The Control of Formation and Recovery of Transverse Tubules in the Heart

The control of formation and recovery of transverse tubules in the heart

A thesis submitted to The University of Manchester for the degree of Doctor of Philosophy in the Faculty of Biology, Medicine and Health

2019

Jessica L. Caldwell

School of Medical Sciences, Division of Cardiovascular Sciences

Table of contents

Table of contents ...... 1

List of figures ...... 9

List of tables ...... 11

List of abbreviations ...... 12

Abstract ...... 15

Declaration ...... 16

Acknowledgments ...... 18

Contributions ...... 19

1. General Introduction ...... 20

1.1. Introduction ...... 20

1.2. Excitation contraction coupling ...... 20

1.3. The transverse tubule network ...... 23

1.3.1. Transverse tubules and EC coupling...... 24

1.3.2. Transverse tubules: Signalling pathways ...... 24

1.3.3. Transverse tubules: Cellular differences ...... 26

1.4. Transverse tubule alterations in disease...... 28

1.4.1. Heart failure ...... 28

1.4.2. Transverse tubule remodelling in heart failure ...... 28

1.4.3. Transverse tubule remodelling in atrial fibrillation ...... 29

1.4.4. Transverse tubule remodelling: altered calcium handling...... 30

1.4.5. Reversal of transverse tubule remodelling ...... 32

1.5. Biogenesis of transverse tubules ...... 34

1.5.1. Phosphatidylinositols (PI) ...... 36

1.5.2. Amphiphysin II (BIN1) ...... 37

1.5.3. Myotubularin (MTM1) ...... 41

1.5.4. Junctophilin (JPH2) ...... 43

1.5.5. Telethonin ...... 45

1.5.6. Proteins of the z-line ...... 47

1.6. Aims ...... 48

2. Methods ...... 50

2.1. Sheep ...... 50

2.1.1. Induction of heart failure in sheep ...... 50

2.1.2. Recovery Sheep ...... 50

2.1.3. Isolation of sheep ventricular and atrial myocytes ...... 51

2.2. Neonatal rats ...... 51

2.2.1. Isolation of neonatal rat ventricular myocytes ...... 51

2.3. Cell culture ...... 53

2.3.1. Neonatal rat ventricular myocyte culture ...... 53

2.3.2. Induced pluripotent stem cell (iCells) culture ...... 53

2.4. Generation of Plasmids and Transfection ...... 54

2.4.1. Plasmid DNA expansion and purification ...... 56

2.4.2. Generation of vectors with fluorescence tag ...... 56

2.4.2.1. Digestion of vectors: ...... 56

2.4.2.2. Ligation of vectors: ...... 57

2.4.2.3. Sequencing of vectors: ...... 59

2.4.2.4. Generated vector expansion ...... 61

2.4.3. Transient Transfection ...... 61

2.5. T-tubule imaging...... 62

2.5.1. Preparation of samples ...... 62

2.5.2. Confocal Microscopy ...... 62

2.5.3. Image analysis ...... 64

2.5.3.1. Distance maps ...... 64

2.5.3.2. Transverse tubule fractional area analysis ...... 66

2.5.3.3. Transverse tubule orientation analysis ...... 66

2.5.3.4. Branching analysis ...... 67

2.5.3.5. Cell width ...... 68

2.6. Measurements of intracellular calcium ...... 69

2.6.1. NRVM intracellular calcium ...... 70

2.6.1.1. Measurements of intracellular calcium concentration ...... 70

2.6.1.2. Analysis of intracellular calcium: ...... 70

2.6.2. Sheep myocyte intracellular calcium...... 72

2.6.2.1. Measurements of intracellular calcium...... 72

2.6.2.2. Analysis of intracellular calcium: ...... 74

2.7. Protein assessment ...... 76

2.7.1. Protein extraction ...... 76

2.7.2. Protein quantification ...... 77

2.7.3. SDS-PAGE ...... 77

2.7.3.1. Protein transfer ...... 78

2.7.3.2. Membrane blocking ...... 78

2.7.3.3. Detection of protein...... 79

2.7.3.4. Analysis of Western blots ...... 79

2.7.4. Immunocytochemistry ...... 83

2.7.5. Co-localisation analysis ...... 83

2.8. Statistics ...... 84

3. Results: Disordered, yet functional, atrial t-tubules on recovery from heart failure...... 85

3.1. Introduction ...... 85

3.1.1. Heart failure model ...... 86

3.1.2. Aims of the chapter ...... 88

3.2. Results ...... 89

3.2.1. Recovery of atrial t-tubules ...... 89

3.2.1.1. Atrial t-tubule density ...... 89

3.2.1.2. Atrial t-tubule structure and organisation ...... 92

3.2.1.3. Atrial t-tubule remodelling ...... 95

3.2.2. Recovered atrial t-tubules were functional ...... 98

3.2.2.1. T-tubules were the main site for calcium release ...... 98

3.2.2.2. T-tubule recovery restored synchronicity of calcium release. .... 101

3.2.2.3. Cellular distribution of calcium handling proteins ...... 102

3.2.3. Expression of t-tubule associated proteins ...... 108

3.3. Discussion ...... 110

3.3.1. Atrial t-tubules can be recovered following loss in a rapid pacing model of induced heart failure...... 110

3.3.2. Recovered atrial tubules were disorganised ...... 111

3.3.3. Recovered atrial t-tubules can trigger synchronous calcium release. 112

3.3.4. Recovery of proteins associated with t-tubule biogenesis ...... 115

3.3.5. Study limitations ...... 116

3.4. Conclusions ...... 118

4. Results: Proteins implicated in the formation and maintenance of t-tubules. 119

4.1. Introduction ...... 119

4.1.1. Regulation of t-tubules by molecular mechanisms ...... 120

4.1.2. T-tubule development ...... 120

4.1.3. Aims of the chapter ...... 121

4.2. Results ...... 122

4.2.1. Over expression of t-tubule associated proteins in NRVMs ...... 122

4.2.1.1. Expression of Amphiphysin II (BIN1) ...... 122

4.2.1.2. Expression of Telethonin (Tcap) ...... 124

4.2.1.3. Expression of Myotubularin 1 (MTM1) ...... 126

4.2.2. Co-expression of t-tubule associated proteins alters BIN1 induced t- tubule structures...... 128

4.2.2.1. Co-expression of proteins in NRVMs ...... 128

4.2.2.2. BIN1 is required for tubule formation...... 132

4.2.2.3. Fluorescent tag positioning alters tubule formation...... 133

4.3. Discussion ...... 135

4.3.1. Transfection with BIN1 led to the development of tubules in NRVMs...... 135

4.3.2. Co-expression with Tcap and MTM1 altered the expression of BIN1 induced tubules...... 137

4.3.3. Other factors regulating t-tubule recovery ...... 138

4.3.4. Study limitations ...... 139

4.4. Conclusions ...... 140

5. Results: The control of t-tubule formation by Amphiphysin II (BIN1) ...... 141

5.1. Introduction ...... 141

5.1.1. Cardiac variants of BIN1 ...... 141

5.1.2. Functional role of BIN1 ...... 144

5.1.3. Aims of the chapter ...... 145

5.2. Results ...... 146

5.2.1. Transfection with BIN1 variants 5, 8 & 9 in cardiac cells ...... 146

5.2.1.1. Expression of several variants of BIN1 led to the formation of tubules in NRVMs...... 146

5.2.1.2. Tubules developed rapidly following expression of BIN1 in NRVMs ...... 150

5.2.1.3. Expression of BIN1 in human iPSC-CMs...... 151

5.2.1.4. The fluorescent tag on BIN1 did not interfere with tubule formation in NRVMs...... 153

5.2.2. BIN1 driven tubules were functional ...... 154

5.2.2.1. Expression of BIN1 led to an increase in the amplitude of the systolic calcium transient...... 154

5.2.2.2. Transfection with BIN1 led to more synchronous calcium release...... 158

5.2.3. Cellular distribution of calcium handling proteins ...... 159

5.2.3.1. Immuno staining of z-line proteins ...... 162

5.2.4. Protein expression in transfected NRVMs ...... 164

5.2.4.1. Expression of calcium handling proteins ...... 164

5.2.4.2. Expression of t-tubule associated proteins ...... 164

5.3. Discussion ...... 167

5.3.1. Exon 11 of BIN1 was not required for the development of t-tubules in NRVMs or iPSCs...... 167

5.3.2. The effect of BIN1 driven tubules on the systolic calcium transient. 168

5.3.3. BIN1 driven tubules were highly disordered...... 172

5.3.4. Study limitations ...... 174

5.4. Conclusions ...... 176

6. General Discussion...... 177

6.1. The atrial t-tubule network recovered following loss in a sheep model of heart failure...... 177

6.2. BIN1 led to t-tubule formation in the heart, the structures which were shaped by MTM1...... 180

6.3. A role for BIN1 in t-tubule recovery ...... 182

6.4. Overall Conclusions...... 185

7. Reference List ...... 186

Appendix ...... 208

Word count: 40, 874

List of figures

Figure 1.1. Excitation contraction (EC) coupling...... 22 Figure 1.2. Transverse tubule network...... 23 Figure 1.3. T-tubules localise signaling pathways...... 26 Figure 1.4. T-tubule cellular differences ...... 27 Figure 1.5. T-tubule remodeling in heart failure...... 29 Figure 1.6. Cellular localisation of membrane associate proteins...... 35 Figure 1.7. BIN1 curves the membrane...... 38 Figure 1.8. Myotubularin (MTM1)...... 43 Figure 1.9.Junctophilin (JPH2)...... 45 Figure 1.10. Telethonin (Tcap)...... 46 Figure 2.1. Commercially available expression vectors...... 55 Figure 2.2. Example of digested vector separation...... 57 Figure 2.3. Generated expression vectors with fluorescent tags...... 60 Figure 2.4. Confocal microscope schematic...... 63 Figure 2.5. T-tubule density analysis using distance maps...... 65 Figure 2.6. T-tubule fractional area analysis...... 66 Figure 2.7. T-tubule orientation analysis...... 67 Figure 2.8. T-tubule branching analysis...... 68 Figure 2.9. ROI analysis of NRVM...... 71 Figure 2.10. Calcium transient analysis...... 72 Figure 2.11. MATLAB Calcium analysis...... 75 Figure 2.12. Example Western blots...... 81 Figure 3.1. T-tubules were restored following recovery from heart failure in the sheep atria...... 91 Figure 3.2. Atrial t-tubules were disordered following recovery from heart failure.. 94 Figure 3.3. Recovered atrial myocyte characteristics ...... 96 Figure 3.4. T-tubules were the main site for calcium release...... 100 Figure 3.5. T-tubule recovery restored synchronicity of calcium release...... 101 Figure 3.6. Cellular distribution of NCX...... 103

Figure 3.7. Cellular distribution of RyR...... 106 Figure 3.8. Recovery of t-tubules was associated with restoration of membrane associated proteins in the atria...... 109 Figure 4.1. BIN1 drives tubule formation in NRVMs...... 123 Figure 4.2. Overexpression of Tcap in NRVMs...... 125 Figure 4.3. Overexpression of MTM1 alone is not required for tubule formation in NRVMs...... 127 Figure 4.4. Co-expression of BIN, Tcap and MTM1 in NRVMs...... 131 Figure 4.5. BIN1 is required for tubule formation...... 132 Figure 4.6. BIN1 & MTM1 tubule formation with different fluorescent tag positioning...... 134 Figure 5.1. BIN1 isoforms...... 142 Figure 5.2. Expression of several variants of BIN1 in NRVMs...... 148 Figure 5.3. BIN1 driven tubule time series...... 150 Figure 5.4. Expression of BIN1 in human iPSC-CMs...... 152 Figure 5.5. BIN1 driven tubules without the mKate2 tag...... 153 Figure 5.6. BIN1 transfection led to improved calcium handling...... 157 Figure 5.7. Transfection with BIN1 led to more synchronous calcium release...... 158 Figure 5.8. Cellular distribution of calcium handling proteins...... 161 Figure 5.9. Immuno staining of z-line proteins...... 163 Figure 5.10. Expression of calcium handling proteins in NRVMs...... 165 Figure 5.11. Protein expression of t-tubule associated proteins in transfected NRVMs...... 166

List of tables

Table 1.1. Contribution of others to this Thesis ...... 19 Table 2.1. Dissociation buffer, pH 7.35 with NaOH ...... 52 Table 2.2. Pre-plating medium ...... 52 Table 2.3. Maintenance media ...... 53 Table 2.4. Commercially available expression vectors purchased from Origene ...... 54 Table 2.5. Generation of vectors with fluorescence tag...... 58 Table 2.6. Primer design...... 59 Table 2.7.Normal Tyrode solution, pH 7.34 with NaOH ...... 70 Table 2.8. Pipette solution, pH 7.2 with KOH ...... 73 Table 2.9. DIDS solution, pH 7.34 with NaOH ...... 73 Table 2.10. RIPA Buffer ...... 76 Table 2.11. Protease and phosphatase inhibitors...... 77 Table 2.12. TBS-T buffer...... 78 Table 2.13. Summary of Western blotting conditions ...... 82 Table 3.1. Mean data summarising echocardiographic parameters in sheep...... 88 Table 3.2. Summary of control, heart failure and recovery t-tubule parameters...... 97 Table 3.3. Summary of t-tubule and protein co-localisation parameters in sheep myocytes ...... 107 Table 5.1. Summary of BIN1 expression parameters in NRVMs...... 149 Table 5.2. Summary of control calcium handling data...... 155 Table 5.3. Calcium handling properties in BIN1 transfected NRVMs...... 156 Table 5.4. Co-localisation of tubules and calcium handling proteins in BIN1 transfected NRVMs...... 160

List of abbreviations

AC Adenylate cyclase AM Acetoxymethyl AmpR Ampicillin resistance gene AV node Atrioventricular node BAR Bin/Amphiphysin/Rvs BIN1 Amphiphysin II bp Base pairs bpm Beats per minute BRDU Bromodeoxyuridine BSA Bovine serum albumin Ca Calcium cAMP Cyclic adenosine monophosphate cBIN1 Cardiac BIN1 cDNA Complementary DNA CLAP Clathrin-associated protein CLIP-170 Cytoplasmic linker protein 170 CMV Cytomegalovirus CNM Centronuclear myopathy

CO2 Carbon dioxide Ctrl Control DM Myotonic dystrophy DMEM Dulbecco's Modified Eagle Medium DMSO Dimethyl sulfoxide DNA Deoxyribonucleic acid DTT Dithiothreitol EC Excitation contraction EDID End diastolic internal diameter EDTA Ethylenediaminetetraacetic acid empty mKate2 only ESC-CMs Embryonic stem cell derived cardiomyocytes ESID End systolic internal diameter

F/F0 Normalised fluorescence FACS Fluorescence activated cell sorting FBS Foetal bovine serum GFP Green fluorescent protein

HEK cells Human embryonic kidney cells hESC-CMs Human embryonic stem cell derived cardiomyocytes HF Heart failure hiPSC CMs Human induced pluripotent stem cell derived cardiac myocytes HL-1 Reductionist atrial myocyte HRP Horseradish peroxidise IC Internal standard i.u. International units i.v. Intravenous IDL Interactive Data Language

If Pacemaker current

Ik Fast voltage potassium channels

Ito Calcium activated chloride channels JPH2 Junctophilin 2 KanR Kanamycin resistance gene Kd Low dissociation constant KO Knock out LB Lysogeny broth LDS Lithium dodecyl sulfate LED Light emitting diode LTCC L-type calcium channel MBD MYC-binding domain mBFP Mutated blue fluorescent protein MCU Mitochondrial calcium uniport mFP Mutated fluorescent protein mGFP Mutated green fluorescent protein minK Minimal potassium channel subunit miR-24 microRNA 24 mKate2 Mutated red fluorescent protein mRNA Messenger RNA MTM1 Myotubularin MWM Molecular weight marker N-BAR N- helix within the BAR domain N-helix N-terminal amphipathic helix NCX Sodium-calcium exchanger NeoR Neomycin resistance gene NRVM Neonatal rat ventricular myocyte NT Non-transfected Ori Origin of replication

PBS Phosphate buffered saline pCMV6-AC C-terminal tagged vector pCMV6-AN N-terminal tagged vector PDE5 Phosphodiesterase type 5 PI Phosphoinositide PI3K Phosphoinositide 3-kinase PI3P Phosphatidylinositol monophosphate 3 PI4P Phosphatidylinositol monophosphate 4 PI5P Phosphatidylinositol monophosphate5

PIP2 Phosphatidylinositol (4,5)-bisphosphate

PIP3 Phosphatidylinositol triphosphate PKA Protein kinase A PLN Phospholamban PMCA Plasma membrane calcium-ATPase PMSF Phenylmethanesulphonylfluoride PMT Photomultiplier tube Poly (A) Poly adenylation PSF Point spread function Rec Recovery RIPA Radio-Immuno Precipitation Assay RNA Ribonucleic acid ROI Region of interests rpm Revolutions per minute RyR Ryanodine receptor SEM Standard error of the mean SERCA Sarcoendoplasmic reticulum calcium ATPase SH3 Src-homology 3 SOC Super optimal broth SR Sarcoplasmic reticulum T-tubule Transverse tubule TAE Tris acetate EDTA Tcap Telethonin TF50 50% rise time tGFP Turbo green fluorescent protein v Variant v/v % in solution WGA Wheat germ agglutinin α-actinin Alpha-actinin β-AR β-adrenergic receptors

Abstract

Transverse (t)-tubules are invaginations of the cell membrane vital for the synchronous rise of systolic calcium. In heart failure, t-tubules are lost leading to dys-synchronous calcium release. Nothing is known however, about the mechanisms that control atrial t-tubule formation or whether t-tubules can recover. Thus, this study aimed to investigate atrial t-tubule recovery following loss in heart failure, with emphasis on t-tubule density, calcium handling and proteins that contribute to t- tubule recovery. This study also aimed to gain understanding of how t-tubules form in the heart.

Heart failure, and subsequent t-tubule loss, was induced in sheep by rapid ventricular pacing for ~ 36 days. Following 5 weeks of recovery, prompted by cessation of pacing, cardiac function, cellular hypertrophy and atrial tubule density all recovered. The organization of recovered tubules was altered however, with recovered tubules being predominately longitudinally orientated. Despite changes in orientation, recovered tubules were functional, in that, calcium was initially released along the tubules, followed by propagation to the rest of the cell. Recovery of atrial t-tubules was associated with increased expression of BIN1, Tcap and MTM1. Expressing these proteins in neonatal rat ventricular myocytes (NRVMs) demonstrated that BIN1 was sufficient to drive tubule formation, the structure of which was altered by coexpression with MTM1 and Tcap. Furthermore, NRVMs expressing variants 5, 8 and 9 of BIN1 exhibited increased systolic calcium transients and more synchronised calcium release. This is in contrast to skeletal muscle where tubule formation was dependent on the muscle specific variant 8 of BIN1.

This study has shown that functional atrial t-tubules are restored following recovery from heart failure. The same proteins that were associated with t-tubule recovery were responsible for the formation of new tubule structures in the developing heart. This research identifies several targets, most notably BIN1, to facilitate t-tubule restoration and subsequent changes in calcium handling following loss in disease.

Declaration

No portion of the work referred to in the thesis has been submitted in support of an application for another degree or qualification of this or any other university or other institute of learning.

Jessica Louise Caldwell

Division of Cardiovascular Sciences

School of Medical Sciences

i. The author of this thesis (including any appendices and/or schedules to this thesis) owns certain copyright or related rights in it (the “Copyright”) and s/he has given The University of Manchester certain rights to use such Copyright, including for administrative purposes. ii. Copies of this thesis, either in full or in extracts and whether in hard or electronic copy, may be made only in accordance with the Copyright, Designs and Patents Act 1988 (as amended) and regulations issued under it or, where appropriate, in accordance with licensing agreements which the University has from time to time. This page must form part of any such copies made. iii. The ownership of certain Copyright, patents, designs, trademarks and other intellectual property (the “Intellectual Property”) and any reproductions of copyright works in the thesis, for example graphs and tables (“Reproductions”), which may be described in this thesis, may not be owned by the author and may be owned by third parties. Such Intellectual Property and Reproductions cannot and must not be made available for use without the prior written permission of the owner(s) of the relevant Intellectual Property and/or Reproductions. iv. Further information on the conditions under which disclosure, publication and commercialisation of this thesis, the Copyright and any Intellectual Property and/or Reproductions described in it may take place is available in the University IP Policy (see http://documents.manchester.ac.uk/DocuInfo.aspx?DocID=2442 0), in any relevant Thesis restriction declarations deposited in the University Library, The University Library’s regulations (see http://www.library.manchester.ac.uk/about/regulations/) and in The University’s policy on Presentation of Theses.

Acknowledgments

As always, this work could not have been completed without the generous funding from the British Heart Foundation and Division of Cardiovascular Sciences.

I have been lucky enough to be a member of the Cardiac Physiology lab for the last 10 years and in that time have gotten to meet many wonderful people that have contributed to my journey. To mention every single person who has helped and inspired me in the last decade would take us into a second volume of this thesis. Therefore, I firstly want to collectively give a huge shout out to the people, who unknowingly have made this experience that little bit more manageable, from a gossip in the CTF kitchen, to the students and colleagues who have kept me on my toes and of course to my conference buddies who have shown me that when you work hard you have to party even harder (still the poster jousting champion).

I can’t thank my supervisors Andy and David enough for giving me a chance all those years ago and believing in me. Their and Kat’s constant support and advice has made me want to continue my career in academia. Thank you to Andy for always challenging me, to Kat for her infectious energy and for discussing all things t- tubules and special thanks to David for his unwavering guidance and backing. I would also like to thank my advisor Jaqui for keeping us all in order and my line manager Margot for when I needed an escape.

To my past and present lab members, there are too many of you to mention individually, this experience would not have been half as enjoyable if I wasn’t able to share it with you. It goes without saying that you are a FABulous group of scientists and I couldn’t have done this without you. Special thanks to Becky, Charlotte, Emma, Jessabelle, Margaux, Mark Richards, Mike and Sarah who have taught me everything science and of course to everyone on team sheep. THANK YOU all for the memories, the copious laughs and of course the kebabs.

I would also like to thank my family and friends for their love, encouragement and support and who have embraced “PhD Jess” despite all the chaos.

Saving the best till last, to the man who introduced me to shower beer, thank you for The Martian, the mash, the inspiration and for making me a better scientist. You probably know just as much as I do about t-tubules now! Framförallt, tack för all kärlek, skratt och att du alltid trodde på mig. I can’t wait for us to embark on our next chapter together across the pond.

Finally, I dedicate this thesis to my family, who have been put second to it more times than I dare to admit, thank you for your constant understanding and unconditional love.

Contributions

Name Contribution Section

Dr, Elizabeth Bode, Dr Jessica In vivo sheep work, including Clarke, Dr Margaux Horn , Dr animal handling, surgery, Methods 2.1.1, Charlie Pearman , Dr Emma electrocardiography and Table 3.1 Radcliffe, Prof Andrew Trafford, echocardiography. Dr Amy Watkins

Dr Jessica Clarke, Dr Katharine Dibb, Dr Charlie Pearman, Dr Sheep atrial cell isolation Methods 2.1.3 Mark Richards, Dr Charlotte Smith, Dr Daniel Wrigley

Methods 2.6.2, Prof Andrew Trafford, Dr Patch clamp of sheep atrial Data to generate Katharine Dibb, Dr Jessica Clarke myocytes Figure 3.4

Jessica Caldwell, Dr Rebecca BIN1 primer design Methods 2.4.2 Taylor

Jessica Caldwell, Dr Rebecca NRVM isolation and culture Methods 2.3 Taylor

Table 1.1. Contribution of others to this Thesis

1. General Introduction

1.1. Introduction

Contraction of the heart depends on the synchronous rise of systolic calcium in a process known as excitation contraction (EC) coupling. In the healthy heart the existence of membrane invaginations in individual myocytes, known as transverse (t) tubules, greatly contribute to the synchronous rise of systolic calcium. T-tubules house many of the key proteins and channels involved in EC coupling. Thus, it is the presence of a well-organised t-tubule system that allows for calcium to be released synchronously throughout the myocyte activating contraction1, 2.

Loss of t-tubules is commonly associated with many cardiac diseases, including heart failure, and results in the rise in calcium becoming dys-synchronous2, leading to contractile dysfunction. The control of t-tubules in the heart is not fully understood but could provide key insight into new approaches aimed at t-tubule restoration and normalisation of calcium handling in heart failure. To understand heart failure progression however, it is first essential to appreciate how the heart works under healthy conditions.

1.2. Excitation contraction coupling

Myocyte contraction occurs when the arrival of an action potential at the cell membrane results in membrane depolarisation. This stimulates the opening of L-type calcium channels (LTCC), which are predominately found on the t-tubule membrane3, allowing calcium to enter the cell and generating the L-type calcium current. LTCC on the t-tubules are close, ~12nm4, to the sarcoplasmic reticulum (SR) calcium release channels, ryanodine receptors (RyRs) forming dyads5. The arrangement of LTCC and RyRs clusters are known as couplons4. The close proximity of these channels enables entering calcium to quickly bind to the RyRs, thus increasing the open probability6. As a result, a greater amount of calcium is released from the intracellular calcium store, the SR, through clusters of RyRs as a

calcium spark. Released calcium activates adjacent RyRs in the cluster resulting in a systolic calcium transient7. For synchronous myocyte contraction to occur, all couplons in the cell must be simultaneously activated. Importantly, these calcium release events are spatially isolated, thus calcium release from one RyR cluster does not trigger neighbouring clusters8. This process is known as calcium induced calcium release9. Elevated levels of free intracellular calcium bind to troponin C on the myofilaments, causing a conformational change in the acitin-myosin complex and resulting in contraction of the myocyte. Contractile force of the myocyte is relative to the amplitude of the systolic calcium transient10.

To complete the EC coupling cycle and for relaxation of the myocyte to occur, calcium must be removed from the cytosol. During steady state, a balance exists whereby calcium influx into the cell must equal efflux. A calcium dependent negative feedback scheme exists to maintain beat-to-beat calcium flux balance to prevent SR overload which could lead to diastolic calcium release11. The main route of calcium removal is via the sarcoplasmic reticulum calcium ATPase (SERCA) which pumps calcium back into the SR. SERCA function and activity is predominantly regulated by the protein Phospholamban (PLN)12, which by inhibiting SERCA can reduce calcium reuptake into the SR. SERCA and thus PLN are expressed close to the t-tubule13, indicating that calcium efflux pathways are closely associated with calcium influx pathways. Calcium extrusion from the cell is also in part regulated by the sodium calcium exchange (NCX), which is also predominately located on t-tubules to allow for rapid calcium extrusion14. In efflux mode, increased intracellular calcium activates the exchange of one calcium ion for three sodium ions. This exchange is driven by the energy that is stored in the electrochemical gradient of sodium. The remaining calcium is removed via the plasma membrane calcium-ATPase (PMCA) and mitochondrial calcium uniport, although these systems remove a much smaller fraction of cellular calcium15, 16. Calcium is then dissociated from the myofilaments, resulting in relaxation of the myocyte. The process of EC coupling is shown in the schematic in Figure 1.1 and has been extensively reviewed in8, 17(Appendix).

Figure 1.1. Excitation contraction (EC) coupling. Schematic diagram showing calcium cycling in a ventricular myocyte. Calcium (Ca) enters the cell via the LTCC. Ca influx triggers sarcoplasmic reticulum (SR) calcium release by the RyR. Free Ca binds to the myofilaments for contraction of the cell. Ca is removed via SERCA, NCX, PMCA and MCU. Note how key EC coupling proteins are located on the t-tubule membrane. β-AR = β-adrenergic receptors. Adapted from 17 (Appendix).

It is evident that cardiac contraction is intricately linked to the organisation of the t- tubule network. Many of the proteins required for EC coupling are expressed on or adjacent to the t-tubules. Thus, by enabling close apposition of calcium handling proteins, t-tubules create specialist sites for regulation of EC coupling.

1.3. The transverse tubule network

In the ventricle, calcium rises synchronously and rapidly throughout the cell due to the presence of a well organised t-tubule system. T-tubules are sarcomeric extensions of the surface sarcolemma membrane approximately 100-300nm in diameter that penetrate deep into the centre of cardiac myocytes18 (Figure 1.2). T-tubules are found in the ventricles of all mammals studied and in the atria of larger mammals.

Figure 1.2. Transverse tubule network. 3D schematic of the t-tubule network in a ventricular myocyte. T-tubules are located adjacent to the z-line of the sarcomere. SR forms a continuous membrane surrounding the t-tubule which facilitates EC coupling.

1.3.1. Transverse tubules and EC coupling.

T-tubules facilitate cardiac EC coupling by enabling the close apposition of LTCC and RyRs throughout the cell, which form junctions known as dyads. Dyad density and organisation plays an important role in facilitating the synchronous and rapid rise of systolic calcium. In the ventricle, LTCC and, by association, dyads are found at higher concentrations on the t-tubule membrane compared with the surface sarcolemma5, 19. As a result calcium entry has been shown to be more concentrated at t-tubules20, 21, which coincides with enhanced expression of LTCC on t-tubules and their co-localisation with RyRs5, 22. Therefore, by enabling all points in the cell to be close to membrane, t-tubules ensure calcium is released simultaneously across the cell, allowing rapid and synchronous contraction.

The role of t-tubules in facilitating synchronous calcium release has been further demonstrated in cells that have t-tubules chemically detached from the surface sarcolemma, by treatment with formamide3, 21. Loss of t-tubules in these cells was associated with decreased LTCC and RyR coupling, reduced L-type calcium current and delayed and decreased calcium transients.

Alongside calcium entry, t-tubules are also important for calcium removal. SERCA has been shown to be expressed in close proximity to t-tubules13 and both NCX and PMCA are expressed on t-tubules5, 14, 23. Furthermore, calcium extrusion via the NCX pathway occurs predominately in the t-tubule14, and in the case of PMCA, exclusively in the t-tubules24. This data highlights the important role t-tubules play in EC coupling by enabling close coupling of calcium entry, calcium release and calcium removal sites.

1.3.2. Transverse tubules: Signalling pathways

β-adrenergic signalling, in response to sympathetic stimulation, increases cardiac output by the regulation of several proteins known to be involved in EC coupling. G- protein-coupled β-adrenergic receptors (β-ARs) regulate cardiac function by stimulation of adenylyl cyclase (AC) which, in turn activates cyclic adenosine

monophosphate (cAMP) dependent protein kinase A (PKA) (Figure 1.3). Phosphorylation by PKA of the proteins that mediate EC coupling, such as LTCC, RyRs and PLN8, 25, 26, enhances the inotropic and lusitropic response of the heart27.

Cardiac myocytes express both β1-AR and β2-ARs, but whilst β1-ARs are expressed equally on the surface sarcolemma and t-tubules, β2-ARs are predominately found on t-tubules28. Consequently, many of the proteins important in the β-adrenergic signalling cascade are localised in the t-tubule membrane (Figure 1.1 and Figure 1.3), in close proximity to the proteins they phosphorylate28-30. The exclusive localisation of β2-AR-cAMP signalling to the t-tubule allows for compartmentalised control and selective phosphorylation of calcium handling proteins31. This compartmentalisation has been demonstrated in work from Davare et al32, who showed that a complex between the proteins of the β2-AR pathway and the LTCC existed on the t-tubule membrane, which permitted local regulation of the LTCC. Further support of the view that t-tubules are specialist sites for β-adrenergic control, is provided in a study that showed β-adrenergic stimulation resulted in a greater increase in calcium current in cells with t-tubules than those that have had their t- tubules chemically removed33. Alongside protein localisation, another mechanism for local control of β-AR signalling at the t-tubules is through recruitment of key ion channels to the t-tubule by β-AR signalling. It has recently been shown that β-AR activation induced PKA dependent increase of LTCC in the t-tubules, enhancing calcium influx34 and led to recruitment of RyR to dyads35. This data demonstrates, in addition to assisting in EC coupling, t-tubules are the main site for regulating the β- adrenergic signalling response in the heart.

Figure 1.3. T-tubules localise signaling pathways. Schematic diagram emphasising how t-tubules enable compartmentalised β-adrenergic control of calcium handling proteins. P = phosphorylation.

1.3.3. Transverse tubules: Cellular differences

In ventricular myocytes t-tubules have a sarcomeric distribution, located approximately every 2µm at each z-line36 (Figure 1.2). Ventricular t-tubules form a tightly linked network consisting mostly of transverse elements with longitudinal connections (Figure 1.4 A). This system ensures that all points inside the cell are close to the surface or t-tubule membrane, thus allowing a rapid and synchronous rise of intracellular calcium. In the mammalian ventricle, cell dimensions are uniform across species36; it is thought however that t-tubule density varies and that this variability is dependent on heart rate37. Species with a higher heart rate require calcium cycling to occur more quickly and therefore have a denser t-tubule network.

A well-developed t-tubule network is not found in all cardiac myocytes. T-tubules are absent in neonatal cells38 and atrial myocytes, from some small mammals, have been shown to lack a well-developed t-tubule network39. The absence of a t-tubule

system in these cells results in dys-synchronous calcium release. Calcium rises at the cell periphery and propagates to the cell centre39. In neonates, the appearance of t- tubules during postnatal development coincides with the development of EC coupling and more synchronous calcium release40, 41. These data highlight the importance of t-tubules in maintaining synchronous calcium release. Compared with the ventricle, atrial myocytes from small mammals are small in size (Figure 1.4 B), thus it is thought that t-tubules are not required as calcium is able to quickly propagate to the cell centre42. This is supported by the discovery that atrial cells from some larger mammals, including human, have well developed t-tubules and these mammals have wider cells42, 43 (Figure 1.4C). It has since been discovered however, that alongside cell size differences, regional differences in atrial t-tubule density also exist. Frisk et al44, showed a higher density of atrial t-tubules in the epicardium compared to the endocardium. The authors of this study proposed that regional differences exist to promote synchronised contraction across the heart as the arrival of the action potential is delayed in the epicardium in comparison to the endocardium44. As with the ventricle, atrial t-tubules have been shown to contribute to the synchronous rise of systolic calcium42, 44. Whereby, early work from our lab demonstrated that, calcium rise sites correlated with the location of t-tubules in the sheep atria. In contrast to the ventricle, in atrial myocytes that possess t-tubules, LTCC occur at similar frequency in both t-tubules and surface sarcolemma45. However, distinct subpopulations of LTCC exist on atrial t-tubules giving rise to higher amplitude and open probability in the t-tubules45.

Figure 1.4. T-tubule cellular differences 2D confocal imaging of (A) rat ventricular, (B) rat atrial and (C) sheep atrial myocytes stained with di-4-ANEPPS to image t-tubules and surface sarcolemma. Scale bar denotes 10µm. Adapted from46.

1.4. Transverse tubule alterations in disease

T-tubule alterations have commonly been associated with many cardiac diseases, including heart failure, myocardial infarction, atrial fibrillation and pressure overload47-51. The consequence of t-tubule remodelling in these diseases are calcium handling alterations which can be associated with a decrease in contractile function52 and diminished cardiac output.

1.4.1. Heart failure

Heart failure, characterised by the inability of the heart to meet the metabolic demands of the body, remains one of the leading causes of morbidity and mortality53, 54. Remodelling of the heart occurs in response to extracellular stress and stimulus. The purpose of cardiac remodelling is to maintain cardiac output, sometimes however, these compensatory alterations prove to be maladaptive and lead to diseases such as heart failure. Reduced contractile function of the heart in both the muscle and single myocyte is commonly associated with heart failure55. Calcium handling abnormalities play a major role in contributing to the phenotype of the failing heart. Evidence has shown, in failing myocytes, reduced contractility is a consequence of slower to rise and decreased amplitude of the calcium transient56-58. Factors that are likely to contribute to these changes are a reduction in calcium influx via LTCC, alterations to SR calcium release and reduced SR content, for a detailed review of each of these factors, see our recent study by Eisner et al59. Changes, both structural and functional, to major contractile proteins and signalling pathways are likely to be a major cause of impaired calcium handling. As many of these proteins are housed on or adjacent to t-tubules, calcium handling alterations in heart failure may result from t-tubule remodelling.

1.4.2. Transverse tubule remodelling in heart failure

In the failing heart in particular, loss and disorder of cardiac t-tubules has been demonstrated in both human60-62 and animal models46, 50, 51, 63, 64 in both ventricular

and, where applicable, atrial myocytes42, 46, 48, 50, 63, 65-67 (Figure 1.5). Numerous studies have also shown that t-tubule remodelling in heart failure can be associated with delayed and dys-synchronous calcium release42, 48, 50, 65, 67. A link between t- tubule loss and altered calcium handling in disease, has been demonstrated in studies that showed areas of altered calcium release co-localised with areas of the cell absent of t-tubules42, 48. This is consistent with studies that have shown altered calcium homeostasis, in heart failure, is associated with a decrease in the co-localisation of the LTCC and RyRs50, 65. Although less is known about t-tubule remodelling in human heart failure, it has also been shown that t-tubules became disorganised and their density was reduced60, 61, 68. Moreover, in the failing human heart, akin to the animal studies, LTCC and RyRs became uncoupled, potentially resulting in impaired calcium homeostasis68, 69.

Figure 1.5. T-tubule remodeling in heart failure. Representative 2D confocal images of sheep ventricular (top row) and atrial myocytes (bottom row) from control (left) and heart failure (right) animals stained with di-4-ANEPPS to image t-tubules; Scale bar denotes 10µm. Adapted from46.

1.4.3. Transverse tubule remodelling in atrial fibrillation

In the healthy heart, atrial contraction contributes approximately 20% to cardiac output70. During the development of diseases such as heart failure however, atrial contraction is reduced and thus further exacerbates the reduction of cardiac output associated with these diseases71. In failing atrial myocytes it has been shown that

virtually all t-tubules were lost42, 46 (Figure 1.4). Additionally, remodelling of atrial myocytes in response to disease is a major risk factor for the development of atrial arrhythmias, the most common being atrial fibrillation. Atrial fibrillation is often associated with stroke, congestive heart failure and mortality100, 101. Despite this, the mechanisms contributing to the initiation and maintenance of atrial fibrillation are poorly understood.

Atrial fibrillation has been characterised by both reduced contractility and reduced systolic calcium transients102, 103. Patients with prolonged atrial fibrillation showed 75% reduction in contractile force of atria tissue strips104. Whilst the likely causes of atrial fibrillation are ion channel remodelling and afterdepolarisation triggered activity72, calcium handling abnormalities could be attributed to t-tubule remodelling. Reduced L-type calcium current has previously been associated with atrial fibrillation49, which may result from alterations to the t-tubule network. Both t- tubule loss and decreased LTCC density were observed in a sheep model of persistent atrial fibrillation, which was associated with dys-synchronous calcium release49. Additionally, in a dog model of atrial tachycardia, t-tubule density was also reduced56. These studies demonstrate that the reduced systolic calcium release associated with atrial fibrillation could be attributed to deficient coupling between LTCC and RyR2, a result of t-tubule loss.

1.4.4. Transverse tubule remodelling: altered calcium handling.

Disruption of t-tubules during cardiac disease has major implications for the proteins and channels that are localised on and are adjacent to the t-tubule membrane. Altered calcium handling, as observed in heart failure, is likely in part to be due to disruption of these proteins. As already discussed, one of the major consequences of t-tubule remodelling in heart failure is uncoupling of the LTCC and RyRs68. This structural alteration decreases the ability of entering calcium, via the LTCC, to trigger calcium release from the SR61. Thus it is not surprising that impaired calcium transients in the failing heart have been attributed to dyad uncoupling48, 65, 73, 74. Increased distance between the LTCC and RyRs could also lead to dys-synchronous calcium release65.

As observed in cells that lack t-tubules, where calcium was initially released at the cell periphery which led to less synchronous calcium release39.

Another consideration of t-tubule loss is the location and density of the channels found on the membrane. When compared to the surface sarcolemma, several studies have shown that LTCC and therefore current, were more concentrated on the t-tubule membrane3, 5, 21, 75. Therefore, it is not surprising that in some, but not all, models of heart failure L-type calcium current was decreased58, 76, resulting in a decrease in the trigger of calcium. Furthermore, decreases in L-type calcium current have been observed in cells that have been experimentally detubulated, by treatment with formamide3, 21. This data suggests that loss of t-tubules directly affects calcium current. Since inhibiting L-type calcium influx reduced the systolic calcium transient77, loss of t-tubules is likely to contribute to the decreased calcium transient amplitude observed in heart failure.

It is conceivable that t-tubule disruption in heart failure will also alter β-adrenergic signalling. As discussed, in healthy cardiac myocytes, β2-AR are exclusively localised to the t-tubules whilst β1-AR are found throughout the cell. In a rat model of heart failure however, β2-AR were redistributed from the t-tubules to the surface sarcolemma, which led to altered signalling28. Impaired responsiveness to β- adrenergic signalling is a hallmark of the failing heart and it is possible that reductions in t-tubule density contribute to the impairment of this signalling cascade in heart failure. Reorganisation of these channels in heart failure could result in changes to the inotropic response to catecholamine and/or predispose the heart to calcium dependent arrhythmias.

Whilst the above data demonstrates that t-tubule dysfunction is largely associated with heart failure, it is still largely unknown if t-tubule remodelling is a cause or consequence of the development of this disease. Data in support of the former shows that during pathological hypertrophy, t-tubule density was reduced before heart failure developed50. Moreover, during the progression from hypertrophy to heart failure, further loss of t-tubules occurred along with reduced myocyte contractility50. Even without the influence of a cardiac insult, cells that had t-tubules experimentally

removed, showed similar calcium handling alterations those that have been observed in the failing heart78. It is therefore probable that t-tubule loss is, at least, in part responsible for the impaired contractility, either due to spatial disorganisation or impaired calcium entry, observed in cardiac disease.

Which mechanism trigger t-tubule remodelling however, is still largely unknown. A common feature of the development of heart failure is pathological hypertrophy that leads to maladaptive remodelling, such as t-tubule loss. Still, physiological hypertrophy is not linked with t-tubule loss. In rats where myocyte volume increased, as a result of exercise, relative t-tubule density was preserved79. This data suggests that t-tubule loss only occurs as a result of pathological hypertrophy. Thus, the mechanisms that activate t-tubule loss during heart failure are likely to be the result of signalling pathways that mediate pathological remodelling. Therapeutics that target these pathways may be imperative to reverse t-tubule remodelling and ultimately reduce the progression of heart failure.

1.4.5. Reversal of transverse tubule remodelling

It is evident that alterations to calcium handling, a hallmark of heart failure, can be partly attributed to t-tubule loss. It is therefore unsurprising, that more recently, the focus of several research groups has been to determine if t-tubule loss can be prevented or reversed, thus providing a novel approach for the treatment of cardiac disease. Several experimental approaches aimed at restoring cardiac function, have resulted in t-tubule recovery or prevention of t-tubule loss in ventricular myocytes. These interventions include gene therapy80, drug treatment81-83 and mechanical unloading of the heart84.

Stolen et al85 first demonstrated reversal of t-tubule remodelling in mice with diabetic cardiomyopathy that underwent exercise training. Alongside restored t- tubule density, contractile function and calcium handling was also restored. It has since been shown that exercise training also resulted in partial t-tubule recovery in a rat model of heart failure79. More clinically relevant studies have shown that t-tubule loss was prevented in animal models of heart failure, using treatments with

therapeutics that are used clinically, such as β-blockers81 and PDE5 inhibitors Sildenafil82 and Tadalafil83. PDE5 inhibition has also been demonstrated to result in cardio-protective effects in patients with left ventricular hypertrophy86 and with systolic heart failure87. In the above mentioned studies however, therapeutics were given either before disease onset or alongside other clinical interventions, making it difficult to determine if these treatments would be useful for established disease. Importantly, we have recently shown, that PDE5 inhibition with Tadalafil, after the onset of heart failure in sheep, improved contractile function, reversed t-tubule loss and restored calcium transient amplitude and catecholamine responsiveness83 (Appendix). In this study, we proposed that PDE5 inhibition was cardio-protective through negative inotropic effects. By reducing myofilament calcium sensitivity, blunting β-adrenergic signaling, and reducing L-type calcium current; PDE5 inhibition reduced load on the heart. T-tubule recovery has also been noted following resynchronization therapy in a canine model of heart failure88 and mechanical unloading of failing rat hearts84. Both studies resulted in improved t-tubule structure and normalised calcium signalling. Similarly, treatment of heart failure using SERCA2a gene therapy showed, alongside normalisation of β-adrenergic signalling and of the systolic calcium transient, recovery of t-tubules80. In this study, t-tubule recovery was associated with restoration of β-adrenergic receptors on the t-tubule membrane and recovery of proteins thought to play roles in t-tubule biogenesis (discussed later).

The above studies demonstrate that t-tubule maintenance and restoration plays a key role in the recovery from heart failure. Another common feature of these interventions is that t-tubule recovery appears to result from a (direct or indirect) reduction of cardiac load. It is therefore not only imperative to identify the mechanisms that drive t-tubule formation, but also those that mediate load sensitivity and maintain the t-tubule system. In doing so, this knowledge may be used for future treatments aimed at reducing the incidence of cardiac disease.

1.5. Biogenesis of transverse tubules

As discussed, t-tubule structure and function varies between species and cardiac chambers, and is altered during disease. This variability has considerable implications for calcium handling. Importantly however, it has been shown that maladaptive t-tubule alterations can be reversed following recovery from heart failure, leading to normalised calcium release. It is therefore vital that we gain a better understanding of the mechanisms that control and regulate t-tubules in the heart as it is likely that these factors may provide new therapeutics for the treatment of heart failure. Given the complexity of this system, it is probable that the control of t-tubules is regulated by several mechanisms. A considerable amount of evidence highlights a role for membrane scaffolding proteins that are associated with the z- lines in t-tubule control. Some of these proteins, which will be discussed in more detail, include; Amphiphysin II (BIN1), Myotubularin (MTM1), Junctophilin 2 (JPH2) and Telethonin (Titin [T]cap). It is also important to consider factors that regulate these proteins and the membranes that house them, such as micro-RNAs, microtubules and signalling molecules to fully elucidate control of this system. The location of these molecules and how they interact with the t-tubules are represented in Figure 1.6.

Figure 1.6. Cellular localisation of membrane associate proteins. Schematic diagram showing t-tubule regulatory proteins in the myocyte. BIN1 binds to the cell membrane causing tubular invaginations. Interactions between BIN1 and CLIP-170 stabilise microtubules enabling delivery of LTCCs to the t-tubule via the microtubules. JPH2 forms junctions between the SR and t-tubules holding the t-tubule in place. Tcap anchors the t-tubule to the myofilaments. MTM1 modulates phosphatidylinositols (PI) in the cell membrane promoting membrane regulation.

1.5.1. Phosphatidylinositols (PI)

T-tubule membrane structure may in part be regulated by membrane lipids. The cell membrane, including the t-tubule, of a myocyte consists of negatively charged lipids making up a phospholipid bilayer. Membrane turnover is dependent on continuous regulation and recycling of these phospholipids and the channels and proteins that contribute to the membrane. This process is modulated by phosphatidylinositol (PI) signalling molecules. PI can be phosphorylated to form multiple combinations of second messenger phosphoinositides including; PI monophosphates (PI3P), (PI4P), 89 (PI5P); PI bisphosphates (PIP2) and PI trisphosphates (PIP3) (reviewed in ). Phosphoinositides are signaling lipids found in the cell membrane that play important roles in cell signalling, membrane trafficking and are anchoring molecules. 89- In particular, PI3P and PIP2 regulate the recruitment of proteins to the membrane 91, and have been shown to interact with t-tubule associated proteins BIN1 and JPH2 92 through electrostatic interactions . It is not surprising therefore, that PIP2 lipids have been shown to modulate membrane structure by enhancing the membrane binding and penetration capability of N-BAR domain proteins such as BIN190. Likewise, depletion of PIP2 from the plasma membrane of HEK293 cells stopped membrane 92 90 bending . PIP2 is preferentially concentrated on t-tubule membranes where it is thought to directly regulate t-tubule ion channels and regulate calcium signalling93-95.

Thus, alongside membrane structure modulation, PIP2 also has important implications for myocyte contraction.

PI are regulated by several kinases, the most common, phosphoinositide 3-kinase (PI3K) has been shown to regulate cardiac function in numerous ways. Alongside phosphorylation of membrane lipids, PI3K sub-units have also been demonstrated to play roles in regulating cardiac myocyte growth and targeting proteins, such as LTCC, to the t-tubule membrane96. More recently, evidence has suggested that some PI3Ks may play a role in maintaining the t-tubule network, possibly through interactions with BIN1 and JPH2. Wu et al97, demonstrated that PI3K knockout (KO) mice, despite having normal t-tubule networks at birth, developed t-tubule alterations and reduced myocyte contraction. It was also shown in this model that

P13K KO led to JPH2 mis-localisation. This may have resulted from deletion of P13K preventing membrane lipid phosphorylation of phosphoinositides, thus reducing the binding capacity of JPH2 to the t-tubule. Interestingly however, in mice deficient in the lipid phosphatase MTM1, a signalling lipid known to modulate phosphoinositides, PI3K inhibition was beneficial and prolonged survival98. This data suggests that a fine balance exists between the phosphatases and kinases that are known to regulate and maintain lipid membranes and more work is required to determine the roles these play in the regulation of t-tubules. Despite this, the data does indicate that, alongside several other mechanisms, signalling pathways such as P13K play a role in t-tubule maintenance, through interactions with proteins of the t- tubule membrane. Thus, alongside post-transcriptional regulation of the proteins that are thought to control t-tubule biogenesis, it is also possible that cell signalling pathways play a role in controlling this system.

1.5.2. Amphiphysin II (BIN1)

Of the known t-tubule associated proteins, there is considerable evidence linking BIN1 (aka Amphiphysin II), a Bin/Amphiphysin/Rvs (BAR) domain protein, to t- tubule formation. BAR domains are ubiquitously expressed scaffold proteins that bind to the cell membrane and play roles in endocytosis, sensing membrane curvature and actin filament regulation99, 100. It has also been shown that BIN1 is involved in the formation of t-tubules in striated muscle. BAR domain proteins are positively charged crescent shaped dimers that are able to bind to negatively charged 92 lipids, such as PIP2, that make up the cell membrane . BAR domain membrane binding can be further enhanced by the presence of the muscle specific isoform 8 of

BIN1, which contains an additional PI domain that binds PIP2 and PI4P to promote membrane tubulation92. Membrane curvature is initiated as a result of the membrane being forced to fit the positively charged curved surface of the protein, this in known as the scaffold mechanism. Additional membrane curvature can be caused by the wedge mechanism, this is where insertion of the N-terminal amphipathic helix (N- helix) of the BAR domain acts as a wedge between lipids in the membrane (Figure

1.7). PIP2 also play a role in the formation of the N-terminal amphipathic helix, such

92 that N-BAR domains, preferentially bind to PIP2 lipids in the membrane . It is thought that these mechanisms are in part responsible for t-tubule formation in the heart.

Figure 1.7. BIN1 curves the membrane. Schematic diagram of BIN1 curving the membrane. (A) The scaffold mechanism; BIN1 α-helix forms a positively charged crescent shaped dimer (i) that binds to the negatively charged membrane inducing membrane curvature (ii). (B) The wedge mechanism; a combination of BIN1 dimer and N-helix (i) inserts into the membrane acting as a wedge (ii) resulting in increased membrane curvature.

BIN1, a member of the BAR domain protein family, has multiple splice isoforms (discussed in more detail in Chapter 5); two of which have been shown to be highly expressed in muscle and localise to t-tubules90, 101. There remains some uncertainty however over which isoforms of BIN1 are expressed in cardiac muscle. Work by Hong et al and Laury-Kleintop et al. identified four BIN1 isoforms, one specific to t- tubule folding, but none of which contained the muscle specific exon in the mouse heart102, 103. In contrast to the work of Hong and Laury-Kleintop et al, we have

recently demonstrated that the muscle specific variants 4 and 8 of BIN1 were expressed within the sheep myocardium, with variant 8 being the dominantly expressed isoform (Appendix)83.

Unlike other BAR domain containing proteins, BIN1 does not play a role in endocytosis and is thought to play a more prominent role in t-tubule formation. This notion is supported by several research groups that have shown, BIN1 transfection induced t-tubule formation in non-muscle cells that do not ordinarily have t- tubules90, 104, 105. Furthermore, whilst BIN1 is not expressed in embryonic stem cell derived cardiomyocytes (ESC-CMs) that do not possess t-tubules106, recently published data demonstrated that overexpression of BIN1 variant 8 promoted the formation of tubules in both human (h)ESC-CMs107 and NRVMs (Appendix)83. Alongside, t-tubule biogenesis, BIN1 also traffics LTCC to the t-tubules and when expressed in non-muscle cells, BIN1 induced membrane was enriched with LTCC104. Together with channel trafficking, BIN1 has also been shown to regulate the movement of RyRs to t-tubule micro-domains in response to β-AR activation35. Thus by recruiting dyadic proteins to the t-tubule membrane and maintaining LTCC-RyR interactions, BIN1 may play a vital role in calcium regulation in the heart.

Mutations in the BIN1 gene can be associated with the inherited condition, centronuclear myopathy (CNM)108, which leads to myocyte disarray and cardiomyopathy109, 110. Furthermore, total KO of BIN1 in mice was lethal due to embryonic cardiomyopathy111 and conditional cardiac deletion of BIN1 resulted in heart failure103, suggesting that this protein is vital for maintaining a healthy heart. The consequences of altered BIN1 expression has been investigated in more detail by Hong et al112, who firstly demonstrated that BIN1 KO led to reduced calcium transients and contractile dysfunction in zebrafish heart. This group more recently showed a decrease in t-tubule intensity and L-type calcium current in BIN1 cardiac conditional KO mice102. Similar structural defects were observed by Razzaq et al, who found that disruption of BIN1 in Drosophila prevented t-tubule formation and led to altered dyad junctions in the flight muscles113. Following transfection with BIN1 cDNA, these structural defects were reversed. In addition, my own work using

siRNA gene silencing in cultured adult ventricular myocytes has shown that BIN1 is required for cardiac t-tubule maintenance46.

The above studies highlight the requirement of BIN1 for t-tubule maintenance in the heart and disruptions to BIN1 led to EC coupling alterations similar to those observed in heart failure. This raises the question; do alterations to BIN1 expression underlie reduced contractile function associated with heart failure? A decrease of BIN1 has been shown in association with t-tubule loss in both the failing human heart112 and in animal models of heart failure46, 80, 83. It is therefore likely that loss of BIN1 in the failing heart may lead to disruption of the t-tubule network and a reduction in L-type calcium current, as observed in some models of heart failure. Crucially, we and others have shown that BIN1 protein expression is correlated with t-tubule density during the development of heart failure and following treatment with both Tadalafil (Appendix)83 and SERCA2a gene therapy80.

Despite the proposed role BIN1 plays in t-tubule biogenesis, it is still largely unknown how this protein regulates cardiac t-tubules. It is most likely that this process is multifaceted and involves a number of protein interactions. In particular, attention should be paid to those molecules that provide anchors to the membrane and determine cellular structure, such as microtubules and membrane lipids. BIN1 regulation via membrane lipids has already been discussed in detail in section 1.5.1, thus the next section will consider a role for microtubules. Microtubules are, in part, responsible for cytoskeleton formation; they play key roles in maintaining the structure of the cell and in intracellular vesicle transportation. One possible molecular partner for BIN1 is CLIP-170. CLIP-170 is a tracking protein involved in microtubule stability and has been identified as an BIN1 interacting molecule114. Consistent with previous studies, Meunier et al114 demonstrated that over-expression of BIN1 in non-muscle cells led to the formation of tubule structures. It was also shown that these structures aligned with microtubules and moreover, both deletion of CLIP-170 and disruption of microtubules in these cells led to a decrease in tubule like structures114. This study not only suggests that interactions between BIN1 and CLIP-170 are required for tubule formation, but also that these interactions are required for microtubule stability. This was supported by Hong et al104 who showed

that microtubules were tethered to the membrane by BIN1 which allowed targeted delivery of LTCC to the t-tubule. So far, co-immunoprecipitation studies have shown that associations can be made between BIN1, CLIP-170 and LTCC, indicating that they are present in the same protein complex. Further work is needed however, to establish other proteins that interact with BIN1 and to determine what role they play in t-tubule control and how in turn they are regulated.

1.5.3. Myotubularin (MTM1)

The MTM1 gene encodes a ubiquitously expressed lipid phosphatase, MTM1, which acts on PI in the cell membrane to regulate membrane curvature and trafficking115, 116. MTM1 is predominately located at the t-tubule-SR junction in skeletal muscle and has been proposed to play roles in SR and t-tubule remodelling. MTM1 117 dephosphorylates PI3P and PIP2 , and as a result, loss of MTM1 in muscle has been shown to decrease phosphatase activity116 and led to decreased levels of PIs118.

As previously discussed, PI3P and PIP2 are signaling lipids found in the cell membrane that regulate the recruitment of proteins to the membrane89. It is thought that the mechanism by which MTM1 regulates these signaling lipids is by directly modulating PI3K activity, PI3Ks phosphorylate membrane lipids producing second messenger PIs, thus by down-regulating PI3K activity, MTM1 can control PI3P and

PIP2. In summary, MTM1 contributes to the constant recycling and regulation of myocyte membranes through PI regulation.

Alongside membrane trafficking, MTM1 has also been implicated in regulating membrane shape in skeletal muscle through interactions with PI. In both MTM1 deficient mice and patients with a MTM1 mutation, junctional SR was severely disorganised and t-tubule architecture impaired116, 119. In these studies, skeletal membrane structure was highly dependent on phosphatase activity119. Amoasii et al119 demonstrated that SR membrane and t-tubule shape was variable between active phosphatase WT MTM1 and dead phosphatase MTM1 mutant mice. Despite MTM1 mutant mice having increased levels of PI3P, these membranes were not correctly formed and t-tubule structures did not connect with extracellular space119. This data indicates that the tight regulation of PI3P by MTM1 may be a mechanism

to control membrane formation. MTM1 KO in mouse and zebrafish has also been associated with t-tubule alterations in skeletal muscle116, 120. The t-tubule alterations in these models were associated with impaired calcium handling and contractile force. This work indicates that MTM1 plays an essential role in SR and t-tubule remodelling and therefore maybe critical for calcium homeostasis.

Mutations in the MTM1 gene have been identified as being responsible for X-linked myotubular myopathy121, a type of CNM, characterised by skeletal muscle weakness. Alterations in MTM1 are thought to affect the assembly of muscle triads; leading to weakened contraction and ultimately muscle weakness. Consequently, alterations in the regulation of the membrane by MTM1 may be a cause for the development of CNM. In support of this, Dowling et al116 demonstrated that zebrafish with decreased MTM1 had EC coupling abnormalities and membrane disruption. As previously described, CNM are also associated with mutations in the BIN1 gene, where a similar phenotype of skeletal muscle weakness is observed. Thus, a potential functional link between MTM1 and BIN1 in the development of CNM has become apparent. In healthy skeletal muscle, MTM1 interacts directly with BIN1, altered interaction between this complex however, has been shown to lead to CNM, which has been associated with cardiomyopathies121, 122.

More recently, another potential functional link between MTM1 and BIN1 has come to light. Along with MTM1 being identified as a binding partner of BIN1 in skeletal muscle, MTM1 has also been found to enhance BIN1 driven membrane tubulation in muscle cells121. In further support of this finding, phosphatase inactive MTM1 mutants were unable to enhance BIN1 membrane tubulation121. This indicates that MTM1 phosphatase activity is important in the regulation of BIN1 membrane tubulation in skeletal muscle. The role of MTM1 in the heart however, remains to be determined.

MTM1

BIN1

Figure 1.8. Myotubularin (MTM1). MTM1 is a BIN1 binding partner and regulator of BIN1. By leaving BIN1 in an open confirmation, MTM1 is thought to increase the membrane tubulating ability of BIN1.

1.5.4. Junctophilin (JPH2)

JPH2 is a junctional anchor protein that plays a role in forming junctions between the SR and t-tubule membranes. These junctions are formed by JPH2 directly binding to the RyRs and to membrane lipids, such as PIP3, on the t-tubules. Junctional complexes not only help maintain efficient EC coupling but are also thought to maintain t-tubule orientation by interactions between the SR and z-lines. JPH2 spans the junctional membrane domain, thus maintaining constant spacing between the SR and t-tubules and determining the distance calcium must travel to activate the RyRs123. Alongside membrane anchoring, it is also thought that JPH2 plays a role in t-tubule development124. This was demonstrated by Ziman et al40 who showed, during early development, when t-tubules were absent, JPH2 was predominately found at the surface sarcolemma. As the t-tubule system developed, the distribution of JPH2 coincided with the developing t-tubules and JPH2 appeared adjacent to the z-lines in close proximity to the RyRs. Likewise, t-tubule development was prevented in embryonic mice that had knockdown of JPH2125, but following overexpression of JPH2, t-tubule development occurred earlier than in the wild type mice. Furthermore, in the mice that did not develop t-tubules in this study, calcium release was altered and these mice developed heart failure, indicating that loss of JPH2 can also be associated with the onset of heart failure.

Genetic KO of JPH2 in mice has been shown to be embryonically lethal due to failure of proper cardiac development126. Isolated myocytes from the heterozygous mice displayed dys-synchronous calcium transients and deficient formation of the junctional membrane complexes. Similar alterations have been shown using gene silencing techniques in various cell types where knockdown of JPH2 was associated with t-tubule alterations46, 50 and impaired calcium signalling124, 125, 127-129. These structural and functional changes were likely to result from uncoupling of the RyRs and LTCC. These data suggest that JPH2 loss results in t-tubule alterations similar to that seen in the failing heart.

Loss of JPH2 has been reported in some animal models of heart failure50, 80, 130 and human heart failure has been associated with reduced JPH2 expression and t-tubule loss 131. In other cases however, JPH2 expression did not change in the failing heart46, 83. Furthermore, t-tubule restoration, following recovery or treatment from heart failure, was not associated with recovery of JPH2 protein levels80, 83, 84, indicating that JPH2 is not required for the recovery of the t-tubule system following depletion during cardiac disease. The above data suggest that JPH2 plays a role in t- tubule development but is not essential for the recovery of cardiac t-tubules. It is therefore likely that JPH2 plays a more specialist role within the heart. This is supported by work from several groups that have shown that JPH2 was present in cells that either lack or have not developed t-tubules46, 126, 132. Furthermore, whilst JPH2 knockdown prevented the development of t-tubules, it had no effect on longitudinal tubule development124. Thus, indicating that JPH2 may play more of a role in the organisation / anchoring of the t-tubules in the heart.

