunen et al., 2000; Rysevaite et al., 2011).

The influx of Ca2<sup>+</sup> in ventricular myocytes is driven by the activation of L-type Ca2<sup>+</sup> channels during an AP depolarization. In large mammals including humans, the prolongation of the AP plateau phase (phase 2) increases the time L-type Ca2<sup>+</sup> channels are open leading to a sustained influx of Ca2<sup>+</sup> into the myocyte. Small mammals like mice have a high metabolic demand. Thus, the heart needs to beat at higher rates to handle their metabolic needs. To cope with a higher heart rate, the mouse heart has developed a short duration ventricular AP. These characteristic kinetics promoted the idea that mouse ventricular AP lacks a plateau phase (phase 2) (Nerbonne and Kass, 2005; Dilly et al., 2006). Recently, our group presented experimental evidence showing that in situ mouse ventricular APs display a well-defined phase 2 (Ferreiro et al., 2012; Ramos-Franco et al., 2016). Interestingly, mouse AP phase 2 was more hyperpolarized than in large mammals (Kornyeyev et al., 2010; Valverde et al., 2010; Ferreiro et al., 2012) and it was driven by an influx of Na<sup>+</sup> through the Na+-Ca2<sup>+</sup> exchanger (NCX) (Ramos-Franco et al., 2016). However, up to now, it has not been possible to rule out the effect of AP phase 2 kinetics on intracellular Ca2<sup>+</sup> dynamics in mouse hearts.

As sympathetic and parasympathetic drives impact the kinetics of both phase 1 and phase 2 (Litovsky and Antzelevitch, 1990) mimicking these autonomic regulations could be a physiological way to assess the role of these AP phases on cardiac contractility across the ventricular wall. Consequently, our goal is to test the hypothesis that in mouse cardiac ventricles the influx of Ca2<sup>+</sup> that triggers excitation–contraction coupling (ECC) does not occur during phase 2. Our results reveal for the first time the transmural effects of autonomic regulation in intact mice hearts and confirm our previous observation that Ca2<sup>+</sup> influx that triggers ECC occurs in the AP phase 1 and not in phase 2.

## MATERIALS AND METHODS

#### Heart Preparation

Experiments were conducted on 8-week-old, male Balb/c mice (Charles River Labs, Wilmington, MA, United States). Mice were maintained in accordance to the National Institutes of Health Guide for the Care and Use of Laboratory Animals (NIH Publication No. 85–23, Revised 1996) and the Institutional Animal Care and Use Committee guidelines of the University of California, Merced (Protocol # 2008–201). Mice were injected intraperitoneally with sodium heparin 15 min before euthanasia. Hearts were extracted by thoracotomy and perfused in a Langendorff apparatus with tyrode solution containing (in mM): 140 NaCl, 5.4 KCl, 2 CaCl2, 1 MgCl2, 0.33 NaPO4H2, 10 HEPES, and 10 glucose, pH 7.4 and equilibrated with 100% O2. Experiments were conducted at physiological temperatures of 35–37◦C using a Peltier unit. The myosin ATPase inhibitor, blebbistatin (10 µM) was added to the tyrode solution to inhibit the heart's mechanical activity. Blebbistatin was continuously perfused throughout the whole duration of the experiment to avoid introducing movement artifacts, breaking the glass electrode and photolability. All salts and drugs were obtained from Sigma (St. Louis, MO, United States). Tertiapin was purchased from Tocris (Minneapolis, MN, United States).

#### Pressure Recordings

A new experimental approach, the micro-hemodynamic analyzer, was developed to measure left ventricular pressure in the whole intact heart (López Alarcón et al., 2019). A common device used to obtain this measurement consists of a small latex or plastic balloon coupled to a cannula that is introduced through the mitral valve. Usually, this balloon-cannula device is sutured to the atrial tissue and may cause damage. To avoid this problem, we introduced a tiny vitrectomy ophthalmic-valved trocar (Alcon, Fort Worth, TX, United States) in the apex of the left ventricle. The ventricular tissue, itself, provided a tight seal around the trocar avoiding any kind of leak from the left ventricular chamber. A 23-gage coupling valve was connected to the trocar, and the pressure at the outlet of the valve was measured with a solid-state integrated differential pressure

transducer (Honeywell, Morris Plains, NJ, United States). The pressure transducer has an expanded dynamic range and can linearly measure pressures up to 200 mmHg. All the pressure measurements were performed under isotonic constant load conditions.

# Pulsed Local Field Fluorescence Microscopy

The pulsed local field fluorescence microscopy (PLFFM) technique has been previously used by our group (Mejía-Alvarez et al., 2003; Valverde et al., 2010; Ferreiro et al., 2012; Kornyeyev et al., 2012; Aguilar-Sanchez et al., 2017). Briefly, the PLFFM technique is capable of assessing physiological parameters by exciting exogenous probes present in the tissue and detecting the light emitted by these fluorescent indicators. The excitation (532 nm Yag laser) and emitted light propagate through a multimode fiber optic (200 µm diameter, 0.67 NA) placed on the surface of the intact heart. The emitted light then travels back through the multimode fiber, dichroic mirrors, and filters (610 nm) and is focused on an avalanche photodiode (Perkin Elmer, Waltham, MA, United States) with the aid of a microscope objective. The signal is digitized by an A/D converter (National Instruments) and acquired by a PC. Here, the PLFFM technique was modified to perform a comparative study, between endocardium and epicardium at the intact heart. We included a beam splitter which allowed us to use two fiber optics. This experimental arrangement facilitated the simultaneous recording of fluorescence from two distinct anatomical regions of the heart. For endocardial measurements, a small incision was made on the surface of the ventricular wall to gain access to the endocardial layer as previously described (Aguilar-Sanchez et al., 2017). Rhod-2 AM and Di-8-ANEPPS (Thermo Fisher Scientific, Waltham, MA, United States) were used to record intracellular Ca2<sup>+</sup> transients and membrane potential, respectively. Rhod-2 AM and Di-8-ANEPPS were prepared with 20% Pluronic (Biotium, Hayward, CA, United States) in tyrode solution as previously described (Ferreiro et al., 2012; Aguilar-Sanchez et al., 2017). All dyes were sonicated for 20 min and then recirculated into the heart for 30 min at room temperature using peristaltic pumps.

# Electrical Recordings

Epicardial electrical recordings of the action potentials (APs) were obtained by means of sharp glass microelectrodes filled with 3M KCl that were connected to a high input impedance differential amplifier (WPI, Sarasota, FL, United States). Glass microelectrodes were fabricated with a micropipette puller (Flaming/Brown Micropipette Puller, Sutter Instrument Co., CA, United States) and had a resistance of 10–20 M. All the AP recordings were performed in hearts that were externally paced at 6–7 Hz.

#### Loose-Patch Photolysis

Our developed LPP technique was used to measure membrane currents during a triggered AP, at the intact heart level (Ramos-Franco et al., 2016). LPP combines three major elements: loose patch recordings, local field optical measurements, and a fiber optic laser-driven flash photolysis. The loose patch was performed using macro patch pipettes with a tip diameter of 250 µm. A fiber optic (200 µm) was placed in the interior to photolyze different drugs locally. This fiber delivered UV pulses to break photosensitive compounds on the membrane patch surface, making it possible to record the immediately resulting currents from 25 to 50 myocytes. Using this procedure, we were able to record membrane currents (capacitive and ionic) during an externally triggered AP before and after a UV flash. To dissect the UV-induced current, these sequential recordings were subtracted using pCLAMP. As we previously reported (Ramos-Franco et al., 2016), the capacitive current was not modified before and after the flash due to the electrotone imposed by the surrounding tissue on the small 200 µm UV flash site. Thus, the only current changing was the ionic current. We used the UVsensitive compound, nifedipine (10 µM), such that the resulting currents were Ca2+-driven. The LPP was used to measure any changes in the Ca2+-driven currents in the epicardium under either continuous β-adrenergic or cholinergic stimulation. In addition, for transient β-adrenergic stimulation, we used the UVsensitive caged cAMP (4,5-dimethoxy-2-nitrobenzyl adenosine 3 0 ,50 -cyclic monophosphate).

## RNA Analysis

Hearts were perfused with RNAlater (Qiagen) to inhibit the RNAses. Then, left ventricular square pieces of 4 × 4 mm were included in agarose to facilitate slicing. Specimens were sliced in 200 µm sections using a vibratome (Leica VT1000S). For RNA extraction, individual slices were placed in a 1.5 mL Eppendorf containing 0.5 mL of Trizol reagent. Tissue was homogenized with an ultrasonic cell disruptor. The homogenate was incubated for 5 min at room temperature and 0.2 mL of chloroform per mL of Trizol was added. After 2–3 min of incubation, the sample was centrifuged at 12,000 × g 15 min 4◦C. The clear phase was transferred to a new tube and 500 µL of isopropanol per mL of Trizol was added and centrifuged at 12,000 × g for 10 min at 4 ◦C. The supernatant was removed and 1 mL of 75% ethanol per mL of Trizol was added and centrifuged at 7,500 × g for 5 min at 4◦C. The dried pellet was dissolved in 30 µL of RNAse free water. For real time PCR, RNA was quantified (NanoDrop 1000 Spectrophotometer) and 2 µg was used for the reverse transcription reaction. The 2 µg of RNA was mixed with 1 µL of random primers (200 ng/µL) and water up to 12.5 µL. 7.5 µL of a MIX solution containing 4 µL of reaction buffer, 0.5 µL of RNAse free water, 2 µL dNTPs mix and 1 µL of RevertAid was added before the thermocycle cycles began. cDNA samples were diluted with water, 6.25 µL of Mix (FastStart Roche), 0.37 µL of forward primers, and 0.37 µL of reverse primers (see **Supplementary Table 1**). The reaction samples were placed in a 96-well plate and real time qPCR cycles were started. Finally, mRNA levels were normalized to GAPDH mRNA and analyzed by the Pfaffl method.

#### Statistical Analysis

It is important to mention that in whole heart experiments there are two main sources of variance. First, the hearts are not identical between each other and second, although we are

measuring Ca2<sup>+</sup> transients and APs in the same region (the mid region of the left ventricle), it is impossible to perform the recordings in the same precise location between different hearts. Thus, the data is presented as the measured times with their standard error (SEM) in Tables and then as percentage change in Graphs, where data was normalized to the control values. To assess electrical changes, AP traces were evaluated at certain repolarization times. Specifically, the time it takes for the AP to reach 30%, half of phase 2, or 90% repolarization was assessed and referred to as APD30, APD half phase 2, or APD90, respectively. To evaluate the kinetics of the recorded Ca2<sup>+</sup> transients, these were normalized between zero (minimum fluorescence) and one (maximum fluorescence). The kinetics parameters of the Ca2<sup>+</sup> transients evaluated were the rise time (time for the transient to rise between 10% and 90%), time-topeak (time for the transient to reach a maximum amplitude), half duration (time for half of the Ca2<sup>+</sup> transient to be complete), and fall time (time taken for the transient to fall between 10% and 90% of the relaxation). Each recorded parameter for AP and Ca2<sup>+</sup> transient kinetics, control and non-control experiments, were evaluated and normalized to the control values for each heart used. After this normalization, data was compiled and statistical analysis was performed. Data is presented as multiple measurements (n; dot cloud) recorded on different hearts (N) with the mean ± SEM (solid lines). The statistical significance was tested using a two-sample Kolmogorov–Smirnov test (OriginPro 2019). The difference was considered to be significant if the p-value was <0.01.

#### RESULTS

In the heart, both excitability and contractility are tightly controlled by the ANS. Although there is substantial information about sympathetic and parasympathetic regulation of the electrical and mechanical function, it is not clear how autonomically driven changes in the AP kinetics impact contractility in the mouse heart, which is often used as a model system.

In a beating heart, the classical way to assess the degree of autonomic regulation is to evaluate the developed pressure during the cardiac cycle. Using intact perfused mouse hearts, we measured the effect of catecholamines in the developed pressure during the cardiac cycle. **Figure 1** illustrates the time course of the developed pressure using the micro-hemodynamic analyzer (µHA) (López Alarcón et al., 2019). As was described in the "Materials and Methods" section, this newly developed method does not use an intraventricular balloon. Specifically, an ophthalmic-valved trocar was introduced through the apex of the left ventricle to gain access to the intra chamber pressure. The pressure at the outlet of the electro-valve was then measured with a solid-state integrated pressure transducer. This method permits an easy access to the ventricular chamber with no need of a more intrusive intraventricular balloon. **Figure 1A** illustrates how a sympathetic drive can increase the developed pressure in Langendorff perfused hearts. Indeed, when the hearts were coronary retro-perfused with 500 nM isoproterenol, we were able to record systolic pressures that were significantly larger than in control conditions. **Figure 1B** summarizes the pressure data obtained with our device, before and after the coronary perfusion with isoproterenol. The amplitudes of the developed pressures were normalized to the mean control. Isoproterenol induced a significant increase in the developed pressure (1 ± 0.06 for the control vs. 1.62 ± 0.05 for isoproterenol, N = 4 hearts p < 0.01), showing that the perfused hearts used here respond to a β-adrenergic drive.

To assess the β-adrenergic regulation in the endocardial and epicardial layers, we examined the properties of Ca2<sup>+</sup> transients in those particular regions using the PLFFM technique. **Figure 2A** illustrates normalized Ca2<sup>+</sup> transients recorded simultaneously at the endocardial and epicardial layer. Perfusion of the hearts with 500 nM isoproterenol increased the amplitude of Ca2<sup>+</sup> transients in both layers (**Figures 2B,C**). A summary

FIGURE 2 | β-Adrenergic stimulation modulates Ca2<sup>+</sup> transients in endocardium and epicardium. (A) Normalized fluorescent Ca2<sup>+</sup> transients recorded, simultaneously, from endocardium and epicardium using Rhod-2 and the pulsed local field fluorescence microscopy technique. (B) Representative fluorescent traces of the endocardial Ca2<sup>+</sup> transients before (black) and after 500 nM isoproterenol (ISO; orange). Right panel: Ca2<sup>+</sup> transient traces normalized to their corresponding maximum fluorescence. (C) Representative Ca2<sup>+</sup> transients recorded from the epicardium before (black) and after isoproterenol (cyan). Right panel: Ca2<sup>+</sup> transients normalized to their maximum fluorescence. (D) Summary of the normalized amplitude of the Ca2<sup>+</sup> transient changes, from the endocardium (circles) and epicardium (squares), before and after isoproterenol. (E) Data showing the kinetic changes in the normalized Ca2<sup>+</sup> transients before and after isoproterenol. Means ± SEM are represented as the solid horizontal lines. <sup>∗</sup>p < 0.01, N = 4 hearts.

TABLE 1 | Isoproterenol effects on Ca2<sup>+</sup> transient kinetics and AP durations.


