Identification and Characterization of the Role of REEP5 in Sarco-Endoplasmic Reticulum Formation, Maintenance, and Function in Cardiac Muscle

Frank Shin-Haw Lee

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy

Department of Physiology University of Toronto

Frank Shin-Haw Lee

Doctor of Philosophy

Department of Physiology University of Toronto

2021

ABSTRACT

Heart failure (HF) remains the most rapidly rising cardiovascular disease and the leading cause of inpatient hospitalization worldwide, with costs exceeding $30 billion dollars annually in North America. While many key regulators of heart function have been identified, yet effective therapies aimed at healing or reversing the progression of HF remain restricted due to the complex nature of the disease and a lack of understanding of the functional membrane proteome of the heart. Here, we created a blueprint of all critical membrane and membrane-associated proteins in the heart by mapping several transcriptomic- and proteomic-based datasets against our mass spectrometry dataset of membrane-enriched protein clusters from human fetal and mouse neonatal cardiomyocytes. We identified 173 membrane-associated proteins that are conserved among eukaryotic species, cardiac-enriched, and have not been previously linked to a cardiac phenotype. These poorly annotated cardiac-enriched membrane proteins represent excellent candidates in follow-up studies aimed at elucidating the underlying molecular mechanisms of cardiomyopathies and HF. One of the highly ranked, poorly

annotated, and cardiac-enriched membrane proteins was REEP5, a sarco-endoplasmic reticulum (SR/ER) membrane protein. In a follow-up study we tested the hypothesis that physiological REEP5 expression is important for cardiac SR/ER organization and heart function. In vitro REEP5 depletion in isolated functional adult mouse cardiomyocytes resulted in SR/ER membrane vacuolization and impaired cellular processes including activated cardiac ER stress pathways and dysregulated Ca2+ cycles. Subsequent in vivo

CRISPR/Cas9-mediated REEP5 loss-of-function zebrafish mutants showed sensitized cardiac dysfunction upon pharmacological HF induction. Similarly, in vivo adeno- associated viral (AAV9)-induced REEP5 depletion in the mouse resulted in lethal diastolic cardiac dysfunction with dilated cardiac chambers and reduced ejection fraction.

Altogether, these results demonstrated 1. Our cardiomyocyte membrane proteome dataset proves instrumental to studies aimed at characterizing novel regulators of heart function and identifying potential heart disease markers and/or therapeutic targets, 2. The critical role of REEP5 in cardiac SR/ER organization, embryonic heart development, and heart function.

iii

ACKNOWLEDGEMENTS

I would like to thank my supervisor Dr. Anthony Gramolini for his guidance and support throughout my time as a part of your research group. I deeply appreciate the opportunity to learn and grow as an independent scientist in your lab. You have been a great supervisor, teacher, and mentor for me and thank you for guiding me both professionally and personally. To my committee members, Drs. Scott Heximer and Christine Bear, whose valuable input and assistance were essential to the success of this work.

I would like to show my profound appreciation to my colleagues in the lab and at the Ted Rogers Centre for Heart Research for all your support and encouragement. Special thanks go to my lab family members, Sina, Uros, and Jake for being an inspiration to me. I would also like to acknowledge my inspiring collaborators, Harsha and Neal, who contributed enormous time and effort, this research could not have been possible without all your endeavours and commitment to science.

Further, I would like to acknowledge the financial support from NSERC, CIHR, Government of Ontario, the Peterborough K.M. Hunter Charitable Foundation, and the Ted Rogers Centre for Heart Research. Thank you for your generous contributions, these awards have made a great difference in my graduate education by recognizing the importance of my work and allowing me to pursuit and achieve my academic and research goals effectively.

Lastly, I would like to thank my parents and my brother for their unconditional support and encouragement throughout the past few years. You have always been there for me every step of the journey, celebrate with me at my best and stand by me at my worst times. I am truly lucky and grateful to have you in my life.

TABLE OF CONTENTS

ACKNOWLEDGEMENTS ...... IIV

TABLE OF CONTENTS ...... IV

LIST OF COMMON ABBREVIATIONS ...... VIII

LIST OF FIGURES & TABLES ...... X

CHAPTER 1: INTRODUCTION ...... 1

1.1 BACKGROUND & RATIONALE ...... 1

1.2 RESEARCH OBJECTIVES ...... 2

1.3 PRIMARY HYPOTHESES ...... 3

CHAPTER 2: LITERATURE REVIEW ...... 5

2.1 OVERVIEW OF CARDIAC PHYSIOLOGY ...... 5

2.2 MOLECULAR ASPECTS OF CARDIAC FUNCTION ...... 10

2.3 MODEL ORGANISMS IN CARDIOVASCULAR BIOLOGY ...... 26

2.4 HEART FAILURE ...... 37

CHAPTER 3: BIOINFORMATIC ANALYSIS OF MEMBRANE AND ASSOCIATED

PROTEINS IN MURINE CARDIOMYOCYTES AND HUMAN MYOCARDIUM ...... 42

CONTRIBUTIONS ...... 43

3.1 ABSTRACT ...... 44

3.2 INTRODUCTION ...... 45

3.3 RESULTS ...... 46

3.4 DISCUSSION ...... 68

3.5 METHODS ...... 73

3.6 DATA AVAILABILITY ...... 80

3.7 REFERENCES ...... 80

3.8 SUPPLEMENTARY INFORMATION ...... 87

CHAPTER 4: REEP5 DEPLETION CAUSES SARCO-ENDOPLASMIC RETICULUM

VACUOLIZATION AND CARDIAC FUNCTIONAL DEFECTS ...... 107

CONTRIBUTIONS ...... 108

4.1 ABSTRACT ...... 110

4.2 INTRODUCTION ...... 111

4.3 RESULTS ...... 112

4.4 DISCUSSION ...... 139

4.5 METHODS ...... 142

4.6 DATA AVAILABILITY ...... 151

4.7 REFERENCES ...... 152

4.8 SUPPLEMENTARY INFORMATION ...... 160

CHAPTER 5: DISCUSSION & CONCLUSION ...... 174

5.1 UNIFYING DISCUSSION ...... 174

5.2 LIMITATIONS ...... 182

5.3 RECOMMENDED FUTURE DIRECTIONS ...... 186

CHAPTER 6: APPENDICES ...... 189

6.1 APPENDIX I: LIST OF PUBLICATIONS ...... 189

6.2 APPENDIX II: REPRINT LICENSES ...... 192

CHAPTER 7: REFERENCES ...... 197

vii

LIST OF COMMON ABBREVIATIONS

AAV9 Adeno-associated virus serotype 9

ATL Atlastin

CKAP4 Cytoskeleton-associated protein 4

Ca2+ Calcium ions

Cas9 CRISPR associated protein 9

COX20 Cytochrome c oxidase assembly protein COX20

CRISPR Clustered regularly interspaced short palindromic repeats

DCM Dilated cardiomyopathy

ECC Excitation-contraction coupling

EF Ejection fraction

FAM162A Protein FAM162A

GO Gene ontology

HCM Hypertrophic cardiomyopathy

HF Heart failure

HFpEF Heart failure with preserved ejection fraction

HFrEF Heart failure with reduced ejection fraction hpf Hours post fertilization

ICD Intercalated discs

ICM Ischemic cardiomyopathy

IF Immunofluorescence

KD Knockdown kDa Kilodaltons

viii

MCT1 Monocarboxylate transporter 1

MO Morpholino

MS Mass spectrometry

NFC Non-failing control

REEP5 Receptor expression enhancing protein 5

ROS Reactive oxygen species

RTN Reticulon

SEM Standard error of the mean shRNA Short hairpin RNA

SR/ER Sarco-endoplasmic reticulum

TEM Transmission electron microscopy

LIST OF FIGURES & TABLES

CHAPTER 2

Figure 2.1. The cardiac cycle. Schematic illustration of a complete heartbeat..…………7

Figure 2.2. The cardiac conduction system. Schematic illustration of the electrical conduction system in the heart…………………………………………………………………8

Figure 2.3. Cardiac myofibrils and myofilaments. Schematic illustration of a cardiac myofibril containing sarcomeres between Z-discs…………………………………………. 11

Figure 2.4. Overview of cardiac excitation-contraction coupling. Schematic illustration of the excitation-contraction coupling mechanism…………………………….. 12

Figure 2.5. Overview of the unfolded protein response (UPR). Schematic illustration of the UPR signaling cascade upon ER stress……………………………………………....17

Figure 2.6. Evolution of cardiovascular proteomics. Schematic timeline of major milestones in the field of cardiovascular proteomics………………………………………..20

Figure 2.7. Trends in the capacity of proteomics studies of the human and mouse heart. Total protein coverage of proteomic studies of the (a) human and (b) mouse heart from 1990-2020………………………………………………………………………………...22

Figure 2.8. Systolic and diastolic dysfunction in HF patients. Schematic illustration of the two different types of HF affecting the left ventricle…………………………………..37

Table 2.1. Timeline of major cardiac events during heart development in the human, mouse, and zebrafish heart.………………………………………………………………...30

Table 2.2. Cardiac echocardiographic parameters in the human, mouse, and zebrafish heart.………………………………………………………………………………..32

Table 2.3. Common interventions and genetic manipulations of the cardiovascular system to model human heart diseases.………………………………………………….35

CHAPTER 3

Figure 3.1. Membrane proteomic analysis of human fetal and murine ventricular cardiomyocytes.…………………………………………………………..….…...……….... 47

Figure 3.2. Analysis of cardiac membrane proteins within various subcellular domains.………………………………………………………………………………………. 50

Figure 3.3. Transcriptomic analysis of novel cardiomyocyte-enriched membrane proteins shows cardiac enrichment across various tissues.…………………………52

Figure 3.4. Immunohistochemistry analysis of top ranked cardiac-enriched candidates.…………………………………………………………………………………….54

Figure 3.5. Immunoblot analysis of protein expression across multiple tissues and microsomal fractions.………………………………………………………………………..57

Figure 3.6. Membrane topology and prediction analysis of FAM162A, MCT1, and COX20.……………………………………………………………….…………………………59

Figure 3.7. Immunofluorescence analysis of FAM162A, MCT1, and COX20 expression in mouse neonatal and adult ventricular cardiomyocytes.……………..61

Figure 3.8. Co-immunofluorescence analysis demonstrates colocalization of FAM162A and COX20 with known mitochondrial marker, COXIV, and MCT1 colocalization with known plasma membrane protein, Gαi, in isolated adult mouse cardiomyocytes.………………………………………………………………………..….....63

Figure 3.9. GEO transcriptomic analysis of FAM162A, MCT1, and COX20 mRNA transcript levels in various mouse and human heart diseases.………………………65

Figure 3.10. FAM162A, MCT1, and COX20 protein expression levels in human adult DCM and ICM heart failure patients.……………………………………………………….68

Table 3.1. Patient baseline and clinical characteristics...... 67

Supplemental Figure 3.1. Transcriptomic analysis of the 550 membrane and membrane-associated protein clusters across various tissues.……………………..87

Supplemental Figure 3.2. Transcriptomic analysis of cardiomyocyte-enriched membrane associated proteins with previous cardiac MGI phenotype.…………….89

Supplemental Figure 3.3. Transcriptomic analysis of non-cardiomyocyte-enriched membrane associated proteins with no previous cardiac MGI phenotype.………...90

Supplemental Figure 3.4. Transcriptomic analysis of non-cardiomyocyte-enriched membrane associated proteins with previous cardiac MGI phenotype.…………….92

Supplemental Figure 3.5. FAM162A and COX20 share high degrees of homology throughout evolution.………………………………………………………………………..93

Supplemental Figure 3.6. A multi-species alignment of MCT1 from selected vertebrates.…………………………………………………………………………………… 95

Supplemental Figure 3.7. Transcriptomic and phylogenetic analyses of the SLC16A/MCT family of proteins.…………………………………………………………...97

Supplemental Figure 3.8. Co-immunofluorescence analysis of FAM162A, MCT1, and COX20 with known mitochondrial marker, COXIV and known plasma membrane protein, Gαi in isolated adult mouse cardiomyocytes.…………………...99

Supplemental Figure 3.9. Original uncropped immunoblots.………………….101-105

CHAPTER 4

Figure 4.1. REEP5 is an evolutionarily conserved, muscle-enriched membrane protein.………………………………………………………………………………………..114

Figure 4.2. REEP5 expression shows consistent SR staining pattern in cardiac myocytes.……………………………………………………………………………….……116

Figure 4.3. In vitro REEP5 depletion in cardiac myocytes results in SR/ER membrane destabilization and dysfunction.…………………………………………...118

Figure 4.4. In vitro REEP5 depletion leads to SR vacuolization and sarcomeric dysfunction in adult mouse cardiac myocytes.………………………………………..121

xii

Figure 4.5. The C-terminal cytosolic domain of REEP5 is required for stabilizing SR/ER morphology.…………………………………………………………………………124

Figure 4.6. Mass spectrometry analysis identifies REEP5 interactions with known cardiac SR shaping proteins.……………………………………………………………..127

Figure 4.7. In vivo CRISPR/Cas9-mediated REEP5 depletion in zebrafish embryos leads to cardiac abnormalities.…………………………………………………………...131

Figure 4.8. Genetic compensation in reep5 CRISPR knockout zebrafish mutant..133

Figure 4.9. In vivo AAV9-mediated REEP5 depletion in mice results in increased cardiac fibrosis, activated cardiac ER stress, cardiac dysfunction, and death.…136

Supplementary Figure 4.1. Phylogenetic analysis of the REEP family of proteins...... 160

Supplementary Figure 4.2. REEP5 immunostaining demonstrates SR staining pattern and co-localization with known j-SR proteins, α-actinin, RyR2, and triadin in cardiac myocytes.………………………………………………………………………..161

Supplementary Figure 4.3. REEP5 depletion-induced SR/ER vacuolization is independent of cellular apoptosis in cardiac myocytes...……………………………163

Supplementary Figure 4.4. Overexpression of dimerized REEP5 robustly marks the ER network in mouse myoblasts.………………………………………………………...165

Supplementary Figure 4.5. Co-immunoprecipitation assays demonstrate dynamic REEP5 interactions with RTN4/Nogo A/B, ATL3, and CKAP4 in neonatal ventricular cardiac lysate.………………………………………………………………………………. 167

Supplementary Figure 4.6. Morpholino-mediated REEP5 depletion in zebrafish embryos causes severe developmental abnormalities.……………………………...168

Supplementary Figure 4.7. CRISPR/Cas9-mediated REEP5 loss-of-function crispants demonstrates cardiac developmental and functional defects in zebrafish embryos.……………………………………………………………………………………...170

Supplementary Figure 4.8. Increased protein expression levels of REEP5, RTN4, ATL3, and CKAP4 in pressure overload-induced mouse failing hearts...... 172

xiii

CHAPTER 5

Figure 5.1. Human cardiac proteome consists of >17,000 unique proteins in the heart.…………………………………………………………………………………………..175

Figure 5.2. Summary of major functions of REEP5 in cardiac myocytes.………….180

xiv

CHAPTER 1: INTRODUCTION

1.1 BACKGROUND & RATIONALE Cardiac membrane proteins are essential for the regulation of spontaneous and synchronous contractions of cardiomyocytes 1, 2. The cardiomyocyte cell membrane, also known as the sarcolemma, has two major functional domains: the intercalated discs and membrane invaginations knowns as the t-tubules, responsible for the propagation of contractile signaling between cardiomyocytes and the initiation of cardiac excitation- contraction coupling, respectively. To ensure synchronous ventricular contraction, this important process is tightly regulated by several classes of membrane proteins in cardiomyocytes. More importantly, due to their exposed position in the cell, many membrane proteins have proven to be an effective diagnostic and therapeutic intervention to reduce overall risk and mortality rate for heart disease.

To date, many key regulators of heart function as well as heart disease-causing mutations have been identified 3, however, effective therapies aimed at halting or reversing the progression of heart disease remain challenging due to a limited number of functional target proteins available as a result of a lack of understanding of the cardiac membrane proteome and its constituents. Moreover, current clinical therapies and pharmacological interventions for failing heart muscle have mainly targeted the beta-adrenergic receptor and the renin-angiotensin pathway as targets for managing HF 4. These symptom- ameliorating treatment approaches have left millions of HF patients underserved. For the past decade, profiling critical cardiac membrane proteins using a proteomics approach has emerged as a popular and reliable method to identify previously underappreciated membrane proteins in the heart. Specifically, the ability to identify organelle-specific membrane proteomes in cardiomyocytes has allowed the development of new therapies to target protein trafficking and function of cardiac membrane proteins at the subcellular level to regulate cardiac Ca2+ cycling and storage, thereby normalizing the altered Ca2+ transients phenotypes in failing cardiomyocytes 5.

The idea of mapping the proteomic make-up of the heart to further our understanding of the molecular and biochemical basis for cardiac physiology and pathophysiology has been recognized since late 1990s. Pioneering studies using 2D electrophoresis were able to separate approximately 2000 proteins from whole cell and tissue extracts. A recent study led by Doll et al. revealed that the human heart proteome consists of over 10,000 distinct proteins 6, with 3,000 of which were estimated to be cardiomyocyte specific and are likely important for heart muscle performance and capacity. However, proteomic make-up of the cardiomyocyte membrane proteome remains to be elucidated given their great importance in governing synchronous beating of the heart. Identification and enrichment of membrane and membrane-associated proteins is perhaps one of the most challenging yet exciting tasks in molecular biology and biochemistry due to their hydrophobic nature and relative low abundance 7. Nonetheless, recent advancements in proteomic technologies have allowed high-resolution mass spectrometry-based techniques to successfully identify membrane proteins of low abundance 8, 9. Detailed characterization of these newly identified, cardiomyocyte specific membrane proteins could provide breakthrough understanding of cardiomyocyte physiology, signaling networks, and protein-protein interactions, leading to potential diagnostic and therapeutic targets.

Taken together, these findings highlight a knowledge gap in identifying and understanding the cardiomyocyte membrane proteome and its constituents. More importantly, the significance and clinical value of these newly identified, cardiomyocyte specific membrane proteins in the prevention, diagnosis, and treatment of heart disease remain unknown. This dissertation directly addressed the critical knowledge gap in elucidating the cardiomyocyte membrane proteome and the role of its constituents in cardiomyocyte physiology, heart function, and the progression of HF.

1.2 RESEARCH OBJECTIVES Objective 1: The primary objective of this dissertation was to generate a non-biased, comprehensive rank-ordered analysis of recently identified cardiomyocyte-enriched

membrane proteins using detailed bioinformatic analyses of several publicly available transcriptomic, proteomic, and phenotypic ontology datasets.

Objective 2: The second main objective of the current dissertation was to validate the utility of our rank-ordered list of cardiac membrane proteins by characterizing the roles of several highly ranked candidates including REEP5, FAM162A, MCT1, and COX20, in regulating heart function and its potential therapeutic value in vitro in isolated adult mouse cardiomyocytes and in in vivo mouse and zebrafish models of heart disease.

These efforts together would prove instrumental to identifying novel molecular regulators of heart function and create in vivo animal models for further therapeutic investigations. Additionally, this work will help delineate underappreciated molecular and signalling pathways that may be of therapeutic value in the prevention of heart disease progression.

1.3 PRIMARY AIMS AND HYPOTHESES i. Many cardiomyocyte-enriched membrane-associated proteins are previously underappreciated and their role in the heart remains poorly understood. I took a systems-biology approach as a ‘hypothesis-generating’ exercise to identify the most interesting, understudied cardiac membrane proteins. ii. Once these proteins were informatically identified and REEP5 was shown to be the protein with the greatest interest in pursuing, I hypothesized that REEP5, a highly ranked novel cardiac membrane-associated protein, was essential in regulating normal heart function. Specifically, I hypothesized that expression of REEP5 was required for the normal formation and maintenance of cardiac SR/ER organization and function, and ultimately cardiomyocyte contractility.

To test my hypotheses, two separate studies were undertaken. In the first study (Chapter 3), entitled “Bioinformatic analysis of membrane and associated proteins in murine cardiomyocytes and human myocardium” (Lee et al., 2020 Scientific Data 7, 425), a

systems-biology approach using a combination of several transcriptomic, proteomic, and phenotypic ontology datasets to identify novel cardiomyocyte-enriched membrane proteins was performed. A non-biased ranking strategy was applied to all candidates and subsequent preliminary informatics and immunohistochemical characterization of selected top ranked proteins including FAM162A, MCT1, and COX20 were carried out to prove the utility of this dataset. In the second study (Chapter 4), entitled “REEP5 depletion causes sarco-endoplasmic reticulum vacuolization and cardiac functional defects” (Lee et al., 2020 Nature Communications 11, 965), a detailed functional characterization of REEP5, the highest ranked cardiac-enriched membrane protein in the previous study, was performed at multiple levels including subcellular localization characterization, cardiomyocyte functional assessment, and proteomic profiling of REEP5 interactome in vitro in functional isolated adult mouse cardiomyocytes. Further, in vivo CRIPSR/Cas9- mediated REEP5 loss-of-function zebrafish mutants as well as in vivo adeno-associated viral (AAV9)-induced REEP5 depletion in the mouse were generated to assess and characterize the role of REEP5 in vivo in the heart.

CHAPTER 2: LITERATURE REVIEW

2.1 OVERVIEW OF CARDIAC PHYSIOLOGY

2.1.1 Anatomy, Physiology, and Function of the Heart The human heart is a complex, muscular organ that pumps blood to supply oxygen and nutrients to the rest of the body via the systemic circulation. This seemingly simple process is regulated by several physiological systems to meet the demands of the body in response to external stimuli such as exercise, eating, and sleeping. The mammalian heart consists of four chambers and four valves to ensure one-directional blood flow in the vascular system. The atria are smaller chambers that sit on top of the ventricles and pump blood directly into the ventricles. The right side of the heart consists of the right atrium and right ventricle connected by the tricuspid valve. The right ventricle is also connected to the outflow tracks through the pulmonary valve to direct blood flow to the lungs to be oxygenated. Anatomically, the right ventricle is situated directly behind the sternum, rendering it the most anteriorly positioned chamber with a normal free wall thickness of 3-5mm compared to 5-9mm in the left ventricle due to lower resistance and greater vascular surface area of the pulmonary circulation 10, 11, 12. For this reason, the right ventricular muscle mass is estimated to be a quarter of that of the left ventricle. The left side of the heart, on the other hand, consists of the left atrium and left ventricle connected by the mitral valve. The left atrium receives oxygenated blood from the pulmonary circulation and pumps it directly into the left ventricle through the mitral valve. Similarly, the left ventricle is connected to the aorta by the aortic valve to pump oxygenated blood returned from the pulmonary circulation out to the systemic circulation.

Taken together, the primary function of the left ventricle is to maintain physiological arterial pressure in the systemic circulation through force generation, ensuring constant and efficient delivery of oxygen and nutrients to other organs in the body. In comparison, the primary function of the right ventricle is to receive deoxygenated blood form venous return in the systemic circulation and direct it to the lungs to be oxygenated. The right and

left side of the heart together makes the heart a powerful organ that supplies continuous blood flow throughout the body starting from day 22 in utero.

2.1.2 The Cardiac Cycle It is estimated that your heart will beat around three billion times in your lifetime assuming an average lifespan to be 80 years. Each heartbeat goes through a complete cardiac cycle and this can be broken down into systole and diastole. A systole happens when one chamber contracts to push blood into another chamber or outflow tracts and a diastole is defined as the relaxation of the chambers to allow blood to fill the chambers. In a normal heart, the atria and ventricles contract and relax in turn and all four chambers of the heart work together in a continuous and coordinated fashion (Figure 2.1). Atrial systole pumps blood directly into the ventricles through the open tricuspid and mitral valves, filing the ventricles in about 100ms while the ventricles are in the diastole phase. During ventricular systole, increased pressure in the ventricles causes the pulmonary and aortic valves to open, ejecting blood into the pulmonary trunk and the aorta. This process takes place in about 270ms. Lastly, ventricular diastole is characterized by relaxation of the ventricles and closing of the pulmonary and aortic valves to prevent backflow of blood into the heart. This phase of the cardiac cycle lasts about 430ms, making a complete cardiac cycle last about 800ms or 0.8 seconds in the human heart. Both clinically and experimentally, the amount of blood pumped out by the left ventricle to the aorta is also known as the stroke volume, which is normally in the range of 70-80ml in the human heart. Another common measurement of how much blood is pumped out by the left ventricle is the ejection fraction of the heart, reported as a percentage. Reduced ejection fraction below the normal range (<40%) is a common clinical indication of HF. Additionally, a common clinical measurement to describe the health status of the heart is the cardiac output, defined as how much blood is pumped out in one minute (mL/min). Cardiac output can simply be calculated by multiplying stroke volume by heart rate (CO = SV x HR). Depending on your body size, the cardiac output indicates how well the heart is functioning knowing that a normal adult’s heart pumps about 5 liters of blood per minute at rest.

Figure 2.1. The cardiac cycle. Schematic illustration of a complete heartbeat. Atrial systole pumps blood into the ventricles, subsequent ventricular systole then pumps blood out to the outflow tracks (pulmonary and aortic trunks), completing the cardiac cycle. Red arrows indicate oxygenated blood and blue arrows indicate deoxygenated blood. This figure is a modified version of materials from Servier Medical Art (https://smart.servier.com) by Servier licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit https://creativecommons.org/licenses/by/3.0/.

2.1.3 The Cardiac Conduction System The driving force of the cardiac cycle is electrical impulses generated by specialized cells in the right atrium leading to muscle contraction. These specialized groups of muscle-like nervous cells are known as the sinoatrial (SA) and atrioventricular (AV) nodes. Autonomous electrical impulses generated by the nodes control the timing of the heartbeats in a coordinated fashion. The SA node is also known as the pacemaker of the heart, firing electrical impulses at about 60-70 times per minute. When the membrane

potential of the SA reaches its threshold, an action potential is fired that travels from the SA node to the AV node (Figure 2.2). Electrical signals from the AV node are then relayed to both atria, causing both atria to undergo systole and pump blood into the ventricles. As electrical impulses travel from the AV node down the heart to the ventricles via the bundle of His located in the septum separating the two ventricles, they are delayed for approximately a tenth of a second due to a slower electrical conduction rate at the AV node. This 0.1 second delay has great importance in cardiac physiology as it allows the ventricles to be filled with blood before ventricular systole takes place. As electrical impulses travel down the bundle of His, the bundle of His branches out into Purkinje

Bundle SA node of His

AV node Purkinje Fibers Purkinje Fibers

Figure 2.2. The cardiac conduction system. Schematic illustration of the electrical conduction system in the heart. Electrical impulses generated by the SA node travel to the AV node, bundle of His, and Purkinje Fibers to generate coordinated atrial and ventricular contractions. This figure is a modified version of materials from Servier Medical Art (https://smart.servier.com) by Servier licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit https://creativecommons.org/licenses/by/3.0/.

Fibers towards the apex of the heart, extending electrical impulses rapidly to the left and right ventricles. This results in ventricular systole, allowing blood to be pumped out into the pulmonary circuit by the right ventricle and the systemic circuit by the left ventricle.

At the cellular level, rapid electrical propagation between cardiomyocytes throughout the heart is made possible by gap junctions. Cardiomyocytes are interconnected at the subcellular level by tight junctions and gap junctions, creating a continuous mechanical and electrical network of cardiomyocytes. It has been shown that any given cardiomyocyte in the myocardium is connected to an average of 11 adjacent cardiomyocytes at intercalated discs 13. Gap junctions are made of connexin proteins that allow intercellular communication of small molecules and electrical impulses. Connexin 43, a widely studied gap junction protein and the predominant connexin in the heart, is mainly responsible for electrical impulse propagation from excited to non-excited cardiomyocytes. Alterations in the expression levels or localization of connexin 43 often results in cardiac arrhythmias and myopathies in the human heart.

2.1.4 The Myocardium in Context There are more than just cardiomyocytes and pacemaker cells in the heart. A recent study has shown that the adult human heart has three dominant cell populations: cardiomyocytes, fibroblasts, and endothelial cells. Cardiomyocytes account for approximately 30-40% of total cell number in heart while occupying 70-85% of the heart volume 14. While non-cardiomyocyte cell populations such as fibroblasts, endothelial cells, and various progenitor cells only account for a small heart volume fraction, they are essential support cells in the heart that are required for rhythmic contractions of cardiomyocytes. It is estimated that cardiac fibroblasts account for approximately 65% of total cell number in the heart 15. Fibroblasts in the heart are mainly responsible for producing extracellular matrix proteins, however, upon activation in response to biomechanical stress under pathological conditions can lead to excessive fibroblast deposition, resulting in fibrosis in the myocardium. Lastly, the adult human heart is estimated to contain approximately 2-3 billion mature, terminally-differentiated cardiomyocytes 15. These terminally-differentiated cardiomyocytes are lost once they are

damaged and cannot be replaced or regenerated in humans. However, cardiomyocytes can undergo hypertrophy in response to external stimuli such as exercise, pressure overload, and pathological signals. Exercise- and pregnancy-induced hypertrophy are considered a physiological adaptation of the heart to meet the changing demands of the body whereas pathological hypertrophy is usually accompanied by devastating heart diseases such as hypertrophic cardiomyopathy, dilated cardiomyopathy, and heart failure.

2.2 MOLECULAR ASPECTS OF CARDIAC FUNCTION

2.2.1 Cardiomyocyte Organization Cardiomyocytes are composed of bundles of myofibrils made up with myofilaments. The myofibrils have distinct striated patterns as a result of the orderly arrangement of sarcomeres, the individual contractile units within cardiomyocytes. One sarcomere is defined as the region of myofilaments between two Z-lines, specialized regions formed for myofilament attachment. Specifically, cardiac Z-discs are composed of layers of  sarcomeric actinin aligned perpendicular to the myofilaments, cross-linking the ends of the thin actin filaments of adjacent sarcomeres 16. At the core of a sarcomere, a “thick” myosin and “thin” actin filaments are chemically and physically interacting with each other to lengthen or shorten the sarcomere, giving rise to myofibril contraction within cardiomyocytes (Figure 2.3). Proper formation and maintenance of cardiac sarcomeres are therefore important for heart function. Additionally, each myofibril contraction requires a significant amount of ATPs hydrolyzed by the myosin ATPase at the heads of each myosin filament to fuel the sliding motion between the thick and thin myofilaments. For this reason, cardiomyocytes contain a large number of mitochondria to support the ATP demand, occupying approximately 30% of a cardiomyocyte volume alone. On the other hand, the thin filaments are composed of actin, tropomyosin, and troponin. More importantly, the troponin complex (troponin T, C, and I subunits) contains binding sites for a very important secondary messenger, Ca2+. Without Ca2+, actin-myosin binding is inhibited in cardiomyocytes. In the presence of Ca2+, binding of Ca2+ molecules to troponin C reveals the myosin binding site on the actin filaments, permitting myosin-actin binding (Figure 2.3). The resulting interaction leads to the formation of a cross-bridge between

the two filaments. Taken together, proper Ca2+ cycling and tight regulation within cardiomyocytes is crucial to the Ca2+-dependent actin-myosin binding process and therefore muscle contraction in the heart. In fact, it is well-established that dysregulated Ca2+ cycling is a pathological hallmark of HF.

Sarcomere Myofibril

Actin

Myosin

Z-disc In the presence of Ca2+ Tropomyosin

Actin

Troponin complex Myosin

Myosin head

Figure 2.3. Cardiac myofibrils and myofilaments. Schematic illustration of a cardiac myofibril containing sarcomeres between Z-discs (upper panel). The bottom panel shows the binding of cardiac myosin and actin myofilaments in the presence of Ca2+ molecules. This figure is a modified version of materials from Servier Medical Art (https://smart.servier.com) by Servier licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit https://creativecommons.org/licenses/by/3.0/.

2.2.2 Cardiac Excitation-Contraction Coupling Cardiomyocyte action potentials are depolarizations (increases in membrane potential above threshold) of the myocyte membrane regulated by many types of ion channels. Membrane depolarizations are electrical events that drive and coordinate the mechanical events of chamber contractions in the heart. This electric-mechanical process is known as the Excitation-Contraction Coupling (ECC) of the heart. The origin of electrical depolarization begins in the SA node. This depolarizing current then travels to all

cardiomyocytes through the electrical syncytium formed by gap junctions between adjacent cardiomyocytes. The main purpose of ECC is to initiate cardiomyocyte contraction and in order to do that, Ca2+ release into the cytoplasm of cardiomyocytes is required for the formation of cross-bridges between actin and myosin filaments. The heart has 4 main types of Ca2+ channels: 2 L-type (long-lasting) Ca2+ and 2 T-type (transient) Ca2+ channels.

Regulation of Ca2+ release and uptake can be conceived as a 4-step process in the heart. First, depolarization of the cardiomyocyte membrane activates the voltage-dependent L- type Ca2+ channels in the sarcolemma, dihydropyridine receptor (DHPR), resulting in an influx of extracellular Ca2+ into the cytoplasm of cardiomyocytes (Figure 2.4). However, this small influx of Ca2+ is not enough to initiate myofilament contraction. Instead, this influx of Ca2+ triggers an event known as the Ca2+-induced Ca2+ release (CICR) mechanism in cardiomyocytes. CICR takes place at sarcolemmal invaginations known as cardiac t-tubules in cardiomyocytes. These membrane invaginations allow the regulation of Ca2+ signaling to be in close proximity to the contractile machinery in cardiomyocytes. Moreover, DHPR is located in close proximity to the ryanodine receptor (RyR2) in the

Figure 2.4. Overview of cardiac excitation-contraction coupling. Schematic illustration of the excitation-contraction coupling mechanism. The sequence of events leading to cardiac ECC is 1) influx of extracellular Ca2+ via DHPR, 2) CICR mechanism via RyR2, 3) intracellular [Ca2+] surge causes myofibril contraction, 4) reducing intracellular [Ca2+] by SERCA2 (SR/ER Ca2+ ATPase) pumps and NCX (Na+/Ca2+ exchanger). Arrows represent the flow of Ca2+ inside the myocyte. This figure is a modified version of materials from Servier Medical Art (https://smart.servier.com) by Servier licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit https://creativecommons.org/licenses/by/3.0/. sarcoplasmic reticulum membrane. Extracellular Ca2+ influx triggers the release of intracellular Ca2+ stored in the SR lumen, causing a surge in intracellular [Ca2+] by nearly 100-fold (Figure 2.4). These Ca2+ molecules released from the SR then act as an important effector leading to myofibril contraction. Lastly, excessive Ca2+ in the cytoplasm is recycled back to the SR lumen by SERCA2 pumps in the SR membrane as well as NCX exchangers in the plasma membrane, uncoupling cross-bridges between myofilaments and undergo muscle relaxation (Figure 2.4). Cardiac ECC is a tightly regulated process, alterations in Ca2+ cycling efficiency as well as the expression levels of Ca2+ regulating membrane proteins can result in deleterious heart diseases in humans. In fact, a reduced SERCA2 expression and a leaky RyR2 channel have been associated with the development and progression of HF 17, 18.

2.2.3 The Sarcoplasmic Reticulum The endoplasmic reticulum (ER) is the largest membrane-bound organelle in eukaryotic cells. The ER is responsible for many essential cellular processes including protein synthesis, transport, and modification as well as organelle-organelle communication. The ER represents a continuous membrane system spanning from the outer nuclear envelope and turns into a network of interconnected tubules and sheets in the cytoplasm. In the heart, cardiomyocytes have adapted a specialized ER network known as the sarcoplasmic reticulum (SR) to regulate Ca2+ cycling and signaling in the cytoplasm of highly differentiated muscle cells. The main function of the SR in cardiomyocytes is therefore maintaining proper balance between Ca2+ release, storage, and reuptake. The

SR surrounds each muscle myofibril like a water jacket with the tubular SR network surrounding each myofibril in a cuff-like structure and the terminal SR cisternae perpendicular to the myofibrils at t-tubules. Additionally, the junctional-SR is the region of the SR cisternae that is closely tethered to the t-tubule of the plasma membrane. This structural domain contains proteins highly specialized for Ca2+ release including DHPR, RyR2, SERCA2, and phospholamban (PLN).

The formation and maintenance of the longitudinal tubular SR structure are regulated by several highly-conserved classes of membrane proteins including the reticulon (RTN), receptor expression-enhancing protein (REEP), and atlastin (ATL) families of proteins as well as CKAP4/Climp-63 19. Members of the RTN family of proteins possess a core reticulon homology domain (RHD) that has been shown to be essential for high membrane curvature formation for SR/ER tubules 20. The RHD domain is a ~200 amino acids region forming two hairpin integral domains connected by an intervening hydrophilic loop. Moreover, RHD domains have been shown to form immobile oligomers at regions of high curvature in the tubular ER network 21. This suggests that the oligomeric status of RTN proteins likely determines the tubular diameter of ER tubules which is normally in the range of 30-100nm. Moreover, the REEP family of proteins represents the mammalian homologs of the protein Yop1p in yeast. Deletion of Yop1p in yeast has been shown to result in the loss of ER tubules and overexpression of Yop1p can lead to long, unbranched ER tubules 20. Similar to RTNs, members of the REEP family of proteins also possess an integral RHD domain occupying more space in the outer leaflet of the membrane thereby forming high membrane curvature in ER tubules. Interestingly, deletion of both Rtn1p and Yop1p in C. elegans resulted in embryonic lethality, suggesting the importance of an intact ER network during early development 22.

The generation of high membrane curvature ER tubules is only half of the story. In order to form the characteristic polygonal array of ER tubules, membrane fusion of ER tubules and the formation of three-way junctions is required. The formation of three-way junctions is facilitated by members of the ATL family of proteins, a class of dynamin-like GTPases 23, 24. ATLs are large GTPases that interact with RTNs and REEPs to drive homotypic

membrane fusion in a GTP-dependent manner. Upon GTP hydrolysis, ATLs in opposing membranes dimerize to destabilize the membranes and undergo a conformational rearrangement for membrane fusion 25. Specifically, the C-terminus amphipathic domain of ATLs plays an important role in membrane destabilization by thinning the membrane through direct interaction with the lipid bilayers 26, 27. Furthermore, stabilization of three- way junctions post-membrane fusion is mediated by the protein Lunapark (Lnp). Lnp has been shown to preferentially localize to three-way junctions and is thought to antagonistically regulate ATLs. Deletion of Lnp in mammalian cells results in reticulated ER network and unstable three-way junctions, leading to rapid ring closure 28, 29. Taken together, proper formation and maintenance of the tubular ER network requires a fine balance between RTNs, REEPs, ATLs, and Lnp. Although Lnp is not required for ER tubule formation, Lnp deficient cells have transient three-way junctions followed by a tubule-to-sheet conversion of the ER 30, 29. However, the precise role of RTN, REEP, ATL, and Lnp and the physiological implications of their interactions in SR tubule formation in highly differentiated cells such as cardiomyocytes remain poorly understood as ER structural biology and function studies have mainly focused on yeast models and tissue culture cells. Lastly, the integrity of the SR/ER network is extremely vulnerable to functional impairment. Defects in the junctional SR/ER space in cardiomyocytes can result in blunted ECC and decreased contractility of cardiac muscle. Similarly, defects in SR/ER tubules in the longitudinal SR network paves the way for impaired Ca2+ regulation and storage, ultimately results in HF.

Interestingly, most of longitudinal SR proteins in cardiomyocytes appear to be localized in alignment with the Z-disc region. One explanation could be that an increased density of SR membranes is observed and therefore its associated proteins at cardiac Z-discs. In alignment with this theory, a recent study utilizing super-resolution microscopy revealed what thought to be conventional ER sheets are composed of highly convoluted, dense ER tubules, termed ER matrices 31.

Taken together, the SR consists of structurally distinct microdomains that are functionally unique than non-muscle cells. The delineation separating the smooth vs rough ER is less

obvious in muscle cells. Immunofluorescence studies of rough ER proteins including BiP and PDI have been shown to localize to both longitudinal SR and the Z-disc region in muscle cells 32. One explanation could be that locally-regulated protein synthesis near the myofibrils in the longitudinal SR as well as synthesis of Z-disc proteins in the junctional SR are required for the tightly regulated process of muscle contraction. Additionally, cytoskeletal interactions have been shown to play an important role in SR organization. Specifically, the interaction between ATL and spastin, a microtubule-severing ATPase, is required for proper maintenance of the ER 33. Therefore, a unifying model for SR organization in mature cardiomyocytes requires the SR network to be formed and maintained by structural proteins such as RTNs, REEPs, ATLs, and likely interactions with the cytoskeleton to maintain a highly organized yet dynamic SR network in muscle cells. However, the precise mechanisms defining the SR development and organization remains to be elucidated especially in mature muscle cells with a highly differentiated SR network.

The SR/ER is a dynamic organelle that is constantly adapting to a variety of physiological conditions in cardiomyocytes. In eukaryotic cells, the close proximity of the SR/ER to the nucleus allows the SR/ER to feedback information regarding protein synthesis and protein quality control. Increased demands on the SR/ER protein production and folding machinery in response to environmental or pathological stimuli can result in an accumulation of misfolded or unfolded proteins in the SR/ER lumen, a condition known as ER stress 34. In an attempt to restore SR/ER homeostasis, a signaling cascade known as the unfolded protein response (UPR) is activated in stressed cells. The UPR consists of three branches initiated by three separate ER transmembrane proteins: inositol requiring 1 (IRE1), PKR-like ER kinase (PERK), and activating transcription factor 6 (ATF6) that act as cellular sensors and sense the protein folding status and capacity in the SR/ER lumen (Figure 2.5). Under unstressed conditions, the three master regulators of UPR are inactivated by GRp78; its dissociation upon ER stress activates the signaling cascade in stressed cells. IRE1, an ER transmembrane kinase, is activated upon ER stress through homodimerization and autophosphorylation. Activated IRE1 results in the splicing of the x-box binding protein 1 (XBP1) mRNA and activation of XBP1 in turn

upregulates the expression of molecular chaperones and protein processing enzymes to mitigate ER stress (Figure 2.5). Similarly, PERK is also an ER transmembrane kinase that oligomerizes and autophosphorylates upon ER stress. Activated PERK then phosphorylates the  subunit of the eukaryotic initiation factor 2 (eIf2), resulting in translational attenuation to reduce ER workload by preventing the formation of ribosomal complexes. In addition, p-eIf2 activates the transcription factor ATF4, an upstream regulator of several UPR genes including C/EBP homologous protein (CHOP) (Figure 2.5).