Even though it has been shown that JPH2 plays an important role in regulating calcium signalling in the heart, it is largely unknown what regulates JPH2 expression. Data does suggest, that JPH2 is a direct target of microRNA 24 (miR- 24)129. By binding to JPH2 mRNA, miR-24 suppresses JPH2 expression. In agreement with this, miR-24 has been shown to be up regulated in heart failure129, 131, 133, which coincided with reduced JPH2 expression and disrupted t-tubule and SR junctions131. Moreover, overexpression of miR-24 was associated with reduced JPH2 protein expression, which resulted in both reduced t-tubule junctional complexes and

calcium release. These effects of miR-24 overexpression, are similar to those observed when JPH2 is knocked down128, 129, suggesting that miR-24 directly regulates JPH2. Furthermore, following treatment with a miR-24 antagomir, to prevent binding of JPH2 to miR-24, calcium signalling and t-tubule structure was preserved in a mouse model of heart failure129. The above data suggests that targeting the pathways that regulate these mechanisms may provide a new target against heart failure.

Figure 1.9.Junctophilin (JPH2). JPH2, a scaffold protein forms junctions between the SR and t-tubule. JPH2 binds directly to RyR on the SR membrane and membrane lipids in the t-tubule membrane.

1.5.5. Telethonin

Telethonin (Tcap) is a stretch sensitive protein that is located at the z-discs and binds to proteins in the t-tubule membrane134 (Figure 1.10). Tcap associates with the membrane associated potassium channel subunit minK which localises at the t- tubules135. Tcap also interacts with the sarcomeric protein Titin, which is responsible for modulating passive stiffness. Thus, Tcap links the contractile apparatus of the cell to the membrane by anchoring t-tubules to the z-lines. It is therefore not surprising that Tcap gene mutations have been associated with altered stretched sensitivity. Mutations of Tcap in humans have been shown to result in dilated

cardiomyopathy or hypertrophic cardiomyopathy, both conditions can be associated with altered response to stretch136. Furthermore, these conditions have also been linked with t-tubule alterations. In addition to stretch sensitivity, Tcap appears to also be required for t-tubule development137. Knockdown of Tcap in zebrafish led to muscular dystrophy associated with altered t-tubule development137. It is also thought that Tcap plays a role in the response to cardiac overload. This is supported by several groups that have shown, using Tcap KO mice, cardiac defects only developed when the heart underwent mechanical overloading138, 139. Following mechanical overload, Tcap KO mice showed calcium handling defects and t-tubule loss. It is therefore not surprising that during heart failure, where t-tubule alterations have been shown to be load sensitive50, 140, Tcap protein expression was reduced 80. Furthermore, heart failure recovery has been associated with t-tubule restoration and Tcap recovery, which is likely to be due to decreased mechanical stress139. It is therefore thought that Tcap plays a load sensitive role in t-tubule regulation.

As with the other membrane proteins discussed here, little is known about the mechanisms that regulate Tcap function. Candasamy et al141, have shown that Tcap interacts with both protein kinase D and calcium/calmodulin dependent kinase, where it is bis-phosphorylated. Induction of a mutant non-phosphorylated Tcap led to disorganisation of t-tubules, indicating that phosphorylation of Tcap by protein kinase D may play a role in t-tubule organisation. Critically however, it has been shown that inhibition of protein kinase D does not inhibit phosphorylation of Tcap, thus indicating that other mechanisms must play a role in Tcap regulation.

Figure 1.10. Telethonin (Tcap). Tcap is a load dependent regulator of t-tubules located at the z-discs. It links the z-disc to the t-tubule membrane.

1.5.6. Proteins of the z-line

The cardiac z-lines (z-discs) make up the lateral boundaries of the sarcomere (Figure 1.2). Z-lines are anchoring sites for myofilaments, and therefore play major roles in sarcomeric organisation, myocyte shape and mechanical stretch. Importantly, in the heart, z-lines also anchor t-tubules135, 142. The z-lines are made up of numerous components that provide a link between the contractile apparatus of the cell and the cytoskeleton143. Some of the components that make up the z-lines include: sarcomeric proteins, such as actin which is important for organisation and function of the contractile apparatus144; cytoskeletal proteins, e.g. desmin that connects sarcomeres to z-lines145; integrins, connecting the z-lines to the extracellular matrix146 and; membrane associated proteins that anchor the myofibrils to the cell membrane and contribute to t-tubule stability147, 148. Notably, z-line proteins have also been shown to interact with scaffold proteins, such as BIN1, at the t-tubule thus providing membrane stabilisation102. Disruption to some of these components, have been observed in heart failure149 and have been associated with t-tubule loss, during prolonged cell culture150. Furthermore, inherited conditions leading to mutations to z-line proteins have been shown to cause cardiomyopathies136, which can be linked with altered response to stretch and t-tubule alterations.

Alongside structural support, a growing body of evidence now implicates a role for the z-lines in signal transduction in cardiac myocytes, which may also regulate t- tubules. For example, N-WASP is a z-line protein that is involved in transduction of signals from receptors to the actin cytoskeleton151. Interactions between this protein and BIN1 however, have been shown to stabilise the t-tubule membrane102. Furthermore, z-line proteins have also been shown to interact with several molecules involved in signalling pathways141, 143, 152, 153, including kinases and phosphatases, thus providing a link between cardiac cellular signalling and contractile force. Thus, it is thought that the components of the z-line are involved in mechano-transduction. This is where mechanical stress is sensed by the myocyte and translated into a transcriptional response, resulting in cellular hypertrophy. It is therefore possible that z-lines are the main site for cellular adaption to an increase in cardiac stress. The mechano-sensors and signalling pathways that lead to pathological remodelling and

t-tubule alterations during cardiac disease however remain unknown. As already disused, alterations to the t-tubule network in heart failure are likely a maladaptive result of increased mechanical load. It is therefore possible that t-tubule alterations result from activation or deactivation of z-line signalling pathways which also regulate t-tubule biogenesis.

1.6. Aims

In the last several years, there has been an increasing interest in research to enhance our understanding and knowledge of cardiac t-tubules. The work discussed in this chapter has explored some of this research and demonstrated that cardiac t-tubules are vital for maintaining normal contractility of the heart through the tight regulation of cardiac EC coupling. Crucially, the fundamental control of myocyte contraction by t-tubules has been further emphasised in the diseased heart, in conditions such as heart failure, where t-tubule loss is closely associated with dys-synchronous calcium release and impaired myocyte contractility. More recently it has been demonstrated that “recovery” from heart failure can result in restoration of t-tubules and, more importantly, normalised calcium handling. This data shows that while t-tubules may provide a useful target for treatments aimed at increasing contractile performance, there are still many unanswered questions. It is still not clear how t-tubules, are formed and maintained, or how t-tubule structure, and function change during conditions such as heart failure and atrial fibrillation. Therefore, if we gain a greater understanding of the mechanisms that control t-tubule regulation, it is possible that these could be used therapeutically for the treatment of cardiac disease.

Evidence suggests the regulation of t-tubules is multifaceted and several protein complexes may be involved. There is strong support however, to suggest that the BAR domain protein BIN1 controls t-tubule formation in cardiac muscle and thus, may play a vital role in calcium regulation. Several studies have shown that KO or silencing of BIN1 caused t-tubule loss in both skeletal and cardiac muscle46, 102, 113. Furthermore, in several cells types that usually lack t-tubules, induction of BIN1 has led to t-tubule formation83, 90, 104, 105, 107. This project will therefore aim to extend

these observations and determine if there is a role for BIN1 in the control of formation and restoration of cardiac t-tubules by addressing the following aims;

(1) To determine if t-tubule loss in a sheep model of heart failure can be recovered and to identify candidate proteins that lead to this recovery.

(2) To investigate proteins implicated in the formation and maintenance of t- tubules in the heart, using neonatal rat ventricular myocytes (NRVMs).

(3) To explore the role of the BAR domain protein AmphiphysinII (BIN1) in t- tubule biogenesis.

2. Methods

All procedures involving animals were carried out in accordance to the United Kingdom (Scientific Procedures) Act of 1986.

2.1. Sheep

Female Welsh Mountain sheep aged approximately 18 months were randomly assigned to control (non-instrumented), heart failure or recovery groups. Contributions of others to in vivo sheep work, including animal handling, surgery, electrocardiography, echocardiography and myocyte isolation are listed in Table 1.1.

2.1.1. Induction of heart failure in sheep

Under isoflurane anaesthesia (1–4% v/v in oxygen) sheep were instrumented with a pacing lead (St Jude Medical or Medtronic) that was fixed transvenously at the right ventricle apex and attached to a cardiac pacemaker (Medtronic) and buried subcutaneously. Post-operative analgesia (meloxicam 0.5 mg/kg) and antibiosis (enrofloxacin 5mg/kg) were provided for 24 hours and animals were allowed to recover post-operatively for one week before tachypacing was induced. Heart failure was induced by rapid ventricular pacing (210bpm) as described previously42, 58, 154, 155. Animals were monitored daily for onset of clinical signs of heart failure including lethargy, dyspnoea and weight loss and in vivo functional assessments recorded.

2.1.2. Recovery Sheep

Following the onset of heart failure in the ‘recovery’ group of animals, pacing was stopped and animals were allowed to recover for 5 weeks. The designated endpoint for the study was either continued / worsening signs of heart failure or 5 weeks of recovery (cessation of tachypacing).

2.1.3. Isolation of sheep ventricular and atrial myocytes

Sheep were killed by pentobarbitone (200 mg kg−1 i.v.). Heparin (10,000 i.u.) was used to prevent blot clots. The heart was quickly excised and washed in cold isolation solution (similar to the solution in Table 2.1) to clear the heart of blood. After separation of the ventricles and atria, the left atrium was cannulated via the small vessel of the left circumflex coronary artery. The atria was then mounted onto a Langendorff perfusion setup at 37°C. Using a collagenase and protease digestion technique as described previously42, 58, 154 single myocytes were isolated from the left auricle.

2.2. Neonatal rats

In order to study t-tubule formation in the heart, an animal model was needed in which t-tubules had not yet developed. Thus, for this part of the study two day old neonatal Wistar rats (Charles River UK Ltd) were used.

2.2.1. Isolation of neonatal rat ventricular myocytes

A litter of two day old neonatal Wistar rats (Charles River UK ltd) were killed by cervical dislocation and then decapitated. Bodies were rinsed in 70% ethanol and hearts excised and placed in ice cold disassociation buffer (Table 2.1). In a microbiological safety cabinet (SC-R, Labcaire), the atria and fat were removed and ventricles placed in fresh dissociation buffer. The ventricles were cut longitudinally and placed into dissociation buffer containing 0.75mg/ml Collagenase A (Roche), and 1.3mg/ml Pancreatin (Sigma) (enzyme solution) at 37°C. The enzyme solution containing the ventricles was spun at 120 rpm at 37°C for 7 minutes. The ventricles were then titrated with a 25ml stripette; this first digestion was discarded as this contains fibroblasts. Serial digestions were performed until the ventricles were completely digested. After each digestion, the cell suspension was collected, filtered through a cell strainer, placed in 2ml Foetal Bovine Serum (FBS) and maintained at 37°C. After all digestions, the cell suspension was centrifuged at 1000 rpm, 21°C for

5 minutes. The supernatant was discarded and cells re-suspended in pre-plating media (Table 2.2). Cells were then pre-plated on tissue culture dishes and incubated in a 5% CO2 -95% air atmosphere at 37°C for 75 minutes to allow fibroblasts to attach. After pre-plating, the supernatant was collected and myocytes counted using a haemocytometer. Myocytes were then diluted into 4x105cells/ml using plating medium (same constituents as pre-plating medium with 100uM Bromodeoxyuridine (BRDU) to stop fibroblast proliferation).

Substance Concentration (mM) NaCl 116 HEPES 20 Glucose 5.6 KCl 5.4

NaH2PO4 1

MgSO4 0.83 Table 2.1. Dissociation buffer, pH 7.35 with NaOH

Substance Concentration (%) DMEM 68 Medium 199 17 Horse serum 10 FBS 5 Fungizone 1 Penicillin streptomycin (10,000 units Pen/10mg/ml Strep) 1 Table 2.2. Pre-plating medium

2.3. Cell culture

2.3.1. Neonatal rat ventricular myocyte culture

Following neonatal rat ventricular myocyte (NRVM) isolation, cells were plated onto either 6 well culture slides (pre-coated with 1% gelatin in phosphate-buffered saline (PBS) for 30 minutes, aspirated off and air dried for 2 hours) or 8 well pre- coated µ-slide (Ibidi) at approximately 4 x 105/ml. Myocytes were maintained in maintenance media (Table 2.3) in a 5% CO2 -95% air atmosphere at 37°C. Maintenance media was changed every 2-3 days until cells were ready to be transfected.

Substance Concentration DMEM 80% Medium 199 17% FBS 1% Fungizone 1% Penicillin streptomycin (10,000 units Pen/10mg/ml Strep) 1% BRDU 100 μM Table 2.3. Maintenance media

2.3.2. Induced pluripotent stem cell (iCells) culture

Human induced pluripotent stem cell derived cardiac myocytes (hiPSC CMs) were purchased (Cellular Dynamics International, USA) and stored in liquid nitrogen until use. Cells were thawed and re-suspended in iCell Cardiomyocyte Plating Medium according to manufactures instructions (Cellular Dynamics). Following re- suspension, cells were plated onto either 6 well culture slides (pre-coated with 1% gelatin in Milli Q H2O for 1 hour at 37°C) or 8 well pre-coated µ-slides (Ibidi) at approximately 6 x 104 cells / cm2. Cells were maintained in iCell Cardiomyocytes Maintenance Medium (Cellular Dynamics) in a 5% CO2 -95% air atmosphere at 37°C. Medium was changed every 2 days until cells were ready to be transfected.

2.4. Generation of Plasmids and Transfection

Commercially available mammalian expression vectors, used to express the BIN1, Tcap or MTM1 gene were obtained from Origene (Table 2.4 and Figure 2.1). A cytomegalovirus (CMV) promoter was used to drive expression of these human genes. The vectors also contained kanamycin or ampicillin resistant genes and the epitope tag Myc-DDK or turbo (t) GFP to allow protein identification. Whilst the commercially available Tcap and MTM1 vectors already contained a fluorescence tag, tGFP, for protein identification, tGFP forms dimers that may affect protein localisation and interactions. Consequently, the MTM1 and Tcap fragments were removed and cloned into vectors containing mutated (m) GFP or mBFP tags. Mutated fluorescent proteins (mFP) are specifically optimised for protein localisation studies and have a monomeric structure156. pCMV6 cloning vectors expressing the fluorescence tag mKate2, mBFP or mGFP, containing ampicillin resistant genes (Table 2.4 and Figure 2.1), were used to generate fluorescently tagged BIN1, Tcap or MTM1 vectors (section 2.4.2, Table 2.5 and Figure 2.3). As C- terminal tagging could possibly interfere with the function of MTM1157, MTM1 was cloned into a pCMV6-AN vector. The pCMV6-AC-mKate2 vector was used as a control.

Fusion tag / Protein Vector Antibiotic resistance (Excitation/emission (nm)) BIN1 variant 5 (RC220682) Kanamycin Myc (n/a) BIN1 variant 8 (RC220616) Kanamycin Myc (n/a) BIN1 variant 9 (RC202423) Kanamycin Myc (n/a) Tcap (RG203158) Ampicillin mGFP (483/506) MTM1 (RG205306) Ampicillin tGFP (483/506) pCMV6-AC-mKate Ampicillin mKate2 (588/633) pCMV6-AC-mBFP Ampicillin mBFP (402/457) pCMV6-AN-mGFP Ampicillin mGFP (483/506) Table 2.4. Commercially available expression vectors purchased from Origene

Figure 2.1. Commercially available expression vectors. (A) pCMV6-AC-mKate control cloning vector; genes cloned in this vector will be expressed as a tagged protein with a C-terminal mKate2 tag. (B) BIN1 expression vector containing a Myc-DDK tag for protein identification. (C) Tcap expression vector containing a tGFP tag for protein identification. (D) MTM1 expression vector containing a tGFP tag for protein identification. AmpR = ampicillin resistance gene. NeoR/KanR = kanamycin resistance gene. Ori = origin of replication. CMV = cytomegalovirus (transcription promoter). Poly (A) = Poly adenylation (transcription terminator). tGFP = turbo green fluorescent protein. Restriction sites in bold. Primers used for gene identification in purple. Vector maps created using SnapGene.

2.4.1. Plasmid DNA expansion and purification

Plasmid DNA vectors were expanded via transformation as follows; under aseptic conditions, 50µl XL1-Blue competent cells (Agilent Technologies) were added to ~500 ng vector. Samples were incubated on ice for 30 minutes, heat shocked for ~45 seconds at 42°C to incorporate the plasmid DNA and incubated for a further 2 minutes on ice to recover. 200 µl super optimal broth with catabolite repression (SOC) (Fisher) was added to the cell/ligation mix and incubated at 37°C at 230 rpm for 1hr. The mixture was then spread evenly across Lysogeny broth (LB), antibiotic agar plates containing; either 100 µg/ml ampicillin or 50 µg/ml kanamycin (see Table 2.4 for vector antibiotic resistance). The plates were incubated overnight at 37°C. Individual colonies were picked from the plates and added to LB broth with antibiotics (100 µg/ml ampicillin or 50 µg/ml kanamycin). Cultures were incubated at 37°C at 230 rpm overnight. Plasmid DNA was then purified using QIAprep spin Miniprep kit (QIAgen) as per manufactures instructions. The concentration of samples was determined using a Nanodrop.

2.4.2. Generation of vectors with fluorescence tag

To be able to visualise successfully transfected cells when experimenting, vectors expressing BIN1 variants 5, 8 & 9, Tcap or MTM1 with a fluorescent tag were generated (Table 2.5 and Figure 2.3).

2.4.2.1. Digestion of vectors:

Purified plasmid DNA (~1 µg) from each colony were digested with restriction enzymes (1U) (for sites see Table 2.5) using 10X fast digestion buffer (Thermo) and incubated for 2 hours at 37°C. 6 X DNA loading buffer (Bioline) was added to the samples which were separated by electrophoresis on a 1 % agarose gel in Tris acetate EDTA (TAE) buffer (containing (mM); 40 Tris base; 10 EDTA and 0.35% acetic acid (pH 8.5)). 1 Kbp hyper-ladder (Bioline) was loaded alongside the samples as a molecular weight marker. Following separation, samples were visualised under UV

light and bands of the correct molecular weight (Figure 2.2 and Table 2.5) were cut out from the gel. DNA was purified using QIAEX II gel extraction kit (QIAgen), as per manufactures instructions.

Figure 2.2. Example of digested vector separation. Purified plasmid DNA was digested and separated by weight on an agarose gel. Restriction enzymes were used to target the gene insert. In this case MTM1 was digested using enzymes Sgf I, Rsr II.

2.4.2.2. Ligation of vectors:

The digested inserts and fluorescent vectors (Table 2.5) were ligated at a 1:1 ratio (containing;1 x T4 buffer; 400 U T4 DNA ligase; 1 µl digested insert (~50 ng); 1 µl digested vector (~50 ng); made up to 10 µl with RNAse free H2O). A sample with no insert was used as a negative control. Samples were incubated overnight at 4°C. The ligation mix was then transformed using 50 µl competent cells and ~500 ng ligation mix and purified using QIAprep spin Miniprep kit as described in section 2.4.1.

Generated Fluorescent vector Weight Insert Insert weight Restriction sites vector (in which gene to be inserted) pCMV6-AC-mKate2-BIN1v5 pCMV6-AC-mKate2 6582 bp BIN1 v5 1367 bp Sgf I, MIu I

pCMV6-AC-mKate2-BIN1v8 pCMV6-AC-mKate2 6582 bp BIN1 v8 1367 bp Sgf I, MIu I

pCMV6-AC-mKate2-BIN1v9 pCMV6-AC-mKate2 6582 bp BIN1 v9 1367 bp Sgf I, MIu I

pCMV6-AC-mBFP-Tcap pCMV6-AC-mBFP 6595 bp Tcap 504 bp Sgf I, MIu I

pCMV6-AN-mGFP–MTM1 pCMV6-AN-mGFP 6628 bp MTM1 1815 bp Sgf I, Rsr II

Table 2.5. Generation of vectors with fluorescence tag.

2.4.2.3. Sequencing of vectors:

To determine that the digested inserts had correctly incorporated into the fluorescent vectors (Figure 2.3), the purified plasmid DNA was digested with restriction enzymes (Table 2.5 and Figure 2.2) and visualised using electrophoresis as described in section 2.4.2.1. The purified plasmid DNA was then sequenced by The University of Manchester Sequencing Facility. The primers used for sequencing were purchased from Origene (VP1.5 forward primer: GGACTTTCCAAAATGTCG; XL39 reverse primer: CCCACCAGCCTTGTCCTAAT). Additional primers were used for BIN1 and MTM1 (Table 2.6). The generated sequence was aligned with the known gene sequence for BIN1 (v5 [NM_139347], v8 [NM_004305], v9 [NM_139350]), Tcap (NM_003673) or MTM1 (NM_000252) using MultAlin.

Gene Primer name Primer BIN1 BIN1 - Primer 1 - forward GTTCCCCGACATCAAGTCAC BIN1 BIN1 - Primer 2 - reverse AGACCATGAGAATCAAGGCG MTM1 MTM1 - Primer right TGATTGTGTGTGGCAAATGTCA MTM1 MTM1 - Primer left 1 AGTTTGGAAACGGATTCTTCTCT MTM1 MTM1 - Primer left 2 GGAGCATTGAAGGGTTCGAAA Table 2.6. Primer design. Additional primers used in the generation and characterization of pCMV6-AC-mKate-BIN1 and pCMV6-AN-mGFP–MTM1 expression plasmids.

Figure 2.3. Generated expression vectors with fluorescent tags. (A) pCMV6- AC-mKate2-BIN1 vector containing the mKate2 tag. SgfI and MIuI restriction sites used to clone the BIN1 insert into the pCMV6-AC-mKate2 entry vector. (B) pCMV6-AN-mGFP-MTM1 vector containing the mGFP tag. SgfI and RsrII restriction sites used to clone the MTM1 insert into the pCMV6-AN-mGFP entry vector. (C) pCMV6-AC-mBFP-TCap vector containing the mBFP tag. SgfI and MIuI restriction sites used to clone the Tcap insert into the pCMV6-AC-mBFP entry vector. AmpR = ampicillin resistance gene. Ori = origin of replication. CMV = cytomegalovirus (transcription promoter). Poly (A) = Poly adenylation (transcription terminator). Restriction sties shown in bold. Primers used for gene identification in purple. Vector maps created using SnapGene.

2.4.2.4. Generated vector expansion

The generated pCMV6 vectors (Figure 2.3) were expanded as described in section 2.4.1 (using 1 µl of mixture to grow bacterial colonies). Following overnight incubation at 37°C, individual colonies were picked from the plates and added to 50 ml LB broth containing 100 µg ampicillin. Cultures were incubated at 37°C at 230 rpm overnight. Plasmid DNA was then purified using Hi Pure plasmid Miniprep kit (Life Tech) as per manufacturers instructions. The concentration of DNA samples was determined using a Nanodrop.

2.4.3. Transient Transfection

NRVMs (2-4 days after isolation) and iCells (5 days after plating) were transiently transfected with either human BIN1 (variant 5,8 or 9), Tcap or MTM1 (Origene Inc, USA) cloned into pCMV6 entry vectors (Origene Inc, USA) or with pCMV6-AC- mKate2 as a negative control as per manufactures instructions. Briefly; Fugene 6 Transfection Reagent (Promega) was added to Opti-MEM (Life Tech) reduced serum media at a ratio of 1:10 and incubated for 5 minutes. 6 µg DNA (3:1 Transfection Reagent: DNA ratio) was added to the reagent/media mix and incubated for 15 minutes. The reagent/media/DNA mix was then overlaid onto the cells and incubated in a 5% CO2 -95% air atmosphere at 37°C for 48 hours.

2.5. T-tubule imaging

Freshly isolated adult sheep atrial myocytes were used to investigate the restoration of transverse tubules and cultured NRVMs and hiPSC-CMs (iCells, Cellular Dynamics International, USA) were used to investigate t-tubule formation in the heart.

2.5.1. Preparation of samples

To obtain images of the tubules, myocytes were stained with 2 µM FM 4-64 or 4 µM di-4-ANEPPS (both Molecular Probes) 10 minutes prior to imaging. Both indicators are voltage sensitive amphiphilic dyes that fluoresce in response to changes in membrane potential. If the cells were contracting spontaneously 2 mM of the calcium buffer ethylenediaminetetraacetic acid (EDTA) was also added to the cells.

2.5.2. Confocal Microscopy

Stained myocytes were imaged for t-tubules by confocal microscopy. Adult myocytes were imaged on a Leica SP2 or Zeiss 7Live; (excitation, 488 nm; emission >515 nm, 63X, 1.2 numeric aperture objective) at 100 nm xy pixel dimensions and 162 nm z-step size. NRVMs and hiPSC-CMs were imaged on a Nikon A1R+ confocal microscope (excitation, 488, 561 or 647 nm; emission 500-530 nm, 553- 618 nm or 663-738 nm, 100X, 1.4 numeric aperture) at 100 nm xyz pixel dimensions.

Confocal microscopy was used as it enabled detailed visualisation of structures within the cells by eliminating both background fluorescence and fluorescence from other regions in the sample. In contrast to widefield microscopy, light from a confocal can be focused on a specific spot and plane within the sample, and fluorescence is excited at exactly this point. Out of focus fluorescence is minimised by the presence of a pinhole in the optical pathway that only allows through light that is close to the imaging focal plane, resulting in a narrow depth of field (x-y ~200 nm, z ~500 nm). Images are acquired when emitted fluorescence is detected in a

photomultiplier tube (PMT) and the image visualised on a PC. Imaging in the z direction is made possible by the presence of the pinhole before the PMT. Thus, many thin sections of the cell to be imaged in the z direction, which can be assembled to create a 3D image (Figure 2.4).

For NRVMs, only cells that had been successfully transfected, thus expressed the fluorescent tag mKate2, were imaged and stacks obtained. A light emitting diode (LED) matching the excitation wavelength of mKate2 (~588 nm) was used to image mKate2.

Figure 2.4. Confocal microscope schematic. Blue lines denote excitation from the confocal laser. Green lines denote emitted fluorescence which is collected in the photomultiplier tube (PMT). The presence of pinholes minimises out of focus fluorescence.

2.5.3. Image analysis

Following image acquisition, confocal stacks were digitally deconvolved to correct for blurring caused by out of focus fluorescence, using either Huygens Professional (Scientific Volume Imaging) or NIS elements (Nikon) software. Deconvolution was achieved using the point spread function (PSF) of the microscope. The PSF is the 3D diffraction of light emitted from any point in the sample. The PSF was calculated ‘theoretically’ (NIS elements, Nikon) using known microscopic parameters or ‘measured’ (Huygens Professional,Scientific Volume Imaging) using 100 nm diameter polystyrene beads (TetraSpeck, ThermoFisher Scientific), imaged in the same way as the cells. Using samples of a known size enables the amount of out of focus fluorescence to be accurately determined which can be applied to the experimental data for deconvolution.

After deconvolution, samples were thresholded in ImageJ (National Institute of Health, USA) to remove background fluorescence. Using three central sections from each cell imaged, the average fluorescence intensity was determined and subtracted from all images in the confocal stacks.

2.5.3.1. Distance maps

T-tubule density was firstly assessed by calculating distance of each pixel within the cell to the nearest membrane (sarcolemma or t-tubule) using Interactive Data Language (IDL) routines (Exelis VIS, UK) written by David Eisner42, 46. Whilst this measurement is not a true assessment of t-tubule density, it does highlight the importance of t-tubules in minimising the distance of any point in a cell from a membrane. Distance between t-tubules is important when considering the role t- tubules play in ensuring a synchronous rise in calcium throughout the cell.

Firstly, central images from thresholded confocal stacks were selected avoiding the top and bottom images to prevent measurement of the surface of the cell. Blank sections were added above and below the confocal stack to prevent inaccurate measurements either side of the cell. Images acquired on the Leica confocal had

different z to xy spacing, which caused voxels to look distorted. Therefore, to achieve accurate 3D (x-y-z direction) measurements of the distance of each voxel to nearest membrane, confocal stacks were resampled using IDL routine “congrid” to make the z distance the same as the x-y. The distance of each voxel to the nearest membrane was then calculated using IDL routine “morph_distance”, this was shown on a distance map, whereby, colour intensity represents the distance. Dark areas on the distance map represent voxels in the section that are closest to membranes, light areas represents voxels that are far away from membrane (Figure 2.5 Aii).

Using the central section from each IDL distance map, colour distribution was determined in Image J and plotted cumulatively. The distance at which 50% of all voxels inside the cell were from the nearest membrane was calculated to give the ‘half distance’ (Figure 2.5 B).

Figure 2.5. T-tubule density analysis using distance maps. (Ai) Confocal image of cell showing staining of the surface sarcolemma and t-tubules, (ii) distance map of the cell produced in IDL. The intensity of pixels relates to distance, dark pixels represent areas closer to a membrane, as denoted on the scale bar. (B) Using the distance map, the distance at which 50% of voxels were from the nearest membrane, the half distance, was calculated.

2.5.3.2. Transverse tubule fractional area analysis

The second assessment of t-tubule density, which was independent of the effects of cell width, was calculated using ‘fractional area’ analysis, where the area of the cell occupied by tubules was measured. Central sections of the cell making up a 2 μm thick section were merged to create a single projection (Figure 2.6 Ai) and binary converted (Figure 2.6 Aii) in Image J. The fraction of black pixels (membrane stain, excluding surface sarcolemma) was expressed as a fraction of the total pixels in the cell to calculate t-tubule fractional area46 (Figure 2.6).

Figure 2.6. T-tubule fractional area analysis. (Ai) confocal image of cell showing staining of the surface sarcolemma and t-tubules, (ii) Image from (i) is binary converted.

2.5.3.3. Transverse tubule orientation analysis

Orientation analysis of the t-tubule network was performed on confocal stacks. Central sections of the cell (adding up to 2 μm in depth) were merged to create a single projection (Figure 2.7 Ai) in Image J. Images were rotated to ensure the cell was horizontal and the t-tubules lay perpendicular to the surface membrane. Surface membranes of the cells were then eliminated to ensure that there were no discontinuities (Figure 2.7 Aii). The images were binary converted, to make the tubules black and background white, before being “skeletonized” using the Image J, Fiji plug-in (Figure 2.7 Aiii and Aiv). The skeletons were analysed using the Fiji plug-in “directionality” which produced frequency plots of t-tubule angular information (Figure 2.7 B). To determine t-tubule orientation, the sum of t-tubules

corresponding to two principal cell directions (longitudinal 0°±15° versus transverse 90°±15°) was measured and the ratio between these orientated tubules was calculated46.

Figure 2.7. T-tubule orientation analysis. (Ai) Confocal image of cell showing staining of the surface sarcolemma and t-tubules. The surface membrane from image in (i) is eliminated (ii); binary converted (iii) and skeletonized (iv) using the Image J, Fiji plug-in. (B) Frequency plots of t-tubule angular information.

2.5.3.4. Branching analysis

To gain t-tubule branch information, 2 μm deconvolved and thresholded confocal stacks were selected in Fiji (Image J). The surface sarcolemma membranes were removed and the Fiji plug-in algorithm “Ridge detection” was used to define tubules (lines of less than 0.2 μm were excluded). The images were then “skeletonized” and

Fiji plug-in algorithm “analyse skeleton” was used to obtain information on t-tubule structures (Figure 2.8).

Figure 2.8. T-tubule branching analysis. (Ai) Original confocal image of cell showing staining of the surface sarcolemma and t-tubules; the surface sarcolemma membrane is eliminated (ii); tubules were defined using the Image J, Fiji plugin ‘ridge detection’ (iii) and (iv). (Av) Images were skeletonized.

2.5.3.5. Cell width

Cell width was calculated using the central image from each t-tubule stack. Firstly, in Image J, the scale of the cell was set and then a line drawn across a region of the cell representing the centre and the measurement recorded.

2.6. Measurements of intracellular calcium

To obtain measurements of intracellular calcium concentration, isolated NRVMs and sheep atrial myocytes were loaded with the intracellular indicator Fluo-8 Acetoxymethyl (AM) (Molecular Probes) that changes its fluorescence in response to changes in calcium concentration. Fluo-8 is a high affinity cell permanent calcium indicator that is excited at 488 nm with peak emission 520 nm. Fluo-8 has a low dissociation constant (Kd) of 389 nm allowing it to bind to calcium efficiently, thus, compared with other indicators, lower concentrations of dye are needed for high levels fluorescence. In comparison to other high affinity indicators such as Fluo-4, Fluo-8 has been found to be 2x brighter and offers improved cell loading158, 159. 4 µM Fluo-8 AM in 1µM pluronic dimethyl sulfoxide (DMSO), (20% w/v DMSO/pulronic F127) was added to the cells for 30 minutes. Fluo-8 is an AM ester and the addition of Pluronic DMSO solubilises the AM esters allowing them to cross the cell membrane. Once inside the cell, AM esters are cleaved by esterases into charged fluorophores, which are no longer able to cross back through the cell membrane trapping the free dye in the cytoplasm. To prevent further loading, the indicator was replaced with either fresh maintenance media (Table 2.3) containing 2 mM probenecid for NRVMs or normal Tyrode solution (Table 2.7) for sheep myocytes for 30 minutes prior to experimentation for de-esterification of the indicator. The addition of probenecid prevents indicator loss by blocking organic- anion transporters in the cell membrane. Organic-anion transporters can lead to efflux of intracellular dyes resulting in poor loading160.

Substance Concentration (mM) NaCl 140 Glucose 10 HEPES 10

KCL 4 Probenecid 2

CaCl2 1.8

MgCl2 1 Table 2.7.Normal Tyrode solution, pH 7.34 with NaOH

2.6.1. NRVM intracellular calcium

2.6.1.1. Measurements of intracellular calcium concentration

Loaded NRVMs were stimulated at 1 Hz with platinum electrodes using an ION OPTIX (C-PACE EP) stimulation box. 2000 frame sequences of beating myocytes were recorded at 200 frames/second using the Zeiss 7Live confocal microscope, fluorescence was excited at 488 nm and emitted at >515 nm as described in section 2.5.2. Experiments were performed at room temperature.

2.6.1.2. Analysis of intracellular calcium:

To quantify changes in intracellular calcium in transfected NRVMs, regions of interest (ROI) were selected at four points on the fluorescence image series using Image J. As shown in Figure 2.9, the following regions were selected; background (no cell); non-transfected cell; transfected cell (not on membrane); transfected cell (on tubule membrane). A LED matching the excitation wavelength of mKate2 (~588 nm) was used to determine which cells were transfected. In BIN1 transfected cells, mKate2 was found on the tubule membrane. For each ROI, fluorescence intensity was measured across the whole time series in Image J.

Figure 2.9. ROI analysis of NRVM. (A)Time series of transfected NRVMs loaded with a calcium indicator. ROIs representing, no cell (red), non-transfected cell (black/grey), transfected cell (pink) and on a tubule in a transfected cell (green), were selected and calcium transients plotted.

Fluorescence was normalised, firstly, by subtracting no cell (background fluorescence) from all measurements, then using the pseudo ratio (F/F0), where; F is raw fluorescence, F0 is fluorescence at diastolic levels. Following normalisation of fluorescence, the calcium transient amplitude was determined from the maximum intensity value, the peak of systolic calcium, for the first transient in each time series (Figure 2.10 A). The maximum rate of rise of calcium was calculated by dividing the difference in fluorescence (F/F0) between frames (Δ Ca) by the difference in time (Δ T) between frames. To measure the rate of calcium removal, macros were written in Microsoft Excel by Dr Charlie Pearman using the ‘Solver’ add-in. In brief, a single exponential curve was fitted to the decay phase of the calcium transient between the peak fluorescence and return to baseline fluorescence (Figure 2.10 B). The rate constant of decay (1/TAU) was measured from the reciprocal of the time constant of the fitted exponential.

Figure 2.10. Calcium transient analysis. (A) Red arrow denotes calcium transient amplitude, blue dashed line denotes peak fluorescence. (B) Red dashed line represents a single exponential curve fitted to the decay phase of the calcium transient.

2.6.2. Sheep myocyte intracellular calcium

2.6.2.1. Measurements of intracellular calcium.

The contributions of others to intracellular calcium release measurements in sheep myocytes are listed in Table 1.1 In brief, loaded myocytes were voltage-clamped using the perforated patch clamp technique, AxoClamp2B (Axon Instruments) and pCLAMP software (Molecular Devices, UK) with amphotericin-B (240 μg/ml) in the pipette solution42, 154 (Table 2.8). The addition of amphotericin-B causes pores to form in the cell membrane allowing access to the cell. After stabilisation, myocytes were superfused with DIDS solution (Table 2.9) via a gravity fed solution changer and heated tip set at 37°C. DIDS solution was used as it contained DIDS, 4-AP and

BaCl2 which block calcium activated chloride channels (Ito) and fast voltage 161 potassium channels (Ik1) . Thus calcium currents could be measured without interference of other channels. A glass capillary attached to a vacuum pump was used to maintain the level of solution in the bath. Cells were stimulated at 0.5 Hz using a voltage clamp protocol with a ramp from -60 to – 40mV and step to +10 mV to activate the L-type calcium current. Using high-speed xyt confocal imaging, fluorescence was excited at 488 nm; emitted light was collected at >515 nm and used to calculate intracellular calcium. Following the acquisition of the time series, 20 µM

wheat germ agglutinin (WGA) was applied to the cells via the gravity fed solution changer to visualize the t-tubule membranes and confocal z-stacks were recorded at 210 nm xyz pixel dimensions as described previously in section 2.5.2. Experiments were performed at 37°C.

Substance Concentration (mM)

KCH3O3S 125 KCL 20 HEPES 10 NaCl 10

MgCl2 5

K2EGTA 0.1 Amphotericin-B 0.26 Table 2.8. Pipette solution, pH 7.2 with KOH

Substance Concentration (mM) NaCl 140 Glucose 10 HEPES 10 4-AP 5

KCL 4 Probenecid 2

CaCl2 1.8

MaCl2 1 DIDS 0.2

BaCl2 0.1 Table 2.9. DIDS solution, pH 7.34 with NaOH

2.6.2.2. Analysis of intracellular calcium:

Script written in MATLAB (Mathworks, UK) by Professor Andrew Trafford was used to quantify changes in intracellular calcium (Figure 2.11). A ROI encompassing the whole cell was selected on the image time series (Figure 2.11 A). The programme then calculated the fluorescence (calcium) intensity over time for this ROI which was plotted as a series of transients (Figure 2.11 B). On some transients, the baseline fluorescence decayed over time, due to high powered continuous imaging, this baseline deviation is known as “baseline drift”. Therefore, a correction algorithm was applied to the time series to correct for baseline drift between transients. The time series was then spilt into 5 individual transients of 400 frames each (Figure 2.11 C). To allow the programme to accurately calculate the pseudo ratio (F/F0) to normalise fluorescence, frames representing resting fluorescence (F0) were inputted into the algorithm and images thresholded to remove background fluorescence. The ‘Dyssynchrony’ analysis algorithm (MATLAB) was then used to process the individual transients, where for each pixel of the ROI selected, a transient was generated. From these transients, the time it took to reach 50% peak fluorescence or 50% rise time (TF50), represented by the green dot on Figure 2.11 C, was calculated. The TF50 data points for each pixel within a cell were represented as a colour map (Figure 2.11 D). The standard deviation of the TF50 values (calcium rise time) for each pixel in the cell was used to represent the degree of dys- synchrony of the systolic rise of intracellular calcium. To determine the relationship between t-tubules and calcium rise time, the xy coordinates were obtained in ImageJ for these paired data points, distance maps (section 2.5.3.1) and TF50, were plotted against each other. Data from the three central calcium transients in the time series 2- 4 were averaged and used for analysis.

Figure 2.11. MATLAB Calcium analysis. (A) Confocal time series showing rises in fluorescence in a sheep myocyte loaded the calcium indicator Fluo 8 AM, a ROI for calcium analysis was selected as show in red. (B) Fluorescence intensity for the ROI in (A) is calculated over time. (C) The transients in (B) are spilt into 5 transients and the time course and fluorescence intensity profile of each pixel from each transient is plotted. (D) 50% peak fluorescence for each pixel (represented by the green dot on (C)) is represented on a colour map; warmer colours represent slower rise times.

2.7. Protein assessment

2.7.1. Protein extraction

For tissue samples, sheep were killed by heparin (10,000 i.u.) and pentobarbitone (200 mg kg−1 i.v.), as described in section 2.2. After hearts were removed and cleared of blood, the ventricles and atria were separated. Approximately 1 cm2 tissue samples were collected and stored in liquid nitrogen until use. For each sample, approximately 200 mg of frozen tissue was dissected into 1 mm cubes. The samples were homogenised (24,000 rotations/minute) in 2 ml ice cold Radio-Immuno Precipitation Assay (RIPA) buffer (Table 2.10) with added protease and phosphatase inhibitors (Table 2.11) for 3 x 5 seconds. The addition of inhibitors prevents protein degradation and phosphorylation of samples. Between each sample, the homogeniser probe was cooled to prevent the samples from overheating, residual tissue was removed and the homogeniser probe was washed with milli-Q H20 and RIPA buffer. The homogenised samples were then centrifuged at 5,445g for 10 minutes at 4 °C to remove insoluble tissue and samples were stored at -80 °C until use.

For protein extraction in NRVMs, cells were collected from 6 well culture plates using a cell scraper and were homogenised in 100 μl RIPA buffer containing protease and phosphates inhibitors (Table 2.11). Samples were centrifuged at 16,000g for 10 minutes at 4 °C to remove insoluble tissue such as lipids, the supernatant was aliquoted and stored at -80 °C until use.

Substance Concentration PBS Total Volume Igepal CA360 1% Sodium deoxycholate 0.5% SDS 0.1% Table 2.10. RIPA Buffer

Substance Concentration Phenylmethanesulphonylfluoride (PMSF) 0.1 mg/ml Sodium orthovanadate 100 mM Aprotinin 1 mg/ml Leupetin 1 mg/ml Table 2.11. Protease and phosphatase inhibitors.

2.7.2. Protein quantification

Protein concentrations were determined using the Bio-Rad DC protein colour metric assay kit according to manufacturer’s instructions. The reaction between protein and an alkaline copper tartrate solution cause a reduction in Folin reagent which led to colour development, representing protein quantity. Absorbance of each sample was measured at 750 nm using an absorbance plate reader (Bio Tek, ELX 800). Each sample was measured in triplicate and averaged and protein was quantified using the Bio Tek Gen5TM analysis software.

2.7.3. SDS-PAGE

For western blotting, NuPAGE® Novex 4-12% Bis-Tris (ThermoFisher Scientific) pre-cast polyacrylamides separating gels were used. Firstly, protein samples were prepared for separation by heating at 70°C for 10 minutes with the addition of 25% NuPAGE® Lithium dodecyl sulfate (LDS) Sample Buffer and 10% NuPAGE® Sample Reducing Agent (ThermoFisher Scientific). The sample reducing agent contains dithiothreitol (DTT), which breaks covalent disulphide bonds between protein structures. Heating the samples further reduces the disulphide bonds. Samples (see Table 2.13 for protein concentration) and 5 µl Biorad Prestained molecular weight marker were run at 200 V at room temperature in 1X NuPAGE® MES SDS Running Buffer (ThermoFisher Scientific) until the dye front reached the bottom of the gel. 0.25% Antioxidant (ThermoFisher Scientific) was also added to the gel tank to help maintain the proteins in a reduced state during electrophoresis by preventing re-oxidation of amino acids.

2.7.3.1. Protein transfer

Separated proteins were transferred from the gel to nitrocellulose membrane using the XCell II™ Blot Module (ThermoFisher Scientific). To prepare for protein transfer, Amersham Hybond-C Extra nitrocellulose membrane (GE Healthcare), filter paper and sponge transfer pads were equilibrated in 1X NuPAGE® transfer buffer (ThermoFisher Scientific), containing 20% methanol and 0.1% Antioxidant. Methanol was added to the transfer buffer as it minimises gel swelling thus increases the binding of protein to the membranes during transfer. Following gel electrophoresis, gels were removed from the gel cassette and were sandwiched with nitrocellulose membrane between a cathode and anode core in the XCell Surelock® Mini-Cell kit (ThermoFisher Scientific), assembled according to manufacturer’s instructions. The blot module was filled with transfer buffer and the outer chamber with Milli Q water. Transferred was performed on ice at 30 V for 75 minutes. Following transfer, the membranes were stained with either Ponceau-S (Sigma) or REVERT Total Protein Stain (LI-COR Biosciences) to visualise protein bands and to check the efficiency of the transfer. The membranes were then washed in Tris- buffered saline with 0.1% Tween 20 (TBS-T) (Table 2.12) to remove residual Ponceau-S or REVERT Wash Solution to remove Total Protein Stain (LI-COR Biosciences).

Substance Concentration Tris 20 mM NaCl 150 mM Tween 20 0.1%

Table 2.12. TBS-T buffer. pH 7.6 with HCL.

2.7.3.2. Membrane blocking

To reduce non-specific antibody binding to the nitrocellulose membranes, the membranes were blocked. Blocking buffers are dilute protein solutions that reduce

noise by binding to all areas of the membrane that is not occupied by protein. Following transfer, the nitrocellulose membranes were blocked using 5% non-fat milk or 5% Bovine serum albumin (BSA) (ThermoFisher Scientific) diluted in TBS- T for 1 hour at room temperature.

2.7.3.3. Detection of protein

After blocking, membranes were incubated with primary antibody diluted in TBS-T overnight at 4°C (see Table 2.13 for antibody concentrations) on a rotary shaker. The membranes were then washed with three changes of TBS-T before being incubated with either a horseradish peroxidise (HRP)-conjugated or fluorescent (IRDye) secondary antibody diluted in TBS-T specific for the species the primary antibody was raised in (Table 2.13). Membranes were incubated with the secondary antibody for 1 hour at room temperature on a rotary shaker; fluorescent (IRDye) membranes were kept in the dark. Following secondary antibody incubation, membranes were washed with three changes of TBS-T before detection of protein. For HRP labelled membranes, protein was detected using Pierce Super Signal ® West Pico Chemiluminescent system. Signal is generated as the HRP-conjugated secondary antibody catalyses the oxidation of a chemiluminescent substrate (luminol) which emits light during decay. The signal is further enhanced by the presence of an enhancer solution. These membranes were incubated with equal amounts of luminol solution and enhancer solution for 5 minutes and visualised after 10 minute exposure on a Syngene Chemiluminescent Bio Imaging System (Alpha, Innotech). For membranes incubated with IRDye fluorescent secondary antibodies, signal detection was visualised on an Odyssey® CLx Imager (LI-COR Biosciences) using the 800 channel.

2.7.3.4. Analysis of Western blots

Densitometric analysis of the protein bands was carried out using either the Syngene Gene Tools analysis program or Image Studio (LI-COR Biosciences). For each protein band, a lane was fitted and the pixel density calculated, the higher the pixel

density, the more protein expressed. Background fluorescence was subtracted using either the Syngene rolling disk function or the Image Studio average background function.

Tissue samples were repeated in triplicates, averaged and protein quantity measured as mean protein level normalised to an internal standard (IC). As certain ‘housekeeping’ proteins have been shown to change in heart failure, including in the sheep model used in this study46, protein levels were normalised to an IC which was loaded on all blots. The IC used was kept the same for each blot to eliminate differences in staining between each gel. Additionally, visualisation of Ponceau-S stained membranes ensured even gel loading and transfer (Figure 2.12). For NRVM samples, the ratio between mean target protein levels to Total Protein Stain was used to measure protein quantity.

Figure 2.12. Example Western blots. (A) Showing (from left to right): Bright field image to determine molecular bands using a molecular weight marker (MWM); Ponceau staining to identify the total protein; Dark field image showing protein band of interest following HRP-conjugated antibody incubation imaged on a Syngene Chemiluminescent Bio Imaging System. (B) Total protein stain in the 690 channel to visualise MWM and to ensure even loading; 800 channel fluorescence showing protein of interest following IRDye antibody incubation, signal was on detection on an Odyssey® CLx Imager.

Secondary Target Protein Primary Species Blocking antibody (all protein (µg) antibody 1:20,000) 1:500 HRP-conjugated sc23918 (SCBT), sc2005 (SCBT) or Sheep atria, BIN1 20 5% BSA ab54764 (Abcam) 925-32210 NRVMs mouse IRDye 800CW monoclonal (LI-COR) HRP-conjugated 1:1000 Sheep atria, sc2020 (SCBT) or JPH2 10 5% Milk sc51313 (SCBT) NRVMs 925-32214 IRDye goat polyclonal 800CW (LI-COR) 1:200 925-32210 ab2864 (Abcam) LTCC NRVMs 10 5% Milk IRDye 800CW mouse (LI-COR) monoclonal 1:200 HRP-conjugated Sheep atria, ab128318 sc2004 (SCBT) or MTM1 20 5% Milk NRVMs (Abcam) 925-32211 IRDye rabbit polyclonal 800CW (LI-COR) 1:200 925-32210 R3F1 (Swant) NCX NRVMs 10 5% Milk IRDye 800CW mouse (LI-COR) monoclonal 1:5000 A010-14 925-32210 PLN NRVMs 10 5% Milk (Badrilla) IRDye 800CW mouse (LI-COR) monoclonal 1:1000 925-32210 ab2827 (Abcam) RYR NRVMs 20 5% Milk IRDye 800CW mouse (LI-COR) monoclonal 1:1000 925-32210 sc73022 (SCBT) SERCA NRVMs 10 5% Milk IRDye 800CW mouse (LI-COR) monoclonal 1:500 HRP-conjugated Sheep atria, ab133646 sc2004 (SCBT) or Tcap 20 5% Milk NRVMs (Abcam) 925-32211 IRDye rabbit polyclonal 800CW (LI-COR) Table 2.13. Summary of Western blotting conditions

2.7.4. Immunocytochemistry

Immunocytochemistry was used to determine the cellular distribution of proteins. NRVMs or freshly isolated sheep atrial myocytes were plated onto 8 well pre-coated µ-slides (Ibidi), washed with 1X Phosphate buffered saline (PBS, Sigma) and fixed in 4% paraformaldehyde for 10 minutes. Following fixation, all samples were washed with three changes of PBS. If t-tubule visualization was required, samples were incubated with WGA Alexa Fluor® 488 conjugate (1:50, ThermoFisher Scientific) for 2 hours at 4°C. Following incubation with WGA, sections were washed with three changes of PBS. All samples were then permeabilised with 0.25% Triton x-100 (Sigma) for 10 minutes at room temperature. Samples were blocked with 10% goat serum in PBS containing 0.25% triton-X 100 (PBS-T) for 90 minutes, followed by incubation with antibodies for NCX (Swant), RyR, α-actinin, Troponin I and Vimentin (Abcam) (1:100, diluted in 1% goat serum in PBS-T) overnight at 4°C. The samples were then washed with three changes of PBS before incubation with an Alexa Fluor® (Molecular Probes) conjugated secondary antibody diluted in 1% goat serum in PBS-T for 1 hour at room temperature in the dark. Following incubation with secondary antibody the samples were washed with three changes of PBS. Cells were left in PBS and slides were imaged on either a Nikon A1R+ or LeicaSP2 confocal microscope as described in section 2.5.2.

2.7.5. Co-localisation analysis

Immunocytochemistry images were digitally deconvolved, as described previously, using Huygens Professional (Scientific Volume Imaging) or Nikon Elements (AR) imaging software. Following image deconvolution, images were thresholded to eliminate background noise and co-localisation analysis was performed (Huygens Professional or Nikon A1). To assess the fraction of overlap between t-tubules (WGA- Alexa Fluor®) and NCX or RyR in atrial myocytes, Manders co-localisation coefficients162 were used. Manders overlap, which indicates co-occurrence of two channels163, was used to determine the co-localisation of BIN1 with Tcap and MTM1, and tubules (WGA- Alexa Fluor®) with NCX and RyR in NRVMs.

2.8. Statistics

Data are presented as mean ± standard error of the mean (SEM), n cells from N animals. Where data was not normally distributed, data was transformed (log10 , reciprocal or square) using means appropriate to the skew of data164. Data have been compared using linear mixed model analysis to account for multiple cells from the same animal (SPSS Statistics; IBM, USA). Mixed model analysis accounts for both fixed i.e. treatment, and random i.e. animal, effects in the same analysis. The inclusion of random effects prevents pseudo-replication. For example, if random effects have not been considered and the data set includes more cells from one affected animal, treating individual cells from that animal as independent could lead to a false positive. Where effects were within the same animal or comparisons were made on cells only, differences were assessed using Student t test or One way ANOVA (SigmaPlot, Systat Software, USA). Two Way Repeated Measured ANOVA (SigmaPlot) was used to test multiple comparisons. GraphPad Prism (GraphPad Software, USA) was used to test for data correlations. Data were considered significant when p<0.05.

3. Results: Disordered, yet functional, atrial t- tubules on recovery from heart failure.

3.1. Introduction

As discussed in the General Introduction (Chapter 1), t-tubules are vital for maintaining normal contractility of the heart through the tight regulation of EC coupling. In the healthy heart, the localisation of key ion channels on or adjacent to the t-tubule allows for the rapid release and synchronous rise of systolic calcium. Whilst it was previously thought that t-tubules were absent in the atria, it is now well established that, at least in large mammals, atrial t-tubules are well developed and are able to trigger calcium release.

Moderate t-tubule loss in the ventricle has long been associated with the progression of heart failure50. The extent of ventricular t-tubule remodelling is tightly associated with the severity of left ventricular dysfunction50, with disruptions to the t-tubule network resulting in a decrease in the amplitude and synchrony of systolic calcium release48, 61, 67, 74. It is not until more recently however that cardiac disease such as heart failure and atrial fibrillation have been associated with almost complete loss of atrial t-tubules42, 49. Similarly, to the ventricle, atrial t-tubule loss has been associated with alterations to systolic calcium handling. Most notably decreased systolic calcium transient amplitude and delayed calcium release has been shown to occur in areas of the cell absent of t-tubules42, the consequence of which is impaired contraction of the myocyte. Which is why more recently, t-tubule and with it calcium transient amplitude restoration has become therapeutically desirable. Previous studies have shown, at least in the ventricle, that it was possible to partially restore t- tubules, and with them the systolic calcium transient, in failing hearts80, 82-84, 88. Given the extent of atrial t-tubule remodelling in disease (compared with the ventricle) however, it is unknown if it is possible to restore atrial t-tubules.

Despite the importance of the t-tubule network in controlling cellular contraction and the consequences of t-tubule loss on cardiac output, the mechanisms that control t- tubule formation and recovery remain largely unknown. Understanding the biogenesis and maintenance of the t-tubule system is vital as this knowledge may reveal novel mechanisms underlying the development of heart failure. In the ventricle, several proteins have been implicated in t-tubule biogenesis, including BIN1, JPH2, MTM1 & Tcap, these potential mechanisms will be discussed in more detail in later chapters. If these proteins play a role in atrial t-tubule formation or recovery however remain to be determined.

3.1.1. Heart failure model

To fully elucidate the mechanisms that control t-tubule dynamics during heart failure and recovery from heart failure, a clinically relevant model of heart failure is required. Whilst historically, rodent models have been useful to study many diseases and have greatly contributed to our understanding of cardiac disease, one major disadvantage is that small mammals often lack atrial t-tubules39, 42. Our lab, amongst others, has demonstrated that large mammals, such as sheep, have extensive atrial t- tubule networks that mimic what is found in the human43. Furthermore, large mammals are physiologically much more comparable to humans; heart size, heart rate and action potential shape are all much more similar than a rodent8, 165. Given this, in this part of the study we have chosen to use a sheep model of rapid right ventricular pacing induced heart failure.

Sustained rapid ventricular pacing leads to dys-synchronous activation of the ventricles resulting in severe diastolic and systolic dysfunction, thus representing what happens in tachycardia induced cardiomyopathy. Several models of rapid pacing have been used to date in the general understanding of heart failure166-168. End stage heart failure is characterised by elevated ventricular wall stress, cardiac dilatation and reduced contractility167, 168, all sympotoms produced by rapid pacing. The rapid pacing model has been used in this study as it not only provides a predictable and stable progression to a dilated cardiomyopathy phenotype which is comparable to clinical end stage heart failure166, it also shows similarities to the

progression of atrial fibrillation induced heart failure169. Furthermore, not only can the progression of cardiac dysfunction be monitored using this model, reducing the risk of sudden death, heart failure induced by this method is reversible after cessation of pacing170.

Our lab has previously used the ovine rapid pacing model of induced heart failure to characterise changes in calcium handling and t-tubule remodelling in heart failure42, 46, 58, 154, 155. We have reported that heart failure induced by this model was associated with well-defined clinical symptoms such as cardiac dilatation (fractional area change) and reduced contractility (fractional shortening)5, 18-21 and these observations were confirmed by echocardiography in the present study (Table 3.1). Moreover, allowing animals to ‘recover’ via cessation of pacing led to reversal of these symptoms and partial restoration of left ventricular fractional shortening, fractional area change and end diastolic and systolic dimensions (Table 3.1).

Pre pacing Heart failure Recovery

Fractional shortening (long 0.58 ± 0.03 0.21 ± 0.03 *** 0.39 ± 0.03 ###, *** axis view)

Fractional area change (short axis 0.69 ± 0.01 0.30 ± 0.04 *** 0.58 ± 0.03 ###, * view)

End diastolic 2.88 ± 0.15 4.18 ± 0.22 *** 3.56 ± 0.20 #, * dimensions (cm)

End systolic 1.24 ± 0.11 3.32 ± 0.25 *** 2.11 ± 0.19 ##, * dimensions (cm) Table 3.1. Mean data summarising echocardiographic parameters in sheep. Short and long axis echocardiogram images were used to calculate left ventricular fractional area change; fractional shortening, end diastolic internal diameter (EDID) and end systolic internal diameter (ESID). Fractional shortening = (EDID – ESID) / EDID. Measurements were taken at each time point for n =12 animals (8 paired); *p<0.05, ***p<0.001 vs Control; # p<0.05, ## p<0.01, ### p<0.001 vs heart failure; using One way ANOVA with repeated measures. All echocardiogram data collection and analysis performed by Amy Watkins, Jessica Clarke & Margaux Horn (see Table 1.1 for details).

3.1.2. Aims of the chapter

This chapter will begin to investigate the possibility of t-tubule and calcium transient amplitude restoration following heart failure by addressing the following aims:

1. Can atrial t-tubules be restored following their loss in heart failure? 2. Does atrial t-tubule restoration alter the systolic calcium transient? 3. Are candidate t-tubule associated proteins involved in t-tubule loss and recovery in the atria?

3.2. Results

3.2.1. Recovery of atrial t-tubules

To determine if t-tubules can be recovered in the atria following heart failure, a sheep model of rapid ventricular pacing was used, where pacing can be terminated, and recovery investigated170. Animals were paced to a point approaching end stage heart failure (35.7 ± 5.6; range 16 - 69 days) as assessed by echocardiography and the development of clinical signs of heart failure. We have previously demonstrated that a shorter pacing period of 4 weeks led to loss of virtually all atrial t-tubules42. At the stage of heart failure, pacing was stopped and a subset of animals allowed to recover for approximately 5 weeks (36 ± 0.8; range 31 - 39 days). This recovery period resulted in recovery of ventricular cardiac function (Table 3.1) and, consistent with our previously published data42, 154, atrial cell width recovered from the hypertrophy seen in heart failure (Table 3.2). T-tubule density was assessed two fold in left atrial myocytes from control, heart failure and recovery sheep; firstly, the fractional area of the cell occupied by t-tubules and the distance to t-tubule membranes as discussed in Methods, Section 2.8. T-tubule characteristics such as orientation and branching were also measured (results are summarised in Table 3.2).

3.2.1.1. Atrial t-tubule density

Figure 3.1 Ai shows typical planar (x-y) confocal images and an expanded view from sheep myocytes stained with di-4-ANEPPS. Consistent with our published data42, 46, a well-developed, organised t-tubule network was present in the control sheep atria. This is also evident in the bottom panels of Figure 3.1 Aii which show reconstructed transverse (y-z) sections from the cells. In the control cell, t-tubules can be seen projecting into the interior of the cell from the cell surface. This t-tubule network was almost completely lost following pacing induced heart failure, which is evidenced in the middle panels of Figure 3.1 A showing (i) planar (x-y) and (ii) end on (y-z) confocal sections. Following cessation of pacing (recovery) t-tubules were once again present in atrial myocytes but appeared disorganised compared with

control t-tubules. The density of the t-tubule network was quantified by two methods, firstly by determining the fraction of the cell occupied by tubules (Figure 3.1 C). Heart failure resulted in 81 ± 5.6% decrease of t-tubule fractional area compared with control. The fractional area of t-tubules in the recovered cells was no different to control.

Secondly, distance maps were used to determine the distance each point inside the cell was to the nearest membrane (t-tubule or surface sarcolemma) (Figure 3.1 B). As demonstrated using the grayscale map in Figure 3.1 B, dark areas on the distance map represent voxels in the image that are closest to membranes and light areas represents voxels that are far away from membrane. T-tubules in sections above and below the representative image are also accounted for. Distance maps of cells from (Ai) are shown in Figure 3.1 B. It is clear that the dark areas of the distance map correspond with membrane staining from (A). Distance plots generated from the distance maps were used to calculate the half distance (as described in Methods section 2.5.3). Here, the distance to membrane (t-tubule or surface sarcolemma) was calculated for each voxel intensity and plotted against the cumulative number of pixels in each distance map. The dotted lines on Figure Di shows that 50% of voxels were less than the distance indicated from the nearest membrane; this is known as the half distance. It is apparent from the representative distance plots (Figure 3.1 Di), in control and recovery (black and green lines) most voxels in the cell were much closer to a membrane than in heart failure (red line). Figure 3.1 Dii shows consistent with decreased t-tubules, half distance was increased by 129 ± 30.7% following rapid pacing and returned back to control levels in the recovery group, suggesting that the observed loss of t-tubules in heart failure resulted in a greater distance of points inside the cell to membrane.

Figure 3.1. T-tubules were restored following recovery from heart failure in the sheep atria. (Ai) Sheep atrial myocytes stained with di-4-ANEPPS to show t- tubules in control and recovery and loss in heart failure (HF). (Aii) Reconstructed transverse yz images showing a cross section through the cell. (B) Distance maps of myocytes in (Ai) showing the distance of any voxel in the cell to the nearest membrane. (C) Summary data for t-tubule fractional area. (Di) Representative distance plots of the cells in (B), dashed line shows that that 50% of voxels were less than the distance indicated from the nearest membrane. (Dii) Mean data summarising half distance to nearest t-tubule and surface membrane. n= 24-61 cells, N= 3-6 animals per group. *** p<0.001 vs control, ## p<0.01 vs heart failure tested by linear mixed model analysis. Scale bars = 10µm.