Endo, endocardium; Epi, epicardium; Iso, isoproterenol; RT, rise time; TP, time to peak; HD, half duration; FT, fall time. APD, AP durations measured at 30, half phase 2, and 90% repolarization. <sup>∗</sup>Kolmogorov–Smirnov test, p < 0.01, N = 4.

of the effect of isoproterenol on the amplitude of the Ca2<sup>+</sup> transients is presented in **Figure 2D** where it is possible to observe that there is a significant increase in both the endocardium (32.2 ± 0.83%) and the epicardium (32.9 ± 0.85%). This result is consistent with the effect of isoproterenol on the developed pressure as shown in **Figure 1**. Isoproterenol not only has an inotropic effect but also a lusitropic action. In **Figures 2B,C** it is possible to observe that the relaxation of the Ca2<sup>+</sup> transients is faster in the presence of isoproterenol. Different kinetic parameters for endocardial and epicardial Ca2<sup>+</sup> transients are compiled from 4 different hearts in **Table 1** (measured times) and **Figure 2E** (normalized percentages). Isoproterenol induced significant changes in all the measured parameters of endocardial and epicardial Ca2<sup>+</sup> transients. The rise time of the Ca2<sup>+</sup> transients marginally increased (6.5 ± 0.83% for endocardium and 11.6 ± 3.1% for epicardium). As expected, the time to peak of the Ca2<sup>+</sup> transients follow the same trend (11.1 ± 1.3% for endocardium and 5.9 ± 1.70% for epicardium). Isoproterenol induced a reduction in Ca2<sup>+</sup> transients' half duration. The half duration is statistically significant shorter in the presence of isoproterenol (5.9 ± 0.42% endocardium and 8.8 ± 0.45% epicardium). However, the changes in the fall time of the relaxation (20.3 ± 0.96% endocardium and 26.3 ± 0.92% epicardium) were larger than those observed in the half durations. The large effect of isoproterenol on the fall time compared to the half duration could be explained by the fact that isoproterenol not only has a positive lusitropic effect but also changes the morphology of the Ca2<sup>+</sup> transients as shown in **Figures 2B,C**. The changes in fall time for the epicardial were significantly faster (p < 0.01) than those in the endocardial layer (**Figure 2E**, bottom panel). It is known that cardiac myocytes from the endocardial, midmyocardial and epicardial layers are characterized by molecular heterogeneities at the level of transcription and expression of proteins. Thus, the faster fall time observed in epicardium can be molecularly defined by a higher expression of SERCA in this layer. Indeed, when we assessed the levels of mRNA transcription by qPCR, there were significant changes in the expression of RyR2, SERCA and Kv4.3, a K<sup>+</sup> channel responsible for Ito that consistently shows a higher expression in the ventricular epicardium (**Figure 3**). However, we did not find significant differences in the expression of L-type Ca2<sup>+</sup> channels, NCX, PKA, M2 muscarinic receptors and β-ARs between both layers. Since the expression of both RyR2 and SERCA is unevenly distributed across the ventricular wall, we expected to have a differential response between endocardium and epicardium.

Another remarkable difference that distinguishes the endocardial and epicardial layers is their electrical properties. One of the main differences is in the phase 1 repolarization of the AP due to the higher expression of Kv4.3 in the epicardial layer (**Figure 3**). Thus, we designed experiments to evaluate the hypothesis that β-adrenergic stimulation will differentially affect the time course of endocardial and epicardial APs. **Figure 4A** compares optically recorded ventricular APs in the endocardium and epicardium, where all phases are distinguished. The APs in both layers have a well-defined phase 2. However, the repolarization rate of phase 1 is faster in epicardium than in endocardium (**Table 1**). The perfusion with

500 nM isoproterenol produced changes in the AP morphology (**Figures 4B,C**). Unexpectedly, isoproterenol decreased the repolarization rate of phase 1, in both the endocardium (**Figure 4B**, inset) and the epicardium (**Figure 4C**, inset). **Table 1** and **Figure 4D** summarizes different endocardial and epicardial AP parameters from four hearts. The most salient features of β-AR activation were an increase in the duration of phase 1 (APD30) and a shortening of the late phases of the AP (APD half phase 2 and APD90). Indeed, isoproterenol induced an increase in APD30 of 21.2 ± 2.40% in endocardium and 27.6 ± 2.18% in epicardium. As in other mammalian species (Litovsky and Antzelevitch, 1990), isoproterenol induced a shortening in APD half phase 2 duration in endocardium and epicardium, 6.9 ± 0.90% and 12.6 ± 0.59%, respectively. Isoproterenol also shortened APD90, 11.5 ± 0.96% in the endocardium and of 9.8 ± 0.73% in epicardium (**Figure 4D**).

The decrease of the repolarization rate of phase 1 with isoproterenol can be attributed to an increase in the amplitude of L-type Ca2<sup>+</sup> currents. We recently demonstrated that most of the Ca2<sup>+</sup> influx that triggers Ca2+-induced Ca2+-release (CICR) occurs during phase 1 (Ramos-Franco et al., 2016) and that the influx of Ca2<sup>+</sup> through L-type Ca2<sup>+</sup> channels slows down the rate of repolarization during phase 1 (López Alarcón et al., 2019). To directly evaluate any changes in the L-type Ca2<sup>+</sup> currents during isoproterenol perfusion, we used the novel Loose Patch Photolysis (LPP) technique. We simultaneously acquired Ca2+-driven currents with the corresponding APs before and after isoproterenol. In **Figure 5A**, the AP showed an increase in APD30 (inset) as well as in phase 2. However, due to the presence of nifedipine, this increase in APD30 is smaller than in its absence (**Table 1** and **Figure 4D**). **Figure 5B** shows a typical experiment where Ca2<sup>+</sup> currents, activated by the photolysis of nifedipine on the membrane patch surface (see "Materials and Methods" section for details), were recorded before and after the heart was perfused with isoproterenol. We observed a clear increase in the amplitude of the Ca2+-driven currents, displayed as an early fast component (iearly) and a late slower component (ilate). Unfortunately, the simultaneous measurement of Ca2<sup>+</sup> driven currents in the endocardium and epicardium was not possible due to the size of the patch pipettes. To reach the endocardial layer with the patch pipette would require a large size hole that would result in electrical conduction alterations. **Figure 5C** shows the effect of isoproterenol on the amplitude of iearly that increased from 13.9 ± 0.2 nA/nF to 27.6 ± 0.2 nA/nF. In a recent work (Ramos-Franco et al., 2016), we showed that Ca2<sup>+</sup> released from the sarcoplasmic reticulum (SR), activated the NCX in the forward mode, resulting in ilate (Ferreiro et al., 2012; Ramos-Franco et al., 2016). This ilate was abolished (**Figure 5D**) when we blocked SR Ca2<sup>+</sup> release with a combination of ryanodine and thapsigargin. The family of traces correspond to consecutive current recordings, where the Ca2<sup>+</sup> currents were evoked locally by progressively photolyzing greater fractions of nifedipine by using UV laser pulses of increasing beam energy (fluence). To test the hypothesis that β-adrenergic stimulation directly increases the amplitude of the Ca2<sup>+</sup> current occurring during phase 1, we performed an identical experiment to the one presented in **Figure 5D** but in the presence of isoproterenol (**Figure 5E**). We observed that isoproterenol dramatically increased the amplitude of iearly. The simultaneous recording of the corresponding APs shows that isoproterenol only altered the kinetics of phase 1 (**Figure 5E**, right panel). **Figure 5F** shows the peak values of iearly as a function of the fluence, before and after isoproterenol. It is possible to observe that iearly saturates at high fluence values (>16 J/cm<sup>2</sup> ) indicating that most of the nifedipine was photolyzed.

Expanded view of AP phase 0 and 1 to visualize the isoproterenol effect. (D) Summary of the normalized APD30, APD half phase 2 and APD90 changes from the endocardium (circles) and epicardium (squares), before and after isoproterenol. Means ± SEM are represented as the solid horizontal lines. <sup>∗</sup>p < 0.01, N = 4 hearts.

**Figure 5** shows that isoproterenol induced an increase in the amplitude of iearly. It is well known that catecholamines activate PKA, which in turn leads to the phosphorylation of multiple targets involved in Ca2<sup>+</sup> signaling (i.e., the L-type Ca2<sup>+</sup> channels, RyR2 and PLN). In order to determine which target is phosphorylated first, the effect of catecholamines on the time course of Ca2+-driven currents was evaluated. We performed experiments in which we mimicked a transient β-adrenergic stimulation by photolyzing caged 3<sup>0</sup> ,50 -cyclic adenosine monophosphate (cAMP). A typical response is illustrated in **Figure 6A** where upon immediate photolysis (purple arrow), both components of the Ca2+-driven current, iearly and ilate, began to increase over time. To better illustrate the changes of iearly and ilate, in **Figure 6B,** we overlapped the current traces at different time points: before, 1 s and 10 s after the uncaging. Plotting the amplitude peaks of iearly and ilate as a function of time during consecutive cardiac cycles, highlighted that both currents changed over time, but with a greater effect

on iearly (**Figure 6C**). When we normalized these currents (**Figure 6D**), it became apparent that iearly also responded sooner to the transient β-adrenergic stimulation.

Sympathetic stimulation is highly antagonized by the other branch of the ANS, the parasympathetic nervous system. Parasympathetic stimulation is thought to generally predominate over an existing sympathetic activation. Next, we evaluated the response of endocardium and epicardium to muscarinic cholinergic agonists. **Figure 7A** shows that upon perfusion with 5 µM carbachol, there were no measurable changes in the developed ventricular pressure. Our pressure data (**Figure 7B**) shows there was no significant differences before and after carbachol (0.99 ± 0.15 vs. 0.99 ± 0.07, N = 3 hearts). These results suggest that in the absence of a β-adrenergic stimulus, carbachol by itself does not induce a decrease in the developed ventricular pressure during the cardiac cycle.

To assess the cholinergic regulation of contractility, we examined the properties of Ca2<sup>+</sup> transients across the ventricular

Ca2+-driven currents recorded after the local photolysis of nifedipine using the Loose Patch Photolysis technique showing the iearly and i late components upon β-adrenergic stimulation. (C) Data from the iearly component, representing the L-type Ca2<sup>+</sup> currents, before (gray) and after isoproterenol (pink). The means ± SEM are represented as the solid horizontal lines. <sup>∗</sup>p < 0.01, N = 4 hearts. (D) Representative traces of Ca2+-driven currents recorded consecutively at increasing UV energies (fluences) in control conditions in the presence of ryanodine (Ry, 10 µM) and thapsigargin (Tg, 2 µM). Right panel: Simultaneously recorded AP in the same conditions. (E) Ca2+-driven currents recorded consecutively at increasing UV energies (fluences) with isoproterenol in the presence of ryanodine (Ry, 10 µM) and thapsigargin (Tg, 2 µM). Right panel: Simultaneously recorded AP in the same conditions. (F) Peak values of iearly as a function of the fluence, before (black) and after isoproterenol (blue) in the presence of Ry and Tg.

wall in the presence of a cholinergic agonist and antagonist (**Table 2**). **Figure 8** illustrates normalized Ca2<sup>+</sup> transients recorded simultaneously at both the endocardial (**Figure 8A**) and epicardial (**Figure 8B**) layer with carbachol and carbachol and atropine. Perfusion of the hearts with 5 µM carbachol induced a minor but statistically significant decrease in the amplitude of the Ca2<sup>+</sup> transients by 3.5 ± 0.35% in the endocardium, and 7.5 ± 0.58% in epicardium (**Figure 8C**). However, this effect was smaller than the increase in the amplitude of Ca2<sup>+</sup> transients observed when hearts were perfused with isoproterenol. The effects of carbachol on various kinetic parameters of the Ca2<sup>+</sup> transients are summarized (N = 5 hearts) in **Figure 8D**. Although, carbachol did not significantly change the normalized rise time in endocardium or epicardium, the time to peak of the Ca2<sup>+</sup> transients were significantly increased in endocardium (5.5 ± 1.17%) but not in epicardium (2.5 ± 1.83%). Carbachol induced a significant increase in the half duration (endocardium 1.9 ± 0.3%; epicardium 3.0 ± 0.57%) and fall time of the Ca2<sup>+</sup> transients (endocardium 3.3 ± 0.69%; epicardium 4.1 ± 1.07%). Although the changes in the fall times

are significantly different, they were smaller than the kinetic effects observed with isoproterenol (**Figure 2**). Thus, we conclude that perfusion with carbachol did not have a large impact on the amplitude and the kinetics of endocardial and epicardial Ca2<sup>+</sup> transients, compared to those observed with the perfusion of isoproterenol. When the hearts were perfused with both agonist and antagonist (5 µM carbachol plus 40 µM atropine) it was possible to observe significant values above the control, for rise time (epicardium 5.3 ± 3.14%), time to peak (endocardium 6.7 ± 1.18%; epicardium 6.5 ± 2.01%), half duration (endocardium 9.4 ± 0.34%; epicardium 10.3 ± 0.64%) and fall time (endocardium 4.2 ± 0.74%; epicardium 6.8 ± 1.29%).

Since the effects induced by carbachol in the contractility were minor, we decided to explore how the activation of a parasympathetic pathway affects the electrophysiological behavior of the endocardial and the epicardial layers. In **Table 2** and **Figure 9** we compared ventricular APs optically recorded in the endocardium (**Figure 9A**) and epicardium (**Figure 9B**). The perfusion with 5 µM carbachol produced significant changes in the AP morphology. **Figure 9C** summarizes different endocardial

the solid horizontal lines, N = 3 hearts.


Endo, endocardium; Epi, epicardium; CCh, carbachol; RT, rise time; TP, time to peak; HD, half duration; FT, fall time. APD, AP durations measured at 30, half phase 2, and 90% repolarization. <sup>∗</sup>Kolmogorov–Smirnov test, p < 0.01, N = 5.

Epi Control 3.24 ± 0.11 (n = 425) 68.90 ± 0.15 (n = 581) 81.62 ± 0.18 (n = 426) Epi CCh 3.22 ± 0.12 (n = 442) 52.15 ± 0.19 (n = 544)<sup>∗</sup> 65.89 ± 0.27 (n = 405)<sup>∗</sup> Epi CCh + Atropine 3.08 ± 0.27 (n = 612)<sup>∗</sup> 66.05 ± 0.17 (n = 628)<sup>∗</sup> 79.46 ± 0.18 (n = 629)<sup>∗</sup>

and epicardial AP parameters from five hearts. Carbachol did not alter APD30 in epicardium, but had a very modest effect on the endocardial layer (6.9 ± 1.16%). Interestingly, the APD half phase 2 (endocardium 33.4 ± 0.62%; epicardium 24.9 ± 0.42%) and APD90 (endocardium 20.1 ± 0.68%; epicardium 19.4 ± 0.43%), were significantly reduced by the muscarinic stimulation. An interesting effect by the addition of atropine in the presence of carbachol was the recovery of the tested parameters. The APD half phase 2 in endocardium recovered to 90.0 ± 1.12% and the epicardium to 95.7 ± 0.29%. The APD90 for endocardium and epicardium recovered 106.6 ± 0.67% and 100.8 ± 0.26% respectively, from the control values. This suggests that atropine, aside from blocking carbachol's action, may also be blocking an intrinsic, existing cholinergic tone in the isolated heart that modulates the time course of the ventricular AP. Moreover, the fact that carbachol clearly modified both APD half phase 2 and APD90 without changing the kinetic properties of Ca2<sup>+</sup> transients indicates that the Ca2<sup>+</sup> influx that triggers Ca2<sup>+</sup> induced Ca2<sup>+</sup> release may not occur during phase 2. To further explore this possibility, we then evaluated the properties of Ca2<sup>+</sup> currents driven by its own AP, when the heart was exposed to a parasympathetic drive. Indeed, the decrease of APD half phase 2 with carbachol could be due to a decrease in the amplitude of L-type Ca2<sup>+</sup> currents. For example, a reduction in the L-type Ca2<sup>+</sup> current will lead to a decrease in the Ca2<sup>+</sup> released from the SR. In turn, this will result in less activation of the NCX in its forward mode. To test this scenario, we evaluated the effect of carbachol on the time course of the L-type Ca2<sup>+</sup> current during an evoked AP. In **Figures 10A,B**, we recorded the epicardial Ca2+-driven currents that were activated by the photolysis of nifedipine before and after carbachol perfusion. **Figure 10C** shows the superposition of two Ca2+-driven current traces before and after the carbachol perfusion. The Ca2+-driven currents

displayed an iearly and ilate component, but they were not affected by the presence of carbachol. **Figure 10D** shows that carbachol did not significantly modify the amplitude of iearly.

