Investigating the Role of Mitochondrial Dynamic Proteins Mid49 and Mid51, As Novel Targets of Cardioprotection

INVESTIGATING THE ROLE OF MITOCHONDRIAL DYNAMIC PROTEINS MID49 AND MID51, AS NOVEL TARGETS OF CARDIOPROTECTION

Thesis submitted by Parisa Samangouei BSc, MSc (First class with Honours) For the degree of Doctor of Philosophy University College London, UK

Primary supervisor: Professor Derek Hausenloy Secondary supervisor: Professor Derek Yellon Tertiary supervisor : Dr Andrew Hall

Institute of Cardiovascular Science Hatter Cardiovascular institute, University College London, 67 Chenies Mews, London WC1E 6HX.

March 2018

DECLARATION

I, Parisa Samangouei confirm that the work presented in this thesis is my own. Where information has been derived from other sources, I confirm that this has been indicated in the thesis. Data gathered from pilot studies conducted by other investigators, within our research group, presented in the introduction sections, has been acknowledged and noted. Any collaborations and assistance provided by other investigators in aiding to generate experimental results, presented in this thesis, has been indicated in the respective methods and results sections.

Parisa Samangouei

ABSTRACT

Background

Acute myocardial infarction and the heart failure that often ensues are the leading causes of death and disability worldwide. As such, novel therapeutic strategies are required to protect the heart from acute ischaemia-reperfusion injury (IRI). Mitochondria play a central role in determining the fate of cardiomyocytes, during acute myocardial IRI. Genetic and pharmacological inhibition of Drp1-mediated mitochondrial fission has been shown to reduce cardiomyocyte death, during acute IRI. Accordingly, we investigated the role of the newly described mitochondrial- specific Drp1 receptors, MiD49 and MiD51, as novel targets for cardioprotection. We hypothesised that inhibition of MiD49 and MiD51 would render the heart more resistant to acute IRI, and provide novel therapeutic targets for cardioprotection.

Methods and Results

1. In cardiac cell-lines, the genetic knockdown (KD) of both MiD49 and MiD51 promoted mitochondrial elongation, inhibited mitochondrial permeability transition pore opening, reduced cell death and prevented mitochondrial calcium overload during simulated IRI, when compared to control cells.

2. In adult mice, MiD49 deficiency had no significant effect on mitochondrial morphology, cardiac size and function, or myocardial infarct size, when compared to wild-type littermates. The effect of dual cardiac-specific genetic ablation of MiD49 and MiD51 in the heart on the susceptibility to acute IRI is currently being investigated.

3. We have developed a biophysical assay to high-throughput screen for novel small molecule inhibitors of the interaction between Drp1 and MiD49 and MiD51, as a therapeutic strategy for inhibiting mitochondrial fission.

Conclusions and Further studies

For the first time, we provide evidence for the role of MiD49 and MiD51 as novel therapeutic targets for cardioprotection. Studies are ongoing to validate their roles in the adult heart. In order to provide a novel therapeutic strategy for inhibiting

2 mitochondrial fission as a cardioprotective strategy, we aim to identify novel small molecule inhibitors of the interaction between Drp1 and MiD49/MiD51.

CONTENTS

DECLARATION ...... 1

ABSTRACT ...... 2

CONTENTS ...... 4

ACKNOWLEDGEMENTS ...... 9

PUBLICATIONS ...... 10

FIGURES...... 11

TABLES ...... 14

ABBREVIATIONS ...... 15

CHAPTER 1: General introduction ...... 19

1.1 Epidemiology of ischaemic heart disease ...... 19 1.1.1 Atherosclerosis ...... 20 1.1.2 Myocardial infarction ...... 21 1.2 Cardiac metabolism ...... 23 1.2.1 Oxidative phosphorylation ...... 23 1.2.2 Cardiac Ischaemia ...... 25 1.2.3 Cardiac Reperfusion ...... 27 1.3 Mitochondrial Dynamics and Cardioprotection ...... 30 1.3.1 Mitochondrial Fusion ...... 30 1.3.2 Mitochondrial fission ...... 32 1.4 Mitochondrial arrangement and dynamics in the heart ...... 35 1.4.1 Observing mitochondrial dynamics in cardiomyocytes ...... 36 1.4.2 Inhibition of mitochondrial fission in the heart during IRI ...... 37 1.5 Mitochondrial fission proteins MiD49 and MiD51 as targets of cardioprotection ...... 39 1.5.1 Discovery ...... 39 1.5.2 Cellular expression of MiD49 and MiD51 ...... 41 1.5.3 MiD49 and MiD51 are Drp1 adaptors involved in the mitochondrial fission machinery ...... 42 1.5.4 The molecular structure of MiD49 and MiD51 in relation to their function .....47 1.5.5 MiD-ER interaction ...... 52 1.5.6 Regulation of MiD49 and MiD51 activity during physiological and stressed conditions ...... 54 CHAPTER 2: Research Objectives ...... 56

2.1 Overall objective ...... 56 2.2 Overall hypothesis ...... 56 2.3 Aims ...... 56 CHAPTER 3: Investigating the role of MiD49 and MiD51, as targets of cardioprotection: in vitro studies ...... 58

3.1 Introduction ...... 58 3.2 Research objectives ...... 59 3.2.1 Aim1: Investigating the effect of MiD49 and MiD51 overexpression: in vitro studies ...... 59 3.2.2 Aim 2: Investigating the effect of MiD49 and MiD51 genetic ablation: in vitro studies ...... 60 3.3 Materials and Methods ...... 60 3.3.2 The general method of cell passage and seeding ...... 61 3.3.3 Freezing cell stocks ...... 62 3.3.4 Thawing of frozen cell lines ...... 63 3.3.5 Transfection of cell lines ...... 63 3.3.6 Cell fixation ...... 65 3.3.7 Fluorescence activated cell sorting ...... 65 3.3.8 Fluorescence microscopy ...... 66 3.3.9 Confocal microscopy ...... 66 3.3.10 MPTP opening ...... 67 3.3.11 Determining cell survival after SIRI ...... 69 3.3.12 Real-time simulated ischaemia-reperfusion injury confocal set-up ...... 70 3.3.13 Statistical analysis ...... 71 Aim 1: Investigating the effect of MiD49 and MiD51 overexpression: in vitro studies ...... 72

3.4 Results ...... 72 3.4.1 Overexpression of MiD49 and MiD51 produces hyperfused mitochondria ...... 72 3.4.2 Mitochondria overexpressing MiD49 and MiD51 are protected from fragmentation during SIRI ...... 73 3.4.3 MiD49 and MiD51 overexpression delays mitochondrial permeability transition pore opening during oxidative stress...... 78 3.4.4 MiD49 and MiD51 overexpression reduces cell death post SIRI ...... 80 Aim 2: Investigating the effect of MiD49 and MiD51 genetic ablation, as targets of cardioprotection: in vitro studies ...... 82

3.5 Results ...... 82 3.5.1 Transfection efficiency of cardiac and non-cardiac cell lines ...... 82

3.5.2 Knockdown of MiD49 and MiD51 induces mitochondrial elongation ...... 87 3.5.3 Knockdown of MiD49 and MiD51 delays MPTP opening in the presence of high mitochondrial matrix ROS ...... 92 3.5.4 Mitochondrial elongation by alteration of MiD49 and MiD51 expression protects cells against ischaemia-reperfusion injury ...... 96 3.5.5 Knockdown of MiD49 and MiD51 attenuates mitochondrial calcium overload during simulated ischaemia-reperfusion injury ...... 107 3.6 Discussion ...... 110 3.6.1 Transfection efficiency of cell lines ...... 110 3.6.2 Overexpression of MiD49 and MiD51 in cell lines ...... 110 3.6.3 Knockdown of MiD49 and MiD51 in cardiac cell lines ...... 114 3.6.4 Knockdown of MiD49 and MiD51 in cardiac cell lines protects the cells against SIRI ...... 115 3.6.5 Summary ...... 117 CHAPTER 4: Investigating the role of MiD49 and MiD51, as targets of cardioprotection: in vivo studies ...... 118

4.1 Introduction ...... 118 4.2 Research objectives and aims ...... 119 4.2.1 Aim 1: Investigating the effect of MiD49 genetic ablation, as targets of cardioprotection: in vivo studies ...... 119 4.2.2 Aim 2: Investigating the effect of cardiomyocyte MiD49 and MiD51 genetic ablation, as targets of cardioprotection: in vivo studies ...... 120 Aim 1: Investigating the effect of MiD49 ablation, as a target of cardioprotection: in vivo studies ...... 121

4.3 Materials and Methods ...... 121 4.3.1 Experimental use of animals ...... 121 4.3.2 Transgenic mouse background ...... 121 4.3.3 Genotyping ...... 122 4.3.4 Cardiac mitochondrial phenotyping by electron microscopy ...... 126 4.3.5 Cardiac phenotyping by echocardiography ...... 128 4.3.6 Non-recovery acute Ischaemia-reperfusion in vivo model ...... 132 4.4 Results ...... 136 4.4.1 Cardiac mitochondrial phenotyping of whole body MiD49 KO mice ...... 136 4.4.2 Cardiac phenotyping of MiD49 KO mice during basal and cardiac stress conditions ...... 138 4.4.3 Susceptibility of MiD49 KO mice to ischaemia-reperfusion injury ...... 142 Aim 2: Investigating the effect of cardiomyocyte MiD49 and MiD51 ablation, as targets of cardioprotection: in vivo studies ...... 144

4.5 Background ...... 144

4.5.2 Recombinant AAV production ...... 146 4.5.3 The cardiotropic AAV9 serotype ...... 148 4.6 Materials and Methods ...... 149 4.6.1 MiD49 and MiD51 RNAi Knockdown ...... 150 4.6.2 MiD49 and MiD51 knockdown in mouse embryonic fibroblasts using RNAi vectors ...... 153 4.6.3 Development of MiD49 and MiD51 cardiac-specific AAV9 constructs ...... 154 4.6.4 Knockdown of MiD49 and MiD51 in C57BL/6 mice ...... 164 4.7 Results ...... 166 4.7.1 Identifying the best RNAi oligonucleotide to knockdown MiD49 and MiD51 in mouse embryonic fibroblasts ...... 166 4.7.2 Transfection efficiency of rAAV9 vectors ...... 166 4.7.3 rAAV9 purification and injection ...... 169 4.8 Discussion ...... 170 4.8.1 The role of MiD49 in determining the cardiac mitochondrial morphology .... 170 4.8.2 Cardiac phenotype of MiD49 KO mice ...... 171 4.8.3 Susceptibility of MiD49 KO mice to ischaemia-reperfusion injury ...... 173 4.8.4 Cardiac-specific knockdown of MiD49 and MiD51 in mice ...... 175 4.8.5 Summary ...... 176 CHAPTER 5: Identification of a chemical inhibitor of the Drp1-MiD49/MiD51 interaction 178

5.1 Introduction ...... 178 5.1.1 Biophysical fragment-based approaches for drug discovery ...... 179 5.2 Research objective and Aims ...... 182 5.3 Materials and Methods ...... 182 5.3.1 Synthesis of pure Drp1 and MiD proteins ...... 182 5.3.2 Computational modelling of MiD proteins’ Drp1 binding loops ...... 183 5.3.3 Thermal shift assay ...... 184 5.3.4 Surface Plasmon Resonance ...... 184 5.4 Results ...... 185 5.4.1 Protein purification of human MiD49, MiD51 and Drp1 ...... 185 5.4.2 Computational analysis of Drp1 and MiD proteins ...... 187 5.4.3 Thermal Shift Assay of Drp1 and MiD proteins ...... 189 5.4.4 Optimisation of surface plasmon resonance assay ...... 194 5.5 Discussion ...... 198 5.5.1 Thermal Shift Assay of Drp1 and MiD proteins ...... 198 5.5.2 Optimisation of surface plasmon resonance assay ...... 199 5.5.3 Future work ...... 200

CHAPTER 6: Overall Discussion and future work ...... 202

6.1 Ongoing and Future work ...... 205 6.1.1 Cardiac-specific knockdown of MiD49 and MiD51 ...... 205 6.1.2 Development of small molecule inhibitors of MiD49 and MiD51 ...... 205 6.1.3 Investigating the role of MiD49 in human fibroblasts and induced Pluripotent Stem Cell-Derived Cardiomyocytes ...... 206 6.1.4 Conclusion ...... 206 CHAPTER 7: References ...... 208

ACKNOWLEDGEMENTS

My sincerest thanks go to my supervisors Professor Derek Hausenloy, Professor Derek Yellon and Dr Andrew Hall, for giving me the opportunity to undertake my PhD under their supervision. I would like to especially thank Professor Hausenloy for his support over the past 4 years and giving me the opportunity to work on various projects that have broadened my skill set. I would like to extend my sincere appreciation to Professor Yellon for his mentorship and advice during my PhD. I would like to thank Dr Andrew Hall, for his incredible dedication to the team and always making the time to provide valuable training or share his expertise. Thanks are of course extended to the whole team at the Hatter Institute. Although people come and go, I have been fortunate enough to become friends with some truly great individuals from all over the world. I am very grateful to have had the opportunity to work in an environment full of trust, support and mentorship. I would also like to thank my MSc Tutors Prof Philippa Talmud and Dr Andrew Cook for their support throughout my studies at University College London.

I would like to thank all my collaborators, for their continuous support; especially Dr Laura Osellame at Monash University, for sharing her immense knowledge of the MiD proteins and mitochondrial morphology. My particular thanks to Dr Jessica Maeve Elder for providing the preliminary data, used to create this project, and Assistant Prof Sang- Bing Ong, for his guidance in re-establishing the RT IRI model. Particular thanks to Dr Kelvin See, Dr Sauri Hernandez Resendiz, Miss Nicole Gui Zhen Tee and Miss Khairunnisa Binte Katwadi for providing their expert advice and specialised skills, in aiding to generate experimental results, presented in this thesis. I would also like to thank the research team at Duke-NUS and National Heart Centre Singapore for providing invaluable advice and training during my time in Singapore.

Finally, my thanks to my parents and my brother for their endless love and support throughout this time; this would have been undoubtedly impossible without you three, as the foundation of everything that I am today.

I was fortunate to be awarded the three-year British Heart Foundation PhD studentship and would like to thank the BHF for their funding support.

PUBLICATIONS

Research publications

Samangouei P, Crespo-Avilan GE, Cabrera-Fuentes H, Hernández-Reséndiz S, Ismail NI, Katwadi KB, Boisvert WA, Hausenloy DJ. (2018). MiD49 and MiD51: New mediators of mitochondrial fission and novel targets for cardioprotection. Cond Med. Aug;1(5):239-246.

Ong S-B, Samangouei P, Beikoghli Kalkhoran S & Hausenloy DJ. (2015). The mitochondrial permeability transition pore and its role in myocardial ischaemia-reperfusion injury. J Mol Cell Cardiol. 78:23-34.

Ong S-B, Kalkhoran SB, Hernández-Reséndiz S, Samangouei P, Ong S-G & Hausenloy DJ. (2017). Mitochondrial-Shaping Proteins in Cardiac Health and Disease – the Long and the Short of It! Cardiovascular Drugs and Therapy 31: 87-107.

Abstract

M. Elder J, Samangouei P, Burke N, Hall A, D. Osellame L, T. Ryan M & J. Hausenloy D. (2014). The novel mitochondrial fission proteins MiD49 and MiD51: new therapeutic targets for cardioprotection, vol. 100.

Oral presentations

Samangouei P., Hall A.R.,Yellon D.M., Hausenloy D.J. Targeting the Mitochondrial Fission Proteins, MiD49 and MiD51, as a therapeutic strategy for Cardioprotection. Singapore Cardiac society, 29th Annual Scientific Meeting (April 2017).

Poster abstracts

Elder J.M., Samangouei P., Burke N., Hall A.R., Hausenloy D.J.,MiD49 and MiD51 as novel therapeutic targets for cardioprotection. Mitochondrial Dynamics and Physiology (Q5), Keystone symposia. Santa Fe (USA). 2014.

Samangouei P., Hall A.R., Burke N., Yellon D.M., Hausenloy D.J. MiD49 and MiD51 as novel therapeutic targets for cardioprotection. 5th Annual UCL Cardiovascular Science Symposium, University College London, London (UK). 2016.

Samangouei P., Hall A.R., Burke N., Osellame L.D.,Ryan M.T. Targeting the mitochondrial fission proteins, MiD49 and MiD51, as a therapeutic strategy for cardioprotection. Mitochondrial Medicine: developing new treatments for mitochondrial disease. Wellcome Genome Campus, Cambridge, (UK). 2016.

FIGURES

Figure 1.3.1.1 Confocal images of mitochondrial morphology in mouse embryonic fibroblast cells ...... 31 Figure 1.4.1.1 Electron microscopy image of an adult mouse cardiomyocyte ...... 37 Figure 1.5.1.1 SMCR7L protein mitochondrial localisation ...... 40 Figure 1.5.2.1 mRNA expression of MiD (MIEF) proteins in human foetal and adult organs ...... 41 Figure 1.5.3.1 MiD depletion or overexpression promotes mitochondrial elongation ...... 43 Figure 1.5.5.1 Mitochondrial ER interaction is essential for Drp1 mediated fission . 53 Figure 3.3.7.1 Flow activated cell sorting ...... 66 Figure 3.3.10.1 Rate of MPTP opening quantification ...... 68 Figure 3.3.12.1 Confocal RT SIRI setup ...... 71 Figure 3.4.1.1 Changes in mitochondrial morphology due to the overexpression of different mitochondrial dynamic proteins ...... 73 Figure 3.4.2.1 Cellular mitochondria morphology observed in MiD transfected cells after 5h of normoxia ...... 75 Figure 3.4.2.2 Cellular mitochondria morphology observed in MiD transfected cells after SIRI ...... 76 Figure 3.4.2.3 Mitochondrial morphology of MEFs overexpressing MiD49 and MiD51 ...... 77 Figure 3.4.3.1 MPTP opening using TMRM staining ...... 78 Figure 3.4.3.2 Time taken to induce MPTP opening due to the overexpression of different mitochondrial dynamic proteins in HL-1 cells...... 79 Figure 3.4.3.3 Time taken to induce mitochondrial MPTP opening in MEF cells overexpressing MiD49 and MiD51 ...... 80 Figure 3.4.4.1 MiD49 and MiD51 overexpression protects against SIRI ...... 81 Figure 3.5.1.1 Transfection efficiency of H9c2 cells ...... 84 Figure 3.5.1.2 Transfection efficiency of cell lines used for experiments ...... 85 Figure 3.5.1.3 H9c2 growth rate and cell fusion ...... 86 Figure 3.5.2.1 Changes in mitochondrial morphology in HL-1 cells ...... 88 Figure 3.5.2.2 Mitochondrial morphology of fixed HL-1cells ...... 89 Figure 3.5.2.3 Mitochondrial morphology or fixed H9c2 cells ...... 90 Figure 3.5.2.4 Changes in mitochondrial morphology in H9c2 cells ...... 91

Figure 3.5.3.1 MiD49 and MiD51 KD delays MPTP opening in HL-1 cells ...... 93 Figure 3.5.3.2 MiD49 and MiD51 KD delayed MPTP opening, at a similar level as Drp1k38A overexpression...... 94 Figure 3.5.3.3 Mitochondrial permeability transition pore opening in H9c2 cells ..... 96 Figure 3.5.4.1 MiD49 and MiD51 KD protects against SIRI ...... 97 Figure 3.5.4.2 Characterisation of RT SIRI model ...... 100 Figure 3.5.4.3 Mitochondrial morphology during SIRI of HL-1 cells after MiD49 and MiD51 KD ...... 102 Figure 3.5.4.4 Mitochondrial morphology during SIRI of H9c2 cells after MiD49 and MiD51 KD ...... 104 Figure 3.5.4.5 Mitochondrial morphology of H9c2 VC cells during SIRI ...... 105 Figure 3.5.4.6 Mitochondrial morphology of MiD49 and MiD51 KD H9c2 cell during SIRI ...... 106 Figure 3.5.5.1 RT chamber conditions during SIRI ...... 108 Figure 3.5.5.2 Changes in mitochondrial calcium during SIRI ...... 109 Figure 4.3.3.1 MiD49 KO colony PCR steps ...... 125 Figure 4.3.3.2 Representative genotyping result of the MiD49 KO mice ...... 126 Figure 4.3.5.1 Echocardiography imaging modes ...... 131 Figure 4.3.6.1 Representative images of non-recovery murine acute myocardial Ischaemia-reperfusion in vivo surgery model ...... 135 Figure 4.4.1.1 Cardiac mitochondrial morphology of MiD49 KO mice ...... 137 Figure 4.4.2.1 Representative echocardiographic phenotyping of MiD49 KO mice ...... 138 Figure 4.4.2.2 LV anatomical measurements of MiD49 and WT hearts ...... 139 Figure 4.4.2.3 Echocardiography Cardiac phenotyping of WT and MiD49 KO mice ...... 141 Figure 4.4.2.4 LV fractional shortening and LV ejection fraction measurements of WT and MiD49 KO mice ...... 142 Figure 4.4.3.1 MI in MiD49 KO mice ...... 143 Figure 4.5.2.1 Recombinant AAV production methods ...... 148 Figure 4.5.3.1 Cardiac-specific rAAV9 vector ...... 149 Figure 4.6.1.1 MiD pCAG-mir-RNAi-mCherry vector production ...... 151 Figure 4.6.3.1 rAAV9 plasmids used in the helper-free AAV production system ...155 Figure 4.6.3.2 Plasmid maps of the vectors required for MiD rAAV9 production ...157 Figure 4.6.3.3 E.coli transduction and plasmid production protocol ...... 159 Figure 4.6.3.4 Optiprep iodixanol density gradient column ...... 163 Figure 4.6.3.5 Viral titre calculation formula ...... 164 12

Figure 4.7.2.1 Transfection efficiency of MiD ...... 167 Figure 4.7.2.2 Representative image of transfection efficiency of HEK293 cells during rAAV9 production ...... 168 Figure 5.1.1.1 Representative schematic of protein thermal shift assay ...... 180 Figure 5.1.1.2 Representative schematic surface plasmon resonance ...... 181 Figure 5.4.1.1 Purified MiD49, MiD51 and Drp1 proteins ...... 186 Figure 5.4.2.1 MiD49 and MiD51 amino acid sequence ...... 188 Figure 5.4.3.1 Thermal shift assay of purified MiD49 and MiD51 in the presence of nucleotides ...... 191 Figure 5.4.3.2 Thermal shift assay of purified Drp1 in the presence of small MiD peptides ...... 192 Figure 5.4.3.3 Thermal shift assay of purified Drp1 in the presence of GTP and small MiD peptides ...... 193 Figure 5.4.3.4 Thermal shift assay of purified Drp1 in the presence of Mdivi-1 .....194 Figure 5.4.4.1 Drp1 surface plasmon resonance assay optimisation...... 196 Figure 5.4.4.2 MiD51 surface plasmon resonance assay optimisation ...... 197

TABLES

Table 1.1.1.1 World Health Organisation statistics and future predictions of the top cause of mortality worldwide ...... 19 Table 4.3.3.1 Property of MiD49 Primers ...... 123 Table 4.3.3.2 MiD49 WT and KO PCR reaction mix ...... 124 Table 4.3.4.1 Electron microscopy sample dehydration ...... 127 Table 4.4.2.1 Echocardiography cardiac phenotyping summary ...... 140 Table 4.6.1.1 MiD49 and MiD51 targetting oligonucleotide sequences ...... 152 Table 4.6.3.1 Helper-free transfection mixture for rAAV9 production in HEK293 cells ...... 161 Table 4.6.3.2 AAV9 virus titre qPCR reaction mix ...... 164 Table 4.7.3.1 volume of rAAV9 constructs used for in vivo injection into the thoracic cavity ...... 169 Table 5.4.2.1 MiD49 and MiD51 small peptide sequences ...... 187 Table 5.4.3.1 Thermal shift assay of human MiD proteins in the presence of nucleotides ...... 189

ABBREVIATIONS

2D imaging Two-dimensional echocardiography imaging aa Amino acids AAR Area at risk AAV Adeno-associated viruses AAV9 Adeno-associated virus serotype-9 ADP Adenosine diphosphate Akt Protein kinase B (PKB) AMI Acute myocardial infarction ATP Adenotriphosphate BLAST Basic Local Alignment Search Tool bp Base pairs BPM Beats per minute BSA Bovine serum albumin cAMP Cyclic AMP CCCP Carbonyl cyanide m-chlorophenyl hydrazine CDK Cyclin-dependent kinase CO Cardiac output Ct Cycle threshold CVD Coronary artery disease Cyc A Cyclosporin A Cyp D Cyclophilin D Cyc c Cytochrome c ddH2O Double distilled water Drp1 Dynamin-related protein 1 E.coli Escherichia coli ECC Excitation contraction coupling ECM Extracellular matrix EDD End-diastolic diameter EDV End-diastolic volume eNOS Endothelial nitric oxide synthase ER Endoplasmic reticulum ESD End-systolic diameter ESV End-systolic volume ETC Electron transport chain FACS Fluorescence-activated cell sorting

FADH2 Flavin adenine dinucleotide FBS Foetal bovine serum FS Fractional shortening FSC Forward and side scatter GDP Guanosine diphosphate GTP Guanosine triphosphate HEK293 Human embryonic kidney 293 15

HEPES 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid Het Heterogeneous HF Heart failure His Histidine HR Heart rate ICAM-1 Intercellular adhesion molecule-1 IFM Interfibrillar mitochondria IHD Ischaemic heart disease IMAC Immobilized metal affinity chromatography IMM Inner mitochondrial membrane iPSC-CMs Induced pluripotent stem cell-derived cardiomyocytes IPTG Isopropyl β-D-1-thiogalactopyranoside IRI ischaemia-reperfusion injury IS Infarct size ITR Inverted terminal repeats

KD Dissociation constant KD Knockdown KO Knockout LAD Left descending coronary artery LB broth Luria-Bertani/ Lysogeny broth LDL Low-density lipoproteins LV Left ventricle MARCH5 Membrane-associated ring-CH-type finger 5 MCP-1 Monocyte chemotactic protein-1 (MCP-1) MCU Mitochondrial calcium uniporter MEFs Mouse embryonic fibroblasts Mfn1 Mitofusin 1 Mfn2 Mitofusin 2 MI Myocardial infarction MiD49 Mitochondrial dynamics proteins of 49 kDa MiD51 Mitochondrial dynamics proteins of 51 kDa MiDs Mitochondrial dynamics proteins MiD49 and MiD51 MIEF1 Mitochondrial elongation factor 1 MIEF2 Mitochondrial elongation factor 2 MM Mitochondria matrix M-mode Motion mode MPTP mitochondrial permeability transition pore mtBFP Mitochondrial blue fluorescent protein mtDNA Mitochondrial DNA mtGFP Mitochondrial green fluorescent protein mtRFP Mitochondrial red fluorescent protein MΦs Macrophages NADH Reduced Nicotinamide adenine dinucleotide NCLX Sodium Calcium exchanger NEO Neomycin gene sequence 16

NO Nitric oxide NT Non-transfected Ntase Nucleotide transferase Oligo Oligonucleotide OMM Outer mitochondrial membrane Opa1 Optic atrophy protein 1 ORF Open reading frame oxLDL Oxidised LDL OXPHOS Oxidative Phosphorylation PBS Phosphate-buffered saline PCR Polymerase chain reaction PEG Polyethylene glycol PI Propidium iodide pI Isoelectric point PiC Mitochondrial phosphate carrier Pim-1 Proto-oncogene serine/threonine-protein kinase PKA Protein Kinase A PLAX Parasternal long-axis mid-level PNM Perinuclear mitochondria PPP Protein purification platform PUMA p53 upregulated modulator of apoptosis qPCR Quantitative polymerase chain reaction (qPCR) rAAV Recombinant adeno associated virus RI Reperfusion injury RMP Revolutions per minute ROS Reactive oxygen species (ROS) RT Room temperature RU resonance units S Serine SAX Parasternal short axis view SI Simulated Ischaemia SIMH Stress-induced mitochondrial hyperfusion SPR Surface plasmon resonance SR Sarcoplasmic reticulum SSC Side scatter SSM Sub-sarcolemmal mitochondria STS Staurosporine SV Stroke volume TAE Tris-acetate-EDTA buffer TB Terrific Broth TCA cycle Tricarboxylic acid cycle Tm Melting temperature TSA Thermal shift assay TTC Triphenyl tetrazolium chloride solution VC Empty vector control 17

VCAM-1 Vascular cell adhesion molecule-1 VDAC Voltage dependent anion channel Vg Viral genome WT Wild-type ΔΨm Mitochondrial membrane potential/ proton gradient

CHAPTER 1: General introduction

1.1 Epidemiology of ischaemic heart disease

Advances in modern medicine have significantly increased our life expectancy, naturally making us more predisposed to developing chronic and age-related conditions. Ischaemic heart disease (IHD) is the leading cause of morbidity and mortality worldwide, responsible for over 7.5million deaths in 2015 (World Health Organisation statistics), (Table 1.1.1.1). Future projections indicate that IHD will remain the number one cause of morbidity and mortality for many more years to come (Mathers & Loncar, 2006). Although IHD is typically considered to be a health problem predominantly in high-income countries, incident rates are on the decline (Levi et al., 2009; Smolina et al., 2012). However, the adoptions of the western diet and lifestyle in middle to low-income countries have resulted in an alarming increase in the incidence of IHD in these countries. Many of these developing countries now have a significantly higher prevalence of morbidity and mortality due to IHD, than many developed countries, known to have high incidence rates (Finegold et al., 2013).