3.2.1.2. Atrial t-tubule structure and organisation

One striking observation when looking at the recovered t-tubules is that, despite restoration of atrial t-tubules, it is clear that t-tubule organization and structure is dramatically different in recovery cells when compared with control. Examples of this altered organization are shown in Figure 3.2 A where representative (x-y) confocal images (i) have been expanded and “skeletonized” in ImageJ (ii) as described in Methods section 2.5.3. To quantify atrial t-tubule disorganization, t- tubules were firstly categorized visually based on structural disorder (blinded observations). Figure 3.2 B shows that in control cells 90% of t-tubules had either normal or mild disorder but following recovery from heart failure, 81% of tubules showed extreme disorder. As t-tubules were almost completely absent in heart failure, t-tubule structural analysis on heart failure myocytes will not be discussed in this study. Skeleton analysis was used to determine if the orientation of the recovered t-tubules was altered compared with control. Once images had been “skeletonized”, the ImageJ (Fiji) algorithm “directionality” was used to produce frequency plots determining the angle that tubules occur (Figure 3.2 Ci). The number of tubules orientated at the two predominant angles 90 ± 15° (transverse) and 0 ± 15° (longitudinal) were quantified and a ratio between the two, transverse:longitudinal, was calculated. Figure 3.2 Ci shows in the representative cells t-tubule orientation largely shifted from transverse in control to longitudinal in recovery. Moreover, the mean data shows the ratio of transverse:longitudinal tubules was decreased 76.1 ± 6.9% in the recovery cells compared with control, suggesting that less of the recovered tubules were transversely orientated (Figure 3.2 Cii).

Lastly, “skeletonized” images were also used to quantify t-tubule structural alterations by determining t-tubule length and the extent of t-tubule branching obtained from the Fiji plug-in algorithm “analyse skeleton”. Figure 3.2 Di & ii shows that t-tubules were 79.8 ± 34% longer and 63.7 ± 28% more branched in the recovery cells. Furthermore, Figure 3.2 Diii shows decreased numbers of tubules with no branches but 225 ± 118% increase in structures with 5 or more branches in recovery. Whilst the level of t-tubule disorder varied between cells and animals, a correlation existed between tubule branching and t-tubule angle indicating that

animals with the most branched tubule network also contained the most longitudinally orientated tubules (Figure 3.2 E). Together these data suggest that atrial t-tubules were restored but disrupted following recovery from heart failure, where recovered t-tubules were longer, with increased branching and longitudinal elements.

Figure 3.2. Atrial t-tubules were disordered following recovery from heart failure. (A) T-tubule and surface sarcolemma staining showing organised t-tubules in control, loss in heart failure and disorder following recovery. Enlarged section of cells in (i) that have been skeletonised (ii). (B) Mean data showing t-tubules categorized based on disorder in control and recovery cells. (Ci) Control t-tubules were orientated at 90°, following recovery t-tubules were mainly longitudinal (0°). (Cii) Following recovery, the transverse to longitudinal tubule ratio in the atria was decreased. (D) T-tubules in recovery cells were longer (i) and more branched (ii & iii). (E) The level of t-tubule branching correlated with the t-tubule angle. n= 30-54 cells, N= 6 animals per group. * p<0.05, ** p<0.01, *** p<0.001 vs control, tested by linear mixed model analysis or Pearson’s coefficient (R2=0.614, N=6 per group). Scale bars =10µm.

3.2.1.3. Atrial t-tubule remodelling

In addition to the remodelling discussed above, compared with control (Figure 3.3 A), while some t-tubules in recovered cells appeared relatively normal others formed complex structures e.g. longitudinal, ‘doublet’ and ‘lattice’ tubule forms (Figure 3.3 Bi). These structures often appeared as ‘doublets’ where it was not possible to resolve if these structures consist of a single wide t-tubule or 2 t-tubules which move in parallel through the cell (Figure 3.3 Bi).

To determine if the ‘doublet’ membrane staining in recovery cells resulted from a wide single tubule or pair of tubules in close proximity, a cell impermeant fluorescent indicator, fluo-5N, was added to the extracellular solution to fill the tubules from the ‘outside in’. Highly disorganised structures were visible with fluo- 5N showing they were continuous with the cell surface membrane being open to the cell exterior and therefore may be functionally important (Figure 3.3 Bii). Fluo-5N did not fill the space between parallel t-tubule pairs suggesting they were separate, parallel running tubules or doublets (Figure 3.3 Bii). T-tubule width was measured by determining the intensity profile following fluo-5N staining (Figure 3.3 Biii). Summary data in Figure 3.3 Ci shows that where doublets did not occur, t-tubules were 2.2 ± 0.2 m apart which was similar to the spacing between neighbouring doublets and in control cells. However, the spacing between individual tubules forming the doublet was 0.7 ± 0.03 µm suggesting recovered t-tubules were present in locations other than the z-line (Figure 3.3 Ci).

Membrane pairs were also found in the longitudinal plane. Fluo-5N filled the space between some of these longitudinal structures, showing the existence of both very wide longitudinal tubules (Figure 3.3 Bii right) and longitudinal doublets. In summary, recovered t-tubules were often highly disordered and included longitudinal tubules, some of which were very wide and some in pairs of tubules which ran in parallel, possibly around the same z-line.

Figure 3.3. Recovered atrial myocyte characteristics (A) Control atrial myocytes stained with di-4-ANEPPS or Fluo 5N to visualise t-tubules. (B) Atrial t-tubule remodelling following recovery. Staining with di-4-ANEPPS (i) or Fluo 5N (ii) show some recovered t-t-tubules occur in pairs. (iii) Representative intensity plots over distance corresponding to the images in (ii) showing spacing between tubules. (C) Mean data for spacing between tubules. n= 15-42 tubules, N= 5-11 cells per group. *** p<0.001 vs control, ### p<0.001 vs heart failure tested by One way ANOVA. Scale bars = 10µm.

Control Heart failure Recovery

Cell width (µm) 16.31±0.64 18.62±0.20 * 15.81±1.07 #

T-tubule fractional 0.063±0.01 0.012±0.003 *** 0.104±0.02 ## area (µm)

Half distance (µm) 0.82±0.06 1.88±0.22 *** 1.18±0.14 #

Orientation (T:L 2.13±0.27 - 0.65±0.07 *** ratio)

T-tubule length 2.16±0.18 - 3.79±0.66 *** (µm)

T-tubule length 1.62±0.13 - 2.65±0.41*** (µm) n=cells; N=animals n=24, N=6 n=26, N=3 n=61, N=6

Table 3.2. Summary of control, heart failure and recovery t-tubule parameters. *p<0.05 , ***p<0.001 vs control, #p<0.05, ##p<0.01 vs heart failure

3.2.2. Recovered atrial t-tubules were functional

The first section of this chapter demonstrated that t-tubules were restored, but largely disrupted, following recovery from heart failure. The next aim was therefore to determine if atrial t-tubule recovery was also associated with recovery of systolic calcium and if the recovered t-tubules could function and trigger calcium release despite their extreme disorder. These experiments were performed in voltage- clamped cells by Andrew Trafford and Katharine Dibb and the analysis was completed by me (see Table 1.1 for details).

3.2.2.1. T-tubules were the main site for calcium release

The systolic rise in calcium was recorded using high-speed xy confocal imaging as shown in the recovery cell in Figure 3.4 A. Calcium was measured during systole as a representation of Fluo-8 fluorescence. WGA was then applied to the cells to visualise the surface sarcolemma and t-tubules. Figure 3.4 B shows the early rise in calcium following stimulation in control, heart failure and recovery myocytes stained firstly with the calcium indicator Fluo 8AM (i) and then WGA (ii) to visualise t- tubules. In the control cells, the systolic rise of calcium occurred first along the t- tubules, whilst in heart failure cells, where t-tubules were absent; this initial rise of calcium was restricted to the cell surface sarcolemma. It is also clear to see that in the recovery atrial cells, the rise of intracellular calcium occurred first along the disordered t-tubules before propagating to the rest of the cell. Merged images in Figure 3.4 Biii show calcium release sites and t-tubules appeared to be co-localised.

To determine if recovered t-tubules were able to trigger calcium as effectively as control tubules, the time for the systolic calcium transient to reach 50% peak fluorescence (TF50) (Section 2.6.2) for each pixel in a cell was calculated (Figure 3.4 Biv). The summary data in Figure 3.4 C shows in heart failure myocytes, calcium rise time was slowed, with the time taken for the systolic calcium transient to reach TF50, increasing 208 ± 33% from 65 ± 1.9 ms in control cells to 201 ± 21.3 ms, as represented by the darker colours in the heat map in Figure 3.4 Biv.

Following recovery from heart failure, the rate of calcium rise was restored back to control levels (Figure 3.4 C).

To determine if faster rise time following recovery from heart failure was due to t- tubule restoration, Figure 3.4 D shows representative (i) and mean data (ii) correlations between the rate of rise of calcium and the distance to the nearest tubule (in the x,y or z direction). As already discussed in this chapter (Figure 3.1 D), t- tubule half distance was greater in the heart failure cells, indicating a greater distance between tubules. Figure 3.4 Dii illustrates that there was a positive correlation between rise time and distance to t-tubules. This data confirms that calcium was faster to rise in the recovery and control cells where the cell interior was closer to membrane, i.e. t-tubules, than in heart failure. This data suggests that the recovered t- tubules were as effective as control in triggering calcium release.

Figure 3.4. T-tubules were the main site for calcium release. (A) Representative time series showing calcium release over time in a recovered atrial myocyte. (B) Representative atrial myocytes showing; calcium release (i), t-tubule and surface sarcolemma staining (ii), merged image showing calcium release sites and t-tubules were co-localised (iii), calcium rise time represented by a colour map (iv). (C) Summary data for 50% calcium rise time. (D) Representative (i) and (ii) mean data showing a correlation between calcium rise time and distance to t-tubules. n= 4-6 cells, N= 1-3 animals per group. *** p<0.001 vs control, ### p<0.001 vs heart failure tested by ANOVA or Pearson’s coefficient (R2 = 0.8457, n=16).

100

3.2.2.2. T-tubule recovery restored synchronicity of calcium release.

A key measurement of systolic calcium function is the synchronicity in which calcium is released. The standard deviation of the TF50 values (calcium rise time) for each point in the cell was used to represent the degree of dys-synchrony of the systolic rise of calcium as shown in Figure 3.5. In heart failure myocytes, there was a 92 ± 15% increase in the standard deviation of the rise time from 36 ± 1.2 ms in control myocytes to 69 ± 5 ms in heart failure myocytes, representing more dys- synchronous calcium release. Following recovery, the standard deviation of TF50 was no different to control values, indicating that the systolic rise of calcium was as synchronous in recovery cells as observed in control. This likely to be because following recovery, more points of the cell were in contact with t-tubules.

Figure 3.5. T-tubule recovery restored synchronicity of calcium release. Mean data showing calcium rise time was more synchronous in recovered myocytes compared with heart failure (HF). Dys-synchrony was measured using standard deviation of rise time. n= 2-6 cells, N= 1-3 animals per group. *** p<0.001 vs control, ### p<0.001 vs heart failure tested by ANOVA.

101

3.2.2.3. Cellular distribution of calcium handling proteins

The results from the previous section indicate that the initial rise of calcium occurred at the t-tubule in the recovery cells. This suggests that calcium handling proteins which are required for calcium rise were localised to the recovered t-tubule. The distribution of these proteins and their localisation to a t-tubule membrane were assessed in fixed cells using immunofluorescence and confocal microscopy (results summarised in Table 3.3).

Previous studies have shown that both the LTCC and NCX localise to the t-tubule membrane in healthy myocytes3, 14. Whilst we did not have a suitable antibody for the staining of the LTCC, Figure 3.6 shows that the NCX (stained with an Alexa Fluor 488 conjugate) was strongly co-localised with the t-tubules (WGA – Alexa Fluor 647 stained) in the control and recovery cells. In heart failure cells, where t- tubules were almost completely absent, the NCX staining was only present on the cell surface sarcolemma (Figure 3.6 Ai). The fraction of co-localisation of the NCX and t-tubules/surface sarcolemma was assessed using Manders coefficients (Huygens Professional). Figure 3.6 Aii shows that there was a strong co-localisation between NCX and membrane (t-tubules and surface sarcolemma) across all groups (control, 0.85 ±0.03; heart failure, 0.86 ±0.03; recovery, 0.85 ±0.01).

102

Figure 3.6. Cellular distribution of NCX. (Ai) Representative immuno staining of NCX (Green) and tubules (Red) in atrial myocytes. Yellow represents co- localisation. (Aii) Mean data showing Manders co-localisation of protein NCX and membrane (t-tubule and surface) in control, heart failure (HF) and recovery atrial myocytes. n= 4-9 cells, N= 1-2 animals per group, no change between groups, tested by ANOVA. Scale bar =10µm.

103

Dyadic structure and organisation is a key determinant of efficient EC coupling in cardiac myocytes. Therefore the effects of t-tubule remodelling in heart failure and recovery on the localisation of the RyRs to the t-tubules was studied. Consistent with RyRs occurring on the z-line5, immunostaining in Figure 3.7 Ai showed that RyRs had a sarcomere distribution in each group. In control and recovery cells, t- tubule (WGA) staining co-localised with RyRs at the z-line (Figure 3.7 Ai & ii). Co- localisation was not absolute in these cells as t-tubules do not occur at every z-line in the atria. In the heart failure group, where t-tubules were absent, the fraction of co- localisation between tubules and RyR decreased 63.5 ± 14% as shown in the summary data in Figure 3.7 Aii (see Table 3.3 for details).

One striking observation from the RyR stained cells in Figure 3.7 Ai is that the distribution of RyR protein was sparse at surface in control but clearly present at the point of heart failure, suggesting a redistribution of RyRs in heart failure (‘surface merge’ panel Figure 3.7 Ai). This is shown by increased co-localisation with surface membrane when t-tubules were removed. Independent of interior RyR, the fraction of co-localisation between surface RyR and cell membrane increased 92.9 ± 42% in heart failure myocytes, compared with control cells (Figure 3.7 A iv). Following recovery, the amount of RyR co-localising with the surface sarcolemma was reduced by 74.1 ± 4%, indicating that in heart failure there was more surface RyR which was redistributed with the recovery of atria t-tubules (Figure 3.7 Aiv). Additionally, when the surface of the cell was removed, the fraction of co-localistion of interior RyR and t-tubules (‘t-tubule merge’ panel Figure 3.7 Ai) was reduced 97.1 ± 0.7% in heart failure compared with control (Figure 3.7 Aiii).

Finally, compared to control and recovery, where confocal imaging shows continuous RyR staining, in heart failure, RyR staining was fragmented and also occurred between z-lines (Figure 3.7 Bi). RyR distribution was measured in two ways; firstly, the Fiji algorithm “analyse skeleton” was used to determine how many separate areas of RyR staining existed along the z-line. The summary data shows that the average number of RyR stained regions per sarcomere increased 116.8 ± 40% in heart failure and returned to control values following recovery (Figure 3.7 Bii). Secondly, RyR intensity plots, taken from the images in Figure 3.7 Bi, across RyR

104

stained sarcomeres displayed disordered patterns across both the longitudinal (blue dashed line) and transverse directions (orange dashed line) in heart failure (Figure 3.7 Biii). This data shows that RyR staining was more fragmented in heart failure and indicates larger gaps between the RyR regions following rapid pacing in the atria. Recovery from heart failure led to the general distribution of RyRs returning to control levels.

105

Figure 3.7. Cellular distribution of RyR. (Ai) Representative immuno staining of RyR (Green) and tubules or surface membrane (Red) in atrial myocytes. Yellow represents co-localisation between RyR and t-tubules with or without surface membrane. (Aii-iv) Mean data showing Manders co-localisation of RyR and t- tubules with or without surface membrane or independent of interior RyR in control, heart failure and recovery atrial myocytes. (Bi) Representative staining of RyR showing distribution along the sarcomere in both the transverse (orange dashed line) and longitudinal (blue dashed line) orientation. (B) Summary data for number of RyR structures per sarcomere (ii) and representative intensity blots across RyR stained sarcomeres in both longitudinal (blue) and transverse directions (orange) (iii). n= 5-7 cells, N= 1-2 animals per group, * p<0.05, ** p<0.01, *** p<0.001 vs control, ## p<0.01, ### p<0.001 vs heart failure tested by One way ANOVA. Scale bar =10µm

106

Control Heart Failure Recovery

NCX & t-tubule co- 0.85 ±0.03 0.86 ±0.03 0.85 ±0.01 localisation (Manders)

Total RyR & t-tubule co- 0.19 ±0.04 0.07 ±0.02 0.2 ±0.06 localisation (Manders)

Surface RyR & t-tubule co- 0.42 ±0.09* 0.81 ±0.03 0.21 ±0.03# localisation (Manders)

Cytosolic RyR & t-tubule 0.68 ±0.07* 0.02 ±0.004 0.46 ±0.04# co-localisation (Manders)

Number of structures 4.52 ±0.78* 9.80 ±0.61 3.55 ±0.68# per sarcomere n=cells; N=animals n=5, N=2 n=5, N=1 n=5-7, N=2

Table 3.3. Summary of t-tubule and protein co-localisation parameters in sheep myocytes *p<0.05 vs control, #p<0.05 vs heart failure.

107

3.2.3. Expression of t-tubule associated proteins

Having established that recovered t-tubules were functional, the next aim sought to determine how t-tubules might be recovered in the heart and thus identifying mechanisms to restore t-tubules therapeutically. Here the expression levels of proteins that are associated with t-tubule formation were investigated in our recovery model using western blot. Target protein levels were quantified by densitometry and expressed as a fraction of the internal standard (IC).

Expression levels of BIN1 (A), Tcap (B), MTM1 (C) and JPH2 (D) in sheep atrial tissue samples are shown in Figure 3.8. (A) Two BIN1 protein bands were detected corresponding to isoforms of 55 kDa and 37 kDa, this is consistent with previously published work from our laboratory that identified both bands in the sheep & ferret heart46. Therefore, both molecular weights were quantified for total mean BIN1 levels. Total BIN1 protein expression was decreased 24.8 ± 8.6% in heart failure when compared with control. There were no differences in the expression of BIN1 in the recovery samples when compared with control or heart failure. Representative blots also show single protein bands corresponding to the reported molecular weights; 19kDA for Tcap (B) and 55kDA for MTM1 (C) in the sheep atria. Both Tcap and MTM1 expression was decreased in heart failure when compared with control (49.4 ± 9.6% and 24.5 ± 7.9% respectively) and returned to control levels following recovery. There was no change in the expression of Tcap or MTM1 between control and recovery samples. Figure 3.8 D shows a protein band of the reported molecular weight of JPH2, ~94kDa, was detected. In addition, protein bands were also detected at 50kDa and 25kDa. As reported previously46, a negative control experiment confirmed that the additional bands at 50kDa and 25kDa were due to non-specific binding. Therefore only the 94 kDa band was quantified for JPH2 levels. No change in expression was detected between the groups (P=0.91). In summary, BIN1, Tcap and MTM1 protein levels were associated with changes in t- tubule density and structure in the atria.

108

Figure 3.8. Recovery of t-tubules was associated with restoration of membrane associated proteins in the atria. Representative western blots and mean data showing protein expression of t-tubule associated proteins BIN1 (A), Tcap (B), MTM1 (C) & JPH2 (D) in control, heart failure (HF) and recovery sheep atrial tissue. Full length blots showing protein molecular weight confirmed the correct protein was labelled. Positive controls, recommended by antibody manufacturer, were used to confirm antibody selectivity. (A) BIN1 protein expression decreased in heart failure in the atria. Tcap (B) and MTM1 (C) both decreased in heart failure in and increased back to control levels following recovery. (D) JPH2 protein expression was unaltered in atrial tissue. n=7 animals, * p<0.05, ** p<0.01 vs control, ## p<0.01 vs heart failure tested by linear mixed model analysis.

109

3.3. Discussion

The data in this chapter has demonstrated that atrial t-tubule loss in heart failure can be recovered by cessation of rapid pacing. Despite recovered atrial t-tubules being disordered and remodelled in appearance, they were functional in that they were able to trigger rapid and synchronous systolic calcium release. The decreased expression of the t-tubule associated proteins BIN1, MTM1 & Tcap in heart failure and their return to control levels in recovery indicates a potential role for these proteins in the recovery of atrial t-tubules.

3.3.1. Atrial t-tubules can be recovered following loss in a rapid pacing model of induced heart failure.

Consistent with previous studies from this laboratory, the data presented here shows that sheep had a well-developed atrial t-tubule network that was almost completely lost following rapid ventricular pacing42, 46. More notably, this study has shown for the first time that atrial t-tubule density was restored following cessation of rapid pacing. As much less is known regarding the structure and function of atrial t- tubules, it not surprising that, to our knowledge, atrial t-tubule recovery has not been previously studied. It is therefore not possible to make comparisons to other studies of recovered atrial t-tubules in this discussion. Despite this, our finding is consistent with studies that have demonstrated that interventions with the aim to restore cardiac function have resulted in t-tubule recovery79, 80, 83, 84 or prevention of t-tubule loss81, 82 in the failing ventricle. Specifically, t-tubule restoration in failing hearts has been observed following mechanical support84, SERCA gene therapy80, resynchronization therapy88 and treadmill exercise79 or attenuated their loss with β-blockers81 and PDE5 inhibition83. The mechanisms as to how t-tubules are lost and recovered are currently unknown, a common feature of the above-mentioned studies however is that recovery appeared to result from normalisation of cardiac load, induced directly or indirectly by different interventions. Increased cardiac workload has been shown to lead to hypertrophy, which eventually progresses to heart failure171. Whilst t- tubule alterations have long been associated with hypertrophy and heart failure42, 46,

110

48, 50, 60-67, 74 (human heart failure secondary to hypertrophic cardiomyopathy resulted in impaired t-tubule structure60), it is still largely unknown if t-tubule loss is a consequence of the maladaptive alterations associated with cardiac disease or if t- tubule loss is a direct result of cardiac overload. Evidence in support of the latter comes from Wei et al50 that suggest t-tubule remodelling plays a critical role in the transition from hypertrophy to heart failure. In this study, t-tubule remodelling occurred in hearts with normal global function, in response to pressure overload prior to the onset of heart failure50. This work suggests that t-tubule remodelling is an early, possibly causal, event in the progression to heart failure. Other studies looking at the effect of increased cardiac workload on the t-tubule network have demonstrated that factors such as cell strain172 and elevated wall stress173 could be associated with t-tubule remodelling. This is in agreement with the data presented in this study where atrial cell hypertrophy was associated with t-tubule loss in heart failure. Moreover, cessation of pacing led to reduced dilatation of the ventricle thus reducing volume overload on the heart, this in turn likely reduced atrial load and was sufficient to reverse atrial cellular hypertrophy and lead to the recovery of t-tubules. Taken together these data further support the concept of a load-sensitive t-tubule network. The mechanisms thought to mediate load sensitivity will be explored in section 3.3.4.

3.3.2. Recovered atrial tubules were disorganised

In the ventricle, t-tubule disorganisation, including t-tubule sheets69 and longitudinal tubules62, are a common feature of the failing myocyte. Moreover, most ventricular t-tubule recovery studies showed a more organised t-tubule network following interventions to treat the failing heart79, 81, 82, 84. Interestingly, this was not our observation in this study in the atria where recovered atrial tubules were highly disordered and consisted of mainly longitudinal elements. The consequence, if any, this has on calcium handling will be explored in the next section, evidence suggests however, that an increase in longitudinal tubules could be a compensatory mechanism to account for decreasing myocyte function. Swift et al174 demonstrated increased t-tubule density due to an increase in longitudinal tubules in their SERCA

111

KO mouse that had declining SR function. This increase in longitudinal tubules enabled a switch from SR calcium generated to sarcolemma calcium influx driven calcium transients through increased calcium influx via NCX, which were predominantly located on the newly formed longitudinal tubules174. Furthermore, resynchronisation therapy by Sachse et al88 led to improved t-tubule structures in the failing ventricle, which was made up of both longitudinal and transverse elements and led to restoration of the calcium transient amplitude. One of the major findings of this study however, is that since atrial t-tubules were almost completely lost in heart failure, the recovered atrial tubules observed must have been completely new t- tubule structures and not remnants of old t-tubules. Therefore the reason these structures have re-formed in such a disordered fashion is likely to be more than just a compensatory mechanism since we assume the most efficient t-tubule would lie close to RyRs in a transverse manner. Factors determining t-tubule organisation are still largely unknown and will be discussed further in the next chapter of this Thesis. Some data suggests however that the protein JPH2 may play a role in t-tubule organisation46, 50, 124, 131, 173 and a more organised t-tubule network has been linked with restoration of JPH282. In contrast, JPH2 was not required for re-tubulation in all models of t-tubule recovery80, 84 or for normal t-tubule development124 and we observed no changes in this study. One final factor to consider is how tubule organisation during re-growth is influenced by intercellular organelle and other structures. The heart failure phenotype is associated with many myocyte structural alterations130, 175, 176, providing a very different platform to that of a normal healthy developing myocyte for t-tubule formation, which may affect t-tubule organisation.

3.3.3. Recovered atrial t-tubules can trigger synchronous calcium release.

Results from this chapter indicate that recovered atrial t-tubules were able to restore triggered central calcium release and with it the synchronous rise in systolic calcium. These findings are consistent with those in the ventricle that showed improved synchronicity of calcium release following t-tubule recovery80, 82, 84, 88. Despite t- tubule disorder and the occurrence of more longitudinally orientated tubules in this

112

model, it is not surprising that we observed recovery of systolic calcium release following tubule restoration. Although, there has been shown to be less LTCC and RyR localisation at longitudinal t-tubules compared with transverse5, 41, both types of dyads have been shown to trigger rapid calcium release41. It is therefore more likely that dys-synchronous calcium release associated with heart failure is a consequence of t-tubule ‘gaps’, rather than disorder, as has been previously demonstrated by Louch et al48. Alongside t-tubule gaps, RyR fragmentation could also contribute to the dys-synchrony of calcium release177.

Alongside calcium synchronicity, delayed calcium release has also previously been linked to ‘gaps’ in the t-tubule network2, 41, 42, 48, 65, such that, in areas devoid of t- tubules, calcium must propagate slowly to activate RyRs39. Consequently, Ibrahim et al84 suggested that irregular distribution of t-tubules observed in their heart failure model accounted for calcium handling disruptions. Moreover, the authors of this study suggested that the calcium handling improvements they observed in their recovered hearts, following unloading, was due to the reappearance of t-tubules by improved coupling of RyRs to LTCC. This is in agreement with our previously published work showing that in control sheep atrial myocytes, the calcium transient was rapid to rise at points in the interior of the cell near t-tubules and delayed calcium release observed in failing myocytes correlated to areas of the cell absent of t-tubules42. Consistent with these studies, our current study provides further evidence that t-tubules are the main site for the rapid rise of calcium in atrial myocytes and shows for the first time that recovered atrial t-tubules were functional.

The finding that the systolic rise of calcium occurred first along the t-tubules in the recovery atrial cells indicates that functional LTCC were present on these structures. T-tubules have long been considered as a specialist site for calcium entry in the myocyte. In the ventricle, L-type calcium current predominately enters the cell through sites located on the t-tubule membrane3, and in the atria, distinct subpopulations of LTCC exist giving rise to higher amplitude and open probability in the t-tubules45. As with the ventricle, a number of studies have reported decreased atrial L-type calcium current in disease49, 154, 178, which was responsible for a reduction in the atrial systolic calcium transient154. A likely factor of L-type calcium

113

current reduction is t-tubule loss49. Loss of t-tubules has been shown to result in cardiac dyad disruption65 which led to asynchronous calcium release2, 48, 50, 61, 65, due to a decreased co-localisation of LTCC and RyRs. This data suggests that there would also likely to be changes in LTCC expression or distribution in our recovery model. Unfortunately due to unspecific antibody binding it was not possible to investigate the expression or distribution of LTCC in the sheep. Despite this, NCX was re-localised to the t-tubules in recovery cells indicating that ion channels can be restored to recovered t-tubules.

Although we were unable to measure LTCC distribution in our model, immunostaining studies demonstrated restoration of co-localisation between the RyR and t-tubules in recovered atrial myocytes. Improved co-localisation between RyR and t-tubules has previously been shown in ventricular recovery studies69, 84, 88, which not only improved dyad coupling and calcium handling84, but has been correlated to recovered cardiac function69, 88. Furthermore, when RyRs were not able to co-localise with t-tubules in the heart failure myocytes in this study, RyR co- localised with the surface membrane. As surface RyR was not observed in control or recovery myocytes, this finding suggests redistribution of RyRs from the z-line to the surface sarcolemma in heart failure, thus allowing triggered calcium release to occur at the surface as observed in this study. The data from this study therefore suggests that recovered atrial t-tubules led to improved dyad coupling which in turn improved synchronicity of calcium release.

Another possible contributing factor to the restoration of synchronous calcium release in the recovered myocytes is the expression of the protein BIN1. Alongside t- tubule biogenesis, BIN1 is also responsible for localising L-type calcium current to the t-tubule through its association with microtubules104. In this study there was an association between BIN1 expression and the alterations to calcium handling in heart failure, suggesting a possible relationship between the two. Despite this, following recovery, BIN1 levels were not different to control or heart failure suggesting the involvement of other factors in t-tubule restoration.

114

3.3.4. Recovery of proteins associated with t-tubule biogenesis

In this sheep model of heart failure and recovery, atrial t-tubule restoration was associated with recovery of BIN1, Tcap and MTM1, but not JPH2. Previous work indicates BIN1, as the most likely candidate for tubule restoration given that overexpression of BIN1 has been shown to drive tubule formation in cells lacking tubules83, 90, 104, 107, 112 and knockdown caused t-tubule alterations46, 102, 103. Furthermore, we and others have shown in the ventricle, BIN1 loss can be associated with t-tubule loss in heart failure46, 80, 112 and recovery following SERCA gene therapy80 and PDE5 inhibition83. In the atria however, whilst BIN1 decreased in heart failure, expression levels were only partially restored, suggesting the contribution of other proteins. MTM1, a binding partner and regulator of BIN1121 was decreased in heart failure and fully restored to control levels following t-tubule recovery. Although this is in contrast to our results in the ventricle where we observed an increase in MTM1 following heart failure83, we proposed that this increase was responsible for the lateralisation of t-tubules observed in the ventricle. Particularly since MTM1 is crucial for t-tubule organisation in skeletal muscle121. Furthermore, decreased MTM1 expression has been linked to t-tubule loss98, 120 and disrupted EC coupling machinery120 in skeletal muscle, further suggesting a role for this protein in t-tubule regulation. The above data suggests that BIN1, Tcap and MTM1 could all play a role in the restoration of atrial t-tubules, the next chapter of this Thesis will aim to establish a causative role.

There is some controversy surrounding the role of the SR – t-tubule coupling protein JPH2 and t-tubule remodelling in heart failure. Whilst knockdown of JPH2 prevented maturation of t-tubules in embryonic mice, resulting in heart failure125, the expression levels of JPH2 following heart failure are somewhat debated46, 50, 81, 173. These inconsistences may reflect differences between models, species and severity of disease or suggest another role for JPH2. Nevertheless, the majority of studies suggest that JPH2 is not required for ventricular t-tubule recovery80, 83, 84, which is consistent with our finding that changes in atrial t-tubule density in heart failure or recovery were not reliant on JPH2. Further supporting the concept of a load-sensitive t-tubule network in our model, the levels of the load dependent protein Tcap,

115

correlated with t-tubule density in the atria. In agreement with our study, decreased expression of Tcap in the ventricle has been associated with t-tubule loss and increased expression associated with t-tubule recovery following SERCA gene therapy80. Moreover, Tcap KO led to t-tubule abnormalities during development and progressive loss of t-tubules and calcium handling abnormalities following mechanical overload84. Tcap therefore could regulate the t-tubule system in response to load.

3.3.5. Study limitations

Whilst there were major advantages of using a large animal model in our study, there were some limitations of using sheep. One of the foremost limitations is the anatomy of the sheep. Due to the position of the sternum over the cardiac apex we were unable to obtain a four-chambered apical view for a full echocardiographic assessment of atria function. As a result, ventricular function including, dilatation and contractility, was used as an in vivo marker of disease progression. Conversely, it is not known in this model if there were in vivo changes to atrial function as a result of rapid ventricular pacing or recovery. Nevertheless, we did observe changes at a cellular level consistent with decreased contraction in heart failure. Despite these limitations, the aim of this study was to characterise t-tubule recovery following the onset of end stage heart failure, thus all animals studied displayed clinical symptoms of end stage heart failure including dyspnoea, lethargy or weight loss. Other studies have reported in large animals that chronic ventricular rapid pacing led to dilation of both cardiac chambers179 suggesting that rapid ventricular pacing alters atrial function. In support of this, un-published data from our lab showed, in a separate cohort of animals, atrial dilatation in heart failure (personal communication with George Madders). Additionally, our results also showed atrial cell hypertrophy after rapid pacing, which is a marker of end stage heart failure, thus further indicating atrial remodelling as a result of rapid pacing.

A further limitation of not being able to obtain in vivo functional data from the sheep atria is that it was more difficult to monitor atrial recovery following cessation of pacing. Thus, for this study a pre-determined time point for recovery was set. Whilst

116

it is important to keep consistency when looking at the recovery time points, it is worth noting that our animals were not breed in house and may suffer from genetic differences. The considerable variability in the time it took for each animal to reach end stage heart failure 35.7 ± 5.6 (range 16 - 69) days, may be a reflection of differences between animals. It is conceivable therefore, that even though each animal was taken to the same point, to just before end stage heart failure, before recovery commenced, recovery time between animals would also vary. Despite this, the mean data suggests that the animals, a least in part, recovered and most notably for this study, t-tubule density was recovered at the time point used. Nevertheless, future work would benefit from the possibility of recovering animals for longer.

Although it is common to use isolated single myocytes to access t-tubule density, there is some concern that the enzymatic digestion used to isolate the cells may result in t-tubule loss50. Some groups have tried to overcome this problem by using either whole heart imaging46, 50 or fixed tissue sections stained with WGA43, 46 for t-tubule assessments. In our study, whole heart imaging was not an option as the sheep myocytes were being used for the functional studies discussed in this chapter, as well as t-tubule assessments. Whilst we have used WGA stained tissue sections in the past to represent t-tubule density43, 46, we have found no difference between non isolated (0.84 ± 0.03)43 and isolated (0.88 ± 0.04)42 sheep atrial myocyte half distance in our control cells. As this technique has its own set of problems, such as WGA staining of non-tubular intracellular structures180 and cell shrinkage caused by tissue fixation181, it was decided that using isolated cells was more beneficial to this study. Furthermore, any error is likely to be negligible and consistent between comparative groups as myocytes have been isolated the same way in all groups.

An important limitation is the lack of a commercially available viable LTCC antibody. It would have been interesting to investigate the possibility of a functional association between the recovered t-tubules and the LTCC in this model. LTCCs are predominantly concentrated in the t-tubule membrane3, therefore, it could be expected that changes in t-tubule density would be associated with changes in LTCC expression or distribution. It is also possible that the recovery of systolic calcium release and synchronicity observed in this study could be due to changes in LTCC

117

distribution. We have previously reported that the reduction in the atrial systolic calcium transient observed in heart failure was a consequence of reduced L-type calcium current154 possibly due to loss of t-tubules. More recent results from our laboratory suggest that recovered t-tubules, rather than recovery of other factors following termination of rapid ventricular pacing, were responsible for normalisation of systolic calcium in our recovery model. Since detubulation of the recovery cells reduced the calcium transient to heart failure levels (un-published personal communication with Dr. Jessica Clarke). Further investigation is needed to look at the relationship between the recovery t-tubules, the LTCC and other calcium handling proteins.

3.4. Conclusions

Atrial t-tubules were lost in a sheep model of heart failure; this was associated with dys-synchronous rise of systolic calcium. T-tubule associated proteins BIN1, MTM1 and Tcap also decreased during heart failure. Recovery of heart failure led to upregulation of these proteins and restoration of atrial tubules, which were once again the main site for the rise of systolic calcium. The next chapter will aim to investigate the mechanisms which may be responsible for t-tubule recovery.

118

4. Results: Proteins implicated in the formation and maintenance of t-tubules.

4.1. Introduction

The data presented in the last chapter indicates t-tubule restoration is an integral part of cardiac recovery. Little is known however, as to how t-tubules are able to recover. Evidence, in both the literature and first chapter of this study, points towards the proteins BIN1, Tcap and MTM1 as potential candidates as the levels of these proteins were correlated with the density of t-tubules46, 80, 84, 116, 120. These studies however, provide only an association with t-tubule density and protein abundance, therefore the next series of experiments aim to establish a causative link between these proteins and t-tubule restoration.

One way to investigate the mechanisms that may regulate t-tubule growth or recovery is by using protein-targeting techniques. Evidence shows that both knockdown and overexpression studies have been used with great success to assess the role of various proteins on the t-tubule system. For example, in HeLa cells, that lack t-tubules, overexpression of BIN1 was able to induce membrane curvature90, 104, suggesting the involvement of this protein in tubule formation. Furthermore, alongside tubule formation, Hong et al104 demonstrated that full length BIN1 expression in HEK-293 cells resulted in enhanced calcium handling, suggesting a functional association between BIN1 and calcium handling. Thus, these models allow characterisation of the role of specific genes and, in the above cases, highlight treatment targets for t-tubule development. Critically, in the study of t-tubule biogenesis, protein overexpression studies to date have predominately used immortalised cells lines. Whilst using cell lines have many advantages in that they are easy to control and can be studied for long periods of time, cell lines to not necessarily represent in vivo conditions. More recently, De La Mata et al107 used hESC-CMs to link BIN1 expression and tubule maturation in the heart and my own work showed development of tubules in NRVMs and iPSC-CMs following BIN1

119

expression83. Thus these recent studies provide, for the first time, a direct link between protein expression and t-tubule formation in cardiac cells.

4.1.1. Regulation of t-tubules by molecular mechanisms

To date, several molecules have been revealed to influence the structure of t-tubules. As I have already begun to discuss, of these mechanisms there is a growing body of evidence indicating a role for BIN1 in t-tubule regulation 90, 102, 104, 107, 112, and maintenance46 making this molecule a very likely candidate for t-tubule restoration. Alongside the association between BIN1 expression and t-tubule density in our model, we also observed a similar trend with the BIN1 binding partner MTM1. Interestingly, whilst no known studies have been conducted on cardiac MTM1, in skeletal muscle, MTM1 co-expression enhanced BIN1 driven membrane tubulation121. Furthermore, knockdown of MTM1 in skeletal muscle in both mice and Zebrafish led to abnormally orientated tubules 116, 120, 182. Although Tcap expression levels in our model of recovery, and others80, correlated with t-tubule density, there is less evidence to suggest a direct involvement of this protein in t- tubule formation. Tcap knockout in both zebrafish137 and mice84 showed altered t- tubule development, but to our knowledge there are no studies determining if Tcap provides a mechanism for t-tubule biogenesis.

4.1.2. T-tubule development

To gain a greater understanding of the factors that govern t-tubule restoration, it may be possible to draw similarities from what is known in the developing heart. In small mammals t-tubules develop postnatally, but this is not the case in large mammals where tubules are thought to begin to develop in utero in both the human and sheep183, 184. The reasons for these differences are mostly unknown, but could be down to the transition from hyperplasia to hypertrophic growth occurring prior to birth in large mammals. Thus, hypertrophic cell growth could be the trigger for t- tubule development, as wider cells are known to require t-tubules43. Additionally, around the time this transition occurs, thyroid and glucocorticoid hormones, which

120

are critical for cardiac maturation and hypertrophic growth during development, levels surge185, 186. Interestingly, thyroid and glucocorticoid hormone treatment promoted t-tubule development in human iPSC-CM187, suggesting that this surge could initiate t-tubule growth. Furthermore, BIN1 localisation became much more prominent throughout human iPSC-CMs following thyroid and glucocorticoid treatment. In terms of proteins associated with t-tubule development, much less is known. Deletion of BIN1 in mice resulted in perinatal lethality as a result of the embryos developing severe cardiomyopathy111 and thus the effects on t-tubule development could not be determined. There is one study however by Reynolds et al125 that demonstrated BIN1 expression increased during development in mice in line with t-tubule development. Whilst not investigated in this part of the study, due to the results from the previous Chapter having found no link between t-tubule recovery and JPH2, like BIN1, JPH2 protein expression has also be associated with t-tubule development40, 124, 125, 187.

4.1.3. Aims of the chapter

The purpose of this chapter is to identify proteins involved in the formation and maintenance of t-tubules. In doing so, therapeutic targets for the recovery of cardiac t-tubules could be discovered. By investigating t-tubule proteins associated with t- tubule recovery, as seen in the previous chapter, via overexpression in cells that usually lack t-tubules, a potential mechanism for t-tubule growth might be determined. Thus, this chapter aims to address the following:

1. Are BIN1, Tcap or MTM1 implicated in the formation and maintenance of t- tubules in neonatal rat ventricular myocytes (NRVMs)? 2. Do these proteins interact in the heart to influence t-tubule formation and shape?

121

4.2. Results

4.2.1. Over expression of t-tubule associated proteins in NRVMs

To examine the effect of different t-tubule associate proteins on t-tubule development, NRVMs were transfected with vectors expressing BIN1, MTM1 & Tcap with a fluorescent tag (Methods section 2.5 for details). A vector containing only a fluorescent tag (either mKate2 or mGFP) was used as negative control. Tubule formation was assessed qualitatively using the membrane dye WGA 647 and extracellular dye Oregon Green 488 and imaged using confocal microscopy (Methods section 2.7 for details).

4.2.1.1. Expression of Amphiphysin II (BIN1)

Confirmation of BIN1 overexpression in the transfected cells is shown in Figure 4.1A. A representative blot shows BIN1 was expressed in very low quantities in non- transfected (NT) and mKate2 only transfected NRVMs (Figure 4.1Ai). Following transfection with BIN1 variant 8, two BIN1 protein bands were observed in the NRVMs. BIN1 protein was up-regulated compared with non-transfected myocytes and the mKate2 empty vector, confirming successful transfection (Figure 4.1Aii).

Figure 4.1 Bi shows representative images of transfected NRVMs compiled from confocal z-sections from the central 2µm of the cell depth. Cells transfected with a vector expressing mKate2 only (control) showed diffuse cytosolic fluorescence and no t-tubule staining. In comparison, NRVMs transfected with vectors expressing BIN1-mKate2 showed intracellular tubule like structures. These structures were abundant, albeit disorganised, in BIN1 transfected cells. Figure 4.1 Bii shows that the extracellular fluorescent indicator, Oregon Green 488, entered these structures and co-localised with BIN1-mKate2 fluorescence, demonstrating that these BIN1 driven structures were in fact tubular structures that were patent at and connected to the extracellular environment.

122

Figure 4.1. BIN1 drives tubule formation in NRVMs. (A) Representative western blots (i) and mean data (ii) summarising protein expression of BIN1 in non- transfected (NT), mKate2 only and BIN1-mKate2 NRVMs. Protein was normalised to total protein. BIN1 protein expression was upregulated following transient transfection. N=3 isolations. (B) Representative images of NRVMs transfected with mKate2 empty vector (control) and BIN1 expression vector (i). Oregon Green 488 staining of BIN1 transfected cell and merge with the mKate2 channel from (i) showing co-localisation (ii). *** p<0.001 vs BIN1, tested by linear mixed model analysis, presented as mean ± SEM. Scale bars =10µm.

123

4.2.1.2. Expression of Telethonin (Tcap)

Protein expression of Tcap in NRVMs is shown in Figure 4.2. In non-transfected myocytes a single protein band of the reported molecular weight 19kDA of Tcap was detected. Tagging Tcap with mGFP altered the molecular weight of the protein, thus in the Tcap-mGFP tagged cells a protein band of the combined molecular weights ~46kDA was detected (Ai). Tcap expression was increased in Tcap-mGFP transfected myocytes when compared with non-transfected myocytes and the mGFP control vector, confirming overexpression.

Representative images of NRVMs transfected with the control vector expressing mGFP only and the expression vector Tcap-mGFP are presented in Figure 4.2 Bi. As previously shown, transfection with a control vector (this time expressing GFP only) in NRVMs displayed diffuse cytosolic fluorescence and no tubule staining. A similar cytosolic fluorescence pattern was observed in a large number of NRVMs transfected with vectors expressing Tcap-mGFP. Conversely, some myocytes transfected with Tcap-mGFP displayed both cytosolic fluorescence and intracellular fluorescence corresponding to the z-lines (Bi). Figure 4.2 Bii shows that the membrane and tubule marker WGA 647 stained the surface sarcolemma only in the Tcap-mGFP transfected cells indicating no tubule structures were present. In addition to WGA staining, cells tagged with Tcap-mKate2 (a different tag was used to ensure no cross over of fluorescent channels) showed no co-localisation with the extracellular fluorescent indicator Oregon Green 488 suggesting that the fluorescence staining observed at the z-lines in the Tcap transfected cells was not connected to the surface sarcolemma (Figure 4.2 C).

124

Figure 4.2. Overexpression of Tcap in NRVMs. (A) Representative immuno blots (i) and mean data (ii) showing expression of Tcap in non-transfected (NT), mGFP and Tcap-mGFP NRVMs. Expression is normalised to total protein. Following transfection Tcap expression was upregulated (ii). N=3 isolations. (B) Confocal images of NRVMs transfected with mGFP control vector and Tcap-mGFP expression vector (i). (Bii) WGA 647 staining of NRVMs from (i) and the overlay of WGA with the Tcap-mGFP vector. (C) Oregon Green 488 staining of a Tcap- mKate2 transfected cell and merge with the mKate2 channel showing no co- localisation. *** p<0.001 vs Tcap, tested by linear mixed model analysis, presented as mean ± SEM. Scale bars =10µm.

125

4.2.1.3. Expression of Myotubularin 1 (MTM1)

Overexpression of MTM1 following MTM1-mGFP transfection in NRVMs was confirmed using western blot as shown in Figure 4.3. A single MTM1 protein band was detected corresponding to the reported molecular weight 55kDA (Figure 4.3 Ai). MTM1 expression increased compared with non-transfected myocytes and the control vector mGFP, confirming successful transfection (Figure 4.3 Aii).

Figure 4.3 Bi shows representative confocal images of NRVMs transfected with the control vector expressing mGFP only and the expression vector MTM1-mGFP. Neither the empty vector nor MTM1-mGFP over expression produced tubule like structures, with both vectors showing cytosolic fluorescence. Thus staining with WGA 647 led to surface sarcolemma staining only and there was no co-localisation between WGA and MTM1-mGFP (Figure 4.3 Bii). Staining of MTM1-mKate2 transfected cells with the extracellular fluorescent indicator Oregon Green 488, did not lead to any intracellular staining (Figure 4.3 C), confirming that no structures were patent to the extracellular environment.

126

Figure 4.3. Overexpression of MTM1 alone is not required for tubule formation in NRVMs. (A) Representative immuno blots (i) and mean data (ii) summarising protein expression of MTM1 in non-transfected (NT), mGFP only and MTM1- mGFP NRVMs. (Aii) Following transfection MTM1 protein expression was upregulated. N=3 isolations. (Bi) Representative confocal images of NRVMs transfected with mGFP empty vector (control) and MTM1-mGFP expression vector. (Bii) WGA 647 staining of NRVMs from (i) and the overlay of WGA with the MTM1-mGFP vector. (C) Oregon Green 488 staining of a MTM1-mKate2 transfected cell and merge showing no co-localisation. * p<0.05 vs MTM1, tested by linear mixed model analysis, presented as mean ± SEM. Scale bars =10µm.

127

4.2.2. Co-expression of t-tubule associated proteins alters BIN1 induced t-tubule structures.

The first section of this chapter demonstrated that over expression of BIN1 in NRVMs caused tubule like structures to develop, but tubules were absent in cells expressing Tcap or MTM1 alone. This suggests that BIN1 is required for the formation of t-tubules in the heart. Based on data from Chapter 3 however, it is unlikely to just be BIN1 that is involved in the recovery of tubules. The next aim therefore, was to determine if Tcap and MTM1 could alter BIN1 induced tubule structures. NRVMs co-expressing BIN1 with Tcap, MTM1 or both were used to establish if either Tcap or MTM1 altered the density or structure of BIN1 driven tubules.

4.2.2.1. Co-expression of proteins in NRVMs

Representative images of co-transfected NRVMs are presented in Figure 4.4 A. The representative cell in Figure 4.4 panel Ai shows successful co-transfection of BIN1- mKate2 (red) and MTM1-mGFP (green); co-localisation of the two channels is represented in yellow. Consistent with the data presented in Figure 4.1, the cells expressing BIN1-mKate2 showed intracellular tubule like structures. Conversely to the data presented in Figure 4.3 where MTM1-mGFP expression was cytosolic, Figure 4.4 panel Ai shows that following co-expression with BIN1-mKate2, the cellular distribution of MTM1-mGFP was redistributed. To the extent that, consistent with MTM1 being a binding partner of BIN1121, 122, there was a strong co- localisation between MTM1-mGFP and BIN1-mKate2 (Mander’s overlap 0.71 ± 0.03) (Figure 4.4 B). Furthermore, co-expression appeared to have altered the BIN1- mKate2 driven tubules with the structure appearing longer and more organised compared to transfection with BIN1-mKate2 alone.

Figure 4.4 panel Aii shows representative images of a NRVM transfected with vectors expressing both BIN1-mKate2 (red) and Tcap-mGFP (green). A noticeable observation is that whilst the distribution of Tcap-mGFP appears unchanged to the

128

data presented in Figure 4.2, the structure of BIN1-mKate2 driven tubules was almost completely disrupted following dual transfection. Furthermore, there was 58.7 ± 16% less co-localisation between BIN1-mKate2 and Tcap-mGFP (Mander’s overlap 0.35 ± 0.03), shown in yellow on panel Aii and in the mean data on Figure 4.4 B when compared with BIN1-mKate2 and MTM1-mGFP.

To determine the combined effect of Tcap and MTM1 on BIN1 driven tubules, all three proteins were transfected into NRVMs simultaneously. Expression of BIN1- mKate2 (red) with both Tcap-mBFP (blue) and MTM1-mGFP (green) vectors in NRVMs is represented in Figure 4.4 C. The images show that the presence of Tcap- mBFP (blue) in the transfection; did not alter the co-localisation between BIN1- mKate2 (red) and MTM1-mGFP (green). Furthermore, the BIN1-mKate2 expression vector produced intracellular tubule like structures that had a similar appearance to the BIN1-mKate2 and MTM1-mGFP tubules from panel Ai.

To establish if t-tubule characteristics were altered following co-transfection, the density of the tubule network was quantified by determining the fraction of the cell occupied by tubules (Figure 4.4 Di). The summary data in Figure 4.4 Di demonstrates that BIN1-mKate2 expression alone in NRVMs led to an increase in the fractional area of cells occupied by tubules compared with the control vector. Compared with BIN1-mKate2 only, co-expression with Tcap-mGFP decreased tubule density by 28.6 ± 5% whilst co-expression with MTM1-mGFP increased tubule density by 14.1 ± 6 %. Triple transfection resulted in increased tubule density compared to BIN1 alone and BIN1 with Tcap. There was no difference when comparing the triple transfection with BIN1 co-expressed with MTM1, suggesting the MTM1 effect on BIN1 induced tubule density predominates, with or without the expression of Tcap.

To determine if Tcap or MTM1 could alter the structure of BIN1 driven tubules in line with the longer branched tubules seen in the recovery atria (Chapter 3), tubule structural alterations were calculated using the ‘Analyse Skeleton’ algorithm in ImageJ. Firstly, the average length of a single tubule structure (skeleton length) was calculated. The summary data in Figure 4.4 Dii shows that the average skeleton

129

length of BIN1-mKate2 induced tubules was 4.8 ± 0.3 µm. Skeleton length decreased to 3.5 ± 0.3 µm following dual expression with Tcap-mGFP, but increased to 7.2 ± 0.4 µm with MTM1 co-expression and 8.0 ± 0.4 µm with triple transfection.

Finally, the extent of tubule branching was also calculated using the ‘Analyse Skeleton’ algorithm. Figure 4.4 Ciii shows that co-expression of Tcap-mGFP with BIN1-mKate2 decreased tubule branching by 14.8 ± 7 % compared with BIN1- mKate2 alone. Tubule branching was increased by 44.4 ± 12% following transfection of MTM1-mGFP with BIN1-mKate2 and by 48.9 ± 12 % following triple transfection compared to BIN1-mKate2, suggesting MTM1 plays a role in increasing branching in tubules.

130

Figure 4.4. Co-expression of BIN, Tcap and MTM1 in NRVMs. (A) Representative confocal images of NRVMs co-transfected with BIN1 (Red) with either MTM1 (Green) (i) or Tcap (Green) (ii). Yellow represents co-localisation between proteins. (B) Mean data showing Manders overlap co-localisation of BIN1 with MTM1 or Tcap in NRVMs. (C) Triple transfection of BIN1 (Red), Tcap (Blue) and MTM1 (Green) in NRVMs. Yellow represents co-localisation. (D) Mean data summarising fractional area of cells occupied by tubules (i), average skeleton length (ii) and average number of branches in each structure (iii). n=5-8 isolations (mKate2, 45 cells; BIN1, 79 cells; BIN1&Tcap, 98 cells; BIN1&MTM1, 84 cells; Triple, 76 cells). $ p<0.05, $$ p<0.01, $$$ p<0.001 vs BIN1; † p<0.001 vs BIN1 & Tcap, tested by linear mixed model analysis, mean ± SEM. Scale bars =10µm.

131

4.2.2.2. BIN1 is required for tubule formation.

Next, to determine if MTM1 could modify Tcap, and vice versa, in cardiac cells, these proteins were co-expressed in the absence of BIN1. Figure 4.5 shows no change to the distribution of either protein compared to when they were expressed on their own. Both proteins had a predominately cytosolic distribution, as such; accurate co-localisation assessments could not be made.

Figure 4.5. BIN1 is required for tubule formation. (A) Representative confocal images of NRVMs transfected with MTM1 (Green) and Tcap (Blue). Yellow/White represents co-localisation. Lower panel demonstrates that not all cells are successfully transfected with both vectors during dual transfection. Scale bars =10µm.

132

4.2.2.3. Fluorescent tag positioning alters tubule formation.

Finally, to determine if the positioning of the fluorescent tag on the expression vector interferes with co-transfection, BIN1 was co-expression with either C-terminal or N- terminal fluorescently tagged MTM1 in NRVMs. Consistent with the data presented in Figure 4.4, cells expressing BIN1-mKate2 with AN-MTM1-mGFP showed a strong co-localisation (Figure 4.6 Ai). Conversely, co-expression of BIN1-mKate2 with AC-MTM1-mGFP showed mostly cytosolic MTM1 expression and little co- localisation with BIN1 (Figure 4.6 Aii). To establish if fluorescent tag positioning alters tubule density, fractional area analysis was measured (Figure 4.6 B). As already discussed, Figure 4.6 B demonstrates that compared with BIN1-mKate2 transfection, co-expression with AN-MTM1-mGFP increased tubule density. In comparison, co-expression of BIN1 with AC-MTM1-mGFP decreased tubule density compared with both BIN1-mKate2 only and BIN1-mKate2 with AN-MTM1- mGFP (12.2 ± 6 % and 25.9 ± 8% respectively), suggesting that the position of the fluorescent tag on the vector alters protein interactions during co- transfection.

133

Figure 4.6. BIN1 & MTM1 tubule formation with different fluorescent tag positioning. (A) Representative images of NRVMs transfected with vectors expressing BIN1-mKate2 (Red) with either AN-MTM1 (i) or AC-MTM1 (ii) (Green). Yellow represents co-localisation. (B) Mean data summarising fractional area of cells occupied by tubules in BIN1 only and dual transfected cells. n=3-6 isolations (BIN1, 66 cells; BIN1& AN-MTM1, 54 cells; BIN1& AC-MTM1, 31 cells). $ p<0.05 vs BIN1; ˄˄˄ p<0.001 vs AC-MTM1, tested by linear mixed model analysis, presented as mean ± SEM. Scale bars =10µm.

134

4.3. Discussion

The major findings in this chapter have shown that expression of BIN1 in NRVMs led to tubule formation. Whilst expression of MTM1 and Tcap alone did not produce tubules, co-expression of both these proteins altered the density and structure of BIN1 driven tubules. The enhanced effect of MTM1 on BIN1 driven tubule structures existed with or without the presence of Tcap indicating that MTM1 could play a role in increasing tubule density and structure in the heart.

4.3.1. Transfection with BIN1 led to the development of tubules in NRVMs.

The aim of this chapter was to determine which proteins were sufficient to drive tubule formation in NRVMs and thus identify potential mechanisms for t-tubule restoration. Consistent with previous findings83, 188, we have shown that NRVMs (prior to transfection) completely lacked a tubule network. We observed however, that transient transfection with BIN1 led to the development of tubule like structures in NRVMs, suggesting that BIN1 plays a role in the formation of tubules. Whilst not much is known about the role of BIN1 in cardiac muscle, this finding is in agreement with our recently published work showing BIN1 was able to drive tubule formation in NRVMs (Appendix)83. Furthermore, these findings are consistent with work from several groups that have previously reported overexpression of BIN1 in other cell types induced tubule formation90, 104, 107. Formation of tubules by BIN1 has previously been shown to be dependent on the presence of the BAR domain region of the protein that ensures targeting to and curvature of the membrane113, 189. Furthermore, in skeletal muscle, a splice variant (variant 8) containing exon 11 encodes a phosphoinositide binding domain that targets BIN1 to the membrane for the formation of tubules90. We found that overexpression of this variant in NRVMs also led to the formation of tubules. One noticeable feature of the BIN1 induced tubules however, was that they did not have a well ordered sarcomeric distribution as seen in adult myocytes. It is possible that an alternative variant of BIN1, not encoding exon 11, is required for tubule organisation. In support of this, Hong et

135

al102 recently discovered a cardiac alternatively spliced variant of BIN1 (cBIN1) that they propose plays a crucial role in the organisation of t-tubule membrane microfolds in mice. Despite this, work exploring the roles of the different BIN1 variants in the heart is limited, and since, global KO of BIN1 was embryonically lethal due to cardiac hypertrophy, at a stage before t-tubules developed, it is likely BIN1 is also required for functions that are unrelated to t-tubule biogenesis. The next chapter of this study will aim to define the different of roles of the BIN1 isoforms in the heart.

Much less is known regarding the role of Tcap and MTM1 in the biogenesis of tubules and notably, in this study, tubules were absent in cells expressing Tcap or MTM1 alone. We have shown here, that in agreement with previous work, Tcap was present early in development (2 days post birth) in NRVMs190. Moreover, Tcap appeared to be located at the z-lines in NRVMs, prior to the formation of tubules, which suggests that this protein is not required for membrane tubulation. It is possible however that Tcap regulates t-tubules in other ways. Studies in Zebrafish showed that enhanced stretch force was able to induce Tcap expression and blocking either of these factors disrupted t-tubule development137. Thus, studying Tcap in vitro, without the influence of stretch force is unlikely to be physiologically relevant and might be a major drawback with this model. Tcap also binds to t-tubule associated protein minK135 and may act on t-tubules through interactions with this protein. It remains to be determined when during development minK is expressed and if the necessary interactions were present in this model of NRVM transfection.

This study shows for the first time that MTM1 was present early in development in NRVMs and overexpressing this protein did not lead to the development of tubular structures. As MTM1 has not yet been studied in relation to cardiac muscle, it is difficult to interpret this result. MTM1 deficient mice experienced some degree of embryonic lethality and all animals developed a fatal myopathy at approx. 4 weeks, suggesting that MTM1 is an important mechanism for muscle development191. Furthermore, t-tubule structural defects were observed in MTM1 deficient muscle fibres from Zebrafish, mice and human98, 116, 120, 182. Interestingly however, in immature muscle cells, knock down of MTM1 had no effect on BIN1 driven tubules; it was only during muscle maturation that MTM1 was required for tubule positioning

136

through interactions with BIN1121. These findings suggest a role for MTM1 in the regulation of BIN1 membrane structures.

4.3.2. Co-expression with Tcap and MTM1 altered the expression of BIN1 induced tubules.

The results from this chapter indicate that co-expression of BIN1 with Tcap, MTM1 or both, altered the density and structure of BIN1 induced tubules. Although BIN1 and Tcap were both associated with tubule restoration following recovery from heart failure (Chapter 3 and80, 83, 84), this study suggests that these proteins do not cooperate during tubule formation. Little co-localisation between BIN1 and Tcap was observed following co-transfection. Tcap expression was mostly cytosolic, with some sarcomeric expression, from early in development and the distribution did not alter following BIN1 transfection. It is possible that BIN1 and Tcap have more co- localisation in the fully developed heart where t-tubules are aligned at the sarcomeres1, but this was not explored in this study. Preliminary co- immunoprecipitation data from our lab, to determine BIN1 interactions, showed that Tcap was not pulled down with BIN1 (Dr Rebecca Taylor, un-published) suggesting this is not the case however. It is even conceivable that Tcap negatively inhibits BIN1, suppressing tubule formation, as co-expression of BIN1 and Tcap led to decreased tubule density, skeleton length and tubule branching. Nevertheless, it is important to consider, that Tcap is a load sensing molecule and without the influence of load, the impact this molecule has on the t-tubule system cannot be fully elucidated.

Consistent with it being a binding partner121, MTM1 co-localised with BIN1 at tubules following co-expression. This co-expression also led to increased tubule density and longer more branched tubules compared to BIN1 alone. Interactions between MTM1 and BIN1 have previously been demonstrated in skeletal muscle where co-expression of MTM1 altered BIN1 tubulation, leading to longer tubules compared to just BIN1 expression121. Taken together, these data suggest that MTM1 enhances membrane tubulation of BIN1 in muscle. MTM1 has shown to have stronger interactions with the BAR and SH3 domains of BIN1, thus interactions are

137

thought to be dependent on BIN1 conformation. Such that, mutations in the SH3 domain of BIN1, have been shown to impact on membrane remodeling by MTM1121.

Results from the previous chapter of this thesis show in the sheep model of recovery from heart failure that increased tubule density and longer branched tubules during recovery were associated with both increased Tcap and MTM1 expression. Despite this, no tubule differences were observed between BIN1 with MTM1 co-transfected cells and triple (BIN1 and MTM1 expressed with Tcap) transfected cells. These results suggest that the effect of MTM1 on BIN1 induced tubules outweighs any effect had by Tcap and may underlie the increase in tubule density and length of tubules upon recovery from heart failure in the atria. It remains to be understood however, how these mechanisms work together to maintain the t-tubule network and in turn how these proteins are regulated.