Since carbachol did not induce changes in the amplitude of L-type Ca2<sup>+</sup> currents, we examined if a repolarizing outward current (associated with the cholinergic stimulation) was involved in the shortening of the APD half phase 2. Although it is well established in the atrium that the current of an acetylcholine (ACh) activated potassium channel, IKACh, can severely affect the AP repolarization, the role of this current in the ventricle is still a matter of debate. The role of the IKACh was evaluated by perfusing the hearts with tertiapin, a bee venom that can block with high affinity GIRK1/GIRK2 hetero-tetramers (Hashimoto et al., 2006). GIRK1/GIRK2 heterotetramers form the pore of the inward rectifying K<sup>+</sup> channels responsible for IKACh. The perfusion of tertiapin prevented the carbachol shortening of the AP in the epicardium **Table 3** and **Figure 11**. **Figure 11A** illustrates that in the presence of tertiapin, 5 µM carbachol is not able to produce any visible change in the kinetics of the electrically recorded APs. However, in the absence of tertiapin, 5 µM carbachol modified APD half phase 2 and APD90 (**Figure 11B**). The summarized epicardial AP data (**Figure 11D**; N = 5 hearts) shows tertiapin induced statistically significant differences in all of the AP phases (5.2 ± 0.38% for APD30, 0.8 ± 0.56% for APD half phase 2 and 0.5 ± 0.02% for APD90). Although these differences are statistically significant, the magnitude of the change is so small (see **Table 3**) that we do not consider that these changes will have any physiological effect. Interestingly, in the presence of tertiapin, carbachol was not able to induce the same shortening effect, as in the absence of tertiapin, on the total AP duration (APD half phase 2 and APD90, 3 ± 0.71% and 5.3 ± 0.19%, respectively). To corroborate that tertiapin was blocking the carbachol effect, we perfused the heart

FIGURE 8 | Cholinergic stimulation has a low impact on the Ca2<sup>+</sup> transients in endocardium and epicardium. (A) Normalized fluorescent Ca2<sup>+</sup> transients recorded, simultaneously, from endocardium and (B) epicardium using Rhod-2 and the pulsed local field fluorescence microscopy technique before and after 5 µM carbachol (CCh) and in the presence of both carbachol and 40 µM atropine. Overlapping Ca2<sup>+</sup> transients were normalized to their maximum fluorescence. (C) Summary of the normalized amplitude of the Ca2<sup>+</sup> transient changes, from the endocardium (circles) and epicardium (squares), before and after carbachol. (D) Data showing the kinetic changes in the normalized Ca2<sup>+</sup> transients before and after carbachol, and after carbachol plus atropine. Means ± SEM are represented as the solid horizontal lines. <sup>∗</sup>p < 0.01, N = 5 hearts.

FIGURE 9 | Cholinergic stimulation modifies overall AP morphology in endocardium and epicardium. (A) Normalized APs recorded optically with Di-8-ANEPPS from the endocardium and (B) epicardium before and after perfusion with 5 µM carbachol (CCh). (C) Summary of the normalized APD30, APD half phase 2 and APD90 changes from the endocardium (circles) and epicardium (squares), before and after carbachol, and after carbachol plus atropine. Means ± SEM are represented as the solid horizontal lines. <sup>∗</sup>p < 0.01, N = 5 hearts.

with carbachol after tertiapin was washed out (**Figures 11B,E**). Carbachol significantly modified APD30 (5.9 ± 0.42%) above control and decreased both APD half phase 2 (12.4 ± 0.22%) and APD90 (9.9 ± 0.19%). These results were consistent with the ones reported in **Figure 9C**. Overall, these results suggest that the AP shortening produced by a cholinergic stimulation is due to the activation of IKACh instead of a Ca2+-driven mechanism. Finally, we evaluated if in the perfused heart, there is a tonic release of acetylcholine from the postganglionic parasympathetic terminals located in the ventricular wall. **Figure 11C** shows that 40 µM atropine prolongs the late repolarization of the epicardial AP. Indeed, although the increase in APD half phase 2 (6.0 ± 0.11%) and APD90 (4.9 ± 0.02%) are not prominent, the increase is statistically significant and reflects a change >5 ms (**Table 3** and **Figure 11F**). This result indicates that it is likely that an intrinsic parasympathetic tone is present in the isolated heart and can slightly decrease the duration of ventricular APs.

#### DISCUSSION

Although, there is a substantial body of experimental information on sympathetic and parasympathetic regulation of the electrical and mechanical properties of the heart, there is

of the peak values of iearly, before and after carbachol.

no conclusive evidence of how autonomic driven changes affects the relationship between contractility and excitability in the mouse heart. Neither is it known how sympathetic and parasympathetic nervous system regulate those properties across the ventricular wall. In this study, we investigated how either β-adrenergic or muscarinic cholinergic agonists regulate the time course of ventricular APs and how these electrical changes correlate with the left ventricular pressure, Ca2<sup>+</sup> transients, and Ca2+-driven currents at the whole-heart level. Our results indicate that in the mouse heart the influx of Ca2<sup>+</sup> that triggers the ECC happens during AP phase 1 and not phase 2, contrary to what occurs in larger mammals (Boyett, 1986; Antzelevitch et al., 1991; Qu et al., 2013). Indeed, in isolated mice hearts, sympathetic stimulation reduced the rate of repolarization and increased the amplitude of epicardial Ca2<sup>+</sup> currents. The increase in Ca2<sup>+</sup> influx drives an increase in the amplitude of Ca2<sup>+</sup> transients both in the endocardium and the epicardium that finally produces an increase in the developed pressure. In contrast, parasympathetic stimulation dramatically reduced the duration of phase 2 without inducing major changes in Ca2<sup>+</sup> currents, Ca2<sup>+</sup> transients and developed pressure indicating that the Ca2<sup>+</sup> signaling involved in controlling the ECC does not occur during the AP phase 2.

### β-Adrenergic Stimulation Across the Ventricular Wall Regulate Ca2+-Driven Mechanisms

It is well known that β-adrenergic stimulation will accelerate the heart rate (chronotropism), and the higher systemic demands will be met by increasing the systolic pressure (inotropism) as well as the rate of relaxation (lusitropism) of the developed pressure. In our mouse model, exposure to isoproterenol generated an increase in the left ventricular developed pressure (**Figure 1**). This increase in inotropism resulted because of an increase in the amplitude of cytosolic Ca2<sup>+</sup> transients in both endocardial and epicardial layers (**Figure 2**). Although we found no significant changes in the expression of β-adrenergic and muscarinic

#### TABLE 3 | Tertiapin, carbachol, and atropine effects on AP durations.


Epi, epicardium; CCh, carbachol. APD, AP durations measured at 30, half phase 2, and 90% repolarization. <sup>∗</sup>Kolmogorov–Smirnov test, p < 0.01, N = 5.

FIGURE 11 | Cholinergic stimulation acts via IKACh in ventricle. (A) APs recorded from the epicardium using glass microelectrodes before (black), with 100 nM tertiapin (cyan) and after tertiapin with 5 µM carbachol (CCh; yellow). (B) Epicardial APs with 5 µM carbachol (green) after 20 min washout from tertiapin + carbachol (black). (C) Effect of 40 µM atropine (dark green) on the time course of epicardial APs. (D) Data summarizing epicardial APD30, APD half of phase 2, and APD90 of the normalized APs before (gray) and after 100 nM tertiapin followed by the perfusion of 5 µM carbachol plus tertiapin. (E) APD30, APD half phase 2, and APD90 after a 20 min washout from tertiapin + carbachol (gray) and then perfusion with 5 µM carbachol (green). (F) APD30, APD half phase 2, and APD90 after perfusion of 40 µM atropine. Means ± SEM are represented as the solid horizontal lines; <sup>∗</sup>p < 0.01, N = 5 hearts.

receptors between the endocardial and the epicardial layer, there were differences in the expression of downstream targets (**Figure 3**). For example, there was a higher expression of SERCA in the epicardial layer. This molecular difference correlates well with the recorded shorter half durations and fall times of Ca2<sup>+</sup> transients with isoproterenol, in this layer (**Figure 2** and **Table 1**). In addition, the higher expression of RyR2 in the epicardium can explain the faster activation of Ca2<sup>+</sup> transients and the larger β-adrenergic modulation of the rise time in this layer (Mattiazzi et al., 2015).

There are intrinsic functional differences in the sympathetic regulation of the endocardial and epicardial APs. The differences are imposed by the differential expression of ionic channels in endocardium versus epicardium as reflected by their respective AP morphologies (Antzelevitch et al., 1991; Antzelevitch and Fish, 2001; Abd Allah et al., 2012). Indeed, in **Figure 3** we showed a higher expression of Kv4.3 in the epicardial layer. The higher expression of this K<sup>+</sup> channel explains the faster rate of repolarization of AP phase 1 in epicardium. In addition, β-adrenergic stimulation affected the AP's morphology by increasing the duration of the AP phase 1 and the amplitude of phase 2 across the ventricular wall (**Figure 4**). One possibility that may explain the increase in the AP phase 1 duration is a reduction in the amplitude of a repolarizing current, such as Ito. There are few reports of β-adrenergic regulation of Ito. Some experimental evidence (Nakayama et al., 1989; Setién et al., 2013) showed that the catecholamine-induced phosphorylation of the transient outward current (Ito) components increased the amplitude of this fast repolarizing current. This possibility is unlikely to occur in our experimental conditions. Increase of Ito, would accelerate instead of slowing the repolarization rate of phase 1. In contrast, a reduction of the amplitude of Ito by norepinephrine and alpha adrenoceptor stimulation has been reported in healthy rat ventricular myocytes (Ravens et al., 1989; Gallego and Casis, 2001) which may explain our results. However, this is very unlikely in our experiments because isoproterenol is a β-adrenergic and not an α-adrenergic agonist at the concentrations used here. In either scenario, the increase in the duration of phase 1 (APD30) is most likely a consequence of the competition between Ito and the inward L-type Ca2<sup>+</sup> current (López Alarcón et al., 2019). Indeed, a β-adrenergic increase of the L-type Ca2<sup>+</sup> current (i.e., by PKA phosphorylation) during phase 1 will maintain a depolarization of this phase for a longer time, increasing APD30. Additionally, in the presence of nifedipine, the effect of isoproterenol on APD30 was smaller (**Figure 5A**). Altogether, these results support the idea that the increase in duration of AP phase 1 correlates well with larger L-type Ca2<sup>+</sup> currents under isoproterenol stimulation. These findings agree with previous experiments from our group in which we showed that the transmembrane Ca2<sup>+</sup> influx that triggers Ca2+-induced Ca2+-released (CICR) enters during phase 1 (Ramos-Franco et al., 2016).

A final proof of concept was to demonstrate that isoproterenol mediates an increase of the amplitude of the L-type Ca2<sup>+</sup> current during phase 1. Indeed, we corroborated that under β-adrenergic stimulation there is an increase in the amplitude of the L-type Ca2<sup>+</sup> current during phase 1 (**Figure 5**). This could result in (1) the decrease of the rate of AP repolarization during phase 1, and (2) the depolarization induced by the increase in the L-type Ca2<sup>+</sup> current will further activate L-type Ca2<sup>+</sup> channels with the enlargement of this Ca2<sup>+</sup> current. Interestingly, in the presence of a submaximal concentration of nifedipine, isoproterenol increased the duration of phase 2. This effect can be attributed to an increase in the amplitude of the Ca2<sup>+</sup> transients as shown in **Figure 2**. The most likely possibility is that an increase in the amplitude of the Ca2<sup>+</sup> transient will induce a larger activation of the NCX in the forward mode as previously suggested (Ferreiro et al., 2012). The validity of this hypothesis was assessed in **Figure 5B** where we found an increase in the fast and early inward current (iearly) through the L-type Ca2<sup>+</sup> channels and a significant increase in the amplitude of the late NCX current as reflected by ilate. These results suggest that changes in the electrical excitability induced by the β-adrenergic stimulation across the ventricular wall are highly mediated through a Ca2+ driven mechanism.

As we showed, the sympathetic branch of the ANS can induce mechanical changes in the intact heart by altering the underlying plasma membrane Ca2<sup>+</sup> currents and consequently, Ca2<sup>+</sup> release from the SR. However, PKA activation can affect other targets. For example, PLN (Toyofuku et al., 1993; Frank and Kranias, 2000), a protein that regulates the activity of SERCA can be phosphorylated on residue S16 (Chu et al., 2000; Valverde et al., 2006). This phosphorylation will relieve the inhibitory effect PLN has on SERCA, that will result in increased contractility (Luo et al., 1994). On the other hand, RyR2 can be phosphorylated by PKA in the residues S2030 (Huke and Bers, 2008) and S2808 (Obayashi et al., 2006; Chen-Izu et al., 2007). This phosphorylation can increase the RyR2 open probability. However, it is not fully established that the RyR2 phosphorylation by itself will increase contractility mostly because of the changes produced in the intra-SR Ca2<sup>+</sup> content as a consequence of the interplay between release and recapture (Eisner et al., 2004). Although it is accepted that the phosphorylation of these targets will modify Ca2<sup>+</sup> release from the SR, the timing of the PKA induced phosphorylation is unknown. Specifically, it is unclear which phosphorylated target will have a faster effect on the ECC process. **Figure 6** clearly showed that under a global increase in cAMP concentration the L-type Ca2<sup>+</sup> current is the first transport mechanism that is activated by PKA. However, this may not be the case in vivo, due to the documented compartmentation of cAMP resulting from localized cyclic nucleotide phosphodiesterase (PDE) activity (Nikolaev et al., 2006; Leroy et al., 2008). In our experimental conditions (cAMP cell loading), we assumed a homogeneous cellular distribution of cAMP. Based on this, we propose that PKA phosphorylation of the L-type Ca2<sup>+</sup> channels has a dual effect. Immediately after phosphorylation, the increase in L-type Ca2<sup>+</sup> current will increase Ca2<sup>+</sup> release from the SR. Consequently, the increase in the intracellular Ca2<sup>+</sup> will promote a larger uptake of Ca2<sup>+</sup> into the SR mediated by SERCA. Finally, the phosphorylation of PLN will induce a later increase in the intra-SR Ca2<sup>+</sup> content that will further increase SR Ca2<sup>+</sup> release.

#### Cholinergic Stimulation Across the Ventricular Wall Did Not Alter Ca2<sup>+</sup> Dynamics

Although the sympathetic branch of the ANS produced mechanical changes in the intact heart by altering the underlying Ca2<sup>+</sup> currents, the cholinergic stimulation did not. Interestingly, the left ventricular pressure showed no changes when exposed to carbachol (**Figure 7**). However, there were minor changes in the amplitude of the Ca2<sup>+</sup> transients in both endocardium and epicardium (**Figure 8**). Carbachol induced some statistically

significant changes in most of the kinetic parameters of Ca2<sup>+</sup> transients, but these changes were never larger than 4% (**Figure 8D**). This suggest that although parasympathetic regulation has a negative inotropic effect on the cardiac function at the whole animal level, in the Langendorff perfused heart, we were unable to observe any substantial effect in contractility. One difference between the intact animal and the isolated perfused heart relies on the fact that, in the perfused heart, we do not have an intrinsic systemic sympathetic drive. Indeed, in the intact animal, the binding of acetylcholine to muscarinic receptors will activate a G<sup>i</sup> protein that will exert an inhibitory effect on AC, antagonizing the stimulatory effect of the β-AR coupled to a G<sup>s</sup> protein. However, in the isolated heart, if G<sup>s</sup> is not activating AC, the activation of a G<sup>i</sup> will not produce changes in the intracellular cAMP levels. Finally, this would explain why we did not observe any major negative inotropic action of carbachol in the perfused heart in the absence of a sympathetic drive.

## Cholinergic Stimulation Modifies Endocardial and Epicardial Cardiac Excitability

Recently, it was shown that vagal innervation exists in the ventricle, but it is 80% less innervated than atria (Coote, 2013). Parasympathetic action on the AP's waveform has been mostly evaluated in atrial and nodal myocytes (sinus and atrioventricular). In both cell types, acetylcholine induces an increase in the AP repolarization rate mediated by a K<sup>+</sup> current activated by acetylcholine (IKACh; Engelstein et al., 1994; Corey et al., 1998; Wickman et al., 1999; Choisy et al., 2012). There was a general belief that this mechanism is absent in the ventricle (Coote, 2013). Recently, immunofluorescence studies have shown the presence of G-protein-coupled inward rectifier K<sup>+</sup> proteins in mouse, rat, and human ventricle (Liang et al., 2014). Here, we found that, in a mouse ventricle, carbachol had a significant effect on AP repolarization rate (**Figure 9**). The APD half phase 2 and the APD90 were reduced in both the endocardial and the epicardial layers, with a modest effect on APD30. Furthermore, atropine, an antagonist of M2 muscarinic receptors, was able to prevent the effect of carbachol indicating that this effect was mediated by a muscarinic receptor.