Figure 2.5. Overview of the unfolded protein response (UPR). Schematic illustration of the UPR signaling cascade upon ER stress. The UPR is initiated by three major branches of consequent downstream responses leading to activation of UPR genes involved in protein folding (chaperones), protein synthesis (translational arrest), and protein degradation (proteasomes) to restore ER homeostasis. This figure is a modified version of materials from Servier Medical Art (https://smart.servier.com) by Servier licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit https://creativecommons.org/licenses/by/3.0/.

The third branch of the UPR is governed by the transcription factor ATF6. Upon ER stress, ATF6 gets translocated to the Golgi apparatus where it gets cleaved by proteases and becomes activated. The activated form of ATF6 is then translocated to the nucleus to upregulate UPR genes involved in protein folding, processing, and degradation. Failure to restore ER homeostasis or prolonged ER stress can result in UPR-induced cell death, a well-established pathogenic factor for many human diseases. Upregulation of the transcription factor CHOP leads to activation of caspase 12, a SR/ER specific protease enzyme. Increased cleavage of caspase 12 in ER stress-induced apoptosis is considered a maladaptive response and can be selectively activated by ER specific proapoptotic proteins, Bax and Bak, independent of the canonical mitochondrial apoptosis pathway 35. However, the underlying mechanisms responsible for the switch from adaptive to maladaptive UPR remains poorly understood. Interestingly, caspase 12 exists as a truncated, catalytically inactive form in most humans, cell death under ER stress may likely depend on the core mitochondrial-induced apoptosis pathway regulated by the B cell lymphoma 2 (Bcl2) family of proteins in humans 34, 36. Nonetheless, extensive investigations into heart diseases have highlighted the SR/ER as a focal site for the initiation and progression of many heart diseases. ER stress in the heart has emerged as an important contributing factor to cardiomyocyte apoptosis in many heart diseases, contributing to the pathogenesis of HF 37, 38, 39.

2.2.4 Proteomics of the Heart

Sections 2.2.4 and 2.2.5 of the chapter are a modified version of a manuscript that was published in American Journal of Physiology – Heart and Circulatory Physiology. Lee et al. Membrane proteomic profiling of the heart: Past, present and future. Am J Physiol Heart Circ Physiol 320(1): H417-H423 (2021).

Permission to reproduce sections was obtained from The American Physiological Society (Appendix II).

Understanding the physiology and pathophysiology of heart diseases requires us to first identify the underlying molecular mechanisms of cardiac development and causes of cardiac dysfunction. Advances in proteomic techniques have allowed for an assessment

of global alterations in protein expression under various physiological and pathological conditions. Global proteomic analyses of heart disease animal models and patient tissue explants have yielded novel insights into the cellular mechanisms of cardiac dysfunction 40, 41. However, MS-based analyses of whole cardiac tissue samples pose a unique limitation as the resulting datasets represent a combination of all cell types and include a high content of extracellular matrix in the heart. Likewise, overrepresentation of the muscle contractile apparatus is a unique signature of proteomics studies of cardiac samples, contributing to the natural limits of identifying low abundance proteins. In light of this, sample preparation and subcellular fractionation strategies including dilution of interfering substances and ultrafiltration to remove competing molecules and/or detergents in the sample have advanced the identification of the proteomic make-up of individual cellular organelles with reduced sample complexity 42-45. Particularly, a non- ionic MS-compatible detergent to solubilize membrane proteins in combination with lipid removal has been critical for optimized detection. Several recent reviews have put together a ‘decision-tree’ including detailed protocols for maximizing intact membrane protein sample preparation and analysis for general MS 46, 47.

In the last few decades, cardiovascular proteomics has reached a level of maturity where reliable, large-scale datasets of heart tissues can be generated (Figure 2.6). Thanks to the advances in the MS technologies, including matrix assisted laser desorption/ionization and the application of isotopic labeling techniques (iTRAQ and SILAC), the field of cardiovascular proteomics as well as our understanding of the heart proteome have evolved rapidly in the 2000s (Figure 2.6). Pathological effects on the heart (as well as other organs) result from changes in the totality of a system where genetic, epigenetic, gene expression and processing, and protein levels play complex interconnected roles 48. However, it is a well-established concept that gene and protein expression are not the only factors responsible for phenotype determination. The growing list of over 300 different types of post translational modifications (PTMs) currently known appears to provide the highest level of heterogeneity and diversity at the molecular and potentially phenotypic levels 49. Of particular recent interest, the use of MS for post-translational modification analysis has greatly increased our understanding of signaling perturbations

in the diseased heart whereby ~3,600 human phosphoproteins have been recently identified in patients with hypertrophic cardiomyopathy 50. Assuming that even a minor proportion of these modifications have functional or phenotypic relevance by value of their presence or absence on individual proteins alone, the potential diversity is staggering even when not considering the heterogeneity that can occur within individual PTM subgroups, such as N-linked glycosylation and ubiquitination which can vary based on composition, size, and branching patterns on different instances of the same protein 51. It has long been established that other PTMs exert influence and play crucial roles in signaling cascades that determine phenotype under “normal” conditions and pathological disturbances. With such a wide spectrum of PTMs, there is considerable interest concerning the crucial roles they play in this complex web of functional interactions, and PTM cross-talk has been well established as an effector of changes in signaling pathways 52. Several recent publications describe in-depth modern MS-based proteomic techniques to identify and quantify a wide variety of PTMs, including several PTM-focused enrichment strategies in combination with LC-MS/MS, resulting in the identification of PTMs associated with different categories of heart disease 53-55. However, the technical complexity and the low-throughput associated with these analyses are limiting to the field, which is presently only at its nascent stages.

Figure 2.6. Evolution of cardiovascular proteomics. Schematic timeline of major milestones in the field of cardiovascular proteomics.

The complexity of humans as an organism far exceeds the 20,000 genes identified in the human genome 56, 57. Genetic- and protein-editing processes such as mRNA splicing and post-translational protein modifications greatly increase the functional complexity of a gene in the human genome. Large-scale analyses of various model organisms and patient samples have been used to identify disease signatures at the genomic, transcriptomic, and proteomic levels. Although genomic and transcriptomic analyses have been instrumental in providing mechanistic insight to pathophysiology, proteomic studies provide a unique advantage in directly assessing the functioning agents of cellular and molecular processes. Additionally, while still unclear at this point as to the extent of correlation, a clear global correlation between mRNA expression level and the abundance of its associated protein product limits the reliability of transcriptomic studies in assessing direct functional alterations 58. Additionally, the majority of current drugs on the market target effectors at the protein level, with membrane proteins contributing to the majority of all of these protein targets. Proteomic analyses have, therefore, been useful as a global systems-biology approach to reveal novel signaling proteins and study pathway interactions in diseased cardiovascular models and in clinical settings.

The concept of protein separation and proteomic profiling of a sample started with two- dimensional gel electrophoresis (2-DE) more than three decades ago 59. Pioneering 2- DE-based proteomic studies of the heart identified proteins in the scale of tens 60, 61 (Figure 2.7). Later in the 2010s, studies using 2-DE were able to separate approximately 1,000 human and 2,000 mouse proteins from whole cell and tissue extracts (Figure 2.7). However, the diversity of cell populations in the heart represents a major limitation for a complete cardiac proteome coverage using 2-DE due to poor protein separation.

Advances in protein separation technologies such as the development of liquid chromatography (LC) combined with MS analysis have significantly increased the capacity of MS-based proteomics studies. The growing technical capacity of LC-MS/MS studies has allowed the total protein coverage in modern proteomics studies to approach the estimated maximum number of proteins in the human and mouse heart (Figure 2.7)

62. Developments of high-performance LC allowed for dynamic proteome profiling of complex samples and the identification of proteins on a global level. This so-called shotgun proteomics approach involves initial multidimensional protein identification and separation analysis of enzymatically digested peptides from complex tissue samples followed by automated database searching 63. This method has served as a powerful strategy and has greatly improved overall data throughput and reproducibility. It has also enabled the relative quantification of thousands of proteins from myocardial tissues as well as the detection of lower abundance proteins and transcriptional regulators. However, shotgun proteomics studies of patient tissue explants simply demonstrate a snapshot of the proteome at a given time point. The lack of real-time temporal proteome profiling in heart disease patients remains an obstacle to fully understanding the dynamic alterations of the heart proteome in disease. Recently, a study in 2017 led by Doll et al. revealed that the human heart proteome consists of over 10,000 distinct proteins, representing the deepest proteomics study of the human heart to date 6. Doll et al. further demonstrated that 3,000 proteins were estimated to be cardiomyocyte-specific and are likely important for heart muscle performance and capacity. Similarly, Lau et al. revealed the deepest proteome of the mouse heart to date by identifying a total of 8,064 unique proteins across 6 mouse strains as well as global alterations in their expression during pathological cardiac hypertrophy 58. Although advances in proteomics technologies have successfully detected the greatest number of proteins in the heart, effective means to utilize these big proteomics datasets to further delineate sub-proteomes of the heart as well as to investigate proteins of interest at low copy numbers have emerged as new challenges in the field of cardiac proteomics 66.

Figure 2.7. Trends in the capacity of proteomics studies of the human and mouse heart. Total protein coverage of proteomic studies of the (a) human and (b) mouse heart from 1990-2020. The dashed lines represent the estimated maximum number of cardiac proteins based on single nucleus RNA sequencing analysis of >280,000 human cardiac nuclei 64 and 20,000 mouse cardiac nuclei 65.

Integration of the output of proteomics studies with bioinformatics databases often yields a comprehensive interpretation of the data in a systematic manner. A recent study took advantage of the intersection of proteomic data with genomic datasets from genome wide association studies to identify novel candidate proteins associated with mitral valve prolapse, a condition where the mitral valve does not close properly during muscle

contractions 44. Furthermore, a multi-omic integration of datasets has identified approximately 53,000 binary protein interactions in a recent study of human protein interactome 67. This study is an example of a comprehensive study approach that not only enables systematic studies of the association between genomic variation and altered protein expression and cellular function, but also allows for investigations of functional protein-protein interactions within particular physiological and pathological cellular contexts, including the heart. In the field of cardiac proteomics, comparative analysis utilizing a combination of datasets will generate a more comprehensive, robust understanding of the molecular fingerprints of heart disease and provide unique inferences on disease mechanisms. Similarly, bioinformatic analyses such as gene ontology and pathway/process network enrichment analyses applied to proteomic data has allowed for the identification of protein interactions and signaling pathways that are differentially regulated in healthy versus diseased myocardial tissues 68, 69. Taken together, in-depth, quantitative proteomics analysis of the proteomic make-up of the heart and its sub-proteomes is critical in identifying novel therapeutic targets and provides new insights into cardiac dysfunction in human heart diseases.

2.2.5 Cardiac Membrane Proteins Cardiac membrane proteins are essential for the regulation of spontaneous and synchronous contractions of cardiomyocytes. The cardiomyocyte cell membrane, also known as the sarcolemma, has two functionally distinct niche domains: the intercalated discs that are responsible for the propagation of contractile signaling between cardiomyocytes, and membrane invaginations knowns as the t-tubules that effectively communicate electrical signals throughout the myocytes which initiate synchronous cardiac excitation-contraction coupling. Orderly cardiac electrical communication and the synchronous ventricular contraction is tightly regulated by several classes of membrane proteins in cardiomyocytes including ion channels, gap junction channels, and scaffolding proteins linking the contractile apparatus to inter- and intracellular signal transmission. In addition, due to their exposed position in the cell, numerous membrane proteins, such as G-protein coupled receptors (GPCRs) for instance, are effective therapeutic targets to reduce overall risk and mortality rate for heart disease 4.

To date, many key regulators of heart function as well as heart disease-causing mutations have been identified for different clinical categories of cardiomyopathies (HCM, DCM, ARVC) 3, however, the development of effective therapies aimed at halting or reversing the progression of heart disease remains challenging due to a limited number of functional target proteins available. In many ways, these deficiencies in new therapeutic developments result from a lack of understanding of the cardiac membrane proteome and its constituents. For the past decade, profiling of critical cardiac membrane proteins via proteomics approaches have emerged as a popular and reliable methodology to identify previously underappreciated membrane proteins in the heart. Specifically, the ability to identify organelle-specific membrane proteomes in cardiomyocytes has allowed for the development of new therapies to target protein trafficking and function of cardiac membrane proteins at the subcellular level 70. Proteomic profiling of cardiomyocyte mitochondria perhaps represents the most studied example of organelle proteomics, ultimately identifying a sophisticated level of protein regulation as revealed through complex post-translational modifications of cardiac mitochondrial fission and fusion proteins 71.

The delineation of the sub-membrane proteome of cardiomyocytes was difficult due to their complex biochemical nature and low abundance. Given their great importance in governing synchronous beating of the heart, the identification and enrichment of membrane and membrane-associated proteins is perhaps one of the most challenging yet seminal tasks in molecular biology and biochemistry 7. Nonetheless, recent advancements in proteomic technologies, along with advanced informatics, have allowed high-resolution MS-based techniques to reach greater depths of the proteome, while successfully identifying increased numbers of low abundance membrane proteins 63. Detailed characterization of these newly identified, cardiomyocyte specific membrane proteins could provide breakthrough understanding of cardiomyocyte physiology, signaling networks, and protein-protein interactions, leading to potential diagnostic and therapeutic targets.

The functions of cardiac membrane proteins are diverse, in addition to regulating ion channel activity, these membrane-bound proteins also help provide structural integrity to the functionally distinct microdomains within cardiomyocytes and act as anchoring sites for signaling receptors. Since cardiomyocytes are subjected to high contractile forces, an enrichment of structural membrane proteins to maintain the integrity of specialized microdomains such as cardiac t-tubules represents a characteristic subgroup of the myocyte membrane proteome 4, 72. For instance, a recent proteomics study aiming to create a blueprint of the cardiomyocyte ICD proteome demonstrated a dense membrane support network consisting of desmocollin, desmoglein, and desmoplakin to connect adjacent cardiomyocytes at the ICD 73; where mutations in these proteins at the ICD have been linked to ischemic, dilated and arrhythmogenic cardiomyopathies 74, 75.

Studies of membrane proteins have been challenging over the past decades due to their hydrophobic nature and relative low abundance 7. To date, the majority of cardiac biochemical studies have focused on abundant soluble proteins with nearly 100 genes identified to be associated with familial cardiomyopathy and HF 76, however, the underlying molecular mechanisms remain poorly understood in many cases. The complex organization of cardiomyocytes suggests there are likely many more proteins, including membrane peripheral proteins that play a yet to be fully appreciated role in regulating cardiac function. Bridging the knowledge gap between cardiomyocyte membrane proteins and their contribution to cardiac dysfunction in disease may be the key to understanding the complex molecular mechanisms underlying heart diseases.

2.3 MODEL ORGANISMS IN CARDIOVASCULAR BIOLOGY

2.3.1 Introduction to Model Organisms A model organism for research investigations is a preclinical animal model that is widely used to study and understand biological phenomena and disease mechanisms. This is made possible by the conservation of core developmental pathways and genetic materials over the course of evolution. An ideal model organism thereby shares similar biology and physiology as humans as the main objective is to apply translatable

discoveries made in the model organism to understand human physiology and disease. However, we need to keep in mind that no model organism is perfect and often the underlying research context represents the determining factor when selecting a suitable model organism for research studies. Additionally, common biological traits among model organisms such as size, short generation time, ease of genetic manipulation, and accessibility favor the production of high throughput in vivo data. The most widely used model organisms in cardiovascular research include mice, rats, zebrafish, and fruit flies. Here I will mainly compare two model organisms, mice and zebrafish, to available human research data, all of which I have used extensively to complete my thesis work. In this section, I will outline the knowledge of utilizing laboratory mice and zebrafish as a model organism in heart research as well as their advantages and limitations in studying heart function and heart development.

The mouse (Mus musculus) and zebrafish (Danio rerio) have emerged as a popular vertebrate model organism for studies of heart development and heart function over the past few decades. Laboratory mice by far represent the most commonly used model organism for cardiovascular research with approximately 85% similarity between the mouse and human genome. The mouse genome has been fully sequenced and every gene in the human genome has a counterpart in the mouse. Additionally, the ease to breed and care, larger body and litter size, and fast reproductive maturity (~8 weeks) are all popular traits for a wide variety of scientific research 77. Moreover, its relatively longer lifespan (average ~3 years) compared to other model organisms have allowed the study of aging and chronic human diseases including neurodegenerative and cardiovascular diseases as well as obesity and hypertension in preclinical mouse models. An important advantage of utilizing mice for heart research is the ability to generate and maintain genetic models in a relatively short period of time due to their short gestational period (21 days) at a low cost 78. To date, an approximately 8000 strains of mice have been reported in the Mouse Genome Informatics database and are available to facilitate the study of human biology and disease. Interestingly, unlike other mammals including humans, neonatal mouse hearts have been reported to possess transient regenerative capacity within 7 days after birth in response to myocardial injury 79.

On the other hand, zebrafish are tiny freshwater fish (2.5 inches in length) with five horizontal blue stripes on the side of their body. Similar to the laboratory mice, zebrafish and humans share an 82% similarity throughout the course of evolution and the zebrafish genome has been shown to be easily manipulated to mimic human diseases 80, 81. Moreover, zebrafish are known for their high degree of conserved integrative physiology, in some cases more similar to human physiology than mouse models. Specifically, cardiovascular physiology is highly conserved between humans and zebrafish, with many human heart diseases well phenocopied in zebrafish models 82. An example is that human cardiac electrophysiology is well-recapitulated in zebrafish 83, 84, better representing human cardiac electrophysiology than its rodent counterpart even though zebrafish are phylogenetically more distant from humans than rodents. More importantly, the zebrafish possesses unique advantageous characteristics that distinguish them from other model organisms for heart research. First, its transparency during development allows non- invasive optical imaging in real-time throughout development as well as in vivo functional analysis of heart development. Second, its rapid ex utero development and high fertility with all major organs appearing within 36 hours of development allows for high throughput screening and analysis of genetic mutations as well as drug discovery and development. More specifically, the zebrafish model adds an additional two layers of advantages to heart research studies. First, zebrafish are known to be capable of regenerating its heart after damage during the larval stages 85, 86. Second, zebrafish embryos with defective cardiovascular system can survive by acquiring oxygen diffused from water, allowing in vivo drug screening and the development of new therapeutic strategies for many cardiovascular diseases. Despite the widespread use of mice and zebrafish for heart research, it is important to keep in mind that the small body size and short lifespan of small model organisms prevent the full recapitulation of human cardiac biology and physiology. Scientific findings from animal studies therefore must be interpreted carefully as major differences exist between model organisms and humans; research animals and humans have evolved in and become adapted to different environments 78.

2.3.2 Heart Development Heart development is a complex, multistep process and the regulation of heart development is highly conserved throughout evolution. Heart development begins with specification of cardiac progenitor cells in the mesoderm that go on to form bilateral cords knowns as the endocardial tubes. A main event in heart development is the migration of the two endocardial tubes containing heart precursors to the midline to form a beating heart tube. This process takes place around 22 days after fertilization in humans, 8 days post coitum (dpc) in mice, and 19 hours post fertilization in zebrafish (dpf) (Table 2.1) 87, 88. Unlike humans, the zebrafish heart is comprised of a single atrium and ventricle. In zebrafish, the development of the cardiovascular system is very fast, a linear beating heart tube is formed within 24hpf followed by the commencement of systemic circulation. The heart is the first functional organ to develop in all mammals and maintains the autonomous beating mechanism since 21dpf while undergoing complex steps to support further development and growth. The linear beating heart tube quickly specializes into five different regions to form the great arteries, outflow tracks, atria, and ventricles, from head to tail. The beating heart tube will continue to specialize to drive heart morphogenesis that entails regionalized changes including the differentiation of the outer curvature that eventually makes up the working myocardium and the inner curvature that turns into the atrioventricular canals. In summary, this seemingly simple tube undergoes complex morphological and signaling modifications including heart looping and chamber formation followed by chamber septation and valve formation to form a functionally coordinated heart to pump blood throughout the body 89.

Elaboration of the linear heart tube by adding on different chambers and septation through conserved genetic pathways not only allows unidirectional blood flow, but also proper heart muscle formation and maturation. Before then, a S-shaped configuration with very characteristic bending motions in rightward direction known as cardiac looping takes place at 23dpf in humans, 8.5dpc in mice, and 33hpf in the zebrafish heart (Table 2.1). In the zebrafish heart, this process converts the linear heart tube to a complex structure, positioning the ventricle to the right of the atrium. Subsequently, a network of transcriptional factors including Nkx2.5, Gatas, Mef2c, Tbxs, and Hand form regulatory

complexes with one another that are required to make mature heart muscle cells as well as chamber identities 90.

Finally, formation of the ventricles takes place via the process known as ballooning. The apical part of the ventricle balloons off the heart tube to form the future apex and base of the heart. Subsequent trabeculations of this ventricular ring give rise to the characteristic morphology of the ventricular muscle walls. On the other hand, formation of the atria requires formation of the lungs along with ballooning of atrial outpouchings off the heart tube. Subsequently, chamber septation is achieved through the development of endocardial cushions at atrioventricular junctions in addition to the development of atrial and ventricular septa 91. Subsequent valve formation involves epithelial-mesenchymal transitions (EMT) of endocardial cells, a calcineurin/Nfat-regulated signaling communication between myocardial cells and endothelial cells, to form endocardial cushions that drive valve specification and chamber septation 92. This process takes place at around 28dpf in humans, and at a comparable stage of development, 9.5dpc in mice and 48hpf in zebrafish.

Although the process of heart development might be simplified in the mouse and zebrafish, but the steps involved are incredibly conserved throughout evolution, allowing comparative studies in these model organisms to further our knowledge in the regulation of human heart development. Finally, the human adult heart is around 12cm in length and roughly the size of a large fist. The mouse and zebrafish heart grows to approximately 8- 9mm and 1-1.5mm in length, respectively. While species-specific differences clearly exist in heart development and maturation, each organism has its strengths, for instance, mice have a 10x faster heart rate than humans, and for zebrafish, it is definitely their capability to regenerate its heart after damage.

Table 2.1. Timeline of major cardiac events during heart development in the human, mouse, and zebrafish heart. dpc: days post coitum; hpf: hours post fertilization. (adopted and modified from 93, 94)

Cardiac Events Human Mouse Zebrafish Migration of CPCs to 15-16 days 7 dpc 5.5 hpf the midline Formation of a single 22 days 8 dpc 19 hpf heart tube Linear heart tube 23 days 8.5 dpc 22 hpf starts beating Cardiac looping starts 23 days 8.5 dpc 33 hpf Formation of 28 days 9.5 dpc 48 hpf endocardial cushions Cardiac regenerative None reported Within 7 days Lifelong capacity

2.3.3 Heart Function Differences in heart function across different animal models have direct impact on the translational value and generalizability of experimental outcomes. Among all anatomical and physiological differences, the most pronounced is size and body weight. An average adult human heart weighs approximately 250-300g, compared to 0.2g in the adult mouse heart 95. Functional differences follow size differences, for instance, the resting human heart beats around 60-80 times per minute whereas the mouse heart beats 10x faster at approximately 600 times per minute. To facilitate rapid heart rate, the mouse heart cycles more intracellular Ca2+ reserve than humans. The zebrafish heart has been reported to have a resting heart rate around 120 beats per minute 96 (Table 2.2). Anatomically, the human heart rests on the diaphragm in the pericardial cavity, resulting in an overall pyramidal shape. In contrast, the adult heart in four-legged mice beats more flexibly in the pericardial cavity, resulting in an overall ellipsoidal shape. In zebrafish, the functional heart consists of only a single atrium and ventricle looped to the right, resulting in a S- shaped heart tube. Nevertheless, ventricular muscle trabeculations can be found in the post-natal human, mouse, and zebrafish heart. Interestingly, the SA node in the mouse

heart is located in the superior vena cava immediate to the right atrium rather than in the atrium in the human heart 97. At the subcellular level, it has been shown that mice have a higher volume of mitochondria and a denser myofibril network in the myocardium than humans 98. Functionally, an average healthy human heart has an ejection fraction between 50-70% with a stroke volume around 60-100ml, resulting in a cardiac output ranging from 4-8 litres per minute (Table 2.2). Comparatively, common laboratory mouse strains show a normal EF range of 53-55% with a normal stroke volume around 24-40l and a normal cardiac output ranging from 8-16mL/min 99, 100. In vivo cardiac functional assessment in zebrafish can be difficult due to its small size, however, available data from non-invasive echocardiographic studies show that the zebrafish heart has a normal ejection fraction of 30-35% with normal cardiac output ranging from 40-50l/min 101. These measurements show that the human heart is built to withstand great pressure due to bigger body size and to overcome the greater vascular resistance, inherent lower venous pressure, and greater pull of gravity observed in humans.

From an electrophysiological perspective, the cardiomyocyte action potential in zebrafish is more similar to that of the human cardiomyocytes, both consisting a plateau phase triggered by Ca2+ and K+ currents. The mouse cardiomyocyte action potential on the other hand, has significantly shorter intervals. Interestingly, zebrafish cardiomyocytes do not have cardiac t-tubules present, instead, a denser presence of L-type Ca2+ channel is found at the sarcolemma 102. Taken together, structural and functional differences observed among the human, mouse, and zebrafish heart are predominately diverse due to the fundamental differences in body mass, metabolism, and heart rate. The challenge of a comprehensive, translational understanding of the cardiovascular system at any scale remains an important obstacle in animal models. These differences should therefore be taken into consideration when analyzing experimental data to answer the question under study.

Table 2.2. Cardiac echocardiographic parameters in the human, mouse, and zebrafish heart.

Cardiac Properties Human Mouse Zebrafish Heart rate (bpm) 60-80 500-600 120-130 Ejection fraction (%) 50-70 53-55 30-35 Total blood volume 5000 1.5-2.0 0.02 (mL) Stroke volume 60-100 0.024-0.040 0.0003-0.0004 (mL/beat) Cardiac output 4000-8000 8-16 0.04-0.05 (mL/min) Mean ventricular wall 10-11 0.7-0.8 A few cell layers thick thickness (mm)

2.3.4 Modeling Human Heart Diseases and Limitations Surgical and pharmacological manipulations of the cardiovascular system have allowed scientists to mimic pathological conditions of the heart. For instance, ligation of the left anterior descending (LAD) coronary artery in mice has been widely used to mimic human myocardial infarction 103. This surgical intervention deprives local nutrients and oxygen supply to cardiomyocytes near the ligation site, resulting in cardiomyocyte loss, reduced ventricular ejection fraction, and cardiac output. Another common surgical intervention performed in mice is the transverse aortic constriction (TAC) procedure. TAC is a well- established procedure in rodents to induce pressure overload-mediated cardiac hypertrophy and HF by tightening the transverse aorta 104. This procedure represents a great model to study the development and progression of HF as the initial stages mimic the compensatory hypertrophy of the heart observed in HF patients leading to eventual cardiac dilatation and heart failure. Lastly, pharmacological interventions such as the treatment of doxorubicin, a cytotoxic nucleic acid chelator, has been used to induce non- ischemic cardiomyopathy in mice 105. In zebrafish, while surgical interventions to model heart disease is nearly impossible due to its tiny heart size, pharmacological interventions and genetic engineering approaches have been widely used to create zebrafish heart

disease models. For instance, individual treatment of aristolochic acid and verapamil have been shown to induce cardiac hypertrophy and HF in zebrafish 106, 107. Additionally, doxorubicin treatment has also been shown to induce HF in zebrafish through the inhibition of DNA replication 108. However, despite its closer phylogenetic distance to humans, zebrafish may not be an ideal model for ischemic cardiomyopathy as zebrafish embryos have the ability to obtain oxygen through passive diffusion in water as well as their ability to regenerate lost myocardium after injury. Table 2.3 below summarizes common interventions and manipulations of the cardiovascular system in mice and zebrafish to model human heart diseases.

Furthermore, advances in gene therapy technologies have allowed us to generate knock- in and knock-out transgenic animal strains to study human heart diseases in pathological cardiovascular conditions 78. Specifically, genetic deletion of the muscle LIM protein (MLP), an actin-associated cytoskeletal protein, has been used to model the development and progression of human DCM. MLP deficient mice develop progressive disorganization of the cardiomyocyte cytoarchitecture, resulting in dilatation of the myocardium and cardiac dysfunction 109. Interestingly, overexpression of a SR Ca2+ binding protein, casequestrin, has been shown to induce rapid ventricular hypertrophy and increased muscle mass in mice 110. Likewise, another transgenic mouse model developed to mimic human hypertrophic cardiomyopathy is the deletion of cardiac myosin binding protein c (cMyBP-C). cMyBP-C deficient mice show severely weakened cardiac muscle with impaired contractility 111. Lastly, a recent study showed that simultaneous metabolic and hypertensive stress induction via a combined treatment of high-fat diet and Nω-nitro-L- arginine methyl ester (L-NAME) in mice phenocopied the cardiovascular features of human HFpEF 112.

Reverse genetic studies in zebrafish have also been successful in generating several heart disease models. For instance, one zebrafish mutant line with non-contractile heart phenotype is generated from silencing the silent heart (sih) gene 113. This mutant line prevents the production of cardiac contractile thin filaments, resulting in a severe DCM phenotype. Similarly, loss of nexilin has also been shown to result in DCM in zebrafish

due to impaired cardiac Z-discs 114. Further reverse genetic studies in zebrafish have provided direct evidence of DCM phenotypes in -sarcoglycan (sarc)-, lamin A/C (lam)-, muscle LIM protein (mlp)-, desmin (des)-, troponin t (ctnt2)-, and tropomyosin (tpm4)- deficient zebrafish embryos and adult fish 115. Although these genetically modified disease models have allowed scientists to follow the development and progression of human heart diseases at an accelerated pace, however, it is important to note that these animal models do not fully recapitulate all disease phenotypes due to diverse and unknown etiologies of HF in many cases.

Nevertheless, advances in scientific knowledge and techniques have allowed scientists to circumvent many biological challenges. For instance, tomaxifin-dependent transgene expression in mice using the Cre-recombinase/LoxP system has proven to be invaluable and allow temporal and spatial regulation of gene expression to study tissue-specific consequences of genetic manipulations 116. Transgenesis in zebrafish using the Tol2 system has allowed scientists to fluorescently label a gene of interest; a well-known example of this is the cmlc2-myl7:GFP line that specifically labels cardiomyocytes in both heart chambers in zebrafish 117. Besides in vivo approaches, recent refinement of the isolation and culture of functional adult mouse cardiomyocytes has allowed scientists to focus on cardiomyocyte-specific consequences in in vitro single cell cultures and remove potential confounding factors in tissue-based studies such as the contribution of non- cardiomyocyte cell populations in the heart 118.

Obviously, both mice and zebrafish are not a universally ideal model organism, certain limitations must be taken into account when assessing the translational aspects of mouse and zebrafish experiments. It is important to note that mice are evolutionarily distant from humans and significant differences between mouse and human immune responses have been reported. Specifically, known discrepancies in the differential expression of leukocyte subsets as well as T cell and B cell pathway components have been previously documented 119.

Table 2.3. Common interventions and genetic manipulations of the cardiovascular system to model human heart diseases.

Mouse Zebrafish Ischemic Non-ischemic Ischemic Non-ischemic Cardiomyopathy Cardiomyopathy Cardiomyopathy Cardiomyopathy Surgical Interventions Surgical Interventions LAD ligation TAC n/a n/a

Pharmacological Interventions Pharmacological Interventions n/a Doxorubucin, High-fat n/a Aristolochic acid, diet + L-NAME Verapamil, Doxorubicin

Genetic Manipulations Genetic Manipulations n/a MLP deletion, n/a Depletion of sih, Casequestrin nexilin, sarc, lam, overexpression, mlp, des, and ctnt2 cMyBP-C deletion

These immunological differences are further magnified in complex, multifactorial human disease models such as cancer and atherosclerosis 120, 121. Nevertheless, these confounding factors can be minimized by increasing sample size and using outbred mouse strains to include genetic variability in experimental design. Similarly, it has been reported that the zebrafish cardiovascular system is naturally more susceptible to environmental and generic perturbations, resulting in generally worse cardiac phenotypes than those observed in the mouse models 122. Therefore, careful interpretations of zebrafish data are essential to distinguish specific phenotypes from off-target effects. Further, it was reported that 82% of human disease causing mendelian genes identified are related to at least one orthologue in the zebrafish genome 123, suggesting that functional redundancy likely exists in zebrafish, favoring a milder pathological phenotype. Lastly, the lack of standard diet and availability of biological reagents such as antibodies limit the specificity of antibody-based experiments in zebrafish models.

2.4 HEART FAILURE

2.4.1 Heart Failure 101 In Canada, HF is the most common reason for inpatient hospitalization as well as the most expensive healthcare problem 124. Clinically, HF is a chronic disease characterized by dysfunction of the myocardium, manifested by either impaired ventricular filling or reduced ejection fraction with an average survival rate of 2.1 years and a ten-year mortality rate of 93.6% 125, yet there is no effective cure for HF to date. HF is a complex, multifactorial disease with unknown etiologies in many cases. Interestingly, urbanization has been shown to be a contributing factor to increased HF incidence 126, 127. Among many possible etiologies, myocardial infarction represents by far the leading cause of HF. Moreover, HF patients often present with pathological remodeling of the myocardium leading to reduced contractile function and ejection fraction. Common myocardial remodeling can be classified into two categories: systolic and diastolic dysfunction (Figure 2.8). In diastolic dysfunction, patients are characterized by increased left ventricular mass and muscular stiffness, resulting in the inability to relax properly and therefore impaired ventricular filling. HF patients with diastolic dysfunction often preserve the intrinsic ejection fraction of the myocardium. Conversely, adverse structural remodeling associated with systolic dysfunction is characterized by reduced muscle contractility to pump sufficient blood to meet the demands of the body and thereby a reduced ejection fraction.

Cardiac ejection fraction measured by echocardiogram is therefore a common measurement used to differentiate the two different types of HF. The American Heart Association has further classified HF as four different stages (Stage A-D) with clear risks of developing HF in each stage. Notably, the AHA staging of HF takes into account the fact that 5% of the population who are asymptomatic with no structural or functional heart disease but are at high risk for HF. Moreover, although it is believed that males are more likely to develop HF, it has been shown recently that males and females have equal contribution to HF 128, 129.

Normal Diastolic dysfunction Systolic dysfunction

Figure 2.8. Systolic and diastolic dysfunction in HF patients. Schematic illustration of the two different types of HF affecting the left ventricle. This figure is a modified version of materials from Servier Medical Art (https://smart.servier.com) by Servier licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit https://creativecommons.org/licenses/by/3.0/.

As a result, uneven male to female ratio in clinical trials and drug development experiments could potentially create sex differences in treatment effectiveness of HF.

2.4.2 Heart Failure with Reduced Ejection Fraction (HFrEF) HFrEF is clinically defined as a patient’s ejection fraction falls below 40% 130. HFrEF can arise from many causes ranging from coronary artery blockage to disease of the myocardium to congenital heart disease 131. Patients diagnosed with HFrEF often receive diuretic-based medications as well as vasodilators to reduce the heart’s workload and alleviate symptoms such as hypertension and edema. Also, mechanical ventricular assist devices can be used in patients with progressed systolic dysfunction and has been shown to increase the 2-year survival rate to 70% despite of all possible implications such as the development of thromboembolism, stroke, and infection. Without identifying the correct primary underlying etiology, designing a targeted treatment plan for HFrEF remains challenging to improve clinical prognostication. To date, damage of the myocardium

resulted from ischemic and dilated cardiomyopathy (ICM and DCM) represent the two major etiologies of HFrEF, contributing to >30% of global HF cases 132.

ICM is a devastating condition of the heart as a result of an occlusion of the arteries supplying the myocardium. This ischemic insult, often confined and local, to the ventricular myocardium triggers an initial inflammatory response at the site of ischemia, resulting in the removal of damaged cardiomyocytes. Replacement of lost cardiomyocytes with fibrosis and increased extracellular matrix contribute further towards systolic dysfunction, ultimately leading to dilated and weakened myocardium. The formation of fibrotic scar tissues at the site of injury is an irreversible modification of the heart in days following the initial ischemic attack, paving the path to eventual HF. Treatments for ICM often involve surgical interventions such as atherectomy, balloon angioplasty, or coronary artery bypass graft to help improve or restore blood flow to the affected area. On the other hand, DCM is a disease condition of the heart muscle affecting all chambers of the heart. DCM is often manifested by ventricular wall thinning and dilatation, resulting in increased ventricular dimensions and compromised systolic function of the heart. Familial DCM accounts for a third of all reported DCM cases, however, a wide range of non-genetic risk factors including diabetes, hypertension, viral infections, certain toxins, and poor lifestyle can also drive the development of DCM 133. Additionally, DCM is usually accompanied by adverse cardiac remodeling events such as cardiac inflammation and interstitial fibrosis in the myocardium that inform prognosis. Notably, DCM has been associated with increased risk for abnormal heart rhythms and thrombus formation 133.

At the cellular level, molecular remodeling events such as abnormal regulation of cardiac ECC and Ca2+ cycling manifested by decreased Ca2+ amplitude and frequency represents an important hallmark of failing cardiomyocytes. Additionally, perturbed SR network organization and the resulting intraorganellar communication dysfunction is another molecular characteristic of failing cardiomyocytes 134. A shift in the isoform expression levels of myosin heavy chain from alpha- to beta-myosin heavy chain in failing myocardium is also an important molecular determinant of HF 135. Further,

downregulation of 1 adrenergic receptors and SERCA2A in failing cardiomyocytes are also important molecular hallmarks of HF 136, 137. Metabolically, a shift in energy substrate usage from fatty acid oxidation to glucose and glycogen utilization associated with impaired mitochondrial function is also observed in HF patients. Lastly, the molecular mechanisms responsible for pathological remodeling of the myocardium in HF including activation of cardiac myofibroblasts, interstitial collagen deposition, and cardiomyocyte hypertrophy are common downstream molecular consequences in HFrEF of different etiologies, highlighting the molecular complexity of HF 138.

2.4.3 Heart Failure with Preserved Ejection Fraction (HFpEF) HFpEF is known as diastolic HF accounting for approximately half of the HF patient population 139. Impaired ventricular relaxation and reduced ventricular compliance in HFpEF patients significantly reduce stroke volume, resulting in cardiac dysfunction. Unlike HFrEF, common neurohumoral inhibition medications for HF such as beta- blockers and ACE inhibitors have been shown to have limited functional improvement in HFpEF patients 140. Over the past decade, aging has been shown to be an important contributing factor to the pathogenesis of HFpEF 141. Comorbidities such as hypertension, obesity, diabetes, renal abnormalities, and atrial fibrillations have been shown to greatly increase the integrative complexity of the disease in HFpEF patients 142, 143. Clinically, nearly half of HFpEF patients present with near normal blood pressure and without the ventricular hypertrophy phenotype 144. The complex clinical phenotypes and associated comorbidities in patients with HFpEF have left the disease uncharacterized and the underlying molecular mechanisms poorly understood 145. Internationally, the incidence of HFpEF is rapidly rising in urban cities occurring equally in men and women, rendering it a global pandemic in the near future 146.

At the molecular level, the underlying mechanisms responsible for ventricular diastolic dysfunction and myocyte stiffening remain poorly understood. Critical Ca2+ handling protein and the main molecular regulator of cardiac relaxation, SERCA2 pumps in the SR, has been shown to negatively correlate with age, responsible for the progressive diastolic dysfunction associated with aging 147, 148. Cardiac inflammation resulted from excessive

oxidative stress and reduced nitric oxide have been suggested to play a role in the development and progression of HFpEF 149, which leads to decreased protein kinase G activity and that translates into cardiac hypertrophy and cardiomyocyte stiffening via hypophosphorylation of titin, the largest structural protein in cardiomyocyte organization 150. Additionally, myocardial interstitial fibrosis is thought to be responsible for the myocyte stiffening phenotype observed in HFpEF patients with significant deposition of collagen type I and III 151. Interestingly, it has been reported that amyloid deposition (amyloidosis) in the heart can result in diastolic dysfunction and HFpEF 152. Among many failed clinical trials aiming to improve end-diastolic ventricular filling in HFpEF patients, the ALDO-DHF study in 2013 showed improved diastolic function in HFpEF patients using the aldosterone antagonist, spironolactone, as a treatment for HFpEF 153. However, patient exercise capacity was not improved nor did the patient quality of life. A recent study calls for tailored, phenotype-specific treatments in order to successfully address the diverse phenotypes of HFpEF ranging from lung congestion to systemic inflammation to vascular disorders 145. Taken together, these findings highlight the unique mechanistic properties of hearts with preserved ejection fraction, perhaps HFpEF should be treated as an integrative physiology disease and future studies should take on a systems- biology/physiology approach to further our understanding of this rapidly rising heart disease. Nonetheless, the Canadian guidelines for HF has now mandated a systems- biology approach to HF management to fully embrace the complexity of HF.

CHAPTER 3

Bioinformatic Analysis of Membrane and Associated Proteins in Murine Cardiomyocytes and Human Myocardium

This chapter is a modified version of a manuscript that was published in Scientific Data. Lee et al. Sci Data 7, 425 (2020).