4.3.3. Other factors regulating t-tubule recovery

In this thesis I have so far considered the role of BIN1, Tcap, MTM1 and to some extent JPH2, in the control and restoration of t-tubules, there are however likely to be several other mechanisms involved. One possibility is that cell signalling pathways play a role in controlling the t-tubule system. Phosphoinositide 3-kinases (PI3Ks) are signalling proteins that regulate cardiac myocyte growth192 and maintain t-tubule structure97 through phosphorylation of membrane lipids phosphatidylinositol 4,5- bisphosphate (PIP2). Phosphorylation of PIP2 generates second messenger phosphatidylinositol trisphosphate (PIP3), which enhances protein targeting to the t- 96 tubule . As previously discussed, BAR domains such as BIN1 bind to PIP2 lipids in the cell membrane and it is thought that PI3Ks regulate these molecules by increasing the binding capacity to the t-tubule. PIP2 also interacts with Dysferlin, another protein recently implicated in t-tubule maintenance in skeletal muscle193. Dysferlin has been shown to localise to developing tubules and interact with proteins in the t-tubule membrane194, 195. Furthermore, transfection of Dysferlin in non- muscle cells leads to tubule formation193, a process that may be regulated by PI3Ks. Interestingly, PI3K is known to be regulated by MTM198, 117, a protein that in turn interacts with and regulates BIN1. PI3Ks have also been shown to play a role in

138

maintaining the t-tubule network through interactions with JPH2. Wu et al97, demonstrated that deletion of P13K resulted in t-tubule disruption, JPH2 mis- localisation and reduced myocyte contraction. This may have resulted from deletion of P13K preventing membrane lipid phosphorylation of PIP3, thus reducing the binding capacity of JPH2 to the t-tubule. These data suggests that other mechanisms and pathways, alongside the ones studied here, are required for t-tubule development and that further research is needed to fully interpret control of this system.

4.3.4. Study limitations

Using fluorescent tags on transfection vectors has advantages, in that it enables successfully transfected cells to be identified easily. It is possible however, that the tag may interfere with a protein’s structure and function (reviewed in156). Experiments in the next chapter of this thesis show that BIN1 transfection was able to drive tubule formation with or without a fluorescent tag, suggesting no interference on protein function when carrying out this single transfection. Despite this, the relationship, if any, between BIN1, Tcap & MTM1 in the heart is largely unknown and my preliminary work (Figure 4.6) suggests that the position of the fluorescent tags on the vector interferes with these protein interactions during co- transfection. For example, the work presented here shows that the effects of N- terminal fluorescent tagged MTM1 on BIN1 driven tubules were not replicated with C-terminal fluorescent tagging of MTM1, suggesting that C-terminal fluorescent tagging of MTM1 alters interactions with BIN1. Literature does indicate, in other cell types, that MTM1 has a binding domain at its C-terminus region important for protein and membrane interactions157, therefore blocking this site with a fluorescent tag could alter this binding capacity. Despite this, how BIN1 and MTM1 interact in cardiac muscle is largely unknown and it is out of the scope of this study to determine.

139

4.4. Conclusions

In conclusion, the data in this chapter has demonstrated, tubule formation can be associated with BIN1 expression but not Tcap or MTM1 expression. In addition, this study has shown, for the first time in the heart that BIN1 cooperates with MTM1 and interactions between the two caused alterations to both t-tubule density and structure. We therefore suggest increased MTM1 in the recovered atria is important for the increase in t-tubule density, length and branching and may therefore represent an important new therapeutic target for the recovery of cardiac t-tubules.

140

5. Results: The control of t-tubule formation by Amphiphysin II (BIN1)

5.1. Introduction

Our previous work46, 83 and the data presented so far in this thesis has shown an association between reductions in BIN1 protein and t-tubule density in an ovine model of rapid pacing induced heart failure. Additionally, recovery from heart failure was associated with a restoration of t-tubules and BIN1 protein levels. The previous chapter of this thesis established that BIN1 variant 8 isoform was capable of driving the formation of tubular structures in neonatal rat ventricular myocytes (NRVMs). Despite this, our understanding of how t-tubules form in the heart and the role of BIN1 remains limited.

5.1.1. Cardiac variants of BIN1

BIN1 protein is ubiquitously expressed and is most highly expressed in brain and skeletal muscle101, 196. The human BIN1 gene consists of 20 exons that can be spliced into multiple isoforms (Figure 5.1), each with tissue specificity. At least 10 alternatively spliced isoforms have been identified, each playing key roles in cellular functions90, 101, 197, 198. Isoforms 1–7 are predominately expressed in brain, isoform 8 is muscle specific and isoforms 9 and 10 are ubiquitously expressed198. Each BIN1 isoform is made up of protein domains specific to its function; the commonly expressed domains include a N-BAR domain that senses membrane curvature (encoded by exons 1-10) and the Src homology 3 (SH3) domain (encoded by exons 19 - 20)90, 101, 198. The remaining exons are variably expressed and make up the clathrin and AP2 (CLAP) binding domain (exons 13–16) and the tumour suppressor Myc-binding domain (MBD) (exons 17-18)198 . The muscle specific exon 11 (previously known as exon 1090) encodes the phosphoinositide (PI) binding domain that targets BIN1 to the membrane in skeletal muscle90, 101, 198. Exon 11 has only been found in the variant 4 and variant 8 isoforms of BIN1198 (Figure 5.1).

141

Figure 5.1. BIN1 isoforms. (A) Schematic diagram of BIN1 exons based on sheep (accession XM_012147686.2) and human (accession NM_139343.2) sequences83. BAR, Bin-Amphiphysin-Rvs domain; PI, phosphoinositide binding domain; CLAP, clathrin-AP2 domain; MBD, myc-binding domain; SH3, SRC homology domain. (B) Spliced isoforms of BIN1. The work in the chapter will focus on BIN1 variants 5, 8 and 9. Diagram adapted from83

142

The importance of the muscle isoform, variant 8, in t-tubule biogenesis and organisation in skeletal muscle has been well documented90, 113, 182, 199, 200. In particular, several groups have shown overexpression of BIN1 variant 8 led to tubulation whereas overexpression of BIN1 without exon 11 did not90, 108, 114, 199, 201. Skipping of BIN1 exon 11 in dogs and humans has been reported in several muscle dystrophies that involve t-tubule defects and muscle weakness199, 201. Notably, the exon 11 skipping DM mutation of BIN1 resulted in the variant 9 isoform which was incapable of tubulating membranes in skeletal muscle201. Surprisingly, BIN1 KO mice, did not exhibit skeletal muscle abnormalities, but resulted in perinatal lethality associated with onset of a severe hypertrophic cardiomyopathy111 indicating a critical role for BIN1 in the heart. A cardiac conditional KO has since been shown to lead to a reduction in t-tubule density with t-tubule dilatation in both the heterozygote102 and KO 103 mouse.

Early work by Hong et al104 showed that in reductionist atrial myocyte (HL-1) and non-myocyte cell lines, BIN1 variants 5 and 8 were able to generate membrane tubulation and distribute LTCC to these membranes. Several groups have since demonstrated that the muscle variant 8 isoform of BIN1 was important for delivery of LTCC to the t-tubule in the heart103, 104, 107. As already discussed, more recent data showed that BIN1 variant 8 promoted the formation of tubules during hESC-CMs differentiation107. Rather surprisingly however, mRNA work from both Hong et al102 and Toussaint et al182 suggests that in the mouse heart, cardiac isoforms of BIN1 do not contain exon 11, but instead both groups identified variant 6, variant 9, variant 10 and a modified variant 9 (+ exon 13, - exon 17) isoforms (although, expressed to different levels). Moreover, only variant 6 was capable of rescuing t-tubule degradation in BIN1 cardiac conditional KO hearts102. These findings contrast our own work that demonstrated in sheep, exon 11 containing variant 8 was the most dominantly expressed isoform in the heart83. Thus, further work is required to determine the differing roles of BIN1 isoforms in cardiac muscle and how they play a part in t-tubule biogenesis. As with t-tubule density, consideration must also be given for species dependent differences and how these relate to the human heart.

143

5.1.2. Functional role of BIN1

It is well established that changes in t-tubule density and organisation are paralleled by changes in the amplitude and synchronicity of the systolic calcium transient. Thus, in addition to a structural role in t-tubule biogenesis, BIN1 may also play a part in maintaining calcium release at the t-tubules. T-tubules are enriched with LTCC to enable close proximity to RyR and facilitate calcium induced calcium release8. Correct organisation of these proteins at the t-tubule is essential to maintain normal calcium signalling. Work from Hong et al104 showed that in both mouse and human cardiac myocytes BIN1 interacted with LTCC, trafficking them to the t- tubule and that these interactions were vital for normal calcium signalling in the heart. In particular, studies have shown that deletion of BIN1 resulted in loss of surface LTCC and the development of abnormal calcium transients in cardiac myocytes104. Furthermore, a reduction in BIN1 expression, as seen in human failing cardiac myocytes, impaired LTCC trafficking112.

Further linking BIN1 to cardiac function, severe contractile dysfunction was observed in Zebrafish hearts lacking BIN1112, suggesting a cardiac phenotype associated with BIN1 reductions. This observation was paralleled by work from Laury-Kleintop et al103 who showed that conditional cardiac deletion of BIN1 in mice resulted in heart failure. Notably, both Hong et al and Laury-Kleintop et al described decreased t-tubule intensity, paired with altered distribution of LTCC103 or decreased L-type calcium current102 in this model. In addition, we have shown that silencing BIN1 in isolated rat ventricular myocytes caused t-tubule loss and dys- synchronous calcium release, which we proposed was due loss of LTCC46. Taken together, these studies highlight reduced BIN1 contributes to the pathophysiology of heart failure through altered calcium signalling. Importantly, overexpression of BIN1 variant 5 in HL-1 cells has been shown to rescue LTCC expression104. Alongside LTCC trafficking, BIN1 also regulates the movement of RyRs to t-tubule micro- domains in response to β-AR activation35. Thus by recruiting dyadic proteins to the t-tubule membrane and maintaining LTCC - RyR interactions, BIN1 may also play a vital role in calcium regulation in the heart.

144

5.1.3. Aims of the chapter

Evidence suggests that BIN1 controls t-tubule formation in cardiac muscle and in turn plays a role in calcium regulation. The previous chapter of this thesis has shown induction of BIN1 in NRVMs led to tubule formation. This chapter will begin to investigate the roles of the different variants of BIN1 in the heart and determine if BIN1 driven t-tubule synthesis will lead to the alignment of EC coupling proteins to the t-tubule and the effect this has on systolic calcium. The aims of this chapter are;

(1) To determine the role of different BIN1 variants in t-tubule biogenesis in NRVMs. (2) To investigate how BIN1 driven t-tubule formation impacts on cell structure and calcium handling during development. (3) To determine if the key proteins involved in EC coupling align and localise to BIN1 induced tubules.

145

5.2. Results

5.2.1. Transfection with BIN1 variants 5, 8 & 9 in cardiac cells

To determine the role of different BIN1 variants on t-tubule biogenesis, NRVMs were transiently transfected with vectors expressing BIN1 variants (v) 5, 8 and 9 with a C-terminal mKate2 fluorescent tag, as described in the previous chapter (Methods section 2.5 for more details). Whilst overexpression of BIN1 variant 8 has already begun to be explored in this thesis, much less is known about the roles of the exon 11 skipping variants 5 and 9 in t-tubule biogenesis.

5.2.1.1. Expression of several variants of BIN1 led to the formation of tubules in NRVMs.

Western blotting was used to confirm expression of BIN1 in transfected NRVMs. Results are shown in Figure 5.2 A and summarised in Table 5.1. BIN1 was expressed in very low levels in non-transfected (NT) and the control vector, mKate2, transfected NRVMs (Figure 5.2 Ai). Following transfection with BIN1 variants (v) 5, 8 and 9, two BIN1 protein bands were observed in the NRVMs. BIN1 protein was up-regulated in all variants compared with non-transfected myocytes and the mKate2 empty vector following BIN1 transfection. Total protein stain was used to confirm even loading (Figure 5.2 Aii).

NRVMs transfected with vectors expressing BIN1 and stained with the membrane dye FM 4-64 showed an increase in tubule like structures when compared to cells transfected with a vector expressing mKate2 only. Figure 5.2 B shows representative (x-y) confocal images compiled from confocal z-sections, tubules were abundant, albeit disorganised, in BIN1 (variant 5, 8 and 9) transfected cells.

Of successfully transfected cells, determined by the presence of the fluorescent tag mKate2, ~97% of cells transfected with BIN1 had developed tubule like structures (Figure 5.2 C and in Table 5.1). Conversely, tubules were absent in cells only expressing mKate2. Tubule density was quantified by determining the fraction of the

146

cell occupied by tubules (section 2.8.2). Mean data, summarised in Figure 5.2 D and in Table 5.1, show that transfection with BIN1 variants 5, 8 and 9 resulted in an increase in fractional area of the cell occupied by tubules compared with mKate2 transfection. Furthermore, transfection with BIN1 variant 8 led to formation of 17.4 ± 6% and 21.7 ± 6% more tubule structures when compared to variant 5 and 9 respectively, suggesting that variant 8 had a more important role in tubule formation.

Alongside differences in tubule density, to establish if there were also structural dissimilarly in tubules between the different variants of BIN1, tubule structural alterations were calculated using the ‘Analyse Skeleton’ algorithm in ImageJ as described in the previous chapter. There was a decrease in skeleton length (Figure 5.2 Ei) and in the number of tubule branches (Figure 5.2 Eii) in BIN1 v9 transfected NRVMs compared with BIN1 v5 and v8 (summarised in Table 5.1).

147

Figure 5.2. Expression of several variants of BIN1 in NRVMs. (A) Representative western blot and summary data showing protein expression of BIN1 and total protein in non-transfected (NT), mKate2 and BIN1-mKate2 variant (v) 5, 8 and 9 NRVMs. (B) Representative images of NRVMs transfected with mKate2 and BIN1 expression vectors. (C) Summary data showing the % of transfected cells that have tubules. (D) Fractional area of tubules in transfected cells. (E) Average skeleton length (i) and number of branches in each structure (ii). N= 3-5 litters (mkate2, 20 cells; BIN1 v5, 27 cells; v8, 29 cells; v9, 31 cells). $ p <0.05, $ $ $ p <0.001 vs BIN1 v8; *** p <0.001 vs mKate2 (and NT); # p <0.05 vs BIN1 v5, tested by linear mixed model analysis, presented as mean ± SEM. Scale bars =10µm.

148

NT mKate2 BIN1 v5 BIN1 v8 BIN1 v9 Protein 0.001 ± 0.03 0.002 ± 0.03 0.06 ± 0.03*** 0.09 ± 0.03*** 0.09 ± 0.03***

% cells with TT 0 ± 0.0 1.4 ± 0.02 96.3 ± 0.03*** 97.3 ± 0.03*** 97.2 ± 0.03***

TT fractional area (µm) - 0.004 ± 0.0 0.19 ± 0.01***,$$$ 0.23 ± 0.01*** 0.18 ± 0.01***,$

TT length (µm) - - 5.1 ± 0.8 5.5 ± 0.9 2.0 ± 0.4#,$

TT branches - - 2.6 ± 0.2 2.7 ± 0.3 1.7 ± 0.2#,$

n=cells; N=litters n=23, N=6 n=20, N=5 n=27, N=5 n=29, N=5 n=31, N=3

Table 5.1. Summary of BIN1 expression parameters in NRVMs. $ p <0.05, $ $ $ p <0.001 vs BIN1 v8; *** p <0.001 vs mKate2 (and NT); # p <0.05 vs BIN1 v5, tested by linear mixed model analysis

149

5.2.1.2. Tubules developed rapidly following expression of BIN1 in NRVMs

NRVMs were recorded over time during the BIN1 transfection period to try to establish the speed in which tubules develop and the origin of the tubules. Figure 5.3 A shows a ~47 minute time lapse recorded during BIN1 transfection, demonstrating that once tubules started to develop, they developed rapidly. Although this data is preliminary, it appears that tubules developed from the surface sarcolemma towards the centre of the cell, as is highlighted in Figure 5.3 B.

Figure 5.3. BIN1 driven tubule time series. (A) Time series recorded during the transfection period showing overexpression of BIN1-mKate2 variant 8 caused tubules to develop rapidly. Each frame represents 255 seconds (total time 46 minutes 45 sec). (B) Expanded time series showing the development of an individual tubule highlighted by the blue arrow, total time ~20 minutes. N= 3 litters, scale bars =10µm.

150

5.2.1.3. Expression of BIN1 in human iPSC-CMs.

To determine if BIN1 could also lead to the development of tubules in human cardiac cells, commercially available human iPSC-CMs (Cellular Dynamics) were transiently transfected with vectors expressing BIN1 variants 5, 8 and 9. Confocal images in Figure 5.4 A show that BIN1 was also capable of driving tubule formation in human iPSC-CMs and these cells normally lack a tubule network. On average 99 ± 0.3% cells successfully transfected with BIN1 developed a tubular network compared to no tubules in non-transfected or cells transfected with the mKate2 control vector (Figure 5.4 B). Furthermore, expression of BIN1 variants 5, 8 & 9 in human iPSC CMs led to an increase in the fractional area of cells occupied by t- tubules from 0.03 ± 0.01 in mKate2 only cells to: 0.13 ± 0.01 in variant 5 cells; 0.19 ± 0.01 in variant 8 cells and 0.14 ± 0.02 in variant 9 cells (Figure 5.4 C). Moreover, a similar pattern to that of the NRVMs was observed, in that transfection with BIN1 variant 8 lead to formation of more tubule structures when compared to variant 5 and 9 (32.1 ± 8% and 27.8 ± 9% increase respectively).

151

Figure 5.4. Expression of BIN1 in human iPSC-CMs. (A) Representative images of iCell human iPSC derived cardiac myocytes transfected with mKate2 empty vector (control) and BIN1-mKate2 variant (v) 5, 8 and 9 expression vectors (B). Mean data summarizing percentage of non-transfected (NT), mKate2 control and BIN1 expression vectors transfected iCell human iPSC cardiac myocytes with tubule structures. (C) Mean data summarizing fractional area occupied by tubules in iCell cardiac myocytes transfected with mKate2 or BIN1 vectors. N = 2 cell batches (mkate2, 17 cells; v5, 15 cells; v8, 29 cells; v9, 10 cells). *** p < 0.001 vs mKate2; $$ p < 0.01 vs v8; $$$ p < 0.001 vs BIN1 v8; by linear mixed model analysis, presented as mean ± SEM. Scale bars, 10 µm.

152

5.2.1.4. The fluorescent tag on BIN1 did not interfere with tubule formation in NRVMs.

A separate group of NRVMs were transfected and imaged, as described above, with commercial vectors (OriGene) expressing BIN1 variants 5 (RC220682), 8 (RC220616) and 9 (RC202423) without the mKate2 fluorescent tag. These experiments were performed to determine if the mKate2 fluorescence tag interfered with the formation of tubules. NRVMs transfected with vectors expressing BIN1 variants 5, 8 & 9 and stained with the membrane dye FM 4-64 led to the development of tubule like structures (Figure 5.5 A). It has not been determined if these tubule structures differed from pCMV6-AC-mKate BIN1 driven tubules. Once again, expression of BIN1 variant 8 led to 25.6 ± 5 % and 22.7 ± 5 % more tubule structures by when compared to variant 5 and 9 transfected cells (Figure 5.5 B).

Figure 5.5. BIN1 driven tubules without the mKate2 tag. (A) Representative images of NRVMs transfected with vectors expressing BIN1 variants (v) 5, 8 and 9 without the fluorescent tag, stained with FM4-64 to visualise tubule structures. (B) Summary data of fractional area occupied by tubules in NRVMs transfected with empty or BIN1 vectors. N= 3-5 litters (mkate2, 22 cells; v5, 12 cells; v8, 14 cells; v9, 16 cells). *** p < 0.001 vs empty vector; $$$ p < 0.001 vs BIN1 v8; by linear mixed model analysis, presented as mean ± SEM. Scale bars, 10 µm.

153

5.2.2. BIN1 driven tubules were functional

In adult myocytes t-tubules play a vital role in ensuring synchronous calcium release throughout the myocyte; therefore the next section of this study aimed to determine how BIN1 driven tubule formation impacted calcium handling. Calcium handling experiments were carried out on field stimulated NRVMs to investigate the properties of calcium release events following tubule induction with BIN1 variants 5, 8 and 9.

5.2.2.1. Expression of BIN1 led to an increase in the amplitude of the systolic calcium transient.

A representative confocal time series frame of stimulated NRVMs loaded with the calcium indicator Fluo 8 AM is shown in Figure 5.6 A. Each time series recorded calcium release from both transfected and non-transfected cells; transfected cells were determined by mKate2 fluorescence. Region of interests (ROI) on the time series were selected, corresponding to; no cell (red), non-transfected cell (grey), transfected cell (not on membrane) (pink), transfected cell (on tubule membrane) (green). For each ROI, fluorescence intensity was measured across the time series. This produced a series of transients which corresponded to calcium release. Calcium was expressed as F/F0 after subtracting background fluorescence (no cell ROI).

Figure 5.6 B shows typical F/F0 normalised systolic calcium transients from non- transfected (grey), transfected (pink) and tubule (green) ROIs.

The first set of experiments set out to determine if transfection with the control vector mKate2 had any effect on calcium properties compared with non-transfected cells. Summary data in Table 5.2 demonstrates that there were no differences in calcium transient amplitude, rise time and rate constant of decay between non- transfected cells and the control mKate2 vector. Therefore, for the next set of experiments, aimed at determining if BIN1 transfection resulted in larger and faster calcium transients, BIN1 transfected cells were compared to non-transfected cells as these comparisons could be performed on cells from the same experimental bath. The data has been normalised to the non-transfected cells.

154

Non-transfected mKate2

Ca transient amplitude (F/F0) 2.01 ± 0.2 2.01 ± 0.2

Ca rate of rise (ms) 21.0 ± 3.8 19.8 ± 3.3

Rate constant of decay (sec-1) 0.97 ± 0.2 1.07 ± 0.2 n=cells; N=litters n=30, N=5 n=30, N=5

Table 5.2. Summary of control calcium handling data. No differences were found between non-transfected and mKate2 (control) transfected NRVMs, tested by linear mixed model analysis.

Following BIN1 transfection, the mean amplitude of the systolic calcium transient was increased by 19 ± 11% in variant 5, 30 ± 12% in variant 8 and 30 ± 12% in variant 9 transfected cells compared with non-transfected NRVMs (Figure 5.6 Ci and Table 5.3). There was no difference in the systolic calcium transient between tubule and non-tubule ROIs in the BIN1 transfected myocytes (Figure 5.6 Di).

Summary data for the time to peak i.e. the rise time of systolic calcium is shown in Figures 5.5 Cii and Dii. Calcium rise time was 63 ± 43% faster in cells transfected with BIN1 variant 5 compared with non-transfected cells, there was also a trend (p=0.056) towards a faster calcium rise time following BIN1 variant 9 transfection, but there was no difference in the time for calcium to reach peak in cells transfected with BIN1 variant 8. No differences were observed in calcium rise time between the tubule and non-tubule regions of any of the BIN1 transfected cells.

Alongside calcium entry to the cell, t-tubules also assist with calcium extrusion. It was therefore also important to determine if the BIN1 driven tubules in the NRVMs play a role in calcium removal after each contraction. By fitting a single exponential curve to the decay phase of the systolic calcium transient (Figure 5.6 B), calcium removal rate (rate constant of decay) was determined. The summary data in Figure 5.6 Ciii and Table 5.3 shows, in NRVMs, following BIN1 transfection the rate constant of calcium decay was 84 ± 46% faster in variant 5 and 122 ± 95 % faster in variant 9 cells compared with non-transfected cells. Figure 5.6 Diii shows tubule ROI verses non-tubule ROI from had no effect on the rate of calcium removal.

155

normalised BIN1 v5 BIN1 v8 BIN1 v9

Ca transient amplitude (F/F0) 1.19 ± 0.1* 1.30 ± 0.1** 1.30 ± 0.1*

Ca rate of rise (ms) 1.63 ± 0.3* 1.31 ± 0.2 1.18 ± 0.4

-1 Rate constant of decay (sec ) 1.84 ± 0.3* 1.07 ± 0.1 2.22 ± 0.6** n=cells; N=litters n=27, N=5 n=43, N=7 n=16, N=4

Table 5.3. Calcium handling properties in BIN1 transfected NRVMs. Summary data of calcium handling properties between non-transfected and BIN1 variant (v) 5, 8 and 9 transfected NRVMs, * p <0.05, ** p <0.01 vs NT tested by Two Way Repeated Measured ANOVA. Data was normalised to the mean of the non- transfected NRVMs.

156

Figure 5.6. BIN1 transfection led to improved calcium handling. (A) Single frame taken from a time series of transfected NRVMs loaded with a calcium indicator. ROIs representing, no cell (red), non-transfected cell (grey), transfected cell (pink) and on a tubule in a transfected cell (green), were selected and calcium transients plotted. (B) Representative calcium transients from non-transfected and BIN1-mKate2 transfected NRVMs. (C) Compared with non-transfected myocytes, expression of BIN1 variants (v) 5, 8 & 9 increased the amplitude of the systolic calcium transient (i). Transfection with BIN1 variants 5 & 9 led to faster rise (ii) and decay (iii) of the systolic calcium transient. (D) No difference in the amplitude (i), rise time (ii) or rate constant (iii) of the systolic calcium transient between tubule

ROIs and non-tubule ROIs on BIN1 transfected cells. Fluorescence (F/F0) measured using Fluo-8. N= 4-7 litters (BIN1 v5, 27 cells; v8, 43 cells; v9, 16 cells). * p <0.05, ** p <0.01 vs non-transfected ROI; # p <0.05 BIN1 v9 vs v8; by Two Way Repeated Measured ANOVA, presented as mean ± SEM

157

5.2.2.2. Transfection with BIN1 led to more synchronous calcium release.

As previously described, synchronous calcium release is important for normal cellular contractile function and the presence of tubules in cardiac myocytes are a key determinate of calcium synchronicity2, 3. Thus the degree of dys-synchrony of the rise of intracellular calcium was measured (as described in section 3.1.4.2) in BIN1 transfected NRVMs. The standard deviation of systolic calcium rise was used to represent dys-synchrony. The preliminary data in Figure 5.7 shows in BIN1 transfected myocytes (variant 8 only), there was 31.4 ± 13% decrease in the standard deviation, i.e. dys-synchrony index, of the rise time from 59.3 ± 5.2 ms in non- transfected myocytes to 40.7 ± 6.9 ms following BIN1 transfection, representing more synchronous calcium release.

Figure 5.7. Transfection with BIN1 led to more synchronous calcium release. (A) Preliminary data showing calcium release was more synchronous in BIN1- mKate2 variant (v) 8 transfected NRVMs compared with non-transfected myocytes. Dys-synchrony measured using standard deviation of rise time. N= 3 litters (non- transfected, 9 cells; BIN1 v8, 9 cells). *** p < 0.001 vs NT; by students t-test.

158

5.2.3. Cellular distribution of calcium handling proteins

Results from this chapter have shown that BIN1 transfection led to enhanced calcium release in NRVMs. It is not known however if calcium release was enhanced because the distance for calcium diffusion was decreased across the cell or if EC coupling proteins were modulated in response to BIN1 driven tubule formation, forming functional couplons at the tubule and thus triggering calcium release from the SR. The purpose of the next set of experiments therefore was to determine if EC coupling associated proteins localised to the BIN1 driven tubules following transfection.

The calcium efflux channel NCX was found on the surface sarcolemma in non- transfected NRVMs (Figure 5.8 A, top left panel) and in myocytes transfected with the mKate2 control vector. Following transfection with BIN1 to induce tubules (WGA – Alexa Fluor 647 stained) however, the localisation of the membrane bound protein NCX shifted from the surface sarcolemma and was found predominately on the BIN1 induced tubules (Figure 5.8 A, top middle panels). In agreement with this finding, the fraction of co-localisation between the membrane (tubule or sarcolemma) and NCX increased in BIN1 transfected cells compared with mKate2 transfected cells (shown in yellow in Figure 5.8 A right panel and Figure 5.8 Bi).

In non-transfected and mKate2 transfected NRVMs the SR calcium release channels, RyRs, had an ordered sarcomeric distribution throughout the cell as demonstrated in Figure 5.8 A (bottom left panel). Whilst the distribution of RyR appeared unaltered following BIN1 transfection, tubules (WGA – Alexa Fluor 647 stained) were found in the center of the cell following transfection (bottom middle panels). Thus, more RyRs were in close proximity to tubules in BIN1 transfected cells leading to more co-localisation between the two compared with mKate2 transfected cells, as demonstrated in yellow in Figure 5.8 A and in the summary data of Figure 5.8 Bii.

When comparing the fraction of co-localisation (Mander’s overlap) between BIN1 variants, more co-localisation between NCX and membrane (tubule or sarcolemma)

159

was observed in NRVMs transfected with variants 5 and 9 of BIN1, when compared with variant 8 (Figure 5.8 Bi and Table 5.4). A similar co-localisation pattern was observed between RyR and membrane with transfection of BIN1 variant 9 leading to 28 ± 7% more co-localisation. There was also a trend (p=0.07) towards increased co- localisation of RyR and tubules following BIN1 variant 5 transfection compared with variant 8 (Figure 5.8 Bii and Table 5.4). There were no observed differences in the cellular distribution of NCX or RyR between non-transfected cells and the control mKate2 vector.

Manders mKate2 BIN1 v5 BIN1 v8 BIN1 v9 overlap $ *** $$ RyR & Tubules 0.17 ± 0.03 0.50 ± 0.02*** 0.45 ± 0.02 0.58 ± 0.02***

*** *** $$$ NCX & Tubules 0.32 ± 0.04 0.56 ± 0.02 0.47 ± 0.02 0.56 ± 0.02*** n=cells; N=litters n=10, N=2 n=25, N=3 n=34, N=3 n=34, N=3

Table 5.4. Co-localisation of tubules and calcium handling proteins in BIN1 transfected NRVMs. Mander’s overlap of calcium handling proteins and tubules or surface sarcolemma in mKate2 only and BIN1 variant (v) 5, 8 and 9 transfected NRVMs, *** p < 0.001 vs mKate2; $ p < 0.05, $$ p < 0.01; $$$ p < 0.001 vs BIN1 v8; by linear mixed model analysis

160

Figure 5.8. Cellular distribution of calcium handling proteins. (A) Immuno staining of protein (RyR or NCX) (green) and membrane (t-tubule or surface sarcolemma) (red) in non-transfected (left panel) and BIN1-mKate2 transfected NRVMs. Yellow represents co-localisation. (B) Preliminary data showing Manders co-localisation overlap (fraction) of protein (NCX (i) or RyR (ii)) and membrane in mKate2 and BIN1 variant (v) 5, 8 and 9 transfected NRVMs. *** p < 0.001 vs mKate2; $ p < 0.05, $$ p < 0.01, ; $$$ p < 0.001 vs BIN1 v8; by linear mixed model analysis, presented as mean ± SEM.. N= 3 litters (NT, 9-10 cells; BIN1 v5, 22-25 cells; v8, 24-34 cells; v9, 30-34 cells).

161

5.2.3.1. Immuno staining of z-line proteins

Despite the results from the previous section showing there was more co-localisation between BIN1 driven tubules and calcium regulatory proteins, it is still important to consider that the morphology of the tubules produced from BIN1 transfection were different to t-tubules found in an adult myocyte. Thus, these tubules might not fully function in the same way. One factor that may influence tubule shape and organisation is the proteins of the z-line. In adult myocytes z-line proteins lie perpendicular to t-tubules202, so it is thought that the z-lines play a role in anchoring the t-tubules (reviewed in143). It is therefore possible that if the z-lines or z-line anchor proteins were not fully developed in NRVMs, the tubule network would have remained disorganised. The next section of this work will try to determine this.

Alpha-actinin (α-actinin) is a cytoskeletal protein that is localised to the z-lines in cardiac muscle. Figure 5.9 Ai shows α-actinin (green) had an ordered sarcomeric distribution, on the z-lines, in both adult and NRVMs. This distribution was unaltered following BIN1 transfection to induced tubule formation (red), moreover, qualitatively, α-actinin did not appear to co-localise with the BIN1 driven tubules (Figure 5.9 Aii). This suggests that whilst the z-lines were fully developed in NRVMs, they do not align with the newly formed tubules.

Finally, immunostaining was used to access the purity of the NRVM cultures. Myocytes were either stained with the cardiomyocyte marker Troponin I or with the fibroblast marker Vimentin. Figure 5.9 B shows positive staining for Troponin I (i) and very little staining for Vimentin (ii), confirming the culture was predominately made up of myocytes.

162

Figure 5.9. Immuno staining of z-line proteins. (Ai) α-actinin (green) was well ordered and regularly spaced in both adult and neonatal myocytes. (ii) Transfection with BIN1-mKate2 appeared to have no effect on α-actinin distribution in NRVMs. BIN1 driven tubules in red; co-localisation of BIN1 and α-actinin shown in yellow. (B) Immuno staining for myocyte marker Troponin I (i) and fibroblast marker Vimentin (ii) in fixed non-transfected NRVMs. N = 3 NRVM litters; N= 1 adult rat (n=3 cells). Scale bar denotes 10µm.

163

5.2.4. Protein expression in transfected NRVMs

Having established that BIN1 driven tubules were functional and express calcium handling proteins, the next aim sought to determine if transfection with BIN1 in NRVMs altered calcium and t-tubule regulators in the heart. The expression levels of proteins that are associated with calcium handling and t-tubule formation were investigated in non-transfected verses transfected NRVMs.

5.2.4.1. Expression of calcium handling proteins

Expression levels of LTCC, NCX, PLB, RyR and SERCA2a proteins are shown in Figure 5.10. Representative blots show bands for (A) LTCC, (B) NCX, (C) RyR, (D) SERCA2a and (E) PLB corresponding to the reported molecular weights (140kDA, LTCC, 116kDA, NCX; 565kDA, RyR; 110kDA, SERCA2a; 5kDA, PLB) Total protein was used as a loading control for all blots. The summary data in Figure 5.10 demonstrates that LTCC and NCX protein was ordinarily expressed in NRVMs (mKate2 control and NT), with no change in expression following transfection with BIN1 (variants 5, 8 or 9). Figure 5.10 also shows that compared to adult rats, RyR2 protein expression was 77.4 ± 6% less, SERCA protein expression was 95 ± 1% less and PLB was 78.6 ± 9% less in NRVMs. BIN1 transfection had no effect on the protein levels of RyR2, SERCA2a or PLB.

5.2.4.2. Expression of t-tubule associated proteins

JPH2, MTM1 and Tcap protein expression is shown in Figure 5.11. Single protein bands at the reported molecular weights 98kDA of JPH2 (A); 19kDA of Tcap (B); 55kDA of MTM1 (C) were detected in NRVMs. Protein expression was normalised to total protein. As previously reported, JPH2 and MTM1 were expressed in non- transfected NRVMs. No change in expression of JPH2 or MTM1 was detected between the groups following transfection with BIN1. Whilst Tcap was detected in non-transfected NRVMs, surprisingly there was a trend towards decreased Tcap expression following BIN1 transfection (v5 p=0.07; v8 p=0.06; v9 p=0.08).

164

Figure 5.10. Expression of calcium handling proteins in NRVMs. Representative western blots and mean data of LTCC (A), NCX (B), RyR2 (C), SERCA2a (D) PLN (E) and total protein in non-transfected (NT), mKate2 and BIN1- mKate2 variant (v) 5, 8 and 9 NRVMs, and in the adult rat ventricle (C-E). FL = Full length blots. Protein was normalised to total protein. N= 3 litters (biological repeats). ** p <0.01; *** p <0.001 vs adult rat, tested 165 by linear mixed model analysis.

Figure 5.11. Protein expression of t-tubule associated proteins in transfected NRVMs. Representative western blots and mean data showing protein expression of JPH2 (A), Tcap (B) and MTM1 (C) and total protein in non-transfected (NT), mKate2 only and BIN1-mKate2 variant (v) 5, 8 and 9 NRVMs. Full length (FL) blots showing protein molecular weight confirm the correct protein was labelled. JPH2 and MTM1 protein expression was unaltered following transfection with BIN1 in NRVMs. There was a trend towards a decrease in Tcap protein expression in BIN1 transfected NRVMs. N= 3 litters (biological repeats). Tested by linear mixed model analysis, presented as mean ± SEM.

166

5.3. Discussion

The major findings in this chapter have shown that expression of several different BIN1 variants in NRVMs can drive tubule formation. The work discussed here has also demonstrated that transfection with these BIN1 variants results in increased calcium transient amplitude. The enhanced effect on calcium handling in BIN1 overexpressing cells indicates that this protein plays a role in contractility of the myocyte.

5.3.1. Exon 11 of BIN1 was not required for the development of t- tubules in NRVMs or iPSCs.

The aim of this chapter was to identify which BIN1 variants are required t-tubule formation in NRVMs and more specifically if the muscle specific variant 8, encoding exon 11, is required. As already discussed in the previous chapter of this thesis, t-tubules were absent in NRVMs (without BIN1) and expression of BIN1 variant 8 led to tubule formation. In this chapter however, we observed for the first time, that transient transfection with other BIN1 variants (5 and 9) also led to the development of tubule like structures in NRVMs. Furthermore, we also detected this tubular pattern following BIN1 expression in human iPSC-CM. Although previous work has demonstrated that overexpression of BIN1 variants 5 and 9 were able to generate tubule like structures in non-muscle cells104, 114, the variant 9 results were surprising. As already discussed, in skeletal muscle, the exon 11 skipping variant 9 was incapable of causing membrane tubulation201 and was unable to rescue t-tubule degradation in BIN1 cardiac conditional knockout hearts102. Thus, this study has shown that the exon 11 PI domain of BIN1 (present in variant 8) was not required for tubule formation in the heart and that variant 8 was not the only variant capable of tubulation. All the variants studied here commonly share an N-terminal amphipathic helix, BAR and SH3 domains203. As discussed in the introduction, these domains are able to create and maintain membrane curvature and mediate interactions with other membrane associated proteins90, 101, 198. The data present here suggests that in cardiac muscle these commonly expressed domains are all that is

167

required to generate tubule formation. Interestingly, mutations in these domains have been shown to stop membrane tubulation and disrupt interaction with BIN1 binding partners in skeletal muscle108, 203. Further work is needed to determine if mutations in these commonly expressed domains affect the ability of BIN1 to tubulate in the heart. Despite this, the data presented here does suggest that BIN1 variant 8 plays a more important role in tubule formation, with transfection leading to more tubule structures when compared with variant 5 or variant 9 in both NRVMs and iPSC- CMs. Transfection with BIN1 variant 8 also led to increased tubule density and longer more branched tubules compared to variant 9. Recently published data from our lab showed that variant 8 was the most highly expressed BIN1 variant in the sheep ventricle83, but this was in contrast to the mouse heart where it was thought variant 8 was not present102, 182. Critically, all these studies identified BIN1 isoforms based on mRNA data and it is possible that BIN1 can be post-translationally modified, for example, additional isoforms of BIN1 were detected in skeletal muscle when identifying BIN1 at the protein level108. Together these data suggest that fundamental differences in tubule biogenesis exist between cell types and species, and these differences may provide key insight into how t-tubules are formed and maintained.

5.3.2. The effect of BIN1 driven tubules on the systolic calcium transient.

The absence of a t-tubule network in NRVMs results in contraction being initiated via calcium diffusion to the centre of the cell8, 188, 204. One of the consequences of calcium diffusion rather than CICR is slower EC coupling. During maturation of NRVMs, t-tubules and LTCC and RyR couplon development coincide with faster rise of systolic calcium40. Results from this chapter have shown that tubule induction in NRVMs, as a result of transfection with BIN1, led to enhanced calcium release. As one of the key roles of the t-tubule network is to ensure the rapid synchronous rise of systolic calcium, it is not surprising that preliminary data from this study shows that BIN1 transfection led to more synchronous calcium release across the myocyte. This was in agreement with De La Mata et al107, who showed that hESC‐

168

CM expressing BIN1 also exhibited synchronized SR calcium release during EC coupling. The authors propose that BIN1 expression enhanced calcium release by acting as an anchoring point for LTCC – RyR junctions. Whilst it has not been possible to study the effect of BIN1 transfection on the formation of RyR and LTCC couplons, due to a lack of a commercially available LTCC antibody satisfactory for immunocytochemistry, the data presented here suggested that more RyRs were found in close proximity to BIN1 driven tubules. It is very likely a co-localisation between these molecules would increase couplon development across the cell and thus provide an explanation for the homogenous calcium release observed in this model. Furthermore, alongside tubule formation, BIN1 has been shown to traffic the LTCC to the t-tubules104. Therefore it would be expected that the BIN1 induced tubules in this study would have been enriched in LTCC.

Alongside more synchronous calcium release, increased proximity of tubule to RyR couplons would also likely result in faster calcium release. Consequently, a faster rise time of calcium release was observed in cells transfected with BIN1 variants 5 and 9 compared with non-transfected cells. Following, BIN1 driven tubule formation more regions of the cell would have been close to membrane, thus the distance calcium was required to travel to reach the centre of the cell was reduced. Decreased diffusion distance has been shown to decrease calcium diffusion time which led to faster calcium release205. Even sparse disorganised tubule networks, found in a small subset of rat atrial cells, have been shown to contribute to rapid calcium release44, 206, as more areas of the cell were close to membrane. A surprising finding of this section however, was that transfection with BIN1 variant 8 had no effect of the rise time of the systolic calcium transient. Since transfection with BIN1 variant 8 led to the development of more tubule structures compared with the other variants studied, it would be expected that more areas of the cell would be in close proximity to tubules. Despite this however, preliminary immunocytochemistry staining experiments indicated that less co-localisation existed between RyR and membrane in the BIN1 variant 8 transfected cells compared with variants 5 and 9. More in-depth work is required to determine the structural properties and localisation of membrane channels on the BIN1 driven tubules.

169

Although we observed mixed calcium rise time results across the BIN1 variants, transfection with BIN1 variants 5, 8 and 9 in NRVMs led to an increase in the amplitude of the systolic calcium transient when compared with non-transfected myocytes. This was in line with studies showing that BIN1 variant 8 expression increased calcium transients in hESC‐CM107. Whilst there are no studies known to date that have looked at the effects of other BIN1 variants on the calcium transient, Hong et al, showed greater surface calcium current in non-muscle cells transfected with full length BIN1 (that causes membrane invagination) compared with a non- functional mutant BIN1112. An increase in calcium current would likely lead to an increased calcium transient amplitude. Furthermore, our work has previously shown that knocking down BIN1 (in cells with tubules) had the opposite effect, whereby tubule density was decreased and calcium transients were reduced46. Systolic calcium transient amplitude alterations are often a result of L-type calcium current and SR content alterations. We have not measured L-type calcium current or SR content in this model, and while we have measured protein expression of LTCC and SERCA, and observe no change, these are not likely to be accurate measurements of channel activity. As NRVMs are much more reliant on calcium influx through LTCC, than SR calcium release, to generate calcium transients205, 207, it would be interesting to determine if BIN1 transfection led to a shift in this dependence. However, by anchoring LTCC to t-tubules, BIN1 has been shown to contribute to L- type calcium current regulation104. Thus, in line with the earlier work from Hong et al112, it is more likely that increased amplitude following BIN1 transfection was a consequence of increased surface calcium current. Further investigation into factors that govern the increase in the amplitude of the systolic calcium transient following BIN1 transfection is required.

Transfection with BIN1 variants 5 and 9 also led to faster rate of decay compared with non-transfected cells. In the adult myocardium, calcium removal is regulated predominately by uptake into the SR by SERCA. Compared to the adult however, SERCA was expressed at much lower levels in NRVMs (208 and Figure 5.10), such that calcium removal is more dependent on NCX208. Whilst there were no alterations to the expression of either SERCA or NCX protein following BIN1 transfection, NCX was predominately located on the tubule membrane following BIN1

170

transfection, which may have contributed, to a faster rate of calcium removal. Although, the effects of BIN1 transfection on NCX have not been widely studied, data from our lab (Charlotte Smith, un-published) showed NCX to co-localise with developing tubules. This was in agreement with studies from the adult rat which showed that calcium extrusion from the cell is, in part, regulated by the NCX, which was predominately located on t-tubules to allow for rapid calcium extrusion14. It is therefore likely in NRVMs, following transfection of BIN1 to induce tubules, calcium removal was primarily regulated by NCX. If NCX was responsible for the rapid rate of calcium removal in BIN1 transfected cells it is very probable that the L- type calcium current was the main regulator of the calcium transient in these cells as calcium must always be in flux balance. It is worth noting however, that neonates are reliant on NCX for both calcium influx and calcium removal202, so enhanced expression of NCX on the tubules in NRVMs may account for not only faster rate of calcium removal from the cell, but also for increased calcium transient amplitude. This section of this thesis further highlights that the role of the SR, in contributing to the rise and removal of systolic calcium, following BIN1 transfection needs to be investigated. As with calcium rate of rise, transfection with BIN1 variant 8 also had no effect on the speed calcium removal in these cells. This was not surprising however, as previously discussed there was no increase in the rise time of calcium in BIN1 variant 8 transfected cells, thus there was no requirement for faster calcium removal. Furthermore, immunocytochemistry staining experiments once again showed that less co-localisation existed between NCX and tubules in the BIN1 variant 8 transfected cells compared with variants 5 and 9.

Although global changes in the calcium transient were observed in cells transfected with BIN1 compared with non-transfected cells, there appears to be no spatial differences when comparing calcium release between regions on the tubule membrane and those adjacent to it. This finding was not surprising when cell dimensions were taken into account. NRVMs in this study were approximately 10µm wide x 40µm length x 10µm deep. Considering that rat atrial myocytes with tubules are often wider than those without, 13.2 ± 2.8 vs. 11.7 ± 2.0 μm209, suggests that in cells of smaller dimensions t-tubules are not required. Furthermore, in adult ventricular myocytes with t-tubules, the distance between t-tubules is ∼2 μm. Thus,

171

it is likely that very few tubules were needed in NRVMs to ensure synchronous calcium release across the cell. This is in agreement with work from Louch et al48 who demonstrated that t-tubule disorganisation alone was not sufficient to delay calcium release. In this study, delayed calcium release regions were only observed in cells exhibiting pronounced gaps (∼4 μm) between tubules48. The distance between BIN1 induced tubules has not been measured in this study, but it would be interesting to determine the spatial organisation of the tubules in these cells. Together, these data suggest that BIN1 driven tubules were functional in that they increased systolic calcium. These data also indicate that BIN1 variants 5 and 9 had more of an effect on calcium handling. The effect this had on contractility or SR calcium remains to be determined.

5.3.3. BIN1 driven tubules were highly disordered.

Although expression of BIN1 variants 5, 8 and 9 all caused tubule like structures to develop in NRVMs, none of these variants led to the development of organised t- tubule structures as found in the adult myocyte. Whilst it is possible that an alternative variant of BIN1, not studied here, is specifically required for tubule organisation, it is more probable that this disorganisation was a consequence of the NRVMs not being fully mature. Development of the t-tubule network in rodents occurs after birth where tubules are primarily longitudinally orientated before being replaced with transverse tubules as the cell matures40, 41, 210. Longitudinal tubules formed during development have been suggested to play specialist roles in triggering calcium release with dyadic arrangement more suited to the developing heart41. Compared with transverse dyads, longitudinal dyads have been shown to co-localise less with LTCC and more with NCX41. As discussed, NCX activity is prominent in the developing heart. Thus the longitudinally orientated tubules observed in this study might have been a necessity for the immature cells they occupy.

In the healthy adult heart, t-tubules anchor to the sarcomeric z-discs allowing for a well-ordered distribution. T-tubule dis-organisation could therefore be explained by z-line disorder or insufficient ability to ‘anchor’. Consistent with previous studies41, the data presented here shows, α-actinin had a sarcomeric distribution, on the z-lines,

172

in both adult and NRVMs indicating that the z-lines are well ordered early in development. BIN1 transfection did not appear to alter α-actinin distribution, moreover α-actinin did not co-localise with the BIN1 driven tubules. These data suggest that BIN1 driven tubules were not anchored or aligned to z-lines in NRVMs. Lack of tubule - z‐line connections could account for the disordered appearance of the tubules observed in this model, thus it is also important to consider the scaffold proteins required to establish this connection.

As previously discussed, t-tubule membrane associated proteins and scaffolds JPH2, MTM1 and Tcap are thought to be important to maintain t-tubule structure. The data presented here showed that both JPH2 and MTM1 expression were unaltered following BIN1 transfection in NRVMs, which could provide further explanations for the tubule disordered observed. In some studies JPH2 expression has been shown to be critical for the shift between longitudinal to transverse tubules during development124, 125. JPH2 deficiency has also been associated with t-tubule disorganisation in the heart 46, 50, 81. It is worth noting however, that whilst evidence strongly supports a role for JPH2 in t-tubule organisation in the developing heart, there is controversy surrounding the role of JPH2 in the failing heart. MTM1 is also likely to be involved in t-tubule formation in the heart. In agreement with studies showing that interactions between MTM1 and BIN1 were crucial for t-tubule organisation in skeletal muscle121, the previous chapter of this thesis demonstrates that MTM1 was able to shape t-tubule structures in the heart. Thus, it is possible that JPH2 and MTM1 expression was required to assist in the organisation of BIN1 driven tubules. Whilst there was no change in the expression of t-tubule proteins JPH2 and MTM1 in NRVMs following BIN1 transfection, Tcap protein expression trended towards a decrease. Tcap provides a link between the t-tubule and z-lines135, thus it is not surprising that deletion of Tcap has been shown to disrupted sarcomere–membrane interactions137. Decreased Tcap, as partially observed in this study, could therefore inhibit anchoring of tubules to z-lines, thus somewhat explaining the disordered pattern of tubules observed following BIN1 transfection. One final point to consider however, that is in the previous chapter of this research, Tcap overexpression altered BIN1 tubule formation in a way that indicated this protein acts as a negative inhibitor of the BIN1 pathway. In support of this

173

possibility, our unpublished work (Charlotte Smith), demonstrated in the developing sheep atria, BIN1 protein expression decreased as Tcap expression increased. This coincided with prevention of further t-tubule development. As Tcap regulates t- tubules in a load dependent manner, it is possible, in response to load, Tcap is able to modulate protein expression and control t-tubule formation. If Tcap were to act on BIN1 to regulate t-tubule development, it is not unexpected that this protein would be supressed following BIN1 overexpression. These data suggest that several other mechanisms and pathways, alongside BIN1, are required for proper t-tubule development and that further work is needed to fully elucidate control of this system.

5.3.4. Study limitations

NRVMs are a very useful model for cardiac studies, however there are several problems with using spontaneously beating myocytes to study calcium handling. Work by Harary et al211, and others, showed that approximately only 2% of isolated neonatal cells had the ability to initiate spontaneous contraction. The other cells only beat when they came into contact with the beating cells, which then formed a cluster of contracting cells. Moreover, isolated NRVMs beat at different frequencies and only become synchronized after the establishment of contacts through gap junctions212. Therefore in one plate of isolated cells, each beating cluster may contract at a different frequency making it impossible to accurately measure and compare some calcium handling properties. To overcome this, myocytes were electrically stimulated to allow greater control over stimulation rates.

Despite being able to control spontaneous activity using electrical stimulation in NRVMs, I was unable to control spontaneous beating during periods of rest. Thus, I was unable to accurately measure SR calcium in the transfected NRVMs. To measure SR content, we have previously applied caffeine to cells during periods of no stimulation to disable SR function. In the maintained presence of caffeine, the amplitude of the systolic rise of calcium is determined only by sarcolemmal calcium entry and efflux. By comparing the rates of decay of the systolic and caffeine-evoked rises of intracellular calcium, SR and sarcolemmal calcium efflux can be 213 determined . As the pacemaker current (If) is thought to initiate spontaneous

174

activity in NRVMs214, a possible way to overcome this problem would be to use ivabradine to block If in NRVMs thus giving better control over stimulation of the cell. Alternatively, it would also be possible to manipulate the excitability of the membrane by blocking sodium channels and consequently preventing action potential propagation (reviewed in215).

Although confocal microscopy is a powerful imaging tool, protein co-localisation studies can often be limited by the relatively low spatial resolution. Spatial resolution of confocal imaging, ∼250 nm, is often much larger than many of the proteins and structures being studied. Thus protein co-localisation is often overestimated and results in contrasting findings. It cannot be discounted that the co-localisation findings in this study could also be overestimated. The invention of super-resolution fluorescent microscopy imaging techniques has led to improved xy spatial resolution from ∼250 nm with confocal microscopy to 10–20 nm216, 217. This improved resolution has enabled study of cellular structures at the nanometer scale and provided much more accurate co-localisation measurements218. A future direction of this research would therefore be to investigate the relative distribution of calcium handling proteins in the BIN1 transfected myocytes with super-resolution fluorescent microscopy to achieve greater resolution and more accurate measurements.

A final limitation of this study was that protein expression changes following transfection were likely to be underestimated or not detected due to low transfection efficiency. The efficiency of transient transfection is often limited, in this study it was approximately 30%, and the number of successfully transfected cells can vary across experiments. Whilst, it is possible to ensure a pure population of transfected cells by separating out successfully transfected and non-transfected cells using fluorescence activated cell sorting (FACS), this technique requires large cell numbers. In this study it was not possible to separate out the transfected cells and those that were not due to low cell volumes. Thus cells selected for protein expression studies contained both transfected and non-transfected cells. Despite this, in the data presented here, there was a clear increase in the expression of BIN1 protein following transfection, suggesting that cell numbers used were sufficient to detect changes at the protein level.

175

5.4. Conclusions

In summary, this study has shown that BIN1 variants 5, 8 and 9 were involved in the formation of cardiac tubules and therefore indicates that the exon 11 PI domain of BIN1 (present in variant 8) was not required for tubule formation in NRVMs. Whilst tubules were present in the BIN1 transfected NRVMs, they appeared disorganised, the reasons and consequences for this disorder remain unclear but were likely due to cell immaturity. Despite limitations, the data indicates that transfection with BIN1 led to greater systolic calcium transients. These data showed faster calcium release and removal in the BIN1 variant 5 and 9 transfected cells, possibly due to the redistribution of key calcium handling proteins. Whether BIN1 driven tubules led to the formation of functional couplons between the calcium entry channels and SR calcium release sites remains to be determined. The present study however, clearly demonstrates that BIN1 contributes considerably to t-tubule formation in the heart. Thus BIN1 may be an attractive target for restoring t-tubules and thus systolic calcium and contractility in heart failure.

176

6. General Discussion

The aim of this study was to investigate the control of t-tubules in the heart and to determine the mechanisms that contribute to t-tubule recovery. The major findings of this work were; 1) that t-tubules were able to recover in the sheep atria following the onset of heart failure; 2) recovered t-tubules, despite being disorganised, could trigger central calcium release, which was more synchronous and faster to rise than in heart failure. Increased co-localisation of EC coupling machinery in the centre of the cell was thought to help mediate this; 3) t-tubule restoration in this model coincided with increased expression of the membrane-associated proteins BIN1, MTM1 and Tcap; 4) BIN1 overexpression in NRVMs led to tubule formation, the structures which were be shaped by Tcap and MTM1; 5) NRVMs expressing BIN1 exhibited increased systolic calcium transients and more synchronised calcium release. As such, the data presented here suggests that BIN1 is an important determinant of the t-tubule changes seen in response to heart failure and recovery.

6.1. The atrial t-tubule network recovered following loss in a sheep model of heart failure.

Induction of heart failure by rapid ventricular pacing resulted in atrial t-tubule loss and altered calcium handling42, 46. The rapid pacing model was used in this study as it is comparable to clinical dilated cardiomyopathy and it also mimics the progression of tachycardia cardiomyopathy induced heart failure169. This study demonstrated that cessation of rapid pacing in sheep led to t-tubule recovery, after almost complete loss, in atrial myocytes. Recovery of t-tubules has previously been shown in ventricular myocytes following the onset of several pathologies79-84, 88, but this study showed for the first time t-tubules can be recovered in the atria. Moreover, unlike ventricular t-tubule recovery where newly formed t-tubules cannot be distinguished from pre-existing tubules, in the atria, recovered t-tubules were completely new, albeit highly disorganised, structures. Thus, we have shown that damaged atrial cells were able to recover by building new t-tubules.

177

Despite the disorganised nature, newly formed tubules in the recovered atria were able to trigger calcium release in the centre of the cell, which was associated with complete recovery of calcium rise time and synchrony of calcium release. This improvement of calcium handling was also associated with restoration of left ventricular contractility and dilatation. In the ventricle, t-tubule recovery has also been linked with improved synchronicity of calcium release80, 82, 84, 88. Atrial t-tubule loss in heart failure has previously been associated with decreased L-type current49, 154, which resulted in smaller calcium transients release 154. We therefore hypothesize that functional LTCCs exist on recovered t-tubules and are coupled with the SR. Although neither LTCC function nor distribution was examined in this study, recent work by Dr Jessica Clarke showed restoration of L-type calcium current was responsible for recovery of the amplitude of the systolic calcium transient in this recovery model (personal communication, un-published). This coincides with improved co-localisation between RyR and tubules in recovered atrial myocytes. It is important to consider however, that dyadic arrangement has been shown to vary between transverse and longitudinally orientated tubules41, 219, with specialist calcium release sites thought to exist on longitudinal tubules. An interesting future direction of this research therefore, would be to explore this arrangement in recovered atrial myocytes, which exhibited such diverse tubule remodelling, to determine if there is a difference in triggered calcium release between the differentially orientated tubules.

We have identified, BIN1, MTM1 and Tcap as targets for t-tubule restoration, whereby expression of these proteins correlated with t-tubule density in the recovered atria. BIN1 has previously been linked with t-tubule restoration80, 83, 84 and maintenance46 in the ventricle. Conversely, in our recently published work, we did not observe reductions in Tcap and MTM1 in the failing sheep ventricle where we observed t-tubule loss83, suggesting differing roles for these proteins between cardiac chambers. Since MTM1 is crucial for t-tubule organisation in skeletal muscle121 we suggested that the observed increase in MTM1 in the ventricle following heart failure83, was responsible for the lateralization of t-tubules observed in this model. In other models in the ventricle, Tcap expression has been associated with t-tubule loss and t-tubule recovery following both SERCA gene therapy80 and mechanical

178

unloading84. Tcap is thought to regulate t-tubules in a load dependent manner due to its positioning at the myofilaments which links the stretch-sensitive complex of the z-discs with proteins in the t-tubule membrane134. It was therefore possible in our model increased atrial pressure was associated with altered Tcap expression, which in turn controlled t-tubule regulation. In agreement with the data presented here, high wall stress in disease has previously been linked with t-tubule loss and cardiac dysfunction140, 173. It remains to be determined however, if t-tubule loss is a consequence of maladaptive alterations associated with cardiac disease or if t-tubule loss is a direct result of increased cardiac load. Data in support of the latter is provided by Wei et al50 who demonstrated that tubule remodelling in response to pressure overload occurred prior to any detectable ventricular dysfunction and the onset of heart failure. This previous study suggests that t-tubules are lost early in cardiac disease as a result of pressure overload, thus if detected early, t-tubule remodelling could be a marker for disease development. The data presented here in this study also suggests that by reducing volume overload following the onset of heart failure, atrial cellular hypertrophy can be reversed and t-tubule loss and cardiac dysfunction are able to recover, lending support to the possibility of a therapeutic role for t-tubule reverse remodelling in cardiac disease.

Despite evidence suggesting that mechanical load directly alters the t-tubule network, it is important to consider that variation and duration of load is likely to have differing effects. Ibrahim et al140 demonstrated that, although mechanical unloading was initially beneficial in the failing heart, prolonged unloading had detrimental effects on t-tubule remodelling. Furthermore, t-tubule density is preserved during physiological hypertrophy79, 85, suggesting factors other than load determine t-tubule remodelling. Future research could investigate if molecular mechanisms thought to mediate t-tubule growth, such as Tcap, BIN1 and MTM1, could directly alter cardiac load or hypertrophic growth.

In summary, this work has shown that restoration of atrial t-tubules was possible after the onset of heart failure and may provide a new and viable therapeutic strategy for normalisation of calcium handling in the failing heart. Since the discovery of an extensive t-tubule network in the human atria43 and diseases such as heart failure and

179

atrial fibrillation being associated with a reduction in atrial t-tubule density42, 46, 49. T- tubule recovery represents a useful target for future treatments of these cardiac diseases. Whilst it is not yet clear if t-tubule restoration alone, would prevent heart failure progression, there are some possible future applications of this research. Seidel et al69 found recovery of contractile function in patients was more likely when the ventricular t-tubule network was intact at the time of intervention with LVADs. The authors found that a correlation existed between the degree of t-tubule remodelling prior to cardiac unloading and subsequent recovery, suggesting that the t-tubule system could be a predictor of recovery from cardiac disease. Although, this study demonstrates that a reduction in ventricular rate, via cessation of rapid pacing, led to restoration of t-tubules and cardiac function, this is unlikely a practical treatment for t-tubule recovery. Clinically, heart rate regulation has been shown to be beneficial for treatment of arrhythmias via cardiac resynchronisation therapy220 and AV node ablation in patients with atrial fibrillation221. These treatments however, are only useful to treat symptoms caused by rapid heart rate and do not provide patients with full recovery. A future direction of this research therefore, would be to determine other methods to generate t-tubule recovery.

6.2. BIN1 led to t-tubule formation in the heart, the structures which were shaped by MTM1.

Using a sheep model of advanced sustained heart failure with subsequent recovery we were able to identify several novel proteins that could be associated with the formation or restoration of t-tubules in the atria. A more causative link between these proteins, BIN1, MTM1 and Tcap, and atrial -tubule restoration was provided in the second results chapter of this thesis whereby BIN1 transfection led to the development of tubule like structures in NRVMs. Several studies have already demonstrated that BIN1 expression can induce tubule formation in a variety of cell types90, 104, 107, 112 and more recently, my own work shows the development of tubules in NRVMs and iPSC-CMs following BIN1 expression83. The work discussed here builds on these findings and showed for the first time that MTM1 increased BIN1 induced tubule density and structure in the heart. While little is known

180

regarding the role of MTM1 in cardiac muscle, in skeletal muscle MTM1 enhanced BIN1 driven membrane tubulation121. It is therefore possible that in our recovery model, where we observed increased density and length of tubules, interactions between BIN1 and MTM1 were responsible for this t-tubule restoration. To further support this hypothesis, MTM1 mutants, which were unable to enhance BIN1 membrane tubulation in skeletal muscle, resulted in t-tubule disorder and impaired EC coupling121, a similar phenotype to our heart failure model with decreased MTM1. Altered interactions between MTM1 and BIN1 have been shown to lead to cardiomyopathies, but have not yet been linked to cardiac t-tubules. In skeletal muscle however, mutations in these genes, associated with centronuclear myopathies (CNM)108, 121, 122 and myotonic dystrophy (DM)222, affect the assembly of muscle triads, lead to t-tubule defects and weakened contraction. Skeletal DM can at least be partially attributed to mis-splicing of BIN1 (variant 8) with skipping of exon 11 resulting in an isoform of BIN1 (variant 9) that is incapable of binding PIP2 and tubulating membranes199, 201. These conformational changes in BIN1 have also been shown to alter its binding and regulation by MTM1121. Moreover, skeletal defects in mice, as a result of MTM1 KO, can be offset by BIN1 variant 8 expression, which has been shown to improve survival, t-tubule-triad organisation and localisation of the LTCC to the t-tubule223. These results suggest that mis-splicing of BIN1 alters interactions with MTM1, which plays a role in CNM through t-tubule modifications. On the contrary, the data presented in this study demonstrated that the mis-spliced variant 9 of BIN1 was capable of driving t-tubules in the heart (discussed later). Thus, it remains unknown, which, if any, of the mutations associated with CNM or DM affect BIN1-MTM1 interactions and t-tubule biogenesis in the heart. A further application of this research could be to target BIN1-MTM1 interactions to determine if these interactions are key to correct abnormalities in tubulation associated with cardiac CNMs.