Interestingly, Litovsky and Antzelevitch (1990) reported that acetylcholine had little if any effect in canine ventricular endocardium but a concentration-dependent biphasic effect in epicardium. Specifically, acetylcholine induced an APD50 prolongation at 0.1 and 1 µM and a reduction at 10 µM. The discrepancy with our findings could be explained by the difference in species (canine vs. murine) and tissue preparation (ventricular wedges vs. whole heart). Another difference is that these authors found no significant effects on APD30 in either layer (Litovsky and Antzelevitch, 1990). Even though our APD30 results showed a significant difference only for the endocardium, the change induced by carbachol was of 530 ± 250 µs (**Table 2**). This difference is 4 times smaller than the increase that we observed in APD30 when the heart was perfused with isoproterenol (**Table 1**). The fact that carbachol had a profound effect on the AP morphology but minor effects on contractility and Ca2<sup>+</sup> currents, supports the idea that Ca2+influx, that determine the CICR, does not occur during AP phase 2.

# Cholinergic Stimulation Across the Ventricular Wall Was Mediated via IKACh

In general, there are two main hypotheses to explain the muscarinic actions on AP repolarization rate: (1) inhibitory effect on the Ca2<sup>+</sup> influx through L-type Ca2<sup>+</sup> channels or (2) an agonistic effect on a cationic outward current. In principle, the muscarinic effect on the L-type Ca2<sup>+</sup> is very unlikely because, as we showed, there were no effects on the developed pressure and very minor effects (3.5 ± 0.35% in the endocardium, and 7.5 ± 0.58% in epicardium) on the amplitude of the Ca2<sup>+</sup> transients in the presence of carbachol. Even more, this first hypothesis was discarded when epicardial Ca2<sup>+</sup> currents were not affected by carbachol (**Figure 10**). The second scenario where the increase in the rate of AP repolarization is mediated by the activation of an outward K<sup>+</sup> current was tested in **Figure 11**. When the hearts were perfused simultaneously with carbachol and the IKACh blocker (tertiapin) a small shortening of the total duration of the AP was observed. Moreover, this apparent blocking effect was corroborated when the same hearts were perfused with carbachol after tertiapin was washed out (**Figure 11**). Together these results suggest that, in the mouse ventricle, cholinergic-mediated changes in the AP morphology are driven by IKACh.

# CONCLUSION

In conclusion, our results indicate that: (1) in both endocardium and epicardium, the increase in contractility by isoproterenol was driven by an increase in the amplitude of intracellular Ca2<sup>+</sup> currents triggered during the AP phase 1; (2) PKA phosphorylates L-type Ca2<sup>+</sup> channels before the SR Ca2<sup>+</sup> release activates NCX current; (3) cholinergic stimulation by carbachol decreased the duration of the late AP repolarization. However, this cholinergic stimulation did not substantially modify per se the intracellular Ca2<sup>+</sup> signals when compared with β-adrenergic stimulation; (4) cholinergic stimulation decreased the total duration of the ventricular AP through activation of IKACh.

In summary, the results presented here demonstrate that, in a mouse heart, β-adrenergic input acts across the ventricular wall by modulating the L-type Ca2<sup>+</sup> currents that occur in phase 1 but not phase 2 of the APs. In contrast, a cholinergic input does not directly modulate Ca2<sup>+</sup> dynamics, but rather alters inward rectifier potassium channels in the isolated heart.

#### ETHICS STATEMENT

Mice were maintained in accordance to the National Institutes of Health Guide for the Care and Use of Laboratory Animals (NIH Publication No. 85–23, Revised 1996) and the Institutional Animal Care and Use Committee guidelines of the University of California Merced (Protocol # 2008–201).

## AUTHOR CONTRIBUTIONS

AE and JR-F designed the research. YA-S, AR, MA, AE, and JR-F performed the research. YA-S, AR, AE, and JR-F analyzed the data. YA-S, AE, and JR-F wrote the manuscript.

#### FUNDING

This work was supported by grants from the NIH (R01 HL-084487 to AE and R01 GM-111397 to JR-F).

#### REFERENCES


#### ACKNOWLEDGMENTS

We thank Alicia Mattiazzi, Fabiana Scornik, and Michael Fill for their valuable feedback on the manuscript.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphys. 2019.00773/full#supplementary-material



Krebs, E. G. (1972). Protein kinases. Curr. Top. Cell. Regul. 5, 99–133.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Aguilar-Sanchez, Rodriguez de Yurre, Argenziano, Escobar and Ramos-Franco. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Unbalance Between Sarcoplasmic Reticulum Ca2<sup>+</sup> Uptake and Release: A First Step Toward Ca2<sup>+</sup> Triggered Arrhythmias and Cardiac Damage

Marilén Federico<sup>1</sup>† , Carlos A. Valverde<sup>1</sup>† , Alicia Mattiazzi<sup>1</sup> and Julieta Palomeque1,2 \*

<sup>1</sup> Centro de Investigaciones Cardiovasculares "Dr. Horacio E. Cingolani", CCT-La Plata/CONICET, Facultad de Cs. Médicas, Universidad Nacional de La Plata, La Plata, Argentina, <sup>2</sup> Centro de Altos Estudios en Ciencias Humanas y de la Salud, Universidad Abierta Interamericana, Buenos Aires, Argentina

#### Edited by:

Maria Fernandez-Velasco, La Paz University Hospital, Spain

#### Reviewed by:

Sandor Gyorke, The Ohio State University, United States Agustín Guerrero-Hernández, Center for Research and Advanced Studies (CINVESTAV), Mexico

#### \*Correspondence:

Julieta Palomeque jpalomeque@ciclaplata.org.ar; pa.lomeque@hotmail.com

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Membrane Physiology and Membrane Biophysics, a section of the journal Frontiers in Physiology

Received: 28 August 2019 Accepted: 24 December 2019 Published: 23 January 2020

#### Citation:

Federico M, Valverde CA, Mattiazzi A and Palomeque J (2020) Unbalance Between Sarcoplasmic Reticulum Ca2<sup>+</sup> Uptake and Release: A First Step Toward Ca2<sup>+</sup> Triggered Arrhythmias and Cardiac Damage. Front. Physiol. 10:1630. doi: 10.3389/fphys.2019.01630 The present review focusses on the regulation and interplay of cardiac SR Ca2<sup>+</sup> handling proteins involved in SR Ca2<sup>+</sup> uptake and release, i.e., SERCa2/PLN and RyR2. Both RyR2 and SERCA2a/PLN are highly regulated by post-translational modifications and/or different partners' proteins. These control mechanisms guarantee a precise equilibrium between SR Ca2<sup>+</sup> reuptake and release. The review then discusses how disruption of this balance alters SR Ca2<sup>+</sup> handling and may constitute a first step toward cardiac damage and malignant arrhythmias. In the last part of the review, this concept is exemplified in different cardiac diseases, like prediabetic and diabetic cardiomyopathy, digitalis intoxication and ischemia-reperfusion injury.

Keywords: sarcoplasmic reticulum Ca2+-ATPase, ryanodine receptor 2, arrhythimas, phospholamban, apoptosis, diabetic myocardiopathy, isquemia-reperfusión

# INTRODUCTION

Cardiovascular diseases are the leading cause of morbidity and mortality worldwide, being Ca2<sup>+</sup> mishandling one of the most striking abnormalities in the setting of a wide spectrum of pathologies, including cardiac hypertrophy, heart failure, DCM, and I/R damage. Indeed, alterations that constitute the hallmark of these diseases, like contractile dysfunction, cardiac arrhythmias or cell death are in great part a reflection of the impairment in Ca2<sup>+</sup> handling and the altered function of the SR, a pivotal responsible of Ca2<sup>+</sup> cycling within cardiac myocytes.

The purpose of this review is: (1) To summarize the regulation of the main cardiac SR Ca2<sup>+</sup> handling proteins involved in SR Ca2<sup>+</sup> uptake and release, i.e., SERCA2a/PLN and RyR2. Both proteins are highly regulated by additional partners' proteins and/or PTMs that may increase or decrease their activity. (2) To describe how the disruption of the interplay among these proteins

**Abbreviations:** β-ARS, β-adrenergic stimulation; AGEs, advanced glycation end products; AP, action potential; Ca2+, calcium; CaM, calmodulin; CaMKII, Ca2+-calmodulin dependent Kinase II; CASQ2, calsequestrin; CPA, cyclopiazonic acid; DAD, delay afterdepolarizations; DCM, diabetic cardiomyopathy; EAD, early afterdepolarizations; ECC, excitation contraction coupling; ER, endoplasmic reticulum; HRC, histidine-rich Ca2+binding protein; I/R, ischemia/reperfusion; MCU, mitochondrial Ca2<sup>+</sup> uniporter; Na+, sodium; NCX, Na+/Ca2<sup>+</sup> exchanger; NO, nitric oxide; ONOO−, peroxynitrite; P0, open probability; PKA, protein kinase A; PLN, phospholamban; PP1, type 1 phosphatase; PTM, post-translational modification; RNS, reactive nitrogen species; ROS, reactive oxygen species; RyR2, ryanodine receptor 2; SERCA2a, sarcoplasmic reticulum Ca2+-ATPase; SR, sarcoplasmic reticulum; SUMO1, small ubiquitin-like modifier type 1; T2DM, type 2 diabetes mellitus.

may constitute a key determinant of Ca2<sup>+</sup> triggered arrhythmias and cardiac damage. (3) To recapitulate experimental evidence in which alterations of the balance between these key players contribute to SR Ca2<sup>+</sup> mishandling with the ensuing production of Ca2<sup>+</sup> triggered arrhythmias and cell death in different cardiac diseases.

#### EXCITATION-CONTRACTION COUPLING

Cardiomyocyte excitation-contraction coupling (ECC) is the process that links membrane depolarization, at the surface cell level, with myofilament interaction that drives contraction, inside the cell. Ca2<sup>+</sup> ions are the link between the two processes (**Figure 1**).

# REGULATION OF CA2<sup>+</sup> HANDLING PROTEINS INVOLVED IN SR CA2<sup>+</sup> UPTAKE AND RELEASE

The SR orchestrates the ECC, being Ca2<sup>+</sup> ions the main players. Ryanodine Receptors 2 (RyR2) and SERCA2a are highly and precisely regulated by several proteins and kinases, which allow a fine-tuned synchronization of Ca2<sup>+</sup> cycling and therefore of cardiac contraction and relaxation processes. We will mention here several key modulators and PTMs of these two main proteins. For a further review see for instance (Kranias and Hajjar, 2012; Haghighi et al., 2014; Mattiazzi and Kranias, 2014; Eisner et al., 2017; Meissner, 2017; Santulli et al., 2017a,b).

#### Regulation of SR Ca2<sup>+</sup> Uptake SERCa2a Post-translational Modifications

Redox regulations appear to play an important role in SERCA2a function, both in health and disease. In cardiac myocytes it has been shown that oxidative stress reduces contractility with depletion of SR Ca2<sup>+</sup> stores, due to SERCA2a inhibition (Morris and Sulakhe, 1997; Xu et al., 1997; Kaplan et al., 2003; Kuster et al., 2010). Indeed, several pathologies like metabolic syndrome (Balderas-Villalobos et al., 2013) or atherosclerosis (Cohen and Adachi, 2006), which are associated with an increase in oxidative stress, reduced SERCA2a activity and contractility.

Nitric oxide activates SERCA2a activity by a cGMPindependent pathway which involves the direct modification of reactive thiol groups on the protein, not only of vascular smooth muscle but also of cardiac and skeletal muscle. On its own, NO is a weak SH-group oxidant. However, in the presence of O<sup>2</sup> <sup>−</sup> it results in an increased ONOO<sup>−</sup> production. Under physiological conditions, the ONOO<sup>−</sup> produced may react with the thiol groups of proteins producing S-nitrosylation and S-gluthationylation of cysteine residues. S-gluthationylation increases the activity of the pump. The SERCA2a residue mainly involved in this reaction is the reactive thiol group Cys<sup>674</sup> (Adachi et al., 2004). However, an exacerbated increase of ONOO<sup>−</sup> nitrosylates the hydroxyl groups of SERCA2a, producing impairment of cardiac relaxation (Braun et al., 2019).

Sarcoplasmic reticulum Ca2+-ATPase is also regulated by the SUMO1 (sumoylation), by AGEs (glycation) and by acetylation/deacetylation processes. SERCA2a sumoylation appears to prolong the lifetime of SERCA2a as well as to increase its intrinsic activity by SUMO1 binding to Lys480 and Lys585 residues. Indeed, increasing SUMO1 expression restores SERCA2a levels, improves hemodynamic performance, and reduces mortality in heart failure (Kho et al., 2011). AGEs complexes can compromise the pump activity by altering the structural movements required for translocating Ca2<sup>+</sup> from the cytosol to the lumen of the SR (Bidasee et al., 2004). Finally, recent experiments indicated that the acetylation of SERCA2a at K492 site was significantly increased in heart failure (HF) in association with a reduction of SIRT1, a class III histone deacetylase. Acetylation of K492 significantly diminished SERCA2a activity, possibly by interfering with the binding of ATP. Activation of SIRT1 restored SERCA2a activity. This strategy may, therefore, be useful for HF treatment (Gorski et al., 2019).

#### PLN Post-translational Modifications and the PLN Interactome

Undoubtedly, the main regulator of SERCA2a activity is PLN (Tada et al., 1975). PLN is a small protein (52 amino acid residues) that binds to and allosterically inhibits SERCA2a (MacLennan and Kranias, 2003). Dephosphorylated PLN reduces the affinity of SERCA2a for Ca2<sup>+</sup> whereas PLN phosphorylation increases SERCA2a pump activity. There are two PLN phosphorylation sites that are physiologically relevant: Ser<sup>16</sup> residue, phosphorylated by PKA and Thr<sup>17</sup> site, phosphorylated by the Ca2+-calmodulin-dependent protein kinase II (CaMKII) (**Figure 2A**). Phosphorylation of these sites increases the affinity of SERCA2a for Ca2<sup>+</sup> and the rate of SR Ca2<sup>+</sup> uptake. This, in turn, leads to increases in SR Ca2<sup>+</sup> load, SR Ca2<sup>+</sup> release and myocardial contractility (Lindemann et al., 1983; Lindemann and Watanabe, 1985; Mundina de Weilenmann et al., 1987; Mundina-Weilenmann et al., 1996).

The status of PLN phosphorylation, as is the case of any other protein, depends on the dynamic balance between the activity of kinases and phosphatases that phosphorylate and dephosphorylate the protein, respectively. PP1, is the major SR phosphatase that specifically dephosphorylates PLN (Steenaart et al., 1992). Inhibition of PP1 results in increased phosphorylation of PLN and SERCA2a activation (Haghighi et al., 2015; **Figure 2A**). During β-ARS, PKA phosphorylates PLN at Ser<sup>16</sup> site and simultaneously inhibits PP1 through the PKA dependent phosphorylation of two additional proteins, inhibitor-1 (I-1) and the small heat shock protein 20 (Hsp20) (Qian et al., 2011). Under β-ARS, PLN-Thr<sup>17</sup> site is also phosphorylated by CaMKII activation due to the increase in intracellular Ca2<sup>+</sup> and the inhibition of PP1, produced by the activity of PKA. In contrast, phosphorylation of Thr<sup>17</sup> of PLN does not occur when only intracellular Ca2<sup>+</sup> was increased without PKA activation, which is necessary to inhibit PP1 (Mundina-Weilenmann et al., 1996).