This article is licensed under a Creative Commons Attribution 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Shin-Haw Lee1,2,8, Sina Hadipour-Lakmehsari1,2,8, Da Hye Kim1,2, Michelle Di Paola1,2, Uros Kuzmanov1,2, Saumya Shah3,4, Joseph Jong-Hwan Lee1,2, Thomas Kislinger5,6, Parveen Sharma2,7, Gavin Y. Oudit3,4, Anthony O. Gramolini1,2

1Translational Biology and Engineering Program, Ted Rogers Centre for Heart Research, Toronto, Ontario, M5G1M1, Canada. 2Department of Physiology, Faculty of Medicine, University of Toronto, Toronto, Ontario, M5S1M8, Canada. 3Department of Medicine, University of Alberta, Edmonton, Alberta, T6G2R3, Canada. 4Mazankowski Alberta Heart Institute, Edmonton, Alberta, T6G2B7, Canada. 5Princess Margaret Cancer Research Centre, Toronto, Ontario, Canada M5G1L8. 6Department of Medical Biophysics, Faculty of Medicine, University of Toronto, Toronto, Ontario, M5G1L7, Canada. 7Present address: Department of Cardiovascular & Metabolic Medicine, University of Liverpool, Liverpool, L693GE, UK. 8These authors contributed equally: Shin-Haw Lee, Sina Hadipour-Lakmehsari. email: [email protected]

SCIENTIFIC DATA | (2020)7:425 | https://doi.org/10.1038/s41597-020-00762-1 | www.nature.com/scientificdata

CONTRIBUTIONS

Shin-Haw Lee, Sina Hadipour-Lakmehsari, Da Hye Kim, Uros Kuzmanov, Parveen Sharma, Thomas Kislinger, and Anthony Gramolini conceived and designed the study and wrote the manuscript. Experimentally, Shin-Haw Lee performed all bioinformatics experiments including transcriptomic, phenotypic ontology, and membrane topology prediction analyses presented in Figures 1, 2, 3, 4, 6, and Supplemental Figures 1, 2, 3, 4, 5, 6, 7. Uros Kuzmanov provided great intellectual and technical contributions in analyzing and interpreting all bioinformatics data. Sina Hadipour-Lakmehsari helped generate and analyze GEO Profiles data presented in Figure 9. Additionally, Sina Hadipour-Lakmehsari isolated adult mouse cardiomyocytes and performed immunofluorescence and 3D reconstruction imaging analyses presented in Figures 7, 8, and Supplemental Figure 8. Da Hye Kim performed immunoblot analysis of all human heart disease tissues and analyzed the data presented in Figure 10. Shin-Haw Lee and Michelle Di Paola performed all antibody validation experiments and subsequent immunoblot analysis presented in Figure 5. Michelle Di Paola isolated and processed various mouse organ tissues used for the experiments presented in Figure 5. Saumya Shah and Gavin Oudit performed patient myocardial biopsies and provided all human myocardium used in the current project as well as all patient baseline and clinical characteristics data presented in Table 1. Joseph Jong-Hwan Lee helped analyze and interpret our full ranking strategy presented in Supplemental Table 1. Shin-Haw Lee prepared all figures and tables. Shin-Haw Lee, Sina Hadipour-Lakmehsari, Uros Kuzmanov, Da Hye Kim, Thomas Kislinger, Parveen Sharma, Gavin Oudit, and Anthony Gramolini edited and revised the manuscript. All authors reviewed and approved the final version of the manuscript. All persons listed as authors qualify for authorship, and all those who qualify for authorship are listed.

3.1 ABSTRACT In the current study we examined several proteomic- and RNA-Seq-based datasets of cardiac-enriched, cell-surface and membrane-associated proteins in human fetal and mouse neonatal ventricular cardiomyocytes. By integrating available microarray and tissue expression profiles with MGI phenotypic analysis, we identified 173 membrane- associated proteins that are cardiac-enriched, conserved among eukaryotic species, and have not yet been linked to a ‘cardiac’ Phenotype-Ontology. To highlight the utility of this dataset, we selected several proteins to investigate more carefully, including FAM162A, MCT1, and COX20, to show cardiac enrichment, subcellular distribution and expression patterns in disease. We performed three-dimensional confocal imaging analysis to validate subcellular localization and expression in adult mouse ventricular cardiomyocytes. FAM162A, MCT1, and COX20 were expressed differentially at the transcriptomic and proteomic levels in multiple models of mouse and human heart diseases and may represent potential diagnostic and therapeutic targets for human dilated and ischemic cardiomyopathies. Altogether, we believe this comprehensive cardiomyocyte membrane proteome dataset will prove instrumental to future investigations aimed at characterizing heart disease markers and/or therapeutic targets for heart failure.

3.2 INTRODUCTION Cell surface or membrane proteins serve as excellent candidates and potential diagnostic and therapeutic targets given their importance in regulating spontaneous contractions and excitation-contraction coupling of cardiomyocytes 1, 2. Additionally, membrane proteins in the mammalian heart act as important signaling receptors, metabolite transporters, enzymes, cell-cell adhesion anchors, and electrical propagation regulators 3. Therefore, a comprehensive understanding of all cardiac membrane proteins is essential to elucidate and characterize the physiology and pathophysiology of human heart diseases. However, in-depth analysis and identification of the membrane proteome of cardiomyocytes remains poorly characterized due to their hydrophobic nature and relatively low abundance 4.

Recent advancements in proteomic technologies have allowed significant improvements in membrane protein proteome discovery and topology resolution 5, 6. Specifically, advancements in mass spectrometry-based shotgun proteomics have allowed successful identification of less abundant membrane proteins, resulting in substantial improvements in representation of membrane proteins in proteomic datasets 6, 7, 8. Therefore, identification of a more complete membrane proteome in the heart has become possible and will contribute to a more detailed understanding of signaling networks, pathways, and protein-protein interactions in cellular biology of the heart.

At the cellular level, the molecular basis and mechanisms of impaired ventricular function in heart failure patients remain poorly understood. Alterations in membrane protein expression levels have been linked largely to depressed Ca2+ signaling and impaired relaxation in heart failure 9, suggesting important functional relevance of membrane proteins in cardiac muscle biology, thus representing excellent therapeutic targets that modulate cellular signaling and function in failing cardiac muscle. To date, the majority of cardiac proteomic and biochemical studies have focused on abundant soluble proteins. In this study, we carefully investigated membrane proteins identified previously 10, incorporating publicly available microarray dataset, tissue expression profiles, and informatics that resulted in the identification of 173 previously uncharacterized, intrinsic

cardiac-enriched membrane-associated proteins from human fetal and mouse neonatal ventricular cardiomyocytes. As evidence of the utility of this set, we applied a non-biased, systematic ranking strategy to all candidates and selected several highest ranked proteins to validate including FAM162A, MCT1, and COX20 as potentially novel regulators of cardiac function.

3.3 RESULTS Cardiac proteomic data analysis uncovers poorly characterized, cardiac-enriched membrane proteins. To identify the proteome of cell-surface/membrane associated and soluble proteins of human and mouse ventricular cardiomyocytes, we optimized previously the membrane enrichment of human fetal and mouse neonatal ventricular cardiomyocytes to enrich for cell surface, membrane, and membrane-associated proteins via cationic colloidal silica beads 10, 11. Quantitative MudPIT analysis had identified 2,762 human and 3,033 mouse proteins with stringent filtering (<1% FDR) of peptide and protein levels. A subsequent comparative QSpec proteomic analysis revealed 555 orthologue proteins were significantly enriched in the membrane fractions compared to the membrane-depleted counterpart. A recent update of gene names realigned the data to 550 proteins since 5 protein IDs were amalgamated with existing protein accessions (Fig. 1a). Our earlier studies focused on two of these proteins, TMEM65 and REEP5, in detailed biochemical analyses, identifying their specific roles in regulating cardiac conduction and cardiac muscle sarcoplasmic reticulum organization and function, respectively 10, 12. Here, we performed a detailed analysis of all 550 identified cardiac membrane proteins and showed that GO term-associated classifications demonstrated 50% of the identified protein clusters had a predicted transmembrane domain and 60% of them contained the GO term “membrane” (Fig. 1b). Moreover, 70% of the 550 protein clusters were classified as cell surface-associated proteins and only 20% of them were linked to known cardiac phenotype ontologies based on available forward genetics studies in mice in the MGI database (Fig. 1b). All 550 proteins were each scored and ranked on a set of non-biased, defined criteria (Fig. 1c and full ranking criteria listed in Supplemental Table 1) to prioritize proteins that had no previous association to cardiac

function and were enriched >3-fold within cardiac tissue compared to other tissues, according to publicly available microarrays (Supplemental Table 1; individual worksheets represent rank ordered analysis of the 550 membrane associated protein clusters in various subcellular domains). Based on this analysis, this strategy resulted in 173 of these membrane-associated proteins that would be classified as previously uncharacterized, cardiac-enriched membrane-associated proteins. Hierarchical clustering of the 550 membrane-associated protein clusters demonstrated distinct clusters of cardiac enrichment across various healthy human tissues (Supplemental Fig. 1a).

We next mapped these 550 membrane enrichment proteins to the Human Protein Atlas ‘Membrane Proteome’; where approximately 28% of all human protein coding genes are predicted to harbor at least one transmembrane domain (Fig. 1d). This analysis showed that 154 of our identified cardiac-enriched membrane proteins were present in the Membrane Proteome, leaving 396 out of the 550 protein clusters potentially novel membrane protein products. Similarly, we compared our list of 550 proteins against recent studies of whole tissue lysates from monkey heart proteome by Hu et al. 13 and the human heart proteome by Doll et al. 14 (Fig. 1e). Of note, 229 cardiac-enriched membrane proteins were also identified in both of these studies with 12 unique proteins identified only in our analysis (Fig. 1e).

Next, we assessed the organelle distribution of all “membrane-associated” candidates identified by our proteomics and informatics analyses. We analyzed their subcellular localizations based on their Gene Ontology cellular component (GOCC) classification (Fig. 1f) and observed that 228/550 (41%) of all candidate proteins were identified as mitochondrial in origin which was expected as cardiomyocytes are highly oxidative and metabolically active. Interestingly, 125/550 (23%) of identified membrane-associated proteins were classified as localized to the nucleus, likely suggesting the dynamic roles of nuclear membrane proteins in regulating gene expression, cell mobility, and DNA damage repair in cardiomyocytes 15, 16 (Fig. 1f).

Fig. 1 Membrane proteomic analysis of human fetal and murine ventricular cardiomyocytes. (a) Schematic diagram of experimental workflow that resulted in the identification of 550 membrane-enriched protein clusters in human and mouse cardiomyocytes. (b) Heatmap depicting Gene Ontology (GO) term associated distribution and cardiac/heart phenotype ontologies of the 550 membrane-enriched protein clusters. (c) Schematic diagram of the ranking strategy applied to the 550 protein clusters to

identify previously uncharacterized cardiac-enriched membrane proteins. (d) Venn diagram depicting overlap of protein coverage between our membrane-enriched proteome from human and mouse ventricular cardiomyocytes and the Human Protein Atlas Membrane Proteome. (e) Venn diagrams depicting overlap of protein coverage between our membrane-enriched proteome from human and mouse ventricular cardiomyocytes and the monkey (Rhesus macaque) heart proteome from Hu et al. 13 and human heart proteome from Doll et al. 14. (f) Distribution of the 550 membrane-enriched protein clusters across various subcellular organelles based on their GO cellular component classification. Detailed ranking criteria and step-by-step ranking of all candidates were uploaded to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12).

Moreover, 63/550 (12%) of the 550 membrane associated protein clusters identified were classified as ‘cell surface’ or ‘plasma membrane’ in origin and another 65/550 (12%) were attributed to other organelles including the endoplasmic reticulum, golgi apparatus, peroxisomes, and lysosomes. Interestingly, of much less abundance, 6% (32/550) and 4% (22/550) of protein clusters were classified as cytosolic or proteins in the secretory pathway, respectively, suggesting the identification of membrane-associated cytosolic and secreted proteins in our proteomic analysis. However, it is important to acknowledge that Gene Ontology classification has limitations, further detailed biomolecular studies obviously would be needed to determine precise subcellular localization and function in the cell.

Identification of organelle-specific, previously uncharacterized cardiac membrane proteins. Next, we aimed to identify highly scored cardiac-enriched protein clusters that have not been investigated to date in the heart across various organelles in the myocyte. Based on previous GOCC classification, we had identified 228 mitochondrial membrane proteins, 125 nuclear membrane proteins, 63 plasma membrane associated proteins, 65 ER, golgi apparatus, peroxisome, lysosome-associated membrane proteins, 32 cytosolic, and 22 secreted membrane associated proteins (Fig. 1f). Our bioinformatic analysis (Fig. 2) showed that 155 (68%) mitochondrial membrane-associated proteins were at least 3- fold enriched within cardiac tissue according to publicly available BioGPS microarrays; 130 of which were further classified as ‘novel’ cardiac membrane proteins with no previously reported MGI cardiac related phenotype. Similarly, 10% of all identified nuclear

membrane associated proteins were found cardiac-enriched by BioGPS, and 6 of the 13 cardiac-enriched nuclear membrane associated proteins were reported with no previous MGI heart related phenotype (Fig. 2b).

Detailed analysis of the plasma membrane-associated protein clusters in Fig. 2c showed that 39 (62%) of which were found 3-fold enriched within cardiac tissue, 15 of these were further classified as novel PM related cardiac-enriched membrane proteins with no phenotype ontology. Similar informatic analysis of ER, golgi apparatus, peroxisome, and lysosome-related membrane proteins showed that 16 (25%) organellar proteins were considered cardiac-enriched, 10 of which were further classified as novel cardiac- enriched organellar membrane-associated proteins with no previous reported cardiac phenotype (Fig. 2d). Interestingly, AFR3 was solely identified in our membrane proteomic analysis of human and mouse cardiomyocytes. Additionally, 14 (44%) cytosolic and 13 (59%) secreted protein clusters were found to be cardiac-enriched (Fig. 2e,f). Specifically, 7 cytosolic membrane-associated proteins had no previously reported MGI heart related phenotype, and 5 secreted membrane-associated proteins were found with no reported MGI heart phenotype.

Next, tissue-wide transcriptomic analysis of these previously uncharacterized, cardiac- enriched membrane proteins was performed across non-diseased human tissues. Specifically, we employed mRNA transcript data from the Human Proteome Map 17 and showed dominant cardiac expression across various organ tissues for each organellar classification (Fig. 3). Specifically, the gene level expression matrix was downloaded and normalized to total expression across all tissues 17. Figure 3a shows predominant mRNA expression in the fetal and adult heart of most of the 130 mitochondrial related membrane proteins. Notably, COQ3, NDUFB8, NDUFB11, PDHB, ATP5D, APOOL, PDK1, and CPT1B showed nearly exclusive expression in the heart. Moreover, dominant cardiac expression of MRL4, MRPS31, and PDK1 in the fetal heart compared to adult heart was observed. Conversely, expression of SUCLG1, SUCLG2, SDHA, COQ3, NDUFA3, and NDUFB4 was seen in the adult heart compared to fetal heart (Fig. 3a).

Fig. 2 Analysis of cardiac membrane proteins within various subcellular domains. Consort diagrams showing classification of the 550 membrane-enriched protein clusters and analyses of cardiac enrichment and phenotypic novelty based on publicly available BioGPS microarray data and MGI phenotypic analysis, respectively in (a) mitochondrion, (b) nucleus, (c) plasma membrane, (d) other organelles (ER, golgi apparatus, peroxisomes, lysosomes), (e) cytosol, and (f) the secretory pathway.

Heat maps of the plasma membrane proteins showed proteins such as MCT1, DYSF and SORBS1 were enriched in the heart with some expression across the other tissues while OBSCN was detected only in the heart (Fig. 3b). Analysis of the 10 organellar (ER, golgi, peroxisomes, lysosomes) (Fig. 3c), 7 cytosolic (Fig. 3d), 6 nuclear (Fig. 3e), and 5

secreted (Fig. 3f) membrane-associated proteins showed transcript expression across various healthy human tissues. Notably, SYNPO2L, MYOM1, MYL3 showed restricted expression to only the cytosol in the heart (Fig. 3d). However, it is important to acknowledge that transcriptomic data must be interpreted carefully due to the poor correlation between mRNA expression level and the abundance of its associated protein product. To circumvent this issue, we applied a defined set of non-biased criteria to prioritize the candidate proteins for follow-up experiments (Supplemental Table 1). This strategy equally evaluated each criterion in our rank-ordered analysis of all candidates and resulted in a total of 18 highest ranked membrane proteins where existing data demonstrated their subcellular localization to both human and mouse cardiomyocyte membranes in the cell with no previous connection to cardiac function, and are least 3- fold enriched in the heart (Supplemental Table 1). Interestingly, many of the abovementioned cardiac preferential proteins were not among the 18 highest ranked candidate proteins due to non-membrane subcellular localization and/or previous association with cardiac function. These cardiac enriched proteins are quite likely of great interest in future studies to further our understanding of their precise role in heart, although they did not reach a high enough criterion weighting in our particular evaluation.

Nevertheless, transcriptomic analysis of the 67 cardiac-enriched membrane associated proteins with previous MGI cardiac phenotype further demonstrated the utility and potential of our dataset (Supplemental Fig. 2a–f). Specifically, mutations in cell surface protein DSC2 have been linked to arrhythmogenic and hypertrophic cardiomyopathies 18, 19, 20, 21 while POPDC2 has been shown to be essential for heart muscle development 22, 23. Deletion of cardiac cytosolic protein CTNNA3 in the mouse has been associated with dilated cardiomyopathy and actomyosin dysregulation in the heart 24, 25. Lastly, detailed transcriptomic breakdown of the non-cardiac-enriched protein distributions were also analyzed and shown in Supplemental Figs. 3, 4a–f for comparison.

Fig. 3 Transcriptomic analysis of novel cardiomyocyte-enriched membrane proteins shows cardiac enrichment across various tissues. Heatmaps showing mRNA transcript levels of 173 novel cardiac-enriched membrane proteins across clinically defined healthy human tissues; mRNA transcript data were obtained from Human Proteome Map and are presented according to their subcellular classifications in (a) mitochondrion, (b) plasma membrane, (c) other organelles (ER, golgi apparatus, peroxisomes, lysosomes), (d) cytosol, (e) nucleus, and (f) the secretory pathway. All source data input and normalized output files were uploaded to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12).

FAM162A, MCT1, and COX20 are evolutionarily conserved, cardiac ventricular- enriched membrane proteins. To further assess the subcellular localization of the 173 ranked candidates in each subcellular organelle in the heart, we utilized immunohistochemical data in the Human Protein Atlas 26 to investigate their protein distributions and expression patterns in healthy human ventricular tissues (Fig. 4). Specifically, relative protein expression was obtained through the reported human heart muscle immunohistochemical images available at the Tissue Atlas. For these studies, we selected the highest ranked proteins in each subcellular organelle classification (Supplemental Table 1) and, where available, obtained human immunohistochemical images from the HPA. Using this approach, as expected COX20 and FAM162A showed localization to mitochondria of cardiomyocytes with strong mitochondrial expression patterns throughout the sections. Similarly, TMEM65, MCT1, SLC12A7, TLN2, MSN, FLOT1, SNTA1, DYSF, and AOC3 showed localization to the membranes of cardiomyocytes with particularly strong expression at cell surface and the cardiac intercalated disks. Immunohistochemistry analysis of REEP5 showed moderate expression and localization to the ER throughout the cells. Furthermore, STOM and CTNND1 revealed localization to the cytosolic space of cardiomyocytes. Similarly, LAMA2, LAMB2, and BGN showed subcellular localization throughout the cells perhaps in the protein secretory pathway. Immunohistochemistry analysis of MYH7, TNNT2, and TNNI3 were included as positive controls and demonstrated very strong, sarcomeric expression patterns and localization to cardiac muscle fibers in these healthy human ventricular tissues (Fig. 4).

Fig. 4 Immunohistochemistry analysis of top ranked cardiac-enriched candidates. Immunohistochemistry analysis of the top ranked candidates in each subcellular classification (Mitochondrion, plasma membrane, other organelles (ER, golgi apparatus, peroxisomes, lysosomes), cytosol, and the secretory pathway in healthy human ventricular tissues from the Human Protein Atlas. Immunohistochemical stain intensity was assessed independently as either weak, moderate or strong and reported as protein expression score (PES). COX20, moderate PES; FAM162A, weak PES; Tmem65, moderate PES; MCT1, moderate PES; SLC12A7, moderate PES; TLN2, moderate PES; MSN, moderate PES; FLOT1, moderate PES; SNTA1, moderate PES; DYSF, moderate PES; AOC3, negative PES; REEP5, moderate PES; STOM, weak PES; CTNND1, moderate PES; LAMA2, strong PES; LAMB2, strong PES; BGN, moderate PES; MYH7, strong PES; TNNT2, strong PES; TNNI3, strong PES. Scale bar, 25μm. All images shown are representative of approximately 5 distinct regions of interest (ROIs) assessed per sample, n = 3 independent patient tissue blocks.

As evidence of the utility of this refined dataset, we selected several highly ranked cardiac-enriched membrane-associated proteins to determine their protein expression across mouse tissues including FAM162A, MCT1, and COX20. We selected FAM162A, MCT1 and COX20 as prioritized candidates of novel proteins based on their highly muscle-specific tissue profiles along with MGI phenotypic analysis, representing the next logical targets for detailed follow-up biomolecular studies. In addition to the phenotypic analysis using the mouse genome informatics data, no known human mutations of FAM162A, MCT1, and COX20 have been linked to classical human heart diseases. To corroborate previous mRNA tissue expression analysis of FAM162A, MCT1, and COX20, immunoblot analysis of diverse adult mouse tissues (whole heart, ventricle, atria, whole brain, stomach, kidney, skeletal muscle) demonstrated strong protein expression in the whole heart in the heart for FAM162A (Fig. 5a), MCT1 (Fig. 5b), and COX20 (Fig. 5c). Notably, while FAM162A, MCT1, and COX20 demonstrated ventricular muscle enrichment, abundant expression of MCT1 was also detected in the atria and skeletal muscle (Fig. 5b). To examine antibody specificity against FAM162A, MCT1, and COX20, HEK overexpressed cMyc/DDK-tagged human FAM162A, MCT1, and COX20 purified lysates were obtained (Fig. 5a–c, left panels) for subsequent immuno-competition assays. For these experiments regarding FAM162A, immunoblot analysis of the HEK overexpressed human FAM162A lysate showed a clear doublet of FAM162A ~17 and 20 kDa. However, in the mouse tissues, higher molecular weight species of FAM162A were observed ~25, 35 and 48 kDa (Fig. 5a); although the implications of the potential oligomeric/altered status of FAM162A remains to be elucidated. Nonetheless, a clear enrichment of FAM162A antibody reactivity in striated muscle in the heart and skeletal muscle was apparent. Importantly, immunodepletion of the bands in the immunoblots using the FAM162A overexpression lysate demonstrated reduced intensity of bands observed at 17, 25, 35 and 48 kDa across all tissues indicating some degree of specificity of the reactivity (Fig. 5a, middle panel). MCT1 showed much cleaner antibody reactivity with a clear single band at ~42 kDa (predicted size) across nearly all tissues (Fig. 5b, top panels) that showed immunodepletion using purified cMyc/DDK-tagged MCT1 lysate, as the 42 kDa bands showed significant reduction in intensity (Fig. 5b, middle panel). The COX20 reactivity showed multiple band reactivity at 13 (predicted correct size), 25, 50,

58, 60 and 65 kDa (Fig. 5c, top panels). Immunodepletion using purified cMyc/DDK- tagged COX20 lysate was successful, as the multitude of bands visualized showed significant reduction in intensity under the competition conditions (Fig. 5c, middle panel). Lastly, immunoblot analysis of FAM162A, MCT1, and COX20 in isolated cardiac cytosolic fraction and microsomes showed a clear enrichment of MCT1 and COX20 in enriched membrane fractions (microsomes), along with the presence of FAM162A (Fig. 5d, left panels). While COX20 and MCT1 were not detected in the cytosolic fraction, FAM162A showed abundant expression also in this fraction. Control immunoblots were performed using calnexin (ER-enriched membrane protein enriched in microsomes), α-tubulin (cytosolic protein), sodium-calcium exchanger (NCX1; plasma membrane protein), and COXIV (inner mitochondrial membrane protein) (Fig. 5d, right panels) demonstrated successful subcellular fractionation of cardiac cytosolic and membrane fractions.

Given the identification of all three of these proteins in both human and mouse cardiomyocyte membrane proteomic isolations, we performed a detailed multi-species amino acid sequence analysis of FAM162A, MCT1, and COX20, with each of the three candidates exhibited a high degree of conservation throughout eukaryotic evolution. Specifically, FAM162A showed 79% homology between human and mouse FAM162A and 65% peptide conservation throughout evolution (Supplemental Fig. 5a). FAM162 is a family of proteins consisting two members, FAM162A and FAM162B. Our phylogenetic analysis demonstrated two distinct clustering of mammalian FAM162, demonstrating two diverse evolutionary descents within the FAM162 family (Supplemental Fig. 5b). Our RNA-Seq analysis of the FAM162 family further demonstrated that FAM162A shows dominant expression across various tissues in the FAM162 family (Supplemental Fig. 5c).

MCT1 exhibited 86% homology between human and mouse MCT1 and 78% peptide conservation throughout evolution (Supplemental Fig. 6a). MCT1 belongs to proton- linked monocarboxylate transporters consisting of a family of 14 members within the MCT family of proteins as demonstrated by our phylogenetic analysis yielding 14 distinct clustering of mammalian taxa (Supplemental Fig. 7a). While the expression distribution of MCT isoforms has been shown to be organism and tissue specific, our RNA-Seq

Fig. 5 Immunoblot analysis of protein expression across multiple tissues and microsomal fractions. Immunoblot analysis of (a) FAM162A, (b) MCT1, and (c) COX20 protein expression levels in mouse tissues. Blots were incubated with either anti-DDK antibody or FAM162A, MCT1 or COX20. Immunodepletion experiments were carried out by incubating primary antibodies together with HEK overexpression lysates. Coomassie blue gels were performed to visualize protein loading across all samples. (d) Immunoblot analysis of FAM162A, MCT1, and COX20 in isolated adult cardiac cytosolic fraction and enriched membrane microsomes from 10 mouse hearts. Blots were probed with primary antibodies as indicated, and in parallel, samples were run on a Coomassie blue gel. All immunoblots shown are representative of 3 total immunoblots performed, n = 3 independent biological replicates. Original uncropped immunoblots are provided in Supplemental Fig. 9. All original uncropped images were uploaded to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12). analysis of various tissues showed that MCT1 is the predominant MCT in both the fetal and adult heart (Supplemental Fig. 7b) 27. Lastly, from zebrafish to mouse to monkey to human, COX20 demonstrated 81% homology between human and mouse COX20 and 69% peptide conservation throughout eukaryotic evolution (Supplemental Fig. 5d).

FAM162A, MCT1, and COX20 transmembrane sequence prediction and membrane topology. We explored human membrane topology using a multi-algorithm prediction tool (TOPCONS) based on available peptide sequences. We showed that a consensus FAM162A membrane topology model exhibited one transmembrane domain region with the N-terminus region residing in the inter mitochondrial membrane space and the C- terminus in the cytosol (Fig. 6a,d). Membrane topology and hydrophobicity prediction analyses of human MCT1 have proposed a structure with eleven transmembrane domains with the N-terminus in the extracellular space and the C-terminus in the cytosol as well as a large internal cytoplasmic loop between TM5 and TM6 (Fig. 6b,e). This prediction is consistent with the knowledge that MCTs possess ten to twelve transmembrane domains. However, structural labeling and proteolytic digestion studies of rat MCT1 has been shown to possess a total of twelve transmembrane domains with both N- and C-termini within the cytosol 28. Lastly, our consensus membrane topology analysis of human COX20 confirmed two transmembrane domains with the hydrophilic

Fig. 6 Membrane topology and prediction analysis of FAM162A, MCT1, and COX20. (a–c) Prediction of human FAM162A, MCT1, and COX20 protein topography generated by TOPCONS (http://topcons.cbr.su.se). (d–f) Predicted membrane topology model of human FAM162A, MCT1, and COX20A generated by modification of a T(E)Xtopo output. Peptide immunogen sequences for commercially available FAM162A, MCT1, and COX20 antibodies used in the current study are indicated by an antibody icon. domains of COX20 in the mitochondrial intermembrane space (Fig. 6c,f), consistent with the literature.

FAM162A, MCT1, and COX20 endogenous expression and localization in cardiac myocytes. We next performed immunofluorescence analysis of endogenous expression pattern in primary cultured neonatal mouse ventricular cardiomyocytes and isolated adult mouse ventricular cardiomyocytes for FAM162A, MCT1, and COX20. FAM162A demonstrated typical mitochondrial staining pattern in cultured neonatal cardiomyocytes (Fig. 7a) and a very organized mitochondrial staining pattern in isolated adult mouse cardiomyocytes (Fig. 7b). MCT1 showed clear localization to the plasma membrane in cultured neonatal mouse ventricular cardiomyocytes and isolated adult mouse cardiomyocytes (Fig. 7a,b). Endogenous COX20 expression exhibited characteristic mitochondrial staining patterns in cultured neonatal mouse ventricular cardiomyocytes and isolated adult mouse cardiomyocytes (Fig. 7a,b) similar to FAM162A patterns. To validate the specificity of the staining patterns, our immunodepletion assays using cMyc/DDK-tagged FAM162A, MCT1, and COX20 lysates successfully masked the specific signals shown in Fig. 7b top panels in cardiomyocytes. Additionally, three- dimensional reconstructive analysis of our confocal data was performed in the isolated adult cardiomyocytes for all three proteins. FAM162A and COX20 showed longitudinal cardiac mitochondria as well as localization between rows of myofibrils (Fig. 7c). Three- dimensional reconstructive analysis of endogenous MCT1 expression in isolated adult mouse cardiomyocytes showed strong expression in the plasma membrane and cardiac intercalated disks (Fig. 7c).

To further corroborate the specificity of the staining patterns, we performed immunofluorescence co-staining analysis of FAM162A with COX IV, a known

Fig. 7 Immunofluorescence analysis of FAM162A, MCT1, and COX20 expression in mouse neonatal and adult ventricular cardiomyocytes. (a) Immunofluorescence analysis of endogenous FAM162A, MCT1, and COX20 expression (green) in isolated mouse neonatal ventricular cardiomyocytes. Nuclear staining was visualized with Hoechst staining (blue). Scale, 10μm. (b) Confocal imaging of endogenous FAM162A, MCT1, and COX20 expression (green) in isolated adult mouse ventricular cardiomyocytes demonstrate striated mitochondrial staining patterns for FAM162A and COX20, and exclusive plasma membrane localization of MCT1. Lower panels, imaging experiments were carried out by incubating reaction mixes of the primary antibodies with their respective overexpression protein lysates prior to incubating overnight. Scale, 20μm. (c) Three-dimensional reconstructive analysis using Imaris (Oxford Instruments) highlights the endogenous expression patterns of FAM162A, MCT1, and COX20. All images shown are representative of approximately 30–40 total images captured per condition, n = 3 independent biological replicates. All original uncropped microscopy images were uploaded to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12). mitochondrial membrane marker, and demonstrated a high degree of co-localization in the mitochondria in isolated adult mouse cardiomyocytes (Fig. 8a, left panels and Supplemental Fig. 8a). Three-dimensional reconstructive analysis confirmed significant, strong co-localization of FAM162A (Pearson coefficient of 0.76 ± 0.16; p < 0.05) with COX IV in adult cardiomyocytes (Fig. 8a, right panels). Next, we compared the expression of MCT1 with that of a known plasma membrane protein, G protein inhibitory alpha subunit (Gαi) which demonstrated strong expression of both MCT1 and Gαi in their respective subdomains in the plasma membrane and MCT1 in the intercalated disks of cardiomyocytes (Fig. 8b and Supplemental Fig. 8b). Three-dimensional reconstructive analysis further demonstrated strong co-localization between MCT1 and Gαi at the plasma membrane (Pearson coefficient of 0.53 ± 0.03; p < 0.05) (Fig. 8b). Lastly, immunofluorescence analysis of COX20 co-stained with a mitochondrial marker, COX IV, demonstrated strong subcellular mitochondrial co-localization throughout the cell (Pearson coefficient of 0.61 ± 0.12; p < 0.05) (Fig. 8c and Supplemental Fig. 8c).

Differential regulation of FAM162A, MCT1, and COX20 protein expression within the myocardium of heart failure patients. To assess for any correlation to cardiac disease, we performed a query of publicly available GEO RNA-Seq datasets containing

Fig. 8 Co-immunofluorescence analysis demonstrates colocalization of FAM162A and COX20 with known mitochondrial marker, COXIV, and MCT1 colocalization with known plasma membrane protein, Gαi, in isolated adult mouse cardiomyocytes. (a) Immunofluorescence analysis of FAM162A (green) co-stained with mitochondrial protein, COXIV (red) in acutely isolated adult mouse cardiomyocytes. Three-dimensional reconstructive analysis demonstrates regions of colocalization (yellow) with a Pearson’s coefficient p > 0.5. Scale, 10 μm. (b) Immunofluorescence analysis of

MCT1 (green) co-stained with known plasma membrane protein, Gαi (red) in acutely isolated adult mouse cardiomyocytes. Three-dimensional reconstructive analysis demonstrates regions of colocalization (yellow) with a Pearson’s coefficient p > 0.5. Scale, 10μm. (c) Immunofluorescence analysis of COX20 (green) co-stained with mitochondrial protein, COXIV (red) in acutely isolated adult mouse cardiomyocytes. Three-dimensional reconstructive analysis demonstrates regions of colocalization (yellow) with a Pearson’s coefficient p > 0.5. Scale, 10μm. Nuclear staining was visualized with Hoechst staining (blue). All images shown are representative of approximately 30–40 total images captured per condition, n = 3 independent biological replicates. All original uncropped microscopy images were uploaded to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12). data for human and mouse cardiovascular diseases to determine expression levels of FAM162A, MCT1, and COX20. Specifically, FAM162A transcript levels were upregulated in several mouse disease models including transverse aortic constriction-induced (TAC) heart failure, myocardial infarction, and hypertrophic cardiomyopathy (Fig. 9a). Consistent with the mouse data, an increased expression of FAM162A was observed in human dilated cardiomyopathy (DCM), idiopathic cardiomyopathy, and ischemic cardiomyopathy (ICM), albeit not significantly. Interestingly, MCT1 transcript levels changed uniquely in human heart diseases. While a relatively inconsistent trend of MCT1 expression levels was observed with several mouse disease models, a marked reduction in MCT1 transcript levels were detected in human idiopathic and ischemic cardiomyopathies, perhaps suggesting a link in ICM disease progression (Fig. 9b). Similarly, decreased levels of COX20 transcripts were observed in mouse TAC, infarcted, and hypertrophic hearts (Fig. 9c). In agreement with these data, human DCM and idiopathic cardiomyopathy also showed a downregulation of COX20 expression, albeit not significantly (Fig. 9c). However, we acknowledge that these data are observational in nature and do not establish a causal or direct link between the altered expression and cardiac dysfunction.

In an attempt to establish the link between altered expression and human heart diseases, we performed immunoblot analysis of FAM162A, MCT1, and COX20 to investigate protein levels in human hearts from patients with clinical DCM and ICM, the two most common etiologies of heart failure. Patient baseline data and clinical characteristics are

Fig. 9 GEO transcriptomic analysis of FAM162A, MCT1, and COX20 mRNA transcript levels in various mouse and human heart diseases. (a–c) GEO RNA profiles demonstrate alterations in FAM162A, MCT1, and COX20 expression across various mouse and human heart diseases. Data are reported as normalized hybridization signals. Red bars indicate increased expression against normal hearts, blue bars indicate decreased expression against normal hearts, n = 3–5 available biologically independent measurements. Asterisks indicate a statistically significant p value in a Tukey’s multiple comparison analysis where *p < 0.05 and **p < 0.01; data are presented as mean ± SEM. All source data input and analyzed output files were uploaded to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12). reported in Table 1. We first compared their expression levels in human DCM and ICM myocardial tissues to non-failing myocardial controls (NFC). Protein levels of FAM162A were found significantly elevated in DCM and to a greater extent in ICM hearts; whereas MCT1 and COX20 expression levels were significantly reduced in ICM hearts only (Fig. 10a). A near complete depletion of MCT1 expression in ICM hearts, but not DCM hearts, showed apparent disease-specific changes in the expression levels of MCT1 and may provide functional insights to ICM-induced heart failure (Fig. 10a). MCT1 has been reported at 40–48 kDa in previous studies in human muscle tissues 29, 30, and consistent with the literature, we detected molecular bands of MCT1 at 42 kDa. The presence of a higher molecular weight band (~52 kDa) across all our human blots was suspected to be IgG. Next, assessment of protein expression in the distinct regions of the left ventricular myocardium of ICM patients was performed using myocardial tissues from infarct, peri-, and non-infarct regions of the left ventricle to explore intra-cardiac differences (Fig. 10b). Markedly elevated FAM162A protein expression levels were measured in the dynamic remodeling region of the peri-infarct relative, and levels were nearly absent in the infarct zone which is composed of fibrotic scar mostly (Fig. 10b). Furthermore, the relatively low MCT1 expression in ICM hearts compared to NFC and DCM hearts was shown to be consistent within the whole left ventricle of ICM hearts, whereas decreased COX20 protein levels appeared to be restricted to the infarct region (Fig. 10b). These data demonstrate that FAM162A upregulation occurred in both models of heart failure, and that MCT1 and COX20 decreased levels occurred in ICM, suggesting perhaps distinct roles for these protein candidates. However, two major limitations for consideration when

evaluating human samples are the heterogeneous cell populations and inter-individual variability which may have resulted in altered expression ratios observed in our study. Future investigations with a larger patient sample size and follow-up biochemical studies in isolated cardiomyocytes are required to corroborate and extend these initial findings.

Table 1 Patient baseline and clinical characteristics. EF, ejection fraction; Hb, hemoglobin; eGFR, estimated glomerular filtration rate; ACEi/ARB, angiotensin- converting enzyme inhibitor/angiotensin receptor blocker; MRA, mineralcorticoid receptor antagonist.

3.4 DISCUSSION The process of cardiac contraction is precisely controlled by many essential membrane proteins to generate electrical signals and mechanical forces needed to pump the heart. For this reason, membrane biology is perhaps the most important, yet most difficult to study in cell biology due to poor solubility and relatively low expression of this class of protein. Here, we used quantitative mass spectrometry data, informatics and imaging analysis of human and mouse cardiomyocyte to investigate these cardiac-relevant, membrane proteins. We identified 173 membrane-associated proteins as a result of this analysis. We have provided further compelling evidence suggesting that some of our prioritized rank-ordered list of proteins, including FAM162A, MCT1, and COX20 are cardiac enriched, localized within subcellular compartments, and may play a role in the progression of heart failure.

Fig. 10 FAM162A, MCT1, and COX20 protein expression levels in human adult DCM and ICM heart failure patients. (a) Immunoblot analysis of FAM162A, MCT1, and COX20 in human adult DCM and ICM patient samples. Right panel, densitometric quantification of the blots. Significantly higher levels of FAM162A in human adult DCM and ICM patients, and significantly reduced levels of MCT1 and COX20 protein expression in ICM patients were measured. (b) Immunoblot analysis of FAM162A, MCT1, and COX20 in region-specific infarcted hearts. Right panel, densitometric quantification of the blots. All immunoblots shown are representative of at least 3 total immunoblots performed, n = 3 independent biological replicates. Original uncropped immunoblots are provided in Supplemental Fig. 9. Asterisks indicate a statistically significant p value in a Tukey’s multiple comparison analysis where *p < 0.05 and **p < 0.01; data are presented as mean ± SEM. All source data input and analyzed output files were uploaded to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12).

While FAM162B has been linked to congenital colon disease as a result of lacking nerve cells in the colon 31, FAM162A has been shown to function as a pro-apoptotic molecule involved in facilitating hypoxia-induced mitochondrial apoptosis 32, 33. Specifically, in vitro overexpression of FAM162A induces canonical mitochondrial cell death in prostatic

cancer cells and human alveolar epithelial cells. More excitingly, FAM162A targeted suppression has been shown to attenuate apoptosis, demonstrating its anti-apoptosis potential for potential alternative therapeutic strategy 32, 34. However, the specific role and detailed subcellular localization of FAM162A in the myocyte in the heart has not been explored.

MCT1 is best known as a proton-coupled monocarboxylate transporter that mediates the transport of many monocarboxylates such as lactate, pyruvate, ketone bodies, and various amino acids across the plasma membrane. The regulation of these small molecules contributes to a multitude of metabolic pathways during normal physiological as well as pathological conditions. The molecular function of MCT1 and its importance for normal physiological metabolism have been characterized, previously 35. However, considerably less information is available regarding the precise role of MCT1 in cardiac metabolism under normal and pathological conditions. In heart failure, a metabolic shift in energy substrate is established, favoring carbohydrate utilization as well as increased lactate transport to improve heart function in response to an insult 36. In fact, increased MCT1 expression in a rat model of myocardial infarction and congestive heart failure has been observed, suggesting the critical initial compensatory role of MCT1 in transporting lactate as a metabolic resource in the failing heart 37.

Finally, the assembly of cytochrome c oxidase in the mitochondria is clearly essential for mature cellular respiration and aerobic metabolism in bacteria and eukaryotes 38. COX20, an inner mitochondrial membrane protein, has been shown to play an important role in organizing efficient cytochrome c oxidase assembly, functioning as a chaperone for the process 39. Specifically, the hydrophilic domains of COX20 have been shown to be required for interactions with COX2 and COX18, two conserved subunits of the cytochrome c oxidase complex, to ensure successful assembly and respiratory growth 39. Furthermore, mutations in human COX20 have been linked to loss of neurological control of muscle movements resulting in muscle hypotonia and dystrophy 40, 41. However, the phenotypic spectrum of COX20 expression alterations has not been investigated to date despite its high expression levels in the ventricle in the heart.