2015 2030

Deaths Deaths Deaths % Deaths % Rank Cause per Rank Cause per (000s) Deaths (000s) Deaths 100,000 100,000

Ischaemic heart Ischaemic heart 1 7594 13.2 105 1 9245 13.2 112 disease disease 2 Stroke 6700 11.7 92 2 Stroke 8578 12.2 104 Lower respiratory Chronic obstructive 3 3223 5.6 44 3 4568 6.5 55 infections pulmonary disease Chronic obstructive Lower respiratory 4 3217 5.6 44 4 3535 5 43 pulmonary disease infections 5 Diarrhoeal diseases 1808 3.2 25 5 Diabetes mellitus 2464 3.5 30

Table 1.1.1.1 World Health Organisation statistics and future predictions of the top cause of mortality worldwide IHD is currently the number one cause of morbidity and mortality in the world. It is a significant health problem which is predicted to dominate death rates for many more years (Global health estimates summary, World Health Organisation, July 2013).

The major clinical manifestations of IHD are due to the reduction in blood flow to myocardial tissue, resulting from the narrowing of coronary vessels due to coronary atherosclerosis, termed coronary artery disease (CAD) (Ross, 1999). A significant consequence of CAD is an acute myocardial infarction (AMI), as prolonged periods

19 of myocardial ischaemia following coronary atherosclerotic plaque rupture and thrombotic occlusion cause cardiomyocyte death (Thygesen et al., 2007).

1.1.1 Atherosclerosis

Atherosclerosis is a chronic inflammatory disease, that can be initiated as early as childhood, and manifests clinically over time (Ross, 1999). The process begins with endothelial dysfunction (Hahn et al., 2007; Sitia et al., 2010).This single-cell-thick layer, lining the vasculature lumen, acts as more than just a barrier between blood and body tissue. Their response to chemical and hormonal signals, as well as physical stress, is essential for vascular homeostasis. Nitric oxide (NO) production by endothelial nitric oxide synthase (eNOS) in response to shear stress is vital, allowing vasodilation, inhibiting inflammation, and preventing smooth muscle proliferation (Rubanyi et al., 1986; Tousoulis et al., 2012).

Endothelial dysfunction results in the infiltration of low-density lipoproteins (LDL), into the intimal layer of the arterial wall. This movement is bidirectional, but high levels of plasma LDL positively correlates with an increased chance of development and progression of atherosclerosis (Castelli et al., 1992). Accumulated LDL in the intima is oxidised (oxLDL) and stimulates the expression of pro-inflammatory chemokines at the site of damage. Monocyte chemotactic protein-1 (MCP-1) expression on the lumen wall, results in monocyte infiltration into the artery wall (Xiao et al., 1999), where they then differentiate to macrophages (MΦs) (Frostegard et al., 1990). As part of a natural immune response to remove the cholesterol deposits, macrophages engulf the oxLDL and release reactive oxygen species (ROS). This process causes further tissue damage, and the high intake of oxLDL differentiates macrophages into foam cells (Tsukamoto et al., 2002; Yvan-Charvet et al., 2007).

The release of extracellular matrix (ECM) molecules and adhesion molecules, such as vascular cell adhesion molecule-1 (VCAM-1), intercellular adhesion molecule-1 (ICAM-1) and E-selectin trap the cells within the arterial wall (Park et al., 2009). MΦs and foam cells also contribute to the development of an unstable plaque by secreting matrix metalloproteinases, and growth factors which cause smooth muscle cell proliferation (Ross & Glomset, 1973).

The growing plaque starts to protrude into the coronary lumen, reducing the flow of blood to the heart. Atherosclerotic plaques with thick fibrous caps remain stable and 20 are the primary cause of angina, (Haverkate et al., 1997). Unstable plaques are vulnerable to rupturing, causing thrombotic occlusions within the coronary vessels. Subsequently, the myocardium is starved of nutrients and oxygen, triggering a cascade of events, which can lead to an acute MI (Kaandorp et al., 2005). Changes in cardiomyocyte metabolism and function following coronary occlusion are further explained in section 1.2.

1.1.2 Myocardial infarction

Myocardial infarction is an acute event which occurs as a result of a lifelong chronic condition, and its outcome has the most significant impact on the patient (Thygesen et al., 2007). As previously mentioned, this is often caused by the rupture of atherosclerotic plaques, resulting in acute coronary occlusion. MI results in the permanent loss of cardiomyocytes due to cardiac ischaemia, in which there is an insufficient supply of blood to ensure the viability of cells within the affected region (Thygesen et al., 2007). To minimise the risk of cardiac remodelling and heart failure, as a result of permanent cardiomyocyte loss, it is essential to salvage viable cells within the ischaemic zone, by preventing prolonged periods of ischaemia (Reimer et al., 1977; Bolli et al., 1988).

Reperfusion is essential to salvage cells within the ischaemic region. Complete cell recovery is possible following short periods of ischaemia; better clinical outcomes are observed when blood flow is restored soon after ischaemia (Reimer et al., 1977; Bolli et al., 1988; Kloner & Jennings, 2001; Yellon & Hausenloy, 2007). Although reperfusion is essential for myocardial salvage, it is often referred to as a double- edged sword, given that it can contribute up to 50% of the final infarct size (Kloner & Jennings, 2001; Yellon & Hausenloy, 2007). This phenomenon is referred to as ischaemia-reperfusion injury or lethal reperfusion injury (see section 1.2.2 and 1.2.3).

1.1.2.1 Current clinical intervention Currently, the most effective therapy for reducing MI size and preventing heart failure following an acute MI is myocardial reperfusion by primary percutaneous coronary intervention (PPCI) (Nallamothu et al., 2007). This procedure has resulted in remarkable improvements in survival rates and recovery of patients suffering from an acute MI (Smolina et al., 2012). However, the benefits are time-dependent and patients that experience ischaemic periods longer than two hours tend to have lower rates of survival, or modest recovery of left ventricular function (Brodie et al., 1998). 21

1.1.2.2 The need for novel cardioprotective therapies Despite improvements in disease prevention, treatment and rehabilitation, AMI incidence rates are still very high. Medical funding and care are being exhausted by trying to cope with the devastating long-term outcomes of MI, as survival after an event does not necessarily guarantee a prolonged or good quality of life (Bhatnagar et al., 2015). A common prognosis for MI patients is heart failure, approximately 25% of cases, which significantly increases the risk of adverse outcomes and earlier death (Minicucci et al., 2011). There is an urgent need to discover novel methods of cardioprotection, to lower AMI related morbidity and mortality. Currently, there are no clinical therapies to protect against acute myocardial IRI. With more control over the time and method of reperfusion, targeting this stage is the more feasible approach to achieving cardioprotection, as the onset of AMI is unpredictable. It is therefore essential to identify new pharmaceutical and non-pharmaceutical therapeutic strategies that can be administrated during reperfusion; as the reduction in the final infarct size can significantly reduce the manifestation of future cardiovascular complications such as heart failure, and death (Smolina et al., 2012).

1.1.2.3 Mitochondria are key targets of cardioprotection Mitochondria are often described as “the powerhouse of cells”. They are one of the distinguishing features of eukaryotic cells, and their presence allowed the formation of complex multicellular organisms. Due to the high energy demands of cardiomyocytes, mitochondria occupy up to 35% of the cells’ volume and produce ≈90% of the adenosine triphosphate (ATP) required for normal cardiac function. Preserving mitochondrial health is essential for preventing cell damage, especially in organs with high metabolic rates, as dysfunctional mitochondria are responsible for the production of most of the reactive oxygen species (ROS) within cells. Mitochondria are also involved in mediating and regulating cellular activity during physiological and cellular stress conditions (Gustafsson & Gottlieb, 2008; Chan, 2012).

It is now well recognised that mitochondria are central in determining cell survival at the time of myocardial reperfusion (Murphy & Steenbergen, 2008). The cascade of events which leads to mitochondrial damage, dysfunction and the opening of the mitochondrial permeability transition pore (MPTP) during acute myocardial IRI (section 1.2.3.1), is now a well-recognised trigger of the intrinsic apoptotic and necrotic cell death pathways (Braunwald & Kloner, 1985; Crompton et al., 1987; Griffiths & Halestrap, 1995). This makes mitochondria ideal targets for 22 cardioprotection. To better understand the role of mitochondrial function in cardiomyocytes and how they can mediate cell death during IRI, we first need to understand the conditions required for ATP production and which factors affect its generation.

1.2 Cardiac metabolism

1.2.1 Oxidative phosphorylation

In adult hearts, over 90% of ATP production in cardiomyocytes is from aerobic mitochondrial oxidative phosphorylation (OXPHOS) by the electron transport chain (ETC) (Javadov et al., 2009). The remaining 10% of ATP is produced by glycolysis and the tricarboxylic acid (TCA) cycle. Under physiological conditions, adult cardiac cells predominantly utilise fatty acids (50-70%) and to a lesser extent, glucose (≈ 30%) for ATP production (Bartelds et al., 2000; Lopaschuk et al., 2010). There can be metabolic flexibility to meet ATP demands during periods of high activity, low oxygen, or during pathologies such as heart failure (Randle et al., 1963; Taegtmeyer et al., 2010). The use of other substrates, such as amino acids is very low during normal physiological conditions (Jeffrey et al., 1995; Wentz et al., 2010).

Acetyl CoA is formed from fatty acid β-oxidation, or pyruvate oxidation from glycolysis (Kresge et al., 2005; Lopaschuk et al., 2010). Acetyl CoA is metabolised in the TCA cycle, in mitochondria, to produce three NADH, one FADH2 and one GTP per cycle (Kornberg, 2000; Lopaschuk et al., 2010). The NADH and FADH2 molecules, formed during glycolysis, β-oxidation and the TCA cycle, act as electron donors in the ETC, at the inner mitochondrial membrane (IMM) (Akram, 2014).

The transfer of electrons along the ETC, starting with NADH electron donation to complex I and FADH2 electron donation to complex II, is required to create the proton gradient essential for ATP production (Mitchell, 1961). Oxygen is the final electron acceptor at the ETC, essential for oxidative phosphorylation and its consumption is matched with ATP production. The transfer of electrons along the ETC complexes is coupled with the movement of protons (H+), from the mitochondrial matrix (MM) into the mitochondrial intermembrane space (IMS) by complexes I, III and IV (Mitchell, 1961). This creates a mitochondrial membrane potential (Δψm) of ≈ -150 to -180mV (Brand M & Nicholls D 2011; Nickel et al., 2013). The membrane electrochemical gradient and the release of energy from this process are essential for ATP synthase phosphorylation of adenosine diphosphate 23

(ADP), forming ATP (Reid et al., 1966; Junge et al., 2009). The movement of protons down the electrochemical gradient, through ATP synthase, causes the clockwise rotation of ATP synthase’s central stalk. Each full rotation in a clockwise direction produces 3 ATP molecules, requiring the movement of 2.7 protons per ATP (Watt et al., 2010). Cytosolic changes in ADP, inorganic phosphate (Pi) and Ca2+ ions have a regulatory effect on ATP production (Cortassa, 2009).

1.2.1.1 ATP, ADP and Pi Increased workload leads to rapid consumption of ATP, producing ADP and Pi. Fast turnover is required to produce more ATP and meet the energy demands of myocardial contraction. Increased ADP and Pi availability upregulates the flow of electrons through the ETC, and ATP synthase activity, to increase the rate of ATP production (Cortassa, 2009). Previously it was believed that this was enough to increase ATP production (Chance & Williams, 1955). However, many studies have now indicated that this increase is also linked to mitochondrial calcium signalling (Maack & O’Rourke, 2007).

1.2.1.2 Calcium and ATP production Mitochondrial ATP production is continually fluctuating in the heart, and this is linked to the energy demand of cardiomyocytes in response to changes in physical activity (Gertz et al., 1988). Mitochondrial productivity is linked to cardiomyocyte excitation- contraction coupling (ECC). The cardiac action potential depolarises the cardiomyocyte membrane, leading to the entry of extracellular Ca2+ through activated sarcolemmal voltage-dependent L-type channels (Bers, 2002). The increase in calcium concentration causes the opening of ryanodine receptors, through a process of calcium induced calcium release, resulting in the release of Ca2+ from the sarcoplasmic reticulum (SR) via ryanodine receptors (Bers, 2002; Bround et al., 2012). Within cardiomyocytes, mitochondria are in close proximity to sarcoplasmic reticulum (SR). The release of Ca2+ from the SR leads to an increase in mitochondrial Ca2+ uptake primarily by the mitochondrial Ca2+ uniporter (MCU) (Pacher et al., 2000; Kirichok et al., 2004; Baughman et al., 2011).

Basal MM calcium levels are restored by the Na+/Ca2+ exchanger (NCLX) (Palty et al., 2010), but the export occurs at a slower rate than the uptake by the MCU in the heart.

Mitochondria are involved in cellular calcium homeostasis and signalling. An increase in mitochondrial Ca2+ uptake, as a result of SR Ca2+ release, promotes the activity of the three rate limiting dehydrogenases of the TCA cycle, resulting in increased production of NADH (Randle et al., 1974; Wan et al., 1989; Traaseth et al., 2004). This is important for promoting ATP production to meet the energy demands of the cell (Denton, 2009). A rise in NADH and FADH2 production leads to an increase in electron donation to the ETC complexes. The increased supply of electrons to the ETC complexes allows for a higher movement of protons across the IMM, into the intermembrane space, resulting in an increase of the ΔΨm. This promotes an increase in ATP production by the ATP synthase (Mitchell, 1961). Elevated sarcoplasmic Na+ levels result in a reduction of mitochondrial Ca2+, due to increased NCLX activity, creating an energy deficit as a result of reduced NADH production (Maack et al., 2006; Celsi et al., 2009).

1.2.2 Cardiac Ischaemia

As previously mentioned, cardiac ischaemia is often caused by acute coronary occlusion during AMI (section 1.1.2). The word ischaemia originates from the Greek words “iskhein” to hold back or suppression, and “haima” means blood. Cells become ischaemic, as they are deprived of oxygen and nutrients essential for cardiomyocyte function and survival. Although the high volume of mitochondria within cardiomyocytes allows a high production of ATP, the cell pool of ATP is only sufficient to provide energy for a few minutes (Reimer et al., 1981; Mootha et al., 1997). Prolonged periods of ischaemia (>15 minutes), results in irreversible ischaemic damage and eventually necrotic cell death due to ATP depletion (Ferris & Friesen, 1979; Bolli et al., 1988). As ischaemia progresses, irreversible cell death begins in the subendocardium and progresses towards the epicardium, termed the ‘wavefront’ phenomenon (Reimer et al., 1977).

As explained previously, oxygen is the final electron acceptor in the ETC, required for aerobic respiration. ETC activity halts when cardiomyocytes are in an ischaemic state, as the cell can no longer produce ATP by aerobic respiration (Mitchell, 1961). Mitochondria undergo further dysfunction, as protein complexes of the ETC are unable to pump proton ions from the MM into the intermembrane space. This subsequently leads to the loss of mitochondrial membrane potential (Mitchell, 1961; Kloner & Jennings, 2001). The loss of the proton gradient causes further ATP depletion, as ATP synthase activity reverses, hydrolysing ATP to ADP, AMP and Pi

(Rouslin et al., 1990; Noll et al., 1992; Grover et al., 2004). Following the loss of the mitochondrial membrane potential, IF1-mediated inhibition of ATP synthase’s reversed activity during myocardial ischaemia helps to conserve ATP levels in the heart (Rouslin & Broge, 1996; Campanella et al., 2008).

Without ATP production, energy cannot be provided for essential housekeeping processes as well as excitation-contraction coupling (Bers, 2002). Ischaemic cardiomyocytes attempt to compensate for the lack of ATP, by anaerobic respiration (Reimer et al., 1981; Noll et al., 1992). This less efficient method of ATP production causes further cellular dysfunction due to acidosis. The build-up of lactic acid can cause the cellular pH to drop to as low as 6, within only 15 minutes (Murphy & Steenbergen, 2008). The accumulation of proton ions in the sarcoplasm induces further ATP depletion and cellular dysfunction, by a cascade of abnormal ionic regulatory activities (Murphy & Steenbergen, 2008).

The natural cellular mechanism to restore physiological pH is to pump out the excess H+. This occurs via the sarcolemma Na+/H+ exchanger (NHE) which pumps in one Na+ for every H+ removed from the sarcoplasm, resulting in Na+ overload (Pike et al., 1990). Consequently, the increase in sarcolemmal Na+ leads to Ca2+ overload, as for every three Na+ ions removed from the cell, there is an influx of one Ca2+ ion into the cell, via sarcolemmal NCLX (Anderson et al., 1991). Ca2+ overload causes a faster depletion of the remaining ATP stores within cardiomyocytes, as it results in an increase in cardiac contraction and the activity of calcium ATPase pumps (Steenbergen et al., 1990; Ma et al., 2004).

Cardiomyocytes have very efficient intrinsic systems to remove Ca2+ from the sarcoplasm, including the plasma membrane calcium pump ATPase (PMCA), as well as the sarcoplasmic/ endoplasmic reticulum calcium pump ATPase (SERCA2a) (Ma et al., 2004; Brini & Carafoli, 2011). Their function is essential to allow relaxation of the cell after sarcomere contraction, to prevent myofilament hypercontraction in the presence of Ca2+ and ATP. Unfortunately, these pumps require ATP to function, which is rapidly depleted after the onset of ischemia. Therefore, cells which survive a prolonged period of ischaemia are left in an acidic state, overloaded with Pi and Ca2+.

Experimental studies have shown that the inhibition of ion exchangers, such as the NHE, can delay the period of irreversible damage during ischaemia (Murphy et al.,

1991; Gumina et al., 1999). Unfortunately, in a clinical setting, such pharmaceutical treatments have shown to be limited in their cardioprotective abilities against IRI. Many trials have been unsuccessful, mainly due to that fact that these inhibitors must be present at the onset of ischaemia, given that these ion exchange events occur at the early stages of ischaemia (Murphy et al., 1991; Avkiran & Marber, 2002).

1.2.3 Cardiac Reperfusion

Reperfusion is the restoration of blood flow to the ischaemic tissue. Reperfusion is essential to salvage cells within the ischaemic region that have not undergone necrosis or are not yet damaged beyond repair. Reperfusion injury (RI) is the death of cells, viable post ischaemia, as a result of cellular stress caused by reperfusion. Reducing ischaemic time causes smaller infarcts and achieves better recovery (Reimer et al., 1977; Brodie et al., 1998). Unfortunately, the re-flow of blood can itself be the cause for up to 50% of the final infarct size, due to the following form of IRI: myocardial stunning, no-reflow phenomenon (Forman et al., 1989), reperfusion arrhythmias and lethal reperfusion injury (Bolli et al., 1988; Kloner & Jennings, 2001).

With the return of blood flow to cardiac tissue, lactic acid is washed away within a few minutes. This helps to restore basal cellular pH and the normal function of ion exchangers, which gradually restores basal ion levels within cardiomyocytes. The presence of oxygen allows ATP production by the ETC. While the sarcoplasm is still overloaded with calcium, ATP production often leads to hypercontraction of the myofilaments (Yellon & Hausenloy, 2007).

Restoration of ETC activity will re-establish the Δψm, by pumping H+ ions into the IMM space, which is required for coupled oxidative phosphorylation (Mitchell, 1961). The re-established proton gradient restores regular ATP synthase activity (Watt et al., 2010). Unfortunately, this does not occur instantly. During the recovery period, the ETC is dysfunctional and inefficient, producing high levels of ROS (Bolli et al., 1988). A large proportion of the ROS produced upon reperfusion occurs as a result of accumulated succinate from prolonged periods of ischaemia. The rapid reoxidation of succinate to fumarate, by succinate dehydrogenase, within the first few minutes of reperfusion leads to the reversal of the ETC activity through complex I, producing NADH and ROS (Pryde & Hirst, 2011; Chouchani et al., 2014). The restoration of the Δψm, following ischaemia, has other consequences. Although 27 restoration of the Δψm is essential for coupled OXPHOS, it can be detrimental to the cells, whilst cellular Ca2+ levels are still very high. This negative membrane potential leads to the influx of Ca2+ ions into the MM, through the MCU (Silverman & Stern, 1994; Kirichok et al., 2004). As the IMM is impermeable to ions, the only way Ca2+ ions can be removed is by the mitochondrial NCLX (Palty et al., 2010). As the rate of Ca2+ influx through the MCU occurs at a much faster rate than Ca2+ efflux (Pacher et al., 2002), the matrix becomes overloaded with Ca2+ ions; as well as already having high ROS and Pi levels (Silverman & Stern, 1994; Pacher et al., 2000; Palty et al., 2010; Kwong et al., 2015). All these events increase the chance of apoptotic signalling by mitochondrial dysfunction upon reperfusion.

Cytochrome c (Cyc c) is a vital component of the ETC and abundant within cristae folds. Cyc c as well as other mitochondrial proteins, such as Smac/Diablo, are key mediators of intrinsic apoptotic cell death. Their release into the sarcoplasm, from the mitochondrial intermembrane space, causes caspase-9 activation and consequently triggers the caspase-mediated apoptotic pathway (Liu et al., 1996; Zou et al., 1999; Du et al., 2000; Li et al., 2000; Elmore, 2007). The oxidative burst, which occurs while mitochondrial ETC activity is recovering is detrimental for cells; as it affects mitochondrial and SR membrane integrity (Forman et al., 1989), as well as potentially causing DNA damage (Bolli et al., 1988). Nuclear DNA damage can occur due to the close proximity of mitochondria, after perinuclear clustering which takes place during hypoxic conditions (Al-Mehdi et al., 2012). DNA damage leads to Bax and/or Bak aggregates at the outer mitochondrial membrane (OMM), causing its permeabilisation, and the release of these apoptotic factors (Zou et al., 1999; Du et al., 2000; Hengartner, 2000; Kroemer et al., 2007). Cyc c is predominantly encapsulated within cristae folds, but mitochondrial ultrastructure damage, such as Opa1 proteolysis at cristae junctions (see section 1.3.1.2), can result in the enhanced release of Cyc c from the mitochondria (Frezza et al., 2006; Xiao et al., 2014).

1.2.3.1 The mitochondrial permeability transition pore and reperfusion injury MPTP formation and opening, is now well recognised as one of the main mediators of apoptotic and necrotic cell death, upon reperfusion (Crompton et al., 1988; Duchen et al., 1993; Griffiths & Halestrap, 1995; Hausenloy & Yellon, 2003). The MPTP is a non-selective mitochondrial channel, that causes rapid changes in mitochondrial membrane permeability in response to certain conditions (Halestrap, 28

2006). The full components of the pore structure remain to be established; however, studies based on inhibiting the MPTP opening have helped to identify potential proteins within its structure.

Evidence from knockout studies and pharmaceutical targeting of cyclophilin-D (CypD), using the immunosuppressant drug cyclosporine-A (CsA), have provided strong evidence that support the role of CypD as a regulator of MPTP opening, despite not being a component of the pore (Duchen et al., 1993; Baines et al., 2005; Schinzel et al., 2005). Other proposed candidates included the voltage-dependent anion channel (VDAC) and adenine nucleotide translocase (ANT). However, VDAC and the ANT are no longer considered to be essential components as their knockout does not inhibit MPTP opening (Kokoszka et al., 2004; Baines et al., 2007). Other suggested pore forming components include the mitochondrial phosphate carrier

(PiC), and F1F0 ATP synthase (Baines et al., 2007; Leung et al., 2008; Bonora et al.,

2013). However, knockout of F1F0 ATP synthase components the does not have a significant effect on MPTP formation (He et al., 2017).

High levels of MM Ca2+, ROS and Pi at reperfusion increase mitochondrial susceptibility to MPTP opening, which results in the collapse of the Δψm (Crompton et al., 1987; Crompton et al., 1988; Halestrap et al., 1997; Ruiz-Meana et al., 2006). This once again causes ATP depletion due to inhibition of coupled OXPHOS, and the reversed activity of ATP synthase (Crompton et al., 1987; Rouslin et al., 1990). Now that the once impermeable IMM has been compromised, non-selective diffusion of molecules smaller than 1.5kDa will occur between the contents of the mitochondrial matrix and the sarcoplasm (Halestrap, 2010). Due to the higher MM concentration of ions, solutes and water molecules flow into the mitochondria, causing it to swell. The IMM is more resistant to mitochondrial swelling, as the cristae structure can unfold. However, the OMM can only withstand a sufficient amount of swelling before it ruptures, releasing apoptotic factors into the cytosol (reviewed by Hausenloy & Yellon, 2003).

Many have debated the existence of reperfusion injury, and believe that the cell death which occurs upon reperfusion is the continuation of cell damage which has occurred during ischaemia. Despite this, an increasing level of scientific evidence shows that interventions administrated upon reperfusion can reduce final infarct size, providing indirect evidence for the existence of reperfusion injury as a determinant of cell death (Hausenloy & Yellon, 2013). Although the actual molecular

29 structure of the MPTP remains elusive, its inhibition at the early stages of reperfusion, using CsA or sanglifehrin-A (SFA) has been shown to be cardioprotective in cellular and animal models of MI (Crompton et al., 1988; Noll et al., 1992; Duchen et al., 1993; Clarke et al., 2002; Hausenloy et al., 2003). Unfortunately, inhibition of MPTP opening has not proven to be successful in the clinical setting. Administration of CsA, prior to PPCI has not been shown to improve patients’ recovery post-AMI (Cung et al., 2015; Ottani et al., 2016). Clinical studies in the setting of PPCI have many limitations as it is harder to apply the intervention in such a time-sensitive setting. There are also various confounders such as the presence of comorbidities, the expertise of the practitioners and their preference to use other pharmacological agents during the procedure which may also be cardioprotective (Roubille et al., 2012). These findings indicate that although the MPTP is a crucial target for cardioprotection, solely inhibiting MPTP opening may not be an effective method of achieving cardioprotection in the clinical setting. Instead, events which take place within the first few minutes of reperfusion should be targeted to minimise mitochondrial dysfunction, to prevent events leading to intrinsic apoptotic signalling or MPTP formation and opening.

1.3 Mitochondrial Dynamics and Cardioprotection

Mitochondria are dynamic organelles that can be transported to different regions of a cell or undergo morphological changes by two opposing processes, mitochondrial fusion and fission (Hollenbeck & Saxton, 2005; Frazier et al., 2006). Changes in mitochondrial morphology are a vital part of maintaining mitochondrial health, DNA stability, energy production, Ca²+ homeostasis, cellular division, and differentiation (Bach et al., 2003; Frazier et al., 2006; Chan, 2012). Changes in morphology can occur due to the presence of evolutionary conserved mitochondrial dynamic proteins. Defects in the fusion/fission machinery are often embryonically lethal in mice, and are associated with human diseases; such as Parkinson's and Huntington's disease (Waterham et al., 2007; Chen & Chan, 2009). Genetic or pharmaceutical inhibition of mitochondrial fission has been shown to be cardioprotective against IRI, with beneficial outcomes observed in cellular and animal models of AMI (Ong et al., 2010; Wang et al., 2011; Disatnik et al., 2013).

1.3.1 Mitochondrial Fusion

Mitochondrial fusion (Figure 1.3.1.1A) is mediated by the fusion proteins Mitofusin 1 (Mfn1), Mitofusin 2 (Mfn2) and Optic atrophy type 1 (Opa1). The mixing of 30 mitochondrial contents is essential for signal transmission, maintenance of the mitochondrial membrane potential and gaining mitochondrial DNA (mtDNA) stability (Skulachev, 2001; Chan, 2012). Fusion is vital for optimising mitochondrial respiration. ATP production is promoted by mitochondrial fusion, which helps to maximise the mitochondrial oxidative capacity, in response to increased energy demands (Bach et al., 2003; Ishihara et al., 2003; Chen et al., 2005; Tondera et al., 2009). Stress-induced mitochondrial hyperfusion (SIMH) which improves ATP production and reduces autophagy, appears to be a pro-survival response to stressors such as UV irradiation and starvation (Tondera et al., 2009; Gomes et al., 2011; Rambold et al., 2011). Mfn1, Mfn2 and Opa1 defects, which prevent gaining mitochondrial stability by the mixing of matrix contents lead to poor respiration in cells (Chen et al., 2005).

Figure 1.3.1.1 Confocal images of mitochondrial morphology in mouse embryonic fibroblast cells Confocal images of control MEF cells transfected with mitochondrial RFP and empty vector plasmids during basal conditions display a fused (A), and fragmented (B) mitochondrial morphology. Mitochondria are dynamic organelles that can change their morphology by two highly regulated, opposing processes, known as mitochondrial fission and fusion. These mechanisms are essential for various cellular processes as well as maintaining a healthy mitochondrial network. A) Mitochondrial fusion helps to restore mitochondrial membrane potential to improve ATP production and gain stability by the mixing of matrix contents such as DNA and calcium. B) Fission allows easier movement of mitochondrial tubules, as well as removal of damaged mitochondria, that can contaminate the rest of the network. Images were taken from experiments described in section 3.4.

1.3.1.1 Outer mitochondrial membrane fusion Mfn1 and Mfn2 are essential for mitochondrial fusion (Rojo et al., 2002; Chen et al., 2003). These GTPases mediate OMM fusion by forming homo- or heterodimers and tethering adjacent mitochondrial tubules, by the extension of their HR2 domains (Chen et al., 2003; Franco et al., 2016). The loss of both Mfn1 and Mfn2 proteins results in embryonic lethality, indicating their importance during development (Chen et al., 2003). Isolated mouse embryonic fibroblasts (MEFs), from Mfn1 or Mfn2 KO 31 embryos, have predominantly fragmented mitochondria and undergo fusion events at a significantly lower rate, than WT MEFs (Chen et al., 2003). Interestingly, mtDNA levels were unaffected in these cells, although, the function of many individual mitochondria was impaired. The use of MitoTracker showed a significant reduction in mitochondrial staining, indicating a reduction in mitochondrial membrane potential (Chen et al., 2003). Embryonic lethality of dual Mfn KO proteins could also be explained by the pleiotropic properties of the proteins. Mfn2 proteins on the endoplasmic reticulum (ER), allow tethering between the ER and mitochondria. This allows better transmission of calcium signalling between these two organelles, which is especially important for ATP production in the heart (Traaseth et al., 2004; de Brito & Scorrano, 2008).