Post-translational modification of phospholamban by reactive oxygen and nitrogen species (ROS and RNS, respectively) may

also influence SR Ca2<sup>+</sup> uptake (Bigelow and Squier, 2005; Froehlich et al., 2008; Lancel et al., 2009; Ha et al., 2011). Among these PTM it is interesting to mention the one-electron reduction product of NO, nitroxyl (HNO). This molecule has received special attention not only as a possible signaling molecule in the cardiovascular system but also as a potential therapeutic strategy for HF treatment due to its positive inotropic and lusitropic effects in normal and failing canine hearts (Paolocci et al., 2001, 2003; Sivakumaran et al., 2013). PLN played a central role in these effects of HNO, by enhancing SERCA2a activity (Froehlich et al., 2008). Moreover, it has been suggested that S-nitrosylation of PLN at Cys<sup>36</sup> and Cys<sup>41</sup> modulates the PLN-dependent regulation of SERCA2a during β-ARS, i.e., S-nitrosylation of PLN is required for stabilization of the pentameric form of PLN, and consequent SERCA2a activation (Irie et al., 2015). Several additional regulatory proteins are associated with PLN and SERCA2a and contribute to the control of SR Ca2+ transport. These include the hematopoietic lineage cell-specific protein-1 (HS-1) associated protein X-1 (HAX1), a ∼35 kDa protein, which was identified forming a complex with HS-1 in lymphocytes (Suzuki et al., 1997), the intra-luminal histidine-rich Ca2<sup>+</sup> binding protein (HRC), which has been shown to interact with both SERCA2a and triadin on the SR luminal side (see below and **Figure 2C**) and S100A1 on the cytosolic side (Kiewitz et al., 2003). Ca2+-dependent S100A1 binding to SERCA2a results in an increased enzymatic activity which is associated with enhanced SR Ca2<sup>+</sup> uptake and load (Most et al., 2001; Kiewitz et al., 2003; Kettlewell et al., 2005). As will be discussed below, HRC and S100A1 also interact and regulate RyR2 and SR Ca2<sup>+</sup> release (Völkers et al., 2007; for review, see Haghighi et al., 2014; Kranias and Hajjar, 2017; Arvanitis et al., 2018).

# Regulation of SR Ca2<sup>+</sup> Release

Ryanodine receptor 2 is the largest ion channel known in nature and one of the most relevant Ca2<sup>+</sup> handling proteins. RyR forms a homotetrameric assembly comprising four monomers of 565 kDa each (Van Petegem, 2015). There are three mammalian isoforms that share 65% sequence identity: RyR1, predominantly expressed in skeletal muscle; RyR2, the cardiac isoform and RyR3, expressed in several tissues including the brain (Capes et al., 2011). While at the cytosolic portion, the channel contains multiple regulatory domains, such as binding sites for energy sensors (ATP, ADP, and AMP) (**Figure 2B**), and inorganic phosphate, metabolites such as pyruvate, fatty acids and polyamines, and ions (Mg2+, H+, and Cl−, not shown in the Figure for the sake of clarity) (Zucchi and Ronca-Testoni, 1997; Fill and Copello, 2002; Meissner, 2004); the Ca2<sup>+</sup> binding site is located in the core domain of the channel just above the transmembrane domain an involves de carboxylterminal domain region (Murayama et al., 2018). This complex is also regulated and modulated by a diverse array of RyR2 interacting proteins which involve PKA, CaMKII, phosphatases (i.e., phosphatase 1 and 2A), and phosphodiesterase (PDE4D) which are tethered to the channel and held near their target sites by means of anchoring proteins (Marks, 2002; Lehnart et al., 2005). This allows for a tight and spatially confined homeostatic regulation of the balance between RyR2 phosphorylation and phosphatase dependent dephosphorylation.

#### Post-translational Modifications of RyR2

Phosphorylation is possibly the most studied and controversial PTM modification of RyR2. Phosphorylation of the channel modulates the effect of Ca2<sup>+</sup> on the RyR2 without having the inherent ability to open or close the channel per se (Camors and Valdivia, 2014). Until now three phosphorylation sites in the RyR2 have been identified: Ser2808, Ser2814, and Ser<sup>2030</sup> (**Figure 2B**). Serine 2808 (Ser2808, mouse, and Ser2809 in human and canine RyR2 nomenclature) was first described by Witcher and collaborators as a CaMKII site (Witcher et al., 1991). Further in-depth studies of this phospho-site indicated that Ser<sup>2808</sup> is a target for PKA, CaMKII and possibly PKG (Jiang et al., 2002; Rodriguez et al., 2003; Stange et al., 2003; Currie et al., 2004; Ai et al., 2005; Xiao et al., 2005; Carter et al., 2006; Kohlhaas et al., 2006; Ferrero et al., 2007; Huke and Bers, 2008; MacDonnell et al., 2008; Fischer et al., 2013). Experiments by Marx et al. (2000) indicated that PKA-dependent phosphorylation of RyR2 at Ser<sup>2808</sup> site under β-ARS, increases P<sup>0</sup> and SR Ca2<sup>+</sup> release. However, this contention was not supported by different studies and the functional meaning of this phosphorylation is not clear yet (Xiao et al., 2006; Ferrero et al., 2007; Huke and Bers, 2008). This is in part due to the fact that most studies found that Ser<sup>2808</sup> is constitutively phosphorylated under basal conditions (Jiang et al., 2002; Rodriguez et al., 2003; Carter et al., 2006; Ferrero et al., 2007; Huke and Bers, 2008), generating doubts about the relevance of "extra" phosphorylation on this site. Moreover, it

phosphates (AMP, ADP, and ATP) are also bounded to RyR2. (C) Dual role of HRC: The intra-luminal protein histidine-rich calcium binding protein (HRC) interacts with SERCA2a as well as with triadin in a Ca2+-dependent fashion, increasing RyR2 Ca2<sup>+</sup> release and SERCA2a Ca2+-uptake when Ca2<sup>+</sup> increases in SR.

was also showed that RyR2s were hyperphosphorylated in failing hearts from humans and dogs at Ser2809, which was attributed at least in part to a decrease in the amount of PP1 associated to RyR2 (Marx et al., 2000; Wehrens et al., 2006). However, several subsequent experiments by other groups failed to reproduce these findings (Li et al., 2002), turning the attention to serine 2814 (Ser2814) site as the primary phosphorylation site responsible for SR Ca2<sup>+</sup> leak and arrhythmogenic events in HF (Li et al., 2002; Respress et al., 2012). The role of Ser<sup>2808</sup> phosphorylation was further complicated by the finding that both, minimum and maximum RyR2 phosphorylation at Ser2808, increase RyR2 activity, suggesting a U-shaped of RyR2 activity according to the PKA phosphorylation level (Carter et al., 2006). A clear revision of these controversial results is given by Bers (Bers, 2012) and Camors and Valdivia (Camors and Valdivia, 2014).

Ser<sup>2814</sup> site was described by Wehrens et al. (2004) as a CaMKII site and further evidence confirmed that this site seems to be exclusively phosphorylated by CaMKII. In single-channel experiments, the P<sup>0</sup> of the RyR2s was generally found to be increased upon phosphorylation by CaMKII (Lokuta et al., 1995; Wehrens et al., 2004; Yang et al., 2007). In line with these results, either activation or overexpression of CaMKII was associated with the positive inotropic effect of β-ARS (Ferrero et al., 2007), an increase of Ca2<sup>+</sup> spark frequency (Guo et al., 2006) and the susceptibility to arrhythmias (Maier et al., 2003; Dybkova et al., 2011; Respress et al., 2012; Mazzocchi et al., 2016; Valverde et al., 2019). In contrast, animals in which Ser<sup>2814</sup> was replaced by Alanine (S2814A mice) were protected from arrhythmias and cardiac dysfunction induced by several diseases (van Oort et al., 2010; Di Carlo et al., 2014; Mazzocchi et al., 2016).

Serine 2030 (Ser2030) was characterized as a PKA phosphorylation site using classical phospho-epitope mapping (Xiao et al., 2005). Whereas in quiescent cardiac myocytes the RyR2 appears to be completely unphosphorylated (Huke and Bers, 2008), this site has been suggested as the major phosphorylation site in RyR2 responding to PKA activation upon β-ARS in normal and failing hearts (Xiao et al., 2006). In this context, it has recently been described that phosphorylation of RyR2 at Ser<sup>2030</sup> is required for a complete effect of β-ARS (Potenza et al., 2019) in mouse lines with genetic ablation of this site (RyR2-S2030A).

Interestingly, recent work reports crystal structures of the RyR2 phosphorylation domain with the PKA catalytic subunit (PKAc), showing Ser<sup>2808</sup> captured within the active site of PKA. The results further demonstrated that the addition of a phosphomimetic at the CaMKII site (S2814D), results in structural changes in the RyR2 phosphorylation domain that enhance the interaction with PKAc. These findings strongly suggest that phosphorylation of Ser<sup>2814</sup> site may affect the activity of PKA and impact on Ser2808, i.e., nearby phosphorylation sites might influence one each other (Haji-Ghassemi et al., 2019). This possible interaction among the different residues sharing the phosphorylation "hotspot" region of RyR2, might clarify previous controversial findings on the role of Ser<sup>2808</sup> site on different physiological and disease situations. Since RyR2 phosphorylation by PKA and CaMKII may not be independent, the authors suggest that the phosphorylation status of Ser<sup>2808</sup> may

be altered in studies that have used S2814D mice. Of note, in contrast with this prediction, previous experiments indicate that isoproterenol-induced phosphorylation of RyR2-Ser<sup>2808</sup> site did not vary when isoproterenol was administrated in the absence and presence of a CaMKII inhibitor (KN-93), to avoid the simultaneous phosphorylation of Ser<sup>2814</sup> residue (Ferrero et al., 2007), i.e., phosphorylation of Ser<sup>2814</sup> did not influence the extent of phosphorylation of Ser<sup>2808</sup> site. However, the isoproterenol concentration used in these experiments was rather high and may not allow any further PKA-dependent phosphorylation of this site. Therefore, it would be important to perform similar experiments in the presence of lower isoproterenol concentrations to investigate the possible influence of Ser<sup>2814</sup> phosphorylation on the isoproterenol-induced phosphorylation of Ser<sup>2808</sup> site predicted by the crystal structure studies.

The role of phosphatases activity on RyR2 phosphorylation was recently emphasized. It was shown that PP1 activation counteracts the increased kinase activity in human heart failure reducing SR Ca2<sup>+</sup> leak as well as cellular arrhythmias without significant changes in SR Ca2<sup>+</sup> load and contractility (Fischer et al., 2018).

Oxidative conditions generally increase the RyR2 P0, while reducing agents do the opposite (Marengo et al., 1998; Xu et al., 1998; Salama et al., 2000; Sun et al., 2008). The functional consequence of a moderate cellular oxidative/nitrosative stress could result in an immediate enhancement of Ca2<sup>+</sup> release from the SR in response to a given physiological trigger. However, severe oxidative stress can cause irreversible and persistent activation of RyR2s (Xu et al., 1998), increasing SR Ca2<sup>+</sup> leak. It has been reported that NADPH oxidase 2 (NOX2) is the predominant isoform expressed in T-tubules and SR membranes of adult cardiomyocytes. Therefore, it is strategically positioned to modulate the activity of the RyR2s. ROS produced by NOX2 stimulates SR Ca2<sup>+</sup> release via at least two pathways: direct oxidation or S-glutathionylation of RyR2s or indirectly through CaMKII activation (Palomeque et al., 2009), followed by phosphorylation of the RyR2s. In healthy cardiac muscle neuronal nitric oxide synthase (nNOS) is mainly located in the SR membrane, linked to the RyR2s, which would favor direct RyR2 nitrosation. The role of this PTM of RyR2 on cardiac ECC has been previously reviewed (Lim et al., 2008; Gonzalez et al., 2009).

#### The RyR2 Complex

Ryanodine receptor 2 modulation has been shown to involve also several key proteins (**Figure 2B**), most importantly CASQ2. CASQ2 not only acts as a Ca2<sup>+</sup> buffer, but it also mediates the responsiveness of the RyR2 channel to luminal Ca2<sup>+</sup> by serving as a Ca2<sup>+</sup> sensor (Gyorke et al., 2004; Gyorke and Terentyev, 2008). This function is effective through protein-protein interactions, with junctin and triadin (Zhang et al., 1997; Shin et al., 2000). However, a direct interaction between CASQ2 and RyR2 has recently been described (Handhle et al., 2016). Junctin is a 26 kDa transmembrane protein forming a complex with triadin, CASQ2, and RyR2. It has been proposed that junctin is in direct contact with RyR2 and works as an anchor for CASQ2 (Zhang et al., 1997; Gyorke et al., 2004).

Calstabin 2 or FKBP12.6, is a peptidyl-prolyl cis/transisomerase of 12.6 kDa, associated with RyR2 with a stoichiometry of 4:1 (**Figure 2B**). The role of FKBP12.6 has been a matter of controversy for over the past two decades (Kaftan et al., 1996; Timerman et al., 1996; Barg et al., 1997; Marks, 2000; Doi et al., 2002; Jiang et al., 2002; Stange et al., 2003; Wehrens et al., 2003; Xiao et al., 2007; Guo et al., 2010). FKBP12.6 has been considered as RyR2 "stabilizer," since some results indicate that FKBP12.6 dissociation from RyR2 produces RyR2 subconductance states and increase the Po of the channel. However, other piece of evidence indicates that FKBP12.6 failed to show any effect of FKBP12.6 on RyR2 gating. The recent identification of FKB12 binding site on RyR2 may help to understand the controversial matters encompassing the FKBP-RyR2 interactions. FKBP12 binds to RyR2 with lower affinity than FKBP12.6, but has a higher cardiac expression level (Jeyakumar et al., 2001). Recent results revealed that only 20% of RyR2 proteins are associated with FKBP12.6 in myocytes (Guo et al., 2010) which could explain why RyR2 is unaffected in FKBP12.6-KO mice (Xiao et al., 2007). Acute overexpression of FKBP12 in adult rabbit ventricular myocytes showed a reduction in the gain of ECC and a decrease in Ca2<sup>+</sup> spark frequency, suggesting that FKBP12 reduces RyR2 sensitivity to cytosolic Ca2+(Seidler et al., 2007). In contrast, more recent results show that FKBP12 activates the RyR2 and competes with FKBP12.6. The last study, would suggest that rather than a direct stabilization of the channel, an increase in FKBP12.6-RyR2 binding competes with FKBP12 at the same binding site and blunts the activation of RyR2 promoted by FKBP12 (Galfre et al., 2012).

Apo-Calmodulin or Ca2+-free CaM has an inhibitory effect on RyR2 channel. The Ca2+-bound CaM is named Ca2+-CaM. Although Ca2+-CaM is the usual form that binds to target proteins, CaM can also bind to RyR2. CaM shifts the Ca2+ dependence of RyR2 activation to higher Ca2<sup>+</sup> concentrations (Fruen et al., 2000; Balshaw et al., 2001; Yamaguchi et al., 2005). The role of CaM on RyR2 regulation was highlighted by results that indicate that mutations in CaM are associated with RyR2 mediated cardiac arrhythmias (Nomikos et al., 2014; Sondergaard et al., 2017). High resolution cryo-electron microscopy recently provided new insights into the modulation of RyR2 channel gating by CaM (Gong et al., 2019). These data indicate that Ca2+- CaM changes RyR2 conformation differently under different situations. Whereas Ca2+-CaM can reverse RyR2 opening by Ca2<sup>+</sup> and PCB95, a potent channel opener (Samso et al., 2009), it cannot counteract the activation of the channel by a mixture of Ca2+, ATP and caffeine. These results emphasize that the P<sup>0</sup> of RyR2 is critically determined by a strict balance between different activators and inhibitors of the channel (Van Petegem, 2019).

Several proteins that interact with SERCA2a regulating SR Ca2<sup>+</sup> uptake, also modify the RyR2 function. As already mentioned, the HRC protein is not only associated with SERCA2a but also with triadin. The interaction of HRC with triadin increases with increasing Ca2<sup>+</sup> concentration (Sacchetto et al., 1999; Arvanitis et al., 2007). This interaction is believed to modulate RyR2 function and SR Ca2<sup>+</sup> release by conferring refractoriness to SR Ca2<sup>+</sup> release. In turn, HRC-SERCA2a interaction is also Ca2+-dependent. In this case, the maximal HRC-SERCA2a association occurs at low Ca2<sup>+</sup> concentration and diminishes with increasing Ca2<sup>+</sup> concentrations (Arvanitis et al., 2007). The different Ca2<sup>+</sup> dependence of the interaction HRC-SERCA2a and HRC-triadin determines the HRC effects on SR Ca2<sup>+</sup> handling (**Figure 2C**): At low SR Ca2+, HRC interacts with SERCA2a inhibiting SR Ca2<sup>+</sup> uptake. When SR Ca2<sup>+</sup> concentration increases, HRC dissociates from SERCA2a and enhances its binding to triadin, regulating SR Ca2<sup>+</sup> release (for review, see Arvanitis et al., 2018).