Elucidating the complete membrane proteome of cardiomyocytes furthers our understanding of myocyte biology in both the healthy and diseased heart. Our in-depth membrane proteomic analyses followed by in vitro validation in functional adult mouse cardiomyocytes as well as detailed immunoblot analysis from human heart biopsies have provided compelling evidence and suggested a link between these previously uncharacterized, cardiac-enriched membrane proteins to heart disease progression in the human heart. However, it is important to acknowledge that abundance may not be a sufficient indicator of translational significance in function. In this study, we have prioritized candidate proteins that show cardiac enrichment as a first step in characterizing previously understudied cardiac membrane-associated proteins. The role and contribution of many of the non-cardiac-enriched candidate proteins, with no previous reported cardiac phenotype shown in Supp. 3, will likely be clarified with detailed future studies.

Furthermore, we have characterized their subcellular localization and expression patterns in the myocyte and performed detailed, region-specific protein expression analysis in dilated cardiomyopathy and ischemic cardiomyopathy. Detailed comparison of the dynamic regional subgroups, such as non-infarct, peri-infarct, and infarct regions of the ischemic hearts provides key insights into understanding the basis and transition between healthy and diseased myocardium. A marked increase in FAM162A expression level in the peri-infarct regions of the infarcted hearts suggest its role in the progression of ICM; we speculate where it may be involved in inducing mitochondrial apoptosis and thus establishing the pathological phenotypes in response to ischemic insults in the heart. Similarly, an upregulation of FAM162A in human DCM hearts suggests its specific role in modulating the etiologies of this specific heart disease characterized by dilated and impaired ventricular contraction. Defects in mitochondrial proteins in the heart are a frequent cause of cardiac myopathies as cardiomyocytes are mitochondria enriched and the myocyte relies on the mitochondria to generate sufficient ATP to fuel beat-to-beat contractions. Future studies investigating the therapeutic potential of FAM162A depletion in the failing heart would thus be invaluable. Moreover, a near complete deletion of MCT1

expression in infarcted hearts – irrespective of ventricular region – may indicate a crucial importance in maintaining normal cardiac physiology and its direct contribution to ICM. However, the specific mechanisms will require ongoing investigations, with future conditional, tissue-specific knockout studies in the heart providing immeasurable insights and quality control into their specific roles in regulating heart function. Until then, ongoing work will continue refine the role of FAM162A, MCT1, and COX20 in the heart.

Lastly, it is important to acknowledge that this study is primarily explorative and observational in its nature. Obviously, future studies must focus on the precise molecular functions of these identified proteins in cardiac health and disease to determine their specific role(s) in regulating heart development and/or function. Moreover, although we attempted necessary controls where possible, antibody-based experiments should always be assessed and interpreted very carefully. The antibodies used in this study were carefully selected, with previous validations for specificity, however, accuracy of these subcellular localization data of FAM162A, MCT1, and COX20 in cardiomyocytes depend largely on the quality of the primary antibodies used in our confocal studies. Despite such limitations, our co-immunofluorescence studies with known mitochondrial and plasma membrane markers and consistent data from three-dimensional reconstructive imaging analysis along with high sample numbers were designed to minimize any potential confounding factors in these studies.

Taken together, our study represents an integrated analysis of cardiomyocyte cell surface proteome to elucidate promising novel membrane candidates of heart function and potential therapeutic targets for heart disease. Our human and mouse cardiomyocyte membrane proteomes can be used as a reference for future studies aiming to identify and characterize heart disease markers and therapeutic targets for heart failure. Future deep membrane proteomics studies of cardiomyocytes with combined proteomic investigations of cardiac membrane protein PTM (post-translational modifications) will help complete the membrane proteome of the human heart and provide molecular mechanistic insights into the various disease etiologies of heart failure.

3.5 METHODS Cell-surface membrane proteome and identification of previously uncharacterized, evolutionarily conserved, cardiac-enriched membrane proteins. Previously, we enriched for membrane fractions from mouse neonatal ventricular cardiomyocytes and human fetal cardiomyocytes using cationic colloidal silica beads 10, 11. In those experiments, a membrane-depleted homogenate (removed after centrifugation at 1000 g for 5 min, labelled ‘H’) as well as a membrane-enriched fraction (pellet resuspended in NET buffer (400 mM NaCl, 25 mM Hepes pH7.4, and 1% Triton-X100) for 2 hours at 4 °C). were then subjected MudPIT 42 (Multidimensional Protein Identification Technology) analysis with stringent statistical filtering (<1% false discovery rate) as previously described 10. This analysis resulted in the identification of 2,762 human and 3,033 mouse proteins. The original raw mass spectrometry files were searched using X!TANDEM (version 2010.12.01.1; ftp://ftp.thegpm.org/projects/tandem/source/2010-12-01/) against human or mouse UniProt database (version 2011_03; ftp://ftp.uniprot.org/pub/databases/uniprot/previous_releases/release-2011_01/) with the following search parameters: Parent ion Δ-mass of 4 Da, fragment mass error of 0.4 Da, and complete carbaminomethyl modification of cysteine by iodoacetic acid. The resulting tryptic peptides matching these criteria made up the final list of identified proteins. Next, to identify conserved membrane-associated proteins, orthology mapping analysis of all identified human and mouse proteins was performed using the MGI orthology database (http://www.informatics.jax.org) taking into account only one-to-one orthology to yield the identification of human and mouse one-to-one orthologues 43. Comparative proteomics analysis using Qspec identified proteins that were enriched in the TX-100 fraction compared to the homogenate fraction, resulting in the identification of the membrane- enriched orthologue proteins (Supplementary Table 1).

The identified 555 proteins were converted to 550 as 5 IDs (TFRC_human P02786, mouse Q62351; TIMM50_human Q3ZCQ8, mouse Q9D880; PSPC1_human Q8WXF1, mouse Q8R326; GNAS_human P84996, mouse Q6R0H6; TMPO_humanP42167, mouse Q61029) were amalgamated into unique IDs. To rank the 550 membrane-enriched proteins identified, we first carried out prediction of transmembrane domains of the 550

identified proteins using TMHMM version 2 (http://www.cbs.dtu.dk/services/TMHMM) to identify membrane localized protein clusters. To further select novel cardiac membrane- enriched proteins, MGI phenotypic analysis (http://www.informatics.jax.org/phenotypes.shtml) was performed to individually map against our datasets to select and rank novel cardiac membrane proteins. Finally, we manually annotated the 550 protein clusters to select cardiac enriched transcripts by employing the publicly available microarray database, BioGPS microarray data (http://biogps.org/) to complete the rank ordered analysis of the 550 identified proteins. Excel output file containing full ranking criteria along with step-by-step ranking of the 550 identified proteins (labelled as “Ranking Criteria + Step-by-step ranking) was uploaded to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12).

Human myocardium procurement. Cardiac tissue samples were collected from donors with no history of heart disease and patients with dilated cardiomyopathy or ischemic heart disease during their cardiac transplantation through the HOPE program (Human Organ Procurement and Exchange program, University of Alberta, Edmonton, Canada) and the Human Explanted Heart Program (HELP) respectively under the approval of the Human Research Ethics Board at the University of Alberta 44, 45. Specifically, following cold cardioplegia, myocardial samples were excised from the left ventricular anterior free wall and immediately flash-frozen to be stored at −140 °C. Clinical review of 12-lead ECG and coronary angiogram were used to identify the infarct, peri-, and non-infarct regions of the left ventricle in ischemic heart disease patients. Patient samples from each category (Non-failing control; n = 3, ICM; n = 3, DCM; n = 3) were collected and used in this study. Patient baseline and clinical characteristics including age, gender, duration of heart failure, ejection fraction, hemoglobin levels, eGFR, and medication history are presented in Table 1.

Adult and neonatal mouse cardiomyocyte isolation. All experimental procedures involving animals were approved by the University of Toronto Animal Use and Care Committee; all animal experiments were conducted in accordance with the animal care guidelines. Adult mouse ventricular cardiomyocytes were isolated from adult male 6–8

week old CD1 mice (Jackson Labs) 46, 47. The outbred CD1 mouse strain was used due to their genetic diversity, larger litter size, and we have used these stains in our ongoing research programs where we have refined cardiomyocyte isolation from CD1 mouse hearts for in vitro genetic and small-molecule manipulation in our lab 47. Briefly, mice were euthanized with isoflurane and hearts perfused with EDTA buffer containing 15 μmol/L of blebbistatin with the heart clamped at the ascending aorta, followed by tissue digestion with collagenase type II (525 units/mL; Worthington Biochemical Corp.). The isolate was then filtered through a 70μm strainer and gravity-settled for 15 min to allow rod-shaped cardiomyocytes to settle by gravity and form a pellet. Dissociated ventricular cardiomyocytes were seeded on Geltrex-coated glass-bottom dishes for subsequent experiments. Primary neonatal mouse ventricular cardiomyocytes were isolated and cultured from 1–3 day old CD1 pups as described previously 46. Isolated ventricles were enzymatically digested in 0.05% trypsin overnight at 4 °C, followed by collagenase type II digestion (450 units/mL) at 37°C. Collected ventricular cardiomyocytes were seeded and maintained on gelatin-coated glass-bottom dishes for 48 hours prior to any experiments.

Publicly available RNASeq and proteomic datasets. Publicly available data sources of mRNA Affymetrix transcript data from the GEO Profiles in NCBI including several mouse and human heart diseases were downloaded and analyzed. FAM162A: 2d, 10d, 21d TAC (GPL81, 160218_at), MI (GPL5371, 20160), alpha-tropomyosin mild HCM (GPL339, 1451385_at), Emery-Dreifuss cardiomyopathy (GPL1261, 1443339_at), heart transplant (GPL96, 220942_x_at), Inflammatory DCM (GPL570, 220942_x_at), DCM (GPL6244, 8082066), and idiopathic and ischemic cardiomyopathies (GPL570, 220942_x_at); MCT1/SLC16A1: Severe DCM (GPL32, 101588_at), 2d, 10d, 21d TAC (GPL81, 101588_at), MI (GPL5371, 12952), alpha-tropomyosin mild HCM (GPL339, 1415802), Emery-Dreifuss (GPL1261, 1415802), heart transplant (GPL96, 202235), Inflammatory DCM (GPL570, 202235_at), DCM (GPL6244, 7918622), and idiopathic and ischemic cardiomyopathies (GPL570, 1557918_s_at); COX20: 2d, 10d, 21d TAC (GLP81, 160315_at), MI (GPL5371, 1309), alpha-tropomyosin mild HCM (GPL339, 1428619_at), Emery-Dreifuss (GPL1261, 1428619_at), heart transplant (GPL96, 206848_at), Inflammatory DCM (GPL570, 224820_at), DCM (GPL6244, 7911085), and idiopathic and

ischemic cardiomyopathies (GPL570, 224820_at). Data were reported as normalized hybridization signals. Comprehensive RNA-Seq based transcript levels of multiple mouse organs were obtained from the Human Protein Atlas 26. Data were reported as the abundance in ‘Transcripts per Million’ (TPM) in the heart, relative to the sum of the TPM values across all tissues 26. All source data input and analyzed output files (labelled as “Fig. 3_Cardiac-enriched + noMGI”, “Fig. 9 GEO profiles raw data”, “Supp. 1_HPM raw data”, “Supp. 2_Cardiac-enriched + MGI”, “Supp. 3_Noncardiac-enriched + noMGI”, “Supp. 4_Noncardiac-enriched + MGI”, “Supp. 5_HPM raw data”, and “Supp. 7_HPM raw data”) were uploaded to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12).

To assess the subcellular localization in subcellular organelles in the heart, we utilized immunohistochemical data available in the Human Protein Atlas 26 to investigate their protein distributions and expression patterns in healthy human ventricular tissues. Images for proteins were downloaded from the Atlas which contains images of histological sections from normal and cancer tissues. Specific antibodies were labeled with DAB (3,3’- diaminobenzidine) with a resulting brown staining indicative of antibody reactivity, with counterstained hematoxylin to enable visualization of nuclei. Images were all available at https://www.proteinatlas.org/humanproteome/tissue/heart 26, and images used in Fig. 4 were uploaded (labelled as “Fig. 4_HPA original uncropped images”) to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12).

Antibodies and reagents. Primary rabbit polyclonal anti-FAM162A antibody (PA5-24470, ThermoFisher; IB: 1:1000 dilution, IF: 1:800 dilution), rabbit polyclonal anti-COX20 antibody (25752-I-AP, Proteintech; IB: 1:1000 dilution, IF: 1:200 dilution), mouse monoclonal anti-MCT1 antibody (MA5-18288, ThermoFisher; IB: 1:1000 dilution, IF: 1:500 dilution) and rabbit polyclonal anti-MCT1 antibody (PA5-72957; ThermoFisher; IB: 1:1000 dilution, IF: 1:500 dilution) were used for immunofluorescence and immunoblot studies. These antibodies were carefully selected where previous validation data have been reported. Specifically, the MCT1 antibody was validated in CRISPR/Cas9 siRNA- mediated knockdown HEK293 cells where a complete depletion of MCT1 expression was observed in immunoblot analysis (ThermoFisher, P14612), the COX20 antibody was

validated in immunohistochemistry and immunofluorescence studies of various human tissues and cell lines (Proteintech, 25752-1-AP), and lastly, the FAM162A antibody was validated in immunoblot experiments where a clear protein band was detected at the expected molecular weight with minimal nonspecific bands (ThermoFisher, PA5-24470). The following immunogen sequences for the abovementioned commercially available antibodies are indicated in Fig. 6d–f: peptide sequence between 118–146 amino acids from the C-terminal region of human FAM162A; peptide fragment within amino acids 436– 500 of human MCT1; peptide sequence corresponding to amino acids 1–118 amino acids of human COX20. Additionally, primary mouse monoclonal anti-COX IV antibody (ab14744, abcam; IF: 1:1000 dilution) and rabbit polyclonal anti-Gαi antibody (#21006, NewEast Biosciences; IF: 1:800 dilution) were used for immunofluorescence co-staining studies. Lastly, mouse anti-alpha actin antibody (JLA20, DSHB) was used for immunoblot studies at 1:1000 dilution.

Similarly, antibody sensitivity and specificity are of essential importance in ensuring correct interpretations for expression patterns in healthy human myocardial tissues presented in Fig. 4. All antibodies used for the presented immunohistochemistry data were validated previously where a unique protein epitope signature tag with high specificity was selected for each antibody used in immunohistochemistry studies followed by immunoblot analysis from limited tissues (liver, tonsil) and cell lines (RT4, U-251 MG). Finally, the observed staining pattern was assigned a reliability score based on concordance with experimental protein characterization in the UniProt database 48, 49. In addition to the standard antibody validation procedures, enhanced antibody validation was performed for MCT1, TLN2, MSN, FLOT1, DYSF, AOC3, CTNND1, LAMA2, BGN, MYH7, TNNT2, and TNNI3 antibodies where genetic and orthogonal validations were performed to confirm antibody specificity for immunohistochemistry assays.

Immunoblot analysis. Protein lysates from mouse tissues were harvested in radioimmunoprecipitation assay buffer (RIPA) (50 mM Tris-HCl; pH7.4, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 150 mM NaCl, and 2 mM EDTA), supplemented with EDTA-free protease inhibitor cocktail (Roche), for 30 minutes on ice, spun down at

15,000 × g at 4 °C as previously described 50. Human myocardium samples were harvested in 8 M urea supplemented with protease and phosphatase inhibitor cocktails (Roche) and spun down at 15,000xg at 4°C for 15 minutes to remove tissue debris. Soluble fractions (supernatant) were then saved for immunoblotting studies. Standard SDS-PAGE chemiluminescent analysis was performed and tissue lysates were run on 10–12% polyacrylamide gels and transferred onto 0.45μm nitrocellulose membranes. Primary antibodies were added after blocking with 5% milk in 0.05% TBS-Tween20 for 1 hour and incubated at 4°C overnight. Secondary antibodies of anti-mouse (W402B, Promega) and anti-rabbit (W401B, Promega) HRP-conjugated IgG were used at 1:2500 dilution. Original uncropped immunoblots are available and provided as Supplemental Fig. 9. Immunodepletion experiments were performed by incubating primary antibodies with 10 μg of purified cMyc/DDK-tagged human FAM162A (Origene, LY415331), cMyc/DDK- tagged human MCT1 (Origene, LY401069), or cMyc/DDK-tagged human COX20 (Origene, LY405090) lysates in 10 mL of 5% milk (1μg /mL) for 3 hr at 4°C prior to probing the membrane at 4°C overnight.

Immunofluorescence and confocal microscopy. Acutely isolated adult mouse ventricular cardiomyocytes and cultured neonatal mouse ventricular cardiomyocytes were seeded on imaging glass-bottom dishes (MatTek, MA, USA) for immunofluorescence analysis. All immunofluorescence procedures were performed as described previously 46. Briefly, cardiomyocytes were fixed with 4% paraformaldehyde for 30 minutes for adult cardiomyocytes and 15 minutes for neonatal cardiomyocytes on ice followed by permeabilization with 0.5% Triton X-100, 0.2% Tween-20, and 5% FBS in PBS) for 30 minutes at room temperature. Cardiomyocytes were then incubated with primary antibodies in permeabilization overnight at 4°C. Samples were then incubated with fluorophore-conjugated secondary antibodies the next day (Alexa Fluor 488 or Alexa Fluor 647, Molecular Probes) at 1:1000 dilution. Samples were visualized using a Zeiss spinning-disk confocal microscope. Immunodepletion experiments were performed by incubating primary antibodies with 2 μg of purified cMyc/DDK-tagged human FAM162A (Origene, LY415331), cMyc/DDK-tagged human MCT1 (Origene, LY401069), or cMyc/DDK-tagged human COX20 (Origene, LY405090) lysates in 1 mL of

permeabilization buffer for 3 hr at 4°C prior to incubating cardiomyocytes with primary antibodies at 4°C overnight.

Image processing and analysis. Image processing and analysis of acquired images were performed using the Zeiss ZEN software (ZEN 2012 blue edition). Specifically, the unsharp masking tool was applied to all exported images to enhance subtler structures using ‘auto-thresholding’. All three-dimensional reconstructive analysis of z-stack images were performed using Imaris 8.1 (Bitplane, Switzerland). Three-dimensional reconstruction and co-localization models were built using surpass surface reconstruction, an intensity-based rendering of three-dimensional volume of z-stack images with ‘auto- thresholding’ set at Pearson’s coefficient p > 0.5.

Experimental design and statistical rationale. Searched mass spectrometry data was downloaded from Sharma et al. 10, whereby three independent biological replicates including three technical replicates within each biological replicate were used for further informatic analyses. For GEO RNA-Seq and immunoblot studies, statistical analyses were analyzed using the GraphPad Prism 8 software. Descriptive statistics are reported and shown as mean ± SEM. All bar graphs showing individual data points with error bars are graphically presented using the GraphPad Prism 8 software. Normally distributed data were analyzed using one-way ANOVA followed by post hoc Tukey’s multiple comparison test for each mean comparison. Experimental mean-fold protein intensities, relative to controls from triplicate assays, were considered different from controls at the p ≤ 0.05 significance level. For visual representation, a p value ≤ 0.05 is denoted by *; a p value ≤ 0.01 is denoted by **, and a p value ≤ 0.001 is denoted by ***. All immunoblots shown are representative from a minimum of three independent biological replicates; all original uncropped immunoblots are provided in Supplemental Fig. 9. All data figures and tables as well as all supporting information were uploaded to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12).

3.6 DATA AVAILABILITY All data records and supporting materials have been deposited in proteomeXchange and figshare. Specifically, MudPIT excel output files obtained from the Supplementary Table files of Sharma et al. 10 containing original unique peptides and total spectra detected files (labelled as “Sharma et al. ncomms_2015_peptides and spectra in human ventricular cardiomyocytes” and “Sharma et al. ncomms_2015_peptides and spectra in mouse ventricular cardiomyocytes”) were uploaded to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12) 51. Additionally, all excel source input and analyzed output files including original uncropped images were uploaded to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12) 51. Lastly, the original step-by-step raw mass spectrometry proteomics data for each biological replicate of mouse and human cardiomyocytes including both homogenate (labelled ‘H’) and membrane enriched (labelled TX-100 for human and NET for mouse) fractions were deposited to the ProteomeXchange Consortium with the dataset identifier (PXD017732) 52.

Received: 28 February 2020; Accepted: 17 November 2020; Published online: 01 December 2020

3.7 REFERENCES 1. Maltsev, V.A. & Lakatta, E.G. Normal heart rhythm is initiated and regulated by an intracellular calcium clock within pacemaker cells. Heart Lung Circ 16, 335-348 (2007). 2. Korzick, D.H. From syncitium to regulated pump: a cardiac muscle cellular update. Adv Physiol Educ 35, 22-27 (2011). 3. Savas, J.N., Stein, B.D., Wu, C.C. & Yates, J.R. Mass spectrometry accelerates membrane protein analysis. Trends Biochem Sci 36, 388-396 (2011). 4. Whitelegge, J.P. Integral membrane proteins and bilayer proteomics. Anal Chem 85, 2558-2568 (2013). 5. Wu, C.C. & Yates, J.R. The application of mass spectrometry to membrane proteomics. Nat Biotechnol 21, 262-267 (2003).

6. Eichacker, L.A. et al. Hiding behind hydrophobicity. Transmembrane segments in mass spectrometry. J Biol Chem 279, 50915-50922 (2004). 7. Speers, A.E., Blackler, A.R. & Wu, C.C. Shotgun analysis of integral membrane proteins facilitated by elevated temperature. Anal Chem 79, 4613-4620 (2007). 8. Chen, E.I., Cociorva, D., Norris, J.L. & Yates, J.R. Optimization of mass spectrometry-compatible surfactants for shotgun proteomics. J Proteome Res 6, 2529-2538 (2007). 9. Roe, A.T., Frisk, M. & Louch, W.E. Targeting cardiomyocyte Ca2+ homeostasis in heart failure. Curr Pharm Des 21, 431-448 (2015). 10. Sharma, P. et al. Evolutionarily conserved intercalated disc protein Tmem65 regulates cardiac conduction and connexin 43 function. Nat Commun 6, 8391 (2015). 11. Meyers, A.B., Towbin, A.J., Serai, S., Geller, J.I. & Podberesky, D.J. Characterization of pediatric liver lesions with gadoxetate disodium. Pediatr Radiol 41, 1183-1197 (2011). 12. Lee, S.H. et al. REEP5 depletion causes sarco-endoplasmic reticulum vacuolization and cardiac functional defects. Nat Commun 11, 965 (2020). 13. Hu, H.L. et al. Region-resolved proteomics profiling of monkey heart. J Cell Physiol 234, 13720-13734 (2019). 14. Doll, S. et al. Region and cell-type resolved quantitative proteomic map of the human heart. Nat Commun 8, 1469 (2017). 15. Oza, P., Jaspersen, S.L., Miele, A., Dekker, J. & Peterson, C.L. Mechanisms that regulate localization of a DNA double-strand break to the nuclear periphery. Genes Dev 23, 912-927 (2009). 16. Lee, J.S. et al. Nuclear lamin A/C deficiency induces defects in cell mechanics, polarization, and migration. Biophys J 93, 2542-2552 (2007). 17. Kim, M.S. et al. A draft map of the human proteome. Nature 509, 575-581 (2014). 18. Castellana, S. et al. Sudden death in mild hypertrophic cardiomyopathy with compound DSG2/DSC2/MYH6 mutations: Revisiting phenotype after genetic assessment in a master runner athlete. J Electrocardiol 53, 95-99 (2019). 19. Lorenzon, A. et al. Homozygous Desmocollin-2 Mutations and Arrhythmogenic Cardiomyopathy. Am J Cardiol 116, 1245-1251 (2015).

20. Hall, C.L. et al. Frequency of genetic variants associated with arrhythmogenic right ventricular cardiomyopathy in the genome aggregation database. Eur J Hum Genet 26, 1312-1318 (2018). 21. Liu, J.S., Fan, L.L., Li, J.J. & Xiang, R. Whole-Exome Sequencing Identifies a Novel Mutation of Desmocollin 2 in a Chinese Family With Arrhythmogenic Right Ventricular Cardiomyopathy. Am J Cardiol 119, 1485-1489 (2017). 22. Breher, S.S. et al. Popeye domain containing gene 2 (Popdc2) is a myocyte-specific differentiation marker during chick heart development. Dev Dyn 229, 695-702 (2004). 23. Kirchmaier, B.C. et al. The Popeye domain containing 2 (popdc2) gene in zebrafish is required for heart and skeletal muscle development. Dev Biol 363, 438-450 (2012). 24. Li, J. et al. Loss of αT-catenin alters the hybrid adhering junctions in the heart and leads to dilated cardiomyopathy and ventricular arrhythmia following acute ischemia. J Cell Sci 125, 1058-1067 (2012). 25. Vite, A., Zhang, C., Yi, R., Emms, S. & Radice, G.L. α-Catenin-dependent cytoskeletal tension controls Yap activity in the heart. Development 145 (2018). 26. Uhlén, M. et al. Proteomics. Tissue-based map of the human proteome. Science 347, 1260419 (2015). 27. Bonen, A. Lactate transporters (MCT proteins) in heart and skeletal muscles. Med Sci Sports Exerc 32, 778-789 (2000). 28. Poole, R.C., Sansom, C.E. & Halestrap, A.P. Studies of the membrane topology of the rat erythrocyte H+/lactate cotransporter (MCT1). Biochem J 320 (Pt 3), 817-824 (1996). 29. Lund, J., Aas, V., Tingstad, R.H., Van Hees, A. & Nikolic, N. Utilization of lactic acid in human myotubes and interplay with glucose and fatty acid metabolism. Sci Rep 8, 9814 (2018). 30. Bonen, A., Heynen, M. & Hatta, H. Distribution of monocarboxylate transporters MCT1-MCT8 in rat tissues and human skeletal muscle. Appl Physiol Nutr Metab 31, 31-39 (2006). 31. Bergeron, K.F. et al. Male-biased aganglionic megacolon in the TashT mouse line due to perturbation of silencer elements in a large gene desert of chromosome 10. PLoS Genet 11, e1005093 (2015).

32. Lee, M.J., Kim, J.Y., Suk, K. & Park, J.H. Identification of the hypoxia-inducible factor 1 alpha-responsive HGTD-P gene as a mediator in the mitochondrial apoptotic pathway. Mol Cell Biol 24, 3918-3927 (2004). 33. Kim, J.Y., Kim, S.M., Ko, J.H., Yim, J.H. & Park, J.H. Interaction of pro-apoptotic protein HGTD-P with heat shock protein 90 is required for induction of mitochondrial apoptotic cascades. FEBS Lett 580, 3270-3275 (2006). 34. Qu, Y. et al. MiR-139-5p inhibits HGTD-P and regulates neuronal apoptosis induced by hypoxia-ischemia in neonatal rats. Neurobiol Dis 63, 184-193 (2014). 35. Halestrap, A.P. The monocarboxylate transporter family--Structure and functional characterization. IUBMB Life 64, 1-9 (2012). 36. Jaswal, J.S., Keung, W., Wang, W., Ussher, J.R. & Lopaschuk, G.D. Targeting fatty acid and carbohydrate oxidation--a novel therapeutic intervention in the ischemic and failing heart. Biochim Biophys Acta 1813, 1333-1350 (2011). 37. Jóhannsson, E. et al. Upregulation of the cardiac monocarboxylate transporter MCT1 in a rat model of congestive heart failure. Circulation 104, 729-734 (2001). 38. Castresana, J., Lübben, M., Saraste, M. & Higgins, D.G. Evolution of cytochrome oxidase, an enzyme older than atmospheric oxygen. EMBO J 13, 2516-2525 (1994). 39. Elliott, L.E., Saracco, S.A. & Fox, T.D. Multiple roles of the Cox20 chaperone in assembly of Saccharomyces cerevisiae cytochrome c oxidase. Genetics 190, 559- 567 (2012). 40. Szklarczyk, R. et al. A mutation in the FAM36A gene, the human ortholog of COX20, impairs cytochrome c oxidase assembly and is associated with ataxia and muscle hypotonia. Hum Mol Genet 22, 656-667 (2013). 41. Doss, S. et al. Recessive dystonia-ataxia syndrome in a Turkish family caused by a COX20 (FAM36A) mutation. J Neurol 261, 207-212 (2014). 42. Kislinger, T. et al. Global survey of organ and organelle protein expression in mouse: combined proteomic and transcriptomic profiling. Cell 125, 173-186 (2006). 43. Cox, B. et al. Comparative systems biology of human and mouse as a tool to guide the modeling of human placental pathology. Mol Syst Biol 5, 279 (2009).

44. Sakamuri, S.S. et al. Differential impact of mechanical unloading on structural and nonstructural components of the extracellular matrix in advanced human heart failure. Transl Res 172, 30-44 (2016). 45. Jana, S. et al. Disparate Remodeling of the Extracellular Matrix and Proteoglycans in Failing Pediatric Versus Adult Hearts. J Am Heart Assoc 7, e010427 (2018). 46. Lee, S.H., Hadipour-Lakmehsari, S., Miyake, T. & Gramolini, A.O. Three-dimensional imaging reveals endo(sarco)plasmic reticulum-containing invaginations within the nucleoplasm of muscle. Am J Physiol Cell Physiol 314, C257-C267 (2018). 47. Callaghan, N.I. et al. Functinal culture and in vitro genetic and small-molecule manipulation of adult mouse cardiomyocytes. Comms Bio 3, 229 doi:10.1038/s42003-020-0946-9 (2020). 48. Fagerberg, L. et al. Contribution of antibody-based protein profiling to the human Chromosome-centric Proteome Project (C-HPP). J Proteome Res 12, 2439-2448 (2013). 49. Stadler, C. et al. RNA- and antibody-based profiling of the human proteome with focus on chromosome 19. J Proteome Res 13, 2019-2027 (2014). 50. Hadipour-Lakmehsari, S. et al. Nanoscale reorganization of sarcoplasmic reticulum in pressure-overload cardiac hypertrophy visualized by dSTORM. Sci Rep 9, 7867 (2019). 51. Lee, S.H. et al. Bioinformatic analysis of membrane and associated proteins in murine cardiomyocytes and human myocardium. figshare https://doi.org/10.6084/m9.figshare.11844972.v12 (2020). 52. Bioinformatic analysis of membrane and associated proteins in murine cardiomyocytes and human myocardium. ProteomeXchange Consortium PRIDE Archive https://www.ebi.ac.uk/pride/archive/projects/PXD017732/private (2020).

Acknowledgements This project was funded by the Ted Rogers Centre for Heart Research Innovation Fund to A.O.G; the Heart and Stroke Richard Lewar Centres of Excellence in Cardiovascular Research (A.O.G.); CIHR Grants to A.O.G (PJT-155921 and PJT-166118; MOP-106538; MOP-123320; GPG-102166); University of Toronto’s Medicine by Design initiative which receives funding from the Canada First Research Excellence Fund (A.O.G.); and Alberta Innovates Team Grant to G.Y.O. S-H.L was supported by a NSERC Postgraduate Scholarship, an Ontario Graduate Scholarship, and a Ted Rogers Centre for Heart Research Doctoral Fellowship. S.H-L was supported by a Canada Graduate Scholarship – Master’s Award from CIHR and an Ontario Graduate Scholarship. G.Y.O is a Canada Research Chair in Heart Failure.

Competing interests The authors declare no competing interests.

Additional information Supplementary information is available for this paper at https://doi.org/10.1038/s41597-020-00762-1.

Correspondence and requests for materials should be addressed to A.O.G.

Reprints and permissions information is available at www.nature.com/reprints.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are

included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

3.8 SUPPLEMENTARY INFORMATION

Lee et al.

SCIENTIFIC DATA | (2020)7:425 | https://doi.org/10.1038/s41597-020-00762-1 | www.nature.com/scientificdata

Supplemental Figure 1 Transcriptomic analysis of the 550 membrane and membrane-associated protein clusters across various tissues. (a) Heatmap showing unsupervised clustering of mRNA transcripts of the 550 membrane-associated proteins identified across clinically defined healthy human tissues; mRNA transcript data were obtained from Human Proteome Map. All source data input and normalized output files were uploaded to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12).

Supplemental Figure 2 Transcriptomic analysis of cardiomyocyte-enriched membrane associated proteins with previous cardiac MGI phenotype. Heatmaps showing mRNA transcript levels of 67 cardiomyocyte-enriched membrane proteins with previously identified cardiac MGI phenotype across clinically defined healthy human tissues; mRNA transcript data were obtained from Human Proteome Map and are presented according to their subcellular classifications in (a) mitochondrion, (b) plasma membrane, (c) other organelles (ER, golgi apparatus, peroxisomes, lysosomes), (d) cytosol, (e) nucleus, and (f) the secretory pathway. All source data input and normalized output files were uploaded to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12).

Supplemental Figure 3 Transcriptomic analysis of non-cardiomyocyte-enriched membrane associated proteins with no previous cardiac MGI phenotype. Heatmaps showing mRNA transcript levels of 241 non-cardiomyocyte-enriched membrane proteins with no previously identified cardiac MGI phenotype across clinically defined healthy human tissues; mRNA transcript data were obtained from Human Proteome Map and are presented according to their subcellular classifications in (a) mitochondrion, (b) plasma membrane, (c) other organelles (ER, golgi apparatus, peroxisomes, lysosomes), (d) cytosol, (e) nucleus, and (f) the secretory pathway. All source data input and normalized output files were uploaded to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12).

Supplemental Figure 4 Transcriptomic analysis of non-cardiomyocyte-enriched membrane associated proteins with previous cardiac MGI phenotype. Heatmaps showing mRNA transcript levels of 54 non-cardiomyocyte-enriched membrane proteins with previously identified cardiac MGI phenotype across clinically defined healthy human tissues; mRNA transcript data were obtained from Human Proteome Map and are presented according to their subcellular classifications in (a) mitochondrion, (b) plasma membrane, (c) other organelles (ER, golgi apparatus, peroxisomes, lysosomes), (d) cytosol, (e) nucleus, and (f) the secretory pathway. All source data input and normalized output files were uploaded to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12).

Supplemental Figure 5 FAM162A and COX20 share high degrees of homology throughout evolution. (a) A multi-species alignment of FAM162A from selected vertebrates shows 65% peptide conservation throughout evolution. (b) Phylogenetic analysis of the FAM162 family of proteins displays clustering of mammalian taxa with conservation throughout multiple species. (c) mRNA transcript levels of FAM162A and FAM162B obtained from Human Proteome Map across various healthy human organ tissues in transcriptomic analyses. (d) A multi-species alignment of COX20 from vertebrates demonstrates 69% peptide conservation throughout evolution. TM indicates the transmembrane domains identified. All source data input and normalized output files were uploaded to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12).

Supplemental Figure 6 (a) A multi-species alignment of MCT1 from selected vertebrates demonstrates 78% peptide conservation throughout evolution. TM indicates the transmembrane domains identified.

Supplemental Figure 7 Transcriptomic and phylogenetic analyses of the SLC16A/MCT family of proteins. (a) Phylogenetic analysis of the SLC16A/MCT family of proteins identifies phylogenetically distinct groups and shows evolutionary conservation within clustered mammalian taxa. (b) mRNA transcript levels of SLC16A/MCT family of proteins obtained from Human Proteome Map demonstrates dominant expression of SLC16A1/MCT1 across various healthy human organ tissues in transcriptomic analyses. All source data input and normalized output files were uploaded to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12).

Supplemental Figure 8 Co-immunofluorescence analysis of FAM162A, MCT1, and COX20 with known mitochondrial marker, COXIV and known plasma membrane protein, Gαi in isolated adult mouse cardiomyocytes. (a) Immunofluorescence analysis of FAM162A (green) co-stained with mitochondrial protein, COXIV (red) in acutely isolated adult mouse cardiomyocytes. Scale, 10 m. (b) Immunofluorescence analysis of MCT1 (green) co-stained with known plasma membrane protein, Gαi (red) in acutely isolated adult mouse cardiomyocytes. Scale, 10m. (c) Immunofluorescence analysis of COX20 (green) co-stained with mitochondrial protein, COXIV (red) in acutely

100

isolated adult mouse cardiomyocytes. Scale, 10m. All images shown are representative of approximately 30-40 total images captured per condition, n=3 independent biological replicates. All original uncropped microscopy images were uploaded to figshare (https://doi.org/10.6084/m9.figshare.11844972.v12).

101

102

103

104

105

Supplemental Figure 9 Original uncropped immunoblots.

106

CHAPTER 4

REEP5 Depletion causes Sarco-endoplasmic Reticulum Vacuolization and Cardiac Functional Defects

This chapter is a modified version of a manuscript that was published in Nature Communications. Lee et al. Nat Commun 11, 965 (2020).

This article is licensed under a Creative Commons Attribution 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Shin-Haw Lee1,2,16, Sina Hadipour-Lakmehsari1,2,16, Harsha R. Murthy3,4, Natalie Gibb3,4, Tetsuaki Miyake2,14, Allen C.T. Teng1,2, Jake Cosme1,2, Jessica C. Yu1,5, Mark Moon2,6, SangHyun Lim7,8, Victoria Wong7, Peter Liu6, Filio Billia9, Rodrigo Fernandez- Gonzalez1,5, Igor Stagljar7,8,10,11, Parveen Sharma2,15, Thomas Kislinger12,13, Ian C. Scott3,4 & Anthony O. Gramolini1,2

1Translational Biology and Engineering Program, Ted Rogers Centre for Heart Research, Toronto, ON M5G1M1, Canada. 2Department of Physiology, Faculty of Medicine, University of Toronto, Toronto, ON M5S1M8, Canada. 3Program in Developmental and Stem Cell Biology, The Hospital for Sick Children, Toronto, ON M5G1X8, Canada. 4Department of Molecular Genetics, Faculty of Medicine, University of Toronto, Toronto, ON M5S1M8, Canada. 5Institute of Biomaterials and Biomedical Engineering, University of Toronto, Toronto, ON M5S3G9, Canada. 6Ottawa Heart Institute, Ottawa, ON K1Y4W7, Canada. 7Donnelly Centre, University of Toronto, ON M5S1M8, Canada. 8Department of Biochemistry, Faculty of Medicine, University of Toronto, Toronto, Canada. 9Toronto General Research Institute, University Health Network, Toronto, ON M5G2C4, Canada. 10Department of Molecular Genetics, Faculty of Medicine, University of Toronto, Toronto, Canada. 11Mediterranean Institute for Life Sciences, Split, Croatia. 12Department of Medical Biophysics, University of Toronto, Toronto, Canada. 13Princess Margaret Cancer Centre, Toronto, ON M5G1L7, Canada. 14Present Address: Department of

107

Biology, Faculty of Science, York University, Toronto, ON M3J1P3, Canada. 15Present Address: Institute of Translational Medicine, University of Liverpool, Liverpool L693BX, UK. 16These authors contributed equally: Shin-Haw Lee, Sina Hadipour-Lakmehsari. email: [email protected]

NATURE COMMUNICATIONS | (2020)11:965 | https://doi.org/10.1038/s41467-019- 14143-9 | www.nature.com/naturecommunications

CONTRIBUTIONS

Shin-Haw Lee, Sina Hadipour-Lakmehsari, Harsha Murthy, Ian Scott, and Anthony Gramolini conceived and designed the study and wrote the manuscript. Experimentally, Shin-Haw Lee and Sina Hadipour-Lakmehsari performed all the bioinformatics analysis and initial molecular characterization of REEP5 presented in Figure 1 and Supp. Figure 1 and 2. Additionally, Shin-Haw Lee and Sina Hadipour-Lakmehsari performed all in vitro functional outcome experiments (MTT, ROS, JC-1, ER stress) presented in Figure 3 and all forward and reverse co-immunoprecipitation experiments presented in Figure 6 and Supp. Figure 5. Shin-Haw Lee performed and analyzed the Z-VAD treatment experiment in cultured cardiomyocytes presented in Supp. Figure 3. Sang Hyun Lim, Victoria Wong, and Igor Stagljar performed the membrane yeast two-hybrid experiments presented in Figure 1. Tetsuaki Miyake performed immunofluorescence analysis of REEP5 in isolated adult and neonatal mouse cardiomyocytes presented in Figure 2 and generated all truncated mutants of REEP5 and overexpressed them in C2C12 cells presented in Figure 5 and Supp. Figure 4. Jessica Yu and Rodrigo Fernandez-Gonzalez imaged and analyzed fractional shortening of adult mouse cardiomyocytes under different conditions presented in Figure 4 and provided SIESTA software and custom scripts for movie analysis of zebrafish heart rhythms presented in Figure 7 and Supp. Figure 6. Allen Teng performed and analyzed Ca2+ transient measurement experiment shown in Figure 4. Jake Cosme performed mass spectrometry analysis of CnVA-REEP5 pulldown experiments presented in Figure 6. Mark Moon and Peter Liu provided mouse TAC tissue samples and performed time-course immunoblot analysis of REEP5 in mouse TAC hearts

108

presented in Supp. Figure 8. Harsha Murthy, Natalie Gibb, and Ian Scott performed all zebrafish experiments in the current study. Specifically, Natalie Gibb generated and analyzed the reep5 morphant model using antisense reep5 morpholino sequence in zebrafish embryos presented in Supp. Figure 6. Harsha Murthy generated and analyzed both the reep5 crispants and CRISPR null reep5 mutants in zebrafish embryos using CRISPR/Cas9 presented in Figure 7, 8, and Supp. Figure 7. Jacqueline Pittman performed TEM imaging of isolated adult cardiomyocytes presented in Figure 4 and mouse heart tissues in Figure 9. Shin-Haw Lee performed all AAV9 REEP5 depletion experiments in mice presented in Figure 9 with expert assistance from Yuqing Zhu with in vivo echocardiographic imaging. Shin-Haw Lee prepared all figures and tables. Shin- Haw Lee, Sina Hadipour-Lakmehsari, Thomas Kislinger, Parveen Sharma, Ian Scott, and Anthony Gramolini edited and revised the manuscript. All authors reviewed and approved the final version of the manuscript. All persons listed as authors qualify for authorship, and all those who qualify for authorship are listed.