1.3.1.2 Inner mitochondrial membrane fusion Opa1 is an IMM protein and is predominantly present at cristae junctions. Opa1 is also a GTPase protein and is part of the mitochondrial fusion machinery. Opa1 is the human orthologue of the Mgm1 protein, essential in mitochondrial fusion in yeast cells (Wong et al., 2000). This protein was first identified as its mutation leads to mitochondrial abnormalities and is responsible for autosomal dominant optic atrophy, causing blindness in 1 in 50,000 individuals (Delettre et al., 2000). Opa1 oligomer tethering causes IMM fusion, which allows the mixing of mitochondrial matrix contents (Ishihara et al., 2006). Fusion of the outer and inner membrane does not happen simultaneously, and the loss of the inner Δψm inhibits Opa1 mediated fusion, thereby preventing the mixing of matrix contents (Chen et al., 2005; Malka et al., 2005; Ishihara et al., 2006; Twig et al., 2008). This prevents the mixing of matrix content with potentially damaged mitochondrial tubules. Opa1 at cristae junctions is essential for preserving cristae integrity, helping to reduce the risk of apoptosis by preventing the release of cytochrome c without BAX and BAK interaction (Frezza et al., 2006). Opa1 processing by inner-membrane peptidases YME1L and OMA1 create long and short Opa1 isoforms which affect mitochondrial fusion, fission, and mitochondrial metabolism (Del Dotto et al., 2017).

1.3.2 Mitochondrial fission

Mitochondria cannot be created de novo (Fox, 2012; Shiota et al., 2015); therefore fission is essential for their transmission in dividing cells (Figure 1.3.1.1B). Fission is also vital for many other cellular functions, such as differentiation and mitochondrial transport (Frazier et al., 2006; Chen & Chan, 2009; Chan, 2012). Mitochondrial

32 fission is also a critical step in mitophagy; which allows the selective elimination of dysfunctional and depolarised mitochondria (Twig et al., 2008). Dynamin-related protein 1 (Drp1), is the principal protein required for mitochondrial fission within cells (Smirnova et al., 2001; Yoon et al., 2001; Ingerman et al., 2005), but its function depends on the ER, actin filaments, and the presence of OMM proteins (Otera et al., 2010; Friedman et al., 2011; Palmer et al., 2011; Ji et al., 2015). Mitochondrial fission is also an important component of apoptosis. Promoting mitochondrial fusion by inhibiting Drp1 activity, in the presence of apoptotic stimuli, has been shown to attenuate cell death (Frank et al., 2001).

1.3.2.1 Dynamin-related protein 1 Drp1, also known as DLP1, is a member of the dynamin family of proteins. These large GTPase proteins are structurally similar. They aggregate to form spirals around vesicles, which constrict following GTP hydrolysis to cause scission of mitochondria (Stowell et al., 1999; Praefcke & McMahon, 2004). A large proportion of what is known about Drp1 is due to research carried out on the yeast homolog of Drp1, Dnm1 (Shaw & Nunnari, 2002).

Drp1 mitochondrial activity was first identified in 1998, as the expression of its GTPase deficient mutant (Drp1k38A) in COS-7 cells, only affected mitochondrial structure and distribution. These cells predominantly appeared to have collapsed perinuclear mitochondrial networks, compared to the even distribution of mitochondrial tubules in control cells (Smirnova et al., 1998). The overexpression of wild-type Drp1 caused excessive mitochondrial fragmentation (Labrousse et al., 1999). It was confirmed that Drp1 forms higher order oligomers on the OMM, creating a ring-like structure around mitochondrial tubules to form a division apparatus (Smirnova et al., 2001; Yoon et al., 2001). It has been shown that Drp1 is also involved in peroxisome fragmentation, as the inhibition of Drp1 activity lead to the formation of elongated peroxisomes (Koch et al., 2003; Waterham et al., 2007).

1.3.2.2 Drp1 regulation Drp1 is a cytosolic protein (Smirnova et al., 1998), regulated by post-translational modifications, that can promote or inhibit its translocation to the OMM. Several proteins regulate Drp1 localisation and activity depending on cellular conditions, by phosphorylation of different serine (S) residues within its structure (Hall et al., 2014). Phosphorylation of S616 by cyclin-dependent kinase (CDK) or dephosphorylation at S637 by calcineurin promotes mitochondrial fission, by activating Drp1 translocation

33 to the mitochondria (Taguchi et al., 2007; Cereghetti et al., 2008; Campello & Scorrano, 2010; Marsboom et al., 2012). p53 up-regulated modulator of apoptosis (PUMA) is an upstream activator of apoptosis, which leads to Bax/Bak activation, and also promotes Drp1 recruitment to the mitochondria (Wang et al., 2009). Phosphorylation of S367 by Pim-1, a downstream effector of Akt-mediated cardioprotective signalling, opposes PUMA activity (Din et al., 2013). The inhibition of Drp1 mediated fission, by cardiac overexpression of Pim-1, reduces mitochondrial fission and significantly protects the heart against simulated Ischaemia (sI); whereas it’s inhibition promotes PUMA activity and increases cellular sensitivity to sI (Din et al., 2013). S637 phosphorylation by cyclic AMP-dependent protein kinase, PKA, also inhibits Drp1 activity and promotes its detachment from the OMM, by reducing Drp1 GTPase activity (Chang & Blackstone, 2007).

Dephosphorylation of S656 by calcineurin occurs during high cellular calcium. Pro- survival kinase activity of PKA phosphorylation at S656 has a fission inhibitory outcome. However, unlike PKA phosphorylation at the Drp1 GTPase domain, this does not affect Drp1-GTP binding and hydrolysis (Chang & Blackstone, 2007; Cribbs & Strack, 2007). As previously mentioned, an increase in calcium signalling, stress and increase in workload causes mitochondrial fusion (section 1.3.1). These conditions were found to cause hyperphosphorylation of Drp1 by PKA, inhibiting mitochondrial fission (Cribbs & Strack, 2007).

1.3.2.3 Actin, myosin and ER activity during mitochondrial fission Although Drp1 is the principal mediator of mammalian mitochondrial fission, fragmentation cannot occur without ER-mitochondria interaction (Friedman et al., 2011). Dnm1 helices have a rough diameter of 129nm, which would not be able to form without ER pre-constriction of mitochondria. ER circumscribed mitochondria were shown to have a diameter of 138-146nm; whereas, uncircumscribed tubules had a diameter of 293-215nm (Friedman et al., 2011). Short extensions from the ER wrap around tubules, to reduce this diameter. Mitochondrial fission proteins, other than Drp1 are present at constriction sites, as they are required for Drp1 anchoring. Mitochondrial fission factor (Mff), does not appear to facilitate the formation of sites for ER constriction of mitochondria (Friedman et al., 2011), but the MiD proteins, (mitochondrial dynamics proteins of 49 kDa and mitochondrial dynamics proteins of

51 kDa, MiD49 and MiD51, respectively), have been shown to affect ER- mitochondria interaction (Elgass et al., 2015).

The assembly and disassembly of Drp1 helices have been linked to the presence of actin and actin-binding proteins at constriction sites. The inhibition of actin polymerisation prevents Drp1 recruitment to the mitochondria (De Vos et al., 2005). Actin filaments help Drp1 oligomerisation and increase their GTPase activity, required for scission (Ji et al., 2015). Actin polymerisation and association at the constriction sites relies on the presence the ER localised inverted formin 2 (INF2) (Korobova et al., 2013). It is believed that this process is then followed by myosin II polymerisation at these sites. Inhibition of myosin using blebbistatin reduced Drp1 ring formation and promoted mitochondrial elongation (Korobova et al., 2014).

1.3.2.4 Outer mitochondrial membrane fission proteins Drp1 molecules, even after activation, are not capable of oligomerising at the mitochondria without OMM fission proteins, which act as scaffolds. The OMM fission proteins identified in mammals so far are fission 1 protein (Fis1), Mff, MiD49 and MiD51 (Stojanovski et al., 2004; Otera et al., 2010; Palmer et al., 2011). These proteins help Drp1 to polymerise and form rings around the mitochondrial tubules. Hydrolysis of GTP by Drp1 rings induces a conformational change in their structure to cause constriction, sufficient for mitochondrial scission (Otera et al., 2010; Mears et al., 2011). The role of Fis1 as a fission protein in mammalian cells has been controversial, as their KO has little effect on Drp1 recruitment and mitochondrial morphology, but it instead appears to be involved in mitophagy (Shen et al., 2014). MiD49 and MiD51 are mitochondrial specific, but Mff and Fis1 are also capable of recruiting Drp1 to peroxisomes (Palmer et al., 2013).

1.4 Mitochondrial arrangement and dynamics in the heart

Changes in mitochondrial morphology and its link to neurodegenerative conditions is well established and have been meticulously studied, given that defects in mitochondrial dynamics can be easily identified. On the other hand, research in mitochondrial dynamics in cardiomyocytes has been emerging at a much slower rate, as mitochondrial dynamic events are much harder to identify in such tightly packed and contracting cells (Hom & Sheu, 2009). Many still find the possibility of mitochondrial dynamics in cardiomyocytes a difficult concept to believe. This is despite the high expression of many mitochondrial dynamics proteins in adult

35 cardiomyocytes (Dorn, 2015). It may also be that due to the high importance of mitochondria, genetic conditions affecting mitochondrial dynamics should also be a common cause of cardiac disease, but the heart is often spared compared to other organs (Dorn, 2015). Despite this, mitochondrial dynamics is essential for mitochondrial biogenesis and removal of damaged mitochondrial tubules in order to maintain an efficient rate of ATP production (Piquereau et al., 2013).

1.4.1 Observing mitochondrial dynamics in cardiomyocytes

One challenge when researching mitochondrial dynamics in cardiomyocytes is their natural tendency to beat, even when isolated. This beating makes it exceptionally difficult to image the cells at a high resolution or image specific mitochondrial tubules in real-time. The use of drugs such as Blebbistatin, which directly inhibits the interaction of myosin heads with actin filaments within sarcomeres, can be used to image the cells in a relaxed state (Kovács et al., 2004). The use of such drugs is controversial as particular cellular pathways and activities, such as cellular contraction, macropinocytosis and phagocytosis, may also be altered when compared to normal physiological conditions (Shu et al., 2005).

Another difficulty in identifying mitochondrial dynamics in cardiomyocytes is due to the tightly packed and dense structure of these cells, leaving little space for mitochondrial tubules to manoeuvre their way around the cells (Hom & Sheu, 2009). There are three different subpopulations of mitochondria in adult cardiomyocytes, which help optimise cardiomyocyte function. These three populations are perinuclear (PNM) mitochondria which surround the nuclei, sub-sarcolemmal (SSM) and interfibrillar (IFM) mitochondria, most of which are tightly packed between sarcomeres for instant ATP supplies (Figure 1.4.1.1). These different arrangements possess different biochemical properties as well as potentially different fusion and fission capabilities (Dorn, 2015; Hatano et al., 2015). During hypoxic conditions, oxygen consumption for ATP production is mainly carried out by SSM, to provide energy for contraction. Subsequently, Δψm in IFM drops but remains normal in SSM (Hatano et al., 2015).

Figure 1.4.1.1 Electron microscopy image of an adult mouse cardiomyocyte The adult cardiomyocyte consists of three mitochondrial subpopulations: Perinuclear mitochondria (PNM) which are around the nucleus (N), subsarcolemmal (SSM) and interfibrillar (IFM) mitochondria, which are located between myofibrils.

The use of fluorescent dyes has always been a vital tool in imaging mitochondrial movements and dynamics within cells. Most of these dyes are mitochondrial membrane potential dependent. Therefore, they are not suitable for simulated IRI imaging studies, as the loss of the Δψm, causes the stain to leak out of the mitochondria. The development of Dendra mouse lines, expressing photoactivatable mitochondrial specific fluorescent proteins, allows studying of mitochondrial dynamics in basal conditions, as well as during the loss of the Δψm (Pham et al., 2012).

1.4.2 Inhibition of mitochondrial fission in the heart during IRI

Although the compact arrangement of mitochondria in cardiomyocytes can limit their movement, the high expression of mitochondrial fusion and fission proteins in adult cardiomyocytes suggests that they still play a vital role in these cells. Changes in mitochondrial morphology, more specifically, the inhibition of Drp1-mediated fission upon reperfusion, has been found to be cardioprotective; and linked to a reduction of apoptotic cell death as a result of inhibiting MPTP opening (Ong et al., 2010; Wang et al., 2011; Din et al., 2013).

Interestingly, the genetic ablation of Mfn1/2 and Opa1, which promotes mitochondrial fragmentation, has also been shown to be cardioprotective. However, this may be a consequence of also altering the proteins’ pleiotropic roles within the cells (Papanicolaou et al., 2011; Papanicolaou et al., 2012; Piquereau et al., 2012). Genetic and pharmacological inhibition of Drp1 has been shown to be protective against IRI, in cell lines and adult murine hearts but the precise cause and mechanisms by which this protection is achieved remain to be identified (Ong et al., 2010; Wang et al., 2011; Disatnik et al., 2013). This protection may be linked to a reduction in oxygen consumption rate during IRI which would reduce the oxidative burst upon reperfusion (Chouchani et al., 2014; Zepeda et al., 2014). Upregulation of Drp1 inhibitors within cardiomyocytes has also shown to significantly protect the heart against IRI (Din et al., 2013). These studies highlight the potential of achieving cardioprotection, by inhibiting Drp1 mediated mitochondrial fission during acute IRI.

It is important to note that inhibition of mitochondrial fission is likely only to be protective if achieved acutely, as both mitochondrial fusion and fission are an essential part of maintaining mitochondrial health (Chan, 2012). Chronic manipulation of the mitochondrial dynamics machinery is often detrimental in the setting of IRI. Genetic knockout of Mfn2 and knockdown of Opa1 in mice induces cardiac hypertrophy over time (Papanicolaou et al., 2011; Piquereau et al., 2012). Permanent inhibition of Drp1 activity has been shown to be detrimental in cardiomyocytes, causing accumulation of dysfunctional mitochondria, cardiac remodelling, cardiac failure and premature death in mice (Kageyama et al., 2014; Ikeda et al., 2015; Song et al., 2015). Drp1 mutation in humans leads to a significant increase in mitochondrial and peroxisome fragmentation and causes several severe developmental conditions, that consequently lead to death at a very young age (Waterham et al., 2007).

Newly identified fission proteins, MiD49 and MiD51, are mitochondrial specific Drp1 recruiters on the OMM (Palmer et al., 2011). There is growing evidence identifying the importance of the MiD proteins in the fission machinery and mediating Drp1 activity (section 1.4). This thesis investigates their activity specifically in cardiomyocytes during acute IRI and their potential to serve as targets of cardioprotection, by inhibiting mitochondrial fission upon reperfusion.

1.5 Mitochondrial fission proteins MiD49 and MiD51 as targets of cardioprotection

Novel components of the mitochondrial fission machinery, mitochondrial dynamics proteins of 49 kDa and mitochondrial dynamics proteins of 51 kDa were first identified as fission proteins in 2011 (Palmer et al., 2011). These mitochondrial specific proteins recruit Drp1, independently of other mitochondrial fission proteins, to the OMM (Liu et al., 2013; Palmer et al., 2013). Inhibition of their activity promotes mitochondrial elongation, by unopposed fusion, in cell lines (Palmer et al., 2011; Loson et al., 2013; Palmer et al., 2013). An imbalance in their expression and activity has been linked to human skeletal muscle myopathy and pulmonary arterial hypertension (Bartsakoulia et al., 2018; Chen et al., 2018).

1.5.1 Discovery

As part of a large cellular screening of novel cDNAs coding for unknown human transcripts, and to aid with functional characterisation in support of bioinformatics data, Simpson et al., developed a systematic strategy to identify the localisation of these proteins (Simpson et al., 2000). Full coding cDNAs were amplified by PCR and recombined into GFP expressing vectors (pdEGFP and pdEYFP). To ensure that the localisation observed was independent of GFP tagging, plasmids were created to code for the proteins with carboxyl (C) terminus or amino (N) terminus GFP tags. The open reading frames (ORFs)-GFP plasmids were used to transfect HeLa and Vero cells. Live cell imaging and immunofluorescence analysis of transfected cells, showed a high consistency of protein localisation in both cell lines.

Of the proteins investigated, approximately 5% had mitochondrial localisation. The overexpression of one of these proteins, with a C-terminal GFP tag, caused abnormal changes in the mitochondrial structure. Imaging at 16, 24 and 40 hours post-transfection was carried out, to monitor cellular changes, with increasing levels of protein expression. A positive correlation was observed between the increase in protein levels and an increase in the formation of mitochondrial networks (Figure 1.5.1.1). This effect, along with the proteins’ ability to specifically localise at mitochondria was not observed when the GFP tag was present at the N-terminus.

Figure 1.5.1.1 SMCR7L protein mitochondrial localisation The overexpression of SMCR7L in Hela cells with the protein GFP tag. A) The presence of GFP at the N-terminus caused cytosolic expression of SMCR7L. B) However, the presence of GFP at the C-terminus allowed mitochondrial localisation of SMCR7L, leading to the formation of fused mitochondrial tubules (Simpson et al., 2000).

This protein was first referred to as the Smith-Magenis syndrome chromosome region candidate gene 7 like (SMCR7L) protein, as the protein shares a 45% amino acid (aa) sequence identity with Smith-Magenis syndrome chromosome region candidate gene 7 (SMCR7) protein (Palmer et al., 2011). Although SMCR7’s chromosomal location is within the specific region identified to be linked to Smith- Magenis syndrome (17p11.2) (Smith et al., 1986), the protein is not associated with the condition (Bi et al., 2002; Slager et al., 2003). SMCR7 and SMCR7L genes are located on two separate chromosomes (SMCR7L gene, loci 22q13.1), but they share structural and functional similarities, are highly conserved and vertebrate- specific (Zhao et al., 2011).

SMCR7 and SMCR7L, due to their mitochondrial activity and size, were later renamed mitochondrial dynamics proteins of 49 kDa (MiD49, at 49kDa) and mitochondrial dynamics proteins of 51 kDa (MiD51, at 51kDa), respectively (Palmer et al., 2011). Although there is an agreement within the scientific community, that MiD49 and MiD51 are involved in mitochondrial dynamics, there is still dispute over whether they are mitochondrial fusion or fission proteins (see section 1.5.3). Some research groups refer to MiD49 as mitochondrial elongation factor 2 (MIEF2) and MiD51 as mitochondrial elongation factor 1 (MIEF1), as they believe that the MiD proteins are part of the mitochondrial fusion machinery (Zhao et al., 2011; Liu et al., 2013). In this thesis, the proteins will be referred to as MiD49 and MiD51.

1.5.2 Cellular expression of MiD49 and MiD51

Human MiD49 and MiD51 consist of 454aa and 463aa, respectively, which are 45% identical in their amino acid sequences (Palmer et al., 2011; Zhao et al., 2011; Liu et al., 2013). They are highly conserved and vertebrate-specific (Zhao et al., 2011; Liu et al., 2013). Different levels of their expression were observed from western blots of human cell lines (Liu et al., 2013). Interestingly, real-time PCR of samples from human foetal and adult organs suggests a variation in the proteins’ expression, as well as being differentially expressed with development (Liu et al., 2013). MiD51 levels are significantly higher than MiD49, and β-actin, in foetal organs; but it is vastly reduced by adulthood. MiD49 mRNA levels become marginally higher than MiD51 in most adult cells, with the exception of expression in skeletal and heart cells (Figure 1.5.2.1). MiD49 expression in skeletal and heart cells is dramatically higher in adult cells, compared to foetal cells (Liu et al., 2013).

Figure 1.5.2.1 mRNA expression of MiD (MIEF) proteins in human foetal and adult organs Real-time PCR from human foetal and adult organs of MIEF1 (or MiD51) and MIEF2 (or MiD49) appear to be differentially expressed with development, but the expression of at least one of them remains high in the heart and skeletal muscle. MIEF1 (MiD51) levels dramatically reduce by adulthood and MIEF2 (MiD49) mRNA expression increases significantly, with maturation (Liu et al., 2013).

Interestingly even though mitochondrial dynamics is highly restricted in muscle cells, compared to many other cell types (Hom & Sheu, 2009), the expression of at least one MiD protein remains high in both stages of life (Liu et al., 2013), indicating the importance of their function. Their higher expression does not necessarily correlate with higher mitochondrial volume within the cells, as similar expression levels are not present in other cell types with high mitochondrial volume and energy demands,

41 such as liver and kidney cells. These findings do indicate functional differences between the two proteins. One obvious difference between the foetal and adult heart is the metabolic switch that occurs soon after birth. This may be linked to the change in the MiD proteins’ expression levels in adulthood. Glycolysis and lactate oxidation are the major sources of ATP production during foetal development, which rapidly switches to predominantly fatty acid oxidation at birth (Lopaschuk et al., 1992; Bartelds et al., 2000; Onay-Besikci, 2006; Lopaschuk et al., 2010). (Bartelds et al., 2000; Lopaschuk et al., 2010). Lower expression of MiD51 in the adult heart may also be a protective evolutionary outcome. Unlike MiD49, MiD51 is able to bind ADP and GDP (see section 1.5.4.3) and therefore, its success in Drp1 recruitment may be determined by changes in metabolic activities within the cell (Loson et al., 2014; Richter et al., 2014). If this interaction is necessary for MiD51 activity, cardiomyocytes will be more susceptible to undergo mitochondrial fragmentation during high cardiac activity and the balance in mitochondrial dynamics, required to maintain a healthy mitochondrial network, will be disrupted (Bach et al., 2003; Frazier et al., 2006; Chan, 2012).

1.5.3 MiD49 and MiD51 are Drp1 adaptors involved in the mitochondrial fission machinery

Much of the information available on the role of MiD49 and MiD51 is based on the original studies which showed alterations in mitochondrial morphology, after changes in protein expression levels. Their function as mitochondrial dynamic proteins is still under dispute by some groups, as their overexpression or deletion causes mitochondrial fusion at basal conditions, summarised in Figure 1.5.3.1 (Palmer et al., 2011; Liu et al., 2013; Loson et al., 2013). The rapid mitochondrial fission observed after stress induction of cells, overexpressing the MiD proteins, has provided strong support for their role as mitochondrial fission proteins (Loson et al., 2013).

Figure 1.5.3.1 MiD depletion or overexpression promotes mitochondrial elongation MiD49 and MiD51 are important components of the mitochondrial fission machinery A) The MiD proteins are OMM receptors of Drp1, required for mitochondrial fission. The deletion or the overexpression MiD49 and/or MiD51 induces mitochondrial elongation and the formation of mitochondrial networks (B and C, respectively).

1.5.3.1 Genetic ablation of MiD49 and MiD51 The mitochondrial proteins were first described as mitochondrial fission proteins in 2011 (Palmer et al., 2011). As already mentioned, the MiD proteins must be present at the OMM to induce mitochondrial fission. The deletion of their transmembrane domain prevents OMM foci formation and Drp1 recruitment. The knockdown (KD) of either protein does not affect the expression of other mitochondrial dynamic proteins but significantly reduces Drp1 recruitment to the mitochondria summarised in Figure 1.5.3.1 B (Palmer et al., 2011).

Loson and colleagues found that the fused mitochondrial phenotype of Drp1 null cells was significantly more severe than observed in Fis1/Mff null cells, indicating that residual fission, independent of Fis1 and Mff, could still occur. The additional KD of MiD49 or MiD51 increased mitochondrial fusion, to a similar level as Drp1 null cells. Previous reports had indicated that the KD of both MiD49 and MiD51 was required to induce a significant change in mitochondrial morphology (Palmer et al., 2011). Loson et al., did not observe a further increase in mitochondrial fusion when both MiD proteins were KD in Fis1 and Mff null cells and argued that the reason the KD of one protein was sufficient to induce fusion might be due to a higher level of

43 protein KD achieved during the study. The team observed similar levels of mitochondrial fusion following the KD of the MiD proteins expressing physiological levels of Fis1/Mff, (Loson et al., 2013).

Carbonyl cyanide m-chlorophenylhydrazone (CCCP) causes mitochondrial uncoupling and induces Drp1 S637 dephosphorylation, initiating unopposed Drp1 mediated mitochondrial fission (Legros et al., 2002; Cereghetti et al., 2008). If the MiD proteins were fusion proteins, then their KD should cause a very rapid fragmentation of mitochondria. In fact, the KD of both proteins (80% reduction) significantly reduced mitochondrial fission in the presence of CCCP, at a similar level to Drp1 KD cells, indicating that the MiD proteins are part of the mitochondrial fission machinery (Palmer et al., 2011).

1.5.3.2 Overexpression of MiD49 and MiD51 The overexpression of MiD49 and MiD51 leads to a reduction in mitochondrial fragmentation events, even though their interaction with Drp1 would naturally lead to the assumption that they are mitochondrial fission proteins (Palmer et al., 2011; Zhao et al., 2011; Liu et al., 2013; Loson et al., 2013; Palmer et al., 2013). A low- level increase in the expression of MiD49 or MiD51 does not appear to have a noticeable effect on mitochondrial morphology (Palmer et al., 2011). However, a significant increase in the expression of either protein causes a very distinctive level of mitochondrial fusion, summarised in Figure 1.5.3.1 C (Loson et al., 2013). Their overexpression in 293T cells, of either MiD49 or MiD51, induces mitochondrial elongation and formation of clusters (Zhao et al., 2011; Liu et al., 2013).

A cell fusion assay, using polyethylene glycol (PEG) treatment, of cells expressing mitochondrial-GFP or mitochondrial-DsRed was carried out by Zhao et al., 2011 to investigate the MiD proteins’ dynamic properties. There was a significant increase in mitochondrial fusion, observed as the co-localisation of the two mitochondrial populations, following MiD51 overexpression (Zhao et al., 2011). The research group later identified that the overexpression of MiD49 caused a significantly higher level of fusion than MiD51 in 293T cells, with more cells displaying compact perinuclear fused mitochondrial clusters, suggesting that MiD49 has a more potent fusion-inducing ability (Liu et al., 2013). The combined overexpression of both proteins resulted in an intermediate mitochondrial morphology, compared to their observation with the single overexpression, rather than a more severe level of fusion following MiD49 overexpression (Liu et al., 2013). The increase in mitochondrial

44 fusion events following the overexpression of MiD49 and MiD51 during basal conditions, observed by several groups in early studies, provided further evidence to suggest that the MiD proteins are indeed mitochondrial fusion proteins.

Elongated mitochondria, formed as a result of MiD49 or MiD51 overexpression, are associated with F-actin. The disassociation of these elongated networks by inhibiting actin polymerisation, using cytochalasin D, caused a collapse of the network (Palmer et al., 2011). No significant outcome was observed after the disruption of microtubules (Palmer et al., 2011). The authors note that the filament interactions observed may not be physiologically relevant as the proteins were overexpressed.

Drp1-mediated mitochondrial fission was inhibited when the MiD49 and MiD51 transmembrane domains were deleted, as this led to cytosolic MiD-Drp1 interaction, and thereby reduced Drp1 aggregation at the OMM (Palmer et al., 2011; Zhao et al., 2011; Liu et al., 2013). Liu et al., suggest that although their interaction with Drp1 is weaker in a cytosolic state (MiD49Δ1-49 and MiD51Δ1-48), the resulting mitochondrial elongation observed provides further evidence to support the proteins’ role in mitochondrial fusion (Liu et al., 2013). This is not a strong statement as the outcome observed was due to the overexpression of mutants, unable to be bound to the OMM; therefore, their cytosolic interaction with Drp1 leaves a smaller pool of free Drp1 which can be recruited to the mitochondria, to induce fission. With that in mind, this does not explain why the overexpression of the native proteins induces mitochondrial elongation, as their overexpression increased Drp1 recruitment to the OMM. Given these data, it is understandable why some research groups may consider the MiD proteins to be fusion proteins.

It was previously suggested that the overexpression of MiD proteins might impair the arrangement and the organisation of the Drp1 fission machinery on the OMM (Palmer et al., 2011). Another hypothesis could be that the proteins sequester Drp1 activity, as their overexpression in Mfn2 null cells produced a similar mitochondrial morphology as to cells expressing the Drp1K38A mutants, defective of GTP binding (Palmer et al., 2013). Loson et al., 2013 identified the cause of mitochondrial fusion after MiD49 or MiD51 overexpression, in HeLa cells. As previously mentioned, S637 phosphorylation negatively mediates Drp1 activity, and Drp1 S616 phosphorylation leads to mitochondrial fragmentation (Chang & Blackstone, 2007; Cereghetti et al., 2008; Marsboom et al., 2012). Cells overexpressing MiD49 or MiD51 had higher levels of Drp1 S637-PO4 and a higher preference for recruiting this inactive form or

Drp1 to the OMM, compared to Drp1 S616-PO4 (Loson et al., 2013). Although MiD49 or MiD51 overexpression causes fusion at basal conditions, the opposite is observed during stress. CCCP induces Drp1 S637 dephosphorylation (Cereghetti et al., 2008). CCCP treatment dephosphorylated S637 residues and further enhanced OMM Drp1 recruitment in MiD-overexpressing cells, with a more significant effect observed with MiD51 overexpression, causing a severe fragmentation of mitochondrial tubules (Loson et al., 2013). The same group also found that Fis1/Mff null cells overexpressing MiD49 or MiD51 had a much more rapid mitochondrial fragmentation response after CCCP treatment than cells transfected with an empty vector. This further supports evidence that the MiD proteins are not involved in the mitochondrial fusion machinery.