Sorcin is a 22 kDa penta-EF hand Ca2+-binding protein expressed in many tissues, including the heart. Single-channel studies indicated that when applied to the cytoplasmic region of RyR2s, sorcin inhibits RyR2 activity in a dose-dependent manner by prolonging the mean close time without modifying singlechannel conductance, an effect that is abrogated when sorcin is phosphorylated by PKA. More recent experiments (Farrell et al., 2003) demonstrated that sorcin significantly inhibits both the spontaneous activity of RyR2s in quiescent cells and the Ca2<sup>+</sup> current (ICa)-triggered activity of RyR2s. Moreover, it decreased the amplitude of the Ca2<sup>+</sup> transient without affecting the amplitude or kinetics of ICa, reducing the "gain" of ECC mechanism. Sorcin seems to be a key RyR2-associated protein under stress conditions since its ablation displayed a significantly higher incidence of cardiac arrhythmias and sudden death in sorcin-KO mice when subjected to acute or chronic stress challenge (Chen et al., 2018). It has also been shown that sorcin increases SR Ca2<sup>+</sup> uptake (Matsumoto et al., 2005) and interacts with NCX (Zamparelli et al., 2010) and L-type Ca2<sup>+</sup> channels (LTCC) (Fowler et al., 2008). All these interactions point to an important role of this protein in ECC regulation.

Also, as in the case of SERCA2a, S100A1 modulates RyR2 function under both diastolic and systolic conditions (Kiewitz et al., 2003; Most et al., 2004; Völkers et al., 2007, Volkers et al., 2010). Most et al. (2004), first demonstrated that addition of S100A1 to isolated SR vesicles resulted in diminished <sup>3</sup>H-ryanodine ([3H]Ry) binding to RyR2 at free Ca2<sup>+</sup> concentrations of about 150 nM, while a significantly increased [3H]Ry binding occurred at Ca2<sup>+</sup> concentrations greater than 300 nM. Hypothesizing a reduced RyR2 P<sup>0</sup> at diastolic cytoplasmic Ca2<sup>+</sup> levels, S100A1 would reduce SR Ca2<sup>+</sup> leak in quiescent cardiomyocytes (Völkers et al., 2007). Moreover, S100A1 increases fractional SR Ca2<sup>+</sup> release in voltage-clamped rabbit cardiomyocytes, suggesting that S100A1 enhances the ECC gain under systolic conditions (Kettlewell et al., 2005).

#### Cytosolic and Luminal Ca2<sup>+</sup> Regulation of RyR2

Both cytosolic and luminal Ca2<sup>+</sup> regulate RyR2. It has long been known that the release of SR Ca2<sup>+</sup> in cardiac muscle during ECC is graded by the amount of activating Ca2<sup>+</sup> outside the SR by the Ca2+-induced Ca2<sup>+</sup> release (CICR) process (Fabiato and Fabiato, 1977). Experimental evidence suggested the presence of high and low affinity Ca2<sup>+</sup> binding sites in the cytosolic region of RyR2 and luminal Ca2+-binding sites, whose luminal occupancy depends on SR Ca2<sup>+</sup> load (Fabiato and Fabiato, 1979; Shannon et al., 2000). RyR2s are normally closed at low cytosolic Ca2<sup>+</sup> (100–200 nM); channel activity is maximal at 10–100 µM cytosolic Ca2+, while elevating cytosolic

Ca2<sup>+</sup> beyond this point leads to a reduction in P<sup>0</sup> (Xu and Meissner, 1998). This biphasic behavior implies there are at least two classes of Ca2<sup>+</sup> binding sites: high-affinity activation and low-affinity inactivation sites. The P<sup>0</sup> steep dependence of RyR2 on cytoplasmic Ca2+, which typically exhibits Hill coefficients of 2–4 (Sitsapesan and Williams, 1994) indicates that RyR2 activation resulted from cooperative involvement of one highaffinity (∼1 µM) Ca2<sup>+</sup> binding site on each subunit of the homotetrameric channel (Zahradnik et al., 2005). RyR2s also possess two inhibitory sites in their cytoplasmic domains with Ca2<sup>+</sup> affinities of the order of 1 µM and 1 mM. Mg2<sup>+</sup> competes with Ca2<sup>+</sup> at these sites to inhibit RyR2s (for instance, see Laver, 2018).

In 1994, Sitsapesan and Williams were the first to show that luminal Ca2<sup>+</sup> could directly activate RyR2 (Sitsapesan and Williams, 1994). Since then, different type of evidence supports the notion that luminal Ca2<sup>+</sup> controls RyR2 function (i.e., Bassani et al., 1995; Gyorke and Gyorke, 1998; Shannon et al., 2000; Kong et al., 2007; Gyorke and Terentyev, 2008). Two different mechanisms have been proposed to explain luminal Ca2<sup>+</sup> regulation of RyR2: The "feed-through" hypothesis suggests that luminal Ca2<sup>+</sup> acts on its own cytosolic Ca2<sup>+</sup> binding site during or after Ca2<sup>+</sup> passage through an open RyR2 (Herrmann-Frank and Lehmann-Horn, 1996; Xu and Meissner, 1998). A second mechanism proposes that luminal Ca2<sup>+</sup> regulation of RyR2 is mediated by a Ca2<sup>+</sup> sensing mechanism inside the SR (Gyorke and Gyorke, 1998; Ching et al., 2000; Jiang et al., 2007; Gyorke and Terentyev, 2008). Although CASQ2, may serve as a key SR luminal Ca2<sup>+</sup> sensor (Gyorke et al., 2004), experiments in CASQ2-null mice (Knollmann et al., 2006) and in purified native and recombinant RyR2s that lack CASQ2 (Xu and Meissner, 1998; Kong et al., 2007), indicate that RyR2s are also regulated by a luminal Ca2<sup>+</sup> sensing mechanism that does not require CASQ2. Indeed, a point mutation in RyR2 (RyR2-E4872A) which eliminates Ca2<sup>+</sup> regulation by luminal but not by cytosolic Ca2<sup>+</sup> was recently identified (Chen et al., 2014). Structural analysis, at near-atomic resolution, suggests that in addition to E4872, the E4878 residue may also be involved in luminal Ca2<sup>+</sup> activation of RyR2 (Peng et al., 2016), although the precise mechanism by which each of these different sites promotes luminal Ca2<sup>+</sup> activation of RyR2 is not clear yet.

# HOW CAN THE DISRUPTION OF THE NORMAL INTERPLAY AMONG SR CA2<sup>+</sup> HANDLING PROTEINS EVOKE CA2<sup>+</sup> TRIGGERED ARRHYTHMIAS AND APOPTOSIS?

The normal interplay among the different proteins responsible for the release and reuptake of Ca2<sup>+</sup> by the SR is regulated by different mechanisms as reviewed above. This regulation may be altered and evolve toward different types of cardiac disorders, which include arrhythmias and cell death through apoptotic and necrotic processes. Therefore, regulation and/or alteration of SR Ca2<sup>+</sup> handling proteins (for instance by phosphorylation, redox changes or mutations), have received great attention from physiologists and clinicians. We will describe below two main consequences of the unbalance of SR Ca2<sup>+</sup> uptake and release, i.e., Ca2<sup>+</sup> triggered arrhythmias and cellular apoptosis and necroptosis, with a main focus on those produced by PTM of Ca2<sup>+</sup> handling proteins.

# Ca2<sup>+</sup> Triggered Arrhythmias Mechanisms

As stated above, RyR2s are highly regulated molecules. Genetic or PTM of RyR2 are a main cause of Ca2<sup>+</sup> triggered arrhythmias, i.e., arrhythmias that are originated due to abnormal Ca2<sup>+</sup> handling.

Triggered activity describes impulse initiation that is dependent on the so-called afterdepolarizations, which are oscillations in membrane potential that follow the primary depolarization phase (0) of an AP. Afterdepolarizations are divided into early and delayed afterdepolarizations, EAD, and DAD, respectively. EADs are defined as a slowing or reversal of normal repolarization that occurs before completion of AP, usually in phases 2 and 3 of human AP, whereas DADs occur after AP completion (**Figure 3**). These mechanisms may produce sustained arrhythmias by reentry circuits (Anderson, 2007). EADs occurs usually in the setting of prolonged repolarization and are classically attributed to reactivation of ICa (January and Riddle, 1989; Nuss et al., 1999). However, a second major current that facilitates EADs formation is NCX. Indeed, experimental evidence indicates that these two currents act synergistically to generate EADs, with their relative contributions varying under specific conditions (Weiss et al., 2010). The late component of the Na<sup>+</sup> current (INa), has been recognized as an important player to set up the conditions for EADs, by producing SR Ca2<sup>+</sup> overload, via the reduction of repolarization reserve and the increase in intracellular Na<sup>+</sup> concentration. In addition, experimental evidence and modeling studies indicate that EADs may directly arise from Na<sup>+</sup> channel reactivation (Horvath et al., 2013; Sato et al., 2017; **Figure 3A**).

Delay afterdepolarizations are caused by spontaneous Ca2<sup>+</sup> releases from the SR (Bers, 2006). Under conditions of SR Ca2<sup>+</sup> overload and/or in circumstances which sensitize the RyR2s, the Ca2<sup>+</sup> released by a group of RyR2 activates neighboring RyR2, in such a way that Ca2<sup>+</sup> propagates in a regenerative way traveling along the myocytes in a saltatory fashion from sarcomere to sarcomere (Cheng et al., 1996). Ca2<sup>+</sup> waves have substantial arrhythmogenic potential, since they may trigger Ca2<sup>+</sup> activated currents, such as the NCX current (INCX). This promotes a transient Na<sup>+</sup> current (Iti), that depolarizes cell membrane and may eventually trigger a spontaneous AP (Bers, 2006), which is referred to as triggered AP, leading to spontaneous contraction (Spencer and Sham, 2003; Fujiwara et al., 2008; Mazzocchi et al., 2016; **Figure 3B**).

When referring to a multicellular tissue, e.g., whole heart, spontaneous Ca2<sup>+</sup> releases synchronized in a small group of cells is not enough for triggering an AP. 3D modeling (Xie et al., 2010) estimated that about 800,000 cells are required to trigger a premature ventricular complex, being able to bring

membrane and producing a delay after depolarization (DAD). If this depolarization attains the excitability threshold, a spontaneous AP occurs.

the sink (adjacent tissue in basal conditions) to its activation threshold. A sort of synchronization mechanism must exist for EADs and DADs to overcome the source-sink mismatch. Normally, in an intact tissue, the source-sink mismatch is the main mechanism protecting the heart against spontaneous Ca2<sup>+</sup> release-induced arrhythmias.

#### The Threshold Concept

In the context of Ca2<sup>+</sup> triggered arrhythmias, an intriguing issue to consider is the SR Ca2<sup>+</sup> threshold. As mentioned above, spontaneous SR Ca2<sup>+</sup> leak can occur in the absence of membrane depolarization. It has long been known that several conditions that increase SR Ca2<sup>+</sup> load increase SR Ca2<sup>+</sup> waves and spontaneous contractions (Orchard et al., 1983; Stern et al., 1983; Wier et al., 1987). Diaz et al. (1997) showed that increasing extracellular Ca2<sup>+</sup> in quiescent cells produced spontaneous Ca2<sup>+</sup> release associated with increased SR Ca2<sup>+</sup> content (Diaz et al., 1997). Once spontaneous Ca2<sup>+</sup> release arose, further increase in extracellular Ca2<sup>+</sup> did not affect SR Ca2<sup>+</sup> content because of the proportional increase in SR Ca2<sup>+</sup> leak. The authors conclude "It appears there is a maximum level of SR Ca2<sup>+</sup> content, perhaps because spontaneous release results when the content reaches a threshold"(Diaz et al., 1997). Due to its dependence on the SR Ca2<sup>+</sup> store, this depolarization– independent SR Ca2<sup>+</sup> release has been called "Store Overload-Induced Ca2<sup>+</sup> Release (SOICR)"(Jiang et al., 2004). It has been further shown that the threshold level also depends on the activity of RyR2. One important premise of this mechanism is that ". . .once a threshold level of SR Ca2<sup>+</sup> content is reached, SOICR occurs"(Jiang et al., 2004). More recent experiments by Belevych et al. (2012) challenged the idea of immediacy that encompasses the last concept. These authors demonstrated that SR Ca2<sup>+</sup> leak occurs with a substantial time delay after the attainment of diastolic SR Ca2<sup>+</sup> level, i.e., the attainment of a certain SR Ca2<sup>+</sup> level is not sufficient for spontaneous Ca2<sup>+</sup> release and waves generation. The time factor is necessary. Interestingly, the postrefilling refractory period was shorter in myocytes from infarcted hearts than in control myocytes, even though the rate of SR Ca2<sup>+</sup> content recovery after the stimulus-induced SR Ca2<sup>+</sup> release was similar. Based on these and other results (Sobie et al., 2005), it was concluded that the probability of spontaneous Ca2<sup>+</sup> triggering also depends on the recovery of RyR2 from refractoriness (time and Ca2<sup>+</sup> store–dependent properties of RyR2). In postinfarction myocytes the post-refilling refractory period was reduced, an effect attributed to CaMKII phosphorylation and redox modifications of RyR2. **Figure 4** schematizes a possible interpretation of the experimental results.

# Ca2+-Induced Apoptosis and Necroptosis

It is generally accepted that mitochondria are at the central stage of cell death (Finkel, 2001; Dorn and Maack, 2013; Pan et al., 2013). Indeed, numerous recent investigations revealed the mitochondria are effectors of programed apoptosis or necrosis and sources of damaging ROS.

Mitochondria are organelle in close association with the SR. This proximity allows a cross-talk between mitochondria and SR which is extremely valuable under normal conditions:

Mitochondrial ATP production is crucial for modulation of oxidative phosphorylation and therefore essential to maintain myocyte activity, including SR Ca2<sup>+</sup> cycling, contraction, and relaxation (Denton, 2009). The physical contact between SR membranes and mitochondria are known as sarco/endoplasmic reticulum (SR/ER)/mitochondria microdomains or mitochondria-associated SR/ER membranes (MAMs). A sufficient local Ca2<sup>+</sup> concentration may be achieved in specialized microdomains created by the close association of mitochondria and the SR/ER (Szabadkai et al., 2003). In these microdomains, cytosolic Ca2<sup>+</sup> is predicted to transiently rise to micromolar concentrations, consequently allowing significant Ca2<sup>+</sup> uptake via the MCU. Physical proximity and functional interplay between mitochondria and SR is maintained in part through tethering of these two organelles by different linkers that may contribute to either decrease or maintain the physical gap between the SR and the mitochondria. For an extensive review see Csordas et al. (2018). On the other hand, perturbation of Ca2<sup>+</sup> handling may alter mitochondrial-SR Ca2<sup>+</sup> crosstalk and excessive Ca2<sup>+</sup> can go to the mitochondria which may contribute to apoptosis and necroptosis in different diseases. Stress conditions that lead to Ca2<sup>+</sup> or ROS overload trigger mPTP opening, i.e., the mitochondrial membrane becomes permeable to any molecule less than 1.5 kDa in size. Consequent dissipation of the membrane potential (19m) leads to mitochondrial membrane depolarization, failure to produce ATP and release of mitochondrial proteins such as cytochrome c, which initiate cell death pathways (Bernardi and Di Lisa, 2015). Importantly, it has been shown that mitochondria-initiated cell death is one main mechanism in HF (Nakayama et al., 2007).