To fully appreciate how MTM1 and BIN1 influence t-tubule biogenesis it is important to understand how these pathways are regulated. Whilst the exact the mechanisms that link these molecules in the heart remain unknown, there is a possible role for phosphatidylinositols (PIs). PIs are signalling molecules found in the cell membrane, which contribute to the constant recycling and regulation of

181

myocyte membranes. Both MTM1 and BIN1 are thought to interact with, or modulate, PIs and it is these interactions that are thought to influence membrane shape92, 115. Thus, more work is required to fully appreciate the mechanisms driving t-tubule formation and consideration must be given to signalling pathways such as PIs that may play a role in controlling this system.

The data presented here indicates that BIN1 is the main tubule driving factor in NRVMs, as neither Tcap nor MTM1 alone had any effect on t-tubule formation in the developing heart in the absence of BIN1, it cannot be discounted however that these proteins could play more of a role in the mature heart. It is well established that Tcap expression is highly influenced by stretch137. As cardiac load is likely to increase during development and during disease progression, due to increases in chamber volume causing an increase pressure, the role of Tcap in t-tubule control might not become apparent until later in development or even following the onset of disease. Several groups who demonstrated in Tcap KO mice, cardiac defects only developed when the heart underwent mechanical overloading, support this notion138, 139. The direct effects of these proteins on t-tubules in disease or recovery have not been studied. This study has shown for the first time however, that both BIN1 and MTM1 could contribute towards t-tubule recovery and targeting MTM1 in the atria may facilitate t-tubule restoration.

6.3. A role for BIN1 in t-tubule recovery

A major aim of this work was to characterise the role of BIN1 in t-tubule formation to determine if this protein could play a role in t-tubule restoration. By expressing BIN1 in NRVMs, which do not normally possess tubules, this study has shown that BIN1 was able to drive tubule formation. Although the tubulating properties of BIN1 have been previously studied83, 90, 104, 107, 112, the work presented here has also focused on how BIN1 induced tubules modulated the systolic calcium transient in the heart, an important consideration since t-tubules play such a key role in calcium handling. We have shown upregulation of BIN1 led to increased calcium transient amplitude in NRVMs, which we hypothesised was due to increased co-localisation between proteins of the dyad. Hong et al104 have previously shown that BIN1 was

182

able to traffic LTCC to the t-tubule membrane and recruit RyR to the junctional SR. Thus the data presented here provides further support that BIN1 is not only required for t-tubule formation in the heart but is an important regulator of calcium signalling. Furthermore, De La Mata et al107 suggest that BIN1 plays a role in dyad stabilisation by reducing SR motility. Although not studied here, the previous point highlights an interesting possibility that BIN1 could further enhance calcium signaling by playing a role in modulation of the SR membrane as well as t-tubule membrane. In support of this, we have previously observed decreased BIN1 expression46, 83 and disordered SR224 in our sheep model of heart failure. As other t-tubule associated proteins, such as MTM1 have also been proposed to play roles in SR remodeling, further study on the modulation of SR structure would be interesting.

In agreement with the work presented here, the muscle variant 8 isoform of BIN1 has previously been shown to generate membrane tubulation103, 104, 107. Together with this finding, this study has also shown that other variants of BIN1, including BIN1 variant 9, were able to induce t-tubule formation in NRVMs and iPSC-CMs. This was in contrast to work in skeletal muscle where tubule formation was dependent on the muscle specific variant 8 of BIN1201. As already discussed, in skeletal muscle, variant 9 of BIN1 was a result of mis-splicing of exon 11 which occurs in CNM and has been shown to stop membrane tubulation108, 203. As mutations in BIN1 have been reported as causative of CNM108, 182, 199, 225 and some patients displaying a cardiac phenotype108, 226, the differing roles between the BIN1 variants between cell types is surprising. Thus, an interesting direction of this research would be to determine if other BIN1 mutations that occur in CNM affect tubule formation in cardiac cells and to identify if these mutations are responsible for cardiomyopathy development. Alongside cellular differences, it also is becoming apparent that BIN1 isoform expression differs between species. Despite several studies showing that exon 11 containing variant 8 of BIN1 was able to drive tubule formation in cardiac cells83, 107, in the mouse heart, cardiac isoforms of BIN1 did not contain exon 11, instead mRNA encoding BIN1 variants 6, 9, 10 and a modified variant 9 were identified102, 182. These findings contrast our work in sheep, where we demonstrated exon 11 containing variant 8 was the most dominantly expressed isoform in the heart, we also showed expression of variants 4 and 1083. Interestingly though, BIN1 variant 8

183

protein staining has been observed in mouse cardiac muscle182, suggesting that post- translational modification can occur. Further work is required to determine the significance of species and cellular differences in BIN1 isoforms, which may provide key insight into how t-tubules are formed and maintained.

BIN1 is commonly linked with membrane curvature and t-tubule biogenesis90, 92, 99, 104, 105, 107, but more recently, BIN1 has emerged as a regulator of cardiac function in health and disease (reviewed in142). As discussed already, loss of BIN1 can be associated with human112, 227 and some animal models of heart failure46, 80, 83, and levels have been shown to be restored following recovery from heart failure in the ventricle80, 83. These findings were supported in the work presented here showing that BIN1 expression in the atria following recovery was restored to control levels. BIN1 loss in heart failure is detrimental and has been linked with reduced calcium handling and altered t-tubule density46, 80, 83, 112. Furthermore, KO of BIN1 has been shown to induce cardiomyopathies102, increase risk of arrhythmia103 and lead to contractile dysfunction112. On a cellular level, BIN1 loss led to t-tubule loss46, 108, delayed calcium transients and impaired LTCC trafficking227. Together this data highlights that compromised BIN1 contributes to the phenotype of the failing heart, thus suggesting that BIN1 could be a target for treatment of heart failure. A promising study from Hong et al227 demonstrated that myocardial BIN1 is blood detectable and decreases in patients with heart failure. Furthermore, plasma BIN1 levels predicted heart failure progression and future arrhythmia in cardiomyopathy patients. Thus, the authors proposed that BIN1 could be used as a biomarker to predict cardiac incidences.

Although the results from this study are encouraging in that they demonstrate that cessation of rapid pacing led to recovery of both cellular and whole heart function, it is important to remember that the long term effects of the initial cardiac insult are unknown and the overall damage to the heart maybe irreversible. Alongside calcium handling changes, our previously published work has shown extra cellular matrix remodelling155, altered autonomic control228 and impaired β-adrenergic responsiveness58, 228 in the sheep model of rapid pacing induced heart failure. Whilst we, and others, have demonstrated that t-tubule recovery can be associated with

184

restored catecholamine responsiveness80, 83, factors such as metabolic processes, muscle fibrosis, and susceptibility to arrhythmia have not been studied in this, and many other recovery models. To take this forward, a better approach for t-tubule restoration would be to target t-tubule loss early in disease progression, and BIN1 appears to be a novel therapeutic target for this. For example, if BIN1 could be used as a biomarker to detect early changes to t-tubule morphology, t-tubules could be recovered or maintained prior to any further cardiac dysfunction.

6.4. Overall Conclusions.

Cardiac diseases, such as heart failure, affect over half a million people in the UK alone54 and whilst there are advances in the treatment of this disease, diagnosis is poor and therapeutics remain limited54. A major contributing factor of many of these diseases is impaired calcium handling of the myocyte, which can be attributed to t- tubule remodelling50, 60, 229. Given the near complete loss of atrial t-tubules in heart failure42 and the functional consequences, this study was undertaken to investigate factors that may mediate the recovery of cardiac t-tubules and subsequent changes in calcium handling. We have shown that atrial t-tubule restoration in the sheep occurred following an almost complete loss and that these restored t-tubules were functional. Furthermore, the same proteins that were associated with t-tubule recovery were also responsible for the formation of new t-tubules structures in the developing heart. Our model has provided a unique opportunity to understand the structure and function of restored t-tubules and has also provided a greater understanding of the mechanisms that control t-tubule recovery. Thus, this research identifies several targets for the development of a therapeutic strategy for t-tubule restoration following loss in disease.

185

7. Reference List

1. Brette F, Orchard C. T-tubule function in mammalian cardiac myocytes. Circulation Research. 2003;92:1182-1192

2. Louch WE, Bito V, Heinzel FR, Macianskiene R, Vanhaecke J, Flameng W, Mubagwa K, Sipido KR. Reduced synchrony of Ca2+ release with loss of T- tubules-a comparison to Ca2+ release in human failing cardiomyocytes. Cardiovascular Research. 2004;62:63-73

3. Kawai M, Hussain M, Orchard CH. Excitation-contraction coupling in rat ventricular myocytes after formamide-induced detubulation. American Journal of Physiology. 1999;277:H603-H609

4. Franzini-Armstrong C, Protasi F, Ramesh V. Shape, size, and distribution of Ca2+ release units and couplons in skeletal and cardiac muscles. Biophysical Journal. 1999;77:1528-1539

5. Scriven DRL, Dan P, Moore EDW. Distribution of proteins implicated in excitation-contraction coupling in rat ventricular myocytes. Biophysical Journal. 2000;79:2682-2691

6. Rousseau E, Smith JS, Henderson JS, Meissner G. Single channel and Ca2+ flux measurements of cardiac sarcoplasmic reticulum calcium channel. Biophysical Journal. 1986;50:1009-1014

7. Cheng H, Lederer WJ, Cannell MB. Calcium sparks: Elementary events underlying excitation-contraction coupling in heart muscle. Science. 1993;262:740-744

8. Bers DM. Cardiac excitation-contraction coupling. Nature. 2002;415:198- 205

9. Fabiato A. Simulated calcium current can both cause calcium loading in and trigger calcium release from the sarcoplasmic reticulum of a skinned canine cardiac purkinje cell. Journal of General Physiology. 1985;85:291-320

10. Lewartowski B, Pytkowski B, Janczewski A. Calcium fraction correlating with contractile force of ventricular muscle of guinea-pig heart. Pflugers Archiv : European journal of physiology. 1984;401:198-203

11. Negretti N, Varro A, Eisner DA. Estimate of net calcium fluxes and sarcoplasmic reticulum calcium content during systole in rat ventricular myocytes. Journal of Physiology. 1995;486:581-591

12. Tada M, Inui M. Regulation of calcium transport by the ATPase- phospholamban system. Journal of Molecular and Cellular Cardiology. 1983;15:565-575

186

13. Green AL, Lalli MJ, Ji Y, Babu GJ, Grupp I, Sussman M, Periasamy M. Overexpression of SERCA2b in the heart leads to an increase in sarcoplasmic reticulum calcium transport function and increased cardiac contractility. The Journal of Biological Chemistry. 2000;275:24722-24727

14. Despa S, Brette F, Orchard CH, Bers DM. Na/Ca exchange and Na/K- ATPase function are equally concentrated in transverse tubules of rat ventricular myocytes. Biophysical Journal. 2003;85:3388-3396

15. Bassani RA, Bassani JWM, Bers DM. Mitochondrial and sarcolemmal Ca2+ 2+ transport reduce [Ca ]i during caffeine contractures in rabbit cardiac myocytes. Journal of Physiology. 1992;453:591-608

16. Negretti N, O'Neill SC, Eisner DA. The relative contributions of different intracellular and sarcolemmal systems to relaxation in rat ventricular myocytes. Cardiovascular Research. 1993;27:1826-1830

17. Eisner DA, Caldwell JL, Kistamas K, Trafford AW. Calcium and excitation- contraction coupling in the heart. Circulation Research. 2017;121:181-195

18. Soeller C, Cannell MB. Examination of the transverse tubular system in living cardiac rat myocytes by 2-photon microscopy and digital image- processing techniques. Circulation Research. 1999;84:266-275

19. Bhargava A, Lin X, Novak P, Mehta K, Korchev Y, Delmar M, Gorelik J. Super-resolution scanning patch clamp reveals clustering of functional ion channels in adult ventricular myocyte. Circulation Research. 2013;112:1112- 1120

20. Brette F, Salle L, Orchard CH. Differential modulation of L-type Ca2+ current by SR Ca2+ release at the t-tubules and surface membrane of rat ventricular myocytes. Circulation Research. 2004;95:e1-e7

21. Brette F, Salle L, Orchard CH. Quantification of calcium entry at the t- tubules and surface membrane in rat ventricular myocytes. Biophysical Journal. 2006;90:381-389

22. Asghari P, Schulson M, Scriven DR, Martens G, Moore ED. Axial tubules of rat ventricular myocytes form multiple junctions with the sarcoplasmic reticulum. Biophysical Journal. 2009;96:4651-4660

23. Yang Z, Pascarel C, Steele DS, Komukai K, Brette F, Orchard CH. Na+- Ca2+ exchange activity is localized in the t-tubules of rat ventricular myocytes. Circulation Research. 2002;91:315-322

24. Chase A, Orchard CH. Ca efflux via the sarcolemmal Ca ATPase occurs only in the t-tubules of rat ventricular myocytes. J Mol Cell Cardiol. 2011;50:187- 193

187

25. Li L, Desantiago J, Chu G, Kranias EG, Bers DM. Phosphorylation of phopholamban and troponin I in -adrenergic-induced acceleration of cardiac relaxation. American Journal of Physiology-Heart and Circulatory Physiology. 2000;278:H769-H779

26. Valdivia HH, Kaplan JH, Ellis-Davies GCR, Lederer WJ. Rapid adaptation of cardiac ryanodine receptors: Modulation by Mg2+ and phosphorylation. Science. 1995;267:1997-2000

27. Song LS, Wang SQ, Xiao RP, Spurgeon H, Lakatta EG, Cheng H. - adrenergic stimulation synchronizes intracellular Ca2+ release during excitation-contraction coupling in cardiac myocytes. Circulation Research. 2001;88:794-801

28. Nikolaev VO, Moshkov A, Lyon AR, Miragoli M, Novak P, Paur H, Lohse MJ, Korchev YE, Harding SE, Gorelik J. 2-adrenergic receptor redistribution in heart failure changes camp compartmentation. Science. 2010;327:1653-1657

29. Gao T, Puri TS, Gerhardstein BL, Chien AJ, Green RD, Hosey MM. Identification and subcellular localization of the subunits of L-type calcium channels and adenylyl cyclase in cardiac myocytes. J Biol Chem. 1997;272:19401-19407

30. Laflamme MA, Becker PL. Gs and adenylyl cyclase in transverse tubules of heart: Implications for cAMP-dependent signaling. American Journal of Physiology-Heart and Circulatory Physiology. 1999;277:H1841-H1848

31. Nikolaev VO, Bunemann M, Schmitteckert E, Lohse MJ, Engelhardt S. Cyclic AMP imaging in adult cardiac myocytes reveals far-reaching 1- adrenergic but locally confined 2-adrenergic receptor-mediated signaling. Circulation Research. 2006;99:1084-1091

32. Davare MA, Avdonin V, Hall DD, Peden EM, Burette A, Weinberg RJ, Horne MC, Hoshi T, Hell JW. A 2 adrenergic receptor signaling complex assembled with the Ca2+ channel Cav1.2. Science. 2001;293:98-101

33. Brette F, Rodriguez P, Komukai K, Colyer J, Orchard CH. Beta-adrenergic stimulation restores the ca transient of ventricular myocytes lacking t-tubules. Journal of Molecular and Cellular Cardiology. 2004;36:265-275

34. Ito DW, Hannigan KI, Ghosh D, Xu B, Del Villar SG, Xiang YK, Dickson EJ, Navedo MF, Dixon RE. Beta-adrenergic-mediated dynamic augmentation of sarcolemmal cav 1.2 clustering and co-operativity in ventricular myocytes. Journal of Physiology. 2019;597:2139-2162

35. Fu Y, Shaw SA, Naami R, Vuong CL, Basheer WA, Guo X, Hong T. Isoproterenol promotes rapid ryanodine receptor movement to bridging integrator 1 (BIN1)-organized dyads. Circulation. 2016;133:388-397

188

36. Page E. Quantitative ultrastructural analysis in cardiac membrane physiology. American Journal of Physiology. 1978;235:C147-C158

37. Heinzel FR, Bito V, Volders PGA, Antoons G, Mubagwa K, Sipido KR. Spatial and temporal inhomogeneities during Ca2+ release from the sarcoplasmic reticulum in pig ventricular myocytes. Circulation Research. 2002;91:1023-1030

38. Olivetti G, Anversa P, Loud AV. Morphometric study of early postnatal development in the left and right ventricular myocardium of the rat. Ii. Tissue composition, capillary growth, and sarcoplasmic alterations. Circulation Research. 1980;46:503-512

39. Hüser J, Lipsius SL, Blatter LA. Calcium gradients during excitation- contraction coupling in cat atrial myocytes. Journal of Physiology. 1996;494:641-651

40. Ziman AP, Gomez-Viquez NL, Bloch RJ, Lederer WJ. Excitation-contraction coupling changes during postnatal cardiac development. Journal of Molecular and Cellular Cardiology. 2010;48:379-386

41. Lipsett DB, Frisk M, Aronsen JM, Norden ES, Buonarati OR, Cataliotti A, Hell JW, Sjaastad I, Christensen G, Louch WE. Cardiomyocyte substructure reverts to an immature phenotype during heart failure. The Journal of physiology. 2019;597:1833-1853

42. Dibb KM, Clarke JD, Horn MA, Richards MA, Graham HK, Eisner DA, Trafford AW. Characterization of an extensive transverse tubular network in sheep atrial myocytes and its depletion in heart failure. Circulation: Heart Failure. 2009;2:482-489

43. Richards MA, Clarke JD, Saravanan P, Voigt N, Dobrev D, Eisner DA, Trafford AW, Dibb KM. Transverse tubules are a common feature in large mammalian atrial myocytes including human. American Journal of Physiology-Heart and Circulatory Physiology. 2011;301:H1996-H2005

44. Frisk M, Koivumaki JT, Norseng PA, Maleckar MM, Sejersted OM, Louch WE. Variable t-tubule organization and Ca2+ homeostasis across the atria. American Journal of Physiology-Heart and Circulatory Physiology. 2014;307:H609-620

45. Glukhov AV, Balycheva M, Sanchez-Alonso JL, Ilkan Z, Alvarez-Laviada A, Bhogal N, Diakonov I, Schobesberger S, Sikkel MB, Bhargava A, Faggian G, Punjabi PP, Houser SR, Gorelik J. Direct evidence for microdomain-specific localization and remodeling of functional L-type calcium channels in rat and human atrial myocytes. Circulation. 2015;132:2372-2384

189

46. Caldwell JL, Smith CE, Taylor RF, Kitmitto A, Eisner DA, Dibb KM, Trafford AW. Dependence of cardiac transverse tubules on the BAR domain protein Amphiphysin II (BIN1). Circulation Research. 2014;115:986-996

47. Eisner DA, Dibb KM, Trafford AW. The mechanism and significance of the slow changes of ventricular action potential duration following a change of heart rate. Experimental Physiology. 2009;94:520-528

48. Louch WE, Mork HK, Sexton J, Stromme TA, Laake P, Sjaastad I, Sejersted OM. T-tubule disorganization and reduced synchrony of Ca2+ release in murine cardiomyocytes following myocardial infarction. Journal of Physiology. 2006;574:519-533

49. Lenaerts I, Bito V, Heinzel FR, Driesen RB, Holemans P, D'hooge J, Heidbuchel H, Sipido KR, Willems R. Ultrastructural and functional remodeling of the coupling between Ca2+ influx and sarcoplasmic reticulum Ca2+ release in right atrial myocytes from experimental persistent atrial fibrillation. Circulation Research. 2009;105:876-885

50. Wei S, Guo A, Chen B, Kutschke W, Xie YP, Zimmerman K, Weiss RM, Anderson ME, Cheng H, Song LS. T-tubule remodeling during transition from hypertrophy to heart failure. Circulation Research. 2010;107:520-531

51. He J, Conklin MW, Foell JD, Wolff MR, Haworth RA, Coronado R, Kamp TJ. Reduction in density of transverse tubules and L-type Ca2+ channels in canine tachycardia-induced heart failure. Cardiovascular Research. 2001;49:298-307

52. Hasenfuss G, Reinecke H, Studer R, Pieske B, Meyer M, Drexler H, Just H. Calcium cycling proteins and forcefrequency relationship in heart failure. Basic Research in Cardiology. 1996;91:17-22

53. Lloyd-Jones DM, Larson MG, Leip EP, Beiser A, D'Agostino RB, Kannel WB, Murabito JM, Vasan RS, Benjamin EJ, Levy D, Framingham Heart S. Lifetime risk for developing congestive heart failure: The framingham heart study. Circulation. 2002;106:3068-3072

54. Bhatnagar P, Wickramasinghe K, Wilkins E, Townsend N. Trends in the epidemiology of cardiovascular disease in the uk. Heart. 2016;102:1945- 1952

55. Mulieri LA, Hasenfuss G, Leavitt B, Allen PD, Alpert NR. Altered myocardial force-frequency relation in human heart failure. Circulation. 1992;85:1743-1750

56. Gwathmey JK, Copelas L, MacKinnon R, Schoen FJ, Feldman MD, Grossman W, Morgan JP. Abnormal intracellular calcium handling in myocardium from patients with end-stage heart failure. Circulation Research. 1987;61:70-76

190

57. Pogwizd SM, Schlotthauer K, Li L, Yuan W, Bers DM. Arrhythmogenesis and contractile dysfunction in heart failure : Roles of sodium-calcium exchange, inward rectifier potassium current, and residual -adrenergic responsiveness. Circulation Research. 2001;88:1159-1167

58. Briston SJ, Caldwell JL, Horn MA, Clarke JD, Richards MA, Greensmith DJ, Graham HK, Hall MC, Eisner DA, Dibb KM, Trafford AW. Impaired - adrenergic responsiveness accentuates dysfunctional excitation contraction coupling in an ovine model of tachypacing induced heart failure. Journal of Physiology. 2011;589:1367-1382

59. Eisner D, Caldwell J, Trafford A. Sarcoplasmic reticulum Ca-ATPase and heart failure 20 years later. Circulation Research. 2013;113:958-961

60. Lyon AR, MacLeod KT, Zhang Y, Garcia E, Kanda GK, Lab MJ, Korchev YE, Harding SE, Gorelik J. Loss of t-tubules and other changes to surface topography in ventricular myocytes from failing human and rat heart. Proc Natl Acad Sci USA. 2009;106:6854-6859

61. Cannell MB, Crossman DJ, Soeller C. Effect of changes in action potential spike configuration, junctional sarcoplasmic reticulum micro-architecture and altered t-tubule structure in human heart failure. Journal of Muscle Research and Cell Motility. 2006;27:297-306

62. Crossman DJ, Ruygrok PN, Soeller C, Cannell MB. Changes in the organization of excitation-contraction coupling structures in failing human heart. PLoS ONE. 2011;6:e17901

63. Balijepalli RC, Lokuta AJ, Maertz NA, Buck JM, Haworth RA, Valdivia HH, Kamp TJ. Depletion of t-tubules and specific subcellular changes in sarcolemmal proteins in tachycardia-induced heart failure. Cardiovascular Research. 2003;59:67-77

64. Huang CK, Chen BY, Guo A, Chen R, Zhu YQ, Kutschke W, Hong J, Song LS. Sildenafil ameliorates left ventricular t-tubule remodeling in a pressure overload-induced murine heart failure model. Acta Pharmacologica Sinica. 2016;37:473-482

65. Song LS, Sobie EA, McCulle S, Lederer WJ, Balke CW, Cheng H. Orphaned ryanodine receptors in the failing heart. Proc Natl Acad Sci USA. 2006;103:4305-4310

66. He JQ, Balijepalli RC, Haworth RA, Kamp TJ. Crosstalk of -adrenergic 2+ receptor subtypes through gi blunts -adrenergic stimulation of L-type Ca channels in canine heart failure. Circulation Research. 2005;97:566-573

67. Heinzel FR, Bito V, Biesmans L, Wu M, Detre E, von WF, Claus P, Dymarkowski S, Maes F, Bogaert J, Rademakers F, D'hooge J, Sipido K. Remodeling of t-tubules and reduced synchrony of Ca2+ release in myocytes

191

from chronically ischemic myocardium. Circulation Research. 2008;102:338-346

68. Crossman DJ, Ruygrok PR, Soeller C, Cannell MB. Changes in the organization of excitation-contraction coupling structures in failing human heart. PLoS ONE. 2011;6:e17901

69. Seidel T, Navankasattusas S, Ahmad A, Diakos NA, Xu WD, Tristani- Firouzi M, Bonios MJ, Taleb I, Li DY, Selzman CH, Drakos SG, Sachse FB. Sheet-like remodeling of the transverse tubular system in human heart failure impairs excitation-contraction coupling and functional recovery by mechanical unloading. Circulation. 2017;135:1632-1645

70. Hauser J, Michel-Behnke I, Zervan K, Pees C. Noninvasive measurement of atrial contribution to the cardiac output in children and adolescents with congenital complete atrioventricular block treated with dual-chamber pacemakers. The American Journal of Cardiology. 2011;107:92-95

71. Kono T, Sabbah HN, Rosman H, Alam M, Stein PD, Goldstein S. Left atrial contribution to ventricular filling during the course of evolving heart failure. Circulation. 1992;86:1317-1322

72. Heijman J, Voigt N, Nattel S, Dobrev D. Cellular and molecular electrophysiology of atrial fibrillation initiation, maintenance, and progression. Circulation Research. 2014;114:1483-1499

73. Gomez AM, Valdivia HH, Cheng H, Lederer MR, Santana LF, Cannell MB, McCune SA, Altschuld RA, Lederer WJ. Defective excitation-contraction coupling in experimental cardiac hypertrophy and heart failure. Science. 1997;276:800-806

74. He J-Q, Conklin MW, Foell JD, Wolff MR, Haworth RA, Coronado R, Kamp TJ. Reduction in density of transverse tubules and L-type Ca2+ channels in canine tachycardia-induced heart failure. Cardiovascular Research. 2001;49:298-307

75. Takagishi Y, Yasui K, Severs NJ, Murata Y. Species-specific difference in distribution of voltage-gated L-type Ca2+ channels of cardiac myocytes. American Journal of Physiology-Cell Physiology. 2000;279:C1963-C1969

76. Chen X, Piacentino V, III, Furukawa S, Goldman B, Margulies KB, Houser SR. L-type Ca2+ channel density and regulation are altered in failing human ventricular myocytes and recover after support with mechanical assist devices. Circulation Research. 2002;91:517-524

77. Trafford AW, Díaz ME, Negretti N, Eisner DA. Enhanced ca2+ current and decreased Ca2+ efflux restore sarcoplasmic reticulum Ca2+ content following depletion. Circulation Research. 1997;81:477-484

192

78. Brette F, Despa S, Bers DM, Orchard CH. Spatiotemporal characteristics of SR Ca2+ uptake and release in detubulated rat ventricular myocytes. Journal of Molecular and Cellular Cardiology. 2005;39:804-812

79. Kemi OJ, Hoydal MA, MacQuaide N, Haram PM, Koch LG, Britton SL, Ellingsen O, Smith GL, Wisloff U. The effect of exercise training on transverse tubules in normal, remodeled, and reverse remodeled hearts. Journal of Cellular Physiology. 2011

80. Lyon AR, Nikolaev VO, Miragoli M, Sikkel MB, Paur H, Benard L, Hulot JS, Kohlbrenner E, Hajjar RJ, Peters NS, Korchev YE, MacLeod KT, Harding SE, Gorelik J. Plasticity of surface structures and beta2-adrenergic receptor localization in failing ventricular cardiomyocytes during recovery from heart failure. Circulation Heart failure. 2012

81. Chen B, Li Y, Jiang S, Xie YP, Guo A, Kutschke W, Zimmerman K, Weiss RM, Miller FJ, Anderson ME, Song LS. -adrenergic receptor antagonists ameliorate myocyte t-tubule remodeling following myocardial infarction. FASEB Journal. 2012;26:2531-2537

82. Xie YP, Chen B, Sanders P, Guo A, Li Y, Zimmerman K, Wang LC, Weiss RM, Grumbach IM, Anderson ME, Song LS. Sildenafil prevents and reverses transverse-tubule remodeling and Ca2+ handling dysfunction in right ventricle failure induced by pulmonary artery hypertension. Hypertension. 2012;59:355-362

83. Lawless M, Caldwell JL, Radcliffe EJ, Smith CER, Madders GWP, Hutchings DC, Woods LS, Church SJ, Unwin RD, Kirkwood GJ, Becker LK, Pearman CM, Taylor RF, Eisner DA, Dibb KM, Trafford AW. Phosphodiesterase 5 inhibition improves contractile function and restores transverse tubule loss and catecholamine responsiveness in heart failure. Scientific Reports. 2019;9:6801

84. Ibrahim M, Navaratnarajah M, Siedlecka U, Rao C, Dias P, Moshkov AV, Gorelik J, Yacoub MH, Terracciano CM. Mechanical unloading reverses transverse tubule remodelling and normalizes local Ca2+ induced Ca2+ release in a rodent model of heart failure. European Journal of Heart Failure. 2012;14:571-580

85. Stolen TO, Hoydal MA, Kemi OJ, Catalucci D, Ceci M, Aasum E, Larsen T, Rolim N, Condorelli G, Smith GL, Wisloff U. Interval training normalizes cardiomyocyte function, diastolic Ca2+ control, and SR Ca2+ release synchronicity in a mouse model of diabetic cardiomyopathy. Circulation Research 2009;105:527-536

86. Giannetta E, Feola T, Gianfrilli D, Pofi R, Dall'Armi V, Badagliacca R, Barbagallo F, Lenzi A, Isidori AM. Is chronic inhibition of phosphodiesterase type 5 cardioprotective and safe? A meta-analysis of randomized controlled trials. BMC Medicine. 2014;12:185

193

87. Guazzi M, Vicenzi M, Arena R, Guazzi MD. Pde5 inhibition with sildenafil improves left ventricular diastolic function, cardiac geometry, and clinical status in patients with stable systolic heart failure: Results of a 1-year, prospective, randomized, placebo-controlled study. Circulation. Heart failure. 2011;4:8-17

88. Sachse FB, Torres NS, Savio-Galimberti E, Aiba T, Kass DA, Tomaselli GF, Bridge JH. Subcellular structures and function of myocytes impaired during heart failure are restored by cardiac resynchronization therapy. Circulation Research. 2012;110:588-597

89. De Matteis MA, Godi A. PI-loting membrane traffic. Nature Cell Biology. 2004;6:487-492

90. Lee E, Marcucci M, Daniell L, Pypaert M, Weisz OA, Ochoa GC, Farsad K, Wenk MR, De Camilli P. Amphiphysin 2 (BIN1) and t-tubule biogenesis in muscle. Science. 2002;297:1193-1196

91. Rozelle AL, Machesky LM, Yamamoto M, Driessens MH, Insall RH, Roth MG, Luby-Phelps K, Marriott G, Hall A, Yin HL. Phosphatidylinositol 4,5- bisphosphate induces actin-based movement of raft-enriched vesicles through WASP-ARP2/3. Current Biology. 2000;10:311-320

92. Yoon Y, Zhang X, Cho W. Phosphatidylinositol 4,5-bisphosphate (Ptdins(4,5)p2) specifically induces membrane penetration and deformation by bin/amphiphysin/rvs (BAR) domains. The Journal of Biological Chemistry. 2012;287:34078-34090

93. Kobayashi M, Muroyama A, Ohizumi Y. Phosphatidylinositol 4,5- bisphosphate enhances calcium release from sarcoplasmic reticulum of skeletal muscle. Biochemical and Biophysical Research Communications. 1989;163:1487-1491

94. Chu A, Stefani E. Phosphatidylinositol 4,5-bisphosphate-induced Ca2+ release from skeletal muscle sarcoplasmic reticulum terminal cisternal membranes. Ca2+ flux and single channel studies. The Journal of Biological Chemistry. 1991;266:7699-7705

95. Hilgemann DW, Feng S, Nasuhoglu C. The complex and intriguing lives of PIP2 with ion channels and transporters. Science Signalling 2001;2001:re19

96. Pretorius L, Owen KL, McMullen JR. Role of phosphoinositide 3-kinases in regulating cardiac function. Frontiers in bioscience. 2009;14:2221-2229

97. Wu CY, Jia Z, Wang W, Ballou LM, Jiang YP, Chen B, Mathias RT, Cohen IS, Song LS, Entcheva E, Lin RZ. PI3Ks maintain the structural integrity of t-tubules in cardiac myocytes. PLoS ONE. 2011;6:e24404

194

98. Kutchukian C, Lo Scrudato M, Tourneur Y, Poulard K, Vignaud A, Berthier C, Allard B, Lawlor MW, Buj-Bello A, Jacquemond V. Phosphatidylinositol 3-kinase inhibition restores Ca2+ release defects and prolongs survival in myotubularin-deficient mice. Proc Natl Acad Sci USA. 2016;113:14432- 14437

99. Dawson JC, Legg JA, Machesky LM. Bar domain proteins: A role in tubulation, scission and actin assembly in clathrin-mediated endocytosis. Trends in cell biology. 2006;16:493-498

100. Masuda M, Mochizuki N. Structural characteristics of BAR domain superfamily to sculpt the membrane. Seminars in Cell and Developmental Biology. 2010;21:391-398

101. Butler MH, David C, Ochoa GC, Freyberg Z, Daniell L, Grabs D, Cremona O, Camilli PD. Amphiphysin II (SH3p9; BIN1), a member of the amphiphysin/rvs family, is concentrated in the cortical cytomatrix of axon initial segments and nodes of ranvier in brain and around t tubules in skeletal muscle. The Journal of Cell Biology. 1997;137:1355-1367

102. Hong T, Yang H, Zhang SS, Cho HC, Kalashnikova M, Sun B, Zhang H, Bhargava A, Grabe M, Olgin J, Gorelik J, Marban E, Jan LY, Shaw RM. Cardiac BIN1 folds t-tubule membrane, controlling ion flux and limiting arrhythmia. Nature medicine. 2014;20:624-632

103. Laury-Kleintop LD, Mulgrew JR, Heletz I, Nedelcoviciu RA, Chang MY, Harris DM, Koch WJ, Schneider MD, Muller AJ, Prendergast GC. Cardiac- specific disruption of BIN1 in mice enables a model of stress- and age- associated dilated cardiomyopathy. Journal of Cellular Biochemistry. 2015;116:2541-2551

104. Hong TT, Smyth JW, Gao D, Chu KY, Vogan JM, Fong TS, Jensen BC, Colecraft HM, Shaw RM. Bin1 localizes the l-type calcium channel to cardiac t-tubules. PLoS Biology. 2010;8:e1000312

105. Zhao X, Wang D, Liu X, Liu L, Song Z, Zhu T, Adams G, Gao X, Tian R, Huang Y, Chen R, Wang F, Liu D, Yu X, Chen Y, Chen Z, Teng M, Ding X, Yao X. Phosphorylation of the Bin, Amphiphysin, and RSV161/167 (BAR) domain of ACAP4 regulates membrane tubulation. Proc Natl Acad Sci USA. 2013;110:11023-11028

106. Lieu DK, Liu J, Siu CW, McNerney GP, Tse HF, bu-Khalil A, Huser TR, Li RA. Absence of transverse tubules contributes to non-uniform Ca2+ wavefronts in mouse and human embryonic stem cell-derived cardiomyocytes. Stem Cells and Development. 2009

107. De La Mata A, Tajada S, O'Dwyer S, Matsumoto C, Dixon RE, Hariharan N, Moreno CM, Santana LF. Bin1 iinduces the formation of t-tubules and adult-

195

like Ca2+ release units in developing cardiomyocytes. Stem Cells. 2019;37:54-64

108. Nicot AS, Toussaint A, Tosch V, Kretz C, Wallgren-Pettersson C, Iwarsson E, Kingston H, Garnier JM, Biancalana V, Oldfors A, Mandel JL, Laporte J. Mutations in amphiphysin 2 (BIN1) disrupt interaction with dynamin 2 and cause autosomal recessive centronuclear myopathy. Nature Genetics. 2007;39:1134-1139

109. Bohm J, Yis U, Ortac R, Cakmakci H, Kurul SH, Dirik E, Laporte J. Case report of intrafamilial variability in autosomal recessive centronuclear myopathy associated to a novel bin1 stop mutation. Orphanet journal of rare diseases. 2010;5:35

110. Verhiest W, Brucher JM, Goddeeris P, Lauweryns J, De Geest H. Familial centronuclear myopathy associated with 'cardiomyopathy'. British heart journal. 1976;38:504-509

111. Muller AJ, Baker JF, DuHadaway JB, Ge K, Farmer G, Donover PS, Meade R, Reid C, Grzanna R, Roach AH, Shah N, Soler AP, Prendergast GC. Targeted disruption of the murine BIN1/amphiphysin II gene does not disable endocytosis but results in embryonic cardiomyopathy with aberrant myofibril formation. Journal of Molecular Cell Biology. 2003;23:4295-4306

112. Hong TT, Smyth JW, Chu KY, Vogan JM, Fong TS, Jensen BC, Fang K, Halushka MK, Russell SD, Colecraft H, Hoopes CW, Ocorr K, Chi NC, Shaw RM. Bin1 is reduced and Cav1.2 trafficking is impaired in human failing cardiomyocytes. Heart Rhythm. 2012;9:812-820

113. Razzaq A, Robinson IM, McMahon HT, Skepper JN, Su Y, Zelhof AC, Jackson AP, Gay NJ, O'Kane CJ. Amphiphysin is necessary for organization of the excitation-contraction coupling machinery of muscles, but not for synaptic vesicle endocytosis in drosophila. Genes & Development. 2001;15:2967-2979

114. Meunier B, Quaranta M, Daviet L, Hatzoglou A, Leprince C. The membrane- tubulating potential of Amphiphysin 2/BIN1 is dependent on the microtubule-binding cytoplasmic linker protein 170 (CLIP-170). European Journal of Cell Biology. 2009;88:91-102

115. Nicot AS, Laporte J. Endosomal phosphoinositides and human diseases. Traffic. 2008;9:1240-1249

116. Dowling JJ, Vreede AP, Low SE, Gibbs EM, Kuwada JY, Bonnemann CG, Feldman EL. Loss of myotubularin function results in t-tubule disorganization in zebrafish and human myotubular myopathy. PLoS Genetics. 2009;5:e1000372

196

117. GeBlondeau F, Laporte J, Bodin S, Superti-Furga G, Payrastre B, Mandel JL. Myotubularin, a phosphatase deficient in myotubular myopathy, acts on phosphatidylinositol 3-kinase and phosphatidylinositol 3-phosphate pathway. Human Molecular Genetics. 2000;9:2223-2229

118. Hohendahl A, Roux A, Galli V. Structural insights into the centronuclear myopathy-associated functions of bin1 and dynamin 2. Journal of structural biology. 2016;196:37-47

119. Amoasii L, Hnia K, Chicanne G, Brech A, Cowling BS, Muller MM, Schwab Y, Koebel P, Ferry A, Payrastre B, Laporte J. Myotubularin and Ptdins3P remodel the sarcoplasmic reticulum in muscle in vivo. Journal of Cell Science. 2013;126:1806-1819

120. Al-Qusairi L, Weiss N, Toussaint A, Berbey C, Messaddeq N, Kretz C, Sanoudou D, Beggs AH, Allard B, Mandel JL, Laporte J, Jacquemond V, Buj-Bello A. T-tubule disorganization and defective excitation-contraction coupling in muscle fibers lacking myotubularin lipid phosphatase. Proc Natl Acad Sci USA. 2009;106:18763-18768

121. Royer B, Hnia K, Gavriilidis C, Tronchere H, Tosch V, Laporte J. The myotubularin-amphiphysin 2 complex in membrane tubulation and centronuclear myopathies. EMBO reports. 2013;14:907-915

122. Agrawal PB, Pierson CR, Joshi M, Liu X, Ravenscroft G, Moghadaszadeh B, Talabere T, Viola M, Swanson LC, Haliloglu G, Talim B, Yau KS, Allcock RJ, Laing NG, Perrella MA, Beggs AH. Speg interacts with myotubularin, and its deficiency causes centronuclear myopathy with dilated cardiomyopathy. American journal of human genetics. 2014;95:218-226

123. Garbino A, van Oort RJ, Dixit SS, Landstrom AP, Ackerman MJ, Wehrens XH. Molecular evolution of the junctophilin gene family. Physiological genomics. 2009;37:175-186

124. Chen B, Guo A, Zhang C, Chen R, Zhu Y, Hong J, Kutschke W, Zimmerman K, Weiss RM, Zingman L, Anderson ME, Wehrens XH, Song LS. Critical roles of Junctophilin-2 in t-tubule and excitation-contraction coupling maturation during postnatal development. Cardiovascular Research. 2013

125. Reynolds JO, Chiang DY, Wang W, Beavers DL, Dixit SS, Skapura DG, Landstrom AP, Song LS, Ackerman MJ, Wehrens XH. Junctophilin-2 is necessary for t-tubule maturation during mouse heart development. Cardiovascular Research. 2013

126. Takeshima H, Komazaki S, Nishi M, Iino M, Kangawa K. Junctophilins: A novel family of junctional membrane complex proteins. Molecular Cell. 2000;6:11-22

197

127. Landstrom AP, Kellen CA, Dixit SS, van Oort RJ, Garbino A, Weisleder N, Ma J, Wehrens XH, Ackerman MJ. Junctophilin-2 expression silencing causes cardiocyte hypertrophy and abnormal intracellular calcium-handling. Circulation Heart failure. 2011;4:214-223

128. van Oort RJ, Garbino A, Wang W, Dixit SS, Landstrom AP, Gaur N, De Almeida AC, Skapura DG, Rudy Y, Burns AR, Ackerman MJ, Wehrens XH. Disrupted junctional membrane complexes and hyperactive ryanodine receptors after acute junctophilin knockdown in mice. Circulation. 2011;123:979-988

129. Xu M, Wu HD, Li RC, Zhang HB, Wang M, Tao J, Feng XH, Guo YB, Li SF, Lai ST, Zhou P, Li LL, Yang HQ, Luo GZ, Bai Y, Xi JJ, Gao W, Han QD, Zhang YY, Wang XJ, Meng X, Wang SQ. Mir-24 regulates Junctophilin-2 expression in cardiomyocytes. Circulation Research. 2012;111:837-841

130. Wagner E, Lauterbach MA, Kohl T, Westphal V, Williams GS, Steinbrecher JH, Streich JH, Korff B, Tuan HT, Hagen B, Luther S, Hasenfuss G, Parlitz U, Jafri MS, Hell SW, Lederer WJ, Lehnart SE. Stimulated emission depletion live-cell super-resolution imaging shows proliferative remodeling of t-tubule membrane structures after myocardial infarction. Circulation Research. 2012;111:402-414

131. Zhang HB, Li RC, Xu M, Xu SM, Lai YS, Wu HD, Xie XJ, Gao W, Ye H, Zhang YY, Meng X, Wang SQ. Ultrastructural uncoupling between t-tubules and sarcoplasmic reticulum in human heart failure. Cardiovascular Research. 2013;98:269-276

132. Liang X, Mei Y, Huang X, Shen G, Zhu D, Yu Y, Wang J, Lou Y. Junctophilin 2 knockdown interfere with mitochondrium status in ESC-CMs and cardiogenesis of es cells. Journal of Cellular Biochemistry. 2012;113:2884-2894

133. van Rooij E, Sutherland LB, Liu N, Williams AH, McAnally J, Gerard RD, Richardson JA, Olson EN. A signature pattern of stress-responsive micrornas that can evoke cardiac hypertrophy and heart failure. Proc Natl Acad Sci USA. 2006;103:18255-18260

134. Valle G, Faulkner G, De Antoni A, Pacchioni B, Pallavicini A, Pandolfo D, Tiso N, Toppo S, Trevisan S, Lanfranchi G. Telethonin, a novel sarcomeric protein of heart and skeletal muscle. FEBS Letters. 1997;415:163-168

135. Furukawa T, Ono Y, Tsuchiya H, Katayama Y, Bang ML, Labeit D, Labeit S, Inagaki N, Gregorio CC. Specific interaction of the potassium channel beta-subunit minK with the sarcomeric protein T-cap suggests a t-tubule- myofibril linking system. Journal of molecular biology. 2001;313:775-784

198

136. Hayashi T, Arimura T, Itoh-Satoh M, Ueda K, Hohda S, Inagaki N, Takahashi M, Hori H, Yasunami M, Nishi H, Koga Y, Nakamura H, Matsuzaki M, Choi BY, Bae SW, You CW, Han KH, Park JE, Kn+Âll R, Hoshijima M, Chien KR, Kimura A. Tcap gene mutations in hypertrophic cardiomyopathy and dilated cardiomyopathy. Journal of the American College of Cardiology. 2004;44:2192-2201

137. Zhang R, Yang J, Zhu J, Xu X. Depletion of zebrafish Tcap leads to muscular dystrophy via disrupting sarcomere  membrane interaction, not sarcomere assembly. Human Molecular Genetics. 2009;18:4130-4140

138. Knoll R, Linke WA, Zou P, S M, Kostin S, Buyandelger B, Ku CH, Neef S, Bug M, Schafer K, Knoll G, Felkin LE, Wessels J, Toischer K, Hagn F, Kessler H, Didie M, Quentin T, Maier LS, Teucher N, Unsold B, Schmidt A, Birks EJ, Gunkel S, Lang P, Granzier H, Zimmermann WH, Field LJ, Faulkner G, Dobbelstein M, Barton PJR, Sattler M, Wilmanns M, Chien KR. Telethonin deficiency is associated with maladaptation to biomechanical stress in the mammalian heart. Circulation Research. 2011;109:758-769

139. Ibrahim M, Siedlecka U, Buyandelger B, Harada M, Rao C, Moshkov A, Bhargava A, Schneider M, Yacoub MH, Gorelik J, Knoll R, Terracciano CM. A critical role for telethonin in regulating t-tubule structure and function in the mammalian heart. Human Molecular Genetics. 2013;22:372-383

140. Ibrahim M, Al Masri A, Navaratnarajah M, Siedlecka U, Soppa GK, Moshkov A, Al-Saud SA, Gorelik J, Yacoub MH, Terracciano CMN. Prolonged mechanical unloading affects cardiomyocyte excitation- contraction coupling, transverse-tubule structure, and the cell surface. The FASEB Journal. 2010;24:3321-3329

141. Candasamy AJ, Haworth RS, Cuello F, Ibrahim M, Aravamudhan S, Kruger M, Holt MR, Terracciano CM, Mayr M, Gautel M, Avkiran M. Phosphoregulation of the titin-cap protein telethonin in cardiac myocytes. The Journal of Biological Chemistry. 2014;289:1282-1293

142. Hong T, Shaw RM. Cardiac t-tubule microanatomy and function. Physiological Reviews. 2017;97:227-252

143. Frank D, Kuhn C, Katus HA, Frey N. The sarcomeric z-disc: A nodal point in signalling and disease. Journal of molecular medicine. 2006;84:446-468

144. Vlahovich N, Kee AJ, Van der Poel C, Kettle E, Hernandez-Deviez D, Lucas C, Lynch GS, Parton RG, Gunning PW, Hardeman EC. Cytoskeletal tropomyosin TM5NM1 is required for normal excitation-contraction coupling in skeletal muscle. Molecular Biology of the Cell. 2009;20:400-409

145. Lazarides E, Hubbard BD. Immunological characterization of the subunit of the 100 a filaments from muscle cells. Proc Natl Acad Sci USA. 1976;73:4344-4348

199

146. Terracio L, Rubin K, Gullberg D, Balog E, Carver W, Jyring R, Borg TK. Expression of collagen binding integrins during cardiac development and hypertrophy. Circulation Research. 1991;68:734-744

147. Messina DA, Lemanski LF. Studies of hamster cardiac myofibrillogenesis in vivo with antibodies to spectrin, desmin, and alpha-actinin. The American Journal of Anatomy. 1991;191:85-94

148. Carpenter S, Karpati G, Zubrzycka-Gaarn E, Bulman DE, Ray PN, Worton RG. Dystrophin is localized to the plasma membrane of human skeletal muscle fibers by electron-microscopic cytochemical study. Muscle Nerve. 1990;13:376-380

149. Sharov VG, Kostin S, Todor A, Schaper J, Sabbah HN. Expression of cytoskeletal, linkage and extracellular proteins in failing dog myocardium. Heart failure reviews. 2005;10:297-303

150. Leach RN, Desai JC, Orchard CH. Effect of cytoskeleton disruptors on L- type Ca channel distribution in rat ventricular myocytes. Cell calcium. 2005;38:515-526

151. Takano K, Watanabe-Takano H, Suetsugu S, Kurita S, Tsujita K, Kimura S, Karatsu T, Takenawa T, Endo T. Nebulin and N-WASP cooperate to cause IGF-1-induced sarcomeric actin filament formation. Science. 2010;330:1536- 1540

152. Iwata M, Maturana A, Hoshijima M, Tatematsu K, Okajima T, Vandenheede JR, Van Lint J, Tanizawa K, Kuroda S. PKCepsilon-PKD1 signaling complex at Z-discs plays a pivotal role in the cardiac hypertrophy induced by G-protein coupling receptor agonists. Biochemical and Biophysical Research Communications. 2005;327:1105-1113

153. Frank D, Frey N. Cardiac z-disc signaling network. The Journal of Biological Chemistry. 2011;286:9897-9904

154. Clarke JD, Caldwell JL, Horn MA, Bode EF, Richards MA, Hall MC, Graham HK, Briston SJ, Greensmith DJ, Eisner DA, Dibb KM, Trafford AW. Perturbed atrial calcium handling in an ovine model of heart failure: Potential roles for reductions in the L-type calcium current. Journal of Molecular and Cellular Cardiology. 2015;79:169-179

155. Horn MA, Graham HK, Richards MA, Clarke JD, Greensmith DJ, Briston SJ, Hall MCS, Dibb KM, Trafford AW. Age-related divergent remodeling of the cardiac extracellular matrix in heart failure: Collagen accumulation in the young and loss in the aged. Journal of Molecular and Cellular Cardiology. 2012;53:82-90

156. Crivat G, Taraska JW. Imaging proteins inside cells with fluorescent tags. Trends in biotechnology. 2012;30:8-16

200

157. Maehama T, Taylor GS, Dixon JE. Pten and myotubularin: Novel phosphoinositide phosphatases. Annual review of biochemistry. 2001;70:247- 279

158. Paredes RM, Etzler JC, Watts LT, Zheng W, Lechleiter JD. Chemical calcium indicators. Methods. 2008;46:143-151

159. Lock JT, Parker I, Smith IF. A comparison of fluorescent Ca2+ indicators for imaging local Ca2+ signals in cultured cells. Cell calcium. 2015;58:638-648

160. Di Virgilio F, Steinberg TH, Silverstein SC. Inhibition of Fura-2 sequestration and secretion with organic anion transport blockers. Cell calcium. 1990;11:57-62

161. Zygmunt AC, Gibbons WR. Properties of the calcium-activated chloride current in heart. Journal of General Physiology. 1992;99:391-414

162. Manders EMM, Verbeek FJ, Aten JA. Measurement of co-localization of objects in dual-colour confocal images. Journal of Microscopy. 1993;169:375-382

163. Dunn KW, Kamocka MM, McDonald JH. A practical guide to evaluating colocalization in biological microscopy. American journal of physiology. Cell physiology. 2011;300:C723-742

164. Owen F, Jones R. Statistics. London: Pitman; 1994.

165. Yarbrough WM, Spinale FG. Large animal models of congestive heart failure: A critical step in translating basic observations into clinical applications. Journal of Nuclear Cardiology. 2003;10:77-86

166. Moe GW, Armstrong P. Pacing-induced heart failure: A model to study the mechanism of disease progression and novel therapy in heart failure. Cardiovascular Research. 1999;42:591-599

167. Byrne MJ, Raman JS, Alferness CA, Esler MD, Kaye DM, Power JM. An ovine model of tachycardia-induced degenerative dilated cardiomyopathy and heart failure with prolonged onset. Journal of Cardiac Failure. 2002;8:108-115

168. Wilson JR, Douglas P, Hickey WF, Lanoce V, Ferraro N, Muhammad A, Reichek N. Experimental congestive-heart-failure produced by rapid ventricular pacing in the dog - cardiac effects. Circulation. 1987;75:857-867

169. Vaziri SM, Larson MG, Benjamin EJ, Levy D. Echocardiographic predictors of nonrheumatic atrial fibrillation. The framingham heart study. Circulation. 1994;89:724-730

201

170. Howard RJ, Stopps TP, Moe GW, Gotlieb A, Armstrong PW. Recovery from heart failure: Structural and functional analysis in a canine model. Canadian journal of physiology and pharmacology. 1988;66:1505-1512

171. Katz AM. Maladaptive growth in the failing heart: The cardiomyopathy of overload. Cardiovascular drugs and therapy. 2002;16:245-249

172. McNary TG, Bridge JH, Sachse FB. Strain transfer in ventricular cardiomyocytes to their transverse tubular system revealed by scanning confocal microscopy. Biophysical journal. 2011;100:L53-55

173. Frisk M, Ruud M, Espe EK, Aronsen JM, Roe AT, Zhang L, Norseng PA, Sejersted OM, Christensen GA, Sjaastad I, Louch WE. Elevated ventricular wall stress disrupts cardiomyocyte t-tubule structure and calcium homeostasis. Cardiovascular research. 2016;112:443-451

174. Swift F, Franzini-Armstrong C, Oyehaug L, Enger UH, Andersson KB, Christensen G, Sejersted OM, Louch WE. Extreme sarcoplasmic reticulum volume loss and compensatory t-tubule remodeling after SERCA2 knockout. Proc Natl Acad Sci U.S.A. 2012;109:3997-4001

175. Pinali C, Bennett HJ, Davenport JB, Caldwell JL, Starborg T, Trafford AW, Kitmitto A. Three-dimensional structure of the intercalated disc reveals plicate domain and gap junction remodeling in heart failure. Biophysical journal. 2015;108:498-507

176. Kaprielian RR, Stevenson S, Rothery SM, Cullen MJ, Severs NJ. Distinct patterns of dystrophin organization in myocyte sarcolemma and transverse tubules of normal and diseased human myocardium. Circulation. 2000;101:2586-2594

177. Kolstad TR, van den Brink J, MacQuaide N, Lunde PK, Frisk M, Aronsen JM, Norden ES, Cataliotti A, Sjaastad I, Sejersted OM, Edwards AG, Lines GT, Louch WE. Ryanodine receptor dispersion disrupts Ca2+ release in failing cardiac myocytes. Elife. 2018;7

178. Bond RC, Bryant SM, Watson JJ, Hancox JC, Orchard CH, James AF. Reduced density and altered regulation of rat atrial L-type Ca2+ current in heart failure. American Journal of Physiology-Heart and Circulatory Physiology. 2017;312:H384-H391

179. O'Brien DW, Fu Y, Parker HR, Chan SY, Idikio H, Scott PG, Jugdutt BI. Differential morphometric and ultrastructural remodelling in the left atrium and left ventricle in rapid ventricular pacing-induced heart failure. Canadian Journal of Cardiology. 2000;16:1411-1419

180. Gadeberg HC, Bond RC, Kong CH, Chanoit GP, Ascione R, Cannell MB, James AF. Heterogeneity of t-tubules in pig hearts. PLoS ONE. 2016;11:e0156862

202

181. Ross KFA. Cell shrinkage caused by fixatives and paraffin-wax embedding in ordinary cytological preparations. Quarterly Journal of Microscopical Science. 1953; s3-94:125-139

182. Toussaint A, Cowling BS, Hnia K, Mohr M, Oldfors A, Schwab Y, Yis U, Maisonobe T, Stojkovic T, Wallgren-Pettersson C, Laugel V, Echaniz- Laguna A, Mandel JL, Nishino I, Laporte J. Defects in Amphiphysin 2 (BIN1) and triads in several forms of centronuclear myopathies. Acta neuropathologica. 2011;121:253-266

183. Kim HD, Kim DJ, Lee IJ, Rah BJ, Sawa Y, Schaper J. Human fetal heart development after mid-term: Morphometry and ultrastructural study. Journal of Molecular and Cellular Cardiology. 1992;24:949-965

184. Munro ML, Soeller C. Early transverse tubule development begins in utero in the sheep heart. Journal of Muscle Research and Cell Motility. 2016;37:195- 202

185. Li M, Iismaa SE, Naqvi N, Nicks A, Husain A, Graham RM. Thyroid hormone action in postnatal heart development. Stem cell research. 2014;13:582-591

186. Rog-Zielinska EA, Craig MA, Manning JR, Richardson RV, Gowans GJ, Dunbar DR, Gharbi K, Kenyon CJ, Holmes MC, Hardie DG, Smith GL, Chapman KE. Glucocorticoids promote structural and functional maturation of foetal cardiomyocytes: A role for PGC-1alpha. Cell death and differentiation. 2015;22:1106-1116

187. Parikh SS, Blackwell DJ, Gomez-Hurtado N, Frisk M, Wang L, Kim K, Dahl CP, Fiane A, Tonnessen T, Kryshtal DO, Louch WE, Knollmann BC. Thyroid and glucocorticoid hormones promote functional t-tubule development in human-induced pluripotent stem cell-derived cardiomyocytes. Circulation Research. 2017;121:1323-1330

188. Delcarpio JB, Claycomb WC, Moses RL. Ultrastructural morphometric analysis of cultured neonatal and adult rat ventricular cardiac muscle cells. American Journal of Anatomy. 1989;186:335-345

189. Razzaq A, Su Y, Mehren JE, Mizuguchi K, Jackson AP, Gay NJ, O'Kane CJ. Characterisation of the gene for drosophila amphiphysin. Gene. 2000;241:167-174

190. Ewen EP, Snyder CM, Wilson M, Desjardins D, Naya FJ. The mef2a transcription factor coordinately regulates a costamere gene program in cardiac muscle. The Journal of Biological Chemistry. 2011;286:29644-29653

191. Buj-Bello A, Laugel V, Messaddeq N, Zahreddine H, Laporte J, Pellissier JF, Mandel JL. The lipid phosphatase myotubularin is essential for skeletal

203

muscle maintenance but not for myogenesis in mice. Proc Natl Acad Sci USA. 2002;99:15060-15065

192. Luo J, McMullen JR, Sobkiw CL, Zhang L, Dorfman AL, Sherwood MC, Logsdon MN, Horner JW, DePinho RA, Izumo S, Cantley LC. Class Ia phosphoinositide 3-kinase regulates heart size and physiological cardiac hypertrophy. Molecular and Cellular Biology. 2005;25:9491-9502

193. Hofhuis J, Bersch K, Bussenschutt R, Drzymalski M, Liebetanz D, Nikolaev VO, Wagner S, Maier LS, Gartner J, Klinge L, Thoms S. Dysferlin mediates membrane tubulation and links t-tubule biogenesis to muscular dystrophy. Journal of Cell Science. 2017;130:841-852

194. Klinge L, Harris J, Sewry C, Charlton R, Anderson L, Laval S, Chiu YH, Hornsey M, Straub V, Barresi R, Lochmuller H, Bushby K. Dysferlin associates with the developing t-tubule system in rodent and human skeletal muscle. Muscle Nerve. 2010;41:166-173

195. Ampong BN, Imamura M, Matsumiya T, Yoshida M, Takeda S. Intracellular localization of dysferlin and its association with the dihydropyridine receptor. Acta Myologica. 2005;24:134-144

196. Sakamuro D, Elliott KJ, Wechsler-Reya R, Prendergast GC. BIN1 is a novel myc-interacting protein with features of a tumour suppressor. Nature Genetics. 1996;14:69-77

197. Prokic I, Cowling BS, Laporte J. Amphiphysin 2 (BIN1) in physiology and diseases. Journal of molecular medicine. 2014;92:453-463

198. Wechsler-Reya R, Sakamuro D, Zhang J, Duhadaway J, Prendergast GC. Structural analysis of the human BIN1 gene. Evidence for tissue-specific transcriptional regulation and alternate rna splicing. The Journal of Biological Chemistry. 1997;272:31453-31458

199. Bohm J, Vasli N, Maurer M, Cowling BS, Shelton GD, Kress W, Toussaint A, Prokic I, Schara U, Anderson TJ, Weis J, Tiret L, Laporte J. Altered splicing of the BIN1 muscle-specific exon in humans and dogs with highly progressive centronuclear myopathy. PLoS Genetics. 2013;9:e1003430

200. Tjondrokoesoemo A, Park KH, Ferrante C, Komazaki S, Lesniak S, Brotto M, Ko JK, Zhou J, Weisleder N, Ma J. Disrupted membrane structure and intracellular Ca2+ signaling in adult skeletal muscle with acute knockdown of bin1. PLoS ONE. 2011;6:e25740

201. Fugier C, Klein AF, Hammer C, Vassilopoulos S, Ivarsson Y, Toussaint A, Tosch V, Vignaud A, Ferry A, Messaddeq N, Kokunai Y, Tsuburaya R, de la Grange P, Dembele D, Francois V, Precigout G, Boulade-Ladame C, Hummel MC, Lopez de Munain A, Sergeant N, Laquerriere A, Thibault C, Deryckere F, Auboeuf D, Garcia L, Zimmermann P, Udd B, Schoser B,

204

Takahashi MP, Nishino I, Bassez G, Laporte J, Furling D, Charlet- Berguerand N. Misregulated alternative splicing of BIN1 is associated with t tubule alterations and muscle weakness in myotonic dystrophy. Nature medicine. 2011;17:720-725

202. Bers DM. Excitation-contraction coupling and cardiac contractile force. Dordrecht/Boston/London: Kluwer Academic Publishers; 2001.