109

4.1 ABSTRACT

The sarco-endoplasmic reticulum (SR/ER) plays an important role in the development and progression of many heart diseases. However, many aspects of its structural organization remain largely unknown, particularly in cells with a highly differentiated SR/ER network. Here, we report a cardiac enriched, SR/ER membrane protein, REEP5 that is centrally involved in regulating SR/ER organization and cellular stress responses in cardiac myocytes. In vitro REEP5 depletion in mouse cardiac myocytes results in SR/ER membrane destabilization and luminal vacuolization along with decreased myocyte contractility and disrupted Ca2+ cycling. Further, in vivo CRISPR/Cas9-mediated REEP5 loss-of-function zebrafish mutants show sensitized cardiac dysfunction upon short-term verapamil treatment. Additionally, in vivo adeno-associated viral (AAV9)- induced REEP5 depletion in the mouse demonstrates cardiac dysfunction. These results demonstrate the critical role of REEP5 in SR/ER organization and function as well as normal heart function and development.

110

4.2 INTRODUCTION The sarco-endoplasmic reticulum (SR/ER) is a multifunctional organelle responsible for many essential cellular processes in eukaryotic cells, including protein translation, lipid synthesis 1, Ca2+ cycling 2, protein trafficking 3, and organelle–organelle communication 4. The ER is a lipid-bilayer extension of the outer nuclear membrane consisting of a continuous peripheral ER network of tubules and interspersed sheets 5. ER tubules are generated mainly by stabilizing membrane curvature, and previous studies have identified a few key evolutionary conserved protein families involved in the formation of these tubular structures 6. However, many aspects of its structural organization and function remain largely unknown, particularly in highly differentiated muscle cells. Members of the reticulon and Yop1p/DP1/REEP protein families contain a reticulon-homology domain (RHD) which is essential for inducing and stabilizing high membrane curvature in cross- sections of ER tubules 7–9. In addition to membrane curvature stabilization, ER tubules are also stabilized by forming a characteristic polygonal network through membrane fusion mediated by the atlastin family of dynamin-related GTPases 10, 11.

RHD proteins contain two closely-spaced hairpin integral membrane structures (~35 amino acids in length), which have been proposed to form arc-shaped oligomers, thus essential for bending and stabilizing membrane curvature around the tubules 7, 12. Of all RHD proteins, Yop1p from Saccharomyces cerevisiae has been the most well-studied. Vertebrate homologs of Yop1p are the family of receptor expression-enhancing proteins (REEPs) and previous studies demonstrate their vital roles in trafficking the odorant receptor 13 and G-protein coupled receptors to the plasma membrane 14. Despite the association of REEPs’ RHD domains to ER network formation, the precise role of REEPs in ER formation, maintenance, and responses to ER stress remains poorly understood. So far, six mammalian REEP homologs have been identified, REEP1 and REEP2 are neuro-enriched in mice 15 and have been linked to hereditary spastic paraplegia in patients and transgenic mice 16, 17. REEP3 and REEP4 are required for mitotic spindle organization in proliferative cells 18. Mutations in REEP6 have been linked to human retinopathies 19, 20. The role of REEP5, in comparison, remains largely unknown.

111

Instabilities in ER structure and function lead to ER stress, unfolded protein response, ER-associated degradation, and autophagy 21. In excitable muscle cells, their ER structures have adapted to handle a large concentration of Ca2+, important for regulated release of Ca2+ into the cytoplasm for muscle contraction. This specialized smooth ER, termed the SR, evolved to function in striated muscle 22. However, differences in protein expression and function between the ER and SR have not been fully determined, resulting in poor characterization and understanding of the formation and function of SR in muscle 22. The SR has been loosely divided into at least two structural and functional domains termed the longitudinal SR and the junctional SR 23. Furthermore, different regions of the SR have specialized to perform specific functions with respect to the control of the excitation–contraction coupling 24. It is recognized in patients and animals that longitudinal and junctional SR undergo significant transformation following heart failure 25, 26. While a great deal is known about SR structure and function in terms of cardiac muscle contraction, considerably less is understood about how the SR is formed and maintained.

4.3 RESULTS REEP5 is a conserved cardiac-enriched membrane protein. Our previous proteomic experiments of mouse and human cardiac myocytes, integrated with microarray tissue expression profiles and phenotype ontology information identified poorly characterized, evolutionary conserved, cardiac-enriched membrane proteins 27. Rank-ordered evaluation of these protein candidates identified that REEP5 was one of these most highly ranked proteins. Accordingly, we investigated the role of REEP5 in the cardiac myocyte.

Given its identification in both mouse and human myocyte proteomic membrane isolations 27, we first performed a detailed multispecies amino-acid sequence analysis of REEP5 which showed 96% homology between human and mouse REEP5 and 73% between human and zebrafish (Fig. 1a). Bioinformatics analysis of extensive phylogeny further demonstrated significant clustering of mammalian REEP5 within the REEP family (Supplementary Fig. 1).

112

To verify human REEP5 membrane topology, we used the multi-algorithm prediction tool TOPCONS (http://topcons.cbr.su.se). All six prediction algorithms within this tool determined that the N-terminus was cytosolic for human REEP5, with 2 or 4 predicted transmembrane helices (Fig. 1b). To confirm these predictions for human REEP5, we employed a membrane yeast two-hybrid (MYTH) system, as previously described 27, 28 (Fig. 1c); which takes advantage of the ability of ubiquitin fragments (NubG and NubI) reconstitution, N-terminal TF transcription factor tagged (TF-Cub) bait protein and ubiquitin fragments tagged (NubI) prey protein were tested for protein–protein interactions. To determine REEP5 membrane topology at the ER, a noninteracting yeast ER integral membrane protein Ost1p was fused at its C-terminus with NubI and NubG (mutated form) which exposed its C-terminus to the cytosol. N-terminally tagged TF-Cub-REEP5 interacted with Ost1p-NubI but not with Ost1p-NubG, indicating that the N-terminus of REEP5 resides in the cytosol (Fig. 1c). Thus, REEP5 consists of four hydrophobic, hairpin transmembrane domains connected by a hydrophilic segment with both the N- and C- termini facing the cytosol (Fig. 1d). This membrane topology model is entirely consistent with models of Yop1p and other REEP homologs 6, 29, 30.

To assess cardiac enrichment of REEP5, we used publicly available comprehensive tissue specific human RNA-Seq datasets (Human Proteome Map) to determine REEP5 expression levels across different tissues 31. These data demonstrate broad REEP5 expression across several different tissues at both fetal and adult stages, with an apparent increase in expression during development. Importantly, of all the REEP family members, REEP5 was significantly enriched in the fetal and adult heart despite its broad expression across many mouse tissues (Fig. 1e). Immunoblots of diverse tissues in the adult mouse using solubilization buffer containing 6 M urea further demonstrated highest expression in the ventricle (Fig. 1f). Abundant expression was also detected in skeletal muscle (gastrocnemius) and kidney. Apparent REEP5 monomers (17 kDa) and trimers (51 kDa) were predominantly expressed in the ventricle, while REEP5 dimers (34 kDa) were seemingly the predominant form in the atria, suggesting functional difference in REEP5 oligomerization status. Since antibody specificity must always be evaluated

113

carefully, anti-REEP5 antibody specificity was confirmed by blocking antibody with 10μg of a bacterially expressed, purified 6xHis-tagged REEP5 (Fig. 1g).

We next queried publicly available GEO RNA-Seq datasets containing data for human and mouse cardiovascular diseases to determine changes in REEP5 expression. As shown in Fig. 1h, REEP5 levels changed dramatically with several cardiac diseases. A marked decrease in REEP5 mRNA levels was seen in severe dilated cardiomyopathy (DCM) in the mouse and in one human dataset, albeit not significantly. Myocardial infarction (MI) in the mouse and ischemic cardiomyopathy in the human showed a downregulation of REEP5. Similarly, pressure overload-induced heart failure via transverse aortic constriction (TAC) led to a downregulation of REEP5 at 2 and 10 days post-surgery; however, a moderate increase was seen after 21 days. By contrast, inflammatory DCM in humans showed an upregulation of REEP5 RNA levels whereas rejected cardiac allografts in human showed the greatest decrease in REEP5 levels. These results showed that REEP5 levels appear to be dynamic under certain cardiac conditions and suggest a potential role of REEP5 in the progression of heart diseases, although the specific mechanisms and directions of changes are not well defined.

Next, we performed immunoblots using mouse cardiac tissue from a phospholamban (PLN) R9C DCM model 32, 4-week post TAC heart failure model 33, and 2-week post LAD ligation MI model 34. A marked increase in REEP5 levels was seen in the 4-week TAC mouse hearts whereas REEP5 protein levels were decreased in the LAD ligation MI model (Fig. 1i). Brain natriuretic peptide (BNP), a robust biomarker for heart failure, dramatically increased in the MI model. PLN R9C mutant DCM mice hearts showed decreased REEP5 levels. BNP levels were very slightly increased in the R9C hearts. Immunoblotting analysis was also carried out in human cardiac samples of idiopathic and ischemic cardiomyopathy. In idiopathic cardiomyopathy, REEP5 and myosin heavy chain β (MF20) levels were elevated. In contrast, REEP5 was downregulated in ischemic cardiomyopathy; in agreement with the GEO data (Fig. 1j).

114

115

Fig. 1 REEP5 is an evolutionarily conserved, muscle-enriched membrane protein. a A multispecies alignment of REEP5 from vertebrates. b Prediction of human REEP5 protein topography generated by TOPCONS. c Membrane yeast two-hybrid assay of REEP5 membrane topology. SD-WL is yeast media that lacks tryptophan and leucine and selects for cells that contain both bait and prey plasmids. SD-WLAH +10 nM 3-AT is yeast media that lacks tryptophan, leucine, adenine, and histidine and selects for cells in which bait and prey are interacting. d Predicted membrane topology model of human REEP5 generated by modification of a T(E)Xtopo output in Protter (http://wlab.ethz.ch/protter). e REEP5 mRNA transcript levels obtained from Human Protein Atlas across various mouse tissues. f Immunoblot of REEP5 protein expression in mouse tissues. Asterisks to the right indicate the number of predicted REEP5 oligomers detected based on anticipated molecular weight. g Immunodepletion of REEP5 antigens with bacterially expressed 6xHis-REEP5. h GEO RNA-seq datasets demonstrate changes in REEP5 expression across various mouse and human cardiovascular diseases; * indicates a statistically significant p < 0.05 in a Tukey’s multiple comparison analysis. Data are presented as mean ± SEM with n = 3 biologically independent measurements. i Immunoblot analysis of REEP5 and BNP expression from cardiac tissue of hypertrophic cardiomyopathy (HCM), myocardial infarction (MI), and dilated cardiomyopathy (DCM) mouse models. j Immunoblot analysis of REEP5, BNP, and MHC expression from human cardiac samples of normal, idiopathic, and ischemic cardiomyopathy. Source data containing original uncropped immunoblots are provided as a Source Data file.

REEP5 localizes to the myocyte SR and j-SR membrane. We next performed IF analysis of endogenous REEP5 in cultured mouse neonatal cardiac myocytes (CMNCs) (Fig. 2a) and acutely dissociated primary adult mouse cardiac myocytes (Fig. 2b, c). As shown in Fig. 2a, REEP5 demonstrated a consistent SR/ER striated staining pattern in CMNCs, with sarcomeric doublet REEP5 patterns observed flanking the z-disks highlighted by F-actin staining, suggesting its localization to the junctional-SR. IF co- staining between REEP5 and SERCA2, a known cardiac SR protein, demonstrated a very high degree of co-localization between the two proteins (Fig. 2b). Further orthogonal and three-dimensional reconstructive analyses confirmed strong co-localization with SERCA2 (Pearson co-localization coefficient of 0.59±0.02; data presented as mean±SEM) in isolated adult ventricular cardiac myocytes (Fig. 2c), indicating REEP5 localization to the SR in cardiac myocytes. In adult mouse ventricular cardiac myocytes, three-dimensional reconstructive analysis demonstrated a Pearson’s co-localization coefficient of 0.64 ±

116

117

Fig. 2 REEP5 expression shows consistent SR staining pattern in cardiac myocytes. a Immunofluorescence analysis of endogenous REEP5 expression (green) and phalloidin-rhodamine staining (red) in CMNCs. Right panel, line-scan analysis (from yellow line) demonstrates SR expression pattern in CMNCs. Scale, 10 μm. b Immunofluorescence of isolated adult mouse cardiac myocytes with REEP5 (blue) co- stained with SERCA2 (green) and phalloidin-rhodamine (red). Scale, 20 μm (left panel), 10 μm (right panel). c Orthogonal projection, three-dimensional reconstructive and line- scan analyses demonstrates co-localization between REEP5, SERCA2, and phalloidin signal. (Left panel): the top panels represent cell imaging in the x–z plane, while side panels represent cell imaging in the y–z plane. All images shown are representative of approximately 40–50 total images captured per condition, n = 3 independent biological replicates.

0.04 between REEP5 and α-actinin, indicating a strong co-localization (Supplementary Fig. 2a, b). Further statistical analysis revealed an average of 99.8 ± 0.1% of α-actinin co- localized with REEP5 while 58.4 ± 1.0% of REEP5 co-localized with α-actinin, suggesting REEP5 localization to the j-SR, but also in patterns indicative of longitudinal SR. Additional analyses revealed strong co-localization between REEP5 and Ryanodine Receptor 2 (RyR2) (Pearson coefficient of 0.76 ± 0.04), and between REEP5 and triadin, with a Pearson coefficient of 0.68 ± 0.12 (Supplementary Fig. 2c, d). These results validate REEP5’s localization to the j-SR membrane that is closely tethered to the cell membrane and contractile machinery in adult striated muscles.

In vitro REEP5 depletion results in SR dysfunction. To characterize the function of REEP5 in vitro, CMNCs were transduced with a lentiviral REEP5-shRNA construct. Temporal analysis of REEP5 protein levels by immunoblot showed ~60% knockdown was achieved 48h post transduction with knockdown of 70% following 96h (Fig. 3a, b). Confocal imaging of transduced cells revealed pronounced vacuolization and disorganization of the SR/ER visualized with ER-Tracker (Fig. 3c). Quantitatively, vacuolization of the SR/ER was observed as soon as 24h post transduction and observed in nearly all cells at 96h. To confirm that the observed vacuolization was not a result simply of dying cells and cell death, we treated cultures with the caspase inhibitor, 100μM z-vad-fmk, as described previously 35. Under these conditions, luminal vacuoles were

118

119

Fig. 3 In vitro REEP5 depletion in cardiac myocytes results in SR/ER membrane destabilization and dysfunction. a Immunoblotting analysis of lentiviral-mediated REEP5 depletion in CMNCs at 0, 24, 48, 72, 96 h post transduction. b Quantitative analysis of REEP5 expression levels showed 60% reduction 48 h after REEP5 shRNA transduction. Quantification was done under ×40 objective lens and approximately 40–50 cells were scored for each experimental condition, n = 3 independent biological replicates; data are presented as mean ± SEM. c Confocal imaging of CMNCs stained with ER- tracker showed SR/ER vacuoles (yellow arrows) 48 h post viral transduction. Vacuoles were observed starting 24 h post transduction, peaking after 96 h post lentiviral infection with REEP5 shRNA. Quantification was done under ×40 objective lens and approximately 30–40 cells were scored for each experimental condition, n = 3 independent biological replicates; data are presented as mean ± SEM. d Confocal imaging of CMNCs stained with CellROX oxidative stress dye 48 h post viral transduction with REEP5 shRNA. Scale, 10 μm. Spectrophotometric analysis showed a marked increase in ROS levels following shRNA-mediated REEP5 depletion in the presence and absence of tunicamycin, n = 40– 50 cells examined under ×40 objective, n = 3 independent biological replicates; data are presented as mean ± SEM. e Cardiac myocyte cell viability levels 48 h post viral transduction with REEP5 shRNA measured by MTT assay, n = 3 independent biological replicates; data are presented as mean ± SEM. f Immunoblotting analysis of REEP5, ER stress markers (GRp78, GRp94, and ATF4), and ER-dependent apoptosis marker (caspase 12) expression levels upon REEP5 depletion in the presence or absence of tunicamycin. g Confocal imaging of CMNCs stained with mitochondrial membrane potential dye JC-1 48 h post viral transduction with REEP5 shRNA. Scale, 10 μm, n = 40– 50 cells examined under ×40 objective lens over 3 independent experiments. Asterisks indicate a statistically significant p value in a Tukey’s multiple comparison analysis where *p < 0.05, **p < 0.01, and ***p < 0.001; data are presented as mean ± SEM. Source data containing original uncropped immunoblots are provided as a Source Data file. All images shown are representative of approximately 30–40 total images captured per condition. still observed in REEP5 shRNA-infected CMNCs in the presence of z-vad-fmk, ruling out that these morphological changes are simply a result of cell death (Supplementary Fig. 3), rather they appear upstream of any activated caspase activities.

ER stress plays important roles in contributing to cardiac diseases 36. To determine if REEP5 depletion-induced structural abnormalities were related to functional impairment, we first assayed for ER stress activity and cell viability. CMNCs were transduced with REEP5-shRNA and intracellular oxidative stress was measured via staining for reactive oxygen species (ROS), reflective of ER stress (Fig. 3d). CellRox fluorescence

120

demonstrated a twofold increase in ROS levels with REEP5 depletion (Fig. 3d). Tunicamycin (5 μg/ml) was used as a cell stress inducer, which increased ROS levels further compared to wild-type CMNCs treated with tunicamycin. Further analysis of ER stress markers, GRp78, GPp94, and ATF4 indicated increased ER stress with REEP5 depletion with/without tunicamycin (Fig. 3f). To assess cell viability, MTT assays were conducted to measure cell metabolic state upon REEP5 depletion and demonstrated a significant decrease in cell viability in cardiac myocytes upon REEP5 depletion (Fig. 3e). Similarly, increased cleaved caspase-12 levels upon REEP5 depletion were observed, indicating activation of ER-dependent apoptosis 37 (Fig. 3f). In addition, we performed mitochondrial membrane potential (MMP) measurements in cardiac myocytes and showed that REEP5 depletion resulted in significant dissipation of the mitochondrial inner membrane electrochemical potential, indicating impaired cellular health and function in the cardiac myocyte (Fig. 3g).

REEP5 depletion leads to compromised myocyte contractility. IF of virally mediated REEP5 depletion in adult cardiac myocytes revealed prominent SR vacuoles as shown by REEP5 and RyR2 immunostaining (Fig. 4a). Three-dimensional reconstruction of the SR marked by RyR2 staining further demonstrated the vacuolated SR in REEP5 depleted myocytes as opposed to the striated, interconnected SR network in the control myocytes (Fig. 4b). Transmission electron microscopy (TEM) analysis revealed that REEP5 depleted adult cardiac myocytes showed significant disruption of SR integrity, with deformed SR membranes and apparent vacuoles, along with deformed cardiac t- tubules, compared to control cells (Fig. 4c). Next, we performed optical imaging edge analysis of REEP5 depleted adult cardiac myocytes to measure myocyte contractility of spontaneously contracting cells (Fig. 4d). REEP5 depleted myocytes demonstrated a significant impairment in fractional shortening of the myocytes (0.3 ± 0.1% compared to 12.4 ± 1.2% fractional shortening per contractile event in the control myocytes; data presented as mean ± SEM), suggesting severe functional impairment of myocyte contractility (Fig. 4d, right) with a significant reduction in the recorded events (0.27 ± 0.07 Hz in scram compared to average null values (below detectable threshold) in REEP5 KD;

121

122

Fig. 4 In vitro REEP5 depletion leads to SR vacuolization and sarcomeric dysfunction in adult mouse cardiac myocytes. a Immunofluorescence of adult mouse cardiac myocytes stained with REEP5 and RyR2 48 h post viral transduction with REEP5 shRNA. Scale, 20 μm. b Three-dimensional reconstruction analysis of RyR2 staining of the SR in Scram and REEP5-depleted myocytes. c Transmission electron microscopy in REEP5-depleted adult mouse cardiac myocytes revealed SR vacuoles and disrupted SR membranes compared to scram controls. M mitochondria, SR sarcoplasmic reticulum, TT T-tubule, V vacuoles. Scale, 0.5 nm. d Optical measurements of spontaneous myocyte contractility in scram and REEP5-depleted adult mouse cardiac myocytes revealed a significant decrease in both fractional shortening measurements and frequency in REEP5-depleted myocytes. Left: still images of representative cardiac myocytes. Scale, 20 μm. Red line indicates region of image used to generate kymographs shown above contractile pulses tracings. n = 30 cells examined over three independent experiments; data are presented as mean ± SEM. e Ca2+ imaging of myocytes showing the frequency of Ca2+ waves and Ca2+ transients amplitude 48 h post viral transduction with REEP5 shRNA, n = 40–50 cells examined per condition over 3 independent experiments. Asterisks indicate a statistically significant p value in a Tukey’s multiple comparison analysis where *p < 0.05 and ***p < 0.001; data are presented as mean ± SEM. All images shown are representative of approximately 30–40 total images captured per condition. p = 0.0194). In addition, Ca2+ sensor Fura2-AM measurements were performed to characterize Ca2+ cycling in REEP5 depleted myocytes (Fig. 4e). A significant decrease in the number of spontaneous Ca2+ transients (1.07 ± 0.05% in scram vs. 0.47 ± 0.03% in REEP5 KD; p = 0.0006) and an increased peak amplitude (25 ± 1.5% in scram controls vs. 58.6 ± 1.8% in REEP5 KD; p = 0.0002) of these transients were observed in REEP5 depleted myocytes (Fig. 4e, right).

The C-terminal domain of REEP5 contributes to SR integrity. To study the biochemical properties of REEP5, we created two tagged REEP5 constructs (Fig. 5a). REEP5 was fused to a V5 tag followed by 6xHis tag for affinity purifications using Ni-NTA beads; alternatively, CnVA-tagged 38 (StrepII-6xHis, 3xFlag) REEP5 was generated and used. Immunoblot analysis showed robust expression of both constructs in HEK293 cells (Fig. 5b, d). Interestingly, addition of increasing dithiothreitol (DTT) concentrations led to dose-dependent decrease of observed REEP5 dimers (Fig. 5c), suggesting disulfide bridge(s) may be required for this dimerization. Moreover, overexpression of recombinant proteins CnVA-REEP5 and REEP5-V5-6xHis followed by immunoprecipitation using anti-

123

V5 antibody identified CnVA-REEP5 in the elution, indicating homodimerization of REEP5 molecules (Fig. 5d). Our results are consistent with previous yeast experiments that showed the formation of DP1/Yop1p immobile oligomers 39. This, however, does not rule out the presence of heterodimers of REEP5 with other proteins, or other REEP isoforms. Furthermore, expression of a fluorescently tagged dimerized REEP5 construct (REEP5-

REEP5-EYFP) in C2C12 myoblasts marked the ER network more robustly, suggesting functionality of REEP5 dimers (Supplementary Fig. 4a).

The N- and C-termini cytosolic domains represent the major differences between REEP1- 4 and REEP5-6 subfamilies 30. To determine the importance of various domains in REEP5 to its function, we constructed various EYFP tagged REEP5 truncated mutants (Fig. 5g). Immunoblotting revealed that stable, sodium dodecyl sulfate (SDS)-resistant dimers were detected in C2C12 myoblasts expressed with REEP5-EYFP, REEP5X2-EYFP, and Δ1–36 REEP5-EYFP (Fig. 5e). Interestingly, loss of REEP5 dimers was observed in cells expressing Δ114–189 REEP5-EYFP and Δ1–36/114–189 REEP5-EYFP, suggesting that the cytosolic carboxyl terminal domain is required for the formation of stable dimers. Expression of the carboxyl-terminal truncated mutant of REEP5 (Δ114–189 REEP5-

EYFP) in C2C12 myoblasts led to a vacuolated ER network as observed with REEP5 depletion in cardiac myocytes (Figs. 3c and 4a). To confirm the origins of the vacuoles, recombinant mCherry fusion proteins with mitochondria targeting peptide (mCherry-mito) or with a ER lumen retention peptide (mCherry-KDEL) were co-expressed with Δ114–189

REEP5-EYFP in C2C12 myoblasts (Fig. 5h). Co-expression of mCherry-KDEL, as an ER– luminal protein, occupied the entire vacuoles generated by expression of Δ114–189 REEP5-EYFP, indicating that the observed vacuoles originate from the ER membrane. Consistent with this hypothesis, expression of Δ1–36 REEP5-EYFP did not affect the morphology of the ER and co-localized strongly with mCherry-KDEL, Δ114–189 REEP5- EYFP and Δ1–36/114–189 REEP5-EYFP significantly disrupted the peripheral ER morphology and caused prominent luminal swelling and vacuolization (Fig. 5i). To test whether disruption of the ER network affects microtubule dynamics, we co-expressed

Δ114–189 REEP5-EYFP and mCherry-α-Tubulin in C2C12 myoblasts and showed intact

124

125

Fig. 5 The C-terminal cytosolic domain of REEP5 is required for stabilizing SR/ER morphology. a Schematic diagrams of REEP5 constructs used for REEP5 dimerization study shown in Fig. 5b–d. b Immunoblot analysis of endogenous REEP5 monomer (~17 kDa) and REEP5 dimer (~34 kDa), and exogenous REEP5 monomer (~22 kDa) and exogenous REEP5 dimer (~44 kDa) in HEK293 transfected with REEP5-V5-6xHis construct. c Immunoblots of REEP5 dimer dissociation in response to increasing DTT concentration. Diamonds to the left indicate the detection of endogenous (orange) and exogenous (blue) REEP5 monomers and dimers. d Co-immunoprecipitation assays in HEK293 cells. REEP5-V5 and CnVA-REEP5 were transfected into cells and precipitated with V5 or flag antibody. Left, control immunoblot experiments. Right panel, CnVA-REEP5 was immunoprecipitated via anti-V5 antibody in REEP5-V5-6xHis transfected HEK293 cells. e Immunoblot analysis of C2C12 cells transfected with REEP5 truncation mutants shows depletion of REEP5 dimers in the absence of the carboxyl terminal domain. f Confocal imaging analysis of EYFP (green) and DAPI (blue) staining shows truncation of the carboxyl terminal domain of REEP5 (Δ114–189) causes ER luminal vacuolization in transfected C2C12 myoblasts. Scale, 10 μm. Right, image intensity analysis along axis. g Schematic diagram of truncated REEP5 mutant constructs fused with GFP. Amino acid sequences are shown. h Live-cell confocal imaging of C2C12 myoblasts expressing the carboxyl-terminus truncated mutant of REEP5 (Δ114–189) and recombinant mCherry- fused mitochondrial targeting signal (mCherry-mito) or luminal ER marker (mCherry- KDEL). Scale, 10 μm. i Live-cell confocal imaging of C2C12 myoblasts expressing four truncated REEP5 mutants co-expressed with mCherry-KDEL suggests the importance of the C-terminal cytosolic domain of REEP5 in stabilizing ER membrane curvatures. Scale, 10 μm. Source data containing original uncropped immunoblots are provided as a Source Data file. All images shown are representative of approximately 40–50 total images captured per condition, n = 3 independent biological replicates. microtubule organization in Δ114–189 REEP5-EYFP transfected cells, suggesting the SR/ER shaping role of REEP5 is independent of microtubule polymerization (Supplementary Fig. 4b).

REEP5 interacts with known SR/ER shaping proteins. Mass spectrometry-based studies identified REEP5-associated proteins as nickel-His purification of CnVA-REEP5 in HEK293 (Fig. 6a) resulted in the identification of several members of the reticulon (RTN) and atlastin (ATL) families of proteins with a marked increased fold-change enrichment in purified REEP5-associated membrane complexes (Fig. 6b). Assessment of ACTC1, MYL6, and GAPDH as negative controls in both conditions demonstrated similar average

126

peptide intensity with minimal fold-change difference detected. Notably, we identified RTN4 (Nogo-A/B), RTN3, and ATL2 solely in the REEP5 transfected cells, while ATL3 and CKAP4, were detected under non-transfected and transfected conditions, although at a 15-fold and 5-fold enrichment in the REEP5 transfected cells were detected, respectively.

To identify which RTN, ATL, and CKAP isoforms are highly expressed in cardiac myocytes, we again analyzed publicly available transcriptomic datasets from the Human Proteome Map 31 for RTN, ATL, and CKAP families of proteins. Of particular interest, Nogo-A/B/RTN4, ATL3, and CKAP4/Climp-63 were significantly enriched in the fetal and adult heart, similar to REEP5 (Fig. 6c).

Previous experiments demonstrated the interaction between RTN1 and DP1/Yop1p in yeast and showed subsequent oligomer formation was essential for tubular ER localization and ER tubule formation 39. Moreover, immunoblots of the REEP5 nickel-His purification eluate confirmed the presence of RTN4, ATL3, and CKAP4 (Fig. 6d). Here, we overexpressed CnVA-REEP5 and Myc-RTN4 (Nogo-B) plasmids in HEK cells (Fig. 6e), followed by co-immunoprecipitation with anti-REEP5 (Fig. 6f, left) or anti-Nogo-A/B antibodies (Fig. 6f, right). In both experiments, Nogo isoforms were coprecipitated with REEP5 confirming the physical interaction between these two proteins, at least in an overexpression system (Fig. 6f). Next, co-immunoprecipitations using mouse ventricular membrane fractions (microsome) were performed. Immunoprecipitating REEP5, followed by immunoblots, showed REEP5 co-precipitated Nogo-A, Nogo-B, ATL3, an ATL3 apparent dimer, and CKAP4 in these lysates (Fig. 6g). REEP5 in a dimer, and oligomer state were detected (Fig. 6g, lower panel). Co-immunoprecipitations in a reverse strategy were also performed (Fig. 6h–j). In these studies, we showed that all four antibodies were able to coprecipitate REEP5, mostly as higher molecular weight dimer and oligomer, along with their respective antigens. Finally, as parallel experiments, co- immunoprecipitation assays performed using total neonatal ventricular cardiac lysate also revealed similar REEP5 interactions with Nogo-A/B, ATL3, and CKAP4 (Supplementary Fig. 5a).

127

128

Fig. 6 Mass spectrometry analysis identifies REEP5 interactions with known cardiac SR shaping proteins. a Immunoblot analysis of nickel-His purification of CnVA- REEP5 from transfected HEK293 cells shows stable expression of CnVA-REEP5 and successful immunoprecipitation of CnVA-REEP5 monomer (~34 kDa) and dimer (~68 kDa). b Identification of known SR/ER-shaping proteins as REEP5 interacting proteins by mass spectrometry analysis. Average precursor MS1 peak areas (peptide m/z signal) as defined by iBAQ (intensity based absolute quantification) are shown, n = 3 independent mass spectrometry runs. To calculate fold change, average null values (n.d.—not detected) were inputted with a value of 10. RTN reticulon, ATL atlastin, CKAP4 cytoskeleton-associated protein 4, ACTC1 alpha cardiac muscle actin 1, MYL6 myosin light chain 6, GAPDH glyceraldehyde-3-phosphate dehydrogenase. Asterisks indicate a statistically significant p value in a Tukey’s multiple comparison analysis where *p < 0.05. c RNASeq analysis of the RTN, ATL families of proteins, and CKAP4 in human fetal heart and adult heart tissue data using data from Human Protein Atlas. d Immunoblot analysis of nickel-His REEP5 immunoprecipitation lysates for RTN4/Nogo-A/B, ATL3, and CKAP4, n = 3. e HEK293 cells were transfected with myc-tagged Nogo-A and Nogo-B plasmids and detected with myc and α-tubulin antibodies. f Co-immunoprecipitation assay of cotransfected HEK293 cells with anti-REEP5 antibody (left panel) and anti-RTN4/Nogo- A/B antibody (right panel) demonstrated interaction between REEP5 and RTN4, n = 3. Eluates were collected on ice and loaded directly for blotting, or samples were boiled prior to blotting. g–j Co-immunoprecipitation and reverse order co-immunoprecipitation assays with g anti-REEP5, h anti-RTN4/Nogo-A/B, i anti-ATL3, and j anti-CKAP4 antibodies in adult mouse cardiac microsomes (input), followed by immunoblots analysis, n = 5. Source data containing original uncropped immunoblots are provided as a Source Data file.

In vivo phenotype analysis of REEP5 depletion in zebrafish. We next used the zebrafish embryonic model, with its rapid ex utero development, to investigate the role of REEP5 in vivo. We performed IF analysis at 48 and 96h post fertilization (hpf) to examine the localization of REEP5 expression within cardiac myocytes. Co-staining with zn-8 (localized to the membrane) and DAPI revealed REEP5 expression within both ventricular and atrial cardiac myocytes, with minimal nuclear or membrane localization evident (Fig. 7a).

To determine the function of REEP5 during zebrafish heart development, embryos were injected with a titrated minimum functional dosage of 1 ng of REEP5 morpholino (MO) at the one-cell stage to block REEP5 translation. Reduced REEP5 expression was observed in MO embryos (Supplementary Fig. 6a). Using Tg(myl7:EGFP) embryos, demarcating

129

differentiated cardiac myocytes, REEP5 morphant embryos demonstrated abnormalities in heart morphology with a lack of looping at 48 hpf (Supplementary Fig. 6b, c). To determine the specificity of the REEP5 MO sequence, REEP6 (92% homology to REEP5) MO sequence was used as a control. Investigations into REEP6 MO embryos revealed no observable cardiac phenotype, supporting the specificity of REEP5 MO associated phenotypes (Supplementary Fig. 6d). Optical imaging and IF of the REEP5 morphants showed defects in cardiac looping where a linear-oriented 2-chambered heart morphology in REEP5, compared to a fully looped heart in the sham controls (Supplementary Fig. 6e–h). These morphological defects were accompanied by irregularities in heart rhythm suggestive of an atrioventricular heart block. Further optical imaging analysis of cardiac conduction patterns demonstrated arrhythmogenic beating rhythms with asynchronous and skipped heart beats in REEP5 MO embryos compared to controls (Supplementary Fig. 6i and Supplementary Movies 1–3). The regular heartbeat of a control embryo by 96 hpf is a 1:1 ratio of atrial:ventricular contractions. Upon loss of REEP5 function, this ratio became offset with the atrium beating 3× to every ventricular contraction. TEM analysis of control embryos revealed cardiac myocytes with organized sarcomeres displaying well-ordered actin–myosin filaments and intact SR membranes (Supplementary Fig. 6j, k). However, TEM analysis of REEP5 morphants revealed apparent SR membrane vacuolization in zebrafish ventricular tissue along with structural discontinuity between sarcomeres (Supplementary Fig. 6l, m).

In vivo functional rescue experiments by injecting wild-type REEP5 mRNA in REEP5 MO zebrafish embryos showed co-injection efficiently rescued the cardiac dysfunction phenotypes observed in REEP5 MO embryos, resulting in significantly improved heart rate (140 ± 11 bpm in control vs. 56 ± 4 bpm in REEP5 MO vs. 134 ± 4 bpm in REEP5 MO + REEP5 mRNA injected embryos; p = 0.004. Data presented as mean ± SEM.) and heart morphology in the rescued morphants (Supplementary Fig. 6n–p). Further analysis showed reduced AV dys-synchrony phenotype with an apparent healthy ratio of 1:1 atrial:ventricular contractions in the rescued embryos (0.3 ± 0.3% in control vs. 83 ± 4% in REEP5 MO vs. 10 ± 2% in REEP5 MO+REEP5 mRNA injected embryos; p < 0.0001). Immunoblots of REEP5-interacting proteins identified in Fig. 6 were also performed and

130

showed a marked increase in the protein expression levels of ATL3 (0.48 ± 0.23 in control vs. 0.89 ± 0.19 in REEP5 MO embryos; densitometric ratio normalized to GAPDH, p = 0.04) (Supplementary Fig. 6q).

To address potential off-target effects of the REEP5 MO mutant and to validate the REEP5 MO model, we used CRISPR/Cas9 to generate a REEP5 loss-of-function mutant in zebrafish 40. High resolution melt (HRM) analysis of REEP5-targeted gRNA and Cas9 mRNA injected zebrafish embryos (F0) showed a marked temperature shift in the melt curves comparing controls and REEP5 gRNA injected embryos, confirming the cutting efficiency of REEP5-targeted gRNA (Supplementary Fig. 7a). Depletion of REEP5 protein expression was confirmed by subsequent immunoblotting and IF analyses of mutant hearts (Supplementary Fig. 7b, c). To determine the functional and cutting specificity of our designed gRNA, we performed dose response gRNA injection conditions (1×, 2×, 3×) (Fig. 7b–e). We determined 2× gRNA provided high phenotypic penetrance and low toxicity and these injected embryos were used for all subsequent functional analysis. Optical imaging of cardiac morphology and contraction in REEP5 gRNA injected fish showed cardiac developmental defects, displayed as a lack of cardiac looping, as well as atrio-ventricular conduction defects with dysynchronous beating rhythms (Fig. 7f–h).

Further quantification showed that more than 40% of the 2× and 3× gRNA injected F0 embryos showed cardiac developmental defects with a linearized heart (Fig. 7i). In all experimental conditions there was reduced heart rate and increased atrio-ventricular dyssynchrony associated with REEP5 gRNA injection. Increasing gRNA concentration led to a greater number of embryos displaying more severe developmental defects with reduced heart rate and impaired trunk circulation (Fig. 7j, k and Supplementary Movies 4–8). These defects were associated with poor survival since 60% of all analyzed REEP5 morphant and crispant mutant embryos did not survive past 7 days, highlighting the importance of REEP5 during embryonic heart development.

Next, we generated germline reep5 CRISPR homozygous maternal zygotic (MZ) zebrafish mutants. Our sequencing analysis showed successful editing of the reep5 gene

131

Fig. 7 In vivo CRISPR/Cas9-mediated REEP5 depletion in zebrafish embryos leads to cardiac abnormalities. a Immunofluorescence analysis of REEP5 expression in wild- type zebrafish hearts at 48 and 96 hpf. Co-staining for REEP5 (green), zn-8 (general plasma membrane; red) and DAPI (nuclear; blue) revealed REEP5 expression in both ventricular and atrial cardiac myocytes, with no nuclear or membrane localization evident. Scale, 50 μm. b–e CRISPR/Cas9-mediated depletion of REEP5 with varying gRNA concentrations reveals the contribution of REEP5 to embryonic heart development in

132

zebrafish embryos. Images shown are representative of 331 control, 410 2× REEP5 gRNA, and 296 3× REEP5 gRNA total injected embryos, n = 3 independent biological replicates. Scale, 500 μm. f, g Optical imaging analysis of control and 2× REEP5 gRNA injected embryos shows cardiac looping defects associated with REEP5-targeted gRNA injection. Scale, 10 μm. h Movie analysis of REEP5 gRNA injected embryos shows dyssynchronous atrio-ventricular beating rhythms compared to control zebrafish embryos. Area profiles were smoothened with a Gaussian curve with σ = 0.2 s. i–k Bar graphs showing i phenotypic penetrance represented as percentage, j heart rate represented as beats per minute (bmp), and k atrioventricular beating dyssynchrony represented as percentage from 331 control, 410 2× REEP5 gRNA, and 296 3× REEP5 gRNA total injected embryos, n = 3 independent biological replicates. Asterisks indicate a statistically significant p value in a Tukey’s multiple comparison analysis where *p < 0.05, **p < 0.01, and ***p < 0.001; data are presented as mean ± SEM. at the gRNA target site, resulting in the generation of reep5 homozygous CRISPR mutants (Fig. 8a). However, analysis of F4 reep5 germline CRISPR MZ mutant embryos showed minimal morphological abnormalities nor developmental defects and did not present with the complete lethality we had observed previously (Fig. 8b–e and Supplementary Movie 9).

Short-term treatment with verapamil, a voltage-gated Ca2+ channel blocker, has been proposed as a rapid heart failure model 41. Zebrafish at 48 hpf were treated with 0.6 mg/mL verapamil for 30 min to induce a rapid decrease in heart rate and contractility, ultimately leading to heart failure 41. The reep5 germline mutants treated with verapamil displayed sensitized cardiac dysfunctional phenotypes highlighted by a marked reduction in heart rate compared to control embryos (133 ± 5 bpm in control vs. 125 ± 4 bpm in reep5 CRISPR fish under resting conditions, compared to 131 ± 4 bpm in control vs. 99 ± 4 bpm in the reep5 CRISPR fish with verapamil treatment; p = 0.003). Data presented as mean ± SEM (Fig. 8f). Compensatory network induced by deleterious mutations in CRIPSR germline mutants, but not in zebrafish knockdown models such as morphants or crispants, has been described previously 42.

133

134

Fig. 8 Genetic compensation in reep5 CRISPR knockout zebrafish mutant. a Sequencing analysis generated by Sanger Synthego shows reep5 gRNA targeted and edited homozygous CRISPR mutant sequences in the region around the guide sequence. The horizontal black underlined region represents the guide sequence and the horizontal red underline indicates the protospacer adjacent motif (PAM) site with the vertical black line representing the cut site. b, c CRISPR/Cas9-mediated reep5 homozygous mutant compared to control zebrafish embryos, n = 3 independent biological crosses. Scale, 500 μm. d, e Optical imaging analysis of control and reep5 CRISPR mutant embryos shows normal cardiac chamber orientation and development in the reep5 CRISPR mutant hearts. Scale, 10 μm. f Heart rate analysis of control and reep5 homozygous mutant embryos upon 600 μM verapamil treatment for 30 min; a total of 46 control, 24 reep5 CRISPR mutants, 32 control + verapamil, and 36 reep5 CRISPR mutants + verapamil embryos were analyzed, n = 3 independent biological crosses; data are presented as mean ± SEM. g, h reep5 CRISPR mutants injected with REEP5 MO (1 ng) appears phenotypically normal compared to overt cardiac abnormalities in the REEP5 morphant embryos. Images shown are representative of 287 REEP5 MO and 232 reep5 CRISPR mutants injected with REEP5 MO embryos, n = 3 independent biological crosses. Scale, 500 μm. i, j Optical imaging analysis of REEP5 MO and reep5 CRISPR mutants injected with REEP5 MO embryo hearts shows normal cardiac chamber orientation and development in the reep5 CRISPR mutant hearts. Scale, 10 μm. k Bar graph showing phenotypic penetrance represented as percentage from 287 REEP5 MO and 232 reep5 CRISPR mutants injected with REEP5 MO embryos, n = 3 independent biological crosses; data are presented as mean ± SEM. l Immunoblot analysis of cardiac-enriched ER structure proteins (RTN4, ATL3, and CKAP4) in reep5 CRISPR homozygous mutants shows upregulation of RTN4 in reep5 CRISPR mutant hearts. Source data containing original uncropped immunoblots are provided as a Source Data file. Asterisks indicate a statistically significant p value in a Tukey’s multiple comparison analysis where **p < 0.01 and ***p < 0.001.