# The Interplay Between SR Ca2<sup>+</sup> Uptake and Leak

As discussed above, the properties of RyR2 are a main factor in determining the magnitude of SR Ca2<sup>+</sup> leak. However, can an increase in RyR2 P0, increase SR Ca2<sup>+</sup> leak by itself? Experiments by Venetucci et al., clearly demonstrated that the potentiation of RyR2 produces only a transient increase in SR Ca2<sup>+</sup> leak, because once SR Ca2<sup>+</sup> leak initiates, SR Ca2<sup>+</sup> load decreases below the threshold for SR Ca2<sup>+</sup> leak (Venetucci et al., 2007). Only a simultaneous enhancement of SR Ca2<sup>+</sup> uptake would be able to maintain the necessary level of SR Ca2<sup>+</sup> content to attain the threshold for SR Ca2<sup>+</sup> leak. This conclusion is in agreement with clinical facts showing that patients with catecholaminergic polymorphic ventricular tachycardia (CPVT) due to RyR2 mutation only suffer arrhythmias after β-ARS. Experiments performed in S2814D myocytes, in which Ser<sup>2814</sup> was mutated to aspartic acid and behaves as pseudo constitutively phosphorylated (van Oort et al., 2010), also showed that an increase in the P<sup>0</sup> of RyR2 produced by CaMKII phosphorylation was not able to evoke an SR Ca2<sup>+</sup> leak higher than the one observed in wild type (WT) myocytes unless they are challenged by increasing extracellular Ca2<sup>+</sup> or β-ARS (Mazzocchi et al., 2016).

Sarcoplasmic reticulum Ca2<sup>+</sup> leak is also critically dependent on SR Ca2<sup>+</sup> load. As stated above, SR Ca2<sup>+</sup> overload triggers spontaneous Ca2<sup>+</sup> release via the activation of the RyR2 luminal Ca2<sup>+</sup> sensor (Gyorke and Terentyev, 2008). Therefore, the potential effect of increasing SR Ca2<sup>+</sup> load on Ca2<sup>+</sup> triggered arrhythmias, seems obvious. However, PLN knock-out mice (PLNKO), which have a fully loaded SR, have not proven to be prone to arrhythmias under basal conditions (Santana et al., 1997). Although at first sight these results might suggest that the increase in SR Ca2<sup>+</sup> load per se does not increase arrhythmia propensity, several studies have provided evidence that the increase in SR Ca2<sup>+</sup> load produced by PLN ablation does produce a dramatic increase SR Ca2<sup>+</sup> leak (Santana et al., 1997; Huser et al., 1998; Mazzocchi et al., 2016). Unexpectedly, it hardly evokes SR Ca2<sup>+</sup> waves (Huser et al., 1998; Mazzocchi et al., 2016). **Figure 5A** shows the results of experiments performed in S2814D mice. As discussed above, this mutation confers the hearts a high propensity to SR Ca2<sup>+</sup> waves and arrhythmias when submitted to stress (van Oort et al., 2010), and in double mutant mice resulting from cross-breeding S2814D mice with PLNKO mice, to increase SR Ca2<sup>+</sup> reuptake (SDKO mice) (Mazzocchi et al., 2016; Valverde et al., 2019). In the presence of high extracellular Ca2+, the frequency of Ca2<sup>+</sup> sparks and SR Ca2<sup>+</sup> leak were higher in SDKO than in S2814D myocytes, consistent with the overall higher SR Ca2<sup>+</sup> load in SDKO cells (Mazzocchi et al., 2016). Similar results were obtained when SR Ca2<sup>+</sup> load was increased in both strains subjected to an I/R protocol (Mazzocchi et al., 2016; Valverde et al., 2019; **Figure 5B**). Unexpectedly, whereas S2814D myocytes displayed full propagating Ca2<sup>+</sup> waves when exposed to high extracellular Ca2<sup>+</sup> (**Figure 6A**) or I/R (**Figure 6B**), SDKO myocytes mostly show non-propagating Ca2<sup>+</sup> events, known as mini-waves. Thus, in spite of the higher SR Ca2<sup>+</sup> leak observed in SDKO myocytes vs. S2814D, the proportion of fully propagating events in SDKO myocytes was significantly less (Mazzocchi et al., 2016). This seeming paradox can be clarified by the acknowledgment that two main factors intervene in the production of arrhythmogenic Ca2<sup>+</sup> waves: 1. Increased SR Ca2<sup>+</sup> leak and 2. Cytosolic Ca2<sup>+</sup> wave propagation.

An increase in SR Ca2<sup>+</sup> leak associated with a decrease in propagating SR Ca2<sup>+</sup> waves indicates a limitation in cytosolic Ca2<sup>+</sup> diffusion (**Figure 7**). PLN ablation interrupts cell-wide propagating Ca2<sup>+</sup> waves, converting them into non-propagated events, like mini-waves or groups of Ca2<sup>+</sup> sparks (Huser et al., 1998; Mazzocchi et al., 2016; Valverde et al., 2019), supporting the contention that by decreasing cytosolic Ca2+, PLN ablation would increase cytosolic Ca2<sup>+</sup> buffer capacity, hampering Ca2<sup>+</sup> wave propagation and preventing the arrhythmogenic susceptibility produced by an enhanced SR Ca2<sup>+</sup> load. Further

support to this idea is given by the experiments in which decreasing SR Ca2<sup>+</sup> uptake by the SERCA2a inhibitor CPA, converts non-propagating mini-waves into full propagating Ca2<sup>+</sup> waves (Valverde et al., 2019). These results indicate therefore that an increase in SR Ca2<sup>+</sup> content does increase the propensity to arrhythmias. However, the mechanism by which the increase in SR Ca2<sup>+</sup> occurred may conspire against the arrhythmogenic effect of the high SR Ca2<sup>+</sup> load, i.e., an enhancement of SR Ca2<sup>+</sup> sequestration, if high enough, would increase SR Ca2<sup>+</sup> load and leak but also preclude Ca2<sup>+</sup> wave propagation (Huser et al., 1998; Mazzocchi et al., 2016; Valverde et al., 2019).

In contrast, the increase in SR Ca2<sup>+</sup> leak evoked by increasing SR Ca2<sup>+</sup> uptake was unable to prevent but rather enhanced heart attack. It was speculated that increasing SR Ca2<sup>+</sup> uptake was not efficient to hamper the excessive flow of SR Ca2<sup>+</sup> to the mitochondria, aggravating cardiac damage (Valverde et al., 2019).

#### Experimental Evidence

In the following sections, we will give experimental evidence that highlight the importance of Ca2<sup>+</sup> handling misbalance in the production of Ca2<sup>+</sup> triggered arrhythmias and cell death. Although the mechanisms of cardiac arrhythmias and apoptosis/necroptosis are usually multifactorial, we will concentrate on experimental examples that emphasize the role of PTM of SR Ca2<sup>+</sup> handling proteins. When possible, the interplay between Ca2<sup>+</sup> uptake and release in determining arrhythmias and cardiac damage will be also discussed.

#### Abnormal RyR2 Regulation in the Development of Diabetic Cardiomyopathy

The current typical definition of DCM comprises structural and functional abnormalities of the myocardium in diabetic patients independently of other risk factors, as coronary artery disease or hypertension (Aneja et al., 2008). DCM is the last stage of cardiac damage in Type 1 and 2 Diabetes Mellitus (T1DM and T2DM, respectively) and of metabolic syndrome associated with insulin resistance. The subcellular mechanisms involved in this last stage of DCM have been discussed in several previous reviews (Bugger and Abel, 2014; Cox and Marsh, 2014; Fuentes-Antras, Picatoste et al. 2015, Fuentes-Antras, Picatoste et al. 2015). Most of them concluded that mayor alterations of Ca2<sup>+</sup> handling, protein expressions, and activities in overt DCM mimic those of HF from different etiologies, including a decrease in SERCA2a activity, SR Ca2<sup>+</sup> load, systolic Ca2+, rate of Ca2<sup>+</sup> decay, and increase in SR Ca2<sup>+</sup> leak (Lebeche et al., 2008). Protein glycosylation (Clark et al., 2003) and oxidized CaMKII are significantly upregulated in the DCM (Jay et al., 2006), therefore CaMKII activation and RyR2 phosphorylation has been proposed as a potential mechanism of heart failure, ventricular arrhythmias and apoptosis in this disease (Daniels et al., 2015). In rats with T2DM, the opening of mPTP in ventricular myocytes was shown to be mainly influenced by the increased ROS and decreased ATP content. It was suggested that Ca2<sup>+</sup> mishandling due to the slow rate of SR Ca2<sup>+</sup> uptake could play a role in increasing mPTP opening that might further exacerbate mitochondrial dysfunction and induce cell death (Riojas-Hernandez et al., 2015). Moreover, ROS derived from hyperglycemia trigger myocardial apoptosis by

mitochondrial cytochrome c release and consequent activation of the caspase-3 pathway (Cai et al., 2002). Although the underlying mechanism of cell death in DCM is not clear yet, these results suggest that a possible pathway underlying apoptosis in DMC is linked to Ca2<sup>+</sup> mishandling and mitochondrial-SR Ca2<sup>+</sup> crosstalk, as described above (For further review about the role of mitochondria on cardiac arrhythmogenesis in DCM see (El Hadi et al., 2019).

Much less is known about Ca2<sup>+</sup> handling and mishandling and the potential occurrence of arrhythmias and apoptosis in the first stages of this disease, in which subclinical events develop for years before the clear emergence of HF symptoms. A possible cause for this lack of information may lie in the poor diagnosis of this stage of the disease and in the different models used to study prediabetic molecular events (King, 2012; Graham and Schuurman, 2015; King and Bowe, 2016). Indeed, the results that reveal the underlying mechanisms of the pathogenesis of DCM, are different according to the model, the degree of evolution of the disease before reaching DCM and the gender explored. Examples of conflicting results at the level of Ca2<sup>+</sup> handling and particularly of RyR2 are observed in the metabolic syndrome model, (db/db mice), which lacks leptin receptors (Chen et al., 1996). In male db/db mouse hearts, the levels of RyR2 were found to be depressed and the RyR2 phosphorylation at the CaMKII site was not altered. However, RyR2 phosphorylation at the PKA site was found to be increased (Pereira et al., 2006, 2014). Intriguingly, db/db female mouse hearts showed no changes in RyR2 expression associated with a decrease in PKA and CaMKII RyR2 phosphorylation sites (Pereira et al., 2014). In a model of fructoserich diet (FRD) applied to male rats and mice, our laboratory has described that this hypercaloric diet, which also induces insulin resistance, increases CaMKII activation and RyR2 dysfunction due to Ser<sup>2814</sup> phosphorylation (Sommese et al., 2016; Federico et al., 2017). On the other hand, experiments in vitro revealed that during acute hyperglycemia, RyR2 activity can also be altered. Hyperglycemia leads to O-Glc-NAcylation of proteins such as CaMKII. Erickson et al. elegantly showed that the acute increase of glucose or O-linked N-acetylglucosamine is directly responsible for CaMKII-dependent diastolic RyR2 Ca2<sup>+</sup> leak and SR Ca2<sup>+</sup> load depletion in hyperglycemia (Erickson et al., 2013). Additionally, we recently demonstrated that in an early diabetic stage, prevention of CaMKII activation by ROS avoided

SR Ca2<sup>+</sup> leak evoked by CaMKII-dependent phosphorylation of RyR2 (Sommese et al., 2016). Thus, the available results indicate that a main disturbance of Ca2<sup>+</sup> handling in prediabetic hearts occurs at the level of RyR2 phosphorylation with the consequent increase in SR Ca2<sup>+</sup> leak and the possibility of triggered arrhythmias and cell death. Indeed, during prediabetic states, the risk of cardiovascular events is already increased and myocardial abnormalities might appear before the diagnosis of T2DM. These alterations are thought to root irregularities at the cardiac myocyte level, ultimately contributing to structural and functional anomalies observed in DCM (Hajat et al., 2004). Actually, in the mouse model of impaired glucose tolerance (IGT) mentioned above (FRD-induced insulin resistance), we found serious cardiac disorders (Sommese et al., 2016; Federico et al., 2017). The fact that this model lacks the more frequent co-morbidities of DCM, supports the metabolic origin of the alterations at the cell level. These animals develop remarkable cardiac remodeling (Sommese et al., 2016). Also, ventricular myocytes exhibit cardiac arrhythmogenic events leading to ventricular arrhythmias which can be prevented in transgenic mice expressing the CaMKII inhibitor AIP targeted to the SR membranes, avoiding phosphorylation of SR proteins (PLN and RyR2) by CaMKII (SR-AIP mice) (**Figure 8**). Moreover, a ROS scavenger as tempol could avoid RyR2 phosphorylation and SR Ca2<sup>+</sup> leak, preventing the arrhythmogenic pattern of the prediabetic cells (Sommese et al., 2016). Of note, the increase in RyR2 phosphorylation observed in FRD myocytes decreases SR Ca2<sup>+</sup> content. This decrease occurs in spite of the increase in SERCA2a activity which contributes to preserving SR Ca2<sup>+</sup> load. This increase, which may be due to the CaMKII-dependent increase in Thr<sup>17</sup> phosphorylation of PLN, would contribute to avoiding a further decrease in SR Ca2<sup>+</sup> load but would also favor SR Ca2<sup>+</sup> leak and arrhythmogenic Ca2<sup>+</sup> waves.

Apoptosis is also an early sign of myocardial dysfunction in the evolution of the diabetic disease, preceding the increase in collagen which may lead to structural and irreversible alterations (Federico et al., 2017). We described a cascade of events initiated by a CaMKII-induced increase in SR Ca2<sup>+</sup> leak which is linked to mitochondrial membrane depolarization and cardiac damage. A particularly striking finding was the CaMKII-induced remodeling of SR-mitochondria microdomains. The latter would strongly support SR–mitochondria dialogue, facilitating Ca2<sup>+</sup> drain to the mitochondria and cell death, in the scenario of an increased SR Ca2<sup>+</sup> leak (Federico et al., 2017). The elucidation of the intracellular signaling pathway of this altered SR-mitochondria relationship would further contribute to the knowledge of DCM molecular alterations. Further investigations are needed to examine the proteins involved in the SRmitochondria communication (like mitofusin-2, Mfn2, and the chaperone glucose-related protein 75, GRP-75), a completely unexplored field in this disease.

#### Ca2<sup>+</sup> Triggered Arrhythmias Induced by Digitalis Intoxication

Cardiac glycosides have been used for the treatment of HF over the last 200 years due to their inotropic properties (Altamirano et al., 2006; Gonano et al., 2011). Although many doubts about their safety in HF treatment have emerged mainly at the end of the last millennium, it is still considered a valuable cardiac tool in some particular scenarios (see for review Whayne, 2018). Unfortunately, these compounds have a very narrow therapeutic range due to their toxic effects that include an enhanced propensity to arrhythmias. The arrhythmic effects of cardiac glycosides have been traditionally attributed to an increase in SR Ca2<sup>+</sup> load which, by leading to an increase in Ca2<sup>+</sup> leak, would evoke cytosolic Ca2<sup>+</sup> waves and triggered arrhythmias (Wier and Hess, 1984; Fujiwara et al., 2008; Eisner et al., 2009; Weiss et al., 2011). More recent experiments indicated that a change in RyR2 may be also involved in cardiac glycosides–induced arrhythmias. Experiments by Gyorke's group (Ho et al., 2011) indicate that the arrhythmogenic effect of cardiotonic glycosides is linked to NOX2-dependent ROS release from mitochondria. The increase in ROS was initially thought to produce RyR2 thiol oxidation that would increase the sensitivity of the channel to luminal Ca2+, thus lowering the critical SR Ca2<sup>+</sup> content at which spontaneous Ca2<sup>+</sup> waves occur (Terentyev et al., 2008). However, simultaneous experiments by Gonano et al., indicated that ouabain-induced arrhythmias requires CaMKII activation: Chronic administration or high–toxic doses of ouabain administered acutely, increased CaMKII activity in mouse hearts (Gonano et al., 2011). Moreover, inhibition of CaMKII was able to prevent spontaneous contractions in isolated myocytes and arrhythmias in intact mouse hearts, without affecting ouabain inotropic action. These experiments also showed that CaMKII phosphorylates both, RyR2 and PLN, which would increase SR Ca2<sup>+</sup> leak and SERCA2a activity. Although this later effect would add to the increase in SR Ca2<sup>+</sup> load resulting from Na+-K+-ATPase inhibition, these experiments concluded that CaMKII-dependent PLN phosphorylation might not contribute to ouabain–induced increase in SR Ca2<sup>+</sup> content and inotropic effect since they were of similar magnitude in the absence and presence of CaMKII inhibition. However, since CaMKII dependent phosphorylation of RyR2 was also inhibited, it might be that the resultant similar SR Ca2<sup>+</sup> load observed after CaMKII inhibition was due to the prevention of SR Ca2<sup>+</sup> leak. Moreover, the fact that CaMKII inhibition prevents arrhythmias without affecting the ouabain-induced increase in SR Ca2<sup>+</sup> load would suggest that the increase in SR Ca2<sup>+</sup> content produced by the drug is not enough to reach the necessary threshold to trigger, by itself, ouabain-induced arrhythmias (Gonano et al., 2011). An increase in RyR2 sensitivity is needed.