203. Peter BJ, Kent HM, Mills IG, Vallis Y, Butler PJ, Evans PR, McMahon HT. Bar domains as sensors of membrane curvature: The amphiphysin BAR structure. Science. 2004;303:495-499

204. Tohse N, Seki S, Kobayashi T, Tsutsuura M, Nagashima M, Yamada Y. Development of excitation-contraction coupling in cardiomyocytes. The Japanese journal of physiology. 2004;54:1-6

205. Korhonen T, Hanninen SL, Tavi P. Model of excitation-contraction coupling of rat neonatal ventricular myocytes. Biophysical Journal. 2009;96:1189- 1209

206. Woo SH, Cleemann L, Morad M. Diversity of atrial local Ca2+ signalling: Evidence from 2-D confocal imaging in Ca2+-buffered rat atrial myocytes. Journal of Physiology. 2005;567:905-921

207. Cohen NM, Lederer WJ. Calcium current in isolated neonatal rat ventricular myocytes. Journal of Physiology. 1987;391:169-191

208. Kaufman TM, Horton JW, White DJ, Mahony L. Age-related changes in myocardial relaxation and sarcoplasmic reticulum function. American Journal of Physiology. 1990;259:H309-H316

209. Kirk MM, Izu LT, Chen-Izu Y, McCulle SL, Wier WG, Balke CW, Shorofsky SR. Role of the transverse-axial tubule system in generating calcium sparks and calcium transients in rat atrial myocytes. Journal of Physiology. 2003;547:441-451

210. Hamaguchi S, Kawakami Y, Honda Y, Nemoto K, Sano A, Namekata I, Tanaka H. Developmental changes in excitation-contraction mechanisms of the mouse ventricular myocardium as revealed by functional and confocal imaging analyses. Journal of pharmacological sciences. 2013;123:167-175

211. Harary I, Farley B. In vitro studies on single beating rat heart cells. I. Growth and organization. Experimental cell research. 1963;29:451-465

212. Yamauchi Y, Harada A, Kawahara K. Changes in the fluctuation of interbeat intervals in spontaneously beating cultured cardiac myocytes: Experimental and modeling studies. Biological cybernetics. 2002;86:147-154

213. Díaz ME, Trafford AW, O'Neill SC, Eisner DA. Measurement of sarcoplasmic reticulum Ca2+ content and sarcolemmal Ca2+ fluxes in isolated

205

rat ventricular myocytes during spontaneous Ca2+ release. Journal of Physiology. 1997;501:3-16

214. Cerbai E, Pino R, Sartiani L, Mugelli A. Influence of postnatal-development on If occurrence and properties in neonatal rat ventricular myocytes. Cardiovascular Research. 1999;42:416-423

215. Eisner DA. Intracellular sodium in cardiac muscle: Effects on contraction. Experimental Physiology. 1990;75:437-457

216. Hell SW, Wichmann J. Breaking the diffraction resolution limit by stimulated emission: Stimulated-emission-depletion fluorescence microscopy. Optics Letters. 1994;19:780-782

217. Rust MJ, Bates M, Zhuang X. Sub-diffraction-limit imaging by stochastic optical reconstruction microscopy (STORM). Nature Methods. 2006;3:793- 795

218. Wong J, Baddeley D, Bushong EA, Yu Z, Ellisman MH, Hoshijima M, Soeller C. Nanoscale distribution of ryanodine receptors and caveolin-3 in mouse ventricular myocytes: Dilation of t-tubules near junctions. Biophysical Journal. 2013;104:L22-24

219. Brandenburg S, Kohl T, Williams GS, Gusev K, Wagner E, Rog-Zielinska EA, Hebisch E, Dura M, Didie M, Gotthardt M, Nikolaev VO, Hasenfuss G, Kohl P, Ward CW, Lederer WJ, Lehnart SE. Axial tubule junctions control rapid calcium signaling in atria. Journal of Clinical Investigation. 2016;126:3999-4015

220. Kass DA, Chen CH, Curry C, Talbot M, Berger R, Fetics B, Nevo E. Improved left ventricular mechanics from acute vdd pacing in patients with dilated cardiomyopathy and ventricular conduction delay. Circulation. 1999;99:1567-1573

221. Gasparini M, Auricchio A, Metra M, Regoli F, Fantoni C, Lamp B, Curnis A, Vogt J, Klersy C, Multicentre Longitudinal Observational Study G. Long- term survival in patients undergoing cardiac resynchronization therapy: The importance of performing atrio-ventricular junction ablation in patients with permanent atrial fibrillation. European heart journal. 2008;29:1644-1652

222. Suominen T, Bachinski LL, Auvinen S, Hackman P, Baggerly KA, Angelini C, Peltonen L, Krahe R, Udd B. Population frequency of myotonic dystrophy: Higher than expected frequency of myotonic dystrophy type 2 (DM2) mutation in finland. European Journal of Human Genetics. 2011;19:776-782

223. Lionello VM, Nicot AS, Sartori M, Kretz C, Kessler P, Buono S, Djerroud S, Messaddeq N, Koebel P, Prokic I, Herault Y, Romero NB, Laporte J, Cowling BS. Amphiphysin 2 modulation rescues myotubular myopathy and

206

prevents focal adhesion defects in mice. Science translational medicine. 2019;11

224. Pinali C, Bennett H, Davenport JB, Trafford AW, Kitmitto A. Three- dimensional reconstruction of cardiac sarcoplasmic reticulum reveals a continuous network linking transverse-tubules: This organization is perturbed in heart failure. Circulation Research. 2013;113:1219-1230

225. Bohm J, Biancalana V, Malfatti E, Dondaine N, Koch C, Vasli N, Kress W, Strittmatter M, Taratuto AL, Gonorazky H, Laforet P, Maisonobe T, Olive M, Gonzalez-Mera L, Fardeau M, Carriere N, Clavelou P, Eymard B, Bitoun M, Rendu J, Faure J, Weis J, Mandel JL, Romero NB, Laporte J. Adult-onset autosomal dominant centronuclear myopathy due to bin1 mutations. Brain : a journal of neurology. 2014;137:3160-3170

226. Mejaddam AY, Nennesmo I, Sejersen T. Severe phenotype of a patient with autosomal recessive centronuclear myopathy due to a bin1 mutation. Acta Myologica. 2009;28:91-93

227. Hong TT, Cogswell R, James CA, Kang G, Pullinger CR, Malloy MJ, Kane JP, Wojciak J, Calkins H, Scheinman MM, Tseng ZH, Ganz P, De MT, Judge DP, Shaw RM. Plasma bin1 correlates with heart failure and predicts arrhythmia in patients with arrhythmogenic right ventricular cardiomyopathy. Heart Rhythm. 2012;9:961-967

228. Horn MA, Bode EF, Borland SJ, Kirkwood GJ, Briston SJ, Richards MA, Dibb KM, Trafford AW. Temporal development of autonomic dysfunction in heart failure: Effects of age in an ovine rapid-pacing model. The journals of gerontology. Series A, Biological sciences and medical sciences. 2016;71:1544-1552

229. Louch WE, Sejersted OM, Swift F. There goes the neighborhood: Pathological alterations in t-tubule morphology and consequences for cardiomyocyte Ca2+ handling. Journal of Biomedicine and Biotechnology. 2010;2010:503906

207

Appendix

The attached PDFs include reprints of some of my published work during this PhD, which I think strengthen this study by further highlighting the importance of t- tubules in the heart and the regenerative capacity of t-tubules.

(1) The original article “Phosphodiesterase 5 inhibition improves contractile function and restores transverse tubule loss and catecholamine responsiveness in heart failure” published in Scientific Reports in 2019.

(2) The review article “Calcium and Excitation-Contraction Coupling in the Heart” published in Circulation Research in 2017.

208

www.nature.com/scientificreports

OPEN Phosphodiesterase 5 inhibition improves contractile function and restores transverse tubule loss and Received: 29 December 2017 Accepted: 26 March 2019 catecholamine responsiveness in Published: xx xx xxxx heart failure Michael Lawless1, Jessica L. Caldwell 1, Emma J. Radclife 1, Charlotte E. R. Smith1, George W. P. Madders 1, David C. Hutchings1, Lori S. Woods1, Stephanie J. Church2, Richard D. Unwin2, Graeme J. Kirkwood1, Lorenz K. Becker 1, Charles M. Pearman 1, Rebecca F. Taylor1, David A. Eisner1, Katharine M. Dibb1 & Andrew. W. Traford 1

Heart failure (HF) is characterized by poor survival, a loss of catecholamine reserve and cellular structural remodeling in the form of disorganization and loss of the transverse tubule network. Indeed, survival rates for HF are worse than many common cancers and have not improved over time. Tadalafl is a clinically relevant drug that blocks phosphodiesterase 5 with high specifcity and is used to treat erectile dysfunction. Using a sheep model of advanced HF, we show that tadalafl treatment improves contractile function, reverses transverse tubule loss, restores calcium transient amplitude and the heart’s response to catecholamines. Accompanying these efects, tadalafl treatment normalized BNP mRNA and prevented development of subjective signs of HF. These efects were independent of changes in myocardial cGMP content and were associated with upregulation of both monomeric and dimerized forms of protein kinase G and of the cGMP hydrolyzing phosphodiesterases 2 and 3. We propose that the molecular switch for the loss of transverse tubules in HF and their restoration following tadalafl treatment involves the BAR domain protein Amphiphysin II (BIN1) and the restoration of catecholamine sensitivity is through reductions in G-protein receptor kinase 2, protein phosphatase 1 and protein phosphatase 2 A abundance following phosphodiesterase 5 inhibition.

Impaired responsiveness to catecholamines is a hallmark of the failing heart and is detectable both in vivo and in isolated ventricular myocytes1,2. Te mechanisms responsible for the attenuated catecholamine efects in heart failure (HF) are varied and include reduced adenylate cyclase activity and enhanced G-protein receptor kinase (GRK2) and intracellular protein phosphatase activity (PP1 and PP2A) which together lead to a decrease in cAMP-dependent signaling and impaired PKA-dependent target phosphorylation1,3,4. Given the functional distribution of β-adrenergic receptors (β-ARs) and G-proteins across the surface sarcolemma and transverse tubule (TT) membrane5–7 a further factor proposed to contribute to impairment of the β-adrenergic signaling cascade in HF is the reduction of transverse tubule (TT) density seen in many pre-clinical models and human HF8–11. In addition to the classical cAMP-dependent process, the myocardial response to catecholamine stimulation is also regulated by the cGMP-PKG signaling axis consisting of the β3-AR/soluble guanylate cyclase (sGC) and

1Division of Cardiovascular Sciences, Unit of Cardiac Physiology, School of Medical Sciences, Faculty of Biology, Medicine and Health, The University of Manchester, Manchester Academic Health Science Centre, 3.24 Core Technology Facility, 46 Grafton Street, Manchester, M13 9NT, United Kingdom. 2Division of Cardiovascular Sciences, Centre for Advanced Discovery and Experimental Therapeutics, School of Medical Sciences, Faculty of Biology, Medicine and Health, The University of Manchester, Manchester Academic Health Science Centre, 3.24 Core Technology Facility, 46 Grafton Street, Manchester, M13 9NT, United Kingdom. Michael Lawless & Jessica L. Caldwell and Emma J. Radclife & Charlotte E. R. Smith contributed equally. Correspondence and requests for materials should be addressed to A.W.T. (email: [email protected])

Scientific Reports | (2019)9:6801 | https://doi.org/10.1038/s41598-019-42592-1 1 www.nature.com/scientificreports/ www.nature.com/scientificreports

natriuretic peptide/particulate guanylate cyclase (pGC) pathways (reviewed by Tsai and Kass12). Te outcome of cGMP-dependent activation depends on the source of activating cGMP; that activated by sGC inhibiting the β-AR response and, pGC-derived cGMP having no efect13,14. Beyond the role of PKG, GRK2 and protein phosphatases in determining the outcome of β-AR stimulation, the intracellular pools of cAMP and cGMP are also diferentially regulated by phosphodiesterases (PDEs) suggesting highly compartmentalized regulation of the cyclic nucleotides and thus catecholamine responsiveness of the healthy ventricular myocardium e.g.15–18. Given the negative impact of acute PDE5 inhibition on the inotropic and lusitropic response to catecholamines and the established loss of catecholamine reserve in HF it is somewhat surprising that an emerging body of evidence suggests PDE5 inhibition is clinically cardioprotective in type II diabetes19, lef ventricular hypertrophy20 and in patients with HF with reduced ejection fraction (systolic HF)21. Similarly, in experimental models, PDE5 inhibition shows cardioprotective efects in pulmonary hypertension22, myocardial infarction23–26 and following aortic banding27. However, in each of these cases PDE inhibition was commenced either before or given concurrently with the disease intervention. Such an experimental approach complicates interpretation of whether the intervention is therapeutically useful in a setting of established disease or is acting by preventing disease development. In most28,29, but not all30 experimental studies where PDE5 inhibition has been commenced once some degree of lef ventricular remodeling has occurred the fndings remain supportive of a cardioprotective efect. However, in the ‘positive’ studies the extent of disease progression to symptomatic HF is unclear and data on survival outcomes is generally missing. Given these considerations, the hypothesis examined is that PDE5 inhibition is benefcial in systolic HF through restoration of catecholamine responsiveness. As such, the primary aim of the present study was to determine if PDE5 inhibitor treatment, instigated at an advanced disease stage once contractile dysfunction and attenuated catecholamine responsiveness are established, is capable of reversing these efects. Te secondary aim of the study was to determine if changes in contractile and catecholamine responsiveness were associated with structural (TT) remodeling and to elucidate the underlying molecular mechanisms of such TT remodeling. Te fnal aim of the study was to determine the underlying mechanisms that contribute to the restoration of catecholamine responsiveness. Te major fndings are that PDE5 inhibition with tadalafl restored catecholamine responsiveness and partially reversed contractile dysfunction in vivo. At the cellular level, PDE5 inhibition restored the amplitude of the systolic calcium transient to control levels. Tese changes were associated with a restoration of TT density but β-AR abundance (β1 and β2) remained unaltered in HF and following tadalafl treatment. However, downstream regulators of catecholamine signalling including GRK2, PKG, PDE2, PDE3, PP1 and PP2A were all diferentially regulated by HF and PDE5 inhibition. Our fndings also implicate the BAR domain protein amphiphysin II (AmpII aka BIN1) as a key driver of the TT changes seen in response to HF and PDE5 inhibitor treatment. Additionally, we found that tadalafl treatment reversed myocardial changes in BNP expression and that this was associated with the prevention of the development of subjective HF symptoms. Results PDE5 inhibition improves cardiac contractility in vivo and systolic calcium in vitro. We frst sought to determine if tadalafl treatment was associated with any changes in cardiac contractility or cardiac dimensions. In line with our previous fndings1,31–33 lef ventricular dilatation measured at end diastole (end diastolic internal dimension; EDID) and the end of the systolic phase (end systolic internal dimension; ESID) as well as free wall thinning were observed following tachypacing. In a subset of animals where echocardiographic assessments were repeated every week, tadalafl treatment did not reverse lef ventricular dilatation (Fig. 1A, all values p < 0.01 vs pre-pacing). Similarly, systolic wall thinning occurred in tachypaced animals; an efect not reversed by tadalafl treatment (p < 0.05 vs pre-pacing). However, afer three weeks of tadalafl treatment, diastolic wall thickness was indistinguishable from pre-pacing levels (Fig. 1B; p = 0.09). In sheep the cardiac apex is positioned over the sternum and it is difcult to obtain four chamber apical views in adult animals. Terefore, we used the short axis fractional area change measured at mid papillary muscle level as a surrogate for ejection fraction. Following 4-weeks of tachypacing, fractional area change was reduced from pre-pacing values by 43 ± 7% (Fig. 1C) in the untreated group and to the same extent in the tadalafl group before treatment commenced (43 ± 3%, both groups p < 0.001 vs pre-pacing values). However, tadalafl treatment increased fractional area change such that by the end of the study fractional area change was 16 ± 8% greater than at 4-weeks (p < 0.05) and 35 ± 24% greater than in the end-stage untreated group (p < 0.01). In agreement with the in vivo contractility fndings and our previous study1, the amplitude of the systolic calcium transient was reduced by 66 ± 14% in HF (Fig. 1D,E, p < 0.05). Importantly, the reduction in calcium transient amplitude was established by 4 weeks of tachypacing (by 74 ± 8%, p < 0.05). However, following tadalafl treatment the amplitude of the systolic calcium transient was 399 ± 150% of HF values (p < 0.001) and indistinguishable from control (132 ± 32% of control; p = 0.49). Terefore, tadalafl treatment restored systolic calcium to a greater extent than in vivo contractility. Whilst tadalafl treatment augmented cardiac contractility and systolic calcium, the additive efects of tadalafl treatment on blood pressure were minimal. Previously we have reported that systolic, diastolic and mean blood pressure decrease in HF34; an observation repeated here (Table 1). However, tadalafl treatment had no further efect on blood pressure which was indistinguishable from both the 4-week tachypaced and HF groups. Tus, the changes in cardiac function associated with tadalafl treatment are not due to changes in peripheral blood pressure. PDE5 inhibition restores the inotropic and chronotropic response to catecholamine stimulation. Since HF is associated with a reduced responsiveness to β-adrenergic stimulation1,31 and impaired catecholamine responses are associated with reduced survival in HF patients35, we next determined if tadalafl treatment was also associated with restoration of catecholamine responsiveness. We frst examined cardiac contractile responses to intravenous infusion of the exogenous β-adrenergic agonist dobutamine (Fig. 1F,G). In

Scientific Reports | (2019)9:6801 | https://doi.org/10.1038/s41598-019-42592-1 2 www.nature.com/scientificreports/ www.nature.com/scientificreports

Figure 1. Tadalafl treatment restores indices of cardiac function in heart failure. (A) Changes in lef ventricular dimensions with tachypacing and response to tadalafl treatment (paired data, N = 4 each time point; **p < 0.01 by RM-ANOVA). ‘pre’ denotes before commencement of tachypacing, ‘4wk’ denotes 4-weeks of tachypacing and T1, T2 and T3 denote 1, 2 & 3 weeks of tadalafl treatment. (B) Changes in lef ventricular free wall thickness in response to tachypacing and tadalafl treatment (paired data, N = 4 each time point; *, **, *** respectively p < 0.05, 0.01 & 0.001 by RM-ANOVA). (C) Echocardiographic assessment showing short-axis fractional area change in the absence of exogenous catecholamine stimulation: *p < 0.05 vs. 4-week time point (tadalafl arm); **p < 0.01 vs. HF group at 7 weeks; ***p < 0.001 vs. pre-pace function (both groups). Statistics by mixed models analysis. N = HF, 5 at all time points; tadalafl, 20 at pre-pace, 19 at 4-weeks and 18 at 7 weeks. (D) Representative systolic calcium transients for control (black), 4-week paced (green), HF (red) and tadalafl (blue) groups. (E) Summary data for systolic calcium transients; *p < 0.05 vs. control; #p < 0.05 vs. tadalafl. (n/N = 34/16 control, 6/2 following 4 weeks tachypacing, 7/3 HF, 24/6 Tadalafl). (F) Summary data of echocardiographic assessments following 5 µg/kg/min dobutamine infusion: **p < 0.01 vs. HF by one-way ANOVA. N = pre-pacing, 4; 4-weeks, 3; HF, 4; tadalafl, 3. (G) Summary data showing dobutamine induced change in fractional area change derived from paired (repeated measures) data in panel F. Data is expressed as a percentage of the maximum possible response in each group. (H) Summary data showing RR

Scientific Reports | (2019)9:6801 | https://doi.org/10.1038/s41598-019-42592-1 3 www.nature.com/scientificreports/ www.nature.com/scientificreports

intervals normalized to pre-dobutamine (5 µg.kg/min) RR interval: **p < 0.01 vs. pre-dobutamine RR interval by t-test. N = pre-pacing, 5; 4-weeks, 5; HF, 6; tadalafl, 5. (I) Paired data from 5 animals showing RR interval – dobutamine dose response relationship. *p < 0.05 by mixed models analysis. (J) Kaplan-Meier plot showing fraction of animals free from subjectively assessed signs of HF (N = HF, 42; tadalafl, 27). Tadalafl treatment commenced 28 days afer tachypacing (dashed line): ***, denotes p < 0.001 (by Cox regression-based test).

paired samples (pre-pacing, 4-weeks and tadalafl treated), there was a greater response to dobutamine infusion afer tadalafl treatment than afer 4-weeks of tachypacing (Fig. 1G) indicating that chronic PDE5 inhibition in HF augments catecholamine sensitivity. As we have reported previously in a separate study31, HF attenuated the chronotropic response to dobutamine. In pre-paced animals, 5 µg/kg/min dobutamine increased heart rate and shortened RR interval by 36 ± 10% (Fig. 1H; p < 0.01) yet had no efect on RR interval following 4-weeks of tachypacing or in end-stage HF (the change in RR interval being 5.4 ± 8% and 3.8 ± 2% respectively). Conversely, at this infusion rate following tadalafl treatment, heart rate increased resulting in RR interval shortening by 26 ± 6% (p < 0.01). To further understand the attenuated chronotropic response, we also investigated the dose-response to dobutamine in a cohort of animals at three time points; pre-pacing, 4-weeks of tachypacing (but before tadalafl treatment commenced) and following 3-weeks of tadalafl treatment (Fig. 1I). It is clear that the sensitivity to dobutamine was reduced at 4-weeks of tachypacing and restored toward control values following tadalafl treatment. Assuming that the maximal achievable heart rate is equal to that in pre-paced animals treated with 10 µg/kg/min dobutamine, the calculated EC50 concentration of dobutamine is 3.6, 34.4 and 10.7 µg/kg/min in the pre-pacing, 4-weeks tachypaced and tadalafl treated groups respectively (p < 0.005). To conclude the in vivo series of experiments, we sought to determine if tadalafl treatment was associated with any improvement in the subjective signs of HF (including lethargy, dyspnoea, cough or loss of bodily condition). In the untreated cohort, 34 of 42 animals developed at least two subjective signs of HF before the designated end of the study (49 days of tachypacing; all 34 animals were dyspnoeic at rest and coughed when restrained in a supine position) and were therefore humanely killed for subsequent in vitro experiments. Conversely, of the 27 animals randomly assigned to the tadalafl treatment group, all remained free from signs of HF to the completion of the study (day 49, Fig. 1J). One tadalafl treated animal was found dead on day 19 of treatment; notably there were no prior subjective signs of HF and on post-mortem examination the pacing lead was found to have perforated the right ventricular wall and pulmonary congestion was not evident. Tus, compared to untreated animals, tadalafl treatment yielded a Cox’s proportional hazard ratio of 0.026 ± 0.03 (95% CI, 0.003–0.19; p < 0.001). Additionally, there was no diference in the initial body weight between the 42 untreated (32.0 ± 1.1 kg) and 26 tadalafl treated animals (31.6 ± 1.3 kg) and, as such, body weight was not a signifcant covariate (p = 0.44). Finally, in a subset of the tadalafl treated HF animals (N = 13) the total plasma tadalafl content determined by mass spectrometry was 269 ± 39 µg/l and plasma tadalafl protein binding measured as 93 ± 0.6% which is in good agreement with that in healthy human subjects36. Tus, the lowest free tadalafl concentration in HF plasma following 3 weeks of daily dosing (20 mg daily) was 51 ± 8 nmol/l. In addition to improving several in vivo measures of cardiac function and catecholamine sensitivity, chronic PDE5 inhibition also normalized the tachypacing-induced increases in myocardial NPPB (B-type natriuretic peptide) mRNA (Fig. 2A). Tus, tadalafl treatment at least partially restores both the inotropic and chronotropic response to catecholamines that are lost in HF and, in further agreement with the improvement in the subjective assessments of HF development, tadalafl treatment also reverses the increase in mRNA abundance of the widely accepted marker of HF, B-type natriuretic peptide.

The impact of PDE5 inhibition on regulators of cGMP and the β-adrenergic signalling cascade. Tere was no change in the abundance of PDE5A mRNA (Fig. 2B) or cGMP levels in myocardial homogenates (Fig. 2C) in HF or following tadalafl treatment. Whilst unaltered PDE5 mRNA levels in HF have been reported previously29,37, the unchanged myocardial cGMP content following tadalafl treatment was initially surprising. However, examining known regulators of cGMP in HF and following tadalafl treatment we found an increase in total, monomeric (75 kDa) and dimerized (150 kDa) PKG (Fig. 2D). Total and 75 kDa PKG abundances were similar between the 4-week paced, HF and tadalafl groups. However, for the 150 kDa dimer of PKG, abundance was greater in the tadalafl group compared to the 4-week paced group and trended to an increase when compared to the HF group (p = 0.053). Moreover, tadalafl treatment augmented the abundance of the cGMP hydrolyzing phosphodiesterases PDE2 and PDE3 (Fig. 2E,F). Tus, the increased abundance of PKG, through inhibition of sGC38 and stimulation of PDE515, coupled with elevated levels of the cGMP hydrolyzing PDEs 2 & 3 provide a plausible explanation for the unchanged myocardial cGMP content despite chronic tadalafl treatment. In the next experiments we sought to elucidate the impact of chronic PDE5 inhibition in HF on key regulators of catecholamine signalling in the heart. We found no diference in β1 or β2 adrenergic receptor abundance in whole heart homogenates following 4-weeks of tachypacing, in HF or following tadalafl treatment (Fig. 3A,B). In agreement with our previous fndings1, in HF the protein abundance of G-protein coupled receptor kinase GRK2 and protein phosphatases PP1 and PP2A was increased (Fig. 3C–E). Te increase in GRK2, PP1 and PP2A was established by 4-weeks of tachypacing, reafrming adrenergic dysfunction is established prior to PDE5 inhibitor treatment in this study. Importantly, chronic tadalafl treatment resulted in a reduction in protein abundance for each of these regulators of β-adrenergic signalling to either the same level (GRK2, PP1) or lower (PP2A) than in control hearts thus mechanistically linking PDE5 inhibition with the enhanced catecholamine responses noted in vivo.

Scientific Reports | (2019)9:6801 | https://doi.org/10.1038/s41598-019-42592-1 4 www.nature.com/scientificreports/ www.nature.com/scientificreports

Systolic BP Diastolic BP Pulse Pressure (mmHg) (mmHg) (mmHg) Mean(mmHg) Control (15) 114 ± 3 68 ± 2 56 ± 3 85 ± 2 4-weeks (10) 105 ± 4 62 ± 3*,§ 43 ± 3 80 ± 4 HF (5) 102 ± 5*,§ 57 ± 5*,§ 45 ± 4 72 ± 5*,§ Tadalafl (5) 97 ± 5*,§ 55 ± 5*,§ 42 ± 4§ 73 ± 4*,§

Table 1. Te infuence of HF and tadalafl treatment on blood pressure measured by tail cuf plethysmography. Summary blood pressure data (mean ± standard error). Values in brackets denote number of animals in each condition. *, denotes p < 0.05 vs. control (One way ANOVA with SNK correction for multiple comparisons). §, denotes P < 0.05 vs. control (paired t-test).

PDE5 inhibition restores the transverse tubule network in HF. Since the β1-AR and β2-AR are functionally distributed on the TT membrane and surface sarcolemma5,7 we sought to determine if the restored response to the β-AR agonist dobutamine following tadalafl treatment was associated with changes in TT density or TT organization in ventricular myocytes (Fig. 4A). To quantify changes in TT density we utilized two approaches; (i) determination of the distance 50% of voxels (3 dimensional pixels) are from their nearest membrane (TT or surface membrane) in all three imaging planes, termed ‘half-distance’8,32 (Fig. 4A.b) and, (ii) determination of the fractional area of the cytosol occupied by TT’s in the imaging plane, termed ‘fractional area’8,32. In agreement with our previous work8 and others e.g.9,10,39 HF resulted in a decrease in ventricular TT density of 19 ± 5% measured using the half distance (Fig. 4B; p < 0.001) and 26 ± 5% using fractional area methodolo- gies (Fig. 4C; p < 0.001). Importantly, these changes in TT density had occurred by 4 weeks of tachypacing and therefore the TT density reduction observed in HF was established before commencement of tadalafl treatment (Fig. 4B,C). Following 3 weeks of tadalafl treatment, with continued tachypacing, there was a complete restoration of TT density back to control levels (half-distance: control, 0.377 ± 0.01 µm; tadalafl, 0.369 ± 0.001 µm; fractional area: control, 0.198 ± 0.01; tadalafl, 0.231 ± 0.01). Despite TT density being fully restored following tadalafl treatment, closer inspection of the skeletonized images (Fig. 4A.c) showed that the TTs are less organized in the tadalafl treated animals than in control myocytes. In control cells TTs are predominantly organized perpendicular to the long axis of the cell (Fig. 4D) having a transverse (perpendicularly arranged) to longitudinal (horizontally arranged) ratio of 2.4 ± 0.2 (Fig. 4E). Afer 4 weeks of tachypacing, in end-stage HF and following tadalafl treatment TTs are nearly equally oriented longitudinally and transversely (4-weeks, 1.1 ± 0.1; HF, 1.0 ± 0.1; tadalafl 1.2 ± 0.1; all p < 0.001 versus control). Tus, the restoration of TT density following tadalafl treatment is not matched by normalization of their orientation. Potential molecular mechanisms responsible for changes in TT density in HF and reversal by PDE5i treatment. Although the control of TT turnover remains poorly understood, we next sought to identify potential molecular mechanisms responsible for the restoration of TTs following tadalafl treatment. We examined protein abundance of four putative regulators of TT turnover: AmpII (BIN1), junctophilin 2 (JPH2), titin capping protein (Tcap) and myotubularin 1 (MTM1). Of these, JPH2 (Fig. 5A) and Tcap (Fig. 5B) showed no change in abundance with disease duration or tadalafl treatment whereas MTM1 abundance was increased equally above control in the 4-week tachypaced, HF and tadalafl treated animals (Fig. 5C, all p < 0.05). Conversely, AmpII protein abundance was reduced from control levels in both the 4-week tachypaced and HF groups where TT density is reduced (Fig. 5D.a; p < 0.05) and restored to control levels in the tadalafl treated group in line with the restoration of TT density. Moreover, AmpII abundance was also greater in the tadalafl group than both the 4-week tachypaced and HF groups (p < 0.05). Te relationship between TT density and AmpII protein abundance is examined in Fig. 5D.b,c. Here, due to the unpaired nature of the protein samples used for Western blotting and cells used for TT density determination we have used a non-replacement random sampling and ftting simulation approach (1000 iterations) to test the correlation between TT density and AmpII protein abundance. As we have shown previously using gene silencing approaches in isolated myocytes8, TT density is correlated with AmpII protein abundance during the development of HF and following treatment with tadalafl (p < 0.001). Whilst our present data (e.g. Fig. 5D) and previous studies8,40 have shown a correlation between AmpII abundance and TT density, there remains some uncertainty over which isoforms of AmpII (BIN1) are expressed in cardiac muscle and drive the formation of TTs in cardiac myocytes. We therefore sought frstly to determine which AmpII isoforms are present within the ovine myocardium. Using RT-PCR approaches and amplifying the variably expressed regions between exons 6 and 18 of AmpII (Supplemental Fig. 3) we identifed 3 major amplicons in the ovine myocardium (Fig. 6A). From amplicon size and subsequent sequence verifcation we identifed the ‘skeletal’, exon 11 containing v8 isoform as the major AmpII isoform expressed in the myocardium with lower levels of expression of v4 and v10 isoforms. By investigating the role of variant 8 AmpII overexpression in NRVMs and iPSC CMs we next sought to establish if the augmented AmpII protein abundance provides a mechanism for the increase in TT density following tadalafl treatment. Importantly, and in contrast to the heterozyous knockout model used by Hong et al.41, both cell types lack TTs. Both non-transfected and mKate2 fuorescent protein transfected NRVMs have low endogenous AmpII protein abundance whereas AmpII-AC-mKate2 transfection increased AmpII protein abundance 4.6 ± 0.3-fold (Fig. 6B, p < 0.05). When transfected NRVMs were imaged confocally (Fig. 6C), those that were transfected with the mKate2 control vector exhibited difuse cytosolic and nuclear fuorescence. Conversely, cells transfected with the AmpII-AC-mKate2 vector exhibited widespread

Scientific Reports | (2019)9:6801 | https://doi.org/10.1038/s41598-019-42592-1 5 www.nature.com/scientificreports/ www.nature.com/scientificreports

Figure 2. Altered abundance of regulators of cyclic GMP in HF and following tadalafl treatment. (A) Summary quantitative PCR assessment of myocardial brain natriuretic peptide (NPPB) mRNA abundance normalized to the housekeeper RPLPO. (B) Summary quantitative PCR assessment of myocardial phosphodiesterase 5 A (PDE5A) mRNA abundance normalized to the housekeepers TBP and YWHAZ. (C) Summary data showing no change in myocardial cGMP content. (D) Changes in protein kinase G abundance assessed by non-denaturing PAGE showing (top) a representative blot, (middle) quantifcation of total and 75 kDa PKG and, (bottom) quantifcation of 150 kDa PKG and % total PKG as dimerized form. (E) Altered PDE2A protein abundance showing (top) representative blot and, (bottom) summary data. (F) Altered PDE2A protein abundance showing (top) representative blot and, (bottom) summary data. For all panels *p < 0.05 vs control; #p < 0.05 vs tadalafl. N = 8 each group.

intracellular fuorescence in the form of an inter-connected network. To determine if the AmpII driven network formed tubules connected to the surface sarcolemma cells were superfused with the extracellular fuorescent marker Oregon Green 488 N. It is clear from Fig. 6C that Oregon Green 488 N enters these tubular structures and co-localizes with mKate2 fuorescence indicating: i) that AmpII forms tubular structures and, ii) these are patent

Scientific Reports | (2019)9:6801 | https://doi.org/10.1038/s41598-019-42592-1 6 www.nature.com/scientificreports/ www.nature.com/scientificreports

Figure 3. Tadalafl normalizes the abundance of key regulators of β-adrenergic signalling. (A) Representative blot (upper) and summary data (lower) showing no change in β1 adrenergic receptor abundance with tachypacing or tadalafl treatment. (B) Representative blot (upper) and summary data (lower) showing no change in β2 adrenergic receptor abundance with tachypacing or tadalafl treatment. (C) Representative blot (lef) and summary data (right) showing tachypacing induced increased and tadalafl mediated normalisation of GRK2 protein abundance. (D) Representative blot (upper) and summary data (lower) showing tachypacing induced increased and tadalafl mediated normalisation of PP1 protein abundance. (E) Representative blot (upper) and summary data (lower) showing tachypacing induced increased and tadalafl mediated normalisation of PP2A protein abundance. For all panels *P < 0.05 vs. control; #p < 0.05 vs. tadalafl. N = 8 per group.

at and connected to the cell surface. On average 97.9 ± 2.1% cells transfected with AmpII-AC-mKate2 developed a tubular network compared to no tubules in non-transfected or cells transfected with the mKate2 control vector (Fig. 6D, p < 0.001). Te summary data of Fig. 6E demonstrates that AmpII-C-mKate2 expression in NRVMs lead to a robust increase in the fractional area of cells occupied by TTs (59 ± 3 fold, p < 0.001). Finally, we also confrmed that AmpII was capable of driving de novo tubule formation in human iPSC CMs and that these cells also normally lack a discernible tubular network (Fig. 6F–H).

Scientific Reports | (2019)9:6801 | https://doi.org/10.1038/s41598-019-42592-1 7 www.nature.com/scientificreports/ www.nature.com/scientificreports

Figure 4. Recovery of transverse tubule density with altered orientation following tadalafl treatment. (A) Representative images from cell types indicated showing membrane staining (a), distance maps (b) and skeletonized images (c) from regions of interest shown in panel a. Note the reduction in TT density (a, b) and lateralization of TTs in 4-week paced, HF and tadalafl treated groups (c). Scale bars, 10 µm. (B) Summary data for TT half-distance measurements. (C) Summary data for TT fractional area measurements. For (B,C): n cells/N hearts; control, 30/6; 4-weeks, 25/4; HF, 29/6; tadalafl, 71/6. (D) Representative orientation analysis for TTs in regions highlighted relative to long-axis of cells. 0° represents longitudinally orientated and 90o transversely oriented tubules. (E) Summary data showing TT transverse to longitudinal orientation ratio. In (B,C,E): *p < 0.05 vs control; **p < 0.01 vs. control; ***p < 0.001 vs. control; #p < 0.05 vs tadalafl; by linear mixed models analysis. Panels B, C & E show mean ± SEM.

Discussion Te main fndings of the present study investigating the impact of chronic PDE5 inhibition on cardiac function and structure in HF are threefold: i) PDE5 inhibition improves indices of cardiac contractility and restores systolic calcium and catecholamine responsiveness; ii) t-tubule density is reduced in HF and fully restored to control levels by PDE5 inhibition and, iii) changes in expression of the putative t-tubule regulator AmpII are correlated with changes in t-tubule density and AmpII drives de novo tubule formation in ventricular myocytes. Importantly, in this study we demonstrate that PDE5 inhibition reverses rather than prevents the HF-dependent changes in cellular structure and catecholamine responsiveness as both the reduction in TT density and loss of catecholamine responsiveness are already present before PDE5 inhibition is commenced.

PDE5 inhibition mediated reverse remodeling in HF. In the present study tadalafl treatment reversed the tachypacing-induced changes in myocardial B-type natriuretic peptide mRNA levels. Consistent with the reversal of B-type natriuretic peptide mRNA we also noted that tadalafl treatment prevented development of a number of subjective signs of HF. Te delayed administration of PDE5 inhibition used here distinguishes this study from the majority of previous studies investigating PDE5 inhibition in HF where treatment was initiated simultaneously with, or prior to, disease initiation22–27. Additionally, where earlier studies have implemented delayed PDE5 inhibitor treatment, these appear to have been at a far less advanced stage of HF than in this study based on contractility measurements or the maintained response to catecholamine stimulation28–30,37. Tus, this study adds to a growing weight of evidence on the efectiveness of PDE5 inhibition in treating systolic HF and demonstrates reversal of key elements of the disease process that are evident when cardiac dysfunction is at an advanced stage prior to initiation of PDE5 inhibition. Whilst changes in B-type natriuretic peptide mRNA are observed, the PDE5 inhibition mediated efects on HF progression occur in the absence of changes in PDE5A mRNA and cGMP levels within the myocardium. Previous studies have also noted no change in PDE5A abundance in preclinical HF models29,37 and Hiemstra et al., using a porcine model, reported no change in myocardial cGMP content following tadalafl treatment37. In the present study we investigated potential mechanisms that could explain the lack of change in the global cGMP content of

Scientific Reports | (2019)9:6801 | https://doi.org/10.1038/s41598-019-42592-1 8 www.nature.com/scientificreports/ www.nature.com/scientificreports

Figure 5. Altered protein abundance of putative regulators of cardiac transverse tubule formation. Representative immunoblots (upper) and summary histograms (lower) for: (A) JPH2; (B) TCAP and (C) MTM1. (D) Changes in AmpII protein abundance showing (upper) representative AmpII immunoblot and (lower) summary data (a); correlation between AmpII protein abundance and TT half distance (b) and TT fractional area (c). Solid lines through data in b. and c. are linear regressions with slopes signifcantly diferent from zero (*p < 0.05). For immunoblots (A–D) N = control, 7; 4-weeks, 6; HF, 7; tadalafl, 8 and mean of 3 technical replicates. *p < 0.05 vs. control; #p < 0.05 vs. tadalafl; by linear mixed models analysis. For regression analysis: n cells/N hearts; control, 30/ 6; 4-weeks, 25/4; HF, 29/6; tadalafl, 71/6. All data presented as mean ± SEM.

the myocardium and found that PKG (monomer, dimer and total levels) and the cGMP hydrolyzing phosphodiesterases PDE2 and PDE3 were augmented following tadalafl treatment. We suggest therefore that the changes in PKG, PDE2 and PDE3 may underlie the lack of a global change in cGMP, yet, there may still remain sub-cellularly restricted pools of elevated cGMP that may infuence distinct signalling pathways and underpin the inotropic, chronotropic and ‘survival outcome’ efects seen with tadalafl treatment.

PDE5 inhibition in HF: efects on cardiac function and catecholamine responsiveness. Te frst concern regarding the impact of tadalafl treatment in HF is that of of-target efects. Tadalafl is structurally distinct from other commonly used PDE5 inhibitors (sildenafl, vardenafl) and shows substantially greater (>1000 fold) specifcity for PDE5 over other PDEs than both sildenafl and vardenafl with the exception of PDE11 where 42–45 the selectivity ratio (PDE5 IC50: PDE11 IC50) is between 7.1 and 40 . However, the absence of PDE11 in cardiac myocytes46 together with the lack of any obvious neurological disturbance or myalgia (PDE11-dependent side-efects) would suggest that despite the plasma concentration of tadalafl used (~50 nmol/l) causing complete 42,47 inhibition of PDE5 (IC50 range 0.94–9.4 nmol/l) , the approximate 40%, inhibition of PDE11 is not underpinning the cardioprotective efects observed in the present study. Te development of PDE11 specifc antagonists or use of lower doses of tadalafl in future studies would enable this to be conclusively determined. In those studies where PDE5 inhibition was started at varying time points following disease initiation there is indeed evidence of attenuated deterioration in function or some reversal of extant dysfunction28,30,48,49. However, in each of these cases catecholamine response data and commentary on the development of signs of HF are absent. Conversely, in those studies where the systolic calcium transient or adrenergic responsiveness have been measured they are either maintained29 or augmented37 in the ‘HF’ groups suggesting that the disease stage in these particular studies is relatively early. In comparison, the present study, using a large animal model of tachycardic end-stage HF, demonstrates that contractile and adrenergic dysfunction are established before the onset of PDE5 inhibitor treatment and these efects are at least partially restored by tadalafl treatment. In the present study we investigated the potential mechanisms underpinning the restoration of catecholamine responsiveness following PDE5 inhibition. Whilst we found no change in β1 or β2 adrenergic receptor density in HF or following tadalafl treatment, it remains possible that the restoration of TT density with tadalafl treatment is important in directing the sub-cellular aspects of β-receptor signaling and thence augmenting cardiac

Scientific Reports | (2019)9:6801 | https://doi.org/10.1038/s41598-019-42592-1 9 www.nature.com/scientificreports/ www.nature.com/scientificreports

Figure 6. Amphiphysin II drives de novo tubule formation in neonatal rat ventricular myocytes and iPSC derived cardiac myocytes. (A) Assessment of sheep myocardial AmpII isoform expression showing (lef to right) representative Qiaxcel run and summary data on sheep cardiac isoform abundance. N = 4 hearts. (B) Representative immunoblot (upper) and summary data (lower) showing up-regulation of AmpII protein abundance following transient transfection of NRVMs. N = 2 isolations. (C) Representative images of NRVMs transfected with mKate2 control (a) vector or AmpII expression vector (b–d). Extracellular Oregon Green 488 imaging showing a transfected and non-transfected (NT) cell (c) and overlay of the mKate2 channel and OG488N channel (from c) showing co-localisation of markers and patency of AmpII driven tubules only in the transfected cell (d). (D) Mean data summarizing percentage of non-transfected (NT), mKate2 control vector and AmpII expression vector transfected cells with tubule structures. N = 6 isolations (non-transfected, 67 cells; mKate2, 23 cells; AmpII, 37 cells) cells in each group. (E) Mean data summarising fractional area of cells occupied by tubules in mKate2 control vector and AmpII expression vector transfected cells. N = 5 isolations (mKate2, 20 cells; AmpII, 29 cells). (F) Representative images showing AmpII driven tubule formation in iCell iPSC derived cardiac myocytes. (G) Mean data summarizing percentage of non-transfected (NT), mKate2 control vector and AmpII expression vector transfected iCell iPSC cardiac myocytes with tubule structures (N = 2 cell batches; non-transfected, 45 cells; mKate2, 17 cells; AmpII, 40 cells). (H) Mean data summarizing fractional area occupied by tubules in iCell cardiac myocytes transfected with mKate2 or AmpII vectors. N = 2 cell batches (makte2, 17 cells; AmpII, 29 cells). **p < 0.01 vs mKate2; ***p < 0.001 vs mKate2; #p < 0.05 vs NT; by linear mixed models analysis. Scale bars, 10 µm. Panels A, C, D, F & G show mean ± SEM.

contractility5,50. However, in line with the restoration of catecholamine sensitivity we observed tadalafl-dependent decreases in GRK2, PP1 and PP2A protein abundance which would increase cAMP levels51 and enhance target phosphorylation including sites on myoflament proteins, phospholamban and the L-type Ca2+ channel52–54.

PDE5 inhibition in HF: efects on cardiac cellular structure. Increasing evidence implicates a reduction in TT density as a major factor contributing to cardiac dysfunction in diverse cardiac diseases8–10,22,32,39,55–57. In the present study, we also note a decrease in TT density in HF and importantly that this is fully reversed by PDE5 inhibition. Our fndings extend the earlier studies of Huang et al.27 where sildenafl treatment commenced concurrently with aortic-banding prevented TT loss and Xie et al.22 where delayed sildenafl treatment partially restored right ventricular TT density in a pulmonary artery hypertension model. Te incomplete recovery of TTs noted by Xie et al. may refect the shorter treatment duration (2 weeks), drug choice (sildenafl) or chamber-wall strain diferences (right versus lef)58–60.

Scientific Reports | (2019)9:6801 | https://doi.org/10.1038/s41598-019-42592-1 10 www.nature.com/scientificreports/ www.nature.com/scientificreports

Whilst tadalafl treatment restores TT density, the TTs in this group are equally distributed in the transverse and longitudinal orientations, rather than predominantly transverse orientation seen in control hearts. Previous studies have noted plasticity in TTs and their ability to be restored to varying extents following interventions to reverse HF27,60–62 although these have not investigated TT orientation in the various intervention arms. Here, we sought to determine if changes in TT density in HF and following tadalafl treatment are correlated with changes in the abundance of proteins implicated in TT biogenesis in striated muscle63–65. Of the putative candidate proteins only one, AmpII, correlates with TT density changes in HF and following tadalafl treatment. AmpII (BIN1) is known to be involved in TT biogenesis in skeletal muscle66 and in cardiac muscle protein abundance is reduced in HF8,40,61. Using siRNA gene silencing in cultured adult ventricular myocytes we have shown that AmpII is required for cardiac TT maintenance8. Hong et al.41,67 and more recently, De La Mata et al.68 demonstrated the importance of BIN1 (AmpII) in delivery of L-type Ca channels to TT membranes and TT membrane folding. In the present study, we extend this earlier work and show that AmpII is capable of driving the de novo formation of tubule like structures in both NRVMs and iPSC CMs; model cell types that lack TTs. Importantly, co-labelling with extracellular dye shows that the AmpII driven tubules are patent at the cell surface. Tus, an emerging body of work highlights the importance of AmpII in the TT biology in the heart and future studies will be required to defne the functional signifcance of AmpII driven TT formation. Moreover, and in stark contrast to the work of Hong et al. using the mouse41, we also demonstrate that exon 11 (previously known as exon 1069) containing variants 8 and 4 of AmpII (BIN1) are expressed within the sheep myocardium with variant 8 mRNA being the dominantly expressed isoform. We also show that variant 8 is capable of inducing de novo tubulation in cardiac cells. Further work will be required to elucidate the signifcance of species diferences in AmpII isoform usage and the critical components of AmpII for cardiac muscle TT formation. It may be that, as with chamber diferences in TT density, important distinctions exist between laboratory rodents and large mammals including human8,32,70. Beyond the correlation between AmpII protein abundance and TT density, we also noted a generalized increase in lipid phosphatase MTM1 protein abundance in the 4-week tachypaced, HF and tadalafl treated animals. In skeletal muscle mutations in MTM1 or altered interaction between MTM1 and a number of binding partners, including AmpII lead to inherited centronuclear myopathies, some of which are associated with cardi- omyopathies71,72. Furthermore, MTM1 augments AmpII driven membrane tabulation72 and MTM1 defciency leads to reductions in skeletal muscle TT density73,74. Tus, it is possible the increase in MTM1 protein abundance noted in the present study is a compensatory mechanism to maintain TTs in the face of cardiac pathology. However, whether the increase in MTM1 abundance is responsible for the lateralization of TTs in HF and tadalafl treated animals remains to be determined although addressing this in the iPSC and NRVM model systems would be problematic given the inherent disorder of the AmpII driven tubules in these cell types. Finally, there remain the unanswered questions as to whether: (i) TT loss is a cause or consequence of HF and, (ii) if tadalafl improves cardiac function and this allows TT density to recover or, tadalafl drives TT recovery and thence improves cardiac function. Irrespective of these potential caveats, the dependence of TT maintenance on AmpII8 and the demonstration in the present study that AmpII drives de novo tubule formation in cardiac myocytes highlights AmpII as a target for future studies. Selectively manipulating AmpII protein abundance in established and evolving cardiac disease states in future studies could resolve this question.

Limitations. Tere are some limitations to the present study that arise primarily from the lack of suitable reagents exhibiting satisfactory cross-reactivity with sheep. As such, we have been unable to serially assess natriurtetic peptide (BNP or NT-proBNP) levels in plasma to monitor the progression and resolution of HF. However, we do observe a strong up-regulation of myocardial NPPB (BNP) mRNA in HF and normalization in the tadalafl treated animals using a single time point assay in each group of experimental animals. Te main operator was not blinded to treatment and the presence of signs of HF (lethargy, dyspnea etc) are subjectively assessed; critically however, the changes in myocardial NPPB (BNP) mRNA validate the qualitative assessments regarding symptom free survival. Additionally, we have been unable to determine PDE5A protein abundance and have relied on quantitative PCR based assessments. Whilst our work shows no change in the myocardial cGMP content we cannot rule out the possibility that cGMP is elevated in certain sub-cellular pools and thus mediates the efects of tadalafl treatment that we observe. FRET based sensors may ofer some potential to assess sub-cellular cyclic nucleotide pools, their use in sheep would require maintenance of primary cardiac myocytes in extended culture to allow transgene expression and thus signifcant concerns would arise from cellular de-diferentiation and ultrastructural changes that inevitably occur when adult cardiac cells are maintained in culture conditions. A further limitation in the present study arises from the anatomical alignment over the cardiac apex over the sternum in adult sheep and inaccessibility to a suitable imaging window to allow four-chamber apical views for a thorough echocardiographic assessment of ejection fraction and E/A ratios. Furthermore, we have found cardiac conductance catheterization to be of limited use given interference from the pacing lead in the right ventricle. Nevertheless, short axis trans-thoracic imaging is possible and clearly shows marked cardiac dilatation and reduced contractility in HF that is at least partially normalized by tadalafl treatment. In summary, using a large animal model of end-stage HF we have shown that chronic treatment with the PDE5 inhibitor tadalafl reverses already established cellular ultra-structural remodeling and impaired catecholamine responses and also restores systolic calcium in HF. Given the localization of β-ARs and G-protein coupled regulators of cardiac catecholamine responses across the TT and surface sarcolemma, the recovery of TTs following PDE5 inhibition provides a cellular mechanism for the restoration of catecholamine reserve in tadalafl treated animals.

Scientific Reports | (2019)9:6801 | https://doi.org/10.1038/s41598-019-42592-1 11 www.nature.com/scientificreports/ www.nature.com/scientificreports

Methods Ethical Approval. All experiments were conducted in accordance with The United Kingdom Animals (Scientifc Procedures) Act, 1986 and European Union Directive EU/2010/63. Local ethical approval was obtained from Te University of Manchester Animal Welfare and Ethical Review Board. Reporting of animal experiments is in accordance with Te ARRIVE guidelines75.

Ovine HF model and drug treatment. HF was induced in 69 female Welsh Mountain sheep (~18 months) by right ventricular tachypacing as described previously1,8,32–34,76,77. In brief, anesthesia was induced and maintained by isofurane inhalation (1–5% in oxygen). Under aseptic conditions using transvenous approaches a single bipolar endocardial pacing lead was actively fxed to the right ventricular apical endocardium and connected to a cardiac pacemaker (e.g. Medtronic Sensia, Medtronic Inc. USA) and buried subcutaneously in the right pre-scapular region. Peri-operative analgesia (meloxicam, 0.5 mg/kg) and antibiosis (enrofoxacin 5 mg/kg or oxytetracycline 20 mg/kg) were administered subcutaneously and animals allowed to recover post-operatively for at least 1 week prior to commencement of tachypacing (210 beats per minute; bpm). Animals were monitored at least once daily for onset of clinical signs of HF including lethargy, dyspnea and weight loss. Following 28 days of tachypacing animals (N = 27) were randomly assigned to receive the PDE5i tadalafl. Treated animals were administered 20 mg tadalafl (Cialis; Lilly, Netherlands) orally once daily for a further 3 weeks and tachypacing (210 bpm) was continued throughout the treatment period. Te designated endpoint for the study was either the onset of signs of HF (dyspnea, lethargy, weight loss) or 49 days of tachypacing (21 days tadalafl treatment).

In vivo assessments of cardiac contractility and catecholamine responsiveness. During echocardiographic and electrocardiographic assessments tachypacing was stopped for 15 minutes prior to data acquisition. Following in vivo functional assessments tachypacing was continued at 210 bpm. Cardiac dimensions and contractility were assessed by right para-sternal transthoracic echocardiography (Vivid 7; GE Healthcare, UK) in non-sedated conscious sheep gently restrained in a supine position. Due to the anatomical alignment of the ventricular apex over the sternum in sheep we are unable to obtain satisfactory 4-chamber long axis views to calculate ejection fraction. However, fractional shortening was measured using m-mode imaging at a level basal to the papillary muscles and fractional area change calculated from short-axis 2d views at the mid-papillary level31. Heart rate was determined from a five-lead ECG (Iox, EMKA Technologies, France). The chronotropic response to β-adrenergic stimulation was assessed as described previously31 by intravenous infusion of dobutamine hydrochloride (0.5–20 µg/kg/min) prepared in 0.9% saline (Baxter, UK) and delivered using an infusion pump (Harvard Apparatus, UK). Heart rate was determined afer 5 mins of dobutamine infusion at each dose and then the infusion rate was doubled to a maximum dose of 20 µg/kg/min or until the heart rate reached 240 bpm.

Transverse tubule staining and quantifcation. Single lef ventricular mid-myocardial myocytes were isolated using a collagenase and protease digestion technique as described in detail previously1,8,32,77. Briefy, animals were killed by an overdose of pentobarbitone (200 mg/kg intravenously). Heparin (10,000 units) was also used to prevent coagulation. Te heart was rapidly excised and the lef anterior descending coronary artery cannulated and perfused with collagenase (Worthington Type II, 0.1 mg/ml) and protease (from Streptomyces gri- seus, 0.02 mg/ml) containing solutions and single myocytes dispersed by gentle trituration in a taurine solution containing (in mmol/l): NaCl, 113; taurine, 50; glucose, 11; HEPES, 10; BDM, 10; KCl, 4; MgSO4, 1.2; Na2HPO4, −1 2+ 1.2; CaCl2, 0.1 and BSA, 0.5 mg.ml (pH 7.34 with NaOH). Changes in intracellular Ca concentration were measured using Fura-2 at 37 °C in cells electrically stimulated at 0.5 Hz as we have described previously1,33,78. TT’s were visualized by staining with di-4-ANEPPS (4 µmol/l. Invitrogen, UK) as described in detail pre- viously8,32,70. Stained myocytes were imaged by confocal microscopy (Leica SP2 or Zeiss 7Live; 488 nm excitation and 505 nm long pass emission settings) at 100 nm XY pixel dimensions and 162 nm Z-step size. Images were deconvolved using the microscope specifc derived point spread function (PSF) using Huygens Sofware (Scientifc Volume Imaging, Netherlands)8,32,70,79. Following background correction and image thresholding the following quantitative assessments of TT’s were calculated: i) TT fractional area derived from binarized images as the pixels occupied by TT’s as a fraction of the planar cell area (ImageJ. NIH, USA), ii) the distance 50% of voxels in the cell are from the nearest membrane (TT or surface sarcolemma), hereafer referred to as the half-distance, was determined using custom algorithms written in IDL (Exelisvis, UK)8,32,70. Images were frst interpolated to give the same X, Y, Z spacing (‘congrid’) and then Euclidean distance maps determined (‘morph_distance’) and, iii) TT orientation was determined from skeletonized binary images as described previously8,57 using the ImageJ plugin ‘Directionality’.

mRNA, protein expression and cGMP content analysis. Changes in protein expression were assessed in samples using approaches described in detail previously1,8,33. Following cardiac excision, a ~1 cm3 region of the posterior lef ventricular free wall was removed and snap frozen and stored in liquid nitrogen until analysis. Whole myocardium homogenates (~100 mg starting material) were prepared in RIPA bufer with protease and phosphatase inhibitors (0.1 mg/ml phenylmethanesulphonylfuroide, 100 mmol/l sodium orhtovanadate, 1 mg/ ml aprotonin, 1 mg/ml leupeptin) and protein content determined (DC Protein Assay, BioRad, UK). Proteins were separated by PAGE, transferred to nitrocellulose membranes and protein expression determined using the antibodies and blotting conditions outlined in Table 2. For PKG protein measurements, sample preparation and blots were conducted under non-reducing conditions. Membranes were visualized by chemiluminescence (Syngene, UK) or IR-Dye labeled secondary antibodies (Licor, UK). Previously we determined that the ‘classical’

Scientific Reports | (2019)9:6801 | https://doi.org/10.1038/s41598-019-42592-1 12 www.nature.com/scientificreports/ www.nature.com/scientificreports

Protein loaded (µg) Blocker Primary Antibody Secondary antibody Detection Amphiphysin II (AmpII) 40 μg 5% BSA 1:200, sc-23918 (Santa Cruz Biotechnology) 1:40,000 HRP conjugated anti-mouse Chemi-luminescence Junctophilin2 (JPH2) 10 μg SEABLOCK 1:1,000, sc-51313 (Santa Cruz Biotechnology) 1:40,000 HRP conjugated anti-goat Chemi-luminescence Telethonin (TCAP) 20 μg 5% Milk 1:500, ab133646 (Abcam) 1:40,000 HRP conjugated anti-rabbit Chemi-luminescence Myotubularin (MTM1) 40 μg 10% Milk 1:500, ab103626 (Abcam) 1:40,000 HRP conjugated anti-rabbit Chemi-luminescence

β1-adrenergic receptor (β1-AR) 20 μg 10% Milk 1:2,000, sc-568 (Santa Cruz Biotechnology) 1:20,000 IRDye 800CW anti-rabbit Fluorescent

β2-adrenergic receptor (β2-AR) 30 μg 10% Milk 1:1000, ab137494 (Abcam) 1:20,000 IRDye 800CW anti-rabbit Fluorescent G-protein receptor kinase 2 (GRK2) 20 μg 10% Milk 1:1,000, sc-562 (Santa Cruz Biotechnology) 1:20,000 IRDye 800CW anti-rabbit Fluorescent Protein phosphatase 1 (PP1) 20 μg 10% Milk 1:1,000, sc-7482 (Santa Cruz Biotechnology) 1:20,000 IRDye 800CW anti-mouse Fluorescent Protein phosphatase 2a (PP2a) 20 μg 10% Milk 1:1500, sc-80665 (Santa Cruz Biotechnology) 1:20,000 IRDye 800CW anti-mouse Fluorescent Protein kinase G (PKG) 20 μg 10% Milk 1:1,000, ADI-KAP-PK005 (Enzo Life Sciences) 1:10,000 IRDye 800CW anti-rabbit Fluorescent PDE2A 20 μg 5% Milk 1:750 ab55555 (Abcam) 1:10,000 IRDye 800CW anti-mouse Fluorescent PDE3A 20 µg SuperBlock 1:1,000, sc-11830 (Santa Cruz Biotechnology) 1:5,000 HRP conjugated anti-goat Chemi-luminescence β-actin (loading ctrl) — — 1: 5,000, sc-47778 (Santa Cruz Biotechnology) 1:40,000 HRP conjugated anti-mouse —

Table 2. Blotting conditions for selected proteins.

housekeeping proteins GAPDH and β-actin have altered expression in various disease states and therefore we normalized expression of our protein of interest to an internal control (IC) common to each blot and total protein transferred to the membrane determined either by Ponceau S staining or REVERT total protein stain (Licor, UK). Te IC was derived from a single control sheep (which itself was not included in the present study). Blots were repeated in triplicate on separate occasions and data averaged to minimize any efects due to pipetting or protein transfer errors. Full length blots and validation controls are shown in Supplemental Information (Fig. S1). Total RNA was isolated using Trizol reagent, purifed using RNAeasy columns (Qiagen, UK) following manufacturers recommendations. RNA yield and integrity were determined using a Nanodrop (TermoFisher, UK) and TapeStation (Agilent, UK) instruments respectively. 1 µg total RNA (RIN scores >7.0) was then reverse tran- scribed using a high capacity RNA to cDNA kit (Applied Biosystems, UK) following manufacturers steps. PDE5A (Oa04859755_m1) and NPPB (Oa04931155_g1) mRNA abundance was determined using Taqman single tube FAM-MGB assays (TermoFisher, UK) following the manufacturers protocols and normalized to either RPLPO −ΔΔC alone (Oa04824512_g1) or TBP (Oa0481075_m1) and YWHAZ (Oa03216375_gH) combined using the 2 t method (Supplemental Information, Fig. 2). All qPCR experiments were repeated 3 times and data averaged. For AmpII (BIN1) isoform determination the variably expressed region between exon 6 and exon 18 (Supplemental Information, Fig. 3) was assessed using standard PCR approaches using primers (Eurofns, Germany; forward) 5’-TACGAGTCCCTTCAAACCGC, (reverse) 5’-AGTGTCAACGGCTCTTCCAG. Phusion high fdelity DNA polymerase with GC rich bufer was used for hot start PCR amplifcation (98 °C, 30 sec; 40 cycles; 98 °C - 10 sec, 62.1 °C - 30 sec, 72 °C - 30 sec; 72 °C - 10 min) and amplicons separated using a Qiaxcel Advanced Separation System using 15–3000 bp markers (Qiagen, UK). Myocardial cGMP content was determined by competitive ELISA (Amersham, UK) following the manufacturers non-acetylation procedure. Tissue was homogenized in 6% (w/v) trichloracetic acid, centrifuged and the supernatant washed 4 times before lyophilisation under nitrogen at 60 °C and proceeding to the immunoassay.