Thus, we next examined the possibility of such genetic compensation by injecting REEP5 MO into reep5 germline mutants and showed that reep5 mutants injected with REEP5 MO were phenotypically normal compared to overt cardiac abnormalities in the REEP5 morphants (Fig. 8g–j and Supplementary Movie 10). Further quantification showed reep5 CRISPR mutants were less sensitive to MO injection, with ~2 ± 1% of reep5 CRISPR mutants vs 65 ± 4% of REEP5 MO embryos showed cardiac developmental defects with a linearized heart (Fig. 8k). Lastly, immunoblots of the REEP5 interacting proteins showed a marked increase in the protein expression levels of RTN4 (0.89 ± 0.16 in control vs. 1.99 ± 0.18 in reep5 CRISPR fish; densitometric ratio normalized to α-actin, p = 0.01),

135

suggesting potential compensatory mechanism regulated by RTN4 upregulation in response to the loss of REEP5 in zebrafish (Fig. 8l).

In vivo AAV9-mediated REEP5 depletion in mice. In parallel to REEP5 depletion in zebrafish embryos, we performed intraperitoneal injections of AAV9 virus carrying scrambled shRNA or REEP5 shRNA in neonatal mice (P10). Immunoblotting analysis of REEP5 4 weeks post viral infection showed near depleted levels of REEP5 in the AAV9 REEP5 shRNA injected mouse hearts (Fig. 9a). All AAV9 REEP5 shRNA injected mice developed lethal cardiac dysfunction at 4 weeks post viral transduction and we never observed animals surviving past 5 weeks post injection, compared to healthy AAV9 scram injected littermates.

Hematoxylin and eosin (H&E) and Masson’s trichrome staining of the REEP5 depleted myocardium showed general disorganization of the myocardium compared to scram controls (Fig. 9b). Masson’s trichrome staining revealed greater fibrotic regions, with significant collagen deposition in the REEP5 depleted ventricular myocardium (Fig. 9b, bottom panels). TEM analysis (Fig. 9c) of control ventricular tissue showed organized sarcomeres displaying well-ordered actin–myosin filaments, whereas REEP5 depleted ventricular myocardium showed disordered sarcomere organization and SR membrane vacuolization, consistent with the observations in REEP5 depleted adult cardiac myocytes and zebrafish embryos (Fig. 9c). Echocardiographic assessment of cardiac function of AAV9 scram shRNA and REEP5 shRNA injected mice was performed (Fig. 9d) and echocardiographic B-mode and M-mode measurements revealed significant myocardial dilatation and cardiac contractile dysfunction (Fig. 9e), with a marked decrease in cardiac ejection fraction 61.0 ± 0.9% in scram vs. 23.1 ± 2.2% in REEP5 KD; p < 0.0001, stroke volume (46 ± 2 μL in scram compared to 18 ± 2 μL in REEP5 KD; p < 0.0001), and cardiac output (19.0 ± 0.5 mL/min in scram vs. 10.1 ± 1.2 mL/min in REEP5 KD; p = 0.0002) in AAV9 REEP5 shRNA injected mice compared to its AAV9 scram injected littermates (Fig. 9e).

136

Fig. 9 In vivo AAV9-mediated REEP5 depletion in mice results in increased cardiac fibrosis, activated cardiac ER stress, cardiac dysfunction, and death. a Immunoblot analysis of AAV9-mediated REEP5 depletion in mice 4 weeks post viral infection. b Histological analyses of AAV9 Scram shRNA and REEP5 shRNA injected mouse hearts at 4 weeks post viral infection. Top panels: H&E, bottom panels: Masson’s trichrome stain. Scale, 20 μm. All images shown are representative of approximately 20 total images captured per condition, n = 3 independent biological replicates. c Transmission electron

137

microscopy analysis of AAV9 REEP5 shRNA-injected myocardium reveals degeneration of muscle fibers and disrupted SR membranes and organization compared to scram controls. M mitochondria, SR sarcoplasmic reticulum, V vacuoles. Scale, 0.5 nm. All images shown are representative of approximately 20 total images captured per condition, n = 3 independent biological replicates. d In vivo echocardiographic heart function assessment of AAV9 scram and AAV9 REEP5 shRNA-injected mice at 4 weeks post viral infection, n = 5. e Echocardiographic M-mode and B-mode measurements showed significantly compromised cardiac function with reduced cardiac ejection fraction and cardiac output, n = 5 independent biological replicates. Asterisks indicate a statistically significant p value in a Tukey’s multiple comparison analysis where ***p < 0.001; data are presented as mean ± SEM. f Immunoblotting analysis of cardiac ER stress markers (GRp78, GRp94, XBP1, ATF4, and CHOP) and ER-dependent apoptosis (caspase 12) upon AAV9-induced REEP5 depletion in vivo in mice, n = 3; p values are shown. g Immunoblots of RTN4, ATL3, and CKAP4 upon AAV9-induced REEP5 depletion in vivo in mice, n = 3 independent biological replicates. Source data containing original uncropped immunoblots are provided as a Source Data file.

REEP5-depleted myocardium had elevated expression levels of ER stress markers GRp78 (1.5 ± 0.2 in scram vs. 2.5 ± 0.5 in REEP5 KD; densitometric ratio normalized to α-actin, p = 0.13. Data presented as mean ± SEM), GRp94 (0.05 ± 0.01 in scram vs. 1.06 ± 0.84 in REEP5 KD; p = 0.27), XBP1 (0.07 ± 0.01 in scram vs. 0.13 ± 0.02 in REEP5 KD; p < 0.05), ATF4 (0.04 ± 0.01 in scram vs 0.79 ± 0.11 in REEP5 KD; p = 0.002), and CHOP (0.02 ± 0.01 in scram vs. 0.17 ± 0.05 in REEP5 KD; p = 0.001) (Fig. 9f). In line with our previous in vitro studies in cardiac myocytes (Fig. 3f), a marked increase in the expression levels of cleaved caspase 12 (0.03 ± 0.01 in scram vs. 0.25 ± 0.06 in REEP5 KD; p = 0.004) was observed, indicating activated ER stress-induced apoptotic activity in the REEP5 depleted ventricular myocardium (Fig. 9f).

Lastly, we measured expression levels of the newly identified REEP5 interacting proteins. Immunoblots revealed significantly increased levels of RTN4 (1.1 ± 0.2 in scram vs. 2.3 ± 0.5 in REEP5 KD; densitometric ratio normalized to α-actin, p = 0.01), ATL3 (0.14 ± 0.04 in scram vs. 0.45 ± 0.07 in REEP5 KD; p = 0.02), and CKAP4 (0.59 ± 0.04 in scram vs. 1.52 ± 0.31 in REEP5 KD; p = 0.04) in REEP5 depleted hearts (Fig. 9g). Interestingly, we also investigated protein levels of these SR/ER shaping proteins in the context of cardiac pressure overload-induced hypertrophy and heart failure and were able to

138

demonstrate consistent and significant elevations in protein levels of REEP5, RTN4, ATL3, and CKAP4 in failing mouse hearts following aortic constriction (see Supplementary Fig. 8a).

4.4 DISCUSSION This study details a fundamental role of REEP5 in the cardiac myocyte, demonstrates that this protein is essential for SR/ER organization and proper function, identify its protein interactions, and shows that its deletion can result in cardiac functional and developmental defects. REEP5 expression was demonstrated to be particularly muscle- specific with the highest protein expression in the mouse ventricles and skeletal muscle. Our results show that REEP5 expression is detected across the SR network in cardiac myocytes, co-localizing with known SR markers (SERCA2, α-actinin, RyR2, and triadin), thereby regulating SR function in cardiac muscle.

Previous experiments in yeast have established the interactions between REEP1, RTN1, and ATL1 coordinate the generation and maintenance of the tubular ER network 30. In both mouse neonatal cardiac myocytes and adult cardiac microsomes, we demonstrated physical interactions between REEP5 and other SR/ER structure proteins including RTN4/Nogo-A/B, ATL3, and CKAP4. A fine balance between SR/ER structure proteins has been proposed to be critical for generating and maintaining a dynamic SR/ER network as overexpression of membrane curvature stabilizing proteins generates long, unbranched tubules with a marked decrease in tubular three-way junctions and absence of membrane fusion were observed with ATL inactivation 43–45. Wang et al. 46 further demonstrated the roles and cooperation between ATLs and RTNs in ER three-way junction morphology and maintenance in Xenopus laevis eggs. Our results support the interactions of these SR/ER structure proteins in primary cardiac myocytes. Moreover, we show that disruption of these interactions through REEP5 depletion resulted in a marked increase in the expression levels of RTN4, ATL3, and CKAP4 in mouse ventricular myocardium. Interestingly, a marked increase in the expression of RTN4 was observed in our germline reep5 CRISPR mutant whereas a clear reduction in the

139

expression levels of ATL3 and CKAP4 were detected. Altogether, these data suggest that REEP5 may act upstream of other ER structure proteins or may possess a more critical role in the heart. While increased RTN4 protein expression may be sufficient to compensate for the loss of REEP5 in CRISPR-based reep5 null zebrafish embryos, this does not appear to be the case in our in vivo studies in the mouse with the rapid knockdown of REEP5 via viral deletion. Furthermore, REEP5 interaction with a previously unassociated sheet-ER stabilizing protein, CKAP4, may explain the importance of REEP5 in stabilizing membrane curvature at the edges of ER sheets 47. This interaction may be important for cardiac SR structure as primary cardiac myocytes are more complex than yeast or X. laevis.

Mutations and altered protein modifications in many ER morphology proteins including ATL1, REEP1, and REEP2 account for the most common forms of a human neurodegenerative disorder known as the hereditary spastic paraplegias 30, 48, 49. In addition, recent studies demonstrate that genetic mutations in REEP6 are responsible for causing retinitis pigmentosa 19, 20, an inherited retinal dystrophy characterized by loss of photoreceptors in the retina. REEP5 protein levels in the context of cardiac disease showed increased REEP5 expression in mouse hypertrophic cardiomyopathy and human idiopathic heart disease, indicating a potential role of REEP5 in the progression and development of heart disease. These results suggest that proper REEP5 protein expression contributes to SR/ER integrity and maintenance in cardiac myocytes and may be responsible for altered SR/ER functions and morphological defects in cardiac disease. In fact, in vitro REEP5 depletion in isolated cardiac myocytes resulted in activation of ER stress and the initiation of ER-dependent cell death as seen by caspase-12 activation.

We observed phenotypic discrepancies between our morphant, crispant, and germline CRISPR reep5 mutant models in zebrafish that we believe can be attributed to the fundamental difference between knockdown and knockout experimental approaches. Morpholino and crispant-driven knockdown experiments in zebrafish embryos allow investigations of acute embryonic phenotypes in embryos whereas the generation of genetically mutated homozygous mutants provides a window for physiological adaptation

140

or genetic compensation. Previously, activation of genetic compensation only in response to deleterious mutations, but not in knockdown-driven morphants or crispants, has been observed in zebrafish 42. Similar compensatory phenomena, as well as differences between knockdown and knockout models, have also been reported in yeast 50. In fact, a significant upregulation of RTN4 in the reep5 germline mutant but not in REEP5 morphants suggests that the consequence of germline reep5 deletion is perhaps masked by the upregulation of RTN4 in the germline reep5 mutant hearts.

Nevertheless, our REEP5 crispant data in the zebrafish embryos are consistent with in vitro REEP5 depletion experiments in the myocyte and in vivo depletion in the mouse. In both cases, altered cell morphology and abnormal heart morphology were observed, highlighting the fundamental role of REEP5 in cardiac biology. Of course, abnormal fish phenotype must always be approached in a very critical manner 51. However, given that we observed consistent phenotypes within the myocytes in vitro, and zebrafish and mouse hearts in vivo, it is highly probable that the observed phenotypes are a direct result of REEP5 deficiency. Our results are supported by a recent study that generated CRISPR/Cas9-mediated inactivation of REEP5 in rat ventricular myocardium with a preliminary characterization that showed depressed cardiac contractility in pressure– volume loops 52. Interestingly, inactivation of REEP5 in rats displayed a weaker and nonlethal cardiac phenotype compared to our REEP5-depleted mouse model. This discrepancy in phenotype severity may be a species-dependent difference, or a result of differences in knockdown vs. knockout approaches similar what we observed in our zebrafish studies.

Taken together, our results suggest an indispensable role of REEP5 in the SR/ER to maintain a highly differentiated SR network in cardiac muscle cells. In conclusion, we believe a more complete understanding of the role of REEP5 in the heart will be critical to understand the normal function of healthy versus diseased SR/ER.

141

4.5 METHODS Human myocardium. All procedures involving human samples were approved by the University Health Network Ethics Review Committee (REB reference #158781- AE, 136800-AE, and 10-0703) and were performed in accordance with the guidelines. Written informed consent was obtained from all patients for use of their samples in research prior to study commencement. Human idiopathic tissue was isolated from a 58-year-old female and the ischemic cardiomyopathy tissue was isolated from a 68-year-old male during implantation of a left ventricular assist device at the Peter Munk Cardiac Centre (Toronto, Ontario). Healthy human left ventricle lysates were obtained from BioChain (Cat#: P1234138).

Mouse heart disease tissues. In all, 4 and 8 week TAC cardiac tissue were collected as described 53. Two-week post LAD induced MI hearts were induced and isolated as described previously 54. Finally, 16-week-old PLN R9C mutant mice, a model of human DCM were analyzed as described 55.

Adult and neonatal mouse cardiac myocyte isolation. All experimental procedures were conducted in accordance with the Animal Care Guidelines approved by University of Toronto Animal Use and Care Committee. Protocol for isolation of adult mouse cardiac myocytes was modified from Ackers-Johnson et al. 56. Briefly, 6–8-week-old male CD1 (Jackson Labs) mouse hearts were perfusion-isolated using 10 mL of EDTA buffer, followed by digestion with warmed collagenase II buffer (525 units/mL; LS004176, Worthington Biochemical Corp., Lakewood, NJ, USA). Following dissociation and washing, myocytes were resuspended in culture media and plated on laminin coated glass-bottom dishes. Primary mouse neonatal cardiac myocytes were isolated from CD1 mice as previously described 57. Pups (p1–3) were decapitated and hearts were washed in ice-cold Hanks-balanced salt solution without calcium and magnesium. Atria were discarded and ventricles were dissected into 2-mm fragments, and enzymatically digested in 0.05% trypsin (Hank’s balanced salt solution) on an end-to-end rotor overnight at 4 °C. Next day, cells were released with 450 unit/mL collagenase II (LS004176, Worthington Biochemical Corp., Lakewood, NJ) at 37 °C (15 min/digestion) until no tissue

142

remaining. Collected cells were filtered through a 70-μm cell strainer (08-771-2, Fisher Scientific), span down at 200g for 5 min at room temperature and pre-plated at 37 ° C for 2 h in DMEM/F12 medium supplemented with 10% fetal bovine serum (FBS)/horse serum. Cells were then seeded on gelatin-coated glass-bottom dishes for 48 h before being maintained in culture medium (insulin-transferrin-sodium selenite media supplement, 10 μM BrdU, 20 μM cytosine β-d-arabinofuranoside, 2.5 μg/mL sodium ascorbate, 1 nM LiCl, 1 nM thyroxine).

Cell culture. Human embryonic kidney (HEK-293T) cells and murine C2C12 myoblasts were maintained in DMEM supplemented with 10% FBS, penicillin and streptomycin. For transgene expression, HEK-293T cells were transfected using polyethylenimine

(Polysciences, Inc., Warrington, PA) and C2C12 myoblasts were transfected with Lipofectamine 2000 (Invitrogen) as described 57.

Membrane yeast two-hybrid experiments. The MYTH system 58 was used to validate REEP5 membrane topology as previously described 27. Briefly, N-terminal TF transcription factor tagged TF-Cub:REEP5 “bait” REEP5 protein and ubiquitin fragments tagged (NubI and NubG) “prey” proteins were tested for protein–protein interactions. Upon reconstitution of TF-Cub and NubI tagged membrane proteins, TF is released to enable reporter activation. A mutant form of the ubiquitin fragment (NubG) that contains a I13G mutation only allows reconstitution with Cub when brought in close proximity by physically interacting proteins. If the bait REEP5 protein is targeted to the membrane, the N-terminal tagged bait will expose its Cub fragment in the cytosol to interact with NubI, but not with NubG. Both Cub and Nub tagged proteins must be exposed in the cytosol for reconstitution and subsequent reporter activation. Activation of the reporter system was measured as growth on media.

Lentiviral transduction of isolated mouse cardiac myocytes. Lentivirus-mediated transduction for shRNA delivery in mouse neonatal and adult cardiac myocytes was conducted as previously described 57. Briefly, lentivectors were created by triple transfections of pLKO.1-mREEP5 (trcn0000106187, RNAi consortium), pSPAX2, and

143

pMD2.G in HEK-293T cells. A lentivector carrying scrambled shRNA was used as control. For transduction of cardiac myocytes, mouse neonatal and adult cardiac myocytes were incubated with 8 μg/mL polybrene for 90 min and incubated with lentiviral solution for 21 h at 37 °C, 5% CO2. Transduced cells were cultured in fresh media for at least an additional 24 h prior to any experiments.

Adeno-associated viral transduction in vivo in mice. All adeno-associated viral (AAV9) parent vectors were obtained from Dr. Roger Hajjar (Mount Sinai School of Medicine, New York) and high titer AAV9 vectors were generated via a triple transfection approach in HEK293 cells as previously described 59, 60. Briefly, HEK293 cells were transfected with 15 μg of pDG9 plasmid and 5 μg desired AAV plasmid per plate. HEK293 cells were harvested 48 h post transfection for subsequent purification by filtration and iodixanol. For in vivo transduction, neonatal (P10) pups from wild-type CD1 mice (Jackson Labs) were injected intraperitonially 5 × 1011 genomic titer of AAV9 virus carrying either mouse REEP5 shRNA or scrambled shRNA sequence (non-expressing control) and were monitored daily.

Echocardiographic assessment of cardiac function in mice. All mice were anesthetized with 1.5% isoflurane and placed on an animal warming pad maintained at 37 °C throughout the procedure. Heart rate, respiration, and body temperature were monitored and maintained consistently throughout the ultrasound imaging session. All mice were imaged in the supine position using Vevo 3100 Imaging system (Visual Sonics Inc., Toronto) with a 30 MHz probe. Standard 2D echocardiographic assessment (B-Mode and M-Mode) was performed through the left para-sternal acoustical window to assess left ventricular dimensions and heart function 61, 62. M-mode recording from the middle segment of the left ventricle was made to measure the left ventricular anterior and posterior wall thickness at peak systole and end diastole. Pulsed Wave Doppler Mode was used to acquire doppler flow profiles and to record velocity time integrals. Subsequent measurements and analyses were performed using the Vevo 3100 2D quantification software to calculate interventricular septal thicknesses, posterior wall thickness, systolic dimension, ejection fraction, and cardiac output.

144

Immunoblotting and IF. Protein lysates from cells and tissues were harvested in radioimmunoprecipitation assay buffer (RIPA, 50 mM Tris-HCl; pH 7.4, 1% NP- 40, 0.5% sodium deoxychloate, 0.1% SDS, 150 mM NaCl, 2 mM EDTA), supplemented with protease and phosphatase inhibitors (Roche), for 30 min on ice, spun down at 15,000g/4 °C. Soluble fractions were saved for immunoblotting. For live imaging, cells were seeded on glass-bottom dishes (MatTek Corp. Ashland, MA), and transfected with plasmids for fluorescence protein expression. Cells were directly visualized 24 and 48 h following transfection. For IF, cells were fixed with 4% paraformaldehyde for 10 min on ice, permeabilized with ice cold 90% methanol at −20 °C for 10 min, and blocked with blocking solution (3% FBS and 0.1% TritonX-100 in PBS) at room temperature for 1 h. Primary antibody staining was carried out overnight at 4 °C and fluorophore-conjugated secondary antibody staining was performed at room temperature for 1 h in the dark. Samples were visualized with Zeiss spinning disk confocal microscopy (Zeiss Spinning Disk Confocal Microscope). Three-dimensional reconstruction of z-stack images was carried out using Imaris 8.1 software (Bitplane, Switzerland).

Transmission electron microscopy. Isolated adult mouse cardiac myocytes were fixed in 2.5% glutaraldehyde in 0.1 M phosphate buffer, while zebrafish were fixed in 2% paraformaldehyde in 0.1 M sodium cacodylate at 4 °C overnight. Samples were postfixed in 1% osmium tetroxide buffer and processed through graded alcohols and embedded in Quetol-Spurr resin. Sections of 90–100 nm were cut and stained with uranyl acetate and lead citrate and imaged at ×20,000 magnification using a Hitachi TEM microscope at the Department of Pathology, St. Michael’s Hospital (Toronto, Canada) or a FEI Tecnai 20 TEM microscope at the Hospital for Sick Children (Toronto, Canada) 35.

Antibodies. Primary rabbit polyclonal anti-REEP5 antibody (IB: 1:1000 dilution, IF: 1:800 dilution; 14643-1-AP; Proteintech), polyclonal anti-Nogo-A/B (IB: 1:1000 dilution, IF: 1:1000 dilution; PA1-41220; ThermoFisher), polyclonal anti-ATL3 antibody (IB: 1:1000 dilution, IF: 1:800 dilution; PA5-24652; ThermoFisher), polyclonal anti-CKAP4 antibody (IB: 1:1000 dilution, IF: 1:800 dilution; PA5- 42926; ThermoFisher) and mouse

145

monoclonal anti-Alpha sarcomeric actinin antibody (IF: 1:500 dilution; MA1-22863; ThermoFisher), monoclonal anti-Triadin antibody (IF: 1:500 dilution; MA3-927; ThermoFisher), monoclonal anti-RyR2 antibody (IF: 1:500 dilution; ab2827; Abcam) were used for immunocytochemistry and immunoblot studies. Anti-alpha tubulin (#2144), anti- GRp78 (#3177), anti-GRp94 (#2104), anti-GAPDH (#2118) antibodies from Cell Signaling; anti-ATF4 (ab216839), anti-Caspase12 (ab62484) antibodies from Abcam were used for immunoblot studies at 1:1000 dilution. Anti-GFP (sc390394) antibody from Santa Cruz Biotechnology was used for immunoblot studies at 1:1000 dilution.

ROS and MMP measurements. Intracellular ROS was measured using CellROX Deep Red reagent (Invitrogen) according to the manufacturer’s instructions. Neonatal cardiac myocytes were cultured and seeded in 96-well plates or glass- bottom dishes. Cells were incubated with the ROS scavenger, Tiron (1 mM, Sigma) for 30 min at 37 °C, 5% CO2 followed by a 30 min incubation with CellROX reagent at a final concentration of 5 μM. MMP was detected in cells using 50 nmol/L of vital mitochondrial dye JC-1 (Molecular Probes). Cells were washed with culture medium before and after a 20 min incubation period with JC-1 at 37 °C, 5% CO2; fluorescence was then measured via plate-reader analysis 35.

MTT viability assay. Cell viability was assessed by performing 3-(4,5-dimethyl- thiazol- 2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (Sigma-Aldrich) according to the manufacturer’s instructions 61. Neonatal cardiac myocytes were seeded in 96-well plates.

Cells were incubated in absence of light with 0.5 mg/ml MTT at 37 °C, 5% CO2 for 1 h. Untransformed MTT was removed and formazan crystals were solubilized with dimethyl sulfoxide (DMSO) followed by fluorescence measurement at 570 nm using Perkin Elmer plate reader.

Protein purification and mass spectrometry analysis. Eluted REEP5 pulldown lysates were concentrated using a 3000 MWCO centrifugal concentrator (Millipore), reduced with 5 mM DTT for 30 min at 55 °C, alkylated with 25 mM iodoacetemide for 30 min at room temperature in the dark. Proteins were trypsinized using 5 μg Trypsin Gold (Promega)

146

overnight at 37 °C and halted via acidification with formic acid to a concentration of 1%. Peptides were then passed through using OMIX C18 Tips (Agilent) and quantified via BCA assay (ThermoFisher). For REEP5-transfected lysates and untransfected controls, three biological replicates were analyzed on a Q Exactive Plus mass spectrometer (ThermoFisher). Briefly, 2 μg of peptides were separated on a 2-h 5–35% acetonitrile gradient on a 10 cm long, 75 μm i.d analytical column packed with 3 μm 100 Å C18 resin (Phenomenex). Data were acquired at R = 70,000 at 400 m/z with one full MS1 scan from 400 to 1500 m/z followed by 12 data-dependent MS2 scans at R = 17,500 with dynamic exclusion set to 60 s. Two biological replicates for each condition was run in technical duplicates. RAW files were searched using MaxQuant version 1.5.4.163 using the UniProt human proteome sequence database. Variable modifications were set for methionine oxidation and protein amino terminal acetylation. False-discovery rate was set to 1% using a target-decoy strategy. Proteins identified in only 2 of 8 MS experiments were excluded for comparison. Expression levels were used using iBAQ values determined by MaxQuant.

In vivo studies in zebrafish. Adult zebrafish were maintained according to Canadian Council on Animal Care (CCAC) and The Hospital for Sick Children Animal Service (LAS) guidelines. Transgenic zebrafish line used in this study was myl7:EGFPtwu 34, 64. Zebrafish embryos were grown at 28.5 °C in embryo medium as previously described 65. Standard techniques were used for microinjection 66. Microinjection was repeated a minimum of five times with consistent results. Morpholino oligos were purchased from Genetools (Oregon, USA). The ATG REEP5 morpholino (5′-GCC GCC ATG ATT TGT CTG AAG GGA T-3′) and ATG REEP6 morpholino (5′-AGT CAC CAT GTT TGC CAT ATT TAC A- 3′) were injected at a concentration of 1 ng/nl at the 1-cell stage of development. Bright- field images were taken using a Zeiss AXIO Zoom V16. Whole-mount immunofluorescence (IF) was carried out as previously described 67. Secondary reagents AF488 goat anti-mouse IgG1 recognizes monoclonal antibody S46 (atrial myosin heavy chain), AF555 goat anti-mouse IgG2b recognizes monoclonal antibody MF20 (Myosin heavy chain), AF488 goat anti-rabbit IgG recognizes antibody REEP5 and goat anti- mouse IgG 568 recognizes monoclonal neuronal cell surface marker (zn-8). MF20, S46

147

and ZN-8 primary antibodies were obtained from the Developmental Studies Hybridoma Bank (University of Iowa, IA, USA). Immunoblotting for REEP5 detection was carried as described above using 40 embryos per experimental condition. IF confocal images were taken with a Nikon A1R laser scanning confocal microscope or using Zeiss spinning disk confocal microscopy (Zeiss LSM 510/AxioImager.M1 Confocal Microscope). For in vivo RNA rescue experiments, Full-length wild-type zebrafish reep5 coding sequence was amplified from 48 hpf wild-type zebrafish cDNA. The coding sequence was modified to avoid interaction with the antisense morpholino while producing the same amino acids (5′ATGGCAGCAGCGTTA...TAA-3′). This modified coding sequence was subcloned into the pCS2+ vector. For mRNA production, the pCS2+ :reep5 construct was linearized via NotI digestion and in vitro transcription was carried out using the mMessage mMachine SP6 kit (Ambion). To validate the morphant phenotype, RNA rescue was performed via subsequent injection of antisense reep5 morpholino and reep5 mRNA. 1 ng/μL (1 nL volume) of reep5 morpholino was injected at the one-cell stage, immediately followed 100 pg (1 nL volume) of reep5 mRNA. Embryos were incubated at 28.5 °C and monitored daily.

CRISPR/Cas9 reep5 mutants were generated as previously described 68. To generate reep5 CRISPR mutants, exon sequences were analyzed for optimal cut sites using the Benchling web-based service (www.benchling.com). gRNA target sequences were ranked by highest off-target score followed by highest on-target score which resulted in a gRNA cut site in exon 3 targeting the beginning of the second transmembrane domain (5′-AAAATGGCTGACCTACTGGG). The REEP5-specific oligo using the T7 promoter with the standard overlap region was ordered from ThermoFisher (5′- TTAATACGACTCACTATAGG-AAAATGG CTGACCTACTGGG-GTTTTAGAGCTAGA). Once the template oligo was assembled, transcription was performed using the T7 Quick High-Yield RNA Synthesis Kit (NEB). The resulting RNA was purified using the standard ethanol/ammonium acetate protocol 69. An injection mix was prepared with gRNA and Cas9 protein in a 1:1 ratio at a concentration of 2 mM (1×). 1–3 nL of the solution was injected into zebrafish embryos at the 1-cell stage of development to achieve 1×, 2×, or 3× amounts of gRNA. Standard microinjection techniques were used 69. CRISPR/Cas9

148

gRNA cutting was evaluated in injected and uninjected 4 dpf embryos using a LightCycler 96 from Roche. Analysis was conducted using the Roche template for the HRM assay. Injections and HRM analysis were repeated 3 times with consistent results. Germline CRISPR reep5 homozygous MZ mutants were generated by performing an additional cross with an identified homozygous female reep5 mutant to rule out maternally- contributed mRNA expression of REEP5. The effects of verapamil were tested in 2 dpf wild-type and reep5 homozygous mutant fish. Concentrations of 10, 5, 2.5, 1.25, and 0.6 mg/mL of verapamil in 1 mL of E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, 0.33 mM MgSO4) were tested, with 0.6 mg/mL having no measurable effect on wild-type embryo heartbeat. Embryos were examined for phenotypes following 30 min of drug exposure using a Zeiss standard dissection microscope. Heart rate was recorded before and after drug treatment with atrial and ventricular heartbeats counted separately over a period of 15 s each.

Plasmids and reagents. Human REEP5 (NM_005669.4) was subcloned into pLD- puro- CnVA 38 and pcDNA-DEST40 (V5/6×-His epitope) plasmids by Gateway cloning technology according to manufacturer’s manual (Invitrogen). Truncation deletions were generated using available restriction digest sites. For the truncation deletion removing the cytosolic C-terminus region, only sequences around amino acid 114 were viable restriction sites, resulting in the 1–114 and Δ114–189 constructs; attempts at generating different C-terminal mutants were not successful. pmCherry-Sec61β was created as previously descripted 70. Briefly, mouse Sec61β (NM_024171) was PCR amplified with primers 5′-CAT CAT AGA TCT ATG CCG GGT CCA ACG CCC-3′ (mSec61 SP, BglII) and 5′-GTA GTA GAA TTC CTA TGA TGA TCG CGT GTA CTT GCC CCA-3′ (mSec61 AP, EcoRI) and subcloned in a pmCherry-C1 plasmid with the same restriction sites. The PCR was performed with following parameters: 1 cycle of 98 °C for 30 s, 35 cycles of 98 °C for 10s and 72°C for 10s, and 1 cycle of 72°C for 2min. All constructs were validated by sequencing at the ACGT Corp (Toronto, ON, Canada).

Movie analysis. Zebrafish bright-field movies were captured using a Zeiss AXIO Zoom_V16 at ×63 magnification. Embryos injected with or without REEP5 MO were

149

subjected to PTU for removal of pigmentation prior to movie capture. To measure atrial and ventricular area, we performed imaging analysis using SIESTA, a custom-built image analysis package 71, 72. Specifically, we applied the LiveWire algorithm implemented in SIESTA for semi-automated delineation of heart chambers. Area profiles were smoothened with a Gaussian curve where σ = 0.1 s and normalized by dividing by the mean area in each profile.

Adult mouse cardiac myocytes time-lapse images were obtained using a Revolution XD spinning disk confocal microscope equipped with an iXon Ultra 897 camera (Andor, Belfast, UK) and a ×1.5 coupling lens. A ×40 oil immersion lens (Olympus, NA 1.35) was used to acquire one 16-bit image every 40 ms. Image analysis was performed using algorithms developed with Matlab (MathWorks, Natick, MA) and DIPImage (Delft University of Technology, Delft, Netherlands). To measure cardiac myocyte area over time, we binarized images of isolated cardiac myocytes using the image mean as the threshold. A Savitzky-Golay filter with second order polynomials and a window size of 75 pixels was used to smooth the cardiac myocyte outlines prior to quantification. To measure fractional shortening, we calculated the equation of the line along the longest axis of each cardiac myocyte and measured the length of the cell under the line over time. We normalized length data to the first time point at which the cell is at rest between contractions, and detrended the signal by subtracting the best-fit sixth order polynomial. We considered a contraction event to correspond to fractional shortening of 5% or more of the initial rest length and the findpeaks function in Matlab was used to identify contractile pulses and averaged their amplitude for each cell. Frequency was calculated as the inverse of the mean period between contractile events.

Statistical analyses and reproducibility. Experimental measurements were analyzed and graphically presented by the GraphPad Prism8 software. Descriptive statistics are shown as mean ± SEM. Data are plotted as mean ± SEM with individual data points as scatter plot overlays. Normally distributed data were analyzed using two-way ANOVA followed by post hoc Tukey’s multiple comparison test for each mean comparison. Experimental mean-fold protein intensities, relative to controls from triplicate assays, were

150

considered different from controls at the p < 0.05 significance level. All immunoblots shown are representative immunoblots from a minimum of three independent biological replicates. Source data containing the raw data underlying all reported averages in graphs and charts and all original uncropped immunoblots are provided as a Source Data file.

Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

4.6 DATA AVAILABILITY The authors declare that all supporting data are available within the article and supplementary files, or from the corresponding author upon reasonable request.

Publically available RNASeq and proteomic data were downloaded and analyzed. Specifically, the following mRNA Affymetrix transcript data from GEO profiles in NCBI for mouse cardiac tissue and disease were obtained: severe DCM (GDS487/96115_at), 2d, 10d, 21d TAC (GDS794/103886_at), MI (GDS3655/14195), alpha-tropomyosin mild HCM (GDS2134/1419398_a_at), and Emery-Dreifuss (GDS2884/1426376_at). Similarly, we obtained human mRNA Affymetrix GEO profiles for human heart transplant rejection (GDS2386/208872_s_at), human inflammatory DCM (GDS2154/208873_s_at), DCM (GDS4772/8113542) as well as human idiopathic and ischemic CM (GDS651/ 208872_s_at). Data were reported as normalized hybridization signals. Comprehensive human RNASeq based transcript levels were obtained from the Human Protein Atlas Project 73. For that normal human tissue, RNA samples were extracted from frozen tissue sections in the Uppsala Biobank. Data were reported as the abundance in “Transcript Per Million” (TPM) as the sum of the TPM values of all its protein-coding transcripts 73.

The source data underlying Figs. 1e, h, 3b–e, g, 4d, e, 6c, 7i–k, 8f, k, 9e, and Supplementary Fig. 3a are provided as a Source Data file.

151

Received: 20 July 2017; Accepted: 17 December 2019; Published online: 19 February 2020

4.7 REFERENCES 1. Markgraf, D. F. et al. An ER protein functionally couples neutral lipid metabolism on lipid droplets to membrane lipid synthesis in the ER. Cell Rep. 6, 44–55 (2014). 2. MacLennan, D. H. & Kranias, E. G. Phospholamban: a crucial regulator of cardiac contractility. Nat. Rev. Mol. Cell Biol. 4, 566–577 (2003). 3. Bauer, B. W., Shemesh, T., Chen, Y. & Rapoport, T. A. A “push and slide” mechanism allows sequence-insensitive translocation of secretory proteins by the SecA ATPase. Cell 157, 1416–1429 (2014). 4. Stoica, R. et al. ER-mitochondria associations are regulated by the VAPB- PTPIP51 interaction and are disrupted by ALS/FTD-associated TDP-43. Nat. Commun. 5, 3996 (2014). 5. Terasaki, M. et al. Stacked endoplasmic reticulum sheets are connected by helicoidal membrane motifs. Cell 154, 285–296 (2013). 6. Voeltz, G. K., Prinz, W. A., Shibata, Y., Rist, J. M. & Rapoport, T. A. A class of membrane proteins shaping the tubular endoplasmic reticulum. Cell 124, 573–586 (2006). 7. Shibata, Y., Hu, J., Kozlov, M. M. & Rapoport, T. A. Mechanisms shaping the membranes of cellular organelles. Annu Rev. Cell Dev. Biol. 25, 329–354 (2009). 8. Shemesh, T. et al. A model for the generation and interconversion of ER morphologies. Proc. Natl Acad. Sci. USA 111, E5243–E5251 (2014). 9. Hu, J. et al. Membrane proteins of the endoplasmic reticulum induce high-curvature tubules. Science 319, 1247–1250 (2008). 10. Chen, S. et al. Lunapark stabilizes nascent three-way junctions in the endoplasmic reticulum. Proc. Natl Acad. Sci. USA 112, 418–423 (2015). 11. Orso, G. et al. Homotypic fusion of ER membranes requires the dynamin-like GTPase atlastin. Nature 460, 978–983 (2009).

152

12. Powers, R. E., Wang, S., Liu, T. Y. & Rapoport, T. A. Reconstitution of the tubular endoplasmic reticulum network with purified components. Nature 543, 257–260 (2017). 13. Saito, H., Kubota, M., Roberts, R. W., Chi, Q. & Matsunami, H. RTP family members induce functional expression of mammalian odorant receptors. Cell 119, 679–691 (2004). 14. Bjork, S., Hurt, C. M., Ho, V. K. & Angelotti, T. REEPs are membrane shaping adapter proteins that modulate specific g protein-coupled receptor trafficking by affecting ER cargo capacity. PLoS ONE 8, e76366 (2013). 15. Hurt, C. M. et al. REEP1 and REEP2 proteins are preferentially expressed in neuronal and neuronal-like exocytotic tissues. Brain Res. 1545, 12–22 (2014). 16. Esteves, T. et al. Loss of association of REEP2 with membranes leads to hereditary spastic paraplegia. Am. J. Hum. Genet. 94, 268–277 (2014). 17. Beetz, C. et al. A spastic paraplegia mouse model reveals REEP1-dependent ER shaping. J. Clin. Invest. 123, 4273–4282 (2013). 18. Schlaitz, A. L., Thompson, J., Wong, C. C., Yates, J. R. III & Heald, R. REEP3/4 ensure endoplasmic reticulum clearance from metaphase chromatin and proper nuclear envelope architecture. Dev. Cell 26, 315–323 (2013). 19. Arno, G. et al. Mutations in REEP6 cause autosomal-recessive retinitis pigmentosa. Am. J. Hum. Genet. 99, 1305–1315 (2016). 20. Veleri, S. et al. REEP6 mediates trafficking of a subset of Clathrin-coated vesicles and is critical for rod photoreceptor function and survival. Hum. Mol. Genet. 26, 2218–2230 (2017). 21. Hetz, C., Chevet, E. & Oakes, S. A. Erratum: proteostasis control by the unfolded protein response. Nat. Cell Biol. 17, 1088 (2015). 22. Doroudgar, S. & Glembotski, C. C. New concepts of endoplasmic reticulum function in the heart: programmed to conserve. J. Mol. Cell Cardiol. 55, 85–91 (2013). 23. Rossi, D., Barone, V., Giacomello, E., Cusimano, V. & Sorrentino, V. The sarcoplasmic reticulum: an organized patchwork of specialized domains. Traffic 9, 1044–1049 (2008).

153

24. Michalak, M. & Opas, M. Endoplasmic and sarcoplasmic reticulum in the heart. Trends Cell Biol. 19, 253–259 (2009). 25. Nahum, J. et al. Impact of longitudinal myocardial deformation on the prognosis of chronic heart failure patients. Circ. Cardiovasc. Imaging 3, 249–256 (2010). 26. Guo, A., Zhang, C., Wei, S., Chen, B. & Song, L. S. Emerging mechanisms of T- tubule remodelling in heart failure. Cardiovasc. Res. 98, 204–215 (2013). 27. Sharma, P. et al. Evolutionarily conserved intercalated disc protein Tmem65 regulates cardiac conduction and connexin 43 function. Nat. Commun. 6, 8391 (2015). 28. Sokolina, K. et al. Systematic protein-protein interaction mapping for clinically relevant human GPCRs. Mol. Syst. Biol. 13, 918 (2017). 29. Brady, J. P., Claridge, J. K., Smith, P. G. & Schnell, J. R. A conserved amphipathic helix is required for membrane tubule formation by Yop1p. Proc. Natl Acad. Sci. USA 112, E639–E648 (2015). 30. Park, S. H., Zhu, P. P., Parker, R. L. & Blackstone, C. Hereditary spastic paraplegia proteins REEP1, spastin, and atlastin-1 coordinate microtubule interactions with the tubular ER network. J. Clin. Invest. 120, 1097–1110 (2010). 31. Kim, M. S. et al. A draft map of the human proteome. Nature 509, 575–581 (2014). 32. Schmitt, J. P. et al. Dilated cardiomyopathy and heart failure caused by a mutation in phospholamban. Science 299, 1410–1413 (2003). 33. Rockman, H. A. et al. Segregation of atrial-specific and inducible expression of an atrial natriuretic factor transgene in an in vivo murine model of cardiac hypertrophy. Proc. Natl Acad. Sci. USA 88, 8277–8281 (1991). 34. Patten, R. D. et al. Ventricular remodeling in a mouse model of myocardial infarction. Am. J. Physiol. 274, H1812–H1820 (1998). 35. Chis, R. et al. α-Crystallin B prevents apoptosis after H2O2 exposure in mouse neonatal cardiomyocytes. Am. J. Physiol. Heart Circ. Physiol. 303, H967–H978 (2012). 36. Yoshida, H. ER stress and diseases. FEBS J. 274, 630–658 (2007). 37. Nakagawa, T. et al. Caspase-12 mediates endoplasmic-reticulum-specific apoptosis and cytotoxicity by amyloid-beta. Nature 403, 98–103 (2000).