It is important to note that the overexpression of the MiD proteins may not necessarily represent normal physiological conditions, and the initial studies were conducted in cells still expressing at least one of the Mfn proteins (Palmer et al., 2011; Zhao et al., 2011; Liu et al., 2013). Zhao et al., were able to show that the mitochondrial fragmentation observed from the KO of only Mfn2 could be reversed after the overexpression of MiD51 (Zhao et al., 2011). Palmer and colleagues believed that knockdown of both Mfn1 and Mfn2 is necessary to identify the true function of MiD49 and MiD51. Indeed the overexpression of MiD49 or MiD51 could not reverse the predominant mitochondrial fragmented state of Mfn1 and Mfn2 null MEF cells, or cause the mixing of mitochondrial contents in cell hybrids formed during cell fusion assays (Palmer et al., 2013).

With more evidence emerging about the MiD proteins cellular functions, it is becoming clear that fusion occurs only in particular circumstances. The overexpression of the MiD proteins causes mitochondrial fusion by recruiting Drp1, in its inactive state (S637-PO4) and therefore reduces the chance of its interaction with other fission proteins (Loson et al., 2013). This leads to unopposed fusion due to the presence of the mitofusin proteins (Loson et al., 2013; Palmer et al., 2013). Stress induces dephosphorylation at Drp1 S637, of the high number of Drp1 proteins at the OMM, leads to rapid Drp1 mediated fragmentation (Loson et al., 2013).

1.5.4 The molecular structure of MiD49 and MiD51 in relation to their function

Although there may be dispute in regard to the proteins’ specific dynamic activity, all studies do agree that MiD49 and MiD51 function, is based on their interaction with Drp1 at the OMM. Initial studies identified the importance of the proteins transmembrane region, with respect to their mitochondrial localisation and function. In 2014, the structure of MiD51 was determined by two independent groups (Loson et al., 2014; Richter et al., 2014). The crystal structure of MiD49 was identified soon after (Loson et al., 2015). It was the identification of their crystal structures that helped to provide a better understanding of their function at the OMM.

1.5.4.1 MiD49 and MiD51 are mitochondrial specific receptors of Drp1 In 2011 Palmer et al., conducted a similar experiment to Simpson et al., (Simpson et al., 2000), this time showing that both C-terminus GFP tagged MiD49 and MiD51 localised at the mitochondria, in COS7 and HeLa cells (Palmer et al., 2011). This was also observed by other groups, following V5 C-terminus epitope tagging of the MiD proteins, in 293T cells (Zhao et al., 2011; Liu et al., 2013).

Mitochondria were isolated by both research groups to further investigate the MiD proteins' precise location at the mitochondria. Western blotting carried out after sodium carbonate extraction, identified incorporation of proteins within the lipid bilayer, and showed the MiD proteins to be membrane bound. Similar to other OMM proteins, such as Mfn2, Tom 20 and Tim 23, the MiD proteins were also sensitive to proteinase K treatment (Palmer et al., 2011; Zhao et al., 2011), thereby confirming that the MiD proteins are integral OMM proteins. Co-immunoprecipitation of GFP- tagged MiD proteins was shown to have a direct association with Drp1 (Palmer et al., 2011).

MiD mutants lacking the transmembrane domain (MiD49Δ1-49 and MiD51 Δ1-48), at the N-terminus, lose their ability to attach to the OMM. They, therefore, interact with Drp1 at the cytosol instead, resulting in mitochondrial elongation (Palmer et al., 2011; Zhao et al., 2011; Liu et al., 2013; Palmer et al., 2013). This was similar to the findings of Simpson et al., after the disruption of the N-terminus (Simpson et al., 2000). The transmembrane residues within the transmembrane domain have been identified to be aa26-47 in MiD49 (454aa in total), and aa24-46 in MiD51 (463aa in total). This sequence is highly conserved in both proteins, essential for anchoring

47 the proteins to the OMM, but is not required for Drp1 interaction (Palmer et al., 2011; Palmer et al., 2013).

1.5.4.2 MiD mitochondrial foci formation MiD proteins form distinct foci on the OMM, at ER-mitochondria contact and constriction sites (Elgass et al., 2015). Foci formation is independent of the presence of Drp1, as they still form in Drp1 KO cells, however, they are required for Drp1 docking (Richter et al., 2014). Low-level overexpression of MiD49 or MiD51 in cell lines, which did not alter normal mitochondrial dynamics, resulted in their collection at discrete foci on the mitochondrial surface (Palmer et al., 2011; Zhao et al., 2011; Liu et al., 2013). There was a significant increase in Drp1 mitochondrial co-localisation at these foci (Palmer et al., 2011). A closer look at these foci revealed that they are present in ring formations around the mitochondrial tubules, similar to Drp1 rings (Smirnova et al., 2001; Legesse-Miller et al., 2003), and were able to form at potential constriction sites even in the absence of Drp1 (Palmer et al., 2011).

MiD49 and MiD51 proteins were not found to have peroxisome association, unlike Mff and Fis1, but their overexpression also caused similar peroxisome elongation to Drp1 KD cells (Palmer et al., 2013). This was identified to be due to the reduction of Drp1 peroxisomes association, and Drp1 overexpression further increased their presence at the OMM, rather than reverse peroxisome fusion (Palmer et al., 2013). Zhao et al., were the first to show that MiD51 mitochondrial foci formation and Drp1 recruitment could still occur in cells after siRNA KD of Fis1, Mff or Mfn1 (Zhao et al., 2011).

The absence of MiD proteins at the peroxisome further indicated that they are able to function independently of Fis1 and Mff. Artificially expressing the MiD proteins on lysosomes successfully redirected Drp1 there; however, this was weakly observed with Mff and not at all with Fis1 (Palmer et al., 2013). This provided further evidence that they are able to recruit Drp1 independent of Fis1 and Mff, and potentially with a higher affinity.

1.5.4.3 The inactive nucleotidyltransferase fold MiD49 and MiD51 are compact globular proteins, comprising of two α-helical regions held together by a central β-strand region (Loson et al., 2014; Richter et al., 2014; Loson et al., 2015). The membrane-proximal regions of the proteins are not

48 as compact in structure. For this reason, to produce high-resolution crystal structures of MiD49 and MiD51, deletion at the proteins’ N-terminal was carried out to remove the membrane-proximal region (Loson et al., 2014; Ma & Sun, 2014; Richter et al., 2014; Loson et al., 2015).

Human MiD51 consists of 463 amino acids. Residues 23-48 are predicted to contain the transmembrane domain, and residues 50 to 123/8 are part of the membrane- proximal region (Ma & Sun, 2014; Richter et al., 2014). Richter et al., identified that the expression of human MiD51 lacking the disordered region (MiD51Δ50-123), in wild- type and MiD51-null MEFs, had no negative effect on the proteins ability to successfully recruit Drp1 to the mitochondria (Richter et al., 2014). The functional purpose of this region may be to allow more flexibility, whilst the attached Drp1 form rings and constricts, during fission at the OMM. The deletion of residues 1-118, allowed the creation of high resolution crystallised human MiD51ΔN118 (Richter et al., 2014). Similarly, other groups also observed that the deletion of residues 1-128, allowed crystallisation of human MiD51 at a resolution of 3.1 Å (Ma & Sun, 2014); and 1-133 residue deletion of mouse MiD51 (MiD51Δ1–133) permits identification of the native structure at a resolution of 2.2 Å (Loson et al., 2014).

Similarly, to crystallise MiD49, the membrane-proximal region which is also predicted to lack a secondary structure, was deleted. Due to further complications in creating a high-resolution crystal, MiD49 mutants were created. Out of 12 mutants with alanine point substitution mutations, and lacking residues1-125, only one allowed crystallisation (MiD49R218A). The MiD49R218A mouse mutant crystal structure was solved at a resolution of 2.4 Å. This mutant was still capable of recruiting Drp1 (Loson et al., 2015).

Both structures revealed that the proteins belong to the nucleotide transferase (NTase) protein family, which normally bind nucleotide triphosphates, but lacked enzymatic activity. The NTase fold was found between the two alpha-helical regions connected by a central β-strand region (Loson et al., 2014; Richter et al., 2014; Loson et al., 2015). Both groups identified an additional electron density within this region of MiD51, suggesting the presence of a bound molecule. Surprisingly, fluorescence-based shift assay of MiD51 showed the highest level of stability with adenine diphosphate (ADP) binding, and it also weakly interacted with guanosine diphosphate (GDP). No significant structural change could be observed following MiD51-ADP and MiD51-GDP interaction (Loson et al., 2014; Richter et al., 2014).

No ligand binding at the nucleotidetransferase cleft could be identified for MiD49 (Loson et al., 2015). The nucleotidetransferase cleft of MiD49 is too small to allow nucleotide binding. Compared to MiD51, the only nucleotide binding residue conserved within the pocket is histidine (H193 in MiD49) (Loson et al., 2013).

Interestingly MiD51-Drp1 recruitment still occurs in the absence of ADP binding (Loson et al., 2014; Richter et al., 2014). ADP binding is suggested to act as an essential cofactor to stabilise Drp1 spiral formation around mitochondria. Loson et al., found that MiD51-ADP binding can promote Drp1-Drp1 interaction, and its GTP hydrolysis activity, at basal level (Loson et al., 2014). However, Richter et al., showed that Drp1 binding to MiD51 mutants that are unable to bind ADP or GDP, can still form rings around mitochondria and cause fission (Richter et al., 2014). If in fact, ADP is an essential cofactor of MiD51, its activity may facilitate mitochondrial fission events, when cellular ADP levels are high. If indeed MiD51 promotes Drp1 activity, its low expression in the adult heart may be a protective evolutionary outcome (Liu et al., 2013), as this would make the cells very prone to mitochondrial fragmentation during high cardiac activity or ischaemic conditions. The weak binding of GDP at the nucleotidyltransferase fold of MiD51 may also be part of a stabilisation or nucleotide sensing process, due to Drp1 hydrolysis of GTP, to achieve the conformation change required to execute constriction (Mears et al., 2011; Richter et al., 2014).

1.5.4.4 The MiD loop is essential for Drp1 interaction X-ray crystallisation of mouse and human MiD proteins, lacking their transmembrane and membrane-proximal regions, revealed the presence of a highly conserved surface loop on both proteins, essential for Drp1 interaction (Loson et al., 2014; Richter et al., 2014; Loson et al., 2015).

MiD51 is capable of recruiting Drp1 puncta to the mitochondria even if it lacks the ability to bind to ADP or dimerise (Loson et al., 2014). The expression of GFP tagged MiD51 mutants lacking the defined loop at residues 238-242 (MiD51ΔPEYFP), although capable of still being present at the OMM of COS7 and MEF cells, were unable to recruit Drp1 to the mitochondria (Richter et al., 2014). The same was observed in 293T Fis1/Mff null cells, as Drp1 proteins remained cytosolic after the expression of two MiD51 mutants, one with a mutation prior to the loops (R234A R235A and N237A) and the other containing a mutation within the loop (E239A Y240A R243A). Similarly, it was shown that the disruption of the salt bridge (R235-

D249) directly below the loop, produced a similar outcome as the loop deficient mutants (Richter et al., 2014). The artificial expression of MiD51 at the lysosomal surface, established in a previous study (Palmer et al., 2013) showed that mutants, lacking the four amino acid loop, could no longer redirect Drp1 to the lysosome (Richter et al., 2014).

The Drp1-binding motif was also later identified on the MiD49 surface, residues 230- 234 (LEFHP) (Loson et al., 2015). Although MiD49 and MiD51 do not share the same amino acid sequence in this region, there is still a high sequence homology, and the loops are almost identical in shape. The salt bridge that maintains the MiD49 loop’s structural integrity, interaction between R227 and D241 residues, is also present and essential for Drp1 recruitment. Co-immunoprecipitation carried out in 293T cells expressing mutants, lacking the salt bridge (MiD49R227A), led to a significant decline in Drp1 interaction. Live cell imaging of Fis1/Mff null cells expressing the loop mutants, showed that the MiD mutants were incapable of recruiting cytosolic Drp1. These finding highlighted the importance of the loop sequence, and the stability of the structure, in its ability to successfully interact with Drp1.

A recent study by Kalia et al., has produced valuable evidence to further our understanding of the mechanism of the interaction between Drp1 and the MiD proteins. The authors’ in vitro findings provided further evidence to support the importance of the MiD binding loop, within the dynamin recruitment region (DDR), for successful Drp1 binding (schematic representation of DDR shown as yellow regions on MiD49 and MiD51 in Figure 1.5.5.1) (Kalia et al., 2018). The surprising discovery made from cryo-electron microscopy of Drp1 and MiD49/ MiD51 was that each MiD DDR interacts with four Drp1 molecules, at four possible surfaces within each Drp1 chain (receptor interfaces 1-4), to form cofilaments (Kalia et al., 2018). DDR interaction with interfaces 1 and 2 are only possible following Drp1 nucleotide binding which is otherwise inaccessible (Kalia et al., 2018). The presence of stalk loops at interfaces 3 and 4 (LINs and L2 loop, respectively) are also essential components for successful receptor interaction. The authors suggest that the inability to interact with MiD49 or MiD51 may explain the impaired Drp1 activity observed in conditions linked to the LINs loop mutation (Chang et al., 2010; Sheffer et al., 2016; Vanstone et al., 2016; Kalia et al., 2018). GTP hydrolysis by Drp1 results in MiD49 or MiD51 dissociation before constriction of the Drp1 rings. MiD association was found to be structurally incompatible with the Drp1 rings during 51 constriction, further indicating that receptor dissociation is required prior to constriction (Kalia et al., 2018).

1.5.4.5 MiD protein dimerisation The homodimer formation of MiD51 was first identified in 2011 before the protein structure was identified. Wild-type MiD51, as well as transmembrane deficient mutants, were able to form homodimers, but the deletion of residues 49-195 prevented dimerisation (Zhao et al., 2011). Dimerisation of MiD51 is not essential in Drp1 recruitment, but it has been suggested to be necessary to cause fission. Cells expressing mutants, incapable of forming dimers undergo mitochondrial fission to a lesser degree after CCCP or antimycin treatment (Loson et al., 2014). MiD49 did not crystallise as a dimer, as the protein lacks the residues used by MiD51 to dimerise. The authors suggest that there may be weak dimerisations present, which could not be detected in the truncated protein (Loson et al., 2015). These findings also suggest that the MiD proteins could have alternative mechanisms of Drp1 recruitment. Recently, it has been shown that the MiD proteins form cofilaments with Drp1 molecules by binding to four Drp1 proteins, at four receptor interface sites (Kalia et al., 2018). As these studies were predominantly carried out using mutants or performed in vitro, it is still not clear if MiD dimer formation occurs in cells and if this is essential prior to the formation of higher-order structures with Drp1.

1.5.5 MiD-ER interaction

Mitochondrial fission occurs at sites which have first undergone constriction by the ER. Small projections from the ER, wrap around mitochondrial tubules, reducing their diameter by around 30%, to allow the formation of Drp1 helices (Friedman et al., 2011; Elgass et al., 2015).

Recently, the simultaneous presence of MiD49/MiD51 foci and their interaction with the ER has been shown at ER-mitochondria constriction sites, summarised in Figure 1.5.5.1 (Elgass et al., 2015). Unlike Mff, their foci formation does not require the presence of Drp1 (Richter et al., 2014). MiD51 was shown to remain at the ER- mitochondria constriction site throughout fission. Interestingly less than 40% of ER- mitochondria contact at the MiD foci, were located at constriction sites (Elgass et al., 2015). This suggests that the MiD proteins may have pleiotropic roles, such as ER tethering to facilitate inter-organelle signalling.

The single KD of MiD49 or MiD51 had no effect on ER-mitochondria constriction sites. Conversely, the ER-mitochondrial interaction was significantly reduced after the KD of both MiD49 and MiD51 (Elgass et al., 2015). These findings further support the importance of the role of the MiD protein in the mitochondrial fission machinery, and the possible need to KD both of the proteins to prevent mitochondrial fragmentation (Palmer et al., 2013).

Mff foci have also been identified at ER-mitochondria contact sites; however, the ER can still circumscribe mitochondrial tubules in the absence of Mff or Drp1. Interestingly, Mff deletion does not affect ER-mitochondria contact and constriction, but their presence at these sites is believed to have a regulatory effect on the MiD proteins, and promote Drp1 GTPase activity (Friedman et al., 2011; Elgass et al., 2015; Osellame et al., 2016).

Figure 1.5.5.1 Mitochondrial ER interaction is essential for Drp1 mediated fission A) MiD foci interact with ER projections at the OMM. B) The ER circumscribes the mitochondrial tubule and constricts, to permit Drp1 oligomerisation. GTP bound Drp1 molecules interact with the DDR of MiD proteins to form Drp1 rings (shown as yellow regions on MiD49 and MiD51). Mff molecules promote Drp1 GTPase activity, required for receptor dissociation, before constriction. C) After Drp1-mediated fission, Drp1 rings disassemble and remain cytosolic until they are again recruited to the OMM.

1.5.6 Regulation of MiD49 and MiD51 activity during physiological and stressed conditions

CCCP mitochondrial uncoupling induces Drp1 recruitment and causes mitochondrial fission. Alterations to the fission machinery can prevent this response and reduce mitochondrial fragmentation (Palmer et al., 2011; Loson et al., 2013). Osellame et al., investigated the degree to which the level of mitochondrial fission inhibition may protect the cells against apoptotic signalling. Mitochondrial fragmentation was assessed in MEFs with the KO of MiD49, MiD51, Mff or Fis1 and compared to MEFs deficient in Drp1. The KO of single Drp1 adaptors did provide some protection, but the best protection was observed with ΔDrp1, ΔMiD49+ MiD51+ Mff and ΔMiD49+ MiD51+ Mff+ Fis1 (Osellame et al., 2016). The additional KO of Fis1 did not make a significant difference to mitochondrial connectivity or cellular protection, as there was no significant difference in the level of cytochrome c release between, ΔMiD49+ MiD51+ Mff and ΔMiD49+ MiD51+ Mff+ Fis1 (Osellame et al., 2016). In this study, MiD49 and MiD51 were shown to have more prominent role in intrinsic apoptotic signalling sensitivity. This may be linked to their involvement in cristae structure, as MiD49 or MiD51 KO cells are significantly more resistant against cristae remodelling, cytochrome c release and apoptosis, following Drp1-dependent mitochondrial fission (Otera et al., 2016). This outcome was not observed in Mff null cells during the induction of intrinsic apoptotic signalling (Otera et al., 2016). The KD of MiD49 has been recently linked to protection against doxorubicin cardiotoxicity, as this reduced doxorubicin-induced mitochondrial fragmentation and cellular apoptosis (Zhou et al., 2017b).

MiD51 and Mff have been shown to assemble at the same mitochondrial fission sites (Elgass et al., 2015). The two proteins have a Drp1 recruitment regulatory effect on each other. GTPase activity of MiD recruited Drp1 proteins, at mitochondrial constriction sites, is simulated in the close proximity of Mff (Elgass et al., 2015).

MiD protein activity has also been found to be regulated by the membrane- associated ring-CH-type finger 5 (MARCH5, also known as E3 ubiquitin-protein ligase MARCH5). MARCH5 activity leads to the loss of MiD49 during stress, subsequently reducing mitochondrial fragmentation. The KD of MARCH5 restores MiD49 ability to induce mitochondrial fragmentation during stress (Xu et al., 2016). This indicates the presence of a natural protective mechanism against mitochondrial fragmentation that may become active and protect against IRI. Drp1 and Mff 54 negatively regulate this activity to reduce ubiquitination of MiD49 (Cherok et al., 2017). Evidence from these studies provide further support for the involvement of the MiD proteins in intrinsic apoptosis, during stress and highlighting their potential as targets of cardioprotection against IRI.

CHAPTER 2: Research Objectives

2.1 Overall objective

The overall objective of this thesis was to investigate whether the newly discovered mitochondrial fission proteins, MiD49 and MiD51, are novel targets of cardioprotection.

2.2 Overall hypothesis

We hypothesise that genetic ablation of both MiD49 and MiD51 protects the heart against acute ischaemia-reperfusion injury by inhibiting mitochondrial fission.

2.3 Aims

1) In Chapter 3 we will investigate in cardiac cell-lines the effects of genetic ablation of MiD49 and MiD51 on mitochondrial morphology, mitochondrial calcium overload, and susceptibility to mitochondrial permeability transition pore opening and acute ischaemia-reperfusion injury.

Specific hypothesis: Genetic ablation of MiD49 and MiD51 will induce mitochondrial elongation, prevent mitochondrial calcium overload, reduce susceptibility to mitochondrial permeability transition pore opening and attenuate cell death following acute ischaemia-reperfusion injury.

2) In Chapter 4 we will investigate the effects in vivo of genetic ablation of MiD49 and MiD51 on cardiac size and function (including contractile reserve), mitochondrial morphology, and susceptibility to acute ischaemia-reperfusion injury.

Specific hypothesis: Genetic ablation of MiD49 and MiD51 will not affect baseline cardiac size and function, will improve contractile reserve, will induce mitochondrial elongation, and will reduce myocardial infarct size following acute ischaemia- reperfusion injury.

3) In Chapter 5 we will use a high throughput screen to identify a novel small molecule inhibitor of the interaction between Drp1 and MiD49 and MiD51, as a novel cardioprotective strategy for inhibiting mitochondrial fission induced by acute IRI.

Specific hypothesis: Identification of a novel small molecule inhibitor of the interaction between Drp1 and MiD49 and MiD51, will provide a novel therapeutic strategy for reducing myocardial infarct size following acute IRI.

CHAPTER 3: Investigating the role of MiD49 and MiD51, as targets of cardioprotection: in vitro studies

3.1 Introduction

Mitochondria are central in determining the fate of cardiomyocytes during acute IRI (Murphy & Steenbergen, 2008); as their dysfunction can trigger apoptotic and necrotic cell death. Changes in mitochondrial morphology, are an essential part of maintaining mitochondrial health, which occur due to the presence of evolutionarily preserved dynamic proteins (Bach et al., 2003; Frazier et al., 2006; Chan, 2012). Mitochondrial fission often occurs during mitochondrial dysfunction, and is an important step in their removal, by mitophagy (Chen & Chan, 2009). Fission is also an important step during apoptosis, by which fission occurs upon apoptotic stimuli signalling. Inhibition of mitochondrial fission by targeting Drp1 activity can prevent apoptotic cell death (Frank et al., 2001). Inhibition of mitochondrial fission has been shown to be cardioprotective against IRI, with beneficial outcomes observed in cellular and animal models of AMI (Ong et al., 2010; Wang et al., 2011). One way in which this occurs is by reducing the susceptibility of mitochondrial permeability transition pore opening upon reperfusion, which prevents the release of apoptotic factors into the cytosol (Hausenloy et al., 2003).

Drp1 is an essential protein for mitochondrial fission (Stowell et al., 1999; Praefcke & McMahon, 2004). Currently, there are no pharmaceutical inhibitors of Drp1, to prevent its activity during IRI (Bordt et al., 2017). Recently identified proteins MiD49 and MiD51 are outer mitochondrial membrane receptors of Drp1, which allow the latter’s polymerisation around mitochondrial tubules, to cause fission (Palmer et al., 2011). Unlike other proteins that interact with Drp1, the MiD proteins are mitochondrial specific. The reduction in their expression significantly reduces Drp1 recruitment to the mitochondria, resulting in unopposed mitochondrial fusion (Palmer et al., 2011; Loson et al., 2013; Palmer et al., 2013).

In this chapter, we used cardiac and non-cardiac cell lines to investigate the role of MiD49 and MiD51 as novel targets of cardioprotection. We conducted some of the initial experiments using MEFs, which have been commonly used in publications based on the MiD proteins; however, to focus on the cardiac-specific nature of our 58 investigation, HL-1 and H9c2 cardiac cell lines were predominantly used to explore the role and activity of MiD49 and MiD51 in the setting of SIRI. Some of the in vitro experiments were carried out using both cell lines, to examine the reproducibility of our findings.

3.2 Research objectives

Overall objective: In this section, we will investigate in cardiac cell-lines the effects of genetic ablation of MiD49 and MiD51 on mitochondrial morphology, mitochondrial calcium overload, and susceptibility to mitochondrial permeability transition pore opening and acute IRI.

Individual aims:

3.2.1 Aim1: Investigating the effect of MiD49 and MiD51 overexpression: in vitro studies

1) To investigate the changes in mitochondrial morphology after MiD49 and MiD51 overexpression. Cell lines will be transfected with MiD49 and MiD51 plasmids, to achieve overexpression. At the same time, the cells will be co-transfected with mitochondrial fluorescent protein plasmids, to allow identification of transfected cells and confocal imaging of mitochondrial morphology.

2) To investigate the changes in MPTP opening susceptibility after MiD49 and MiD51 overexpression. Cell lines will be loaded with Tetramethylrhodamine, Methyl Ester, Perchlorate (TMRM), and imaged by confocal microscopy to induce the formation of ROS and MPTP opening.

3) To investigate the changes in susceptibility to acute IRI after MiD49 and MiD51 overexpression. Cell lines will be subjected to simulated IRI, and cell viability will be assessed.

3.2.2 Aim 2: Investigating the effect of MiD49 and MiD51 genetic ablation: in vitro studies

1) To investigate the changes in mitochondrial morphology after KD of MiD49 and MiD51. Cell lines will be transfected with MiD49 shRNA and MiD51 shRNA plasmids, to achieve MiD KD. At the same time, the cells will be co-transfected with mitochondrial fluorescent protein plasmids, to allow identification of transfected cells and confocal imaging of mitochondrial morphology.

2) To investigate the changes in MPTP opening susceptibility after KD of MiD49 and MiD51. Cell lines will be loaded with TMRM, and imaged by confocal microscopy to induce the formation of ROS and MPTP opening.

3) To investigate the changes in susceptibility to acute IRI after KD of MiD49 and MiD51. Cell lines will be subjected to simulated IRI, and cell viability will be assessed.

4) To investigate the changes in mitochondrial morphology and mitochondrial calcium overload induced by IRI after KD of MiD49 and MiD51. Cell lines will be subjected to simulated IRI and imaged in real-time by confocal microscopy to assess mitochondrial morphology and calcium levels.

3.3 Materials and Methods

3.3.1.1 HL-1 cardiac cells HL-1 cardiac cells are atrial cardiomyocytes of a cancerous lineage, isolated from AT-1 mice (Claycomb et al., 1998). For these cells to maintain their cardiac contractile phenotype, HL-1 cells were cultured in specially formulated Claycomb medium (51800C, Sigma-Aldrich) and supplemented with 10% foetal bovine serum (FBS), specific for HL-1 cell culture (F2442, batch 11A568, Sigma-Aldrich). The culture medium was also supplemented with 0.1mM of Norepinephrine, 2mM of L- Glutamine (G7513, Sigma-Aldrich) and 100U/ml of Penicillin/Streptomycin (P4458, Sigma-Aldrich). Norepinephrine stocks were made from powder, 30mM ascorbic acid and 10mM norepinephrine in ddH2O, which was filter sterilised before being stored at -20°C. The cells used were from frozen stocks, originally received from Prof Claycomb in 2010. Culture medium was changed once every 48 hours until cells became confluent and cellular beating was observed. At this point, cells were passaged or plated onto culture dishes, for experiments.

As HL-1 cells are non-adherent cells, culture flask and plates were first pre-coated with fibronectin, for at least 1h prior to cell seeding. The coating was prepared by firstly creating 0.02% Gelatin solution in ddH2O, and autoclaved. Once the gelatin solution had reached room temperature, 1ml of fibronectin stock (F1141 Sigma- Aldrich) was added to the solution, in sterile conditions. Aliquots were stored at - 20°C. A very thin layer of coating was required, sufficient to cover the surface of plates or dishes (1ml/35mm wells, 750µl /25mm round glass microscopy coverslips, 1.5ml/T-25 or 4.5ml/T-75 culture flask). Fibronectin coating was aspirated, before cell seeding. Culture flasks and dishes were kept in a 37°C humidified incubator, supplied with 5% CO2.

3.3.1.2 H9c2 cells H9c2 (2-1) cells are adherent neonatal ventricular myoblasts obtained from BD1X rats. Cell culture medium and its supplementation were carried out according to the supplier’s recommendation (Sigma ECACC, England). H9c2 cells were cultured in high glucose (4500 mg/L glucose) Dulbecco's Modified Eagle's medium, (DMEM medium from Sigma-Aldrich (D6546)), supplemented with 10% FBS serum (26140079, Gibco®) and 2mM of L-Glutamine (G7513, Sigma-Aldrich) and 100U/ml of Penicillin/Streptomycin. To minimise cellular fusion of H9c2 cells, they were passaged at 70-80% confluence. Culture flask and dishes were kept in a 37°C humidified incubator, supplied with 5% CO2.

3.3.1.3 MEFs and HeLa cells MEF (mouse embryonic fibroblasts) and HeLa cells (cervical cancer lineage cells) are both adherent cell lines and require very similar care. Cells were cultured in DMEM medium (D6546, Sigma-Aldrich), supplemented with 10% FBS serum (26140079, GibcoTM), 2mM of L-Glutamine (G7513, Sigma-Aldrich) and 100U/ml of Penicillin/Streptomycin. Cells were grown in T-25/T-75 flasks, and their medium was changed every 48 hours. Once confluent, the cells were passaged or seeded onto culture dishes, for experiments. Culture flask and dishes were kept in a 37°C humidified incubator, supplied with 5% CO2.

3.3.2 The general method of cell passage and seeding

Culture medium from confluent flasks were aspirated before the flasks were washed for 10 seconds with pre-warmed (37°C) phosphate-buffered saline (PBS), or 0.25% Trypsin-EDTA phenol red (25200056, Gibco®), to remove any remaining culture medium, (2ml/T-25 and 4ml/T-75 flask). Cells were incubated for 2-3 minutes with 61 pre-warmed trypsin (3-5 minutes for HL-1 cells), to detach the cells (2ml/T-25 and 5ml/T-75 flask). Cell detachment was monitored, every 1-2 minutes, under a light microscope. Gentle taps were applied to the sides of the flasks to reduce the incubation time and cell damage, whilst in trypsin. Once cells had detached, trypsin’s proteolytic activity was inhibited by the addition of FBS (1/5 the volume of trypsin), or the same volume of supplemented cell culture medium. The cell suspension was centrifuged for 2 minutes at 1000 revolutions per minute (RPM), before resuspension in 3ml of pre-warmed culture medium. A quarter to a third of cell population would be added to new culture flask with fresh medium (total volume of 4ml/T-25 and 12ml/T-75 flask) Cell culture medium in flasks were changed every 48-72h, depending on cell density. Transfected cells’ medium was changed 24h after transfection and at most, 48h before an experiment.