Isolation of Neonatal Rat Ventricular Myocyte, induced pluripotent stem cell cardiac myocyte culture and AmpII overexpression. Neonatal rat ventricular myocytes (NRVMs) were isolated from 2-day old Wistar rats and plated onto 8 well plates (Ibidi µ-plates; Tistle Scientifc UK) at a density of ~3 × 105 cells/ml in media (containing: DMEM, 68%, M199, 17%; normal horse serum, 10%; fetal bovine serum, 5%; all Gibco Life Technologies, UK). Fungizone, Penicillin (10,000 units/ml) and streptomycin (10 mg/ml) were added to a fnal concentration of 1% and bromodeoxyuridone (100 µmol/l) added to retard fbroblast proliferation. Cells maintained at 37 °C in a 5% CO2 incubator and media changed every 2–3 days. In brief, rat pups were killed by cervical dislocation, bodies rinsed in 70% ethanol and the ventricles rapidly excised and placed in ice cold dissociation bufer (mmol/l: NaCl, 116; HEPES, 20; glucose, 5.6; KCl, 5.4; NaH2PO4, 1; MgSO4, 0.83; pH 7.35). Single NRVMs were digested using 0.75 mg/mL collagenase A (Roche, UK) and 1.3 mg/ml pancreatin (Sigma, UK) by gentle stirring (120 rpm, 7 minutes, 37 °C) and trituration with a 25 ml Stripette. NRVMs (1–3 days post isolation) were transiently transfected with either variant 8 of human BIN1 (AmpII) (Origene Inc, USA) cloned into pCMV6-AC-mKate2 entry vector (Origene Inc, USA) or pCMV6-mKate2 as a negative control. Plasmid DNA (6 µg) was mixed (1: 3 ratio) with transfection reagent (Fugene 6; Promega, UK) in reduced serum media (OptiMEM; Gibco Life Technologies, UK) and cells transfected and maintained in a 5% CO2 incubator at 37 °C for 48 hours prior to visualization using a Nikon A1R+ confocal microscope (excitation, 561 nm; emission 595 ± 50 nm). Te density (fractional area) of tubular structures was determined as described above for adult ventricular myocytes. Human induced pluripotent stem cell derived cardiac myocytes (iPSC CMs) were purchased (Cellular Dynamics International, USA) and stored in liquid nitrogen until use. Cells were thawed, plated (~6 × 104 cells/cm2) and maintained as per the manufacturer’s instructions and using the manufacturer’s supplied media (iCell cardiomyocyte plating and culture media). Media was changed every 2 days. Five days afer plating cells were transfected as described for NRVMs. Tubular structures were quantifed by determining the fractional area as described above.

Scientific Reports | (2019)9:6801 | https://doi.org/10.1038/s41598-019-42592-1 13 www.nature.com/scientificreports/ www.nature.com/scientificreports

Determination of plasma tadalafl concentration and plasma protein binding. Plasma tadalafl concentration (N = 13) was determined by mass spectrometry. Briefy whole blood from chronically treated and tachypaced animals was collected from the jugular vein in to K2EDTA blood collection tubes (Vacutainer; Becton Dickinson & Co, UK) on the day of euthanasia and stored on ice. Following centrifugation (4000 g for 15 minutes, 4 °C; within 30 minutes of sample collection) plasma was stored in liquid nitrogen until analysis. Tadalafl concentrations were subsequently determined using a liquid chromatography-mass spectrometry method modifed from Kim et al.80. Briefy, following metabolite extraction samples were spiked with heavy stable-isotope containing Tadalafl and Sildenafl (Toronto Research Chemicals, Canada). Each sample was analysed using a Selected Reaction Monitoring method on a TermoFisher Accela UHPLC coupled to a TermoFisher TSQ Vantage, using transitions described by Kim et al.81 with sample blanks and tadalafl spiked samples analysed periodically to ensure quantitative accuracy. Tadalafl concentration was calculated by comparing the peak area of the endogenous compound to that of its heavy labelled counterpart. Tadalafl plasma protein binding was also determined by mass spectrometry following rapid equilibrium dialysis (TermoFisher, UK). K2EDTA plasma samples (100 µl) from tadalafl-treated animals were dialysed in to phosphate bufered saline following manufacturers instructions. Following dialysis equal volumes of control plasma or phosphate bufered saline were added respectively to the dialysate and sample plasma prior to mass spectrometry. Te percentage plasma protein bound tadalafl was calculated as (dialysate concentration/sample concentration) × 100.

Statistics. Data are presented as mean ± standard error of the mean (SEM) for n observations/N experiments (animals). Where multiple observations (n) have been obtained from the same animal (N) linear mixed modeling (SPSS Statistics. IBM, USA) was performed thus accounting for the nested (clustered) design of the experiment e.g. multiple observations from the same heart and replicate Western blots. Data were log10 transformed prior to linear mixed modeling82. Dose response and echocardiographic experiments were assessed using a linea mixed model (‘xtmixed’; Stata, StataCorp USA), one-way analysis of variance with Bonferroni correction for multiple comparisons (‘oneway’; Stata, Stacorp USA) or repeated measures ANOVA with Holm-Sidak multiple comparison correction as indicated in fgure legends. Survival data and Cox’s proportional hazard ratios were also assessed using Stata (‘stcox’). Linear regressions were ftted to non-paired data (Fig. 5) by simulating 1000 randomly sampled (7 points in each condition without replacement) data subsets and ftting each of these with a linear regression. Te mean slope and 95% confdence limits were then obtained and superimposed on the plotted mean ± SEM data points. All values were considered signifcant when p < 0.05. Data & Code Availability Te data that support the fndings of this study are available from the corresponding author on reasonable request. Similarly, computer code (IDL) used for assessment of TT density are available from the corresponding author on reasonable request. References 1. Briston, S. J. et al. Impaired β-adrenergic responsiveness accentuates dysfunctional excitation-contraction coupling in an ovine model of tachypacing-induced heart failure. J Physiol 589, 1367–1382, https://doi.org/10.1113/jphysiol.2010.203984 (2011). 2. Borlaug, B. A. et al. Impaired chronotropic and vasodilator reserves limit exercise capacity in patients with heart failure and a preserved ejection fraction. Circulation 114, 2138–2147, https://doi.org/10.1161/CIRCULATIONAHA.106.632745 (2006). 3. Raake, P. W. J. et al. AAV6.βARKct cardiac gene therapy ameliorates cardiac function and normalizes the catecholaminergic axis in a clinically relevant large animal heart failure model. Eur Heart J. https://doi.org/10.1093/eurheartj/ehr447 (2012). 4. El-Armouche, A., Pamminger, T., Ditz, D., Zolk, O. & Eschenhagen, T. Decreased protein and phosphorylation level of the protein phosphatase inhibitor-1 in failing human hearts. Cardiovasc Res 61, 87–93, https://doi.org/10.1016/j.cardiores.2003.11.005 (2004). 5. Nikolaev, V. O. et al. β2-adrenergic receptor redistribution in heart failure changes cAMP compartmentation. Science 327, 1653–1657, https://doi.org/10.1126/science.1185988 (2010). 6. Lafamme, M. A. & Becker, P. L. Gs and adenylyl cyclase in transverse tubules of heart: implications for cAMP-dependent signaling. Am J Physiol 277, H1841–1848, https://doi.org/10.1152/ajpheart.1999.277.5.H1841 (1999). 7. Cros, C. & Brette, F. Functional subcellular distribution of β1- and β2-adrenergic receptors in rat ventricular cardiac myocytes. Physiol Rep 1, e00038, https://doi.org/10.1002/phy2.38 (2013). 8. Caldwell, J. L. et al. Dependence of cardiac transverse tubules on the BAR domain protein amphiphysin II (BIN-1). Circ Res 115, 986–996, https://doi.org/10.1161/CIRCRESAHA.116.303448 (2014). 9. He, J. et al. Reduction in density of transverse tubules and L-type Ca(2+) channels in canine tachycardia-induced heart failure. Cardiovasc Res 49, 298–307, https://doi.org/10.1016/S0008-6363(00)00256-X (2001). 10. Louch, W. E. et al. T-tubule disorganization and reduced synchrony of Ca2+ release in murine cardiomyocytes following myocardial infarction. J Physiol 574, 519–533, https://doi.org/10.1113/jphysiol.2006.107227 (2006). 11. Crossman, D. J., Ruygrok, P. N., Soeller, C. & Cannell, M. B. Changes in the organization of excitation-contraction coupling structures in failing human heart. PLoS One 6, e17901, https://doi.org/10.1371/journal.pone.0017901 (2011). 12. Tsai, E. J. & Kass, D. A. Cyclic GMP signaling in cardiovascular pathophysiology and therapeutics. Pharmacol Ter 122, 216–238, https://doi.org/10.1016/j.pharmthera.2009.02.009 (2009). 13. Senzaki, H. et al. Cardiac phosphodiesterase 5 (cGMP-specifc) modulates β-adrenergic signaling in vivo and is down-regulated in heart failure. FASEB J 15, 1718–1726 (2001). 14. Takimoto, E. et al. Compartmentalization of cardiac β-adrenergic inotropy modulation by phosphodiesterase type 5. Circulation 115, 2159–2167, https://doi.org/10.1161/CIRCULATIONAHA.106.643536 (2007). 15. Castro, L. R. V., Schittl, J. & Fischmeister, R. Feedback control through cGMP-dependent protein kinase contributes to diferential regulation and compartmentation of cGMP in rat cardiac myocytes. Circ Res 107, 1232–1240, https://doi.org/10.1161/CIRCRESAHA. 110.226712 (2010). 16. Takimoto, E. et al. cGMP catabolism by phosphodiesterase 5A regulates cardiac adrenergic stimulation by NOS3-dependent mechanism. Circ Res 96, 100–109, https://doi.org/10.1161/01.RES.0000152262.22968.72 (2005). 17. Stangherlin, A. et al. cGMP signals modulate cAMP levels in a compartment-specifc manner to regulate catecholamine-dependent signaling in cardiac myocytes. Circ Res 108, 929–939, https://doi.org/10.1161/CIRCRESAHA.110.230698 (2011). 18. Mehel, H. et al. Phosphodiesterase-2 is up-regulated in human failing hearts and blunts β-adrenergic responses in cardiomyocytes. J Am Coll Cardiol 62, 1596–1606, https://doi.org/10.1016/j.jacc.2013.05.057 (2013).

Scientific Reports | (2019)9:6801 | https://doi.org/10.1038/s41598-019-42592-1 14 www.nature.com/scientificreports/ www.nature.com/scientificreports

19. Anderson, S. G. et al. Phosphodiesterase type-5 inhibitor use in type 2 diabetes is associated with a reduction in all-cause mortality. Heart 102, 1750–1756, https://doi.org/10.1136/heartjnl-2015-309223 (2016). 20. Giannetta, E. et al. Is chronic inhibition of phosphodiesterase type 5 cardioprotective and safe? A meta-analysis of randomized controlled trials. BMC Med 12, 185, https://doi.org/10.1186/s12916-014-0185-3 (2014). 21. Guazzi, M., Vicenzi, M., Arena, R. & Guazzi, M. D. PDE5 inhibition with sildenafl improves lef ventricular diastolic function, cardiac geometry, and clinical status in patients with stable systolic heart failure: results of a 1-year, prospective, randomized, placebo-controlled study. Circ Heart Fail 4, 8–17, https://doi.org/10.1161/CIRCHEARTFAILURE.110.944694 (2011). 22. Xie, Y.-P. et al. Sildenafl prevents and reverses transverse-tubule remodeling and Ca2+ handling dysfunction in right ventricle failure induced by pulmonary artery hypertension. Hypertension 59, 355–362, https://doi.org/10.1161/HYPERTENSIONAHA.111.180968 (2012). 23. Salloum, F. N. et al. Sildenafl (Viagra) attenuates ischemic cardiomyopathy and improves lef ventricular function in mice. Am J Physiol Heart Circ Physiol 294, H1398–1406, https://doi.org/10.1152/ajpheart.91438.2007 (2008). 24. Nagy, O., Hajnal, A., Parratt, J. R. & Végh, A. Sildenafl (Viagra) reduces arrhythmia severity during ischaemia 24 h afer oral administration in dogs. Br J Pharmacol 141, 549–551, https://doi.org/10.1038/sj.bjp.0705658 (2004). 25. Ockaili, R., Salloum, F., Hawkins, J. & Kukreja, R. C. Sildenafl (Viagra) induces powerful cardioprotective efect via opening of mitochondrial KATP channels in rabbits. Am J Physiol Heart Circ Physiol 283, H1263–1269, https://doi.org/10.1152/ ajpheart.00324.2002 (2002). 26. Refelmann, T. & Kloner, R. A. Efects of sildenafl on myocardial infarct size, microvascular function, and acute ischemic lef ventricular dilation. Cardiovasc Res 59, 441–449 (2003). 27. Huang, C. K. et al. Sildenafl ameliorates lef ventricular T-tubule remodeling in a pressure overload-induced murine heart failure model. Acta Pharmacol Sin 37, 473–482, https://doi.org/10.1038/aps.2016.13 (2016). 28. Takimoto, E. et al. Chronic inhibition of cyclic GMP phosphodiesterase 5A prevents and reverses cardiac hypertrophy. Nat Med 11, 214–222, https://doi.org/10.1038/nm1175 (2005). 29. Nagayama, T. et al. Sildenafl stops progressive chamber, cellular, and molecular remodeling and improves calcium handling and function in hearts with pre-existing advanced hypertrophy caused by pressure overload. J Am Coll Cardiol 53, 207–215, https://doi. org/10.1016/j.jacc.2008.08.069 (2009). 30. Eskesen, K. et al. Sildenafil treatment attenuates ventricular remodeling in an experimental model of aortic regurgitation. Springerplus 4, 592, https://doi.org/10.1186/s40064-015-1317-8 (2015). 31. Horn, M. A. et al. Temporal Development of Autonomic Dysfunction in Heart Failure: Efects of Age in an Ovine Rapid-pacing Model. J Gerontol A Biol Sci Med Sci, https://doi.org/10.1093/gerona/glv217 (2015). 32. Dibb, K. M. et al. Characterization of an extensive transverse tubular network in sheep atrial myocytes and its depletion in heart failure. Circ Heart Fail 2, 482–489, https://doi.org/10.1161/CIRCHEARTFAILURE.109.852228 (2009). 33. Clarke, J. D. et al. Perturbed atrial calcium handling in an ovine model of heart failure: Potential roles for reductions in the L-type calcium current. J Mol Cell Cardiol 79, 169–179, https://doi.org/10.1016/j.yjmcc.2014.11.017 (2015). 34. Horn, M. A. et al. Age-related divergent remodeling of the cardiac extracellular matrix in heart failure: Collagen accumulation in the young and loss in the aged. Journal of Molecular and Cellular Cardiology 53, 82–90, https://doi.org/10.1016/j.yjmcc.2012.03.011 (2012). 35. Coletta, C. et al. Prognostic value of lef ventricular volume response during dobutamine stress echocardiography. Eur Heart J 18, 1599–1605 (1997). 36. Forgue, S. T. et al. Tadalafl pharmacokinetics in healthy subjects. Br J Clin Pharmacol 61, 280–288, https://doi.org/10.1111/j.1365- 2125.2005.02553.x (2006). 37. Hiemstra, J. A. et al. Saxagliptin and Tadalafl Diferentially Alter Cyclic Guanosine Monophosphate (cGMP) Signaling and Lef Ventricular Function in Aortic-Banded Mini-Swine. J Am Heart Assoc 5, e003277, https://doi.org/10.1161/JAHA.116.003277 (2016). 38. Zhou, Z. et al. Protein kinase G phosphorylates soluble guanylyl cyclase on serine 64 and inhibits its activity. Arterioscler Tromb Vasc Biol 28, 1803–1810, https://doi.org/10.1161/ATVBAHA.108.165043 (2008). 39. Wei, S. et al. T-Tubule Remodeling During Transition From Hypertrophy to Heart Failure. Circ Res 107, 521–531, https://doi. org/10.1161/CIRCRESAHA.109.212324 (2010). 40. Hong, T.-T. et al. BIN1 is reduced and Cav1.2 trafcking is impaired in human failing cardiomyocytes. Heart Rhythm 9, 812–820, https://doi.org/10.1016/j.hrthm.2011.11.055 (2012). 41. Hong, T. et al. Cardiac BIN1 folds T-tubule membrane, controlling ion fux and limiting arrhythmia. Nature Medicine 20, 624–632, https://doi.org/10.1038/nm.3543 (2014). 42. Bischof, E. Potency, selectivity, and consequences of nonselectivity of PDE inhibition. Int J Impot Res 16(Suppl 1), S11–14, https:// doi.org/10.1038/sj.ijir.3901208 (2004). 43. Zhang, X., Feng, Q. & Cote, R. H. Efficacy and selectivity of phosphodiesterase-targeted drugs in inhibiting photoreceptor phosphodiesterase (PDE6) in retinal photoreceptors. Invest Ophthalmol Vis Sci 46, 3060–3066, https://doi.org/10.1167/iovs.05-0257 (2005). 44. Wright, P. J. Comparison of phosphodiesterase type 5 (PDE5) inhibitors. Int J Clin Pract 60, 967–975, https://doi.org/10.1111/j.1742- 1241.2006.01049.x (2006). 45. Weeks, J. L. et al. High biochemical selectivity of tadalafl, sildenafl and vardenafl for human phosphodiesterase 5A1 (PDE5) over PDE11A4 suggests the absence of PDE11A4 cross-reaction in patients. Int J Impot Res 17, 5–9, https://doi.org/10.1038/sj.ijir.3901283 (2005). 46. Loughney, K., Taylor, J. & Florio, V. A. 3’,5’-cyclic nucleotide phosphodiesterase 11A: localization in human tissues. Int J Impot Res 17, 320–325, https://doi.org/10.1038/sj.ijir.3901317 (2005). 47. Mehrotra, N., Gupta, M., Kovar, A. & Meibohm, B. Te role of pharmacokinetics and pharmacodynamics in phosphodiesterase-5 inhibitor therapy. Int J Impot Res 19, 253–264, https://doi.org/10.1038/sj.ijir.3901522 (2007). 48. Chau, V. Q., Salloum, F. N., Hoke, N. N., Abbate, A. & Kukreja, R. C. Mitigation of the progression of heart failure with sildenafl involves inhibition of RhoA/Rho-kinase pathway. Am J Physiol Heart Circ Physiol 300, H2272–2279, https://doi.org/10.1152/ ajpheart.00654.2010 (2011). 49. Westermann, D. et al. Selective PDE5A inhibition with sildenafl rescues lef ventricular dysfunction, infammatory immune response and cardiac remodeling in angiotensin II-induced heart failure in vivo. Basic Res Cardiol 107, 308, https://doi.org/10.1007/ s00395-012-0308-y (2012). 50. Nikolaev, V. O., Bünemann, M., Schmitteckert, E., Lohse, M. J. & Engelhardt, S. Cyclic AMP imaging in adult cardiac myocytes reveals far-reaching β1-adrenergic but locally confned β2-adrenergic receptor-mediated signaling. Circ Res 99, 1084–1091, https:// doi.org/10.1161/01.RES.0000250046.69918.d5 (2006). 51. Katz, M. G. et al. AAV6-βARKct gene delivery mediated by molecular cardiac surgery with recirculating delivery (MCARD) in sheep results in robust gene expression and increased adrenergic reserve. J Torac Cardiovasc Surg 143, 720–726.e723, https://doi. org/10.1016/j.jtcvs.2011.08.048 (2012). 52. Carr, A. N. et al. Type 1 phosphatase, a negative regulator of cardiac function. Mol Cell Biol 22, 4124–4135 (2002). 53. Wijnker, P. J. et al. Protein phosphatase 2A afects myoflament contractility in non-failing but not in failing human myocardium. J Muscle Res Cell Motil 32, 221–233, https://doi.org/10.1007/s10974-011-9261-x (2011). 54. Lei, M., Wang, X., Ke, Y. & Solaro, R. J. Regulation of Ca2+ transient by PP2A in normal and failing heart. Front Physiol 6, 13, https:// doi.org/10.3389/fphys.2015.00013 (2015).

Scientific Reports | (2019)9:6801 | https://doi.org/10.1038/s41598-019-42592-1 15 www.nature.com/scientificreports/ www.nature.com/scientificreports

55. Seidel, T. et al. Sheet-Like Remodeling of the Transverse Tubular System in Human Heart Failure Impairs Excitation-Contraction Coupling and Functional Recovery by Mechanical Unloading. Circulation, https://doi.org/10.1161/CIRCULATIONAHA.116.024470 (2017). 56. Lenaerts, I. et al. Ultrastructural and functional remodeling of the coupling between Ca2+ infux and sarcoplasmic reticulum Ca2+ release in right atrial myocytes from experimental persistent atrial fbrillation. Circ Res 105, 876–885, https://doi.org/10.1161/ CIRCRESAHA.109.206276 (2009). 57. Wagner, E. et al. Stimulated emission depletion live-cell super-resolution imaging shows proliferative remodeling of T-tubule membrane structures afer myocardial infarction. Circ Res 111, 402–414, https://doi.org/10.1161/CIRCRESAHA.112.274530 (2012). 58. Frisk, M. et al. Elevated ventricular wall stress disrupts cardiomyocyte t-tubule structure and calcium homeostasis. Cardiovasc Res 112, 443–451, https://doi.org/10.1093/cvr/cvw111 (2016). 59. McNary, T. G., Bridge, J. H. & Sachse, F. B. Strain transfer in ventricular cardiomyocytes to their transverse tubular system revealed by scanning confocal microscopy. Biophys J 100, L53–55, https://doi.org/10.1016/j.bpj.2011.03.046 (2011). 60. Li, H. et al. Cardiac Resynchronization Terapy Reduces Subcellular Heterogeneity of Ryanodine Receptors, T-Tubules, and Ca2+ Sparks Produced by Dyssynchronous Heart Failure. Circ Heart Fail 8, 1105–1114, https://doi.org/10.1161/CIRCHEARTFAILURE.115.002352 (2015). 61. Lyon, A. R. et al. Plasticity of Surface Structures and β2-Adrenergic Receptor Localization in Failing Ventricular Cardiomyocytes During Recovery from Heart Failure. Circ Heart Fail 5, 357–365, https://doi.org/10.1161/CIRCHEARTFAILURE.111.964692 (2012). 62. Ibrahim, M. et al. Mechanical unloading reverses transverse tubule remodelling and normalizes local Ca2+-induced Ca2+release in a rodent model of heart failure. Eur J Heart Fail 14, 571–580, https://doi.org/10.1093/eurjhf/hfs038 (2012). 63. Ibrahim, M., Gorelik, J., Yacoub, M. H. & Terracciano, C. M. Te structure and function of cardiac t-tubules in health and disease. Proc Biol Sci 278, 2714–2723, https://doi.org/10.1098/rspb.2011.0624 (2011). 64. Dowling, J. J. et al. Loss of myotubularin function results in T-tubule disorganization in zebrafsh and human myotubular myopathy. PLoS Genet 5, e1000372, https://doi.org/10.1371/journal.pgen.1000372 (2009). 65. Al-Qusairi, L. & Laporte, J. T-tubule biogenesis and triad formation in skeletal muscle and implication in human diseases. Skelet Muscle 1, 26, https://doi.org/10.1186/2044-5040-1-26 (2011). 66. Razzaq, A. et al. Amphiphysin is necessary for organization of the excitation-contraction coupling machinery of muscles, but not for synaptic vesicle endocytosis in Drosophila. Genes Dev 15, 2967–2979, https://doi.org/10.1101/gad.207801 (2001). 67. Hong, T.-T. et al. BIN1 localizes the L-type calcium channel to cardiac T-tubules. PLoS Biol 8, e1000312, https://doi.org/10.1371/ journal.pbio.1000312 (2010). 68. Mata, D. L. A. et al. BIN1 Induces the Formation of T-Tubules and Adult-Like Ca2+ Release Units in Developing Cardiomyocytes. Stem Cells 37, 54–64, https://doi.org/10.1002/stem.2927 (2019). 69. Wechsler-Reya, R., Sakamuro, D., Zhang, J., Duhadaway, J. & Prendergast, G. C. Structural analysis of the human BIN1 gene. Evidence for tissue-specifc transcriptional regulation and alternate RNA splicing. J Biol Chem 272, 31453–31458 (1997). 70. Richards, M. A. et al. Transverse (t-) tubules are a common feature in large mammalian atrial myocytes including human. Am J Physiol Heart Circ Physiol 301, H1996–2005, https://doi.org/10.1152/ajpheart.00284.2011 (2011). 71. Agrawal, P. B. et al. SPEG interacts with myotubularin, and its deficiency causes centronuclear myopathy with dilated cardiomyopathy. Am J Hum Genet 95, 218–226, https://doi.org/10.1016/j.ajhg.2014.07.004 (2014). 72. Royer, B. et al. Te myotubularin-amphiphysin 2 complex in membrane tubulation and centronuclear myopathies. EMBO Rep 14, 907–915, https://doi.org/10.1038/embor.2013.119 (2013). 73. Kutchukian, C. et al. Phosphatidylinositol 3-kinase inhibition restores Ca2+ release defects and prolongs survival in myotubularin- defcient mice. Proc Natl Acad Sci USA 113, 14432–14437, https://doi.org/10.1073/pnas.1604099113 (2016). 74. Al-Qusairi, L. et al. T-tubule disorganization and defective excitation-contraction coupling in muscle fbers lacking myotubularin lipid phosphatase. Proc Natl Acad Sci USA 106, 18763–18768, https://doi.org/10.1073/pnas.0900705106 (2009). 75. Kilkenny, C., Browne, W. J., Cuthill, I. C., Emerson, M. & Altman, D. G. Improving bioscience research reporting: the ARRIVE guidelines for reporting animal research. PLoS Biol 8, e1000412, https://doi.org/10.1371/journal.pbio.1000412 (2010). 76. Pinali, C., Bennett, H., Davenport, J. B., Traford, A. W. & Kitmitto, A. 3-D Reconstruction of the Cardiac Sarcoplasmic Reticulum Reveals a Continuous Network Linking T-Tubules: Tis Organization is Perturbed in Heart Failure. Circ Res 113, 1219–1230, https://doi.org/10.1161/CIRCRESAHA.113.301348 (2013). 77. Greensmith, D. J., Galli, G. L. J., Traford, A. W. & Eisner, D. A. Direct measurements of SR free Ca reveal the mechanism underlying the transient efects of RyR potentiation under physiological conditions. Cardiovasc Res 103, 554–563, https://doi.org/10.1093/cvr/ cvu158 (2014). 78. Traford, A. W., O’Neill, S. C. & Eisner, D. A. Factors afecting the propagation of locally activated systolic Ca transients in rat ventricular myocytes. Pfugers Arch 425, 181–183 (1993). 79. Walden, A. P., Dibb, K. M. & Traford, A. W. Diferences in intracellular calcium homeostasis between atrial and ventricular myocytes. J Mol Cell Cardiol 46, 463–473, https://doi.org/10.1016/j.yjmcc.2008.11.003 (2009). 80. Kim, K.-Y. et al. Validated UPLC-MS/MS method for the determination of tadalafl in human plasma and its application to a pharmacokinetic study. Translational and Clinical Pharmacology 25, 21–27, https://doi.org/10.12793/tcp.2017.25.1.21 (2017). 81. Clasen, L. et al. A modifed approach for programmed electrical stimulation in mice: Inducibility of ventricular arrhythmias. PLoS One 13, e0201910, https://doi.org/10.1371/journal.pone.0201910 (2018). 82. Keene, O. N. Te log transformation is special. Stat Med 14, 811–819 (1995). Acknowledgements Tis work was supported by grants from Te British Heart Foundation: FS/12/57/29717, CH200004/12801, FS/15/28/31476, FS/10/71/28563, PG/15/70/31724, IG/15/2/31514, FS/09/002/26487, FS/12/34/29565, FS/15/67/32038, FS/10/52/28678 and Medical Research Council: MR/K500823/1, MR/K501211/1. Te authors also acknowledge technical support from Medtronic UK. and Oscor Inc. (USA) in the form of high rate pacing patches for pacing devices and adaptors for pacing leads respectively. Author Contributions Study Conception & Design: A.W.T.; Supervision: A.W.T., D.A.E. and K.M.D.; Animal models, in vivo data generation and analysis: M.L., A.W.T., E.J.R., C.M.P., G.J.K., G.W.P.M., L.S.W., D.C.H. and L.K.B.; mRNA, protein and cGMP studies C.E.R.S., E.J.R., J.L.C. and R.F.T.; transverse tubule analysis and protein regulator studies, J.L.C.; Mass spectrometry: S.J.C. and R.D.U. All authors contributed to manuscript drafing/editing. Additional Information Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-019-42592-1. Competing Interests: Te authors declare no competing interests.

Scientific Reports | (2019)9:6801 | https://doi.org/10.1038/s41598-019-42592-1 16 www.nature.com/scientificreports/ www.nature.com/scientificreports

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afliations. 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 Cre- ative Commons license, and indicate if changes were made. Te 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/.

Scientific Reports | (2019)9:6801 | https://doi.org/10.1038/s41598-019-42592-1 17 Review

Calcium Signaling Series Donald M. Bers, Guest Editor

Calcium and Excitation-Contraction Coupling in the Heart

David A. Eisner, Jessica L. Caldwell, Kornél Kistamás, Andrew W. Trafford

2+ Abstract: Cardiac contractility is regulated by changes in intracellular Ca concentration ([Ca ]i). Normal function 2+ requires that [Ca ]i be sufficiently high in systole and low in diastole. Much of the Ca needed for contraction comes from the sarcoplasmic reticulum and is released by the process of calcium-induced calcium release. The factors that regulate and fine-tune the initiation and termination of release are reviewed. The precise control of intracellular Ca cycling depends on the relationships between the various channels and pumps that are involved. We consider 2 aspects: (1) structural coupling: the transporters are organized within the dyad, linking the transverse tubule and sarcoplasmic reticulum and ensuring close proximity of Ca entry to sites of release. (2) Functional coupling: where the fluxes across all membranes must be balanced such that, in the steady state, Ca influx equals Ca efflux on every beat. The remainder of the review considers specific aspects of Ca signaling, including the role of Ca buffers, mitochondria, 2+ Ca leak, and regulation of diastolic [Ca ]i. (Circ Res. 2017;121:181-195. DOI: 10.1161/CIRCRESAHA.117.310230.) Key Words: calcium ■ cytoplasm ■ mitochondria ■ ryanodine receptor calcium release channel ■ sarcoplasmic reticulum

he process of excitation–contraction (E–C) coupling links arrangement. The depolarization produced by the action poten- Tthe electric excitation of the surface membrane (action po- tial opens L-type Ca channels situated in the surface membrane tential) to contraction. Since the initial measurements in cardiac and transverse tubules. The resulting entry of a small amount 1,2 2+ muscle, an enormous amount of work has shown the under- of Ca results in a large increase of [Ca ]i in the dyadic space 2+ 3,4 lying changes of cytoplasmic calcium concentration ([Ca ]i). (the region bounded by the t-tubule and sarcoplasmic reticu- 2+ Ca binds to troponin resulting in sliding of the thick and thin lum [SR]). This increase of [Ca ]i makes the SR Ca release filaments, cell shortening, and thence the development of pres- channels (ryanodine receptors [RyR]) open thereby releasing sure within the ventricle and ejection of blood. Force, therefore, a much larger amount of Ca from the SR in a process termed depends on the amount of Ca bound to troponin. This will be a calcium-induced calcium release. The magnitude of the rise of 2+ 2+ function of both the magnitude and duration of the rise of [Ca ]i. [Ca ]i depends not only on the structures mentioned above but It will also depend on the strength of Ca binding, a factor that also on Ca binding to buffers and uptake into organelles includ- can be altered genetically,5 is controlled by factors such as phos- ing mitochondria. For relaxation to occur, Ca must be removed phorylation6 and may form the basis of therapeutic interventions.7 from the cytoplasm. This requires that the RyRs close and then Nevertheless, the major factor that regulates contraction is the that Ca is pumped (1) back into the SR, by the SERCA (SR Ca- level of intracellular Ca. As well as focusing on the increase of ATPase) and (2) out of the cell, largely by the sodium–calcium 2+ [Ca ]i during systole, it is important to remember that proper car- exchange (NCX). 2+ diac function requires that force and [Ca ]i relax quickly to low The remainder of this review is in 3 sections. In the enough levels such that the heart can refill with blood. Therefore, first, we discuss recent studies on the spatial organization 2+ both diastolic and systolic [Ca ]i must be tightly regulated; this of the structures responsible for calcium cycling. The sec- regulation is the subject of the current article. ond addresses the general principles that determine how the The events that occur in E–C coupling are now well estab- amplitude of the Ca transient is controlled, with particular lished (Figure 1). The process depends not only on a combi- reference to understanding the importance of Ca flux bal- nation of the properties of Ca channels and transporters but, ance. In the final section, we consider specific steps in Ca equally importantly, also on their precise locations and spatial signaling.

From the Unit of Cardiac Physiology, Division of Cardiovascular Sciences, Manchester Academic Health Sciences Centre, University of Manchester, United Kingdom. Correspondence to David A. Eisner, 3.18 Core Technology Facility, 46 Grafton St, Manchester M13 9NT, United Kingdom. E-mail [email protected] © 2017 The Authors. Circulation Research is published on behalf of the American Heart Association, Inc., by Wolters Kluwer Health, Inc. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution, and reproduction in any medium, provided that the original work is properly cited. Circulation Research is available at http://circres.ahajournals.org DOI: 10.1161/CIRCRESAHA.117.310230 181 182 Circulation Research July 7, 2017

Major changes occur in dyadic structure in heart failure Nonstandard Abbreviations and Acronyms with a reduction in the number of t-tubules in the ventricle27and, CSQ calsequestrin in the atrium, the loss of virtually all.18 T-tubule loss and the MCU mitochondrial Ca uniporter consequent loss of tight coupling between L-type Ca2+ channels NCX sodium–calcium exchange and RyRs result in the so-called orphaned RyRs and a reduction 2+ 28,29 PLN phospholamban in the synchronicity and amplitude of the Ca transient. RyR ryanodine receptor Until recently, it was unclear how many RyRs make up a SERCA sarco/endoplasmic reticulum Ca-ATPase cluster. The advent of super-resolution imaging methods has SOCE store-operated Ca entry provided estimates for the number of RyRs in each cluster (dyad) from 14 in peripheral couplings to 100 in intracel- SR sarcoplasmic reticulum ≈ ≈ lular sites.10,30,31 The significance of RyR cluster size and cluster homogeneity currently lacks direct experimental evidence. Structural Considerations: The Dyad However, simulation studies have related the number of RyRs Transverse (t-) tubules are 150- to 300-nm-wide8,9 deep invag- per cluster (ie, cluster size) and the uniformity of the cluster inations of the surface sarcolemma occurring at the junction (ie, the presence of gaps between individual RyRs) to the prop- of each sarcomere (z-line). They are observed in ventricular erties of (1) Ca2+ sparks32,33 and (2) the synchronicity of systolic myocytes from all mammalian species studied. E–C coupling Ca2+.34 In particular, larger and more uniformly packed clusters depends on the close association between the SR network and were modeled to be more likely to give rise to a Ca2+ spark, and t-tubule membranes.10,11 The junctional SR makes close con- larger clusters uniformly distributed throughout the cell gave tact with the t-tubule membrane so that RyRs on the SR are rise to a more synchronous rising phase of systolic Ca2+ (see very closely apposed (≈15 nm12) to L-type Ca2+ channels on below). However, bridging the gap between measurements of the t-tubule thus forming the cardiac dyad that is fundamental Ca2+ spark sites, systolic Ca2+ synchronicity in living cells, and to the processes initiating the systolic Ca2+ transient. simultaneous or even sequential super-resolution imaging of When viewed end on, t-tubules are seen to radiate the same RyR clusters is an ongoing challenge. It is worthy of throughout the ventricular cell (Figure 1B). The close asso- consideration as to whether RyR expression, distribution, or ciation between the t-tubule and SR ensures the synchronous function changes in cardiac disease states. Many studies have 2+ 13 rise of [Ca ]i during systole. Indeed, chemical detubulation suggested previously that RyR opening is increased in disease in ventricular myocytes with formamide results in a markedly (see below). Beyond these functional changes, a recent study heterogenous Ca2+ transient commencing at the surface sarco- by Li et al35 suggests that the distribution of RyRs also shifts in lemma with a slowly propagating wave of Ca2+ release travel- heart failure with higher densities of receptors being observed ing to the cell center.14 This has similar spatial properties to the at cell ends where t-tubule density is decreased. Thus, to a first systolic Ca2+ transient in those atrial cells lacking t-tubules.15,16 approximation at least the simulation studies noted previously (Although outside the scope of the present article, it should seem to have some experimental basis although considerable be noted that atrial myocytes from larger mammalian species further study in this area is still required. have a well-developed ventricular-like t-tubule network17–19). Beyond the archetypal L-type Ca2+ channel-RyR dyad, The cardiac dyad is a specialized signaling nexus con- consideration should also be given to other Ca2+ regulatory pro- cerned primarily with the initiation of cardiac contraction. teins that may be localized within the dyad. The limited studies Classically, it consists of clusters of L-type Ca2+ channels on that have systematically investigated the localization of NCX the sarcolemma closely apposed (≈15 nm) across the dyadic to the cardiac dyad give divergent results. For example, using cleft to clusters of RyRs on the SR membrane. In addition confocal approaches (thus with limited resolving power be- to these basic requirements for excitation–contraction cou- low ≈200 nm), Scriven et al36 reported very low colocalization pling, the cardiac dyad may also be considered as containing coefficients between NCX and RyRs in rat ventricle, whereas additional structures that may contribute to or modulate Ca2+ RyR and the L-type Ca2+ channel exhibited high colocalization. release from the SR during systole (Figure 1). Of these, the Thus, the authors concluded that NCX did not form part of most extensively studied is NCX that has been argued via its the dyad in ventricular myocytes. In atrial myocytes, the same reverse-mode action to contribute to Ca2+ influx early during group found that some NCX was in the dyad but its degree of the action potential.20 However, assuming dyadic and cytosolic localization there was less than that of the L-type channel.37 intracellular Na+ are similar during diastole (5–10 mmol/L,21) Conversely, using immuno-gold labeling with electron mi- such reverse-mode NCX is thermodynamically limited lead- croscopy methods, Thomas et al38 were able to identify a popula- ing to the suggestion that Na+ entry via voltage-gated Na+ tion of NCX within 100 nm of RyR clusters and concluded that + 2+ channels (INa) may raise dyadic Na sufficiently early during this dyadic NCX had the potential to regulate Ca fluxes within the action potential to facilitate effective reverse-mode NCX. the dyad and influence systolic Ca2+. Consistent with this, a pro- Indeed, Leblanc and Hulme22 first demonstrated the modu- portion of NCX colocalizing with RyRs has also been noted39 with 2+ 2+ lating effect of INa on Ca release from the SR. Subsequent similar implications for dyadic Ca signaling being suggested. experiments suggested that a subpopulation of neuronal Na+ The second protein of interest is SERCA. Available data on channels are localized to the t-tubule and thence dyadic envi- SERCA2 distribution in the heart is surprisingly sparse; howev- ron23–25; however, Brette et al26 also concluded that although er, confocal studies seem to suggest that a significant proportion neuronal Na+ channels were concentrated on the t-tubule, they of SERCA is localized to the z-line40–42 as well as enveloping the were not required for cardiac excitation–contraction coupling. myofilaments.43 Thus, there is a possibility that both NCX and Eisner et al Ca and Cardiac E–C Coupling 183

Figure 1. Structures involved in Ca cycling. A, Schematic diagram. This shows surface membrane, transverse tubule, sarcoplasmic reticulum (SR), and mitochondria, as well as the various channels and transporters mentioned in the text. B, High-resolution transverse section of a ventricular myocyte showing t-tubule network. Reprinted from Jayasinghe et al39 with permission of the publisher. Copyright ©2009, Biophysical Society. C, Cartoon of dyad emphasizing the major proteins involved in Ca cycling. B-AR indicates beta adrenoceptor; MCU, mitochondrial Ca uniporter; NCX, sodium–calcium exchange; NCLX, mitochondrial Na–Ca exchange; PMCA, plasma membrane Ca-ATPase; RyR, ryanodine receptor; and SERCA, sarco/endoplasmic reticulum Ca-ATPase.

SERCA could be sufficiently close to the SR Ca2+ release sites here we can posit that a dyadic population of SERCA and NCX as to modulate dyadic and thence cytosolic Ca2+. In the normal may act as a firebreak and prevent these localized Ca2+ release myocardium, systolic Ca2+ is tightly controlled, and while Ca2+ events from activating adjacent RyR clusters and leading to trig- sparks may occur, these do not ordinarily form proarrhythmic gering of Ca2+ waves. Indeed, there is evidence that SERCA Ca2+ waves. Subsequent sections will address in more detail how activity can modulate the time course of Ca2+ sparks with in- Ca2+ sparks may be self-terminating localized events; however, creasing SERCA activity accelerating Ca2+ spark decay.44 Given 184 Circulation Research July 7, 2017 this position, reductions in SERCA activity (via either reduced How Is Flux Balance Achieved? expression or hypophosphorylation of phospholamban45,46) in This results from the negative feedback scheme of Figure 2A, the diseased heart, especially if coupled to increased RyR den- which illustrates how the cell responds to a situation in which sity35 and thence an increased probability of Ca2+ spark occur- Ca influx is greater than efflux. (1) The imbalance of fluxes rence,32,33 may impair the protective firebreak and facilitate the increases cell and therefore SR Ca. (2) Ca release is a steep formation of Ca2+ waves and triggered activity. function of SR Ca content51,52 and therefore the amplitude of the Ca transient increases. (3) Increasing the amplitude of the Ca Flux Balance Ca transient increases Ca efflux and decreases Ca entry into In the steady state, on each cardiac cycle, the amount of Ca the cell. This is because of a combination of 2 factors52: (1) entering the cell must equal that pumped out. If not, the cell 2+ 53 Ca efflux on NCX is increased by increasing [Ca ]i and (2) would either gain or lose Ca. Imbalances between Ca entry and 2+ increased [Ca ]i increases Ca-dependent inactivation of the exit can only occur transiently and then result in changes of the L-type Ca current.54 (3) This net loss of Ca from the cell de- amplitude of the Ca transient and thence contractility. One well- creases SR Ca. These events continue until Ca influx and ef- known example is the effect of changing frequency or pausing flux are equal. A good example of this mechanism in operation stimulation. If stimulation is stopped in ventricular muscle from is provided by Figure 2B, which shows what happens when most nonrodent species, Ca leaks out of the SR,47 SR content the SR has been emptied by exposure to 10 mmol/L caffeine. decreases, and therefore the first stimulus results in a small Ca When stimulation is recommenced, the Ca transient is small transient and contraction.48 Because the Ca transient is small, because of the low SR Ca content. Consequently, Ca influx is less Ca is pumped out of the cell than enters and the cell is not much larger than efflux and the SR Ca content increases. This in Ca flux balance. This results in an increase of SR Ca content leads to an increase in the amplitude of the Ca transient until until the Ca transient increases sufficiently that the Ca efflux influx and efflux return to balance. now balances influx and the cell is back in a steady state. In Work on Ca cycling often takes insufficient notice of the the steady state, however, influx and efflux must be equal.49,50 flux balance condition. As discussed below, it is essential that The need for Ca flux balance applies not only to the surface postulated mechanisms and explanations are tested to ensure membrane but also to organelles such as SR and mitochondria that they are compatible with the requirement for Ca efflux to (see below). equal influx such that steady state conditions can prevail.

Figure 2. Mechanisms producing calcium flux balance and controlling sarcoplasmic reticulum (SR) Ca content. A, Flow diagram. This illustrates recovery from a situation where influx is greater than efflux. Boxes show (fromtop to bottom): increase of cell Ca content leading to an increase of SR Ca; increase of the amplitude of the Ca transient (red). The bottom box shows membrane current records in response to a depolarization. The red traces show that increase in size of the Ca transient leads to faster inactivation of the L-type Ca current during the pulse and a larger sodium–calcium exchange current on repolarization (arrowed). B, Illustrative traces. These 2+ show (from top to bottom) [Ca ]i; sarcolemmal fluxes; calculated cell (and SR) Ca gain. At the start of the record, 10 mmol/L caffeine was applied to empty the SR. After removing caffeine, stimulation was commenced. Note that the recovery of the amplitude of the Ca transient is accompanied by a decrease of Ca influx and increase of efflux. Reprinted from Trafford et al84 with permission of the publisher. Copyright ©2001, American Heart Association, Inc. Eisner et al Ca and Cardiac E–C Coupling 185

Examples of the Effects of Flux Balance by one of time course such that efflux is unaffected (see on Ca Handling below for consideration of mitochondrial function). A striking example is provided by considering the effects Effects of Altering Sarcolemmal Ca Fluxes of changing the open probability of the RyR. Adding sub- 2+ A change of Ca entry must be balanced by a change of [Ca ]i. millimolar concentrations of caffeine potentiates the opening Starting off from a steady state, an increase in the L-type Ca 2+ of the RyR (without affecting Ca entry via ICa), increasing current will mean that influx is greater than efflux. This will the amplitude of the systolic Ca transient. After a few beats, increase the amount of Ca in the cell and SR until the resulting however, the amplitude of the Ca transient in caffeine is iden- increase of the amplitude of the Ca transient increases efflux tical to that in control56,57 (Figure 3A). The explanation of this to a level that restores flux balance. What magnitude increase result is that potentiation of RyR opening initially increas- of the Ca transient is required to bring the cell back into flux es the amplitude of the Ca transient making efflux greater balance? At first sight, this appears to be an intractable prob- than influx so the cell is no longer in a steady state. The SR lem as at least 2 factors have to be considered. (1) An increase therefore loses Ca, decreasing the amplitude of the Ca tran- of Ca entry will increase the number of RyRs that open, and sient until a new steady state (influx=efflux) is reached, with the size of this effect will depend on the relationship between a decreased SR Ca content offsetting the potentiation of the dyadic [Ca2+] and RyR opening. (2) Increased Ca entry might RyR produced by caffeine. This occurs when the amplitude be expected to increase SR Ca content (but see below), and of the Ca transient returns to the control (pre-caffeine) level therefore the amount of Ca released from the SR through (Figure 3B). The underlying decrease of SR Ca, responsible each RyR that opens. The analysis is, however, simplified by for the decline of the Ca transient amplitude to the control the requirement for flux balance. Specifically, the increase of level, has been measured directly using a fluorescent indica- Ca entry must be balanced by an equal increase of Ca efflux. tor in the SR58 (Figure 3A). 2+ Assuming that diastolic [Ca ]i does not change, the increased These arguments were originally made with respect to the efflux will be provided by an increase of the amplitude of the effects of low concentrations of caffeine, but similar effects are Ca transient. This is irrespective of the underlying mecha- seen when the RyR is potentiated with BDM (2,3-butanedione nisms. If we make the reasonable assumption that the rate of monoxime).59 Likewise, decreasing RyR opening with tetracaine60 2+ 53 61 NCX is proportional to [Ca ]i, then the amplitude of the Ca or decreased pH produces a transient decrease of contraction or transient must increase by the same proportion as the Ca entry. the Ca transient. This analysis can be generalized to other mecha- Likewise, slowing sarcolemmal extrusion by NCX increases nisms that alter RyR opening. For example, phosphorylation of the the amplitude of the Ca transient to a level that restores the RyR increases its open probability and this has been suggested to efflux to balance the influx.55 contribute to the positive inotropic effects of β-adrenergic stimulation.62 This conclusion has been criticized on other grounds,63 but, Effects of Altering Intracellular Mechanisms in the context of flux balance, seems implausible as any increase What happens if Ca transporters across intracellular mem- of the Ca transient would make efflux exceed influx, resulting in branes such as SR or mitochondria are affected? The simple the Ca transient returning to the control level. As discussed else- answer is that if the Ca influx into the cell is unchanged, then where,64 the positive inotropic effects of β-adrenergic stimulation the Ca efflux must be unaffected. This either means that the can be explained by the well-established effects to (1) increase amplitude and kinetics of the systolic Ca transient are unaf- Ca influx via the L-type current65 and (2) increase SR content by fected or that a change of amplitude is exactly compensated phosphorylating PLN and thereby stimulating SERCA.66

Figure 3. Effects of potentiating ryanodine receptor (RyR) opening. A, Records show measurements of (top) cytoplasmic and (bottom) sarcoplasmic reticulum (SR) [Ca2+]. Caffeine (0.5 mmol/L) was added for the period shown. Reprinted from Greensmith et al58 with permission. Copyright ©2014, The Authors. Published by Oxford University Press on behalf of the European Society of Cardiology. B, Flow diagram of events underlying changes in A. (i) The increase of RyR opening increases the amplitude of the Ca transient (ii) leading to increased Ca efflux (iii) and a decrease of SR Ca content (iv) which returns the amplitude of the Ca transient to control levels (v). Reprinted from Eisner183 with permission. Copyright ©2014, The Author. Published by the Physiological Society. 186 Circulation Research July 7, 2017

2+ What are the effects of altering SERCA activity? An in- [Ca ]i. In some cases, this is genetic. Familial hypertrophic car- crease (in the absence of effects on sarcolemmal fluxes) will in- diomyopathy is caused by mutations in myofilament proteins increase SR Ca thereby increasing the amount of Ca released and cluding thin filament proteins such as troponin and tropomyosin. thence the Ca efflux from the cell. This effect alone would make Many of these mutations result in an increase of Ca binding to efflux greater than influx, thus violating flux balance, but it will, troponin75 and thence increased Ca buffering.76 This increased 2+ however, be offset by the fact that increased SERCA activity buffering was correlated with an increase of both diastolic [Ca ]i increases the rate of decay of the Ca transient thereby allowing and the probability of triggered Ca waves.76 Ca buffering can less time for Ca extrusion. In the steady state, the increase in the also be modified acutely, by phosphorylation. As regards the 2 amplitude of the Ca transient must exactly balance the accelera- major buffers, phosphorylation of troponin decreases its affin- tion of its decay so that efflux is unaltered. Put another way, the ity for binding Ca whereas phosphorylation of PLN increases fraction by which the amplitude of the Ca transient increases is the affinity of SERCA for Ca and presumably its buffering. The determined by the acceleration of SERCA. expected changes of buffering produced by β-adrenergic stimu- This section has given some examples of the importance lation have been demonstrated experimentally. In cells from wild- of flux balance considerations to understanding changes of type mice, β-adrenergic stimulation has no effect on buffering contractility. The consequences of flux balance will also be power as the decrease in buffering by troponin is compensated referred to in subsequent sections. for by the increase because of SERCA.77 The individual effects could, however, be revealed in cells from animals in which either Ca Buffering the regulation of troponin or SERCA by β-adrenergic stimula- 2+ 77 The increase of free Ca ([Ca ]i) during systole is of the order of 1 tion was prevented. A final question on buffering is whether it µmol/L. Ca is, however, strongly buffered; for every free Ca2+ ion, is affected by disease. No changes in buffering were observed in ≈100 to 200 are bound to buffers,67,68 meaning that the total in- ventricular myocytes when heart failure was induced.78 In atrial crease of cytoplasmic Ca is of the order of 100 to 200 µmol/L. The fibrillation, however, an increase of buffering power has been major Ca buffers are troponin and SERCA.3 Buffering by SERCA suggested to decrease propagation of the Ca transient into the requires some discussion as it is generally thought of only as a Ca interior of the atrial myocyte.79 Given the extensive changes in pump. Although this is, indeed, the case, the initial binding of Ca SERCA expression found in many models of heart failure,80,81 by SERCA contributes significantly to buffering.69 This buffering it is perhaps surprising that changes of buffering have not been increases the fluxes of Ca required to produce a given change of reported more generally and this area would warrant study. 2+ [Ca ]i. Similarly, buffering decreases the rate constant of decay of cytoplasmic Ca. The effects of Ca buffers will also depend on the Regulation of SR Ca Content kinetics with which they bind and unbind Ca. A very fast buffer Direct measurements of intra-SR free Ca concentration provide will decrease the amplitude of the Ca transient and slow its de- values of 1 to 1.5 mmol/L at the end of diastole with the con- 2+ 82 cay, as a given Ca flux produces a smaller change of [Ca ]i. This centration decreasing by 50% to 75% during contraction. As 2+ will result in decreased systolic and increased diastolic [Ca ]i, an mentioned above, a major factor controlling the amount of Ca re- effect that can be a problem experimentally when excessive con- leased from the SR and thereby the amplitude of the Ca transient centrations of Ca indicators (buffers) are used.70 If, however, the is the SR Ca content. It is, therefore, important to understand the buffer binds Ca more slowly, the initial amplitude of the Ca tran- regulation of SR Ca content. In brief, SR content is determined sient will be large but will decay with a rate given by the binding by the balance between uptake of Ca into the SR (by SERCA) of Ca to the buffer.71 In skeletal muscle, parvalbumin has similar and efflux (through the RyR). In turn, these fluxes depend, not effects. Binding of Ca2+ is slow because Mg2+ must first dissoci- only on the properties of SERCA and RyR but also on the Ca 2+ 72 ate thereby allowing [Ca ]i to rise (for review see ) resulting in concentration in the cytoplasm and SR. The feedback mecha- a buffer that is most active during diastole. It has been suggested nism shown in Figure 2 to explain cellular Ca flux balance also that incorporating such a buffer into the heart would, therefore, serves to explain regulation of SR Ca content (see83 for review). 2+ preserve the systolic rise of [Ca ]i (when buffering is weak) but SR content will change until the Ca transient is the exact ampli- 2+ lower diastolic [Ca ]i (when it is strong). In support of this, the tude required to produce a Ca efflux that balances the influx. If incorporation of parvalbumin protects against increased diastolic the SR content is below this value, then efflux will be less than 2+ 73 [Ca ]i in the Dahl salt-sensitive rat. Finally, it may well be that influx and the cell (and SR) Ca content will increase. The steady buffering is not uniform throughout the cytoplasm. For example, state value of SR Ca content reached will be altered by changing the dyadic space contains no myofibrils, and therefore troponin the expression or properties of any of the Ca-handling proteins. will not contribute to the buffering here. As reviewed above, increasing SERCA activity or decreasing RyR opening will increase SR content and increasing NCX will Factors That Affect Ca Buffering decrease it. The most complicated factor is the L-type Ca cur- 2+ Ca buffering depends on [Ca ]i; buffer power is greatest at low rent. At first sight, one might think that an increase of the L-type 2+ levels of [Ca ]i and decreases at higher levels as the buffers tend Ca current will load the cell with calcium and thence increase to saturate. This may play a role in determining the rate of relax- SR content. Experimentally, however, even a large increase of ation of the systolic Ca transient. For example, the reduced buff- the L-type current has little effect on SR content, and a decrease 2+ 84 ering at the highest [Ca ]i will mean that a given rate of pumping increases content. Indeed, in a sheep model of heart failure, a 2+ by SERCA will reduce [Ca ]i more quickly at the start of the decrease of L-type current was suggested to cause the observed 2+ 74 85 decay of the Ca transient resulting in a biphasic decay of [Ca ]i. increase of atrial SR content. This is because the L-type current However, other factors affect buffer power, even at a constant plays 2 roles in Ca cycling. The peak amplitude determines the Eisner et al Ca and Cardiac E–C Coupling 187 triggering of Ca release from the SR so an increase of current and does not activate other release sites. In other words Ca will decrease SR content, whereas the rest of the current loads release is controlled locally. Under resting conditions, local- the cell and SR with Ca.86 The net effect on SR Ca content of ized releases of Ca from individual clusters of RyRs are seen a change of L-type current will depend on the relative strength as Ca sparks.98 Depolarization of the surface membrane acti- of the 2 opposing effects on the SR. Under basal conditions, it vates more and more L-type Ca channels resulting in an in- appears that these are matched so that there is little effect on creasing number of sparks until spatially uniform Ca release SR content.84 This may be physiologically useful as it means is observed.99 that an increase of L-type Ca current will produce an immediate As mentioned above, Ca release from the SR is a steep func- increase in the amplitude of the Ca transient without the delay tion of SR Ca content.51,52 This steep dependence is functionally produced by the need to increase SR content.84,87 important, not only does it contribute to regulation of flux balance but also it provides a mechanism to regulate contractility. Need for Adequate Measurement of SR Ca The steep dependence is a consequence of several factors in- Content cluding the fact that an increase of SR Ca content (1) increases Many experimental studies involve addressing whether a the driving force for Ca release through open RyRs and (2) in- change in the amplitude of the systolic Ca transient results creases the number of open RyRs. As noted below, the latter ef- from one of SR Ca content. Undoubtedly, the best way to mea- fect may not be directly because of SR Ca but, rather, secondary sure SR Ca is to use a Ca-sensitive indicator in the SR.82,88 to Ca release from the SR. Too steep a dependence of Ca release Such indicators, however, do not seem to work in all tissues.58 on SR content may result in instability of Ca release resulting There is also a problem with saturation of the indicator at the in such phenomena as alternans.100,101 Finally, when SR Ca con- high [Ca2+] in the SR. A simple way to measure SR content is tent exceeds a certain threshold level, the local regulation of Ca to release all the Ca into the cytoplasm by applying 10 mmol/L release breaks down, propagating Ca waves are observed102 (see caffeine and measuring the amplitude of the resulting increase Calcium Leak section of this article). 2+ 2+ of [Ca ]i. The problem here is that the level of [Ca ]i at the Although the discovery of the spark resolved the question peak of the caffeine response is close to those that saturate of how Ca release could be graded rather than all or none, it commonly used Ca indicators thereby reducing the sensitivity raised another difficulty; how does release terminate so that of the measurement, an issue that is exacerbated by the steep- the SR can refill with Ca? The problem is that Ca released ness of the dependence of Ca transient amplitude on SR Ca. from the SR would be expected to continue to activate RyRs We suggest that, before rejecting the hypothesis that a change in the same cluster thereby maintaining Ca release. Various of Ca transient amplitude results from one of SR Ca content, it explanations have been considered. (1) The release process is essential to see how large a change of content would be re- may inactivate, even in the presence of constant activating quired to explain the effect and whether the measurement has Ca.103,104 A related phenomenon, known as adaptation (where sufficient sensitivity to detect it. Problems of saturation of the the RyR can still be opened but requires a larger stimulus), indicator can be mitigated by using a lower affinity calcium has also been identified.105,106 The rate of this inactivation/ indicator.89 Alternatively, if the experiment can be performed adaptation may, however, be too slow to be the only factor under voltage clamp, then a more accurate estimate of SR Ca involved in terminating Ca release (see107 for review). (2) An content can be obtained by measuring the integral of the NCX alternative, termed stochastic attrition, depends on there being 2+ 90 current activated by the caffeine-evoked increase of [Ca ]i. a probability that all the RyRs in a cluster close by chance such that the Ca outside the SR will fall to levels too low to Calcium Release From the SR open RyRs. Although this would work well if there were only Before considering mechanisms that may control Ca release a small number (up to ≈15) of RyRs in a cluster, it is less from the SR, it is important to remember that the RyR does plausible given experimental data showing that there can be not sit in isolation in the SR membrane but, rather, forms a >100 RyRs per cluster.10,30,31 (3) Another explanation depends complex with triadin, junctin, and CSQ (calsequestrin).91 CSQ on changes of lumenal Ca. The fact that, even at the end of the is the major Ca buffer in the SR but has been suggested to have release, the SR still contains 25% to 50% of its Ca content82 other effects because it, in addition to triadin and junctin, is is inconsistent with the idea that a simple effect on driving required to make RyR open probability respond to luminal Ca, force accounts for the turn off of release. The decrease of SR at least in bilayer studies.92 Ca will also decrease the frequency of RyR opening,108,109 but The phenomenon of calcium-induced calcium release has this effect, alone, is probably not strong enough to terminate been appreciated for ≈50 years.93 A major concern for much of release. A variety of modifications has been made to try to this time was the issue of how it was regulated. As originally account for termination. One, the sticky cluster model110 is described, calcium-induced calcium release is a positive feed- based on the observation that the opening of 2 RyRs can be back system in which one might expect the Ca released from coupled such that they open and close together.111 This makes the SR to trigger further release of Ca until the SR is empty. it easier for stochastic attrition to occur and together with SR This contrasts with the observation that Ca release is graded depletion could account for spark termination. The difficulty with the amplitude of the L-type Ca current94,95 and, indeed, is that coupled gating is not observed in most bilayer studies. the SR only releases ≈50% of its Ca during the Ca transient.96 More recent studies have suggested that luminal Ca may still The resolution of this paradox came from both modeling97 and be the controlling factor but via an indirect mechanism involv- experimental work showing that, under normal conditions, ing effects on cytoplasmic activation of the RyR. Evidence Ca release from one release site of the SR remains localized that luminal concentration per se is not the important factor is 188 Circulation Research July 7, 2017 supported by the fact that large organic cations that decrease tetracaine, indicating that it is because of the unsynchronized Ca flux through the RyR increase SR content while decreasing opening of individual RyRs. Modeling suggests that stochas- spark frequency.112 This is consistent with a model in which tic considerations determine which RyR opening result in a Ca 2+ 123 initially 1 RyR opens; the resulting increase of [Ca ]i leads spark. Finally, SR Ca continued to decrease, even when the to or induces the opening of adjacent RyRs and thence a Ca RyRs were inhibited indicating an additional mechanism for spark. The release of Ca decreases luminal [Ca2+] in the junc- leak efflux from the SR. 2+ tional SR, decreasing release to a level where [Ca ]i is suf- In the experiments described above, changes of Ca leak ficiently low that all RyRs close. This model has been called resulted from those of SR Ca content. However, Ca leak is induction decay113,114 and pernicious attrition115 (see also32). In also a function of the properties of the RyR itself and associ- these models, the start and end of Ca release depend on (1) the ated proteins. SR Ca leak is elevated by single amino acid sensitivity of the single-channel current to luminal Ca and (2) mutations such as those occurring in catecholaminergic the activation by cytoplasmic Ca. Consistent with this model, polymorphic ventricular tachycardia in either the RyR124 or a recent study has demonstrated that increasing cytoplasmic CSQ.125 Leak is also increased in heart failure as was origi- Ca buffering (thereby impeding the activation of neighboring nally shown by measuring the open probability of RyRs in- RyRs) makes sparks terminate at an elevated SR Ca content.116 corporated into bilayers. Those from dogs with heart failure A phenomenon that is related to that of termination of re- had a higher open probability than from control animals, an lease is that after 1 stimulated release, there is a refractory effect that was suggested to result from excessive phosphory- period before another full release can occur. Early attempts lation of the RyR leading to the dissociation of the regulatory to investigate this experimentally suffered from the fact that protein FKBP12.6126. Although there is general agreement of the triggering L-type Ca current itself requires time to recover increased SR Ca leak in heart failure, the precise mechanism from inactivation.117 When this issue was overcome using pho- remains controversial. The concentration of FKBP12.6 in the tolysis of caged Ca to trigger Ca release, it was found that ventricular myocyte is too low to bind to more than a small mi- Ca release recovered with a time constant of ≈300 ms.118 This nority of RyRs and phosphorylation by PKA (protein kinase slow recovery was absent if only a small region of SR was A) has no effect on binding.127 In contrast, there is substantial stimulated (using 2 photon photolysis) leading to the conclu- evidence for a major role for Ca/calmodulin-dependent pro- sion that the refractoriness was because of depletion of SR tein kinase II—dependent phosphorylation128,129 as well as for Ca and presumably would involve the mechanisms described oxidation either directly affecting the RyR130 or indirectly via above. Further support for a role for Ca depletion was pro- Ca/calmodulin-dependent protein kinase II of the RyR.131 For vided by the observation that incorporation of Ca buffers into a recent review of this area, see the study by Bers.132 the SR decreased apparent refractoriness.119 Simultaneous measurements of SR and cytoplasmic Ca suggest, however, Consequences of Ca Leak that refractoriness may result from something in addition to A major effect of Ca leak is to decrease the Ca content of SR Ca. After the first stimulus, SR Ca content (as measured the SR and thence the amplitude of the Ca transient. In this with an intra-SR indicator) recovered fully before maximal context, an important issue concerns the properties of the Ca release could be obtained.120 This dissociation is partly ex- leak. Evidence from the Gyorke group using bilayer studies plained by the fact that the indicator is tending to saturation, has found that in heart failure, there is an apparent sensitiza- but this cannot explain everything. Another explanation might tion of the RyR to activation by luminal Ca.133 This contrasts be that the SR Ca concentration at the release sites recovers with a previous study, suggesting that heart failure locked the more slowly than that in the bulk SR. Finally, it is possible that RyR in a subconducting state.126 The difference is significant there is a genuine Ca-independent refractoriness of the RyR. for the effects of leak on the systolic Ca transient. If the leak results from a mechanism that sensitizes the RyR, then (as for Calcium Leak the effects of low concentrations of caffeine mentioned above) The emphasis above has been on the release of Ca from the sensitization of the RyR will initially compensate for the the RyR in response to triggering by the L-type Ca current. decrease of SR Ca content and no effect will be seen on the However, given that the RyR has a finite open probability even amplitude of the Ca transient. As the leak increases, the SR Ca 2+ at diastolic [Ca ]i, Ca will leak out of the SR. Early evidence content falls to such a low level that even the release of 100% for a leak came from the phenomenon of rest decay where, cannot sustain a normal-sized Ca transient and the amplitude after a pause in stimulation, the first contraction is smaller of the Ca transient declines134 and efflux is maintained by pro- than the steady state and this is associated with a decrease in longation of decay. In contrast, if the leak does not result from total cell121 and SR47 Ca. Some, but not all, of this leak occurs sensitization of the RyR, then the amplitude of the Ca tran- via Ca sparks. Work on rabbit ventricular myocytes that had sient will decline in parallel with SR content. An analogy is been skinned (ie, the surface membrane was removed) found provided by comparing the effects of caffeine (sensitizing-) that inhibiting SERCA with thapsigargin decreased both SR with those of ryanodine (nonsensitizing-) leak. In ryanodine, Ca content and the frequency of Ca sparks.122 A point was SR Ca and Ca transient amplitude decay together, whereas, in then reached when, although SR Ca content continued to de- caffeine, the amplitude of the Ca transient is preserved at low crease, no sparks were observed. The decrease of Ca spark levels of leak.135,136 frequency as SR Ca falls is to be expected from the effect of Increased Ca leak is arrhythmogenic as a result of the oc- SR Ca on RyR opening.108 Most of the spark-independent de- currence of intracellular Ca waves that occur when the Ca crease of SR Ca was inhibited by blockers of the RyR such as spark frequency and flux rises so that Ca spreads beyond the Eisner et al Ca and Cardiac E–C Coupling 189 original site and activates others.137 The waves that activate noted that the mitochondrial Ca transients decayed with a time NCX,138 giving rise to arrhythmogenic delayed afterdepolar- constant of ≈5 seconds. This slow rate is probably a function izations139,140 and resulting arrhythmias, were originally de- of the low level of Ca efflux through the mitochondrial Na–Ca scribed for situations where the SR Ca content was elevated exchange. When stimulation rate was increased from 0.1 to to above a threshold level102 (sometimes referred to as store 0.5 Hz, the beat-to-beat mitochondrial Ca transients disap- overload–induced Ca release141) but also occur when the RyRs peared and were replaced by a virtually tonic increase of mito- are modified as in catecholaminergic polymorphic ventricu- chondrial Ca concentration. Even allowing for the fact that the lar tachycardia and heart failure (see142,143 for reviews). The experiments were performed at room temperature and there- occurrence of these waves also relates to Ca flux balance as fore the mitochondrial Ca transients may decay more quickly they activate a component of Ca efflux from the cell in ad- at 37°C, this result makes it less likely that mitochondrial dition to that produced by the systolic Ca transient. For ex- Ca transients are of physiological importance. This study151 ample, when a cell that has an increased Ca load (eg, because also investigated the question as to whether any flux of Ca of β-adrenergic stimulation) is treated with caffeine, Ca waves into the mitochondria affects the cytoplasmic Ca transient. develop. To maintain flux balance, there is a compensatory This was done by comparing the cytoplasmic Ca transient in decrease of the amplitude of the systolic Ca transient.144 Flux control cells with those incubated with the MCU inhibitor, balance considerations also determine whether making the Ru360. The amplitude of the cytoplasmic Ca transient was RyR leaky, either in a natural disease such as catecholamin- identical in both groups, suggesting that the total flux of Ca ergic polymorphic ventricular tachycardia or experimentally into the mitochondria is small when compared with that re- with caffeine, results in Ca waves. If the Ca influx into the cell leased from the SR and therefore makes little contribution to on each beat is below a certain level, then making the RyR relaxation.151 This is consistent with a previous study, show- leaky will not produce waves. Only if there is sufficient influx ing that mitochondria have little effect on Ca removal from to balance the extra (wave-associated) Ca efflux, will waves the cytoplasm.154,155 Work on neonatal myocytes also found result.64,144,145 mitochondrial Ca transients that were abolished by knocking Finally, Ca leak occurring during the decay of the Ca tran- down the MCU with siRNA.156 However, knockdown of MCU sient will slow its decay as it adds an additional flux to com- increased the amplitude of the cytoplasmic Ca transient by pete with SERCA.146 This is observed with concentrations of ≈50% to 60% leading to the conclusion that there is a signifi- caffeine above ≈1 mmol/L.135,147 The increased leak can also cant beat-to-beat flux of Ca into the mitochondria that buffers result in a biphasic decay of the Ca transient, an effect attrib- the cytoplasmic Ca transient. As well as being at odds with uted to the flux through the open RyRs being low at the start of the work on adult cells, this result is difficult to interpret in the decay of Ca transient and therefore not competing greatly the context of flux balance considerations. Simply increasing with SERCA. As the SR refills, then the leak efflux increases the amplitude of the Ca transient (as occurs when MCU is and the rate constant of decay slows.136 knocked down) would be expected to increase the efflux of Ca from the cell thereby making it greater than the influx, a Mitochondria and Calcium situation that cannot persist in the steady state. If Ca is taken The mitochondrial inner membrane contains a calcium chan- up into the mitochondria, then it must be released between 2+ nel, the mitochondrial Ca uniporter (MCU), identified a few beats, an effect that should slow the rate of decay of [Ca ]i 148,149 150 2+ years ago (see for recent review). Ca entry into the and elevate diastolic [Ca ]i. Knockdown of the MCU might mitochondrion is driven largely by the inside-mitochondria therefore be expected to accelerate the decay of the Ca tran- negative membrane potential. Mitochondria are often located sient.156 This would compensate for the increased amplitude adjacent to the junctional SR, and it is therefore been suggest- and thereby restore Ca efflux to control levels, achieving flux ed that Ca release will elevate local Ca to high levels resulting balance. No effect on the rate constant of decay was, however, in a large influx.151 Such an influx is important for mitochon- reported, thus leaving open the question as to how the result drial function as many of the mitochondrial enzymes are ac- can be squared with flux balance. Finally, for completeness, it tivated by a rise of matrix calcium concentration leading to should be noted that, although the MCU is required for rapid increased supply of ATP when demand, because of increased regulation of mitochondrial Ca,157 mitochondrial Ca can still 2+ 158 contraction and [Ca ]i, is increased. As far as understanding change, albeit more slowly when this has been deleted, sug- E–C coupling is concerned, it is important to know whether gesting the existence of another mechanism for Ca to enter the on each beat a significant amount of the Ca released from the mitochondria. SR enters the mitochondria. Flux balance conditions require that, in the steady state, if Ca enters the mitochondria on 1 Control of Diastolic Ca beat, exactly the same amount must leave before the next and As mentioned above, the ability of the heart to pump blood 2+ 2+ therefore a transient change of mitochondrial [Ca ] would be depends as much on a low diastolic [Ca ]i as on the systolic expected. Although some early studies reported beat-to-beat elevation. Indeed diastolic heart failure is a major cause of changes of mitochondrial Ca, others did not (see152,153 for re- morbidity with many patients having no apparent impairment views). Earlier studies, particularly in adult myocytes, suf- of systolic function (heart failure with preserved ejection frac- fered from problems of specifically measuring mitochondrial tion).159 Diastolic heart failure may well involve many factors 2+ Ca. A more recent study, using a mitochondrially targeted Ca other than [Ca ]i regulation, with fibrosis being a significant sensor, found beat-to-beat Ca transients that were larger at re- factor.160 That said, it is clearly important to understand the 151 2+ gions of mitochondria near to the SR . It should, however, be regulation of diastolic [Ca ]i. As will become apparent below, 190 Circulation Research July 7, 2017 understanding of the control of diastolic Ca lags behind that limitation of the analysis was the lack of a specific inhibitor of systolic Ca. with only gadolinium having a marked effect. Work using the HL-1 cell line found a store-operated Ca Resting Ca entry (SOCE). Interestingly, inhibiting this entry mechanism 2+ 169 The heart beats continuously. Nevertheless, as a first step in un- also decreased the resting level of [Ca ]i, suggesting that derstanding the mechanisms involved in the control of diastolic SOCE may contribute to resting Ca in unstimulated cells. Such Ca, a considerable amount of work has studied quiescent car- SOCE has been identified in many cell types. Briefly, a decrease diac preparations. In the absence of stimulation, in the steady of endoplasmic reticulum Ca results in the opening of surface state, there must be no net flux across the membranes of the SR membrane channels leading to a refilling of the endoplasmic 2+ 170 and other organelles and the level of resting [Ca ]i is therefore reticulum with Ca (see for review). The mechanism of this controlled entirely by the surface membrane.161,162 Early studies involves an endoplasmic reticulum Ca sensor (STIM1 [stromal 2+ showed that resting [Ca ]i, either measured directly or using interaction molecule 1]) which, when endoplasmic reticulum resting force as a surrogate, was very sensitive to the sodium Ca is decreased, interacts with the surface membrane channel gradient with NCX being the main mechanism responsible Orai.171 This mechanism is best characterized in nonexcitable for pumping Ca out of the cell.163,164 The surface membrane cells, where it may be the major route for Ca entry into the also contains a plasma membrane Ca-ATPase (PMCA), which cell, but is much less well characterized in cardiac myocytes. should also contribute to Ca efflux. In rat ventricular myocytes, Early evidence for SOCE in the heart was obtained in neonatal PMCA has been estimated to make a contribution equal to be- cells,172 and the mechanism was reported not to exist in adult tween 7%165 and 25%166 of that produced by NCX. It is unclear cells.173 Although some subsequent studies have revealed SOCE what is responsible for the 3-fold range of these estimates. in adult myocytes,174,175 a recent study could not find it.176 The 2+ Given that the level of resting [Ca ]i represents the balance general consensus is that the mechanism is much more evident between Ca influx and extrusion, it is important to identify the in the developing heart (see177 for review). It may also be impor- Ca entry mechanism in a quiescent cell. In rat ventricular myo- tant for the development of cardiac hypertrophy178,179 with over- cytes, we estimated a background Ca entry of the order of ≈2 expression of STIM1 leading to cardiomyopathy.180 A further to 5 µmol/L per second.167 This compares with that of the order complication in this field is the report that STIM1 increases SR of 4 µmol/L for the entry through the L-type current,84 which at Ca by activating SERCA secondary to interacting with PLN.176 a typical heart rate of 6 Hz corresponds to 24 µmol/L per second, a considerably larger value. The identity of this Ca entry Diastolic Ca 2+ is unclear. At a normal resting potential, the open probability of When the heart beats, the level of diastolic [Ca ]i is deter- the L-type channel is very low and therefore unlikely to make mined by a combination of sarcolemmal and SR fluxes. The a major contribution. When both the L-type Ca channel and beating cell provides an interesting illustration of the effects NCX were inhibited, this background entry mechanism was re- of flux balance. In much previous work,57 the low rate of stim- 2+ 2+ vealed by the decrease of [Ca ]i on maintained depolarization, ulation results in diastolic [Ca ]i being constant at the resting an effect attributed to a decreased driving force.168 A major level and flux balance is determined by systolic fluxes. When