The role of CaMKII-dependent induced increase in SR Ca2<sup>+</sup> leak and ventricular arrhythmias was later confirmed by experiments by Gyorke's group, which revealed that replacement of Ser<sup>2814</sup> site of RyR2 by Ala -a non-phosphorylatable amino acid-prevents ouabain-induced Ca2<sup>+</sup> leak and arrhythmias. These results definitively confirmed that phosphorylation, rather than RyR2 oxidation, was required for the increase in channel spontaneous activity and arrhythmogenesis in the context of digitalis toxicity. Instead, the increase in ROS would contribute to CaMKII activation that in turn produces the observed RyR2 phosphorylation (Palomeque et al., 2009; Ho et al., 2014).

Importantly, other studies have shown that mitochondria are also involved in the toxic and arrhythmogenic effects

of cardiotonic glycosides. The results of these investigations indicate that ouabain-induced increase in cytoplasmic Na<sup>+</sup> compromises mitochondrial energetics and redox balance by blunting mitochondrial Ca2<sup>+</sup> accumulation. Improving mitochondrial Ca2<sup>+</sup> retention by inhibition of mitochondrial NCX, can mitigate these effects, suppress Ca2+-triggered arrhythmias and improve the positive inotropic effects of cardiac glycosides (Liu et al., 2010).

#### Abnormal Ca2<sup>+</sup> Handling in Ischemia/Reperfusion

Ischemic heart disease is a leading cause of mortality worldwide. Cardiac ischemia reduces cardiac output and promotes arrhythmias and cell death. Reperfusion therapies are the standard treatment for patients suffering myocardial infarction, however, re-establishing blood flow is associated with additional cell damage (I/R injury), and exacerbating the effect of the preceding ischemia. Indeed, it was shown that reperfusion may trigger life-threatening arrhythmias, accounting for up to half of the total I/R-induced infarcts (Braunwald and Kloner, 1985; Yellon and Hausenloy, 2007; Said et al., 2011; Garcia-Dorado et al., 2012). Although the factors contributing to I/R injury are complex (see for review (Murphy and Steenbergen, 2008). I/R injury constitutes another example in which experimental evidence reveals that disturbed Ca2<sup>+</sup> handling and mitochondria ROS production are main responsible for reperfusion arrhythmias and cardiac damage (Valverde et al., 2006, 2010; Mattiazzi et al., 2015; Bagheri et al., 2016). The increase in cytosolic Ca2<sup>+</sup> during ischemia was associated with an enhancement of SR Ca2<sup>+</sup> load (Valverde et al., 2010), which is released at the onset of reperfusion and produces an abrupt rise in cytosolic Ca2<sup>+</sup> (Ca2<sup>+</sup> bump) and the consequent decrease in SR Ca2<sup>+</sup> content and Ca2<sup>+</sup> transient. Moreover, a major mechanism for the ischemia-induced increase in diastolic Ca2<sup>+</sup> is an increase in the frequency of Ca2<sup>+</sup> sparks which may switch to arrhythmogenic Ca2<sup>+</sup> waves during reperfusion (Mattiazzi et al., 2015).

Reactive oxygen species/reactive nitrogen species are generated during reperfusion by several different cellular sources, being the mitochondria the more important one. Mitochondrial Ca2<sup>+</sup> overload and subsequently ROS production trigger mitochondrial permeability transition pore and ROS production via ROS-induced ROS release mechanisms (Zorov et al., 2000). Both, Ca2<sup>+</sup> mishandling and ROS production set an ideal intracellular milieu for activation of CaMKII, which play a main role in I/R arrhythmias, apoptosis, and necroptosis. Ex vivo and in vivo experiments described an increase in phosphorylated-CaMKII (p-CaMKII) and oxidized-CaMKII (ox-CaMKII) at the onset of reperfusion (Said et al., 2011; Becerra et al., 2016), which was associated to a significant increase in the phosphorylation of Thr<sup>17</sup> site and RyR2 Ser<sup>2814</sup> (Vittone et al., 2002; Salas et al., 2010; Ling et al., 2013), as well as redox changes of RyR2 (Becerra et al., 2016). Reperfusion arrhythmias are largely dependent on SR Ca2<sup>+</sup> leak evoked by these PTM of RyR2 (Said et al., 2008; Becerra et al., 2016). In this scenario the role played by a substantial increase in SR Ca2<sup>+</sup> uptake was similar to that observed in stress-induced Ca2+-triggered arrhythmias, i.e., increasing SR Ca2<sup>+</sup> uptake by PLN ablation protects against reperfusion arrhythmias. This protection was achieved by alteration of Ca2<sup>+</sup> wave propagation, which were transformed in non-arrhythmogenic mini-waves and reconverted in full Ca2<sup>+</sup> waves in the presence of SERCA2a inhibition (Figure 6C; Valverde et al., 2019). Of note, PLN ablation is equivalent to a situation of permanent maximal PLN phosphorylation as stated above. In ischemic reperfused WT hearts, PLN phosphorylation is highly but transiently increased at the onset of reperfusion (Vittone et al., 2002; Said et al., 2008). Under these conditions, we did not observe Ca2<sup>+</sup> mini-waves (Valverde et al., 2010). This would mean that Thr<sup>17</sup> phosphorylation at the onset of reperfusion is unable to prevent SR Ca2<sup>+</sup> waves progression. In contrast, it might help to increase SR Ca2<sup>+</sup> load and maintain SR Ca2<sup>+</sup> leak. This contention is supported by experiments in which hearts were subjected to a short I/R protocol to produce stunning. In the stunned heart, the transient phosphorylation of Thr<sup>17</sup> of PLN is essential for contractile recovery upon reperfusion, even though phosphorylation of RyR2 also occurs and induces reperfusion arrhythmias (Said et al., 2003, 2011).

After a prolonged ischemic period, reperfusion evokes irreversible cardiac injury. Under these conditions, myocytes die by apoptosis, autophagy, and necrosis. The rise in Ca2<sup>+</sup> during I/R leads to mitochondrial Ca2<sup>+</sup> accumulation, which is greatly favored by the close association between mitochondria and SR and constitutes a main event in the initiation of cell death (Rizzuto and Pozzan, 2006). Experimental evidence consistently indicates that CaMKII inhibition is protective in the irreversible I/R injury (Zhang et al., 2005; Vila-Petroff et al., 2007; Salas et al., 2010). Although the mechanisms for myocardial protection by CaMKII inhibition are still unclear, one of the CaMKII deleterious pathway in I/R certainly involves the SR and the mitochondria (Vila-Petroff et al., 2007; Salas et al., 2010; Valverde et al., 2010; Joiner et al., 2012). In previous experiments, we showed that I/R damage was diminished in hearts from S2814A mice. Conversely, in the hearts of S2814D mice (constitutively pseudo-phosphorylated), cardiac damage increased (Di Carlo et al., 2014). A decrease in the expression of RyR2 described in I/R (Salas et al., 2010), compatible with a degradation/damage of these channels (Pedrozo et al., 2010) and changes in RyR2 activity induced by redox alterations, may contribute to increase SR Ca2<sup>+</sup> leak (Hidalgo et al., 2004; Said et al., 2011; Di Carlo et al., 2014). These alterations in RyR2 would add to the deleterious action of RyR2 phosphorylation favoring the increase in mitochondria Ca2<sup>+</sup> content and greatly contributing to necroptosis and apoptosis in reperfusion cardiac damage (Salas et al., 2010; Di Carlo et al., 2014). This cascade would be further stimulated by the increase in CaMKII-dependent phosphorylation of MCU described by Joiner, Koval et al.(Joiner et al., 2012). However, recent experiments do not support a relevant role of CaMKII for mitochondrial Ca2<sup>+</sup> uptake in cardiac myocytes at least under physiological conditions (Nickel et al., 2019).

As stated above, the transient increase in CaMKII-dependent PLN phosphorylation plays a beneficial role in the stunned heart. The role of PLN phosphorylation in I/R injury is less clear. After prolonged ischemia, we showed that preventing PLN

phosphorylation exacerbates the functional and structural heart damage after myocardial infarction, suggesting that CaMKIIdependent phosphorylation of PLN observed during reperfusion favors post-ischemic recovery and protects from I/R cardiac damage (Di Carlo et al., 2014). However, the fact that I/R does produce cardiac damage indicates that even being beneficial, CaMKII-dependent PLN phosphorylation results insufficient to counteract the effect of simultaneous detrimental mechanisms that take place during I/R. On the other hand, several groups have tested the effects of increasing SR Ca2<sup>+</sup> uptake on cardiac damage during I/R injury, by different maneuvers. The outcome of these experiments is controversial. For instance, Yang et al. (2006) demonstrated that the protective effect of chronic CaMKII inhibition in AC3I mice was lost when they were interbred with PLNKO mice and submitted to myocardial infarction, supporting a detrimental effect of enhancing of SR Ca2<sup>+</sup> uptake. Similar results were obtained in our laboratory. The ablation of PLN in SDKO mice increases Ca2<sup>+</sup> leak upon reperfusion (**Figure 6A**). This increase was associated with an increase in infarct size and mitochondrial dysfunction. Therefore, these experiments demonstrated that an important increase in SR Ca2<sup>+</sup> uptake as that produced by PLN ablation, was able to prevent reperfusion arrhythmias, but failed to prevent, and even enhance, cardiac damage (Valverde et al., 2019). The important increase in SR Ca2<sup>+</sup> uptake would favor the unbalance between SR Ca2<sup>+</sup> uptake and leak, promoting mitochondrial Ca2<sup>+</sup> overload and cell death. Other studies demonstrated that accelerating SR Ca2<sup>+</sup> uptake by different means (i.e., overexpressing SERCA1a, with higher kinetics than SERCA2a, or expressing a repressor of PLN activity, PP1 H-1), alleviated post-ischemic cardiac injury (Talukder et al., 2007, 2008; Nicolaou et al., 2009a), supporting a beneficial effect of accelerating SR Ca2<sup>+</sup> uptake. These controversial results seem not to arise from species differences since most of the experiments mentioned above referred to rodents. It is possible that the final beneficial or detrimental outcome of increasing SR Ca2<sup>+</sup> uptake might tightly depend on the extent of Ca2<sup>+</sup> uptake and SR Ca2<sup>+</sup> load achieved during ischemia and at the onset of reperfusion. For instance, moderate increases in SR Ca2<sup>+</sup> content have been associated with beneficial effects (Nicolaou et al., 2009b), whereas more important increases, like those expected in PLNKO mice, were associated with detrimental actions (Valverde et al., 2019).

#### Is RyR2 Activation Always Detrimental?

We have previously associated the increase in RyR2 activation, for instance by CaMKII-dependent phosphorylation, with cardiac damage and arrhythmias due to exacerbated diastolic SR Ca2<sup>+</sup> leak, as discussed above. However, it is important to bear in mind that potentiation of RyR2 activity persists during systole and enhances systolic fractional Ca2<sup>+</sup> release, bringing the heart to a new state of higher efficiency. This would allow the heart to maintain a given contractility despite a decrease in SR Ca2<sup>+</sup> content or to enhance contractility if SR Ca2<sup>+</sup> content is simultaneously preserved (for further discussion, see Lascano et al., 2017).

#### CONCLUDING REMARKS

The normal interplay among the proteins involved in SR Ca2<sup>+</sup> uptake and release is a main determinant of the regular beat to beat contractile function of cardiac myocytes. Regulation and deregulation of these proteins are crucial to understanding the balance between SR Ca2<sup>+</sup> uptake and leak, responsible for SR Ca2<sup>+</sup> content and myocardial contractility, as well as its unbalance, which determines an excess of SR Ca2<sup>+</sup> leak, able to produce arrhythmias and cardiac damage. Posttranslational enhancements of SR Ca2<sup>+</sup> uptake have a beneficial effect resulting in a detectable increase in contractility when the unbalance between SR Ca2<sup>+</sup> uptake and leak favors the uptake. This is the case of PLNKO mice and stunned hearts. When SR Ca2<sup>+</sup> leak is increased by the enhancement of RyR2 activity, the increase in SR Ca2<sup>+</sup> uptake may not be enough to counteract SR Ca2<sup>+</sup> leak, resulting in two opposite effects: (a) beneficial, by opposing to SR Ca2<sup>+</sup> leak and rescuing at least in part SR Ca2<sup>+</sup> load and contractility; (b) detrimental, because the increase in SR Ca2<sup>+</sup> load would favor SR Ca2<sup>+</sup> leak, arrhythmias, and cardiac damage. Interestingly, a greatly exacerbated increase in SR Ca2<sup>+</sup> uptake, as that produced by PLN ablation, does contribute to increasing SR Ca2<sup>+</sup> leak and cardiac damage by incrementing SR Ca2<sup>+</sup> load and leak. Paradoxically, and in spite of the exacerbated SR Ca2<sup>+</sup> leak, this increase may prevent Ca2<sup>+</sup> triggered arrhythmias, by a different mechanism, i.e., diminishing cytosolic Ca2<sup>+</sup> and avoiding Ca2<sup>+</sup> wave propagation. Post-translational activation of RyR2 activity would produce a deleterious effect by increasing SR Ca2<sup>+</sup> leak and predisposing to cardiac damage and arrhythmias. When this modification occurs only at the level of SR Ca2<sup>+</sup> release/leak, the short-lived enhancement of SR Ca2<sup>+</sup> leak may produce a decrease in contractility due to the decrease in SR Ca2<sup>+</sup> load, without any further detrimental effect. Moreover, the increased activity of RyR2 during systole tends to preserve contractility, even at lower SR Ca2<sup>+</sup> loads. The detrimental effect of RyR2 activation (i.e., arrhythmias and cardiac damage) can only take place when it occurs associated to an increase in SR Ca2<sup>+</sup> uptake, able to maintain SR Ca2<sup>+</sup> leak. This is the typical case of Ser2814D myocytes, with constitutive pseudo-phosphorylation of RyR2 at the CaMKII site.

#### AUTHOR CONTRIBUTIONS

JP and AM generated the idea. JP, AM, MF, and CV wrote the manuscript, designed, and edited the figures.

#### FUNDING

This work was supported by PICT 2015-3009 (FONCyT, Argentina) and PS-1 (UAI, Argentina) to JP, PICT 2015- 3210 (FONCyT, Argentina) to CV, and PICT 2524 (FONCyT, Argentina) and PIP 0305 (CONICET, Argentina) to AM. MF is a doctoral fellow from CONICET.

#### REFERENCES

fphys-10-01630 January 21, 2020 Time: 17:47 # 16



cardiac muscle sarcoplasmic reticulum calcium pump by nitroxyl. Biochemistry 47, 13150–13152. doi: 10.1021/bi801925p


arrhythmias. J. Mol. Cell Cardiol. 89(Pt B), 349–359. doi: 10.1016/j.yjmcc.2015. 10.004



type 2 diabetic mice. Diabetes Metab. Res. Rev. 55, 608–615. doi: 10.2337/ diabetes.55.03.06.db05-1284


heterozygous SERCA2a knockout mice. Am. J. Physiol. Heart Circ. Physiol. 294, H1426–H1434. doi: 10.1152/ajpheart.01016.2007



**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Federico, Valverde, Mattiazzi and Palomeque. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

digital media

of impactful research

article's readership