As the cell size and division rate of the cell lines varied, it was important that an appropriate and a consistent number of cells were seeded onto culture plates for their relevant experiment. 10µl of trypan blue (15250061, Thermo Fisher) was added to 10µl cell suspension (1:1 ratio). Cell density was measured by adding 10µl of this mixture to the grid of a freshly cleaned Haemocytometer. Three counts of live cells within the 16 square boundaries of the haemocytometer (1mm2) were taken, using a light microscope. The average value was then doubled, to account for the 2x dilution in trypan blue, and then multiplied by 10,000, to give the cell density/ml. Due to the long transfection protocol, cells were seeded at a low density onto 35mm wells or 25mm glass coverslips (HL-1 80,000, H9c2 40,000, MEFs and HeLa cells 40 000 cells/well).

3.3.3 Freezing cell stocks

Confluent cell culture flasks were washed twice with sterile PBS and detached using pre-warmed 0.25% trypsin-EDTA, phenol red (Gibco®) at 37°C. The suspended cells were then centrifuged at 1200 RPM, and re-suspended in either 10% DMSO in medium/FBS or Cellbanker2 (Zenoaq). Cell line stocks were made from cells at low passage (usually P2 or pP3). The cells were frozen and stored at -80°C. Better cell viability was achieved with Cellbanker2, and this became the permanent method of cryopreservation.

3.3.4 Thawing of frozen cell lines

Frozen stocks of cells, vials were removed from a -80°C freezer or liquid nitrogen bank, and instantly placed in a 37°C water bath. Once no more ice crystals remained in a vial, its contents were gently transferred to a 15ml falcon, before 5ml of the pre-warmed supplemented medium was gently added to the cells. Once cells were suspended in fresh medium, they were centrifuged at 800-1000 RPM for 2 minutes. The supernatant was discarded, and cells were once again re-suspended in fresh medium and transferred to T-75 flasks (10-12ml medium). Due to some cell death, during the thawing process, and a slow growth rate during the first few passages, a higher density was generally used than when normally passaging cells. Cells were used for experiments after 3 passages, after thawing.

3.3.5 Transfection of cell lines

3.3.5.1 Plasmids The MiD49GFP, MiD51GFP, shRNAMiD49 and shRNAMiD51 plasmids were a generous gift of Dr Laura Osellame and Prof Michael Ryan from the University of Monash. The Empty vector (EV), hFis1, Mfn2, Drp1k38A plasmids were a generous gift of Prof Luca Scorrano from the University of Padua. The mitochondrial calcium- sensitive protein plasmids (mitoGCamP6) was the generous gift of Prof Gyorgy Szabadkai, from University College London. Mito-BFP (49151) and GFP-Mff (49153) plasmids were commercially available and purchased from Addgene.

3.3.5.2 Plasmid amplification Plasmids were amplified by heat shock insertion into competent Escherichia coli cells (E.coli JM109, Promega). The transformed E.coli cells were grown overnight on Luria-Bertani (LB) Agar Petri dishes (Bacto agar dishes containing Tryptone, NaCl Yeast extract, and 50µg/ml ampicillin or 40µg/ml kanamycin). The following day, one antibiotic-resistant colony of transformed E.coli would be selected and grown in LB broth for 24h, containing the relevant antibiotic. The plasmids were harvested using a Midiprep kit (K210005, InvitrogenTM). Plasmids were suspended in Tris-acetate-EDTA buffer (TAE), and their concentrations were measured using a NanoDrop 1000 (Thermo Scientific).

3.3.5.3 Cell transfection As the MiD proteins are mitochondrial specific proteins, expression of the MiD-GFP plasmids allows identification of transfected cells and their mitochondrial

63 morphology. To identify mitochondrial morphology as well as identifying cells that have undergone KD of our proteins of interest, using shRNA plasmids, cells were co-transfected with mtBFP, mtGFP or mtRFP. Fluorescent protein plasmids were used at a 1:2-1:3 ratio of the shRNA plasmid. Cells used as empty vector controls (VC) were transfected with a fluorescent protein plasmid, and an empty vector (EV) control plasmid, at a 1:2-1:3 to EV plasmid ratio. Overall, 2-3µg of DNA was added per well. The amount of X-tremeGENE™ (Fugene) used per well depended on cell sensitivity/ transfection efficiency.

HL-1 cells were co-transfected with a fluorescent protein plasmid, shRNAMiD49 and shRNAMiD51, or EV and fluorescent protein plasmid at a 2.5:1 ratio. The cells were plated onto 6 well dishes 24h prior to transfection (day 0), at a density to prevent overgrowth (80,000 cells). On day one, cell culture medium was changed, 30 minutes prior to the procedure and were transfected at a 1:3 DNA to X- tremeGENE™ 9 DNA transfection reagent ratio (Roche). The following day (day 2), the culture medium was changed, and the cells were transfected again. The culture medium was changed a final time, 72h post seeding (day 3). The cells were used 72h after the initial transfection for SIRI studies and at 96h after the initial transfection (day 5), for the remaining studies. This protocol was previously established and optimised in our laboratory, due to the poor transfection efficiency of HL-1 cells. The same transfection protocol was used for MEF and HeLa cells (50 000 cells seeded on day 0), but these cells were only transfected once and used for experiments 72h after transfection. Transfection efficiency of cells was measured using fluorescence activated cell sorting (section 3.3.7).

Transfection of H9c2 cells was optimised to achieve the highest transfection efficiency, whilst ensuring that the cells remained at a healthy state for experiments. H9c2 cells were very sensitive to the protocol used to transfect HL-1 cells. H9c2 cells were also required to be seeded at a lower density to reduce the fusion of cells. Cells were plated at a 50,000 cell density per 35mm well and transfected 24h after plating. Different DNA to X-tremeGENE™ 9 ratios (1:3, 1:2.5 or 1:2) was tested, but the shRNAMiD49 and shRNAMiD51 or EV to fluorescent protein plasmid ratio was kept at a 2.5:1 ratio. Cells used for experiments were seeded at a 40,000 cells density for all experiments, except SIRI Hypoxia chamber experiments (80,000).

3.3.6 Cell fixation

Cells used for imaging were seeded onto 25mm round microscopy coverslips. 4% paraformaldehyde (PFA) and PBS were pre-warmed to 37°C, to reduce the shock applied to cells during fixing. 1ml of media from each well, containing HL-1 cells, was gently removed and replaced with pre-warmed PFA. The plate was gently shaken and then incubated for 15-20 minutes at 37°C. Other cell lines were fixed by first removing all the culture medium and washed three times with 1ml of pre- warmed PBS. 1ml of pre-warmed 4% PFA was then added to the cells and incubated for 15-20 minutes at 37°C. After incubation, the PFA, or PFA mixture, was gently removed and the cells were gently washed three times with pre-warmed PBS. The coverslips were then stuck, face down, onto microscopy slides using 1 drop of Dako fluorescence mounting medium (S3023). After leaving the slides to dry in the dark for at least 1hour, the fixed cells were ready for imaging.

3.3.7 Fluorescence activated cell sorting

Fluorescence activated cell sorting (FACS) was carried out using the BD Accuri C6 flow cytometer and BD Accuri C6 software. The system allowed the quantification of suspended cells’ size, shape and the detection of fluorescence. Non-transfected (NT) cells were used to set the quantification parameters of the system (Figure 3.3.7.1). Gates P1 or P2 were used to create the boundary of events to include as part of the cell count, excluding cell debris. At least a thousand events within the primary gate were counted for experiments. A secondary gate was then applied, to distinguish between fluorescent (transfected) and non-fluorescent cells within the P1 or P2 cell population. Depending on the fluorescent protein present, the appropriate detector was used to identify fluorescence (FL1 for green fluorescence and FL3 for red fluorescence). As NT cells should not express any fluorescence, the gate was set after all events were counted (gate M1). This ensured that no event was counted as a false positive, or counted due to auto-fluorescence. These two gates were then used to quantify the transfection efficiency of the transfected cells. FACS analysis was also used to detect cell death, using Propidium Iodide (PI) staining.

In most FACS experiments, cells were either transfected with GFP or mtGFP plasmids and therefore the green detector was used to measure transfection efficiency (FL-1, Figure 3.3.7.1 B). This allowed the detection and quantification of dead cells in gates P1, P2 or M1 which were stained red by PI, using the FL-3 red fluorescence detector (Figure 3.3.7.1 C). The laser detection used in the system can

65 also help to identify viable, damaged or dead cells without the need for PI staining. Dead and damaged cells, as well as cellular debris, have a lower forward scatter and higher side scatter, as they tend to be smaller and less granular compared to viable cells (Herzenberg et al., 1976). Two detectors are used to identify the forward and side scatter of the laser beam within the machine (FSC and SSC), as individual cells intercept the laser path. One detector identifies FSC, which is proportional to the size of the cell, and another detects the SSC, which is proportional to the shape of the cell. As the viable cells have a more spherical shape, once they intercept the FACS laser, they are less likely to produce SSC signals. The two distinct populations are shown in Figure 3.3.7.1 A, gate P1 showing the viable cells population, and gate P2 showing the overall cell population. In experiments looking at cell viability, gating similar to P2 was used. (Figure 3.3.7.1 A).

Figure 3.3.7.1 Flow activated cell sorting Transfection efficiency and viability of cell lines were quantified using FACS analysis. Gating was applied using non-transfected cells before experimental samples were analysed. A) Gate P1 and P2 were applied to exclude cell debris. P1 gating includes a predominantly viable cell population, used when measuring transfection efficiency of cells. P2 gating includes the overall cell population, used to analyse the viability of cells. B) M1 gating following P1 or P2 gating was used to quantify the transfection efficiency of cells. The FL1 sensor was used to detect cells expressing green fluorescence. C) A third gate was applied to quantify the red fluorescence emission from PI staining, using the FL3 sensor.

3.3.8 Fluorescence microscopy

Nikon Eclipse TE200 fluorescent microscope was used to image live cells, by phase contrast or fluorescence imaging (green fluorescence excitation at 488nm or red fluorescence excitation at 543nm). Nikon SPOT application suite was used to image, edit and overlap phase contrast and fluorescent images.

3.3.9 Confocal microscopy

Fixed and live cells were imaged using oil immersion objective lenses with a 40x or 63x magnification, on a Leica SP5 confocal microscope and processed on the Leica 66

Application Suite version 2.6.3.8173. The laser excitation settings were set to 488nm, to image green fluorescence protein and 543nm for red fluorescence protein imaging. However, specific fluorescent proteins and probes required more complex settings; ultraviolet excitation was used to observe mtBFP expression (excitation/emission maxima at 402 and 457nm). Laser intensity setting was set as low as possible to reduce laser-induced stress and photo-bleaching, whilst still ensuring to produce a clear image. When possible, laser intensity levels and the zoom were kept the same for each experiment, to prevent laser-induced stress causing confounding effects during experiments.

3.3.10 MPTP opening

High concentrations of TMRM are commonly used in cellular IRI models, to simulate high mitochondrial ROS conditions which lead to MPTP opening (Hausenloy et al., 2004; Bhamra et al., 2008; Ong et al., 2010). The red cationic fluorescent dye is readily sequestered by healthy mitochondria, due to the presence of the mitochondrial membrane potential (Duchen et al., 2003). The accumulation of high concentrations of TMRM in mitochondria leads to the formation of non-fluorescent dye aggregates that cause fluorescence quenching (Bunting et al., 1989; Duchen et al., 2003). Continuous laser illumination of quenched TMRM results in oxidative stress, leading to MPTP opening (Hausenloy et al., 2004; Bhamra et al., 2008; Ong et al., 2010). Mitochondria that are capable of withstanding higher levels of matrix ROS are less susceptible to MPTP opening, and will therefore be able to withstand longer periods of laser-induced oxidative stress (Ong et al., 2010). The opening of the MPTP results in the loss of the ΔΨm, leading to the movement of TMRM out of the mitochondrial matrix. This is observed as an increase in fluorescence intensity as TMRM is dequenched (Bunting et al., 1989; Duchen et al., 2003).

Cells seeded onto 25mm round glass coverslips were placed in confocal metal well adaptors prior to staining. Cells were incubated in TMRM (HL-1 and MEFs, 1µM of TMRM for 30 minutes, 0.5µM of TMRM for H9c2, 20 minutes incubation). MEFs and HL-1 cells were incubated with TMRM in Krebs imaging buffer. H9c2 cells appeared less stressed if incubated and imaged in normoxic buffer. After 20/30 minutes, the TMRM containing buffer was gently removed, and the coverslips were washed twice with 1ml of pre-warmed buffer, before a final addition of 1ml of buffer. The

67 incubation (37°C), wash steps and imaging was carried out in the dark, to prevent TMRM excitation.

Confocal imaging was carried out using the 40x oil immersion lens. Sequential imaging was carried out every 1.3 seconds, at a scan speed of 400Hz, until the red TMRM fluorescence intensity plateaued. This procedure was repeated twice at different regions of a coverslip (due to their high sensitivity, H9c2 cells were only imaged once per coverslip). The rate of MPTP opening was quantified using the Leica application suite (2.6.3. 8173, Figure 3.3.10.1). The midway point from the start of the protocol, to the time taken for a cell to reach its maximum fluorescence intensity was calculated as the estimated time of MPTP opening. It was important for cells to be imaged soon after staining and washing, as the wash step and prolonged periods in TMRM free buffer causes the dye to leak out of the cells and reduce the efficiency of the stain to induce pore opening (Duchen et al., 2003; Dedkova & Blatter, 2012). Cells additionally incubated with 0.2µM CsA, prior to imaging, were used as the positive controls (see section 3.4.3 and 3.5.3), as CsA treatment delays MPTP opening (Crompton et al., 1988; Noll et al., 1992; Duchen et al., 2003; Hausenloy et al., 2003).

Figure 3.3.10.1 Rate of MPTP opening quantification Experimental model used to quantify the rate of MPTP opening. Quantification of MPTP opening experiments was carried out using the Leica application suite. A) Transfected cells selected, expressing fluorescent proteins. B) Red fluorescence intensity is low at the start of the protocol as the TMRM stain is within the mitochondrial matrix. C) The loss of the 68 mitochondrial membrane potential, due to MPTP opening, leads to the migration of TMRM from the mitochondrial matrix and into the cytosol. This results in an increase in red fluorescence intensity as TMRM is dequenched. D) Changes in red fluorescence were tracked, and imaging was stopped once the intensity plateaued (x and y-axis showing time in seconds and mean intensity, respectively).

3.3.11 Determining cell survival after SIRI

3.3.11.1 SIRI of HL-1 cells Transfected HL-1 cells were subjected to a lethal period of simulated ischaemia, followed by a set period of simulated reperfusion. The length of simulated ischaemia and reperfusion were determined by the amount of death caused in VC cells (>45%). Claycomb medium was removed from the 35mm wells, containing HL-1 cells, and washed twice with pre-warmed hypoxic buffer (Na-Lactate 10mM, NaCl

127.8mM, KCl 14.8mM, KH2PO4 1.2mM, MgSO4 1.2mM, CaCl2 1mM and NaHCO3 2.2mM) to simulate ischaemia (Schafer et al., 2000). Prior to use, the buffer was gassed with 5% CO2-95% N2, and the resulting pH was 6.4 (Schafer et al., 2000).

Subsequently, 2ml of hypoxic buffer was added to each well and placed in an airtight hypoxic chamber. The chamber was flushed with N2 gas for 15 minutes to expel residual oxygen, before being sealed. The chamber was placed in a 37°C incubator, for 7 hours. Following simulated ischaemia, the hypoxic buffer was removed, and the cells were washed twice with pre-warmed normoxic buffer (D- glucose 10mM, NaCl 118mM, KCl 2.6mM, KH2PO4 1.2mM, MgSO4 1.2mM, CaCl2

1mM and NaHCO3 22mM). Prior to use, the normoxic buffer was gassed with 5%

CO2-95% N2 to reach a pH of 7.4 (Schafer et al., 2000). After washing, the cells were placed in normoxic buffer containing 3μM of propidium iodide (PI) and kept in a 37°C incubator (simulated reperfusion). Normoxic control cells were kept in normoxic buffer for 7h, followed by the standard simulated reperfusion step. Insulin treatment (0.2µM), upon simulated reperfusion, was used as a positive control.

After 1h, cell death post-SIRI was determined by the percentage of mtGFP transfected cells stained with PI. Imaging was carried out using a Nikon Eclipse TE200 fluorescent microscope. Ten regions were randomly selected within each well, containing approximately 50 transfected cells in total. The experiment was repeated six times (300 cells per treatment group). The analysis was carried out by two independent investigators, blinded to the treatments.

3.3.11.2 Simulated ischaemia-reperfusion injury of MEF cells A similar protocol was used for SIRI of MEF cells. MEF cells underwent 4 hours of simulated ischaemia, followed by 1 hour of simulated reperfusion. The cells were fixed after reperfusion for confocal imaging of mitochondrial morphology.

3.3.12 Real-time simulated ischaemia-reperfusion injury confocal set-up

Real-time confocal microscopy, with the Warner PM-2 heated perfusion chamber (Harvard Apparatus, Holliston, Mass), allowed live imaging of cells in a continuous fluid flow system. The chamber once constructed was mounted onto the Leica SP5 confocal microscope. This set-up allows imaging of the cells with fluorescent proteins or markers in a fixed position as environmental conditions are changed, such as an alteration of the buffers or introduction of pharmacological agents. This confocal set-up has been previously used, in our research group, to characterise the conditions required for real-time simulated ischaemia-reperfusion injury (RT SIRI) of HL-1 cells (Ong et al., 2010) and by myself for isolated adult rat cardiomyocytes (unpublished). Normoxic and hypoxic buffers (Schafer et al., 2000), were freshly prepared, on the day of the experiments. They were kept at 37°C and gassed throughout the experiment (Normox 95%O2-5%CO2, Hypox 95%N2-5%CO2) to maintain a pH of 6.4 and 7.4 respectively. Gas impermeable tubing was used to prevent alterations of the normoxic and hypoxic conditions, and a two-way tap was used to switch between buffers.

Once the chamber was constructed (Figure 3.3.12.1), the peristaltic pump was switched on to pump normoxic buffer through the system at 1ml/min. This was maintained for at least 20 minutes, to allow cells to reach basal conditions. Real- time SIRI method was based on experiments conducted by Ong et al., 2010. The period of ischaemia used for the experiments, using the flow of ischaemic buffer through the system, was determined by the time required to achieve full mitochondria fission in all cells. The re-oxygenation period was determined by the time required for optimal mitochondrial fusion (see section 3.5.4.2 for RT SIRI characterisation). The cells were imaged once at basal level (5 minutes before hypoxia) and then at fixed intervals during hypoxia and re-oxygenation. If any buffer had leaked out of the chamber during the protocol, the images from that experiment were excluded from the analysis. A blood gas analyser was used to measure glucose, oxygen and lactose levels of the buffer flowing out of the chamber, to ensure that hypoxic and normoxic conditions were achieved (ABL90 FLEX PLUS

Blood Gas Analyser, Radiometer). To measure mitochondrial calcium in real-time, cells were additionally transfected with plasmids coding for matrix calcium-sensitive protein, mitoGCamP6.

Figure 3.3.12.1 Confocal RT SIRI setup A) Warner PM-2 perfusion chamber. B) Warner PM-2 stage adaptor. C Warner PM-2 perfusion chamber magnetic heated cover. D) The fully constructed chamber with an inward flow of buffer (left) and runoff of buffer (right). E) Confocal RT SIRI setup. A peristaltic pump was used to allow a continuous flow of buffer through the system.

3.3.13 Statistical analysis

Statistical analysis was carried out using GraphPad Prism 6 (GraphPad Software, La Jolla, CA). Values are expressed as mean±SEM. Data statistical analysis was carried by one-way or two-way ANOVA followed by Bonferroni comparison test or unpaired t-test to compare the values of each variant to the control. Difference between groups and the control condition were considered statistically significant if P<0.05. Statistical significance indicated by asterisks (*P<0.05, **P<0.01 and ***P<0.001). No statistical (NS) significance was reported where P>0.05.

Aim 1: Investigating the effect of MiD49 and MiD51 overexpression: in vitro studies

3.4 Results

Preliminary work was first carried out using plasmids coding for GFP tagged MiD49 and MiD51, to ensure the mitochondrial elongation observed by Palmer et al., could be reproduced in our hands (Palmer et al., 2011). Cells overexpressing the MiD proteins were also used in experiments to identify any signs of protection against SIRI, after mitochondrial elongation, as inhibiting mitochondrial fragmentation has been shown to be cardioprotective against IRI (Ong et al., 2010).

3.4.1 Overexpression of MiD49 and MiD51 produces hyperfused mitochondria

Measurements of mitochondrial morphology were based on the dominant morphology observed (>50%) in each cell, assessed by two independent investigators, blinded to the treatment. The overall mitochondrial morphology of cells was adapted from classifications used by Palmer et al. 2011, defined as elongated (tubular extensions and network-like structures), fragmented (small doughnut-like structures), collapsed (intertwined and concentrated networks) or peri-nuclear collapsed (intertwined and concentrated networks adjacent to the nuclei). Overexpression of MiD49 or MiD51 in HL-1 cells caused the formation of elongated mitochondrial networks which were predominantly collapsed, compared to vector control cells (Figure 3.4.1.1). VC cells were transfected with an empty vector and mtRFP. 49.4±4.9% of VC cells displayed elongated mitochondria and the remaining cells contained predominantly fragmented mitochondria (50.6±4.9%). Mfn2 overexpression caused a significant increase in the presence of elongated mitochondria compared with the vector control (71.8±5.2% elongated mitochondria, P<0.05). The overexpression of human Fis1 protein (hFis1) caused a significant decrease in the percentage of cells containing predominantly elongated mitochondria, compared with the vector control group (14.2±5.2% elongated mitochondria, P<0.05).

MiD49 overexpression caused HL-1 cells to display a highly fused and collapsed mitochondrial morphology (morphology of cells imaged, 9.8±2.5% fragmented, 4.4±2.5% elongated, 55.5±3.1% collapsed mitochondrial networks and 30.3±3.3%

72 peri-nuclear collapsed). The overexpression of MiD51 also caused cells to display a predominantly fused and collapsed mitochondrial morphology (18.7±6.4 fragmented, 1.6±1.0% elongated, 48.4±3.5% collapsed mitochondrial networks and 31.4±4.3% peri-nuclear collapsed). The collapsed and peri-nuclear phenotype was only observed in cells overexpressing MiD49 or MiD51. These cells also appeared larger compared to the other treatment groups, but this was not quantified and should be further investigated in future studies.

Figure 3.4.1.1 Changes in mitochondrial morphology due to the overexpression of different mitochondrial dynamic proteins Mitochondrial elongation was significantly higher in HL-1 cells following Mfn2 overexpression compared to the VC cells; whereas, hFis1 overexpression significantly increased the proportion of cells expressing fragmented mitochondria (P<0.05). There was a significant increase in mitochondrial elongation following MiD49 or MiD51 overexpression, which was predominantly displayed as collapsed or peri-nuclear collapsed clusters (P<0.05). N=6 (total of 120 cells imaged per treatment group). This work was carried out by Dr Jessica Maeve Elder.

3.4.2 Mitochondria overexpressing MiD49 and MiD51 are protected from fragmentation during SIRI

MEF cells, as previously used by Palmer et al., 2011, were transfected with GFP tagged MiD proteins, and co-transfected with mtRFP. These cells were exposed to either 5h of normoxia or 4h of simulated ischaemia, followed by 1h of simulated

73 reperfusion. The cells were then fixed for confocal imaging. Twenty cells were randomly selected and imaged per coverslip (total of 40 cells per condition, Figure 3.4.2.1). As described in previous studies, the MiD proteins are OMM proteins, due to this their GFP signal should co-localise with the mtRFP signal, Figure 3.4.2.1 (Palmer et al., 2011; Zhao et al., 2011). Mitochondrial morphology was analysed independently, by two investigators, blinded from the treatment groups.

The overexpression of the MiD proteins in MEF cells co-transfected with mtRFP, hyperfused abnormally elongated mitochondria, compared to mtRFP control cells (EV and mtRFP) during normoxic or hypoxic conditions (Figure 3.4.2.1 and Figure 3.4.2.3, respectively). Cells overexpressing MiD49, MiD51 or MiD49 and 51 in normoxic conditions, displayed a significant increase in mitochondrial fusion (91.25±1.3%, 92.50±2.5% and 91.25 ±1.3% respectively, N=2, 20 cells per group, P<0.05) when compared to the control group (47.5±2.5%, N=2, 20 cells per group). Fused mitochondria, after overexpression of the MiD proteins, were predominantly in dense peri-nuclear clusters (Figure 3.4.2.1). The remaining cells expressed a fragmented mitochondrial morphology. This degree of fusion and perinuclear clusters were not observed in any of the cells within the control group. Another interesting observation was that cells overexpressing the MiD proteins appeared larger in size, with large nuclei, when compared to the mtRFP empty vector control cells (Figure 3.4.2.1). This observation was not quantified.

Simulated IRI increased mitochondrial fragmentation in all four groups (Figure 3.4.2.2 and Figure 3.4.2.3), however, there was still a significantly higher level of fusion in cells overexpressing the MiD proteins compared to the control group (17.50±2.5% N=2, 20 cells per group, P<0.05). There was no significant difference between MiD49 or MiD51 overexpressing cells (62.5±0% and 65.00±2.5% respectively, N=2, 20 cells per group). However, MiD49 and MiD51 overexpressing cells were more susceptible to fragmentation during SIRI, than when only one MiD protein was overexpressed (*P <0.05). Even with an increase in mitochondrial fragmentation, MiD49 and 51 overexpressing cells significantly expressed a higher level of mitochondrial fusion than the control group (46.25±1.3%, P<0.05, N=2, 20 cells per group).

Figure 3.4.2.1 Cellular mitochondria morphology observed in MiD transfected cells after 5h of normoxia MEF Cells overexpressing the MiD proteins predominantly expressed intertwined fused mitochondria, collected in dense structures adjacent to the nuclei. Scale size indicated 0- 10µM. Cells within the control group (EV and mtRFP), expressed strand like fused mitochondria, or doughnut-shaped fragmented mitochondria, scattered around the cell in an unorganised manner.

Figure 3.4.2.2 Cellular mitochondria morphology observed in MiD transfected cells after SIRI In MEF cells, simulated IRI conditions caused mitochondrial fragmentation in all 4 groups. Scale size indicated 0-10µM. This was most noticeably observed in the control group (EV and mtRFP), with most cells expressing predominantly fragmented mitochondria. There was no significant difference in the mitochondrial morphology of MiD49 and MiD51 overexpressing cells. Fewer MiD49 and MiD51 overexpressing cells remained fused, compared to when only one MiD protein was overexpressed.

Figure 3.4.2.3 Mitochondrial morphology of MEFs overexpressing MiD49 and MiD51 A) MEF cells within the normoxic buffer for 5h. Mitochondrial fusion was significantly increased with the overexpression of MiD49, MiD51 or MiD49 and 51 compared to the EV and mtRFP control group (91.25±1.3%, 92.50±2.5% and 91.25±1.3% respectively. Control 47.5±2.5%, N=2, 20 cells per group, P<0.05). B) There was an increase in mitochondrial fragmentation after 4h of simulated ischaemia, followed by 1h of simulated reperfusion (control 17.50±2.5% remained fused), however the cells overexpressing the MiD49 or MiD51 or bothe MiD proteins were less sensitive to this (62.50±0%, 65.00±2.5% and 46.25±1.3 respectively. N=2, 20 cells per group, P<0.05). Mitochondrial morphology remained predominantly fused when MiD49 or MiD51 was overexpressed with no significant difference between the two groups (NS P>0.05). Overexpression of both MiD49 and MiD51 significantly increased susceptibility to mitochondrial fragmentation during simulated ischaemic stress (P<0.05 compared to MiD49 or MiD51).

3.4.3 MiD49 and MiD51 overexpression delays mitochondrial permeability transition pore opening during oxidative stress

Confocal microscopy was used to assess the rate of mitochondrial permeability transition pore (MPTP), opening in cells overexpressing MiD49 and MiD51 and vector control cells (EV and GFP). Continuous laser stimulation of mitochondrial- specific TMRM dye (Tetramethylrhodamine, methyl ester), was carried out to induce ROS formation in the matrix (Figure 3.4.3.1). The rate of MPTP opening, with increasing levels of ROS, was measured as the halfway point of the time taken for the red stain to be released into the cytosol and achieved maximum fluorescent intensity (Hausenloy et al., 2004).

Figure 3.4.3.1 MPTP opening using TMRM staining Transfected cells and control cells were stained with TMRM. Scale size indicated 0-100µM A) Transfected MEF cells overexpressing GFP tagged MiD49 and MiD51 (laser excitation at 488nm). B) The red intensity of TMRM stained cells, at the start of the MPTP opening assay (laser excitation at 543nm). C) The opening of the MPTP leads to the red fluorescent stain to leak out from the mitochondria, and into the cytoplasm of cells. This results in an increased intensity of red fluorescence as TMRM is dequenched. Sequential imaging was carried out until the red fluorescence intensity plateaued.