154

38. Mak, A. B. et al. A lentiviral functional proteomics approach identifies chromatin remodeling complexes important for the induction of pluripotency. Mol. Cell Proteom. 9, 811–823 (2010). 39. Shibata, Y. et al. The reticulon and DP1/Yop1p proteins form immobile oligomers in the tubular endoplasmic reticulum. J. Biol. Chem. 283, 18892–18904 (2008). 40. Shah, A. N., Davey, C. F., Whitebirch, A. C., Miller, A. C. & Moens, C. B. Rapid reverse genetic screening using CRISPR in zebrafish. Nat. Methods 12, 535–540 (2015). 41. Zhu, X. Y. et al. A zebrafish heart failure model for assessing therapeutic agents. Zebrafish 15, 243–253 (2018). 42. Rossi, A. et al. Genetic compensation induced by deleterious mutations but not gene knockdowns. Nature 524, 230–233 (2015). 43. Hu, J. et al. A class of dynamin-like GTPases involved in the generation of the tubular ER network. Cell 138, 549–561 (2009). 44. Wang, S., Romano, F. B., Field, C. M., Mitchison, T. J. & Rapoport, T. A. Multiple mechanisms determine ER network morphology during the cell cycle in Xenopus egg extracts. J. Cell Biol. 203, 801–814 (2013). 45. Bian, X. et al. Structures of the atlastin GTPase provide insight into homotypic fusion of endoplasmic reticulum membranes. Proc. Natl Acad. Sci. USA 108, 3976–3981 (2011). 46. Wang, S., Tukachinsky, H., Romano, F. B. & Rapoport, T. A. Cooperation of the ER- shaping proteins atlastin, lunapark, and reticulons to generate a tubular membrane network. Elife 5, 1–29 (2016). 47. Osseni, A. et al. Triadin and CLIMP-63 form a link between triads and microtubules in muscle cells. J. Cell Sci. 129, 3744–3755 (2016). 48. Blackstone, C. Cellular pathways of hereditary spastic paraplegia. Annu Rev. Neurosci. 35, 25–47 (2012). 49. Roda, R. H., Schindler, A. B. & Blackstone, C. De novo REEP2 missense mutation in pure hereditary spastic paraplegia. Ann. Clin. Transl. Neurol. 4, 347–350 (2017). 50. Jost, A. P. & Weiner, O. D. Probing yeast polarity with acute, reversible, optogenetic inhibition of protein function. ACS Synth. Biol. 4, 1077–1085 (2015).

155

51. Stainier, D. Y. R. et al. Guidelines for morpholino use in zebrafish. PLoS Genet. 13, e1007000 (2017). 52. Yao, L. et al. REEP5 (Receptor Accessory Protein 5) acts as a sarcoplasmic reticulum membrane sculptor to modulate cardiac function. J. Am. Heart Assoc. 7, 1–15 (2018). 53. Martino, T. A. et al. Disturbed diurnal rhythm alters gene expression and exacerbates cardiovascular disease with rescue by resynchronization. Hypertension 49, 1104–1113 (2007). 54. Yang, J. et al. Proximal cerebral arteries develop myogenic responsiveness in heart failure via tumor necrosis factor-α-dependent activation of sphingosine- 1- phosphate signaling. Circulation 126, 196–206 (2012). 55. Kuzmanov, U. et al. Global phosphoproteomic profiling reveals perturbed signaling in a mouse model of dilated cardiomyopathy. Proc. Natl Acad. Sci. USA 113, 12592– 12597 (2016). 56. Ackers-Johnson, M. et al. A simplified, Langendorff-free method for concomitant isolation of viable cardiac myocytes and nonmyocytes from the adult mouse heart. Circ. Res. 119, 909–920 (2016). 57. Teng, A. C. et al. Metformin increases degradation of phospholamban via autophagy in cardiomyocytes. Proc. Natl Acad. Sci. USA 112, 7165–7170 (2015). 58. Petschnigg, J. et al. The mammalian-membrane two-hybrid assay (MaMTH) for probing membrane-protein interactions in human cells. Nat. Methods 11, 585–592 (2014). 59. Pleger, S. T. et al. Cardiac AAV9-S100A1 gene therapy rescues post-ischemic heart failure in a preclinical large animal model. Sci. Transl. Med. 3, 92ra64 (2011). 60. Fish, K. M. et al. AAV9.I-1c delivered via direct coronary infusion in a porcine model of heart failure improves contractility and mitigates adverse remodeling. Circ. Heart Fail. 6, 310–317 (2013). 61. Wang, D. Y. et al. Endoplasmic reticulum resident protein 44 (ERp44) deficiency in mice and zebrafish leads to cardiac developmental and functional defects. J. Am. Heart Assoc. 3, e001018 (2014).

156

62. Gramolini, A. O. et al. Comparative proteomics profiling of a phospholamban mutant mouse model of dilated cardiomyopathy reveals progressive intracellular stress responses. Mol. Cell Proteom. 7, 519–533 (2008). 63. Cox, J. & Mann, M. MaxQuant enables high peptide identification rates, individualized ppb-range mass accuracies and proteome-wide protein quantification. Nat. Biotechnol. 26, 1367–1372 (2008). 64. Huang, C. J., Tu, C. T., Hsiao, C. D., Hsieh, F. J. & Tsai, H. J. Germ-line transmission of a myocardium-specific GFP transgene reveals critical regulatory elements in the cardiac myosin light chain 2 promoter of zebrafish. Dev. Dyn. 228, 30–40 (2003). 65. Ekker, M., Akimenko, M. A., Bremiller, R. & Westerfield, M. Regional expression of three homeobox transcripts in the inner ear of zebrafish embryos. Neuron 9, 27–35 (1992). 66. Xu, Q., Stemple, D. & Joubin, K. Microinjection and cell transplantation in zebrafish embryos. Methods Mol. Biol. 461, 513–520 (2008). 67. Alexander, J., Stainier, D. Y. & Yelon, D. Screening mosaic F1 females for mutations affecting zebrafish heart induction and patterning. Dev. Genet. 22, 288–299 (1998). 68. Gibb, N. et al. Hey2 regulates the size of the cardiac progenitor pool during vertebrate heart development. Development 145, 1–15 (2018). 69. Varshney, G. K. et al. High-throughput gene targeting and phenotyping in zebrafish using CRISPR/Cas9. Genome Res. 25, 1030–1042 (2015). 70. Zurek, N., Sparks, L. & Voeltz, G. Reticulon short hairpin transmembrane domains are used to shape ER tubules. Traffic 12, 28–41 (2011). 71. Fernandez-Gonzalez, R. & Zallen, J. A. Oscillatory behaviors and hierarchical assembly of contractile structures in intercalating cells. Phys. Biol. 8, 045005 (2011). 72. Leung, C. Y. & Fernandez-Gonzalez, R. Quantitative image analysis of cell behavior and molecular dynamics during tissue morphogenesis. Methods Mol. Biol. 1189, 99–113 (2015). 73. Uhlén, M. et al. Proteomics. Tissue-based map of the human proteome. Science 347, 1260419 (2015).

157

Acknowledgements

We would like to thank Dr. Stephen Strittmatter (Yale University) for providing human Myc-Nogo-A and Myc-Nogo-B plasmids. We thank Aaron Wilson and Wenping Li for providing expert technical support; and Dr. Yu-Qing Zhou for providing expert experimental support for all in vivo echocardiographic experiments. This project was funded by the Ted Rogers Centre for Heart Research Innovation Fund to A.O.G. and I.C.S.; the Heart and Stroke Richard Lewar Centre of Cardiovascular Excellence; and CIHR Award to A.O.G. (PJT-155921 and PJT-166118) and a Discovery Grant from NSERC (RGPIN-#05618). S.H. Lee was supported by a NSERC Postgraduate Scholarship. S.H. Lakmehsari was supported by a CGS-Master’s Award. N.G. was supported by the Philip Witchel Research Fellowship. J.C.Y. was supported by an Ontario Graduate Scholarship.

Competing interests

The authors declare no competing interests.

Additional information

Supplementary information is available for this paper at https://doi.org/10.1038/s41467- 019-14143-9.

Correspondence and requests for materials should be addressed to A.O.G.

Peer review information Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. Peer reviewer reports are available.

Reprints and permission information is available at http://www.nature.com/reprints

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and

158

reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

159

4.8 SUPPLEMENTARY INFORMATION

Lee et al.

NATURE COMMUNICATIONS | (2020)11:965 | https://doi.org/10.1038/s41467-019- 14143-9 | www.nature.com/naturecommunications

160

Supplementary Figure 1 Phylogenetic analysis of the REEP family of proteins using Phylogeny.fr (www.phylogeny.fr) demonstrates clustering of mammalian taxa with conservation throughout multiple species. Analysis identifies two phylogenetically distinct groups: REEP1-4 and REEP5-6.

161

162

Supplementary Figure 2 REEP5 immunostaining demonstrates SR staining pattern and co-localization with known j-SR proteins, α-actinin, RyR2, and triadin in cardiac myocytes. (a) Immunofluorescence analysis of acutely isolated adult mouse cardiac myocytes with REEP5 (green) co-stained with cardiac sarcomeric α-actinin (red). Line scan analysis of yellow bar shows co-localization between REEP5 and α-actinin at the junctional-SR. Scale,10μM. (b) Three-dimensional reconstructive analysis (Imaris) demonstrates 56.78% ± 1.29% area of co-localization (overlay) between REEP5 and α- actinin with a Pearson coefficient of 0.61 ± 0.04. (c) Immunofluorescence and three- dimensional reconstructive analyses of acutely isolated adult mouse cardiac myocytes with REEP5 (green) co-stained with RyR2 (red). Scale, 10μM. (d) Immunofluorescence and three-dimensional reconstructive analyses of acutely isolated adult mouse cardiac myocytes with REEP5 (green) co-stained with triadin (red). Scale, 10μM. All images shown are representative of approximately 30-40 total images captured per condition, n=3 independent biological replicates.

163

Supplementary Figure 3 REEP5 depletion-induced SR/ER vacuolization is independent of cellular apoptosis in cardiac myocytes. Confocal imaging of REEP5 shRNA infected CMNCs stained with ER-Tracker shows consistent SR/ER luminal vacuoles in the presence of caspases inhibitor z-vad-fmk. Treatment of z-vad-fmk significantly reduced ER stress-mediated SR/ER vacuolization, indicating distinct cellular mechanisms involved in REEP5 depletion-mediated SR/ER vacuolization versus prolonged ER stress mediated cytoplasmic vacuolization. All images shown are

164

representative of 30 total images captured per condition, n=3 independent biological replicates. Scale, 10μM. Quantitative analysis was done under 40x objective lens and approximately 30-40 cells were scored and data averaged for each experimental condition. Asterisks indicate a statistically significant p value in a tukey’s multiple comparison analysis where ***p<0.001; data are presented as mean ± SEM.

165

166

Supplementary Figure 4 Overexpression of dimerized REEP5 robustly marks the ER network in mouse myoblasts. (a) Confocal imaging shows dimerized REEP5 extensively marks the ER network in C2C12 myoblasts transfected with REEP5-REEP5- EYFP. Scale, 20μM (left), 10μM (right). (b) Confocal imaging of C2C12 myoblasts cotransfected with Δ114-189 REEP5-EYFP and mCherry-α-Tubulin. Scale, 20μM (left), 10μM (right). All images shown are representative of 30-40 total images captured per condition, n=3 independent biological replicates.

167

Supplementary Figure 5 Co-immunoprecipitation assays demonstrate dynamic REEP5 interactions with RTN4/Nogo A/B, ATL3, and CKAP4 in neonatal ventricular cardiac lysate. Co-immunoprecipitation and reverse-order co-immunoprecipitation with anti-REEP5, anti-Nogo A/B, anti-ATL3, and anti-CKAP4 antibodies in neonatal ventricular cardiac lysate (input), n=3 independent biological replicates. Asterisks to the right indicate the number of predicted REEP5 monomer, dimer, trimer and oligomer detected based on anticipated molecular weight. Source data containing original uncropped immunoblots are provided as a Source Data file.

168

169

Supplementary Figure 6 Morpholino-mediated REEP5 depletion in zebrafish embryos causes severe developmental abnormalities. (a) Immunoblot analysis of REEP5 expression at 48hpf in wild-type and REEP5 MO injected embryos. (b-d) Tg(myl7:EGFP) transgenic embryos injected with 1ng REEP5 and REEP6 MOs and imaged at 48hpf reveal overt cardiac defects in REEP5 MO embryos and no observable cardiac phenotype in REEP6 MO embryos. Scale, 500μm. Images shown are representative of 43 control, 68 REEP5 MO, and 31 REEP6 MO total injected embryos, n=3. (e,f) Optical imaging of REEP5 MO embryos reveals cardiac looping defects and swelling of the atrium, ventricle, and the pericardial sac compared to fully-looped control embryo hearts. A, atrium; V, ventricle. Images shown are representative of 63 control and 84 REEP5 MO total injected embryos, n=3 independent biological crosses. Scale, 10μm. (g,h) Immunofluorescence analysis of atrial (green) and ventricular (red) chambers at 72hpf shows arrested heart looping process in REEP5 MO hearts. Scale, 10μm. (i) Movie imaging analysis showing irregular cardiac beating rhythms with reduced beating frequency and dyssynchronous atrioventricular contractions in REEP5 deficient embryos. Area profiles were smoothened with a Gaussian curve with σ=0.2s. (j) TEM of control zebrafish ventricular tissue showing intact ventricular myocardium. Scale, 1μm. (k) Higher magnification of control zebrafish ventricular myocardium showing intact sarcomeres and SR. Scale, 500nm. (l) TEM of REEP5 MO zebrafish ventricular tissue showing SR vacuolization and disorganized muscle fibers. Scale, 1μm. (m) Higher magnification of REEP5 MO ventricular myocardium showing vacuolated SR. Scale, 500nm. S, sarcomere; M, mitochondria; SR, sarcoplasmic reticulum; D, desmosomes; JS, junctional space, V, vacuole. (n,o) REEP5 mRNA rescue studies co-injecting REEP5 mRNA and REEP5 MO effectively rescued the aberrant heart morphology observed in the REEP5 morphant embryos. 500μm (p) Movie imaging analysis showing corrected cardiac beating rhythms with normal atrioventricular contractions in the rescued embryos. Area profiles were smoothened with a Gaussian curve with σ=0.2s. (q) Immunoblot analysis of cardiac- enriched ER structure proteins (RTN4, ATL3, and CKAP4) in REEP5 MO mutant hearts, n=3 independent replicates from three independent biological crosses.

170

171

Supplementary Figure 7 CRISPR/Cas9-mediated REEP5 loss-of-function crispants demonstrates cardiac developmental and functional defects in zebrafish embryos. (a) HRM analysis of REEP5 gRNA injected embryos shows efficient cutting of the reep5 gene in zebrafish. (b) Immunoblotting analysis of REEP5 and α-Tubulin levels in control and 2X REEP5 gRNA injected embryos. Source data containing origical uncropped immunoblots are provided as a Source Data file. (c,d) Immunofluorescence analysis of REEP5 expression in control and 2X REEP5 gRNA injected Tg(myl7:EGFP) transgenic embryo hearts shows depletion of REEP5 expression and linearized hearts associated with REEP5 gRNA injection. Scale, 50μM. All images shown are representative images of approximately 30 total images captured per condition, n=3 independent biological replicates.

172

Supplementary Figure 8 Increased protein expression levels of REEP5, RTN4, ATL3, and CKAP4 in pressure overload-induced mouse failing hearts. Immunoblots of REEP5, RTN4, ATL3, and CKAP4 in mouse failing hearts at 4-week and 8-week post- transverse aortic constriction, n=3 independent biological replicates. Source data containing original uncropped immunoblots are provided as a Source Data file.

173

CHAPTER 5: DISCUSSION & CONCLUSION

5.1 UNIFYING DISCUSSION

This section of the chapter is a modified version of two manuscripts that were published in American Journal of Physiology – Heart and Circulatory Physiology and Cell Stress.

Lee et al. Membrane proteomic profiling of the heart: Past, present and future. Am J Physiol Heart Circ Physiol 320(1): H417-H423 (2021). Permission to reproduce sections was obtained from the American Physiological Society (Appendix II).

Lee et al. Towards understanding the role of Receptor Expression Enhancing Protein 5 (REEP5) in cardiac muscle and beyond. Cell Stress 4(6): 151-153 (2020). This article is licensed under a Creative Commons Attribution 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Cardiovascular diseases remain the most rapidly rising contributing factor of all-cause mortality and the leading cause of inpatient hospitalization worldwide, with costs exceeding $30 billion dollars annually in North America. Cell surface and membrane associated proteins play an important role in cardiomyocyte biology and are involved in the pathogenesis of many human heart diseases. In cardiomyocytes, membrane proteins serve as critical signaling receptors, Ca2+ cycling regulators, and electrical propagation regulators, all functioning in concert to maintain spontaneous and synchronous contractions of cardiomyocytes. Perturbations in cardiac membrane protein localization and function have been implicated in the progression and pathogenesis of many heart diseases. However, previous attempts at profiling the cardiac membrane proteome have yielded limited results due to poor technological developments for isolating hydrophobic, low abundant membrane proteins. Comprehensive mapping and characterization of the cardiac membrane proteome thereby remains incomplete. This thesis sought to directly identify and characterize previously underappreciated membrane associated proteins in the heart. I pioneered a systems-biology approach to identify novel regulators of heart function by integrating multi-omics and phenotypic ontology datasets, allowing us to

174

prospectively study the link of these novel membrane associated proteins to normal heart function and heart disease.

In the first study, it was hypothesized that a systems-biology-based informatics analysis of a previous proteomics dataset of cardiomyocyte membrane-enriched proteins would reveal poorly annotated, cardiomyocyte-enriched membrane proteins that have the highest potential of being an important regulator of heart function. Indeed, our results revealed 173 cardiomyocyte-enriched conserved membrane associated proteins that have not been studied in the cardiac setting previously. In an attempt to validate the utility of this dataset, we showed differential expression of several highly ranked candidates including REEP5, FAM162A, MCT1, and COX20 in human DCM and ICM heart tissues, suggesting their potential involvement in regulating heart disease progression. Completion of this large-scale informatics analysis could not have been achieved previously due to technological and infrastructure barriers. Over the past decade, the improvements in tissue sample processing including enhanced enzymatic digestion, improved chromatography methods, and tissue fractionation have greatly increased protein resolution and coverage. These advancements along with technology-driven increases in machine sensitivity and accuracy have allowed for a comprehensive protein landscape and expression patterns of the human heart 62.

These studies together have generated in-depth, comprehensive proteomic maps of human cardiac protein expression, resulting in a reference map of human cardiac proteome with over 17,000 proteins identified (Figure 5.1). In addition to the deepest human heart proteome generated by Doll et al. in 2017, a recent quantitative proteomics study of human cardiac sinus node containing highly specialized pacemaker cells in the heart revealed >7,000 unique proteins that are important for cardiac ion channel and Ca2+ cycling activities 154. In addition, proteomic mapping of living human heart samples obtained during valve replacement surgery identified 7,314 proteins in living heart tissues 44. In comparison to these studies, our dataset identified 550 poorly studied, cardiac peripheral membrane proteins by applying systematic bioinformatic analyses to a previous proteomic mapping study of membrane-enriched fractions from human and

175

mouse hearts 155, 156. As mentioned previously, a significant number of these highlighted proteins were cardiac-enriched membrane peripheral proteins without any previously reported cardiac related phenotype in the mouse; 18 of which were considered as highest ranked novel cardiac peripheral membrane proteins based on their transcriptomic, proteomic, and phenotypic ontology expression profiles. We believe that such studies represent a unique resource of the human cardiac proteome and provide detailed insights into the molecular underpinnings of the human heart.

Figure 5.1. Human cardiac proteome consists of >17,000 unique proteins in the heart. Venn diagrams depicting overlap of protein coverage among recent in-depth, quantitative proteomics studies of the human heart.

By direct comparison of some of the recent heart proteome maps, it is clear that the proteomic analysis of the pacemaker cells and the whole heart with extensive, region specific fractionation revealed by far the most distinct, and apparently complex,

176

proteomes (Figure 5.1). While a subset of proteins appears to be unique to acutely analyzed live heart tissues, the enrichment for membrane fractions using silica beads showed limited improved unique coverage since 514 of the 550 cardiac membrane peripheral proteins have been identified in other studies, but likely at significantly lower levels due to the lack of a membrane-enriched strategy (Figure 5.1). Moving forward, as more data is available through online proteomic portals, follow-up cross correlation studies, detailed informatic mining approaches, and comparison to genomic and transcriptomic datasets will significantly increase confidence in our understanding of cardiac protein expression patterns, particularly for the lower abundant and membrane proteins.

The potential and impact of utilizing large-scale proteomic analyses to shed light on mechanistic insight of human diseases have previously yielded positive clinical impacts 157. In our study, the elevated expression of FAM162A in human DCM- and ICM-induced HF patients along with its previously reported function in mitochondrial apoptosis initiation led to the hypothesis that targeted inhibition of FAM162A expression may be an effective approach to attenuate cardiomyocyte cell death in HF patients 158, 159, 34. A global assessment of FAM162A’s binary interactome networks using the Human Reference Interactome (HuRI) database revealed its interaction with HSP90AB1, a molecular chaperone that has been shown to be essential for protein homeostasis under stressed conditions 160. These analyses indicate that there is likely an arms race between FAM162A-mediated cardiomyocyte protein homeostasis restoration and cardiac programed cell death in failing hearts; however, the molecular switch governing the two arms of downstream cellular consequences remains unexplored. While MCT1’s binary interactome has not been mapped out to date, its dominant expression in striated muscles positively correlates with the oxidative capacity of muscle cells and is essential for lactate removal from circulation. Reduced MCT1 expression observed in patients with ischemic cardiomyopathy negatively impacts lactate extrusion and transport in response to an ischemic insult, resulting in altered cellular metabolism during myocardial ischemia. However, there is currently limited evidence on the specific functional role of MCT1 in the heart. Future studies characterizing MCT1 loss-of-function mutants in vivo as well as

177

overexpression of MCT1 in cardiac ischemic disease models would prove invaluable and further our understanding of altered cardiac metabolism under pathological conditions.

Lastly, COX20, a cellular respiration facilitating protein in the mitochondria, has been shown to be important for neurological control of muscle tone and growth 161, 162. Its association with muscle dystrophy in patients with mutated COX20 along with reduced expression in ischemic heart disease suggest the functional importance of this peripheral membrane protein in regulating muscle health and disease in the heart. A global assessment of the binary protein-protein interactome networks demonstrated its association with membrane proteins involved in cellular metabolism and respiration including SLC10A1, SLC10A6, and MAOB. Moreover, a survey of tissue specific interactome networks of COX20 in the heart demonstrated its interactions with several cardiac-enriched membrane proteins, providing context specific information about the functional role of COX20 in the heart. Specifically, its interaction with FCER1G, an adapter protein essential for inflammatory signaling initiation via integrin-ß-mediated neutrophil activation as well as integrin-α-mediated platelet activation may play an important role in limiting ischemic myocardial damage in the heart. Likewise, its interaction with muscle specific DYNC1H1 might contribute to the reduced heart weight and muscular atrophy phenotype observed in heart disease patients. These hypotheses are however preliminary and warrant further discussion.

Nonetheless, the cardiovascular proteomics community has made a tremendous amount of progress to elucidating the complete cardiac membrane proteome. Characterizing the function of these poorly annotated peripheral membrane proteins in the heart is paramount to understanding cardiomyocyte function underlying various physiology and pathophysiology contexts. Current in-depth membrane proteomics studies of healthy and diseased heart tissues have greatly advanced our knowledge of the molecular underpinnings of heart function and heart disease providing an important reference map of the human proteome for future studies aimed at characterizing and identifying novel regulators of heart function and potential therapeutic targets for heart disease. Characterization of poorly annotated, cardiac-enriched membrane associated proteins is

178

highly warranted to provide a comprehensive understanding of cardiac physiology and future development of novel therapies to overcome the current limitations of managing heart diseases. Lastly, it is important to note that effective therapies for HF will not rely on single proteins but rather a group of targets to tackle the molecular complexity underlying cardiac dysfunction. Therefore, an integrated, systems-biology approach is essential for future studies to fully embrace the complexity of HF. As well, future single- cell proteomics and phosphoproteomics studies of healthy and diseased cardiomyocytes will provide important mechanistic insights into the complex signaling interplay in failing hearts.

Having established a cardiac membrane proteome and a prioritized list of novel membrane associated protein candidates in the heart, the second study examined whether the highest ranked candidate, REEP5, contributes to normal heart function and heart disease progression. Given that REEP5 harbours a core RHD domain and its hairpin-shaped topology in the cell membrane, it was hypothesized that physiological expression of REEP5 would be critical for development, maintenance, and function of the cardiomyocyte SR/ER, and ultimately, contraction of the cardiomyocytes. Additionally, our bioinformatic data across various human tissues and immunoblot analysis of multiple mouse organs suggested that although the expression of REEP5 was ubiquitously expressed across organs, it was found preferentially expressed in muscle cells with the most abundant expression in the cardiac ventricle. High-resolution three-dimensional co- immunofluorescence imaging of endogenous REEP5 in cardiac myocytes showed localization of REEP5 to different functional domains of the SR/ER network in the myocyte including the tubular/longitudinal SR/ER network where protein synthesis, trafficking, and modification takes place. REEP5 was also localized to the junctional SR, which is closely tethered to the cell surface and mainly responsible for Ca2+ cycling and the regulation of excitation-contraction coupling in the myocyte.

Consistent with our hypothesis, depletion of REEP5 in cardiomyocytes resulted in SR/ER membrane destabilization and the formation of prominent luminal vacuoles. Morphologically, without REEP5, the SR membrane network failed to form high curvature

179

SR/ER tubules that are required for proper association with t-tubules, and propagation of cardiac excitation-contraction coupling signals. Interestingly, REEP5 interactome analysis using mass spectrometry and co-immunoprecipitation studies elucidated REEP5 interactions with several members of the reticulon and atlastin families of proteins, as well as CKAP4. These interacting proteins have been linked previously to ER membrane- shaping functions including high curvature formation, tubule fusion, and intraluminal spacing. Previous studies into the role of many of these proteins in the ER have mainly focused on yeast and Xenopus laevis. However, it is now apparent that members of these protein families, likely acting in a coordinated manner, will be important for specialized ER structure and function in unique mammalian cell populations and cardiac SR structure and biology as primary cardiac myocytes are fundamentally more complex. For instance, REEP5 is the dominant REEP protein in the heart among its other five members of the REEP family of proteins, named REEP1 to REEP6. Notably, REEP1, a neuro-enriched REEP protein, and its interaction with ATL1 has been shown to be important for maintaining ER tubular membrane in corticospinal neurons; defects in ER organization and this REEP1-ATL1 interaction have been shown to be the predominant pathogenic mechanism of hereditary spastic paraplegia, a neurodegenerative disease of the upper motor neurons. Our work in identifying REEP5-ATL3 interactions in cardiomyocytes would suggest that, similar to REEP1-ATL1 interactions in corticospinal neurons, normal cardiac SR/ER function and disruptions in REEP5 network interactions could be pathogenic and result in the development of cardiomyopathy. Similar insight would be anticipated given the specific cellular distribution and enrichment of REEPs, ATLs, and RTNs in a tissue- and cell-specific manner.

An anecdote saying “structure leads to function” applies well in the fields of cardiac biology and development. We demonstrated that REEP5 depletion-induced vacuolization of the SR/ER membrane correlates with elevated levels of intracellular reactive oxygen species, dissipated mitochondrial membrane potential, dysregulated Ca2+ cycles, and increased expression levels of ER stress markers and ER-dependent apoptotic marker, caspase 12 (Figure 5.2). Importantly, we showed that the C-terminus region of REEP5 in the cytosol is the link to REEP5-induced membrane vacuolization phenotype observed in

180

cardiac myocytes; overexpression of the C-terminus truncated mutant of REEP5 led to severely vacuolated SR/ER. This data supports the existing evidence in the field that the conserved amphipathic helix domain in the C-terminus is essential for membrane stabilization support. These morphological and functional defects translated into lethal cardiac phenotypes which included degenerating muscle fibers and reduced heart rate, ejection fraction, and cardiac output in our preclinical zebrafish and mouse models of REEP5 depletion in vivo.

Figure 5.2. Summary of major functions of REEP5 in cardiac myocytes. REEP5 is a SR/ER membrane protein responsible for generating high membrane curvature to stabilize the highly differentiated SR network in cardiac myocytes. Normal expression of REEP5 under healthy physiological conditions (left) is required for proper formation, maintenance, and function of the SR/ER in the myocyte; loss of REEP5 (right) results in SR membrane vacuolization, excessive intracellular ROS, hyperpolarized mitochondrial membrane potential, activated ER stress pathways, dysregulated Ca2+ cycles, and cardiac dysfunction.

181

Extensive investigations into heart disease have highlighted the SR/ER as a focal site for the initiation and progression of many heart diseases. From our depletion and knockout studies, it is evident that REEP5 is important in the normal functioning of the heart and potentially has a role in cardiac disease. The findings in in vivo REEP5-depleted mice suggest a dilated cardiomyopathy phenotype linked with upregulation of cell stress markers. Heart failure is an increasingly common disease which has several different etiologies, some genetic and some acquired. Although over 100 genes have been associated with familial cardiomyopathy and heart failure, there are still many cases where the underlying cause of the disease remains unknown 76. From the initial data, it would appear likely that REEP5 may be linked to some of these idiopathic cases. Further genetic investigation of these cases would appear warranted, and may provide insight into understanding the mechanisms behind the disease.

These two interrelated studies were, to the best of my knowledge, the first to characterize the importance of membrane associated proteins in the heart and establish an association between altered expression of these peripheral membrane proteins and functional impairment in heart disease. However, we are only at the nascent stages of understanding the physiological significance of all 173 cardiac-enriched peripheral membrane proteins and their proteoforms in regulating heart function, and more importantly, how to translate these findings to heart disease patients and the development of novel therapeutics for HF.

5.2 LIMITATIONS Our study methods provided an excellent model for a greater assessment of the importance and function of membrane associated proteins in the heart. However, characterizing the biochemistry and physiology of these “novel” membrane associated proteins was difficult due to the biochemical nature of membrane proteins as well as the complex, interdisciplinary physiology and pathophysiology of heart disease. Therefore, certain limitations must be acknowledged and discussed. In Chapter 3, while we attempted to provide a deep proteome profile of all membrane proteins in the heart, the

182

use of mouse neonatal cardiomyocytes and human fetal cardiomyocytes in our initial MS experiments does not represent the physiology of differentiated cardiomyocytes in the mature mouse and human hearts. This approach provided only a snapshot of the cardiomyocyte proteome profile during development, and therefore, the number of proteins reported could have been underestimated due to a lack of MS analysis across multiple time points throughout development. This is an important consideration in both of our mouse and human studies, requiring us to assume that proteins that were deemed important for heart development were also important for normal heart function in the mature heart. However, it remains unclear whether membrane proteins with late-onset cardiac expression were assigned a low priority score, and therefore overlooked in our study. Additionally, the use of MGI Phenotypic Ontology dataset as a benchmark to determine cardiac novelty of the identified membrane associated proteins in our dataset warranted caution as the mouse informatics dataset does not include identified human disease mutations associated with our 550 protein candidates. As a result, the 173 novel membrane associated proteins reported in our study represented an overestimation due to the exclusion of human heart disease causing mutation data associated with our identified protein candidates.

Moreover, while our findings suggested that FAM162A, MCT1, and COX20 may be important regulators of heart function, evidence proving a causal relationship between FAM162A, MCT1, COX20 and heart disease remains to be elucidated. Moreover, clinical biopsy samples used in our study were not timepoint specific with a wide range of HF duration including both early-stage (as early as 58 months) and late-stage (as late as 81 months) samples without age-matched controls. This required future investigations incorporating a larger patient sample size of DCM and ICM to further corroborate our current findings and delineate, if any, sex differences among different clinical categories of cardiomyopathy. The addition of animal disease models of DCM and ICM in parallel to the human studies in future investigations would prove invaluable and promote the translational potential of our current findings. Notably, the use of human biopsy samples to determine the expression levels of FAM162A, MCT1, and COX20 in diseased human hearts represented a simple combined measurement of the expression patterns in

183

different cell populations in the heart including cardiomyocytes, cardiac fibroblasts, and endothelial cells. Given their favourable enrichment for the myocytes in the heart, we assumed alterations in the expression patterns observed in human DCM and ICM myocardium compared to healthy controls were a result of altered protein expression in the myocytes. However, it remains unclear whether myocytes were solely responsible for the alterations in the protein expression patterns of FAM162A, MCT1, and COX20 observed in human DCM and ICM in our study.

Antibody-based experiments must always be approached carefully. The selection and validation of antibodies used in our study was critical to generate reliable tissue expression and subcellular localization data of our prioritized membrane proteins in cardiomyocytes. While we attempted co-immunofluorescence studies of FAM162A, MCT1, and COX20 with known plasma membrane and mitochondrial markers and showed a great degree of subcellular colocalization, our immunodepletion experiments using HEK-overexpressed human FAM162A, MCT1, and COX20 lysates were a suboptimal validation system as overexpression lysates have the ability to mask both specific and non-specific signals in immunoblot and confocal imaging analyses, resulting in potential false-positive validation results. To address this limitation, one may perform immunodepletion studies using purified synthetic peptides against the epitopes of FAM162A, MCT1, and COX20 antibodies used in our study. However, another simpler solution would be to perform the same set of immunodepletion experiments with HEK lysates alone as a negative control. While we did not carry out this specific experiment, the addition of this negative control would further strengthen the specificity of the antibodies used in our current study.

This dissertation work aimed to capture low abundant cardiomyocyte membrane associated proteins that are critical contributors to normal heart function. While a main selection criterion employed was abundance based on available transcriptomic data when prioritizing candidates for follow-up studies, it is important to acknowledge that abundance does not always indicate importance in biology due to functional redundancy and context- driven high variability of the human body. In Chapter 4, our systematic ranking approach

184

resulted in the identification REEP5 as the top ranked protein candidate for detailed in vitro and in vivo follow-up studies. While this approach may not have been the perfect ranking system, our REEP5 study provided a detailed description of the fundamental role of REEP5 in SR/ER organization in the myocyte in the heart.

The use of multiple in vitro systems and in vivo animal models in our REEP5 study provided a comprehensive assessment of the role of REEP5 in the heart. However, we noticed that transient acute KD studies using lentiviral-mediated REEP5 KD in vitro in cardiomyocytes, AAV9-mediated REEP5 KD in vivo in the mouse, and CRISPR/Cas9- mediated REEP5 crispants in zebrafish embryos resulted in morphologically worsened phenotypes compared to germline CRISPR reep5 mutants in the zebrafish. While the fundamental differences in the biology and specificity of the KD methods could have accounted for this observation, we showed evidence of genetic compensation activation in germline CRISPR reep5 mutants, not in KD animal models. This suggested that the generation of germline mutants provides a critical window for functional adaptation and compensation in response to deleterious genetic editing. Therefore, characterizing the phenotypes of acute KD-driven studies must always be approached with caution due to potential over-manifestation of phenotypes that may not be substantiated in stable germline animal models or humans. To overcome this concern in in vitro studies in cardiomyocytes, the generation of a stable cell line from wild type cardiomyocytes or isolation of cardiomyocytes from germline KO animal hearts would allow us to circumvent the scientific concerns associated with acute KD studies.

Importantly, since REEP5 exhibited ubiquitous expression across major tissues despite being the dominant REEP protein in the heart, the lack of a conditional cardiac tissue- specific KO animal model or a heart disease causing human mutation in heart disease patients in our study precluded us from claiming that the cardiac phenotypes observed were specific to REEP5 depletion in the heart. Given the special heart-kidney link in regulating blood pressure and hormonal regulation of heart function, we cannot rule out the possibility that impaired renal function as a result of REEP5 depletion may have contributed to the phenotypic severity observed in REEP5 KD and KO models in the heart.

185

Lastly, the ability to translate the findings of animal models to humans in our studies is limited. While we established that REEP5 plays crucial roles in maintaining cardiac SR/ER integrity and function, and therefore heart function, mice and zebrafish models of heart disease do not fully recapitulate most aspects of human heart disease 78, 163. It is important to acknowledge that small mammalian animal models allow us to investigate the complex physiology of the human heart within a reasonable timeline while simplifying the clinical complexity of human heart disease 164. However, mice and zebrafish are not humans. As mentioned before in Chapter 2, fundamental differences between mouse and human heart physiology such as resting heart rate, Ca2+ handling mechanisms, and predominant myosin isoforms in the heart all lead to important differences in the mechanism and timeframe for developing similar phenotypes observed in the mouse in humans. Moving forward, it would be invaluable to use clinically validated endpoints and outcome measurements as well as standard background medical treatment such as beta- blockers and ACE inhibitors to ensure a greater translational value and better model human heart disease.

5.3 RECOMMENDED FUTURE DIRECTIONS

5.3.1 In Vitro and In Vivo Functional Assessments of FAM162A, MCT1, and COX20 While we established an association between the expression levels of FAM162A, MCT1, and COX20 and heart disease progression in patient biopsy samples, it is still unknown whether alterations in their expression represent a direct contributor to functional impairment observed in patients with heart failure. It would be of great value to generate in vivo KD models of FAM162A, MCT1, and COX20 using the AAV9-mediated RNAi KD system to obtain loss-of-function models in mouse, or perhaps guinea pig or other non- human primate models 165. Use of this strategy will allow for functional assessments of FAM162A, MCT1, and COX20 depletion in vivo and in vitro by isolating adult cardiomyocytes directly from FAM162A-, MCT1-, and COX20-depleted hearts 165, 118. Specifically, in vivo cardiac function assessment via electrocardiogram and echocardiography analyses would allow for detailed functional assessment post

186

FAM162A, MCT1, and COX20 depletion in the heart. Subsequent analysis of isolated FAM162A-, MCT1-, and COX20-depleted cardiomyocytes would remove any potential confounding contributions of other cell types, thereby allowing a focused analysis of the roles of FAM162, MCT1, and COX20 in cardiomyocytes. In particular, in vitro functional assessment via Fura2-AM, JC-1, Di-4-ANEPPS, and Annexin V/PI fluorescence analyses in isolated cardiomyocytes as well as human induced pluripotent stem cells (iPSCs) would allow us to confirm whether FAM162A, MCT1, and COX20 depletion leads to dysregulated Ca2+ cycling, mitochondrial membrane potential destabilization, impaired myocyte contractility, and activation of apoptosis in the myocyte in other model systems, including human iPSC models 165, 166, all of which are important hallmarks of HF.

With the evidence reflecting the function of FAM162A as a pro-apoptotic molecule upon hypoxia induction 158, 159, 167 and increased protein expression in human ICM patients, it would be of great interest to investigate whether targeted suppression of FAM162A expression in the heart results in attenuated apoptosis in LAD ligation-induced myocardial infarcted hearts. Similarly, as our findings suggested a potential role of MCT1 in regulating cardiac metabolism under normal and pathological conditions, future studies investigating whether overexpression of MCT1 in the heart can restore metabolic homeostasis and thereby delaying heart disease progression in DCM and ICM patients 168, 169. Lastly, efforts should be made to characterize the phenotypic spectrum of altered COX20 expression levels in heart disease. While it has been shown that mutations in human COX20 resulted in muscle hypotonia and dystrophy 170, 161, 162, it would be invaluable moving forward to identify the mechanism by which COX20 modulates muscle integrity in the heart.

5.3.2 Temporal Proteomic Analysis of REEP5 Depleted Mouse Hearts While we provided evidence of REEP5’s involvement in regulating cardiac ER stress pathways, ROS production, and Ca2+ handling within cardiomyocytes, the precise molecular mechanism of action by which REEP5 may regulate ROS production and the ER stress pathways in the myocyte remains to be elucidated. Future studies are warranted to provide mechanistic insight into the direct roles of ER stress pathway

187

proteins in SR/ER formation and maintenance. In fact, such studies may even help determine the therapeutic potential of REEP5 in heart disease as sustained ER stress has emerged as an important contributor to a wide range of prevalent human diseases including cardiomyopathies and congestive heart failure. Using a temporal approach of analyzing REEP5 depleted mouse hearts 1-week, 2-week, 3-week, and 4-week post KD, global proteomic analysis of myocardial tissues would allow for the identification of phenotypic differences across different timepoints post REEP5 KD as well as the onset of pathological remodeling associated with REEP5 KD in the heart. Subsequent global pathway analysis will offer the opportunity to identify the primary drivers and key regulators involved in initiating the downstream pathological remodeling processes observed in REEP5 KD mouse hearts. Additionally, using this approach will offer an opportunity to compare the phenotypic spectra among different cell populations and/or subcellular organelles in REEP5 depleted mouse hearts throughout the progression of cardiac dysfunction via sample fractionation or cell sorting prior to MS analysis.