2+ Figure 4. Effects of stimulation rate on diastolic and systolic [Ca ]i. A, Original traces showing the effects of stimulation at the rates 2+ indicated. B, Diagrammatic representation of systolic efflux. At 1 Hz left( ) efflux will be activated by the systolic rise of [Ca ]i (dotted area). At 4 Hz (right), systolic efflux has decreased (dotted) whereas diastolic has increased (diagonal lines). Reprinted from Dibb et al181 with permission. Copyright ©2007, The Authors. Published by the Physiological Society. Eisner et al Ca and Cardiac E–C Coupling 191 the cell is stimulated rapidly, then both systolic and diastolic T-tubules-a comparison to Ca2+ release in human failing cardiomyocytes. levels of [Ca2+] .are important. A good example is provided Cardiovasc Res. 2004;62:63–73. doi: 10.1016/j.cardiores.2003.12.031. i 14. Brette F, Komukai K, Orchard CH. Validation of formamide as a detubula- by the effects of increasing stimulation rate. Figure 4 shows tion agent in isolated rat cardiac cells. Am J Physiol Heart Circ Physiol. that increasing rate decreases the amplitude of the systolic Ca 2002;283:H1720–H1728. doi: 10.1152/ajpheart.00347.2002. transient.181 This is, in part, because of the decrease of the am- 15. Mackenzie L, Bootman MD, Berridge MJ, Lipp P. Predetermined recruit- 182 ment of calcium release sites underlies excitation-contraction coupling in plitude of the L-type Ca current. This is accompanied by rat atrial myocytes. J Physiol. 2001;530:417–429. 2+ an increase of diastolic [Ca ]i because, at higher rates, the 16. Lipp P, Hüser J, Pott L, Niggli E. Spatially non-uniform Ca2+ signals in- previous Ca transient has not had time to decay before the next duced by the reduction of transverse tubules in citrate-loaded guinea-pig stimulus. However, the increased level of diastolic [Ca2+] will ventricular myocytes in culture. J Physiol. 1996;497(pt 3):589–597. i 17. Lenaerts I, Bito V, Heinzel FR, Driesen RB, Holemans P, D’hooge J, also increase efflux of Ca from the cell. This will compensate Heidbüchel H, Sipido KR, Willems R. Ultrastructural and functional for the fact that the shorter and smaller Ca transient produces remodeling of the coupling between Ca2+ influx and sarcoplasmic re- less efflux from the cell during systole. ticulum Ca2+ release in right atrial myocytes from experimental persistent atrial fibrillation. Circ Res. 2009;105:876–885. doi: 10.1161/ CIRCRESAHA.109.206276. Conclusions 18. Dibb KM, Clarke JD, Horn MA, Richards MA, Graham HK, Eisner DA, The work reviewed in this article illustrates the enormous Trafford AW. Characterization of an extensive transverse tubular network progress that has been made in understanding calcium signal- in sheep atrial myocytes and its depletion in heart failure. Circ Heart Fail. 2009;2:482–489. doi: 10.1161/CIRCHEARTFAILURE.109.852228. ing in the heart. The next few years should see further rapid 19. Richards MA, Clarke JD, Saravanan P, Voigt N, Dobrev D, Eisner DA, advances, helped to no small extent by technological advances Trafford AW, Dibb KM. Transverse tubules are a common feature in large in areas such as imaging. A major area that should develop mammalian atrial myocytes including human. Am J Physiol Heart Circ greatly is that of the control of diastolic Ca. Physiol. 2011;301:H1996–H2005. doi: 10.1152/ajpheart.00284.2011. 20. Weber CR, Piacentino V 3rd, Ginsburg KS, Houser SR, Bers DM. Na(+)- Ca(2+) exchange current and submembrane [Ca(2+)] during the cardiac Sources of Funding action potential. Circ Res. 2002;90:182–189. This work was supported by the British Heart Foundation (grant number: 21. Cook SJ, Chamunorwa JP, Lancaster MK, O’Neill SC. Regional differ- CH/2000004/12801). ences in the regulation of intracellular sodium and in action potential configuration in rabbit left ventricle. Pflugers Arch. 1997;433:515–522. doi: 10.1007/s004240050307. Disclosures 22. Leblanc N, Hume JR. Sodium current-induced release of calcium from None. cardiac sarcoplasmic reticulum. Science. 1990;248:372–376. 23. Brette F, Orchard CH. Density and sub-cellular distribution of cardiac and neuronal sodium channel isoforms in rat ventricular myocytes. References Biochem Biophys Res Commun. 2006;348:1163–1166. doi: 10.1016/j. 1. Allen DG, Blinks JR. Calcium transients in aequorin-injected frog cardiac bbrc.2006.07.189. muscle. Nature. 1978;273:509–513. 24. Radwański PB, Ho HT, Veeraraghavan R, et al. Neuronal Na(+) Channels 2. Allen DG, Kurihara S. Calcium transients in mammalian ventricular mus- Are Integral Components of Pro-arrhythmic Na(+)/Ca(2+) Signaling cle. Eur Heart J. 1980;Suppl A:5–15. Nanodomain That Promotes Cardiac Arrhythmias During β-adrenergic 3. Bers DM. Excitation-Contraction Coupling and Cardiac Contractile Stimulation. JACC Basic Transl Sci. 2016;1:251–266. doi: 10.1016/j. Force. 2 ed. Dordrecht/Boston/London: Kluwer Academic Publishers; jacbts.2016.04.004. 2001. 25. Duclohier H. Neuronal sodium channels in ventricular heart cells are 4. Bers DM, Kranias EG. Normal and abnormal calcium homeostasis. In: localized near T-tubules openings. Biochem Biophys Res Commun. Walsh RA, ed. Molecular Mechanisms of Cardiac Hypertrophy and 2005;334:1135–1140. doi: 10.1016/j.bbrc.2005.06.203. Failure. Oxon: Taylor & Francis; 2005: 199–220. 26. Brette F, Orchard CH. No apparent requirement for neuronal sodium chan- 5. Robinson P, Griffiths PJ, Watkins H, Redwood CS. Dilated and hypertro- nels in excitation-contraction coupling in rat ventricular myocytes. Circ phic cardiomyopathy mutations in troponin and alpha-tropomyosin have Res. 2006;98:667–674. doi: 10.1161/01.RES.0000209963.02720.70. opposing effects on the calcium affinity of cardiac thin filaments. Circ Res. 27. He J, Conklin MW, Foell JD, Wolff MR, Haworth RA, Coronado R, 2007;101:1266–1273. doi: 10.1161/CIRCRESAHA.107.156380. Kamp TJ. Reduction in density of transverse tubules and L-type Ca(2+) 6. Solaro RJ, Moir AJ, Perry SV. Phosphorylation of troponin I and the channels in canine tachycardia-induced heart failure. Cardiovasc Res. inotropic effect of adrenaline in the perfused rabbit heart. Nature. 2001;49:298–307. 1976;262:615–617. 28. Louch WE, Mørk HK, Sexton J, Strømme TA, Laake P, Sjaastad I, 7. Lehtonen L, Põder P. The utility of levosimendan in the treatment of heart Sejersted OM. T-tubule disorganization and reduced synchrony of Ca2+ failure. Ann Med. 2007;39:2–17. doi: 10.1080/07853890601073346. release in murine cardiomyocytes following myocardial infarction. 8. Forbes MS, van Neil EE. Membrane systems of guinea pig myocardium: J Physiol. 2006;574:519–533. doi: 10.1113/jphysiol.2006.107227. ultrastructure and morphometric studies. Anat Rec. 1988;222:362–379. 29. Song LS, Sobie EA, McCulle S, Lederer WJ, Balke CW, Cheng H. doi: 10.1002/ar.1092220409. Orphaned ryanodine receptors in the failing heart. Proc Natl Acad Sci U S 9. Soeller C, Cannell MB. Examination of the transverse tubular system in A. 2006;103:4305–4310. doi: 10.1073/pnas.0509324103. living cardiac rat myocytes by 2-photon microscopy and digital image- 30. Baddeley D, Jayasinghe ID, Lam L, Rossberger S, Cannell MB, Soeller processing techniques. Circ Res. 1999;84:266–275. C. Optical single-channel resolution imaging of the ryanodine recep- 10. Hayashi T, Martone ME, Yu Z, Thor A, Doi M, Holst MJ, Ellisman MH, tor distribution in rat cardiac myocytes. Proc Natl Acad Sci U S A. Hoshijima M. Three-dimensional electron microscopy reveals new de- 2009;106:22275–22280. doi: 10.1073/pnas.0908971106. tails of membrane systems for Ca2+ signaling in the heart. J Cell Sci. 31. Hou Y, Jayasinghe I, Crossman DJ, Baddeley D, Soeller C. Nanoscale 2009;122:1005–1013. doi: 10.1242/jcs.028175. analysis of ryanodine receptor clusters in dyadic couplings of rat car- 11. Pinali C, Bennett H, Davenport JB, Trafford AW, Kitmitto A. Three- diac myocytes. J Mol Cell Cardiol. 2015;80:45–55. doi: 10.1016/j. dimensional reconstruction of cardiac sarcoplasmic reticulum reveals a yjmcc.2014.12.013. continuous network linking transverse-tubules: this organization is per- 32. Walker MA, Williams GS, Kohl T, Lehnart SE, Jafri MS, Greenstein turbed in heart failure. Circ Res. 2013;113:1219–1230. doi: 10.1161/ JL, Lederer WJ, Winslow RL. Superresolution modeling of calcium CIRCRESAHA.113.301348. release in the heart. Biophys J. 2014;107:3018–3029. doi: 10.1016/j. 12. Fawcett DW, McNutt NS. The ultrastructure of the cat myocardium. I. bpj.2014.11.003. Ventricular papillary muscle. J Cell Biol. 1969;42:1–45. 33. Walker MA, Kohl T, Lehnart SE, Greenstein JL, Lederer WJ, Winslow 13. Louch WE, Bito V, Heinzel FR, Macianskiene R, Vanhaecke J, Flameng W, RL. On the Adjacency Matrix of RyR2 Cluster Structures. PLoS Comput Mubagwa K, Sipido KR. Reduced synchrony of Ca2+ release with loss of Biol. 2015;11:e1004521. doi: 10.1371/journal.pcbi.1004521. 192 Circulation Research July 7, 2017

34. Rajagopal V, Bass G, Walker CG, Crossman DJ, Petzer A, Hickey 55. Meme W, O’Neill S, Eisner D. Low sodium inotropy is accompanied by A, Siekmann I, Hoshijima M, Ellisman MH, Crampin EJ, Soeller C. diastolic Ca2+ gain and systolic loss in isolated guinea-pig ventricular Examination of the effects of heterogeneous organization of RyR clus- myocytes. J Physiol. 2001;530:487–495. ters, myofibrils and mitochondria on Ca2+ release patterns in cardio- 56. O’Neill SC, Eisner DA. A mechanism for the effects of caffeine on Ca2+ myocytes. PLoS Comput Biol. 2015;11:e1004417. doi: 10.1371/journal. release during diastole and systole in isolated rat ventricular myocytes. pcbi.1004417. J Physiol. 1990;430:519–536. 35. Li H, Lichter JG, Seidel T, Tomaselli GF, Bridge JH, Sachse FB. Cardiac 57. Trafford AW, Díaz ME, Sibbring GC, Eisner DA. Modulation of CICR resynchronization therapy reduces subcellular heterogeneity of ryano- has no maintained effect on systolic Ca2+: simultaneous measurements dine receptors, T-tubules, and Ca2+ sparks produced by dyssynchro- of sarcoplasmic reticulum and sarcolemmal Ca2+ fluxes in rat ventricular nous heart failure. Circ Heart Fail. 2015;8:1105–1114. doi: 10.1161/ myocytes. J Physiol. 2000;522(pt 2):259–270. CIRCHEARTFAILURE.115.002352. 58. Greensmith DJ, Galli GL, Trafford AW, Eisner DA. Direct measure- 36. Scriven DR, Dan P, Moore ED. Distribution of proteins implicated in ments of SR free Ca reveal the mechanism underlying the transient ef- excitation-contraction coupling in rat ventricular myocytes. Biophys J. fects of RyR potentiation under physiological conditions. Cardiovasc Res. 2000;79:2682–2691. 2014;103:554–563. doi: 10.1093/cvr/cvu158. 37. Schulson MN, Scriven DR, Fletcher P, Moore ED. Couplons in rat atria 59. Adams W, Trafford AW, Eisner DA. 2,3-Butanedione monoxime (BDM) form distinct subgroups defined by their molecular partners. J Cell Sci. decreases sarcoplasmic reticulum Ca content by stimulating Ca release in 2011;124:1167–1174. doi: 10.1242/jcs.080929. isolated rat ventricular myocytes. Pflugers Arch. 1998;436:776–781. doi: 38. Thomas MJ, Sjaastad I, Andersen K, Helm PJ, Wasserstrom JA, Sejersted 10.1007/s004240050701. OM, Ottersen OP. Localization and function of the Na+/Ca2+-exchanger 60. Overend CL, O’Neill SC, Eisner DA. The effect of tetracaine on stimulated in normal and detubulated rat cardiomyocytes. J Mol Cell Cardiol. contractions, sarcoplasmic reticulum Ca2+ content and membrane current 2003;35:1325–1337. in isolated rat ventricular myocytes. J Physiol. 1998;507(pt 3):759–769. 39. Jayasinghe ID, Cannell MB, Soeller C. Organization of ryanodine recep- 61. Choi HS, Trafford AW, Orchard CH, Eisner DA. The effect of acidosis on tors, transverse tubules, and sodium-calcium exchanger in rat myocytes. systolic Ca2+ and sarcoplasmic reticulum calcium content in isolated rat Biophys J. 2009;97:2664–2673. doi: 10.1016/j.bpj.2009.08.036. ventricular myocytes. J Physiol. 2000;529(pt 3):661–668. 40. Drago GA, Colyer J, Lederer WJ. Immunofluorescence localization of 62. Shan J, Kushnir A, Betzenhauser MJ, Reiken S, Li J, Lehnart SE, SERCA2a and the phosphorylated forms of phospholamban in intact rat Lindegger N, Mongillo M, Mohler PJ, Marks AR. Phosphorylation of the cardiac ventricular myocytes. Ann N Y Acad Sci. 1998;853:273–279. ryanodine receptor mediates the cardiac fight or flight response in mice. 41. Greene AL, Lalli MJ, Ji Y, Babu GJ, Grupp I, Sussman M, Periasamy M. J Clin Invest. 2010;120:4388–4398. doi: 10.1172/JCI32726. Overexpression of SERCA2b in the heart leads to an increase in sarcoplas- 63. MacDonnell SM, García-Rivas G, Scherman JA, Kubo H, Chen X, Valdivia mic reticulum calcium transport function and increased cardiac contractil- H, Houser SR. Adrenergic regulation of cardiac contractility does not in- ity. J Biol Chem. 2000;275:24722–24727. doi: 10.1074/jbc.M001783200. volve phosphorylation of the cardiac ryanodine receptor at serine 2808. 42. Vangheluwe P, Louch WE, Ver Heyen M, Sipido K, Raeymaekers L, Circ Res. 2008;102:e65–e72. doi: 10.1161/CIRCRESAHA.108.174722. Wuytack F. Ca2+ transport ATPase isoforms SERCA2a and SERCA2b 64. Eisner DA, Kashimura T, O’Neill SC, Venetucci LA, Trafford AW. What are targeted to the same sites in the murine heart. Cell Calcium. role does modulation of the ryanodine receptor play in cardiac inot- 2003;34:457–464. ropy and arrhythmogenesis? J Mol Cell Cardiol. 2009;46:474–481. doi: 43. Jayasinghe I, Crossman Dj, Soeller C, Cannell M. Comparison of the or- 10.1016/j.yjmcc.2008.12.005. ganization of T-tubules, sarcoplasmic reticulum and ryanodine receptors 65. Reuter H. Localization of beta adrenergic receptors, and effects of nor- in rat and human ventricular myocardium. Clin Exp Pharmacol Physiol. adrenaline and cyclic nucleotides on action potentials, ionic currents and 2012;39:469–476. doi: 10.1111/j.1440-1681.2011.05578.x. tension in mammalian cardiac muscle. J Physiol. 1974;242:429–451. 44. Gómez AM, Cheng H, Lederer WJ, Bers DM. Ca2+ diffusion and sar- 66. Tada M, Kirchberger MA, Katz AM. Phosphorylation of a 22,000-dalton coplasmic reticulum transport both contribute to [Ca2+]i decline dur- component of the cardiac sarcoplasmic reticulum by adenosine 3’:5’-mono- ing Ca2+ sparks in rat ventricular myocytes. J Physiol. 1996;496(pt phosphate-dependent protein kinase. J Biol Chem. 1975;250:2640–2647. 2):575–581. 67. Berlin JR, Bassani JW, Bers DM. Intrinsic cytosolic calcium buffering 45. Briston SJ, Caldwell JL, Horn MA, Clarke JD, Richards MA, Greensmith properties of single rat cardiac myocytes. Biophys J. 1994;67:1775–1787. DJ, Graham HK, Hall MC, Eisner DA, Dibb KM, Trafford AW. Impaired doi: 10.1016/S0006-3495(94)80652-6. β-adrenergic responsiveness accentuates dysfunctional excitation-con- 68. Trafford AW, Díaz ME, Eisner DA. A novel, rapid and reversible method traction coupling in an ovine model of tachypacing-induced heart failure. to measure Ca buffering and time-course of total sarcoplasmic reticulum J Physiol. 2011;589:1367–1382. doi: 10.1113/jphysiol.2010.203984. Ca content in cardiac ventricular myocytes. Pflugers Arch. 1999;437:501– 46. Periasamy M, Bhupathy P, Babu GJ. Regulation of sarcoplasmic reticulum 503. doi: 10.1007/s004240050808. Ca2+ ATPase pump expression and its relevance to cardiac muscle physi- 69. Higgins ER, Cannell MB, Sneyd J. A buffering SERCA pump in mod- ology and pathology. Cardiovasc Res. 2008;77:265–273. doi: 10.1093/ els of calcium dynamics. Biophys J. 2006;91:151–163. doi: 10.1529/ cvr/cvm056. biophysj.105.075747. 47. Bridge JH. Relationships between the sarcoplasmic reticulum and sarco- 70. Kirschenlohr HL, Metcalfe JC, Morris PG, Rodrigo GC, Smith GA. Ca2+ lemmal calcium transport revealed by rapidly cooling rabbit ventricular transient, Mg2+, and pH measurements in the cardiac cycle by 19F NMR. muscle. J Gen Physiol. 1986;88:437–473. Proc Natl Acad Sci U S A. 1988;85:9017–9021. 48. Allen DG, Jewell BR, Wood EH. Studies of the contractility of mamma- 71. Díaz ME, Trafford AW, Eisner DA. The effects of exogenous calcium buf- lian myocardium at low rates of stimulation. J Physiol. 1976;254:1–17. fers on the systolic calcium transient in rat ventricular myocytes. Biophys 49. Negretti N, Varro A, Eisner DA. Estimate of net calcium fluxes and sarco- J. 2001;80:1915–1925. doi: 10.1016/S0006-3495(01)76161-9. plasmic reticulum calcium content during systole in rat ventricular myo- 72. Falke JJ, Drake SK, Hazard AL, Peersen OB. Molecular tuning of ion cytes. J Physiol. 1995;486(pt 3):581–591. binding to calcium signaling proteins. Q Rev Biophys. 1994;27:219–290. 50. Eisner DA, Trafford AW, Díaz ME, Overend CL, O’Neill SC. The control 73. Rodenbaugh DW, Wang W, Davis J, Edwards T, Potter JD, Metzger of Ca release from the cardiac sarcoplasmic reticulum: regulation versus JM. Parvalbumin isoforms differentially accelerate cardiac myocyte autoregulation. Cardiovasc Res. 1998;38:589–604. relaxation kinetics in an animal model of diastolic dysfunction. Am J 51. Bassani JW, Yuan W, Bers DM. Fractional SR Ca release is regulated Physiol Heart Circ Physiol. 2007;293:H1705–H1713. doi: 10.1152/ by trigger Ca and SR Ca content in cardiac myocytes. Am J Physiol. ajpheart.00232.2007. 1995;268:C1313–C1319. 74. Díaz ME, Trafford AW, Eisner DA. The role of intracellular Ca buffers in 52. Trafford AW, Díaz ME, Negretti N, Eisner DA. Enhanced Ca2+ current determining the shape of the systolic Ca transient in cardiac ventricular and decreased Ca2+ efflux restore sarcoplasmic reticulum Ca2+ content myocytes. Pflugers Arch. 2001;442:96–100. doi: 10.1007/s004240000509. after depletion. Circ Res. 1997;81:477–484. 75. Bottinelli R, Coviello DA, Redwood CS, Pellegrino MA, Maron BJ, 53. Barcenas-Ruiz L, Beuckelmann DJ, Wier WG. Sodium-calcium ex- Spirito P, Watkins H, Reggiani C. A mutant tropomyosin that causes hy- change in heart: membrane currents and changes in [Ca2+]i. Science. pertrophic cardiomyopathy is expressed in vivo and associated with an 1987;238:1720–1722. increased calcium sensitivity. Circ Res. 1998;82:106–115. 54. Sipido KR, Callewaert G, Carmeliet E. Inhibition and rapid recovery of 76. Schober T, Huke S, Venkataraman R, Gryshchenko O, Kryshtal D, Ca2+ current during Ca2+ release from sarcoplasmic reticulum in guinea Hwang HS, Baudenbacher FJ, Knollmann BC. Myofilament Ca sen- pig ventricular myocytes. Circ Res. 1995;76:102–109. sitization increases cytosolic Ca binding affinity, alters intracellular Ca Eisner et al Ca and Cardiac E–C Coupling 193

homeostasis, and causes pause-dependent Ca-triggered arrhythmia. Circ 99. Cannell MB, Cheng H, Lederer WJ. The control of calcium release in Res. 2012;111:170–179. doi: 10.1161/CIRCRESAHA.112.270041. heart muscle. Science. 1995;268:1045–1049. 77. Briston SJ, Dibb KM, Solaro RJ, Eisner DA, Trafford AW. Balanced 100. Eisner DA, Choi HS, Díaz ME, O’Neill SC, Trafford AW. Integrative anal- changes in Ca buffering by SERCA and troponin contribute to Ca han- ysis of calcium cycling in cardiac muscle. Circ Res. 2000;87:1087–1094. dling during β-adrenergic stimulation in cardiac myocytes. Cardiovasc 101. Díaz ME, O’Neill SC, Eisner DA. Sarcoplasmic reticulum calcium con- Res. 2014;104:347–354. doi: 10.1093/cvr/cvu201. tent fluctuation is the key to cardiac alternans. Circ Res. 2004;94:650– 78. Hobai IA, O’Rourke B. Enhanced Ca(2+)-activated Na(+)-Ca(2+) 656. doi: 10.1161/01.RES.0000119923.64774.72. exchange activity in canine pacing-induced heart failure. Circ Res. 102. Díaz ME, Trafford AW, O’Neill SC, Eisner DA. Measurement of sarco- 2000;87:690–698. plasmic reticulum Ca2+ content and sarcolemmal Ca2+ fluxes in isolated 79. Greiser M, Kerfant BG, Williams GS, et al. Tachycardia-induced si- rat ventricular myocytes during spontaneous Ca2+ release. J Physiol. lencing of subcellular Ca2+ signaling in atrial myocytes. J Clin Invest. 1997;501(pt 1):3–16. 2014;124:4759–4772. doi: 10.1172/JCI70102. 103. Fabiato A. Time and calcium dependence of activation and inactivation of 80. Hasenfuss G, Reinecke H, Studer R, Meyer M, Pieske B, Holtz J, calcium-induced release of calcium from the sarcoplasmic reticulum of a Holubarsch C, Posival H, Just H, Drexler H. Relation between myocardial skinned canine cardiac Purkinje cell. J Gen Physiol. 1985;85:247–289. function and expression of sarcoplasmic reticulum Ca(2+)-ATPase in fail- 104. Sham JS, Song LS, Chen Y, Deng LH, Stern MD, Lakatta EG, Cheng H. ing and nonfailing human myocardium. Circ Res. 1994;75:434–442. Termination of Ca2+ release by a local inactivation of ryanodine receptors 81. Eisner D, Caldwell J, Trafford A. Sarcoplasmic reticulum Ca-ATPase and in cardiac myocytes. Proc Natl Acad Sci U S A. 1998;95:15096–15101. heart failure 20 years later. Circ Res. 2013;113:958–961. doi: 10.1161/ 105. Györke S, Fill M. Ryanodine receptor adaptation: control mechanism of CIRCRESAHA.113.302187. Ca(2+)-induced Ca2+ release in heart. Science. 1993;260:807–809. 82. Shannon TR, Guo T, Bers DM. Ca2+ scraps: local depletions of free 106. Valdivia HH, Kaplan JH, Ellis-Davies GC, Lederer WJ. Rapid adaptation [Ca2+] in cardiac sarcoplasmic reticulum during contractions leave of cardiac ryanodine receptors: modulation by Mg2+ and phosphoryla- substantial Ca2+ reserve. Circ Res. 2003;93:40–45. doi: 10.1161/01. tion. Science. 1995;267:1997–2000. RES.0000079967.11815.19. 107. Niggli E. Localized intracellular calcium signaling in muscle: calcium 83. Díaz ME, Graham HK, O’neill SC, Trafford AW, Eisner DA. The control sparks and calcium quarks. Annu Rev Physiol. 1999;61:311–335. doi: of sarcoplasmic reticulum Ca content in cardiac muscle. Cell Calcium. 10.1146/annurev.physiol.61.1.311. 2005;38:391–396. doi: 10.1016/j.ceca.2005.06.017. 108. Sitsapesan R, Williams AJ. Regulation of the gating of the sheep car- 84. Trafford AW, Díaz ME, Eisner DA. Coordinated control of cell Ca(2+) diac sarcoplasmic reticulum Ca(2+)-release channel by luminal Ca2+. loading and triggered release from the sarcoplasmic reticulum underlies J Membr Biol. 1994;137:215–226. the rapid inotropic response to increased L-type Ca(2+) current. Circ Res. 109. Terentyev D, Györke I, Belevych AE, Terentyeva R, Sridhar A, Nishijima 2001;88:195–201. Y, de Blanco EC, Khanna S, Sen CK, Cardounel AJ, Carnes CA, Györke 85. Clarke JD, Caldwell JL, Horn MA, Bode EF, Richards MA, Hall MC, S. Redox modification of ryanodine receptors contributes to sarcoplasmic Graham HK, Briston SJ, Greensmith DJ, Eisner DA, Dibb KM, Trafford reticulum Ca2+ leak in chronic heart failure. Circ Res. 2008;103:1466– AW. Perturbed atrial calcium handling in an ovine model of heart failure: 1472. doi: 10.1161/CIRCRESAHA.108.184457. potential roles for reductions in the L-type calcium current. J Mol Cell 110. Sobie EA, Dilly KW, dos Santos Cruz J, Lederer WJ, Jafri MS. Cardiol. 2015;79:169–179. doi: 10.1016/j.yjmcc.2014.11.017. Termination of cardiac Ca(2+) sparks: an investigative mathematical 86. Fabiato A. Simulated calcium current can both cause calcium loading in model of calcium-induced calcium release. Biophys J. 2002;83:59–78. and trigger calcium release from the sarcoplasmic reticulum of a skinned 111. Marx SO, Gaburjakova J, Gaburjakova M, Henrikson C, Ondrias K, canine cardiac Purkinje cell. J Gen Physiol. 1985;85:291–320. Marks AR. Coupled gating between cardiac calcium release channels 87. Eisner DA, Trafford AW. No role for the ryanodine receptor in regulating (ryanodine receptors). Circ Res. 2001;88:1151–1158. cardiac contraction? News Physiol Sci. 2000;15:275–279. 112. Guo T, Gillespie D, Fill M. Ryanodine receptor current amplitude con- 88. Brochet DX, Xie W, Yang D, Cheng H, Lederer WJ. Quarky cal- trols Ca2+ sparks in cardiac muscle. Circ Res. 2012;111:28–36. doi: cium release in the heart. Circ Res. 2011;108:210–218. doi: 10.1161/ 10.1161/CIRCRESAHA.112.265652. CIRCRESAHA.110.231258. 113. Laver DR, Kong CH, Imtiaz MS, Cannell MB. Termination of calcium- 89. Venetucci LA, Trafford AW, Díaz ME, O’Neill SC, Eisner DA. Reducing induced calcium release by induction decay: an emergent property of ryanodine receptor open probability as a means to abolish spontaneous Ca2+ stochastic channel gating and molecular scale architecture. J Mol Cell release and increase Ca2+ transient amplitude in adult ventricular myocytes. Cardiol. 2013;54:98–100. doi: 10.1016/j.yjmcc.2012.10.009. Circ Res. 2006;98:1299–1305. doi: 10.1161/01.RES.0000222000.35500.65. 114. Cannell MB, Kong CH, Imtiaz MS, Laver DR. Control of sarcoplasmic 90. Varro A, Negretti N, Hester SB, Eisner DA. An estimate of the calcium reticulum Ca2+ release by stochastic RyR gating within a 3D model of content of the sarcoplasmic reticulum in rat ventricular myocytes. Pflugers the cardiac dyad and importance of induction decay for CICR termina- Arch. 1993;423:158–160. tion. Biophys J. 2013;104:2149–2159. doi: 10.1016/j.bpj.2013.03.058. 91. Zhang L, Kelley J, Schmeisser G, Kobayashi YM, Jones LR. Complex 115. Gillespie D, Fill M. Pernicious attrition and inter-RyR2 CICR current formation between junctin, triadin, calsequestrin, and the ryanodine recep- control in cardiac muscle. J Mol Cell Cardiol. 2013;58:53–58. doi: tor. Proteins of the cardiac junctional sarcoplasmic reticulum membrane. 10.1016/j.yjmcc.2013.01.011. J Biol Chem. 1997;272:23389–23397. 116. Bovo E, Mazurek SR, Fill M, Zima AV. Cytosolic Ca²⁺ buffering 92. Györke I, Hester N, Jones LR, Györke S. The role of calsequestrin, determines the intra-SR Ca²⁺ concentration at which cardiac Ca²⁺ triadin, and junctin in conferring cardiac ryanodine receptor responsive- sparks terminate. Cell Calcium. 2015;58:246–253. doi: 10.1016/j. ness to luminal calcium. Biophys J. 2004;86:2121–2128. doi: 10.1016/ ceca.2015.06.002. S0006-3495(04)74271-X. 117. Vornanen M, Shepherd N. Restitution of contractility in single ventricu- 93. Endo M, Tanaka M, Ogawa Y. Calcium induced release of calcium from lar myocytes of guinea pig heart. Cardiovasc Res. 1997;33:611–622. the sarcoplasmic reticulum of skinned skeletal muscle fibres. Nature. 118. DelPrincipe F, Egger M, Niggli E. Calcium signalling in cardiac 1970;228:34–36. muscle: refractoriness revealed by coherent activation. Nat Cell Biol. 94. Cannell MB, Berlin JR, Lederer WJ. Effect of membrane potential chang- 1999;1:323–329. doi: 10.1038/14013. es on the calcium transient in single rat cardiac muscle cells. Science. 119. Terentyev D, Viatchenko-Karpinski S, Valdivia HH, Escobar AL, Györke 1987;238:1419–1423. S. Luminal Ca2+ controls termination and refractory behavior of Ca2+- 95. Barcenas-Ruiz L, Wier WG. Voltage dependence of intracellular [Ca2+]i induced Ca2+ release in cardiac myocytes. Circ Res. 2002;91:414–420. transients in guinea pig ventricular myocytes. Circ Res. 1987;61:148–154. 120. Belevych AE, Terentyev D, Terentyeva R, Ho HT, Gyorke I, Bonilla IM, 96. Picht E, Zima AV, Shannon TR, Duncan AM, Blatter LA, Bers Carnes CA, Billman GE, Györke S. Shortened Ca2+ signaling refrac- DM. Dynamic calcium movement inside cardiac sarcoplasmic re- toriness underlies cellular arrhythmogenesis in a postinfarction model ticulum during release. Circ Res. 2011;108:847–856. doi: 10.1161/ of sudden cardiac death. Circ Res. 2012;110:569–577. doi: 10.1161/ CIRCRESAHA.111.240234. CIRCRESAHA.111.260455. 97. Stern MD. Theory of excitation-contraction coupling in cardiac muscle. 121. Lewartowski B, Pytkowski B, Janczewski A. Calcium fraction corre- Biophys J. 1992;63:497–517. doi: 10.1016/S0006-3495(92)81615-6. lating with contractile force of ventricular muscle of guinea-pig heart. 98. Cheng H, Lederer WJ, Cannell MB. Calcium sparks: elementary events Pflugers Arch. 1984;401:198–203. underlying excitation-contraction coupling in heart muscle. Science. 122. Zima AV, Bovo E, Bers DM, Blatter LA. Ca²+ spark-dependent and -in- 1993;262:740–744. dependent sarcoplasmic reticulum Ca²+ leak in normal and failing rabbit 194 Circulation Research July 7, 2017

ventricular myocytes. J Physiol. 2010;588:4743–4757. doi: 10.1113/ 143. Pogwizd SM, Bers DM. Cellular basis of triggered arrhythmias in jphysiol.2010.197913. heart failure. Trends Cardiovasc Med. 2004;14:61–66. doi: 10.1016/j. 123. Williams GS, Chikando AC, Tuan HT, Sobie EA, Lederer WJ, Jafri MS. tcm.2003.12.002. Dynamics of calcium sparks and calcium leak in the heart. Biophys J. 144. Venetucci LA, Trafford AW, Eisner DA. Increasing ryanodine receptor 2011;101:1287–1296. doi: 10.1016/j.bpj.2011.07.021. open probability alone does not produce arrhythmogenic calcium waves: 124. Jiang D, Xiao B, Zhang L, Chen SR. Enhanced basal activity of a cardiac threshold sarcoplasmic reticulum calcium content is required. Circ Res. Ca2+ release channel (ryanodine receptor) mutant associated with ven- 2007;100:105–111. doi: 10.1161/01.RES.0000252828.17939.00. tricular tachycardia and sudden death. Circ Res. 2002;91:218–225. 145. Kashimura T, Briston SJ, Trafford AW, Napolitano C, Priori SG, Eisner 125. Terentyev D, Nori A, Santoro M, Viatchenko-Karpinski S, Kubalova DA, Venetucci LA. In the RyR2(R4496C) mouse model of CPVT, β- Z, Gyorke I, Terentyeva R, Vedamoorthyrao S, Blom NA, Valle G, adrenergic stimulation induces Ca waves by increasing SR Ca content and Napolitano C, Williams SC, Volpe P, Priori SG, Gyorke S. Abnormal in- not by decreasing the threshold for Ca waves. Circ Res. 2010;107:1483– teractions of calsequestrin with the ryanodine receptor calcium release 1489. doi: 10.1161/CIRCRESAHA.110.227744. channel complex linked to exercise-induced sudden cardiac death. Circ 146. Guo T, Zhang T, Ginsburg KS, Mishra S, Brown JH, Bers DM. Res. 2006;98:1151–1158. doi: 10.1161/01.RES.0000220647.93982.08. CaMKIIδC slows [Ca]i decline in cardiac myocytes by promot- 126. Marx SO, Reiken S, Hisamatsu Y, Jayaraman T, Burkhoff D, Rosemblit ing Ca sparks. Biophys J. 2012;102:2461–2470. doi: 10.1016/j. N, Marks AR. PKA phosphorylation dissociates FKBP12.6 from the cal- bpj.2012.04.015. cium release channel (ryanodine receptor): defective regulation in failing 147. Belevych A, Kubalova Z, Terentyev D, Hamlin RL, Carnes CA, hearts. Cell. 2000;101:365–376. Györke S. Enhanced ryanodine receptor-mediated calcium leak de- 127. Guo T, Cornea RL, Huke S, Camors E, Yang Y, Picht E, Fruen BR, termines reduced sarcoplasmic reticulum calcium content in chronic Bers DM. Kinetics of FKBP12.6 binding to ryanodine receptors in canine heart failure. Biophys J. 2007;93:4083–4092. doi: 10.1529/ permeabilized cardiac myocytes and effects on Ca sparks. Circ Res. biophysj.107.114546. 2010;106:1743–1752. doi: 10.1161/CIRCRESAHA.110.219816. 148. De Stefani D, Raffaello A, Teardo E, Szabò I, Rizzuto R. A forty-kilo- 128. Curran J, Brown KH, Santiago DJ, Pogwizd S, Bers DM, Shannon TR. dalton protein of the inner membrane is the mitochondrial calcium uni- Spontaneous Ca waves in ventricular myocytes from failing hearts de- porter. Nature. 2011;476:336–340. doi: 10.1038/nature10230. pend on Ca(2+)-calmodulin-dependent protein kinase II. J Mol Cell 149. Baughman JM, Perocchi F, Girgis HS, Plovanich M, Belcher-Timme Cardiol. 2010;49:25–32. doi: 10.1016/j.yjmcc.2010.03.013. CA, Sancak Y, Bao XR, Strittmatter L, Goldberger O, Bogorad RL, 129. Ai X, Curran JW, Shannon TR, Bers DM, Pogwizd SM. Ca2+/calmodu- Koteliansky V, Mootha VK. Integrative genomics identifies MCU as lin-dependent protein kinase modulates cardiac ryanodine receptor phos- an essential component of the mitochondrial calcium uniporter. Nature. phorylation and sarcoplasmic reticulum Ca2+ leak in heart failure. Circ 2011;476:341–345. doi: 10.1038/nature10234. Res. 2005;97:1314–1322. doi: 10.1161/01.RES.0000194329.41863.89. 150. Finkel T, Menazza S, Holmström KM, Parks RJ, Liu J, Sun J, Liu J, 130. Cooper LL, Li W, Lu Y, Centracchio J, Terentyeva R, Koren G, Terentyev Pan X, Murphy E. The ins and outs of mitochondrial calcium. Circ Res. D. Redox modification of ryanodine receptors by mitochondria-de- 2015;116:1810–1819. doi: 10.1161/CIRCRESAHA.116.305484. rived reactive oxygen species contributes to aberrant Ca2+ handling 151. Lu X, Ginsburg KS, Kettlewell S, Bossuyt J, Smith GL, Bers DM. in ageing rabbit hearts. J Physiol. 2013;591:5895–5911. doi: 10.1113/ Measuring local gradients of intramitochondrial [Ca(2+)] in cardiac jphysiol.2013.260521. myocytes during sarcoplasmic reticulum Ca(2+) release. Circ Res. 131. Anderson ME. Oxidant stress promotes disease by activating CaMKII. J 2013;112:424–431. doi: 10.1161/CIRCRESAHA.111.300501. Mol Cell Cardiol. 2015;89:160–167. doi: 10.1016/j.yjmcc.2015.10.014. 152. Hüser J, Blatter LA, Sheu SS. Mitochondrial calcium in heart cells: beat- 132. Bers DM. Cardiac sarcoplasmic reticulum calcium leak: basis and to-beat oscillations or slow integration of cytosolic transients? J Bioenerg roles in cardiac dysfunction. Annu Rev Physiol. 2014;76:107–127. doi: Biomembr. 2000;32:27–33. 10.1146/annurev-physiol-020911-153308. 153. Williams GS, Boyman L, Chikando AC, Khairallah RJ, Lederer 133. Kubalova Z, Terentyev D, Viatchenko-Karpinski S, Nishijima Y, Györke WJ. Mitochondrial calcium uptake. Proc Natl Acad Sci U S A. I, Terentyeva R, da Cuñha DN, Sridhar A, Feldman DS, Hamlin RL, 2013;110:10479–10486. doi: 10.1073/pnas.1300410110. Carnes CA, Györke S. Abnormal intrastore calcium signaling in chronic 154. Bers DM, Bassani JW, Bassani RA. Competition and redistribution heart failure. Proc Natl Acad Sci U S A. 2005;102:14104–14109. doi: among calcium transport systems in rabbit cardiac myocytes. Cardiovasc 10.1073/pnas.0504298102. Res. 1993;27:1772–1777. 134. Belevych AE, Terentyev D, Terentyeva R, Nishijima Y, Sridhar A, 155. Negretti N, O’Neill SC, Eisner DA. The relative contributions of differ- Hamlin RL, Carnes CA, Györke S. The relationship between arrhythmo- ent intracellular and sarcolemmal systems to relaxation in rat ventricular genesis and impaired contractility in heart failure: role of altered ryano- myocytes. Cardiovasc Res. 1993;27:1826–1830. dine receptor function. Cardiovasc Res. 2011;90:493–502. doi: 10.1093/ 156. Drago I, De Stefani D, Rizzuto R, Pozzan T. Mitochondrial Ca2+ up- cvr/cvr025. take contributes to buffering cytoplasmic Ca2+ peaks in cardiomyo- 135. Negretti N, O’Neill SC, Eisner DA. The effects of inhibitors of sarcoplas- cytes. Proc Natl Acad Sci U S A. 2012;109:12986–12991. doi: 10.1073/ mic reticulum function on the systolic Ca2+ transient in rat ventricular pnas.1210718109. myocytes. J Physiol. 1993;468:35–52. 157. Pan X, Liu J, Nguyen T, Liu C, Sun J, Teng Y, Fergusson MM, Rovira 136. Sankaranarayanan R, Li Y, Greensmith DJ, Eisner DA, Venetucci L. II, Allen M, Springer DA, Aponte AM, Gucek M, Balaban RS, Murphy Biphasic decay of the Ca transient results from increased sarcoplas- E, Finkel T. The physiological role of mitochondrial calcium revealed mic reticulum Ca leak. J Physiol. 2016;594:611–623. doi: 10.1113/ by mice lacking the mitochondrial calcium uniporter. Nat Cell Biol. JP271473. 2013;15:1464–1472. doi: 10.1038/ncb2868. 137. Cheng H, Lederer MR, Lederer WJ, Cannell MB. Calcium sparks and 158. Kwong JQ, Lu X, Correll RN, Schwanekamp JA, Vagnozzi RJ, Sargent [Ca2+]i waves in cardiac myocytes. Am J Physiol. 1996;270:C148–C159. MA, York AJ, Zhang J, Bers DM, Molkentin JD. The mitochondrial 138. Mechmann S, Pott L. Identification of Na-Ca exchange current in single calcium uniporter selectively matches metabolic output to acute con- cardiac myocytes. Nature. 1986;319:597–599. doi: 10.1038/319597a0. tractile stress in the heart. Cell Rep. 2015;12:15–22. doi: 10.1016/j. 139. Ferrier GR, Saunders JH, Mendez C. A cellular mechanism for the gen- celrep.2015.06.002. eration of ventricular arrhythmias by acetylstrophanthidin. Circ Res. 159. Owan TE, Hodge DO, Herges RM, Jacobsen SJ, Roger VL, Redfield 1973;32:600–609. MM. Trends in prevalence and outcome of heart failure with preserved 140. Kass RS, Lederer WJ, Tsien RW, Weingart R. Role of calcium ions in ejection fraction. N Engl J Med. 2006;355:251–259. doi: 10.1056/ transient inward currents and aftercontractions induced by strophanthi- NEJMoa052256. din in cardiac Purkinje fibres. J Physiol. 1978;281:187–208. 160. Li AH, Liu PP, Villarreal FJ, Garcia RA. Dynamic changes in myocar- 141. Jiang D, Xiao B, Yang D, Wang R, Choi P, Zhang L, Cheng H, Chen dial matrix and relevance to disease: translational perspectives. Circ Res. SR. RyR2 mutations linked to ventricular tachycardia and sudden death 2014;114:916–927. doi: 10.1161/CIRCRESAHA.114.302819. reduce the threshold for store-overload-induced Ca2+ release (SOICR). 161. Eisner DA, Orchard CH, Allen DG. Control of intracellular ionized cal- Proc Natl Acad Sci U S A. 2004;101:13062–13067. doi: 10.1073/ cium concentration by sarcolemmal and intracellular mechanisms. J Mol pnas.0402388101. Cell Cardiol. 1984;16:137–146. 142. Priori SG, Chen SR. Inherited dysfunction of sarcoplasmic reticulum 162. Ríos E. The cell boundary theorem: a simple law of the control of cyto- Ca2+ handling and arrhythmogenesis. Circ Res. 2011;108:871–883. doi: solic calcium concentration. J Physiol Sci. 2010;60:81–84. doi: 10.1007/ 10.1161/CIRCRESAHA.110.226845. s12576-009-0069-z. Eisner et al Ca and Cardiac E–C Coupling 195

163. Eisner DA, Lederer WJ, Vaughan-Jones RD. The control of tonic tension 174. Hunton DL, Zou L, Pang Y, Marchase RB. Adult rat cardiomyocytes by membrane potential and intracellular sodium activity in the sheep car- exhibit capacitative calcium entry. Am J Physiol Heart Circ Physiol. diac Purkinje fibre. J Physiol. 1983;335:723–743. 2004;286:H1124–H1132. doi: 10.1152/ajpheart.00162.2003. 164. Allen DG, Orchard CH. Measurements of intracellular calcium concen- 175. Kojima A, Kitagawa H, Omatsu-Kanbe M, Matsuura H, Nosaka S. tration in heart muscle: the effects of inotropic interventions and hypoxia. Presence of store-operated Ca2+ entry in C57BL/6J mouse ven- J Mol Cell Cardiol. 1984;16:117–128. tricular myocytes and its suppression by sevoflurane. Br J Anaesth. 165. Bassani JW, Bassani RA, Bers DM. Relaxation in rabbit and rat cardiac 2012;109:352–360. doi: 10.1093/bja/aes212. cells: species-dependent differences in cellular mechanisms. J Physiol. 176. Zhao G, Li T, Brochet DX, Rosenberg PB, Lederer WJ. STIM1 enhances 1994;476:279–293. SR Ca2+ content through binding phospholamban in rat ventricular myo- 166. Choi HS, Eisner DA. The role of sarcolemmal Ca2+-ATPase in the cytes. Proc Natl Acad Sci U S A. 2015;112:E4792–E4801. doi: 10.1073/ regulation of resting calcium concentration in rat ventricular myocytes. pnas.1423295112. J Physiol. 1999;515(pt 1):109–118. 177. Collins HE, Zhu-Mauldin X, Marchase RB, Chatham JC. STIM1/ 167. Choi HS, Trafford AW, Eisner DA. Measurement of calcium en- Orai1-mediated SOCE: current perspectives and potential roles in try and exit in quiescent rat ventricular myocytes. Pflugers Arch. cardiac function and pathology. Am J Physiol Heart Circ Physiol. 2000;440:600–608. 2013;305:H446–H458. doi: 10.1152/ajpheart.00104.2013. 168. Kupittayanant P, Trafford AW, Díaz ME, Eisner DA. A mechanism dis- 178. Luo X, Hojayev B, Jiang N, Wang ZV, Tandan S, Rakalin A, Rothermel tinct from the L-type Ca current or Na-Ca exchange contributes to Ca BA, Gillette TG, Hill JA. STIM1-dependent store-operated Ca²⁺ entry entry in rat ventricular myocytes. Cell Calcium. 2006;39:417–423. doi: is required for pathological cardiac hypertrophy. J Mol Cell Cardiol. 10.1016/j.ceca.2006.01.011. 2012;52:136–147. doi: 10.1016/j.yjmcc.2011.11.003. 169. Touchberry CD, Elmore CJ, Nguyen TM, Andresen JJ, Zhao X, Orange 179. Voelkers M, Salz M, Herzog N, Frank D, Dolatabadi N, Frey N, Gude M, Weisleder N, Brotto M, Claycomb WC, Wacker MJ. Store-operated N, Friedrich O, Koch WJ, Katus HA, Sussman MA, Most P. Orai1 and calcium entry is present in HL-1 cardiomyocytes and contributes to Stim1 regulate normal and hypertrophic growth in cardiomyocytes. J Mol resting calcium. Biochem Biophys Res Commun. 2011;416:45–50. doi: Cell Cardiol. 2010;48:1329–1334. doi: 10.1016/j.yjmcc.2010.01.020. 10.1016/j.bbrc.2011.10.133. 180. Correll RN, Goonasekera SA, van Berlo JH, Burr AR, Accornero F, 170. Parekh AB, Putney JW Jr. Store-operated calcium channels. Physiol Rev. Zhang H, Makarewich CA, York AJ, Sargent MA, Chen X, Houser SR, 2005;85:757–810. doi: 10.1152/physrev.00057.2003. Molkentin JD. STIM1 elevation in the heart results in aberrant Ca²⁺ 171. Prakriya M, Lewis RS. Store-operated calcium channels. Physiol Rev. handling and cardiomyopathy. J Mol Cell Cardiol. 2015;87:38–47. doi: 2015;95:1383–1436. doi: 10.1152/physrev.00020.2014. 10.1016/j.yjmcc.2015.07.032. 172. Hunton DL, Lucchesi PA, Pang Y, Cheng X, Dell’Italia LJ, Marchase 181. Dibb KM, Eisner DA, Trafford AW. Regulation of systolic [Ca2+]i and RB. Capacitative calcium entry contributes to nuclear factor of ac- cellular Ca2+ flux balance in rat ventricular myocytes by SR Ca2+, tivated T-cells nuclear translocation and hypertrophy in cardio- L-type Ca2+ current and diastolic [Ca2+]i. J Physiol. 2007;585:579– myocytes. J Biol Chem. 2002;277:14266–14273. doi: 10.1074/jbc. 592. doi: 10.1113/jphysiol.2007.141473. M107167200. 182. Antoons G, Mubagwa K, Nevelsteen I, Sipido KR. Mechanisms underly- 173. Uehara A, Yasukochi M, Imanaga I, Nishi M, Takeshima H. Store- ing the frequency dependence of contraction and [Ca(2+)](i) transients in operated Ca2+ entry uncoupled with ryanodine receptor and junctional mouse ventricular myocytes. J Physiol. 2002;543:889–898. membrane complex in heart muscle cells. Cell Calcium. 2002;31:89–96. 183. Eisner D. Calcium in the heart: from physiology to disease. Exp Physiol. doi: 10.1054/ceca.2001.0257. 2014;99:1273–1282. doi: 10.1113/expphysiol.2013.077305.