The overexpression of MiD49 or MiD51 significantly delayed the time taken to induce MPTP opening in HL-1 cells, compared to VC cells (208±16 seconds with VC, 339±23 seconds with MiD49 overexpression and 336±19 seconds with MiD51 overexpression, N=6, P<0.05). MPTP inhibition, by the use of CsA, significantly delayed the time taken to induce MPTP opening compared to VC cells (341±27 seconds, P<0.05). Similarly, MPTP opening was significantly delayed in cells expressing elongated mitochondria, as a result of Mfn2 overexpression, compared to VC cells (321±32 seconds, P<0.05). However, overexpression of mitochondrial human fission protein hFis1 caused no significant change in the rate of MPTP opening (179±29 seconds, P>0.05). There was no significant difference in the rate of MPTP opening between CsA treated cells and cells overexpressing Mfn, MiD49 or MiD51 (Figure 3.4.3.2, P>0.05).

Figure 3.4.3.2 Time taken to induce MPTP opening due to the overexpression of different mitochondrial dynamic proteins in HL-1 cells CsA treatment of VC cells, as a positive control, significantly delayed the time taken to induce MPTP opening, compared to untreated VC cells (P<0.05). Similarly, the overexpression of Mfn2, MiD49 or MiD51 significantly delayed the time taken to induce MPTP opening compared to the VC (P<0.05). Overexpression of hFis1 had no significant effect on the rate of MPTP opening, compared to VC cells (P>0.05). N=6 (20 cells imaged per treatment group). This work was carried out by Dr Jessica Maeve Elder.

Correspondingly, in MEF cells, the double overexpression of MiD49 and MiD51 caused a significant delay in mitochondrial MPTP opening (Figure 3.4.3.3), occurring at 408.20±21.6 seconds, compared to the control cells at 270.6±29.3 seconds (mean difference 137.6±36.4 seconds. N=3, P<0.05).

Figure 3.4.3.3 Time taken to induce mitochondrial MPTP opening in MEF cells overexpressing MiD49 and MiD51 Induction of an elongated mitochondrial morphology by overexpressing MiD49 and MiD51 significantly delayed mitochondrial MPTP opening, during the continuous laser-induced oxidative stress of TMRM stain. MPTP opening occurred at 270.60±29.3s in the control group and at 408.20±21.6s in cells overexpressing MiD49 and MiD51. At least 8 cells were imaged per experiment/ per group (N=3, P<0.05).

3.4.4 MiD49 and MiD51 overexpression reduces cell death post SIRI

The overexpression of MiD49 or MiD51 in HL-1 cells significantly reduced the percentage of cell death post SIRI compared with the VC group: 52.0±4.6% with VC cells, 21.5±3.7% MiD49 and 20.0±3.7% with MiD51 overexpression (N=6, P<0.05) (Figure 3.4.4.1). Insulin treatment has been shown to activate the reperfusion injury salvage kinase (RISK) pathway (Hausenloy & Yellon, 2007). The use of Insulin upon reperfusion as a positive control, significantly reduced cell death post SIRI, compared to the VC group (22.0±6.1%, N=6, P<0.05). Similarly, Mfn2, overexpression in HL-1 cells significantly decreased the percentage of cell death post SIRI, compared to the VC group (22.3±4.7%, N=6, P<0.05). The overexpression of hFis1 significantly increased the percentage of cell death post SIRI, compared to VC cells (75.2±2.2%, N=6, P<0.05).

Figure 3.4.4.1 MiD49 and MiD51 overexpression protects against SIRI Insulin treatment of HL-1 cells upon reperfusion or the overexpression of Mfn, MiD49 or MiD51 significantly reduced the percentage of cell death post SIRI, compared to VC cells. The overexpression of hFis1 significantly increased percentage cell death compared with the vector control. N=6 (50 cells per treatment group). P<0.05 compared to vector control. This work was carried out by Dr Jessica Maeve Elder.

In accordance to previously published MiD studies, we were also able to show that the overexpression of MiD49 and MiD51 leads to abnormal mitochondrial fusion in MEFs and additionally in cardiac HL-1 cell lines. Mitochondrial fusion was found to be protective, by delaying MPTP opening and reducing cell death in response to SIRI, however, this response may not be physiologically relevant. As the KD of MiD49, MiD51, or both proteins, also leads to mitochondrial elongation (Palmer et al., 2011; Loson et al., 2013; Palmer et al., 2013), we continued investigating their potential as targets of cardioprotection, following their KD in cardiac cell lines.

Aim 2: Investigating the effect of MiD49 and MiD51 genetic ablation, as targets of cardioprotection: in vitro studies

3.5 Results

3.5.1 Transfection efficiency of cardiac and non-cardiac cell lines

The use of MEF cells at the start of the project was ideal due to their high transfection efficiency (VC 37.34±4.0%, MiD49+51KD 39.80±3.5%, MiD49+51 overexpression 39.78±3.2%, N=5, Figure 3.5.1.2); but it was essential to use cardiac-derived cell lines to investigate the MiD proteins as potential targets of cardioprotection, Although HL-1 and H9c2 cells lines were better suited for SIRI experiments, their transfection was more difficult. The same transfection reagent and plasmids were used during optimisation of the transfection protocols.

The protocol used for HL-1 transfection was previously established in our laboratory. HL-1 cells had a very low transfection efficiency and were required to be transfected at 24h and 48h post plating. This would achieve a transfection efficiency of 10-15%, which decreased as the passage number increased (VC 7.06±0.8%, MiD49+51 KD 7.96±1.2%, N=11, Figure 3.5.1.2). As this would be problematic for some experiments, transfection of H9c2 cells was also optimised for experiments which required a higher cell number. H9c2 are also larger in size, allowing better analysis of mitochondrial morphology.

Using the same transfection protocol as HL-1 cells (transfecting twice with mtGFP and EV at a 1:3 plasmids to transfection reagent ratio), had high toxicity for H9c2 cells (Figure 3.5.1.1). A reduction of this to 1:2 ratio, twice transfected (1:2x2), achieved a good level of transfection efficiency after 24h with a noticeable reduction in cell death. Although this was a significant improvement, transfected cells observed under a fluorescent microscope, still appeared stressed (Figure 3.5.1.1). 1:3 DNA to X-tremeGENE reagent ratio (1:3x1), achieved the same level of transfection efficiency and cells appeared to be at a healthier state. The addition of only X-tremeGENE to wells, caused a significant level of damage to cells, indicating the importance of reducing the time in which cells are exposed to the reagent.

Following the preliminary H9c2 transfection tests, the levels of transfection efficiency were quantified using FACS. The 1:2x2 and 1:3x1 protocols achieved very similar transfection efficiencies (VC1:2x2 33.53±3.0% VC1.3x1 34.00±2.2%, MiD49+51KD 1:2x2 30.84±1.1%, MiD49+51KD 1:3x1 31.84±2.9%). There was no significant difference in transfection efficiency, between the two protocols (VC P=0.9025, MiD49+51KD P=0.7584. N=4). For characterisation experiments, 50,000 cells were seeded onto plates, at day 0 (35mm wells), After characterisation, the 1:2x1 transfection protocol was used for experiments (2µg DNA, 4µl X-tremeGENE /35mm well), as this ensured that the cells remained at a healthy state and the growth rate was not highly affected. At day 0, a cell density of 40,000 was used for RT imaging experiments, to ensure that cells were well spread out, and mitochondrial morphology could be clearly imaged (Figure 3.5.1.3). A cell density of 80,000 was used at day 0, for hypoxia chamber SIRI experiments. As there is some cell death and growth rate is slower post transfection, this did not lead to overcrowding or a high level of cellular fusion.

Figure 3.5.1.1 Transfection efficiency of H9c2 cells Transfection of H9c2 cells was optimised to reduce cell damage while maintaining a high transfection efficiency (mtGFP and EV plasmids). The HL-1 cell transfection protocol was highly toxic to H9c2 cells (1:3x2). Reducing the transfection reagent (X-tremeGENE), reduced cellular toxicity, but there was still a significant level of cellular damage (1:2x2). Transfecting the cells only once, but keeping the 1:3 DNA to X-tremeGENE ratio, achieved a similar level of transfection with less damage. Transfecting the cells only once at a 1:2 DNA to X-tremeGENE ratio, achieved a lower level of transfection efficiency, but the cells remained healthy and their growth rate was less affected. Pictures were taken at 10x magnification.

Figure 3.5.1.2 Transfection efficiency of cell lines used for experiments Transfection efficiency of cell lines was measured 72h (MEFs and H9c2 cells), and 96h (HL- 1 cells) after-transfection, using FACS. A) MEF cell had high transfection efficiency (NT 0.00%, VC 37.34±4.0%, MiD49+51KD 39.80±3.5%, MiD49+51 overexpression 39.78±3.2%, N=5). B) HL-1 cells have a low transfection efficiency compared to MEF and H9c2 cells (VC 7.06±0.8%, MiD49+51 KD 7.96±1.2%, N=11). A small degree of background signal was detected in the NT cells, to give a false positive of 0.01% (N=11). C) The same transfection protocol could not be used for H9c2 cells due to a high level of cellular death. Transfecting the cells only once (1:3x1), or reducing the DNA to X-tremeGENE ratio (1:2x2) achieved the same level of transfection efficiency (VC1:2x2 33.53±3.0%, VC1.3x1 34.00±2.2%, MiD49+51KD 1:2x2 30.84±1.1%, MiD49+51KD 1:3x1 31.84±2.9%), with no significant difference between the two protocols (VC P=0.9025, MiD49+51KD P=0.7584. N=4). Although transfection efficiency was lower using a 1x1:2 protocol, there was less cell death, and the remaining cells were the healthiest post transfection (Figure 3.5.1.1).

Figure 3.5.1.3 H9c2 growth rate and cell fusion A) H9c2 cells were seeded onto 35mm dishes and left to grow for 4 days. They were not transfected, but their culture medium was changed at 24 and 48 hours post plating, same as the transfection protocol. Cells plated at a 40,000 density grew well, with minimal cellular fusion. Pictures were taken at 10x magnification. B) Transfection efficiency of cells seeded at 40,000 cell density, achieved the same level of transfection efficiency as cells used during characterisation of the protocol (VC1:2x2 34.55±3.7%, VC 1.3x1 32.72±3.2%, P=0.7238. MiD49+51KD 1:2x2 30.58±1.1%, MiD49+51KD 1:3x1 29.90± 3.8%, P=0.8706. N=3).

3.5.2 Knockdown of MiD49 and MiD51 induces mitochondrial elongation

The KD of MiD49 and MiD51 in numerous cell lines results in mitochondrial elongation (Palmer et al., 2011; Loson et al., 2013; Palmer et al., 2013). The MiD proteins were knocked down in HL-1 and H9c2 cells by transfection if plasmids coding for shRNAMiD49 and shRNAMiD51, to see if mitochondrial elongation would also occur in cardiac cell lines,. The cells were co-transfected with fluorescent tagged mitochondrial proteins (mtBFP, mtGFP and mtRFP), to allow the analysis of mitochondrial morphology. Additionally, this allowed the identification of cells that had undergone successful transfection and MiD protein KD.

3.5.2.1 Knockdown of MiD49 and MiD51 in HL-1 cells MiD49 and MiD51 KD in HL-1 cells was achieved using shRNAMiD49 and shRNAMiD51 plasmids. The cells were co-transfected with mtGFP. The control group was transfected with an EV and mtGFP. Mfn2 was overexpressed as a positive control and hFis1 as a negative control. After 96h from the initial transfection, cells were fixed in 4% paraformaldehyde. 150 cells per condition were selected at random and imaged by confocal microscopy (N=5). The morphology was analysed independently, by two investigators, blinded from the treatment groups. The investigators were asked to identify if the overall mitochondrial morphology of each imaged cell, determining if the cells expressed a predominantly fused (>50% fused mitochondria), or fragmented (<50% fused mitochondria) mitochondrial morphology. The two sets of data were then combined to look at the level of agreement between the two analyses. Overall, there was 83.50% agreement between the two investigators’ analysis (VC 82.67%, MiD49+51 KD 88.67%, Mnf2 83.33%, hFis1 79.33%). A third investigator was asked to analyse the images in which the two original investigators did not agree on.

The KD of mitochondrial proteins MiD49 and MiD51 in HL-1 cells resulted in a significant increase in mitochondrial elongation (Figure 3.5.2.1 and Figure 3.5.2.2), compared to the control group (percentage of cells displaying a predominantly fused mitochondrial morphology, 74.00±3.6% of MiD49 KD cells and 47.33±4.2% of VC cells). A similar level of mitochondrial fusion was present in the positive control group (74.67±4.6% Mfn2 overexpression). Overexpression of hFis1 had no significant effect on mitochondrial morphology, compared to the VC cells (39.33±4.0%). The KD of the proteins appears to cause an increase in cell size which was also observed in HL-1 cells expressing a fused mitochondrial 87 morphology, following MiD overexpression (observation not quantified, see section 3.4.1).

Figure 3.5.2.1 Changes in mitochondrial morphology in HL-1 cells The KD of MiD49 and MiD51 and the overexpression of Mfn2 significantly increase mitochondrial fusion, compared to control cells. The overexpression of hFis1 had no significant effect on mitochondrial morphology. Cells were co-transfected with mtGFP, to allow the analysis of mitochondrial morphology.

Figure 3.5.2.2 Mitochondrial morphology of fixed HL-1cells The KD of MiD49 and MiID51 proteins leads to a significant increase in mitochondrial fusion events (74.00±3.6%), compared to the VC cells (47.33±4.2%) The increase of mitochondrial fusion was similar to overexpression of Mfn2 (74.67±4.6%). There was no significant difference in mitochondrial morphology between hFis1 overexpression (39.33±4.0%), compared to VC cells. P=0.0012 N=5 (150 cells per condition). Statistical significance assessed by one-way ANOVA followed by Bonferroni comparison test.

3.5.2.2 Knockdown of MiD49 and MiD51 in H9c2 cells MiD49 and MiD51 KD in H9c2 cells was achieved using the same shRNAMiD49 and shRNAMiD51 plasmids, as used to transfect HL-1 cells. In addition to the protein groups tested previously (outlined in section 3.5.2.1), the individual KD of MiD49 or MiD51 proteins was also analysed, as well as the overexpression of Mff. The cells were co-transfected with mtGFP, to allow the identification of transfected cells and mitochondrial morphology. H9c2 were found to be very sensitive to the transfection protocol used for HL-1 cells. Therefore, the transfection was only carried out once, and the cells were fixed for imaging 72h after transfection, to reduce damage. H9c2 cells’ mitochondrial morphology had a higher tendency to be fused during basal conditions than observed in HL-1 cells (VC cells, Figure 3.5.2.2 and Figure 3.5.2.3). The KD of MiD49 or MiD51 alone did not cause a significant increase in mitochondrial fusion, when compared to the VC cells (MiD49 KD 63.75±9.3%, N=6, P=0.5488. MiD51 KD 79.91±8.7%, N=6, P=0.0869); however the KD of both proteins significantly increased the proportion of cells expressing a predominantly fused mitochondrial morphology (VC 55.55±9.4%, MiD49+51KD 83.75± 6.7%, N=6,

P=0.0350). MiD49 and MiD51 KD cells also expressed highly fused networks of mitochondria, but this was not to the same degree or as prevalent, as to when both proteins were KD (Figure 3.5.2.3 and Figure 3.5.2.4). The overexpression of Mff, as a negative control, significantly increased mitochondrial fragmentation events in H9c2 cells (Mff 20.00±4.3%, N=6, P=0.0063). Overexpression of Mfn2 did increase mitochondrial fusion events, compared to VC cells. However, this was not to a degree which was statistically significant (Mfn2 70.83±12.6%, P=0.3540). Overexpression of hFis1 had no significant effect on mitochondrial morphology compared to VC (hFis1 50.26±10.7%, N=6, P=0.7190).

Figure 3.5.2.3 Mitochondrial morphology or fixed H9c2 cells H9c2 cells were transfected with various mitochondrial dynamic proteins and mtGFP, fixed, and imaged, by confocal microscopy, for the analysis of mitochondrial morphology. 55.55±9.4% of VC cells expressed a predominantly fused mitochondrial morphology. MiD49 or MiD51 KD increased mitochondrial fusion. However, this was not statistically significant, compared to the VC cells (MiD49 KD 63.75±9.3%, N=6, P=0.5488. MiD51 KD 79.91±8.7%, N=6, P=0.0869). The KD of both MiD49 and MiD51 significantly increased mitochondrial fusion (MiD49+51KD 83.75±6.7%, P=0.0350), Overexpression of Mff significantly increased mitochondrial fragmentation events (Mff 20.00±4.3%, P=0.0063). Overexpression of Mfn2 increase mitochondrial fusion events, but this was not statistically significant (Mfn2 70.83±12.6%, P=0.3540). Overexpression of hFis1 caused no significant change to mitochondrial morphology, compared to VC cells (hFis1 50.26±10.7%, N=6, P=0.7190). Statistical significance assessed by one-way ANOVA (P=0.0007), followed by an unpaired t- test (N=6. 120 cells imaged per group).

Figure 3.5.2.4 Changes in mitochondrial morphology in H9c2 cells Mitochondrial morphology of fixed H9c2 cells expressing mtGFP, after the KD or overexpression of different mitochondrial dynamic proteins. Scale size indicated in whole cell images 0-25µM, and scale size indicated 0-5µM for zoomed in regions of the cells.

3.5.3 Knockdown of MiD49 and MiD51 delays MPTP opening in the presence of high mitochondrial matrix ROS

High levels of mitochondrial matrix ROS during reperfusion induces MPTP opening (Crompton et al., 1987; Crompton et al., 1988), which results in the release of apoptotic factors. The rate of MPTP opening was assessed using the TMRM oxidative stress induced assay. Continuous laser illumination of cells stained with TMRM creates oxidative stress in the mitochondrial matrix, leading to MPTP opening. The opening of the MPTP results in movement of TMRM stain into the cytosol, resulting in a significant increase in the fluorescent signal as TMRM is dequenched (Bunting et al., 1989; Duchen et al., 2003; Hausenloy et al., 2003).

3.5.3.1 MiD49 and MiD51 delay MPTP opening in HL-1 cells HL-1 cells with MiD49 and MiD51 KD were incubated with 1µM of TMRM in Krebs buffer for 30 minutes (37°C). After 30 minutes, the cells were washed and imaged in TMRM-free Krebs buffer. Continuous imaging every 1.3 seconds was used to cause TMRM laser-induced ROS formation. CsA treatment has been identified to delay MPTP opening in cell lines (Crompton et al., 1988; Noll et al., 1992; Hausenloy et al., 2003). As a positive control, TMRM stained cells were pre-treated with 0.2µM of CsA, for 20 minutes, prior to imaging. Six independent experiments were carried out per treatment group (total of 120 cells per group, Figure 3.5.3.1). This work was carried out by Dr Jessica Maeve Elder.

The KD of MiD49 and MiD51 significantly delayed MPTP opening, compared to the VC group (VC 233±20 seconds vs MiD KD 485±61 seconds). A significant delay was also observed in the CsA treated group, compared with the VC group (CsA 362±24 seconds). The over-expression of mitochondrial fusion protein Mfn2 delayed MPTP opening, compared with the VC group (Mfn2 352±31 seconds). The overexpression of hFis1 increased MPTP sensitivity compared to the VC group, but this time reduction was not statistically significant (hFis1 196±29 seconds).

Figure 3.5.3.1 MiD49 and MiD51 KD delays MPTP opening in HL-1 cells MPTP opening was induced by laser-induced breakdown of TMRM, to form high levels of mitochondrial matrix ROS. Time taken to induce MPTP opening was measured in seconds in cells transfected with empty vectors or plasmids encoding for different mitochondrial dynamic proteins. Statistical significance was assessed by unpaired t-test (N=6. 120 cells imaged per group. P<0.05). CsA treatment significantly delayed MPTP opening, compared to the VC group (CsA 362±24s vs VC 233±20s). KD of MiD49 and MiD51 and overexpression of Mfn2 significantly delayed the rate of MPTP opening compared to the VC group (MiD KD 485±61s, Mfn2 352±31s). The increased susceptibility to pore opening was not statistically significant in cells over-expressing hFis1, compared to the VC group (hFis1 196±29 s vs VC 233±20s. P>0.05). This work was carried out by Dr Jessica Maeve Elder.

Following the preliminary work carried out in HL-1 cells, by Dr Jessica Maeve Elder, I repeated this work with the addition of more treatment groups. As a delay in imaging post washing of cells can reduce the rate of MPTP opening, this meant that CsA treatment could be causing a delay in response partially due to the 20 minutes pre-treatment time required before imaging. Due to this, the overexpression of the dominant negative Drp1k38A was used as an additional positive control (Smirnova et al., 1998). Mff over-expression was used as the negative control, instead of hFis1. As well as the double KD of MiD49 and MiD51 (MiD KD), cells with individual KD of MiD49 or MiD51 were also used in the experiments. The experiment was repeated twelve times per group, due to the low HL-1 transfection efficiency. Data was excluded from the analysis if cells appeared unhealthy prior to imaging, or mechanical errors with the confocal microscope caused a delay in imaging. The positive control groups, both showed to have a significant delay in MPTP opening, compared to VC cells (VC 385.90±29.8s N=11 vs VC CsA 703.80±96.9s, N=7, P=0.0017. Drp1k38A 584.10± 47.0s N=12, P=0.0022). The KD of both MiD proteins also caused a significant delay in MPTP opening, compared to VC cells (MiD KD

579.5±50.8s, N=10, P=0.0033). There was no significant difference between MiD KD and CsA treatment, or Drp1k38A over-expression (MiD vs VC CsA P=0.2360. MiD KD vs Drp1k38A P=0.9475). The KD of MiD49 caused a significant delay in MPTP opening. However, this was not observed when MiD51 was knocked down, compared with VC cells (MiD49 KD 509.6±39.5s, N=11, P=0.0212. MiD51 KD 505.9±58.4s N=11, P=0.0818). Mff overexpression did not have a significant effect on the time required to induce MPTP opening, compared with VC cells (Mff 400.6±43.2s N=12, P=0.7855).

Figure 3.5.3.2 MiD49 and MiD51 KD delayed MPTP opening, at a similar level as Drp1k38A overexpression There was a significant delay in MPTP opening in VC CsA, MiD KD and Drp1k38A overexpressing cells, compared to VC cells (VC 385.9±29.8s N=11 vs VC CsA 703.8±96.9s,N=7,P= 0.0017. MiD KD 579.5±50.8s, N=10, P=0.0033. Drp1k38A 584.1±47.0s N=12 P=0.0022). The KD of MiD49, but not MiD51, also caused a significant delay in MPTP opening, compared with VC cells (MiD49 KD 509.6±39.5s, N=11,P=0.0212. MiD51 KD 505.9±58.3s N=11, P=0.0818). Mff overexpression did not have a significant effect on the rate of MPTP opening, compared with VC cells (Mff 400.6±43.2s N=12, P=0.7855). Eight cells were imaged per group, for each experiment.

3.5.3.2 MiD49 and MiD51 delayed MPTP opening in H9c2 cells The MPTP assay protocol had to be slightly altered for H9c2 cells, to maintain cellular health. H9c2 cells incubation with TMRM was carried out in normoxic buffer (20 minutes), as the cells appeared stressed when incubated in Krebs buffer, and

94 became very sensitive to MPTP opening (normoxic buffer: D-glucose 10mM, NaCl

118mM, KCl 2.6mM, KH2PO4 1.2mM, MgSO4 1.2mM, CaCl2 1mM and NaHCO3

22mM, Krebs buffer: D-glucose 11.0mM NaCl 118mM, KCl 4.7mM, KH2PO4 1.2mM,

MgSO4 1.2mM, CaCl2.2H2O 1.8mM, NaHCO3 25.0mM, pH 7.4). TMRM concentration was also reduced to 0.5µM, as 1µM concentration induced a very fast pore opening, making it difficult to assess the true effect of mitochondrial morphology on MPTP opening. Just as previously observed in HL-1 cells (Figure 3.5.3.1 and Figure 3.5.3.2), The KD of MiD49 and MiD51 had no significant effect on the time required to induce MPTP opening in H9c2 cells. Experiments were repeated five times for each treatment group (6-8 cells per experiment). The experiments in which there was a delay in imaging or the cells appeared unhealthy were excluded from analysis.

MPTP opening in H9c2, compared to HL-1 cells, occurred at a much faster rate (Figure 3.5.3.1,Figure 3.5.3.2 and Figure 3.5.3.3). CsA treatment as a positive control, significantly delayed MPTP opening in H9c2 cells, compared to VC cells (VC 278.0±5.71 N=5 vs VC CsA 417.9±65.1s N=4. P=0.0454). Mff overexpression in cells, as a negative control, increased susceptibility to MPTP opening, compared to VC cells (Mff 222.9±21.3s N=5. P=0.0368. The other treatment groups did not have a significant effect on the rate of MPTP opening, compared to VC cells (MiD KD 341.2±42.9s, N=5, P=0.1827. MiD49KD 278.0±5.7s, N=5, P=0.8836. MiD51 KD 252.1±37.9s, N=4, P=0.4688. hFis1 264.8±47.8s, N=4, P=0.7643. Mfn2 278.0±5.7s N=5, P=0.0531). Due to H9c2 cells higher susceptibility to MPTP opening, more experimental repeats may be required.

Figure 3.5.3.3 Mitochondrial permeability transition pore opening in H9c2 cells CsA treatment significantly delayed MPTP opening, and Mff overexpression increased susceptibility to MPTP opening, compared to VC cells (VC 278.00±5.7s N=5 vs VC CsA 417.9±65.1s N=4. P=0.0454, and Mff 222.9±21.3s N=5. P=0.0368). MiD KD and Mfn2 overexpression delayed MPTP opening, compared to VC cells; however, this was not statistically significant (MiD KD 341.2±42.9s, N=5, P=0.1827, Mfn2 278.0±5.7s N=5, P=0.0531). MiD49 KD, MiD51 KD, and hFis1 overexpression had no significant effect on the rate of MPTP opening, compared to VC cells (MiD49 KD 278.0±5.7s, N=5, P=0.8836. MiD51 KD 252.1±37.9s, N=4, P=0.4688. hFis1 264.8±47.8s, N=4, P=0.7643).

3.5.4 Mitochondrial elongation by alteration of MiD49 and MiD51 expression protects cells against ischaemia-reperfusion injury

Two models of SIRI were used to investigate if targeting MiD49 and MiD51 KD could cause cardioprotection, in cardiac cell lines. Simulated ischaemia was achieved by placing the transfected cells in an airtight chamber, deficient in oxygen. The cell medium was removed and replaced by hypoxic buffer (Schafer et al., 2000). Simulated reperfusion was achieved by reoxygenation of the cells. The hypoxic buffer was removed and replaced with normoxic buffer (Schafer et al., 2000). The first set of experiments were carried out to see if mitochondrial elongation will reduce cell death, after SIRI. Following these experiments, cells were imaged in real-time during SIRI, to see how changes in mitochondrial morphology may influence the level of cell death during simulated ischaemia and simulated reperfusion.

3.5.4.1 The KD of MiD49 and MiD51 protects cells against SIRI. HL-1 cells were exposed to 7h of simulated ischaemia, followed by 1h of simulated reperfusion using hypoxic and normoxic buffers, respectively (Schafer et al., 2000). MiD49 and MiD51 were both simultaneously KD in HL-1 cells. This significantly reduced cell death following SIRI compared to VC cells (VC 56.7±2.4% vs MiD KD 27.9±3.0% of transfected dead cells, P<0.05). Insulin-treatment of control cells, or the overexpression of Mfn2, caused a significant reduction in cell death, compared to VC cells (VC+In 17.3±3.0% and Mfn2 22.3±4.7%, P<0.05). In contrast, hFis1 over-expression caused a significant increase in cell death following SIRI, compared to VC cells (hFis1 75.2±2.2%, P<0.05). Experiments were repeated six times (N=6, 300 cells were imaged per condition Figure 3.5.4.1). This work was carried out by Dr Jessica Maeve Elder.

Figure 3.5.4.1 MiD49 and MiD51 KD protects against SIRI VC Insulin-treated cells, Mfn2 overexpressing cells and MiD49 and MiD51 KD caused a significant reduction in cell death following SIRI, compared with VC cells (VC 56.7±2.4% vs VC+In 17.3±3.0%, Mfn2 22.3±4.7% and MiD KD 27.9±3.0%, P<0.05). In contrast, hFis1 over-expression cause a significant increase in cell death following SIRI, compared with VC cells (hFis1 75.2±2.2%, P<0.05). N=6 (300 cells per condition). This work was carried out by Dr Jessica Maeve Elder.

3.5.4.2 Characterisation of real-time SIRI model in different cell lines Several cell lines were tested using the real-time model of SIRI, previously characterised in our laboratory (Ong et al., 2010). Transfected cells were placed in an airtight chamber, which allowed the continuous flow of buffer through the system. The chamber was mounted on a confocal microscope stage, and cells were imaged under different conditions. Images were taken at set time points, to minimise the stress induced by confocal imaging. The cells would first be allowed 20-30 minutes of stabilisation, while the chamber was filled with normoxic buffer. Prior to the fusion of hypoxic buffer, an image of the cells would be taken during basal conditions, to identify the mitochondrial morphology prior to hypoxia. Subsequently, an image was taken at the start of simulated ischaemia (2 minutes) followed by every 20/30 minutes, while the chamber was perfused with hypoxic buffer. The period of hypoxia required, per cell line, was set as the time required to induce mitochondrial fragmentation in VC cells. Simulated reperfusion was then achieved by switching back to normoxic buffer, flowing through the chamber. The primary focus of this period of simulated reperfusion was to observe changes in mitochondrial morphology and cell death, within the early stages of reperfusion. The chamber was found to remain air-tight and free from leaks for up to 4h. If the chamber was identified to be leaking during SIRI, the images taken from that experiment were excluded from the final analysis.