5.3.3 Super-Resolution Microscopy Imaging of REEP5 Depleted Cardiomyocytes In order to directly examine in great detail the role in which REEP5 plays in organelle development, maintenance, and inter-organellar communication, multi-color live-cell super-resolution imaging of REEP5 depleted cardiomyocytes is needed 171, 172. Future studies using live-cell super-resolution imaging techniques will allow for direct assessment of the dynamic structural organization of the SR/ER in cardiomyocytes and provide additional insight into the distribution, cluster density, and number of detection per cluster of REEP5-, RTN4-, ATL3-, and CKAP4-positive signals in cardiomyocytes. Moreover, these studies would shed light on the intricate balance between SR/ER shaping proteins in regulating the SR/ER integrity and function under both normal physiological and pathological conditions. Importantly, these experiments will also confirm whether REEP5-associated structural impairment of the SR/ER may be mechanistically linked to SR/ER dysfunction and genetic compensation observed in our REEP5 deficient mouse cardiomyocytes and zebrafish embryos.

188

CHAPTER 6: APPENDICES

6.1 APPENDIX I: LIST OF PUBLICATIONS FIRST-AUTHORED PEER REVIEWED PUBLICATIONS

1. Lee SH, Kim DH, Kuzmanov U, Gramolini AO. (2021) Membrane proteomic profiling of the heart: Past, present and future. American Journal of Physiology – Heart and Circulatory Physiology 320(1): H417-423.

2. Lee SH, Hadipour-Lakmehsari S, Kim DH, Di Paola M, Kuzmanov U, Shah S, Lee JJH, Kislinger T, Sharma P, Oudit GY, Gramolini AO. (2020) Bioinformatic analysis of membrane and associated proteins in murine cardiomyocytes and human myocardium. Scientific Data 7(1): 425.

3. Lee SH, Hadipour-Lakmehsari S, Gramolini AO. (2020) Towards understanding the role of Receptor Expression Enhancing Protein 5 (REEP5) in cardiac muscle and beyond. Cell Stress 4(6): 151-153.

4. Callaghan NI*, Lee SH*, Hadipour-Lakmehsari S*, Lee XA, Siraj AM, Driouchi A, Yip CM, Husain M, Simmons CA, Gramolini AO. (2020) Functional culture and in vitro genetic and small-molecule manipulation of adult mouse cardiomyocytes. Communications Biology 3(1): 229. * indicates co-first authorship

5. Lee SH, Hadipour-Lakmehsari S, Murthy HR, Gibb N, Miyake T, Teng ACT, Cosme J, Yu JC, Moon M, Lim S, Wong V, Liu P, Billia F, Fernandez-Gonzalez R, Stagljar I, Sharma P, Kislinger T, Scott IC, Gramolini AO. (2020) REEP5 depletion causes sarco- endoplasmic reticulum vacuolization and cardiac functional defects. Nature Communications 11(1): 965.

6. Lee SH & Kim DH. (2019) Synapses in the heart: Sympathetic neuro-cardiac interaction modulates myocardial remodeling in healthy and diseased myocardium. Journal of Physiology 597(17): 4441-4442.

189

7. Lee SH, Murthy HR, Langburt D. (2018) Stem-cell cardiospheres for myocardial regeneration: advancing cell therapy in myocardial infarction and heart failure. Journal of Physiology 596(17): 3839-3840.

8. Lee SH, Hadipour-Lakmehsari S, Miyake T, Gramolini AO. (2018) Three-dimensional imaging reveals endo(sarco)plasmic reticulum-containing invaginations within the nucleoplasm of muscle. American Journal of Physiology – Cell Physiology 314(3): C257-C267.

9. Hadipour-Lakmehsari S*, Lee SH*, Chan SWS. (2017) Possible mechanisms of age- dependent decline in cellular function in c-kit+ cardiac progenitor cells. Journal of Physiology 595(22): 6823-6824. * indicates co-first authorship

10. Lee SH & Hadipour-Lakmehsari S. (2016) Dietary restriction and aerobic exercise attenuate obesity-induced lymphatic dysfunction. Journal of Physiology 595(6): 1855-1856.

CO-AUTHORED PEER REVIEWED PUBLICATIONS

1. Kuzmanov U, Wang EY, Vanderlaan R, Kim DH, Lee SH, Hadipour-Lakmehsari S, Guo H, Zhao Y, McFadden M, Sharma P, Billia F, Radisic M, Gramolini AO, Emili A. (2020) Mapping signaling perturbations in myocardial fibrosis via the integrative phosphoproteomic profiling of tissue from diverse sources. Nature Biomedical Engineering 4(9): 889-900.

2. Bastin G, Dissanayake K, Langburt D, Tam A, Lee SH, Lachhar K, Heximer SP. (2020) RGS4 controls Gai3-mediated regulation of Bcl-2 phosphorylation on TGN38- containing intracellular membranes. Journal of Cell Science 133(12): jcs241034.

3. Hadipour-Lakmehsari S, Driouchi A, Lee SH, Kuzmanov U, Callaghan NI, Heximer SP, Simmons CA, Yip CM, Gramolini AO. (2019) Nanoscale reorganization of

190

sarcoplasmic reticulum calcium regulating proteins in pressure-overload cardiac hypertrophy visualized by dSTORM. Scientific Reports 9(1): 7867.

4. Callaghan NI, Hadipour-Lakmehsari S, Lee SH, Gramolini AO, Simmons CA. (2019) Modeling cardiac complexity: Advancements in myocardial models and analytical techniques for physiological investigation and therapeutic development in vitro. APL Bioengineering 3(1): 011501.

5. Osman W & Lee SH. (2015) The role of muscle sympathetic nerve activity (MSNA) in limiting exercise capacity in heart failure. Journal of Physiology 593(9): 2119-2120.

191

6.2 APPENDIX II: REPRINT LICENSES

THE AMERICAN PHYSIOLOGICAL SOCIETY LICENSE TERMS AND CONDITIONS

Dec 09, 2020

This Agreement between Shin-Haw Lee ("You") and The American Physiological Society ("The American Physiological Society") consists of your license details and the terms and conditions provided by The American Physiological Society and Copyright Clearance Center.

License Number 4963731125283

License date Dec 07, 2020

Licensed Content Publisher The American Physiological Society

Licensed Content Publication Am J Physiol- Heart and Circulatory Physiology

Licensed Content Title Membrane proteomic profiling of the heart: Past, present and future

Licensed Content Author Shin-Haw Lee, Da Hye Kim, Uros Kuzmanov, et al

Licensed Content Date Nov 13, 2020

Licensed Content Volume 320

Licensed Content Issue 1

Type of Use Thesis/Dissertation

192

Requestor type author of original work

Format print and electronic

Portion full article

Will you be translating? No

World Rights English plus translations

Order reference number

Title PhD Student

Institution name University of Toronto

Expected presentation date Jan 2021

Requestor Location Shin-Haw Lee

661 University Avenue, MaRS Phase II 14th Floor, THCHR, Toronto, ON M5G 1M1 Canada

Billing Type Invoice

Billing Address Mr. Shin-Haw Lee

661 University Avenue, MaRS Phase II 14th Floor, TRCHR, Toronto, ON M5G 1M1 Canada

Total 0.00 CAD

Terms and Conditions

Terms and Conditions:

193

this license transaction, you agree to the following terms and conditions that apply to this transaction. At the time you opened your Rightslink account you had agreed to the billing and payment terms and conditions established by Copyright Clearance Center (CCC) available at http://myaccount.copyright.com

The APS hereby grants to you a nonexclusive limited license to reuse published material as requested by you, provided you have disclosed complete and accurate details of your proposed reuse of articles, figures, tables, images, and /or data in new or derivative works. Licenses are for a one-time English language use with a maximum distribution equal to the number of copies identified by you in the licensing process, unless additional options for translations or World Rights were included in your request. Any form of print or electronic republication must be completed within three years from the date hereof. Copies prepared before then may be distributed thereafter

The following conditions are required for a License of Reuse:

Attribution: You must publish in your new or derivative work a citation to the original source of the material(s) being licensed herein, including publication name, author(s), volume, year, and page number prominently displayed in the article or within the figure/image legend.

Abstracts: APS Journal article abstracts may be reproduced or translated for noncommercial purposes without requesting permission, provided the citation to the original source of the materials is included as noted above (“Attribution”). Abstracts or portions of abstracts may not be used in advertisements or commercial promotions.

Non-profit/noncommercial reuse: APS grants permission for the free reuse of APS published material in new works published for educational purposes, provided there is no charge or fee for the new work and/or the work is not directly or indirectly commercially supported or sponsored. Neither original authors nor non-authors may reuse published material in new works that are commercially supported or sponsored including reuse in a work produced by a commercial publisher without seeking permission.

194

Video and photographs: Some material published in APS publications may belong to other copyright holders and cannot be republished without their permission. The copyright holder of photographs must be ascertained from the original source by the permission requestor. Videos and podcasts may not be rebroadcast without proper attribution and permission as requested here. For further inquiries on reuse of these types of materials, please contact [email protected]

Figures/Tables/Images are available to the requestor from the APS journals website at http://www.the-aps.org/publications/journals/. The obtaining of content is a separate transaction and does not involve Rightslink or CCC, and is the responsibility of the permission seeker. Higher resolution images are available at additional charge from APS; please contact [email protected]

Original Authors of Published Works: To see a full list of original authors rights concerning their own published work http://www.the-aps.org/publications/authorinfo/copyright.htm

Content reuse rights awarded by the APS may be exercised immediately upon issuance of this license, provided full disclosure and complete and accurate details of the proposed reuse have been made; no license is finally granted unless and until full payment is received either by the publisher or by CCC as provided in CCC’s Billing and Payment Terms and Conditions. If full payment is not received on a timely basis, then any license preliminarily granted shall be deemed automatically revoked and shall be void as if never granted. Further, in the event that you breach any of these Terms and Conditions or any of CCC’s Billing and Payment Terms and Conditions, the license is automatically revoked and shall be void as if never granted. Use of materials as described in a revoked license, as well as any use of the materials beyond the scope of the license, may constitute copyright infringement and the Publisher reserves the right to take action to protect its copyright of its materials.

The APS makes no representations or warranties with respect to the licensed material. You hereby indemnify and agree to hold harmless the publisher and CCC, and their respective officers, directors, employees and agents, from and against any and all claims

195

arising out of your use of the licensed material other than as specifically authorized pursuant to this license.

This license is personal to you /your organization and may not be sublicensed, assigned, or transferred by you /your organization to another person /organization without the publisher’s permission. This license may not be amended except in writing signed by both parties, or in the case of the publisher, by CCC on the publisher’s behalf.

The APS reserves all rights not specifically granted in the combination of (i) the license details provided by you and accepted in the course of this licensing transaction, (ii) these Terms and Conditions and (iii) CCC’s Billing and Payment Terms and Conditions. v1.0

Questions? [email protected] or +1-855-239-3415 (toll free in the US) or +1- 978-646-2777.

196

CHAPTER 7: REFERENCES

1. Boateng, S.Y. & Goldspink, P.H. Assembly and maintenance of the sarcomere night and day. Cardiovasc Res 77, 667-675 (2008). 2. Hong, T. et al. Cardiac BIN1 folds T-tubule membrane, controlling ion flux and limiting arrhythmia. Nat Med 20, 624-632 (2014). 3. Morita, H., Seidman, J. & Seidman, C.E. Genetic causes of human heart failure. J Clin Invest 115, 518-526 (2005). 4. Xiao, S. & Shaw, R.M. Cardiomyocyte protein trafficking: Relevance to heart disease and opportunities for therapeutic intervention. Trends Cardiovasc Med 25, 379-389 (2015). 5. Marks, A.R. Calcium cycling proteins and heart failure: mechanisms and therapeutics. J Clin Invest 123, 46-52 (2013). 6. Doll, S. et al. Region and cell-type resolved quantitative proteomic map of the human heart. Nat Commun 8, 1469 (2017). 7. Whitelegge, J.P. Integral membrane proteins and bilayer proteomics. Anal Chem 85, 2558-2568 (2013). 8. Gao Y, Yates JR. 3rd. Protein analysis by shotgun proteomics. In Mass- Spectrometry-Based Chemical Proteomics. doi:10.1002/9781118970195. 9. Speers, A.E., Blackler, A.R. & Wu, C.C. Shotgun analysis of integral membrane proteins facilitated by elevated temperature. Anal Chem 79, 4613-4620 (2007). 10. Matsukubo, H., Matsuura, T., Endo, N., Asayama, J. & Watanabe, T. Echocardiographic measurement of right ventricular wall thickness. A new application of subxiphoid echocardiography. Circulation 56, 278-284 (1977). 11. Sandstede, J. et al. Age- and gender-specific differences in left and right ventricular cardiac function and mass determined by cine magnetic resonance imaging. Eur Radiol 10, 438-442 (2000). 12. Kawel, N. et al. Normal left ventricular myocardial thickness for middle-aged and older subjects with steady-state free precession cardiac magnetic resonance: the multi-ethnic study of atherosclerosis. Circ Cardiovasc Imaging 5, 500-508 (2012).

197

13. Saffitz, J.E., Kanter, H.L., Green, K.G., Tolley, T.K. & Beyer, E.C. Tissue-specific determinants of anisotropic conduction velocity in canine atrial and ventricular myocardium. Circ Res 74, 1065-1070 (1994). 14. Zhou, P. & Pu, W.T. Recounting Cardiac Cellular Composition. Circ Res, 118, 368- 370 (2016). 15. Tirziu, D., Giordano, F.J. & Simons, M. Cell communications in the heart. Circulation 122, 928-937 (2010). 16. Frank, D. & Frey, N. Cardiac Z-disc signaling network. J Biol Chem 286, 9897- 9904 (2011). 17. Andersson, D.C. & Marks, A.R. Fixing ryanodine receptor Ca leak - a novel therapeutic strategy for contractile failure in heart and skeletal muscle. Drug Discov Today Dis Mech 7, e151-e157 (2010). 18. Arai, M., Alpert, N.R., MacLennan, D.H., Barton, P. & Periasamy, M. Alterations in sarcoplasmic reticulum gene expression in human heart failure. A possible mechanism for alterations in systolic and diastolic properties of the failing myocardium. Circ Res 72, 463-469 (1993). 19. Westrate, L.M., Lee, J.E., Prinz, W.A. & Voeltz, G.K. Form follows function: the importance of endoplasmic reticulum shape. Annu Rev Biochem 84, 791-811 (2015). 20. Voeltz, G.K., Prinz, W.A., Shibata, Y., Rist, J.M. & Rapoport, T.A. A class of membrane proteins shaping the tubular endoplasmic reticulum. Cell 124, 573-586 (2006). 21. Shibata, Y. et al. The reticulon and DP1/Yop1p proteins form immobile oligomers in the tubular endoplasmic reticulum. J Biol Chem 283, 18892-18904 (2008). 22. Audhya, A., Desai, A. & Oegema, K. A role for Rab5 in structuring the endoplasmic reticulum. J Cell Biol 178, 43-56 (2007). 23. Hu, J. et al. A class of dynamin-like GTPases involved in the generation of the tubular ER network. Cell 138, 549-561 (2009). 24. Orso, G. et al. Homotypic fusion of ER membranes requires the dynamin-like GTPase atlastin. Nature 460, 978-983 (2009).

198

25. Winsor, J., Machi, U., Han, Q., Hackney, D.D. & Lee, T.H. GTP hydrolysis promotes disassembly of the atlastin crossover dimer during ER fusion. J Cell Biol 217, 4184-4198 (2018). 26. Liu, T.Y. et al. Lipid interaction of the C terminus and association of the transmembrane segments facilitate atlastin-mediated homotypic endoplasmic reticulum fusion. Proc Natl Acad Sci U S A 109, E2146-2154 (2012). 27. Faust, J.E. et al. The Atlastin C-terminal tail is an amphipathic helix that perturbs the bilayer structure during endoplasmic reticulum homotypic fusion. J Biol Chem 290, 4772-4783 (2015). 28. Chen, S., Novick, P. & Ferro-Novick, S. ER network formation requires a balance of the dynamin-like GTPase Sey1p and the Lunapark family member Lnp1p. Nat Cell Biol 14, 707-716 (2012). 29. Chen, S. et al. Lunapark stabilizes nascent three-way junctions in the endoplasmic reticulum. Proc Natl Acad Sci U S A 112, 418-423 (2015). 30. Wang, S., Tukachinsky, H., Romano, F.B. & Rapoport, T.A. Cooperation of the ER-shaping proteins atlastin, lunapark, and reticulons to generate a tubular membrane network. Elife 5 (2016). 31. Nixon-Abell, J. et al. Increased spatiotemporal resolution reveals highly dynamic dense tubular matrices in the peripheral ER. Science 354 (2016). 32. Kaisto, T. & Metsikkö, K. Distribution of the endoplasmic reticulum and its relationship with the sarcoplasmic reticulum in skeletal myofibers. Exp Cell Res 289, 47-57 (2003). 33. Rao, K. et al. Spastin, atlastin, and ER relocalization are involved in axon but not dendrite regeneration. Mol Biol Cell 27, 3245-3256 (2016). 34. Hetz, C. The unfolded protein response: controlling cell fate decisions under ER stress and beyond. Nat Rev Mol Cell Biol 13, 89-102 (2012). 35. Zong, W.X. et al. Bax and Bak can localize to the endoplasmic reticulum to initiate apoptosis. J Cell Biol 162, 59-69 (2003). 36. Taylor, R.C., Cullen, S.P. & Martin, S.J. Apoptosis: controlled demolition at the cellular level. Nat Rev Mol Cell Biol 9, 231-241 (2008).

199

37. Xu, C., Bailly-Maitre, B. & Reed, J.C. Endoplasmic reticulum stress: cell life and death decisions. J Clin Invest 115, 2656-2664 (2005). 38. Okada, K. et al. Prolonged endoplasmic reticulum stress in hypertrophic and failing heart after aortic constriction: possible contribution of endoplasmic reticulum stress to cardiac myocyte apoptosis. Circulation 110, 705-712 (2004). 39. Yao, Y. et al. A non-canonical pathway regulates ER stress signaling and blocks ER stress-induced apoptosis and heart failure. Nat Commun 8, 133 (2017). 40. Kislinger, T. et al. Global survey of organ and organelle protein expression in mouse: combined proteomic and transcriptomic profiling. Cell 125, 173-186 (2006). 41. Kuzmanov, U. et al. Global phosphoproteomic profiling reveals perturbed signaling in a mouse model of dilated cardiomyopathy. Proc Natl Acad Sci U S A 113, 12592- 12597 (2016). 42. Huber, L.A., Pfaller, K. & Vietor, I. Organelle proteomics: implications for subcellular fractionation in proteomics. Circ Res 92, 962-968 (2003). 43. Fujinaka, C.M., Waas, M. & Gundry, R.L. Mass Spectrometry-Based Identification of Extracellular Domains of Cell Surface N-Glycoproteins: Defining the Accessible Surfaceome for Immunophenotyping Stem Cells and Their Derivatives. Methods Mol Biol 1722, 57-78 (2018). 44. Linscheid, N. et al. Quantitative proteomics of human heart samples collected in vivo reveal the remodeled protein landscape of dilated left atrium without atrial fibrillation. Mol Cell Proteomics 19, 1132-1144 (2020). 45. Forner, F., Foster, L.J., Campanaro, S., Valle, G. & Mann, M. Quantitative proteomic comparison of rat mitochondria from muscle, heart, and liver. Mol Cell Proteomics 5, 608-619 (2006). 46. Donnelly, D.P. et al. Best practices and benchmarks for intact protein analysis for top-down mass spectrometry. Nat Methods 16, 587-594 (2019). 47. Laganowsky, A., Reading, E., Hopper, J.T. & Robinson, C.V. Mass spectrometry of intact membrane protein complexes. Nat Protoc 8, 639-651 (2013). 48. Franklin, S. & Vondriska, T.M. Genomes, proteomes, and the central dogma. Circ Cardiovasc Genet 4, 576 (2011).

200

49. Beltrao, P., Bork, P., Krogan, N.J. & van Noort, V. Evolution and functional crosstalk of protein post-translational modifications. Mol Syst Biol 9, 714 (2013). 50. Kuzmanov, U. et al. Mapping signalling perturbations in myocardial fibrosis via the integrative phosphoproteomic profiling of tissue from diverse sources. Nat Biomed Eng 4, 889–900 (2020). 51. Thorner, J., Hunter, T., Cantley, L.C. & Sever, R. Signal transduction: From the atomic age to the post-genomic era. Cold Spring Harb Perspect Biol 6, a022913 (2014). 52. Hunter, T. The age of crosstalk: phosphorylation, ubiquitination, and beyond. Mol Cell 28, 730-738 (2007). 53. Fert-Bober, J., Murray, C.I., Parker, S.J. & Van Eyk, J.E. Precision Profiling of the Cardiovascular Post-Translationally Modified Proteome: Where There Is a Will, There Is a Way. Circ Res 122, 1221-1237 (2018). 54. Chung, N.C. et al. Unsupervised classification of multi-omics data during cardiac remodeling using deep learning. Methods 166, 66-73 (2019). 55. Fert-Bober, J. et al. Citrullination of myofilament proteins in heart failure. Cardiovasc Res 108, 232-242 (2015). 56. Lander, E.S. et al. Initial sequencing and analysis of the human genome. Nature 409, 860-921 (2001). 57. Venter, J.C. et al. The sequence of the human genome. Science 291, 1304-1351 (2001). 58. Lau, E. et al. Integrated omics dissection of proteome dynamics during cardiac remodeling. Nat Commun 9, 120 (2018). 59. O'Farrell, P.H. High resolution two-dimensional electrophoresis of proteins. J Biol Chem 250, 4007-4021 (1975). 60. Li, W. et al. Proteomic analysis of metabolic, cytoskeletal and stress response proteins in human heart failure. J Cell Mol Med 16, 59-71 (2012). 61. Roselló-Lletí, E. et al. Cardiac protein changes in ischaemic and dilated cardiomyopathy: a proteomic study of human left ventricular tissue. J Cell Mol Med 16, 2471-2486 (2012).

201

62. Omenn, G.S. et al. Research on the Human Proteome Reaches a Major Milestone: >90% of Predicted Human Proteins Now Credibly Detected, According to the HUPO Human Proteome Project. J Proteome Res (2020). 63. Gao, Y. et al. Constructing Ecological Networks Based on Habitat Quality Assessment: A Case Study of Changzhou, China. Sci Rep 7, 46073 (2017). 64. Tucker, N.R. et al. Transcriptional and Cellular Diversity of the Human Heart. Circulation 142, 466-482 (2020). 65. Hu, P. et al. Single-nucleus transcriptomic survey of cell diversity and functional maturation in postnatal mammalian hearts. Genes Dev 32, 1344-1357 (2018). 66. Perez-Riverol, Y. et al. Quantifying the impact of public omics data. Nat Commun 10, 3512 (2019). 67. Luck, K. et al. A reference map of the human binary protein interactome. Nature 580, 402-408 (2020). 68. Liu, S. et al. In-depth proteomic profiling of left ventricular tissues in human end- stage dilated cardiomyopathy. Oncotarget 8, 48321-48332 (2017). 69. Isserlin, R. et al. Pathway analysis of dilated cardiomyopathy using global proteomic profiling and enrichment maps. Proteomics 10, 1316-1327 (2010). 70. Agnetti, G., Husberg, C. & Van Eyk, J.E. Divide and conquer: the application of organelle proteomics to heart failure. Circ Res 108, 512-526 (2011). 71. Adaniya, S.M., J, O.U., Cypress, M.W., Kusakari, Y. & Jhun, B.S. Posttranslational modifications of mitochondrial fission and fusion proteins in cardiac physiology and pathophysiology. Am J Physiol Cell Physiol 316, C583-C604 (2019). 72. Wright, P.T. et al. Cardiomyocyte Membrane Structure and cAMP Compartmentation Produce Anatomical Variation in β(2)AR-cAMP Responsiveness in Murine Hearts. Cell Rep 23, 459-469 (2018). 73. Soni, S. et al. A Proteomics Approach to Identify New Putative Cardiac Intercalated Disk Proteins. PLoS One 11, e0152231 (2016). 74. Thiene, G., Corrado, D. & Basso, C. Arrhythmogenic right ventricular cardiomyopathy/dysplasia. Orphanet J Rare Dis 2, 45 (2007). 75. Gerull, B. et al. Mutations in the desmosomal protein plakophilin-2 are common in arrhythmogenic right ventricular cardiomyopathy. Nat Genet 36, 1162-1164 (2004).

202

76. McNally, E.M., Barefield, D.Y. & Puckelwartz, M.J. The genetic landscape of cardiomyopathy and its role in heart failure. Cell Metab 21, 174-182 (2015). 77. Milani-Nejad, N. & Janssen, P.M. Small and large animal models in cardiac contraction research: advantages and disadvantages. Pharmacol Ther 141, 235- 249 (2014). 78. Camacho, P., Fan, H., Liu, Z. & He, J.Q. Small mammalian animal models of heart disease. Am J Cardiovasc Dis 6, 70-80 (2016). 79. Porrello, E.R. et al. Transient regenerative potential of the neonatal mouse heart. Science 331, 1078-1080 (2011). 80. Howe, K. et al. The zebrafish reference genome sequence and its relationship to the human genome. Nature 496, 498-503 (2013). 81. Grunwald, D.J. & Streisinger, G. Induction of recessive lethal and specific locus mutations in the zebrafish with ethyl nitrosourea. Genet Res 59, 103-116 (1992). 82. Asnani, A. & Peterson, R.T. The zebrafish as a tool to identify novel therapies for human cardiovascular disease. Dis Model Mech 7, 763-767 (2014). 83. Burns, C.G. et al. High-throughput assay for small molecules that modulate zebrafish embryonic heart rate. Nat Chem Biol 1, 263-264 (2005). 84. Chi, N.C. et al. Genetic and physiologic dissection of the vertebrate cardiac conduction system. PLoS Biol 6, e109 (2008). 85. Poss, K.D., Wilson, L.G. & Keating, M.T. Heart regeneration in zebrafish. Science 298, 2188-2190 (2002). 86. Kikuchi, K. & Poss, K.D. Cardiac regenerative capacity and mechanisms. Annu Rev Cell Dev Biol 28, 719-741 (2012). 87. Kupperman, E., An, S., Osborne, N., Waldron, S. & Stainier, D.Y. A sphingosine- 1-phosphate receptor regulates cell migration during vertebrate heart development. Nature 406, 192-195 (2000). 88. Kawahara, A. et al. The sphingolipid transporter spns2 functions in migration of zebrafish myocardial precursors. Science 323, 524-527 (2009). 89. Moorman, A., Webb, S., Brown, N.A., Lamers, W. & Anderson, R.H. Development of the heart: (1) formation of the cardiac chambers and arterial trunks. Heart 89, 806-814 (2003).

203

90. Olson, E.N. Gene regulatory networks in the evolution and development of the heart. Science 313, 1922-1927 (2006). 91. Anderson, R.H., Webb, S., Brown, N.A., Lamers, W. & Moorman, A. Development of the heart: (2) Septation of the atriums and ventricles. Heart 89, 949-958 (2003). 92. Armstrong, E.J. & Bischoff, J. Heart valve development: endothelial cell signaling and differentiation. Circ Res 95, 459-470 (2004). 93. Fishman, M.C. & Chien, K.R. Fashioning the vertebrate heart: earliest embryonic decisions. Development 124, 2099-2117 (1997). 94. Poon, K.L. & Brand, T. The zebrafish model system in cardiovascular research: A tiny fish with mighty prospects. Glob Cardiol Sci Pract 2013, 9-28 (2013). 95. Wessels, A. & Sedmera, D. Developmental anatomy of the heart: a tale of mice and man. Physiol Genomics 15, 165-176 (2003). 96. Verkerk, A.O. & Remme, C.A. Zebrafish: a novel research tool for cardiac (patho)electrophysiology and ion channel disorders. Front Physiol 3, 255 (2012). 97. Meijler, F.L. Atrioventricular conduction versus heart size from mouse to whale. J Am Coll Cardiol 5, 363-365 (1985). 98. Barth, E., Stämmler, G., Speiser, B. & Schaper, J. Ultrastructural quantitation of mitochondria and myofilaments in cardiac muscle from 10 different animal species including man. J Mol Cell Cardiol 24, 669-681 (1992). 99. Vinhas, M., Araújo, A.C., Ribeiro, S., Rosário, L.B. & Belo, J.A. Transthoracic echocardiography reference values in juvenile and adult 129/Sv mice. Cardiovasc Ultrasound 11, 12 (2013). 100. Tanaka, N. et al. Transthoracic echocardiography in models of cardiac disease in the mouse. Circulation 94, 1109-1117 (1996). 101. Wang, L.W. et al. Standardized echocardiographic assessment of cardiac function in normal adult zebrafish and heart disease models. Dis Model Mech 10, 63-76 (2017). 102. van Opbergen, C.J.M., van der Voorn, S.M., Vos, M.A., de Boer, T.P. & van Veen, T.A.B. Cardiac Ca(2+) signalling in zebrafish: Translation of findings to man. Prog Biophys Mol Biol 138, 45-58 (2018).

204

103. Kolk, M.V. et al. LAD-ligation: a murine model of myocardial infarction. J Vis Exp (2009). 104. deAlmeida, A.C., van Oort, R.J. & Wehrens, X.H. Transverse aortic constriction in mice. J Vis Exp (2010). 105. Agbulut, O. et al. Temporal patterns of bone marrow cell differentiation following transplantation in doxorubicin-induced cardiomyopathy. Cardiovasc Res 58, 451- 459 (2003). 106. Huang, C.C., Chen, P.C., Huang, C.W. & Yu, J. Aristolochic Acid induces heart failure in zebrafish embryos that is mediated by inflammation. Toxicol Sci 100, 486- 494 (2007). 107. Zhu, X.Y. et al. A Zebrafish Heart Failure Model for Assessing Therapeutic Agents. Zebrafish 15, 243-253 (2018). 108. Ding, Y. et al. Haploinsufficiency of target of rapamycin attenuates cardiomyopathies in adult zebrafish. Circ Res 109, 658-669 (2011). 109. Arber, S. et al. MLP-deficient mice exhibit a disruption of cardiac cytoarchitectural organization, dilated cardiomyopathy, and heart failure. Cell 88, 393-403 (1997). 110. Jones, L.R. et al. Regulation of Ca2+ signaling in transgenic mouse cardiac myocytes overexpressing calsequestrin. J Clin Invest 101, 1385-1393 (1998). 111. Harris, S.P. et al. Hypertrophic cardiomyopathy in cardiac myosin binding protein- C knockout mice. Circ Res 90, 594-601 (2002). 112. Schiattarella, G.G. et al. Nitrosative stress drives heart failure with preserved ejection fraction. Nature 568, 351-356 (2019). 113. Sehnert, A.J. et al. Cardiac troponin T is essential in sarcomere assembly and cardiac contractility. Nat Genet 31, 106-110 (2002). 114. Hassel, D. et al. Nexilin mutations destabilize cardiac Z-disks and lead to dilated cardiomyopathy. Nat Med 15, 1281-1288 (2009). 115. Vogel, B. et al. In-vivo characterization of human dilated cardiomyopathy genes in zebrafish. Biochem Biophys Res Commun 390, 516-522 (2009). 116. Sohal, D.S. et al. Temporally regulated and tissue-specific gene manipulations in the adult and embryonic heart using a tamoxifen-inducible Cre protein. Circ Res 89, 20-25 (2001).

205

117. Poon, K.L., Liebling, M., Kondrychyn, I., Garcia-Lecea, M. & Korzh, V. Zebrafish cardiac enhancer trap lines: new tools for in vivo studies of cardiovascular development and disease. Dev Dyn 239, 914-926 (2010). 118. Callaghan, N.I. et al. Functional culture and in vitro genetic and small-molecule manipulation of adult mouse cardiomyocytes. Commun Biol 3, 229 (2020). 119. Mestas, J. & Hughes, C.C. Of mice and not men: differences between mouse and human immunology. J Immunol 172, 2731-2738 (2004). 120. Libby, P., Ridker, P.M. & Hansson, G.K. Progress and challenges in translating the biology of atherosclerosis. Nature 473, 317-325 (2011). 121. Masopust, D., Sivula, C.P. & Jameson, S.C. Of Mice, Dirty Mice, and Men: Using Mice To Understand Human Immunology. J Immunol 199, 383-388 (2017). 122. Stainier, D.Y.R. et al. Guidelines for morpholino use in zebrafish. PLoS Genet 13, e1007000 (2017). 123. Howe, K. et al. The zebrafish reference genome sequence and its relationship to the human genome. Nature 496, 498-503 (2013). 124. Ross, H. et al. Treating the right patient at the right time: access to heart failure care. Can J Cardiol 22, 749-754 (2006). 125. Trends in survival after a diagnosis of heart failure in the United Kingdom 2000- 2017: population based cohort study. BMJ 367, l5840 (2019). 126. Kumar, R. et al. Urbanization and coronary heart disease: a study of urban-rural differences in northern India. Indian Heart J 58, 126-130 (2006). 127. Sliwa, K., Acquah, L., Gersh, B.J. & Mocumbi, A.O. Impact of Socioeconomic Status, Ethnicity, and Urbanization on Risk Factor Profiles of Cardiovascular Disease in Africa. Circulation 133, 1199-1208 (2016). 128. Mosca, L., Barrett-Connor, E. & Wenger, N.K. Sex/gender differences in cardiovascular disease prevention: what a difference a decade makes. Circulation 124, 2145-2154 (2011). 129. Azad, N., Kathiravelu, A., Minoosepeher, S., Hebert, P. & Fergusson, D. Gender differences in the etiology of heart failure: A systematic review. J Geriatr Cardiol 8, 15-23 (2011).

206

130. Yancy, C.W. et al. 2016 ACC/AHA/HFSA Focused Update on New Pharmacological Therapy for Heart Failure: An Update of the 2013 ACCF/AHA Guideline for the Management of Heart Failure: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines and the Heart Failure Society of America. Circulation 134, e282-293 (2016). 131. Ziaeian, B. & Fonarow, G.C. Epidemiology and aetiology of heart failure. Nat Rev Cardiol 13, 368-378 (2016). 132. Global, regional, and national incidence, prevalence, and years lived with disability for 301 acute and chronic diseases and injuries in 188 countries, 1990-2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet 386, 743- 800 (2015). 133. Japp, A.G., Gulati, A., Cook, S.A., Cowie, M.R. & Prasad, S.K. The Diagnosis and Evaluation of Dilated Cardiomyopathy. J Am Coll Cardiol 67, 2996-3010 (2016). 134. Pinali, C., Bennett, H., Davenport, J.B., Trafford, A.W. & Kitmitto, A. Three- dimensional reconstruction of cardiac sarcoplasmic reticulum reveals a continuous network linking transverse-tubules: this organization is perturbed in heart failure. Circ Res 113, 1219-1230 (2013). 135. Nakao, K., Minobe, W., Roden, R., Bristow, M.R. & Leinwand, L.A. Myosin heavy chain gene expression in human heart failure. J Clin Invest 100, 2362-2370 (1997). 136. de Lucia, C., Eguchi, A. & Koch, W.J. New Insights in Cardiac β-Adrenergic Signaling During Heart Failure and Aging. Front Pharmacol 9, 904 (2018). 137. Lai, L.P., Raju, V.S., Delehanty, J.M., Yatani, A. & Liang, C.S. Altered sarcoplasmic reticulum Ca2+ ATPase gene expression in congestive heart failure: effect of chronic norepinephrine infusion. J Mol Cell Cardiol 30, 175-185 (1998). 138. Liu, T. et al. Current Understanding of the Pathophysiology of Myocardial Fibrosis and Its Quantitative Assessment in Heart Failure. Front Physiol 8, 238 (2017). 139. Dunlay, S.M., Roger, V.L. & Redfield, M.M. Epidemiology of heart failure with preserved ejection fraction. Nat Rev Cardiol 14, 591-602 (2017).

207

140. Holland, D.J., Kumbhani, D.J., Ahmed, S.H. & Marwick, T.H. Effects of treatment on exercise tolerance, cardiac function, and mortality in heart failure with preserved ejection fraction. A meta-analysis. J Am Coll Cardiol 57, 1676-1686 (2011). 141. Borlaug, B.A. & Paulus, W.J. Heart failure with preserved ejection fraction: pathophysiology, diagnosis, and treatment. Eur Heart J 32, 670-679 (2011). 142. Owan, T.E. et al. Trends in prevalence and outcome of heart failure with preserved ejection fraction. N Engl J Med 355, 251-259 (2006). 143. Bhatia, R.S. et al. Outcome of heart failure with preserved ejection fraction in a population-based study. N Engl J Med 355, 260-269 (2006). 144. Redfield, M.M. et al. Effect of phosphodiesterase-5 inhibition on exercise capacity and clinical status in heart failure with preserved ejection fraction: a randomized clinical trial. JAMA 309, 1268-1277 (2013). 145. Shah, S.J. et al. Phenotype-Specific Treatment of Heart Failure With Preserved Ejection Fraction: A Multiorgan Roadmap. Circulation 134, 73-90 (2016). 146. Yip, G.W., Ho, P.P., Woo, K.S. & Sanderson, J.E. Comparison of frequencies of left ventricular systolic and diastolic heart failure in Chinese living in Hong Kong. Am J Cardiol 84, 563-567 (1999). 147. Cain, B.S. et al. Human SERCA2a levels correlate inversely with age in senescent human myocardium. J Am Coll Cardiol 32, 458-467 (1998). 148. Schmidt, U. et al. Restoration of diastolic function in senescent rat hearts through adenoviral gene transfer of sarcoplasmic reticulum Ca(2+)-ATPase. Circulation 101, 790-796 (2000). 149. Glezeva, N. & Baugh, J.A. Role of inflammation in the pathogenesis of heart failure with preserved ejection fraction and its potential as a therapeutic target. Heart Fail Rev 19, 681-694 (2014). 150. Paulus, W.J. & Tschöpe, C. A novel paradigm for heart failure with preserved ejection fraction: comorbidities drive myocardial dysfunction and remodeling through coronary microvascular endothelial inflammation. J Am Coll Cardiol 62, 263-271 (2013).

208

151. Westermann, D. et al. Cardiac inflammation contributes to changes in the extracellular matrix in patients with heart failure and normal ejection fraction. Circ Heart Fail 4, 44-52 (2011). 152. Mohammed, S.F. et al. Left ventricular amyloid deposition in patients with heart failure and preserved ejection fraction. JACC Heart Fail 2, 113-122 (2014). 153. Edelmann, F. et al. Effect of spironolactone on diastolic function and exercise capacity in patients with heart failure with preserved ejection fraction: the Aldo- DHF randomized controlled trial. JAMA 309, 781-791 (2013). 154. Linscheid, N. et al. Quantitative proteomics and single-nucleus transcriptomics of the sinus node elucidates the foundation of cardiac pacemaking. Nat Commun 10, 2889 (2019). 155. Sharma, P. et al. Evolutionarily conserved intercalated disc protein Tmem65 regulates cardiac conduction and connexin 43 function. Nat Commun 6, 8391 (2015). 156. Lee, S.H. et al. Bioinformatic analysis of membrane and associated proteins in murine cardiomyocytes and human myocardium. Sci Data 7, 425 (2020). 157. Jiang, Y. et al. Proteomics identifies new therapeutic targets of early-stage hepatocellular carcinoma. Nature 567, 257-261 (2019). 158. Lee, M.J., Kim, J.Y., Suk, K. & Park, J.H. Identification of the hypoxia-inducible factor 1 alpha-responsive HGTD-P gene as a mediator in the mitochondrial apoptotic pathway. Mol Cell Biol 24, 3918-3927 (2004). 159. Kim, J.Y., Kim, S.M., Ko, J.H., Yim, J.H. & Park, J.H. Interaction of pro-apoptotic protein HGTD-P with heat shock protein 90 is required for induction of mitochondrial apoptotic cascades. FEBS Lett 580, 3270-3275 (2006). 160. Haase, M. & Fitze, G. HSP90AB1: Helping the good and the bad. Gene 575, 171- 186 (2016). 161. Szklarczyk, R. et al. A mutation in the FAM36A gene, the human ortholog of COX20, impairs cytochrome c oxidase assembly and is associated with ataxia and muscle hypotonia. Hum Mol Genet 22, 656-667 (2013). 162. Doss, S. et al. Recessive dystonia-ataxia syndrome in a Turkish family caused by a COX20 (FAM36A) mutation. J Neurol 261, 207-212 (2014).

209

163. Vegter, E.L. et al. Rodent heart failure models do not reflect the human circulating microRNA signature in heart failure. PLoS One 12, e0177242 (2017). 164. Breckenridge, R. Heart failure and mouse models. Dis Model Mech 3, 138-143 (2010). 165. Lee, S.H. et al. REEP5 depletion causes sarco-endoplasmic reticulum vacuolization and cardiac functional defects. Nat Commun 11, 965 (2020). 166. Sharma, P. et al. Evolutionarily conserved intercalated disc protein Tmem65 regulates cardiac conduction and connexin 43 function. Nat Commun 6, 8391 (2015). 167. Qu, Y. et al. MiR-139-5p inhibits HGTD-P and regulates neuronal apoptosis induced by hypoxia-ischemia in neonatal rats. Neurobiol Dis 63, 184-193 (2014). 168. Jaswal, J.S., Keung, W., Wang, W., Ussher, J.R. & Lopaschuk, G.D. Targeting fatty acid and carbohydrate oxidation--a novel therapeutic intervention in the ischemic and failing heart. Biochim Biophys Acta 1813, 1333-1350 (2011). 169. Jóhannsson, E. et al. Upregulation of the cardiac monocarboxylate transporter MCT1 in a rat model of congestive heart failure. Circulation 104, 729-734 (2001). 170. Elliott, L.E., Saracco, S.A. & Fox, T.D. Multiple roles of the Cox20 chaperone in assembly of Saccharomyces cerevisiae cytochrome c oxidase. Genetics 190, 559- 567 (2012). 171. Hadipour-Lakmehsari, S. et al. Nanoscale reorganization of sarcoplasmic reticulum in pressure-overload cardiac hypertrophy visualized by dSTORM. Sci Rep 9, 7867 (2019). 172. Kohl, T., Westphal, V., Hell, S.W. & Lehnart, S.E. Superresolution microscopy in heart - cardiac nanoscopy. J Mol Cell Cardiol 58, 13-21 (2013).

210

211