The original times tested for the RT SIRI model were based on the previous characterisation of this model in our laboratory, using HL-1 cells (Ong et al., 2010). It was identified that 60 minutes of hypoxia was sufficient to achieve mitochondrial fragmentation in HL-1 cells (N=3, P<0.0001), and restoration of mitochondrial morphology would occur within the first 30 minutes of reperfusion (Figure 3.5.4.2 A). Due to the low transfection efficiency of HL-1 cells, other cell lines were also tested using similar conditions, to see if similar mitochondrial morphology changes could be observed during RT SIRI. This would help to identify a cell line which could ideally be used as a substitute for future experiments.

Similar to HL-1 cells, H9c2 cells underwent full mitochondrial fragmentation after 80 minutes of hypoxia, which remain fragmented until reoxygenation (N=3, P<0.0001, Figure 3.5.4.2 B). There was a significant difference in mitochondrial morphology, from 20 minutes of hypoxia to 2 minutes of reperfusion, compared to morphology during basal conditions. During this period, the mitochondria of cells remained predominantly fragmented. As mitochondrial morphology beings to recover, during 98 reoxygenation, there is no longer a significant difference in mitochondrial morphology. A small degree of cell death was observed after 100-120 minutes of ischaemia, which increased upon reoxygenation (observed as cells shrinking). H9c2 cells have been shown to be more energetically similar to primary cardiomyocytes, than HL-1 cells, and are potentially more suitable for SIRI experiments (Kuznetsov et al., 2015). This RT SIRI model was suitable for the investigation of SIRI in cardiac-derived cell lines.

HeLa cells expressed a very high level of resistance against hypoxic conditions. Mitochondrial morphology showed a gradual increase in mitochondrial fragmentation; however, the difference between eat time point was not statistically significant when compared to the morphology during basal conditions (Figure 3.5.4.2 C). After 180 minutes of hypoxia, mitochondrial morphology was predominantly of an indeterminate state (same level of elongated and fragmented mitochondria). No significant change in mitochondrial morphology was observed in MEF cells, after 180 minutes of hypoxia. Characterising this model for MEF cells could not be completed, as the length of hypoxia required caused the chamber to leak, distorting the images on several occasions. These experiments indicated that MEFs and HeLa cells required a significantly longer period of hypoxia to express a predominantly fragmented mitochondrial morphology.

Figure 3.5.4.2 Characterisation of RT SIRI model Cells within an airtight chamber were mounted onto a confocal microscope and imaged in real-time, during continuous flow of hypoxic buffer (simulated ischaemia), followed by a period of reoxygenation using normoxic buffer (simulated reperfusion). Cells were transfected with mtGFP, for identification of mitochondrial morphology at each time point. A) HL-1 cells underwent full mitochondrial fragmentation after 60 minutes of hypoxia. There was a significant difference in mitochondrial morphology between cells imaged at 30 minutes of hypoxia until 5 minutes of reperfusion, compared to cells’ mitochondrial morphology during basal conditions (N=3, P<0.0001). B) 80 minutes of hypoxia was required to induce full mitochondrial fragmentation in H9c2 cells (N=3, P<0.0001). There was a significant difference in mitochondrial morphology, from 20 minutes of hypoxia to 2 minutes of reperfusion, compared to the morphology of the cells during basal conditions. C) There was a gradual increase in mitochondrial fragmentation, but full mitochondrial fragmentation during hypoxia could not be achieved in HeLa cells, using this model of SIRI (N=3, P=0.0189). Although there was an increase in mitochondrial fragmentation during simulated ischaemia, this was not found to be statistically significant when compared with the mitochondrial morphology of cells during basal condition.

100

3.5.4.3 The knockdown of MiD49 and MiD51 reduced mitochondrial fragmentation during RT SIRI The KD of MiD49 and MiD51 in HL-1 cell and H9c2 cells induces mitochondrial fusion. The established RT SIRI model was used to identify if MiD KD would prevent or reduce mitochondrial fragmentation, as observed in VC cells. This was investigated in HL-1 and H9c2 cells.

The KD of MiD49 and MiD51 in HL-1 cells protected them undergoing mitochondrial fragmentation during RT SIRI (VC N=10, MiD49+51KD N=8). There was a significant difference between the two groups (interaction P=0.0007). Comparison of individual time points between the two groups revealed a significant difference at 30 and 60 minutes of hypoxia, as well as 2 and 5 minutes of reoxygenation (Figure 3.5.4.3). At basal levels, the high level of mitochondrial fusion is characteristic of MiD49 and MiD51 KD. MiD49 and MiD51KD mitochondrial morphology remained in a predominantly fused state during hypoxia and fully recovered after 30 minutes of reoxygenation. The difference in mitochondrial morphology between each time point of MiD49 and MiD51 KD cells was not statistically significant. VC cells mitochondrial morphology became predominantly fragmented during hypoxia (Figure 3.5.4.3) and began restoration upon reoxygenation. Mitochondria in VC cells appear to undergo hyperfusion upon reoxygenation, a characteristic effect of mitochondrial stress (Tondera et al., 2009; Gomes et al., 2011; Rambold et al., 2011). However, there was no statistical difference between basal and 30 minutes reoxygenation time points. Time points 30 and 60 minutes of hypoxia and 2 minutes of reoxygenation, were significantly different in mitochondrial morphology, compared to the morphology during basal conditions.

101

Figure 3.5.4.3 Mitochondrial morphology during SIRI of HL-1 cells after MiD49 and MiD51 KD The KD of MiD49 and MiD51 (green) in HL-1 prevents mitochondrial fragmentation during RT SIRI. Comparison of individual time points between the two groups revealed a significant difference at 30 and 60 minutes of hypoxia, as well as 2 and 5 minutes of reoxygenation. VC (red) cells undergo mitochondrial fragmentation during hypoxia, which is restored after 30 minutes of reoxygenation (VC N=10, MiD49+51 KD N=8, P=0.0007).

The characterisation of the RT SIRI showed that 100 minutes of hypoxia is sufficient to achieve mitochondrial fragmentation in H9c2 cells (Figure 3.5.4.4). After hypoxia, cells were exposed to 30minutes of reoxygenation. Transfected cells from the same passage were also imaged while the chamber was infused with normoxic buffer only, for the same period of time. This was to ensure that the mitochondrial morphology changes observed were due to hypoxia and reoxygenation.

There was no significant difference in mitochondrial morphology between VC and MiD49 and MiD51 KD, during normoxia (N=6, interaction P=0.2357). Despite this, with an increase in the length of time, there was a significant difference between the two groups at various time points (Figure 3.5.4.4, Figure 3.5.4.5 and Figure 3.5.4.6). Mitochondrial morphology of MiD KD cells remained highly fused throughout the experiment (no significant difference between time points, within the MiD49 and MiD51 KD group, during normoxia). There was a reduction of VC cells

102 expression a predominantly elongated mitochondrial morphology over time. However, the mitochondrial morphology of the cells remained predominantly fused during the experiments (no significant difference between time point, within the VC group, during normoxia).

The KD of MiD49 and MiD51 in H9c2 cells inhibited mitochondrial fragmentation during hypoxia, compared to VC cells (N=6, P<0.0001, Figure 3.5.4.4 and Figure 3.5.4.6). Mitochondria in cells with MiD49 and MiD51 KD remained predominantly fused during hypoxia and reoxygenation (no significant difference between time points within the SIRI MiD49 and MiD51KD group). VC cells underwent a significant level of mitochondrial fragmentation during hypoxia, which started to recover upon reperfusion. The level of mitochondrial fusion of VC cells was significantly different from 20 minutes of hypoxia until 2 minutes of reoxygenation, compared to their morphology during basal conditions. This significant difference was also present at the same time points between VC and MiD49 and MiD51 KD cells.

103

Figure 3.5.4.4 Mitochondrial morphology during SIRI of H9c2 cells after MiD49 and MiD51 KD A) There is no significant difference in mitochondrial morphology between VC and MiD49 and MiD51KD cells, during normoxia (N=6, interaction P=0.0236). There was a significant difference at some time points between the two groups, but the mitochondria remained predominantly fused during normoxic conditions, in both groups. B) The KD of the MiD proteins prevents mitochondrial fragmentation during hypoxia, compared to VC cells (N=6, interaction P<0.0001). There is a significant difference in mitochondrial morphology between the two groups from 20 to 100 minutes of hypoxia and 2 minutes of reoxygenation.

104

Figure 3.5.4.5 Mitochondrial morphology of H9c2 VC cells during SIRI Mitochondrial morphology during basal conditions was predominantly fused in all cells. Mitochondria fully fragmented during exposure to hypoxic conditions. Mitochondrial morphology begins to recover upon reoxygenation, initially becoming hyperfused (2min of normoxia image). Cell death sometimes occurred within the VC group during SIRI, observed as cells drastically shrinking during imaging (100min of hypoxia image).

105

Figure 3.5.4.6 Mitochondrial morphology of MiD49 and MiD51 KD H9c2 cell during SIRI MiD49 and MiD51 KD cells expressed a highly fused mitochondrial morphology at basal conditions, which remained predominantly fused during hypoxic conditions and fully recovered again upon exposure to normoxic conditions.

106

3.5.5 Knockdown of MiD49 and MiD51 attenuates mitochondrial calcium overload during simulated ischaemia-reperfusion injury

The RT SIRI model was used to analyse changes in calcium levels during SIRI. Mitochondrial GCamP6 (mtGCamP6), a calcium-sensitive protein was used to quantify mitochondrial matrix calcium levels. As mtGCamP6 is a green fluorescent protein, VC and MiD49 and MiD51 KD cells were also co-transfected with mtBFP. Buffer collected from the chamber at each time point was used to measure oxygen, glucose and lactic acid levels. This was to ensure that both groups were treated to the same conditions (VC N=7, MiD49 and MiD51 KD N=6, Figure 3.5.5.1). Similar to previous experiments, if hypoxic and reoxygenation conditions were not achieved within the chamber, images were excluded from analysis. On 3 occasions, these parameters could not be measured due to blood gas analyser being under maintenance.

There was a significant difference in mitochondrial matrix calcium levels between VC and MiD49 and MiD51 KD cells during SIRI (N=8, two-way ANOVA of interaction between the two groups P=0.0325, Figure 3.5.5.2). There was no significant change in mitochondrial calcium levels during hypoxia when compared to calcium levels during basal conditions in VC cells. However, there was a significant increase in matrix calcium upon reperfusion, in VC cells during early stages of hypoxia, compared to mitochondrial calcium during reoxygenation (2, 20 and 40 minutes of hypoxia, vs 2 and 5 minutes of reperfusion, P<0.05). There was no significant increase in matrix calcium, in MiD KD cells, until reoxygenation. There was a significant difference between every time point prior to reoxygenation vs, 2 and 5 minutes of reoxygenation (P<0.005), but the difference in calcium levels is no longer statistically significant after 5 minutes of reoxygenations, as matrix calcium levels quickly begin to return to basal levels in MiD49 and MiD51 KD cells.

107

Figure 3.5.5.1 RT chamber conditions during SIRI Buffer collected from the RT confocal chamber was collected at each time point during RT SIRI, to ensure hypoxia and reoxygenation was achieved (VC N=7, MiD49+51 KD N=6). A) There was no significant difference in oxygen (A), glucose (B) and Lactic acid (C) levels between the two groups.

108

Figure 3.5.5.2 Changes in mitochondrial calcium during SIRI Mitochondrial calcium levels were quantified using the normalised fluorescence intensity of mtGCamP6 protein, during normoxia and SIRI. There was a significant difference in mitochondrial matrix calcium levels between VC and MiD49+51 KD cells (N=8, interaction P=0.0325). There was a significant increase in mitochondrial matrix calcium levels in VC cells, during the early stages of hypoxia compared to mitochondrial calcium during reoxygenation (P<0.05). There was no significant increase in matrix calcium in MiD49+51 KD cells until reoxygenation (P<0.005). Better recovery in basal calcium levels was observed in MiD49+51 KD cells, compared to VC cells.

109

3.6 Discussion

The primary objective of this chapter was to investigate the effect of modulating MiD49 and MiD51 expression on mitochondrial morphology and cell survival following SIRI. The overexpression of MiD49 or MiD51 caused a significant increase in mitochondrial elongation compared to VC cells. The hyperfused mitochondrial morphology protected the cells against MPTP opening and cell death after SIRI. The KD of MiD49 and MiD51 was required to cause a significant effect on mitochondrial morphology. The dual KD of both proteins reduced susceptibility to MPTP opening in cardiac cell lines, protected cells against SIRI, prevented mitochondrial fragmentation and calcium overload during RT SIRI, when compared to VC cells.

3.6.1 Transfection efficiency of cell lines

Cardiac and non-cardiac cell lines were used to investigate the role of MiD49 and MiD51 as novel targets of cardioprotection. Experiments were predominantly carried out in cardiac cell lines, due to our desire to investigate the role of MiD49 and MiD51 during IRI in the heart. Transfection of HL-1 cells was previously optimised in our laboratory to achieve a transfection efficiency of approximately 10-15%. This protocol required the cells to be transfected 24h and 48h after plating. Transfection of HL-1 cells with plasmids to KD MiD49 and MiD51, co-transfected with mtGFP achieved a transfection efficiency of approximately 7%. The low transfection efficiency was not suitable for some of our experimental models. Due to this, the transfection of H9c2 cells was also optimised. The same protocol to transfect HL-1 cells was highly toxic to H9c2 cells. Unlike HL-1 cells’ transfection protocol, H9c2 cells were only transfected once, to ensure that cells were healthy prior to experiments. Despite this, H9c2 cells achieved a significantly higher transfection efficiency than HL-1 cells, making them an ideal substitute for HL-1 cells.

3.6.2 Overexpression of MiD49 and MiD51 in cell lines

3.6.2.1 The overexpression of MiD49 and MiD51 promotes mitochondrial elongation during basal conditions and post SIRI Overexpression of MiD49 and MiD51 has been shown to induce mitochondrial elongation (Palmer et al., 2011). Similarly, the overexpression of MiD49 or MiD51 caused a significant increase in mitochondrial elongation in HL-1 cardiac cell lines, compared to control cells (Figure 3.4.1.1). The overexpression of Mfn2, as a positive

110 control also caused a significant increase in mitochondrial elongation; however, the morphology observed in MiD overexpressing cells was distinct from the morphology observed following Mfn2 overexpression. Cells overexpressing MiD49 or MiD51 displayed hyperfused mitochondrial networks which were predominantly collapsed or formed peri-nuclear collapsed clusters. This morphology was similar to the morphology observed by Simpson et al., 2000 and Palmer et al., 2011, indicating that mitochondrial fission has been inhibited, or fusion events have been promoted.

It has been shown that MiD overexpression results in high recruitment of Drp1, the overexpression promotes Drp1 S637 phosphorylation, and the MiD proteins also preferentially recruited this inactive form of Drp1 to the OMM (Loson et al., 2013), which leads to unopposed mitochondrial fusion. The high presence of Drp1 at the OMM makes cells highly susceptible to mitochondrial fragmentation during stress, (Loson et al., 2013). In preliminary studies, we also observed a similar level of mitochondrial fusion, previously reported following the overexpression of MiD proteins in MEF cells at basal conditions (Zhao et al., 2011; Liu et al., 2013). MEF cells overexpressing MiD49, MiD51, or both proteins were subjected to SIRI, or kept in normoxic conditions, before being fixed for imaging and analysis. All cells were co-transfected with mtRFP, and mitochondrial morphology was compared to cells transfected with an empty vector. Normoxic cells, overexpressing one MiD protein or both, showed a significant increase in mitochondrial fusion, compared to VC cells (Figure 3.4.2.1 and Figure 3.4.2.3 A).

MiD overexpression caused mitochondria to distribute in predominantly dense mitochondrial clusters adjacent to the nuclei (Figure 3.4.2.1and Figure 3.4.2.2), similar to previously published studies (Zhao et al., 2011; Liu et al., 2013). These cells also appeared larger in size compared to VC cells. This may be due to the collapsed mitochondria preventing cell division, as mitochondrial fission is an essential step in cell division; required for dividing the mitochondrial population between the two newly formed cells (Frazier et al., 2006; Chan, 2012).

The loss of the mitochondrial membrane potential initiates Drp1-mediated mitochondrial fragmentation (Legros et al., 2002). MiD-overexpressing cells are highly sensitive to CCCP induced membrane potential loss, causing mitochondrial fragmentation, compared to control cells (Loson et al., 2013). Due to this, it would be expected that cells overexpressing the MiD proteins would be highly fragmented after SIRI stress. SIRI resulted in the increase of fragmented mitochondria in all cell

111 groups. However, a significantly higher proportion of cells overexpressing MiD49, MiD51 or both proteins still expressed a predominantly fused mitochondrial morphology, compared to VC cells (Figure 3.4.2.2 and Figure 3.4.2.3). This may be due to these cells being protected against mitochondrial fragmentation during ischaemia, compared to VC cells, or the cells undergo a higher level of stress- induced hyperfusion which occurs during starvation or upon reperfusion (Tondera et al., 2009; Gomes et al., 2011; Rambold et al., 2011). Another explanation may be that cells overexpressing the MiD proteins are less susceptible to mitochondrial fission during stress as the collapsed mitochondrial networks may no longer be able to interact with ER and cytoskeletal filaments required for mitochondrial fragmentation (Otera et al., 2010; Friedman et al., 2011; Ji et al., 2015).

Interestingly a lower proportion of cells overexpressing both proteins displayed a predominantly elongated mitochondria post SIRI. This may have been due to the higher degree of mitochondrial fragmentation which occurs in cells overexpressing the MiD protein during stress conditions, and therefore more time may be required to restore the hyperfused collapsed morphology which is present during basal conditions (Loson et al., 2013). To further understand the cause of this phenomenon during SIRI, further investigation is required. Firstly the experiment should be repeated to increase the N numbers (N=2). Secondly, to identify if cells overexpressing both MiD49 and MiD51 do indeed undergo a higher degree of mitochondrial fragmentation during ischaemia, compared to cells only overexpressing MiD49 or MiD51, cells should also be fixed directly after ischaemia and imaged. This would allow us to compare changes in mitochondrial morphology during simulated ischaemia and reperfusion.

3.6.2.2 The overexpression of MiD49 and MiD51 delays MPTP opening We were able to show that the overexpression of MiD49 or MiD51 significantly delayed the time taken to induce MPTP opening in HL-1 cells (Figure 3.4.3.2). This delay was significantly higher than VC cells, but there was no significant difference between the cells overexpressing MiD49, MiD51 or Mfn2. The double overexpression of MiD49 and MiD51, which also significantly increases the presence of elongated mitochondria, caused a significant delay in mitochondrial MPTP opening in MEF cells (Figure 3.4.3.3).

112

TMRM staining is commonly used to assess the rate of MPTP opening in cells, by ROS induced stress (Huser & Blatter, 1999; Hausenloy et al., 2003; Ong et al., 2010). The limitation of this assay is that TMRM is a charge dependent molecule, and can exit mitochondria following the loss of the mitochondrial membrane potential. For this reason, VC cells were treated with CsA as a positive control during MPTP opening assay, due to the drugs ability to delay MPTP opening (Clarke et al., 2002). For this reason, we are able to conclude that the delay of TMRM release into the cytosol is due to the delayed MPTP opening, which leads to the uncoupling of oxidative phosphorylation and resulting in the collapse of the Δψm (Huser & Blatter, 1999).

There was no change in mitochondrial morphology following CsA treatment. As expected, CsA treatment significantly delayed MPTP opening. There was no significant difference in the rate of MPTP opening between CsA treated cells, and cells overexpressing Mfn2, MiD49 and MiD51. Fragmented mitochondria, in cells overexpressing hFis1, became significantly susceptible to early MPTP opening, compared to VC cells. These findings further support previously published data which showed that elongated mitochondria are less susceptible to MPTP opening, and indicate that elongated mitochondria can accommodate higher levels of ROS prior to pore opening (Ong et al., 2010). Notably, this assay does not accurately represent in vivo IRI conditions that lead to MPTP opening, which results predominantly from the collective effect of ROS, calcium ion and Pi overload (Halestrap, 2010).

3.6.2.3 MiD49 and MiD51 overexpression reduces cell death following SIRI For the first time, we were able to show that the overexpression of MiD49 or MiD51 protects HL-1 cardiac cell lines against SIRI (Figure 3.4.4.1) This result suggests that mitochondrial hyperfusion and the formation of both elongated and peri-nuclear collapsed mitochondrial networks, after MiD overexpression, protects HL-1 cells against SIRI. Similar protection has been previously observed in HL-1 cells, following the overexpression of the dominant negative Drp1 mutant (Ong et al., 2010). It is not clear to what degree a collapsed fused mitochondrial network contributed to the level of protection observed as there was no significant difference between cells with Mfn2 overexpression, which have a predominantly elongated mitochondrial phenotype, compared to cells overexpressing MiD49 or MiD51. Interestingly, despite the fact that hFis1 predominantly expressed fragmented 113 mitochondria, this had no significant effect on cell survival post SIRI, compared to VC cells.

3.6.3 Knockdown of MiD49 and MiD51 in cardiac cell lines

HL-1 and H9c2 cells were transfected with shRNAMiD49 and shRNAMiD51 plasmids, to KD the MiD proteins. The cells were co-transfected with fluorescent tagged mitochondrial proteins (mtBFP, mtGFP and mtRFP), to allow the identification of transfected cells and mitochondrial morphology. We were able to show that the KD of MiD49 and MiD51 KD achieved mitochondrial elongation in HL- 1 and H9c2 cardiac cell lines (Figure 3.5.2.1, Figure 3.5.2.2, Figure 3.5.2.3 and Figure 3.5.2.4). Interestingly, the elongated mitochondrial morphology observed were similar to the morphology observed following Mfn2 or dominant negative Drp1k38A overexpression. Unlike MiD overexpression, the knockdown of MiD49, MiD51 or both proteins did not result in the formation of hyperfused collapsed, or peri-nuclear collapsed clusters. A significant change in mitochondrial morphology compared to the VC group was only observed following the KD of MiD49 and MiD51. Despite the difference in the MiD49 and MiD51 expression in different organs (Liu et al., 2013), and their minor structural differences(Loson et al., 2014; Richter et al., 2014; Loson et al., 2015) suggesting that the proteins may have distinct roles in mitochondrial fragmentation, it is still not clear if there is no interplay between the two proteins’ activity. Our findings from the KD of MiD49 and MiD51 in HL-1 and H9c2 cells suggest that both proteins should be targeted to significantly reduce Drp1-mediated mitochondrial fragmentation in cardiac cells. This however still requires to be investigated further in adult cardiomyocytes, due to their highly complex and compact cellular structure.

3.6.3.1 MiD49 and MiD51 KD delayed MPTP opening The opening of the MPTP causes the collapse of the Δψm, leading to ATP depletion. This is due to the reverse activity of ATP synthase, which occurs due to the loss of mitochondrial membrane potential (Crompton et al., 1987; Rouslin et al., 1990). A more severe consequence of the pore formation and opening is the release of pro-apoptotic factors from the mitochondrial matrix into the sarcoplasm. This process is one of the main mediators of apoptotic cell death upon reperfusion (Griffiths & Halestrap, 1995; Hausenloy & Yellon, 2003). High levels of Ca2+, ROS and Pi within the mitochondrial matrix, are factors which lead to MPTP opening, upon reperfusion (Crompton et al., 1987; Crompton et al., 1988). The rate of MPTP

114 opening was assessed using laser induction TMRM breakdown, to form reactive oxygen species in the mitochondrial matrix. MPTP opening causes a significant increase in TMRM’s fluorescent signal, as the stain is no longer obtained within mitochondria, due to the loss of the Δψm (Hausenloy et al., 2003).

The KD of MiD49 and MiD51 significantly delayed MPTP opening in HL-1 cells. This was at a similar level to the positive controls, CsA treated cells or cells overexpressing mutant Drp1. These findings along with MiD49 and MiD51 overexpression data further support previously published data showing that elongated mitochondria are less susceptible to MPTP opening, suggesting that elongated mitochondria may be able to accommodate higher levels of ROS prior to MPTP opening (Ong et al., 2010).

Although the single KD of MiD49 and MiD51 delayed MPTP opening, this outcome was not statistically significant. Similar to changes in mitochondrial morphology studies, the knockdown of both proteins was required to have a significant effect on the cell, compared to VC cells. This outcome could not be reproduced in H9c2 cells. This may be due to the cells’ high sensitivity to this model, as the rate of MPTP opening occurred at a faster rate in these cells.

3.6.4 Knockdown of MiD49 and MiD51 in cardiac cell lines protects the cells against SIRI

3.6.4.1 MiD49 and MiD51 KD reduces cell death following SIRI Mfn2 overexpression, or insulin treatment of VC cells upon reperfusion, protected cells against SIRI when compared to VC cells. Similarly, inducing mitochondrial elongation following MiD49 and MiD51 KD, which lead to an increase in elongated mitochondria and protected HL-1 cells against SIRI (Figure 3.5.4.1). In contrast, hFis1 overexpression causes a significant increase in cell death following SIRI. This data supports previous findings from MPTP opening studies suggesting that elongated mitochondria are able to withstand a higher level of ROS accumulation, which is one of the factors that leads to MPTP opening during IRI (Halestrap, 2010). To further investigate MiD49 and MiD51 as potential targets of cardioprotection, and how their KD causes cells to become better equipped to deal with IRI, we studied the effects of their KD on mitochondrial morphology and activity in a real-time SIRI model.

115

3.6.4.2 MiD49 and MiD51 KD inhibits mitochondrial fragmentation during SIRI Following SIRI experiments, RT SIRI was characterised for HL-1 and H9c2 cells to analyse changes in mitochondrial morphology (Figure 3.5.4.2). HL-1 and H9c2 VC cells underwent full mitochondrial fragmentation during simulated ischaemia and started to fuse upon simulated reperfusion (Figure 3.5.4.3 and Figure 3.5.4.4). Mitochondria in HL-1 and H9c2 cells with MiD49 and MiD51 KD remained predominantly elongated, throughout simulated ischaemia. This finding further supports previous publications that classify MiD49 and MiD51 as fission proteins, as the reduction in their presence significantly reduced mitochondrial fragmentation in the real-time model of SIRI. As a control, cells were exposed to normoxic conditions for the same period of time as SIRI. Both VC and MiD KD cells remained predominantly elongated. Chamber conditions (oxygen, glucose and lactic acid levels) were monitored at each time point to ensure hypoxic and normoxic conditions were achieved, using a blood gas analyser machine (Figure 3.5.5.1). If these conditions were not achieved, or there was a leak from the chamber, images were excluded from the analysis.

3.6.4.3 MiD49 and MiD51 KD reduces calcium overload during SIRI The RT SIRI model was used to analyse changes in calcium levels during SIRI (Figure 3.5.5.2). Mitochondrial calcium-sensitive protein GCamP6, mtGCamP6, allowed quantification of mitochondrial matrix calcium levels. For the first time, we were able to show that there was a significant overall difference in mitochondrial matrix calcium levels between VC and MiD49 and MiD51 KD cells (N=8, interaction P=0.0325, Figure 3.5.5.2). There was a trend of increasing matrix calcium during hypoxia, in VC cells, however, this was not observed until reperfusion in MiD KD cells. MiD KD caused lower levels of calcium increase upon reperfusion and recovered much faster during reperfusion. Further investigation is required to gain a better understanding of this outcome, including measuring changes in cytosolic calcium levels and mitochondrial membrane potential during SIRI. The lower levels of mitochondrial calcium during RT SIRI in MiD49 and MiD51 KD cells may be due to a reduction in ER/ SR interaction. A reduction in SR and mitochondrial tethering has been shown to be cardioprotective due to a reduction in calcium uptake during IRI, in Mfn1 and Mfn2 deficient cardiomyocytes (Hall et al., 2016). Mitochondrial fission occurs at sites which have first undergone constriction by the ER (Friedman et al., 2011; Elgass et al., 2015). MiD49/MiD51 foci have been identified to interact with the ER, however, less than 40% of the ER-MiD interaction was associated to be 116 at constriction sites, and interaction was significantly reduced after the KD of both MiD49 and MiD51 (Elgass et al., 2015). This suggests that the MiD proteins are also involved in ER tethering. We speculate that MiD49 and MiD51 cells are less likely to undergo mitochondrial calcium overload during IRI, due to a reduction in ER-MiD interaction (Elgass et al., 2015).

3.6.5 Summary

These results revealed that overexpression of the MiD proteins promotes the formation of elongated mitochondrial tubules, which were predominantly hyperfused and in collapsed peri-nuclear clusters. The overexpression of MiD49, MiD51 or both resulted in a delay in MPTP opening and protected the cells against SIRI. As the effects observed by overexpression of MiD proteins is not physiological, our investigation was continued following the KD of the proteins, in cardiac cell lines. The KD of MiD49 and MiD51 induced mitochondrial elongation in cardiac cell lines. These cells were less susceptible to MPTP opening due to high mitochondrial matrix ROS levels. MiD KD in cells inhibits mitochondrial fragmentation during simulated ischaemia makes them less susceptible to cell death following SIRI. Lower levels of mitochondrial calcium in MiD KD cells during simulated ischaemia and the faster recovery of basal mitochondrial calcium may be a contributing factor in lower cell death following SIRI, as lower calcium levels reduce mitochondrial susceptibility to MPTP opening (Halestrap, 2010). Changes in mitochondrial ROS during RT SIRI, following MiD49 and MiD51 KD, remain to be investigated.

117

CHAPTER 4: Investigating the role of MiD49 and MiD51, as targets of cardioprotection: in vivo studies

4.1 Introduction

Mitochondria occupy up to 35% of cardiomyocytes’ volume, in the adult heart (Hom & Sheu, 2009). The abundance of mitochondria is essential for producing high levels of ATP, required for healthy cardiac function. Mitochondria are involved in mediating and regulating cellular activity during physiological or cellular stress conditions and are central in determining the fate of cardiomyocytes during IRI (Gustafsson & Gottlieb, 2008; Murphy & Steenbergen, 2008; Chan, 2012).

Mitochondrial dynamics is vital for preserving mitochondrial health and function (Bach et al., 2003; Frazier et al., 2006; Chan, 2012). Even though movement and changes in mitochondrial shape in cardiomyocytes are highly restricted, mitochondrial dynamic proteins are still highly expressed in the heart, highlighting the importance of their activity in maintaining normal cardiomyocyte function (Hom & Sheu, 2009).

Mitochondrial fragmentation is crucial for the maintenance of a healthy mitochondrial network. Highly damaged mitochondria, which may otherwise contaminate the mitochondrial network, are removed via mitophagy (Chen & Chan, 2009). Although this process is essential, inhibition of mitochondrial fragmentation in certain circumstances have been shown to be beneficial. Mitochondrial fission, mediated by Drp1, is an important step in apoptosis (Stowell et al., 1999; Frank et al., 2001). The expression of the dominant negative Drp1k38A mutants, deficient in GTPase activity, significantly promotes mitochondrial elongation in cells. These cells have been shown to be protected against the loss of ΔΨm, cytochrome c release and cell death following treatment with the pro-apoptotic agent, staurosporine (Li et al., 2000; Frank et al., 2001). Interestingly, similar protective outcomes have been observed in cell lines and animal models of IRI, as a result of inhibiting Drp1 mediated mitochondrial fragmentation (Ong et al., 2010; Wang et al., 2011).

There are currently no pharmaceutical agents capable of directly targeting Drp1, which could be administrated to inhibit its activity during IRI (Bordt et al., 2017). 118

Drp1 polymerisation at the mitochondrial surface requires the presence of OMM fragmentation proteins (Otera et al., 2010; Palmer et al., 2011), making these proteins the ideal targets for cardioprotection. From the mammalian mitochondrial fragmentation proteins identified so far, MiD49 and MiD51 are mitochondrial specific anchors of Drp1 (Palmer et al., 2013); therefore, inhibiting their interaction with Drp1 can prevent mitochondrial fragmentation, without interruption of other Drp1 functions.

The expression of the MiD49 is significantly higher than MiD51 in the adult heart and skeletal tissue (Liu et al., 2013). In part one of this chapter, whole body MiD49 KO C57BL/6 mice were used to investigate the effect of MiD49 genetic ablation on cardiac size and function assessed by echocardiography, and susceptibility to acute IRI, using the in vivo model of MI (Aim 1). In part two of this chapter, rAAV9 constructs were developed to KD of MiD49 and MiD51 specifically in the mouse heart, for in vivo experiments to investigate genetic ablation of both proteins, as targets of cardioprotection (Aim 2).

4.2 Research objectives and aims

4.2.1 Aim 1: Investigating the effect of MiD49 genetic ablation, as targets of cardioprotection: in vivo studies

Specific hypothesis: Genetic ablation of MiD49 will not affect baseline cardiac size and function, will not improve contractile reserve, will induce mitochondrial elongation, and reduce myocardial infarct size following acute IRI.

1) In part one of this chapter, we will investigate the effects in vivo of genetic ablation of MiD49 on cardiac size and function (including contractile reserve), mitochondrial morphology, and susceptibility to acute IRI.

2) To investigate the effect of genetic ablation of MiD49 on cardiac size and function and contractile reserve, using echocardiography imaging.

3) To investigate the effect of genetic ablation of MiD49 on susceptibility to acute myocardial IRI using an acute in vivo murine model of AMI.

119

4.2.2 Aim 2: Investigating the effect of cardiomyocyte MiD49 and MiD51 genetic ablation, as targets of cardioprotection: in vivo studies

We will investigate the effects in vivo of genetic ablation of both MiD49 and MiD51 on cardiac size and function (including contractile reserve), mitochondrial morphology, and susceptibility to acute IRI.

Specific hypothesis: Genetic ablation of both MiD49 and MiD51 will not affect baseline cardiac size and function, will improve contractile reserve, will induce mitochondrial elongation, and reduce myocardial infarct size following acute IRI.

The first objective of this section was to identify RNAi oligonucleotide sequences which can be used to design and produce rAAV9 constructs, to KD of MiD49 and MiD51, specifically in the mouse heart. For achieving this objective, the principle aims were:

1) To identify RNAi oligonucleotide sequences to KD MiD49 and MiD51, using Basic local alignment search tool (BLAST). Transient transfection of MEF cells and Immunohistochemistry will be used to identify the plasmids which induce the highest level of protein KD.

2) To design MiD49 and MiD51 specific AAV-9-cTnT-EGFP-RNAi (pENN.AAV.cTnT.PI.EGFP.RBG(p13720)Q) plasmids, using the oligonucleotide sequenced which achieve the highest level of protein KD.

3) To produce MiD49 and MiD51 targeting, cardiac-specific, rAAV9 constructs, to KD these proteins in the murine heart, using the AAV helper-free production system.

120

Aim 1: Investigating the effect of MiD49 ablation, as a target of cardioprotection: in vivo studies

The global MiD49 KO mouse transgenic strain was used to investigate the effect of MiD49 ablation on cardiomyocyte mitochondrial morphology, cardiac function and myocardial infarct size following acute IRI.

4.3 Materials and Methods

4.3.1 Experimental use of animals

Animal procedures at University College London were carried out after accreditation of Module 1-4 Home office license. As part of this, animal procedures were carried out in accordance with the United Kingdom Animal Scientific Procedures Act of 1986 and local guidelines. Additional guidance was provided by the Named Animal Care Welfare Officer (NACWO) of the biological services and University College London. MiD49 KO mice were housed in individually ventilated cages (IVCs), with standard chow provided ad libitum (maximum of 5 adult mice per cage). Mice were kept in 12 hour light/dark cycles. Daily care was provided by biological services staff at University College London.

The MiD49 KO colony was re-established in Singapore (Duke-NUS medical school), following the transportation of three breeding pairs from University College London’s animal housing facility. Animal work was carried out after Responsible Care & Use of Laboratory Animals (RCULA) license accreditation. All animal work was carried out in accordance with local guidelines, and the procedures were approved by the SingHealth Institutional Animal Care and Use Committee (IACUC). MiD49 KO mice were housed in individually ventilated cages (IVCs), with standard chow provided ad libitum (maximum of 5 adult mice per cage). Mice were kept in 12 hour light/dark cycles, and daily care was provided by biological services staff at Duke-NUS medical school vivarium facility.

4.3.2 Transgenic mouse background

The global MiD49 KO transgenic mouse strain was provided as a kind gift from a collaborator, Prof Michael Ryan (University of Monash). The transgenic colony was first established at La Trobe University, using inbred C57BL/6NTac mice. After a period of acclimatisation the MiD49 heterogeneous (Het,+/-), females received, were 121 set up with WT males of the same colony; however, the mice were unable to reproduce successfully.

To ensure that the colony could still be established before the females were no longer cable of breeding, the Het females were paired with commercially obtained 6- week old wild-type C57BL/6J males (ordered from Jackson laboratory). C57BL/6NTac mice are less commonly used in the U.K and were not available at the time. Subsequently, randomly selected Hets, from different parents were selected as the new breeders, and their progeny was sent to establish the colony in Singapore, for animal experiments.

The mouse MiD49 gene consists of 4554 base pairs, with four exons, located on chromosome 11 (Chromosome 11: 60,728,398-60,732,951,NC_000077.6). The MiD49 KO transgenic mouse has a deletion of 1762 base pairs (bp) within the gene sequence (deletion of exon two, three and four, base 60,730,273-60,732,034 which is replaced with the neomycin gene (NEO), at the site of deletion.

4.3.3 Genotyping

4.3.3.1 Biopsy collection Mouse ear biopsies were obtained soon after weaning (3-5 weeks old), to conduct genotyping of the MiD49 KO mice. Sample identification was carried out in accordance with the cage ID and the number of ear punches made in each ear. Five different ear punch codes were used (left, right, left and right, two left or two right) as no more than five mice were housed per IVC cage after weaning. Ear biopsies were collected using an ear puncher (ID001C, VetTech UK), which was wiped clean with 70% ethanol after each mouse, and the ear biopsies were placed in individual sterile eppendorfs, marked with the mouse ID. If processing could not be carried out immediately post collection, the ear biopsies were kept at -20ºC for a maximum of two weeks. The ear lysates were stored at -20ºC until completion of the study.

4.3.3.2 Crude DNA preparation Ear samples were lysed by the addition of 150µl of Direct PCR Lysis Reagent (401- E, Viagen Biotech), containing 0.3 mg/ml Proteinase K (p6556, Sigma). The ears, submerged in lysis buffer, were incubated overnight, in a heating block set to 55ºC. Post incubation, samples were vortexed and checked to ensure full lysis was accomplished. If further lysis was required, incubation for an additional 1 hour was carried out, at 55ºC. Post lysis, Proteinase K was denatured by incubation at 85ºC 122 for 45 minutes. Hairs within the samples were pelleted down (30s centrifugation, at 1000 RPM), and the supernatants containing DNA, were placed in new sterile tubes and stored at -20ºC.The Direct PCR lysis reagent allows storage of samples at - 20°C for one year without loss of efficacy.

4.3.3.3 DNA Amplification by Polymerase chain reaction Amplification of DNA, collected from ear biopsies, was carried out using polymerase chain reaction (PCR). PCR was carried out using a commercially available PCR kit (201223, Qiagen), containing Taq DNA polymerase and the Veriti Thermal Cycler (Applied Biosystems). Before PCR, the DNA concentration of each sample was measured (NanoDrop 1000 Thermo Scientific), to create a 50ng/µl aliquots in autoclaved ddH2O.

DNA primers, required for amplification, were designed using the MiD49 WT and MiD49 KO template DNA sequences. Primer sequences were designed with a high G/C content (48-60%, Table 4.3.3.1). Primers were purchased from Eurofins (MG Operon, Germany) and diluted to a stock concentration of 100 μM in autoclaved ddH2O (stored at -20ºC).

Primer Expected Reaction Primer Primer Sequence Tm Length product 5' to 3' (°C) (base) length (bp) WT Forward CTTCCTCTTGGCTAATGCTCGATTG 25 63.0 1 WT Reverse CAGTAGCTAGGGGCTGGCTGAAG 23 66.0 415 NEO Forward ATCTCCTGTCATCTCACCTTGC 22 60.3 2 NEO Reverse ATGATATTCGGCAAGCAGGC 20 57.3 300

Table 4.3.3.1 Property of MiD49 Primers Property of primers used for the genotyping of the MiD49 KO transgenic colony. The successful annealing of the WT forward, and reverse primers to the WT template DNA produces DNA products at an expected size of 415 base pairs. The successful annealing of the NEO forward, and reverse primers, to the KO template DNA, produces DNA products at an expected size of 300bp.

Samples and reagents were kept on ice during the preparation of PCR samples, and kept at 4°C post PCR. The master mix, consisting of the Qiagen PCR kit reagents (Qiagen, 2010), and autoclaved water, was prepared freshly on the day of the experiment. The optimised concentration of each reagent, primer and template DNA gave a final volume of 25µl/reaction (Table 4.3.3.2). PCR steps were based on primer’s Tm, PCR kit reaction temperatures, as well as gradient PCR reaction results (Table 4.3.3.1,Figure 4.3.3.1). With each experiment, one DNA negative control reaction was also carried out (22.5µl master mix with 2.5µl of autoclaved

123 ddH2O). Two separate PCR reactions were carried out for each mouse, using the WT and NEO primers (Figure 4.3.3.1).

Reagent Volume per sample (μl) Final concentration 10x CoralLoad PCR buffer 1x 2.50 (contains 15mM MgCl2) (1.5mM MgCl2) dNTP (10mM of each) 0.50 200μM of each Forward primer (10uM) 0.75 0.3μM Reverse Primer (10uM) 0.75 0.3 μM Taq Polymerase 0.125 - 5x Q Solution 5.00 1x

Autoclaved ddH2O 12.88 - Mouse DNA (50ng/µl) 2.50 5ng/µl

Table 4.3.3.2 MiD49 WT and KO PCR reaction mix PCR reaction mixtures were prepared fresh on the day of the experiment, using the Qiagen PCR kit. One DNA null sample was also created per experiment, to ensure reagents and primers were free of DNA contaminants.

124

Figure 4.3.3.1 MiD49 KO colony PCR steps During PCR, denaturation of the template DNA separates the double strands, exposing the nucleotides. The annealing of primers with the complementary template DNA nucleotide sequence, by the formation of hydrogen bonds, permits the amplification of the region of interest. Mouse samples were used for two separate PCR reactions, with two different annealing temperatures; A) PCR reaction with the WT primers, and B) PCR reaction with the NEO primers.

4.3.3.4 Gel electrophoresis Standard horizontal agarose gel electrophoresis was used for the separation and visualisation of PCR products, according to size. The CoralLoad buffer, within the PCR mix, permits direct loading of PCR samples into electrophoresis gels, and the visualisation of migrating DNA bands. Agarose gels were freshly prepared on the day of the experiment and left to set at RT (1.5% agarose in TAE buffer and 1x GelRed Nucleic Acid Gel Stain, 41003, Biotium). Once set, gels were placed in the Bio-Rad wide mini-sub cell GT system and submerged in TAE running buffer (Tris 40 mM, acetic acid 20mM and EDTA 1 mM). The first lane of each row was always used for the DNA molecular weight ladder (4µl of DNA ladder, SM0333

ThermoFisher, 3µl of Loading Dye, G190A Promega and 8µl ddH2O). 15µl of each PCR sample was loaded directly into the wells, with WT and NEO reactions for each

125 mouse placed next to each other (Figure 4.3.3.2). Once all samples were loaded onto the gels, the gels were run for 45-60 minutes, at 70V. DNA bands were imaged using the Bio-Rad ChemiDoc XRS+ (Figure 4.3.3.2). If DNA bands, other than the primers, were observed in the DNA null samples, PCR experiments were re-run with a new set of reagents and primers.

Figure 4.3.3.2 Representative genotyping result of the MiD49 KO mice The DNA samples from both PCR reactions from each mouse, containing WT or NEO primers, were loaded into two adjacent wells. Genotyping of the transgenic MiD49 KO mouse colony was determined by the presence/ absence of DNA products; WT DNA products at 415bp and NEO DNA product at 300bp. DNA bands: lane 1) DNA ladder, 2) WT band and 3) KO band. Reactions from a Het mouse sample produces DNA bands from both reactions (lane 2 and 3). A KO mouse sample only produces products of 300bp, in the presence of NEO primer (lane 5), and no products in the presence of the WT primer (lane 4); however, a WT mouse sample only produces DNA products of 415bp, in the presence of the WT primer (lane 6), and no products in the presence of the NEO primer (lane 7).

4.3.4 Cardiac mitochondrial phenotyping by electron microscopy

Cardiomyocyte baseline mitochondrial morphology of WT and KO mice, from the MiD49 KO transgenic mouse colony, was analysed using electron microscopy (EM) imaging.

Subsequent processing and EM imaging described here were undertaken with the assistance of Miss Khairunnisa Binte Katwadi (Duke-NUS medical school, Singapore).

126

4.3.4.1 Preparation of EM heart samples Five MiD49 WT and five KO mice, aged 8-12 weeks were anesthetised using isoflurane. Cervical dislocation was carried out, upon achieving anaesthesia, and hearts were excised.

The hearts were immediately placed in cold PBS (4ºC), to remove residual blood, and fixed by immersion in EM fixative buffer (2% Glutaraldehyde, 2% Paraformaldehyde (PFA) in 0.1 M sodium cacodylate buffer, in PBS). Hearts were stored at 4°C in EM fixative buffer, and processed for imaging within 72h. Post fixation, the samples were washed three times with 0.1M cacodylate buffer solution and trimmed down. For cell membrane preservation, the trimmed samples were incubated in 2% osmium tetroxide and potassium ferrocyanide, for 3h on a rotary shaker. The hearts were washed three times with ddH2O, before gradual dehydration, using increasing strengths of ethanol (Table 4.3.4.1). Resin infiltration was achieved by the incubation of samples in freshly prepared acetone and EPON resin mixture (Poly/Bed® 812 - BDMA Embedding Kit, Polysciences), followed by overnight incubation in pure EPON resin. The samples were transferred to the cutting edge of an EM mould, before the addition of resin. Subsequently, the samples in resin were left to set for 48h, at 65°C.

Once the hearts were set in resin blocks, excess resin was trimmed away with a razor. Ultra-thin sectioning at 90nm was carried out using a microtome diamond knife, at a speed of 1.2mm/second (Leica EM UC7). Sections floating in the trough were collected, using a mesh copper grid, and dried before uranyl acetate and lead citrate staining.

Dehydration step Incubation time (minutes) 1 25% Ethanol 5 2 50% Ethanol 10 3 75% Ethanol 10 4 95% Ethanol 10 5 100% Ethanol 10 6 100% Acetone 10 (twice)

Table 4.3.4.1 Electron microscopy sample dehydration Dehydration of samples, with increasing strengths of ethanol, permits the addition of resin to cells. The replacement of water is required to help preserve cellular structure during sectioning.

127

4.3.4.2 Electron microscopy Two-dimensional EM imaging was carried out using the JEOL JEM 2100 transmission electron microscope. Five randomly selected regions were imaged at a 4000x and 6000x magnification, displaying the interfibrillar mitochondria, at the longitudinal cross-section of cardiomyocytes.

4.3.4.3 Analysis of mitochondrial morphology Mitochondrial morphology of all interfibrillar mitochondria within 5 images, of randomly selected regions of the mice LV was assessed based on mitochondrial length, in comparison to sarcomere length at a relaxed state (2µm). Mitochondrial length >2µm was defined as elongated, equal to 2µm was define to have an intermediate mitochondrial morphology, and mitochondrial <2µm were defined as fragmented mitochondria (Ong et al., 2010).

4.3.4.1 Statistical analysis of mitochondrial morphology The mean percentage of each specific mitochondrial morphology per heart, based on interfibrillar mitochondrial length, was calculated and expressed as mean ±SEM. Statistical analysis was carried out by unpaired t-test comparison of the average MiD49 KO and WT control mitochondrial length and morphology (GraphPad Prism 6 Software, La Jolla, CA). The difference between groups and the control condition were considered statistically significant if P<0.05. Statistical significance indicated by asterisks (*P<0.05, **P<0.01 and ***P<0.001). No statistical significance (NS) was reported where P>0.05.

4.3.5 Cardiac phenotyping by echocardiography

The cardiac function of MiD49 KO and WT control mice, during basal and acute stress conditions, was assessed using echocardiography. The assessment of the cardiac phenotype by echocardiography was carried out to identify any overt cardiac phenotypes which could be caused as a result of MiD49 genetic ablation.

Echocardiography and analysis of raw data were carried out with the assistance of Miss Nicole Gui Zhen Tee (National Heart Research Institute | National Heart Centre Singapore).

4.3.5.1 Preparation of mice for echocardiography imaging Prior to echocardiography, mice were anesthetised with 2% vaporised isoflurane in 1l/min oxygen, using an induction chamber. Whilst unconscious the mice were 128 weighed, and body hair covering the chest was shaved. Mice were placed in a supine position, on a heated electrocardiogram (ECG) monitoring platform (Vevo Imaging Station). Thereafter, the mice were kept in a steady state of anaesthesia by the administration of 0.6-1.2% isoflurane, with 1l/min of oxygen, using a rodent face mask. Acoustic coupling gel was applied to the legs and feet, and secured onto the ECG electrodes, using surgical tape. Further hair removal of the chest region was carried out, using hair removal cream.

4.3.5.2 Echocardiography imaging during basal conditions Mice from the MiD49 KO colony underwent echocardiography imaging, between the ages of 8-13 weeks old. Echocardiography was undertaken using a Vevo 2100 echocardiography machine with an MS 400 18-38MHz probe (Fujifilm VisualSonics). The minimal concentration of isoflurane was administrated (0.6-1.2%, in 0.8-1.2 l/min oxygen), to maintain a steady heart rate (HR, 410-510 beats per minute (BPM)). Imaging was only carried out when a constant HR was stable. Acoustic coupling gel was applied to the thorax and two-dimensional (2D) imaging was carried out using the parasternal short axis view at papillary level (SAX, M-mode), and two-dimensional parasternal long-axis mid-level view (2D PLAX, visualising the aortic valve and apex). Pulsed wave Doppler was used for measuring the blood flow velocity, at the aortic arch (Figure 4.3.5.1). Measurements for the following were collected from images captured on cine loops to assess the cardiac phenotype of WT and MiD49 KO mice:

1. Anatomical measurements – The left ventricle (LV) internal dimensions, anterior and posterior wall thickness, at during end diastole and systole, and LV mass was measured from SAX images (M mode).

2. Heart Rate – This is the average number of heart beats per minute, calculated from the ECG readings, once a steady HR was achieved under anaesthesia.

3. Stroke volume (SV) – This is the measure of the volume of blood ejected out of the LV per contraction, calculated by the subtraction of the blood remaining in the ventricle at the end of systole (end-systolic volume, ESV), from the volume of blood that was in the ventricle at the end of diastole (end-diastolic volume, EDV).

푆푉 = 퐿푉퐸퐷푉 − 퐿푉퐸푆푉

129

4. Cardiac output (CO) – This is the measure of the volume of blood, pumped out of the heart per minute. This is calculated by the multiplication of BPM by the SV.

퐶푂 = 퐻푅 푥 푆푉

5. LV Ejection fraction (EF) – This is the measure of the percentage of blood leaving the left ventricle with each contraction. This is calculated the using the LV end- diastolic volume (LVEDV) and the end-systolic volume (LVESV).

퐿푉퐸퐷푉 − 퐿푉퐸푆푉 퐿푉퐸퐹 = ( ) × 100 퐿푉퐸퐷푉

6. LV Fractional shortening (FS) – This is the reduction of the LV diameter, from the LV end-diastolic diameter (LVEDD) to the LV end-systolic diameter (LVESD).

퐿푉퐸퐷퐷 − 퐿푉퐸푆퐷 퐿푉퐹푆 = ( ) × 100 퐿푉퐸퐷퐷

7. Peak aortic blood flow velocity- This is the highest velocity in which the blood is ejected from the left ventricle. This was measured as the average maximum velocity from 3 velocity-time traces, from pulsed wave Doppler images.

130

Figure 4.3.5.1 Echocardiography imaging modes A) SAX (parasternal short axis at mid-level) M-mode was used for anatomical measurements during diastole (D) and systole (S): D1) anterior LV wall thickness during end diastole, D2) LV internal diameter during end diastole, D3) posterior wall thickness during end diastole, S4) anterior LV wall thickness during end systole, S5) LV internal diameter during end systole, S6) posterior LV wall thickness during systole and 7) echocardiography papillary muscle artefact. Average HR was measured using ECG readings. B) Pulsed wave Doppler was used for measuring the peak aortic blood flow velocity, at the aortic arch. C) PLAX (parasternal long-axis mid-level) view of the LV chamber, visualising the aortic valve and apex. D) SAX (parasternal short axis at mid-level) view of the LV chamber.

4.3.5.3 Echocardiography imaging during cardiac stress Isoproterenol is a β-adrenergic agonist, capable of causing a significant increase in HR and ventricular contractility as a result of β1 and β2-receptor activation (Hoit et al., 1997; Davel et al., 2014). Isoproterenol is often used in chronic high doses to induce cardiac remodelling, in animal models of heart failure (Carll et al., 2011). The use of the drug in low doses can be used to identify irregular cardiac functions and contractile reserve, which may not be evident at basal conditions, in transgenic mice (Rottman et al., 2007).

131

Post-echocardiography at basal level, the mice received a 4ng/g; intraperitoneal bolus of isoproterenol (16504, Sigma) in sterile normal saline solution (0.9% NaCl). This protocol was previously established in our laboratory, using WT C57BL/6J mice, and was capable of causing cardiac stress in different transgenic mouse strains (Dongworth et al., 2014; Hall et al., 2016). The echocardiography protocol was repeated, 5 minutes post injection, to measure the same parameters outlined in section 4.3.5.2, under isoproterenol-induced cardiac stress.

4.3.5.4 Statistical analysis of echocardiography analysis Echocardiography measurements, mentioned in section 4.3.5.2, were collected as triplicates, from each mouse. The mean of each echocardiography measurement was calculated and expressed as mean ±SEM. Statistical analysis was carried out by one-way ANOVA followed by Bonferroni comparison test (GraphPad Prism 6 Software, La Jolla, CA). Difference between groups and the control condition were considered statistically significant if P<0.05. Statistical significance indicated by asterisks (*P<0.05, **P<0.01 and ***P<0.001). No statistical (NS) significance was reported where P>0.05.

4.3.6 Non-recovery acute Ischaemia-reperfusion in vivo model

The non-recovery Ischaemia-reperfusion in vivo model was used to assess infarct size after left anterior descending (LAD) occlusion of 45 minutes, followed by 2 hours of reperfusion, of WT and MiD49 KO mice.

Subsequent in vivo surgical procedures described here was undertaken by Dr Sauri Hernandez Resendiz (Duke-NUS medical school, Singapore).

4.3.6.1 Surgical preparation and anaesthetics Male mice from the MiD49 KO colony, 8-13 weeks old, were injected with an intraperitoneal bolus of an anaesthetic cocktail containing: ketamine 50mg/kg (Vetalar, BoehringerIngelheim, Bracknell, UK), xylazine 10 mg/kg (Rompun, Bayer, Newbury, UK) and Heparin sodium 8mg/kg. Once anaesthesia was confirmed (loss of the toe-pinch reflex), hair at the site of surgery was removed using a small electrical clipper. Loose hair was removed using moistened sterile gauze, and the surgical site was disinfected with a sterile cotton-tipped applicator, soaked in 70% ethanol or Betadine surgical scrub. The mice were then placed on a heated surgical platform. The legs and arms were secured onto the surgical platform’s ECG

132 sensors, using surgical tape, and the core body temperature was measured with a rectal probe.

4.3.6.2 Non-recovery surgical procedure Intubation – Anaesthesia was maintained during surgery by the inhaled isoflurane (0.5%-1% in 1l/min of oxygen) via endotracheal Intubation, to achieve a stable heart rate of 250-300 BPM. To ensure that the ventilation cannula was inserted correctly into the trachea, a small skin incision was made at the midline of the salivary glands, to expose the trachea. Once the cannula could be directly visualised inside the trachea and regular chest movements were observed, its position was secured with a tying a surgical knot around the trachea (Figure 4.3.6.1 A). Artificial respiration was provided using a PhysioSuite® for Mice & Rats, (Kent Scientific), to keep the respiration within 100-150 strokes/min, and the tidal volume of 120-160μl. Mice were intubated within 20 minutes of receiving the single intraperitoneal aesthetic bolus, which was injected prior to surgical preparation. After intubation, the cut was loosely closed by bringing the skin back together, and covered with a moist cotton pad, to prevent tissue desiccation. Mice were covered with a sterile disposable surgical blanket, only leaving the site of surgery exposed, to help maintain constant body temperature during surgery, (Figure 4.3.6.1 B). Five minutes of stabilisation was required, to achieve a stable tidal volume, before the first chest incision was made.

Thoracotomy – A 1cm incision was made on the left side of the thorax, between the Xiphoid and the Axilla. Next, another small incision was made to separate the muscles directly above the thoracic cavity, exposing the ribs. Surgical scissors were used to pierce a small hole into the intercostal muscle between the left third and fourth ribs, directly above the LV. To reduce the risk of puncturing the heart, the opening into the thoracic cavity was enlarged, by gently opening and closing a blunt Dumont forceps within the cut (Figure 4.3.6.1 C). 0.5cm chest retractors were used to keep the chest open during the procedure (Figure 4.3.6.1B).

Ischaemia-reperfusion – The occlusion of the left anterior descending coronary artery (LAD) is often used in mouse models of MI, which has been optimised since it was first established in 1995 (Michael et al., 1995; Fisher & Marber, 2002; van Laake et al., 2007). To aid the identification of the LAD (approximately 2mm below the left atrium), the pericardium was removed, with blunt forceps, while viewing the heart with a standard dissection microscope (Figure 4.3.6.1 C). A curved 3/8 circle suture needle was inserted into the LV, using a small needle holder (roughly 2mm

133 below the tip of the left atrium), and the 8-0 silk suture was passed under the LAD. To induce LAD occlusion, a small polyethylene tube was placed on the surface of the LV (1mm in diameter and 5mm in length) and was tightly tied down with the suture (Figure 4.3.6.1 D). The use of a small tube helps to protect tissue integrity and allow successful reperfusion once it is removed. Achieving LAD ischaemia resulted in the pallor pink/grey appearance of cardiac tissue and a reduction of LV contraction below the site of occlusion, as well as ECG ST-segment elevation. LAD occlusion was maintained for a period of 45 minutes, as this has been previously characterised to achieve an infarct size of 45-50% in our laboratory, with WT C57BL/6 mice. During this period the mice were monitored, and the chest opening was covered with a moist cotton pad, to prevent tissue desiccation.

Reperfusion was achieved by the release of the polyethylene tube, placed above the LAD. Upon reperfusion, the rate of LV contraction increases and the return of blood to the ischaemic region restores the red colour of the myocardium. Upon reperfusion, the chest opening was once again covered, with a moist cotton pad to prevent drying and damage to the tissue. Mice were monitored during the two hours of reperfusion, and small volumes of saline solution would be applied to the exposed tissue and cotton pad. The open suture was not removed from the heart, as it was essential to maintain its position for in vivo histological staining.

In vivo Histological staining of hearts – Histological staining was carried out in vivo, via the inferior vena cava. Two sutures were placed around the inferior vena cava, both above the renal vessels. The suture positioned just above the renal vessels was used to occlude the vessel, whilst the other suture was used to secure the 20 gauge needle in place during staining. Prior to staining, the inferior vena cava was tied off, and a small 45° incision was made into the vessel to insert the needle between the two sutures. Once the top suture was tightened to secure the needle within the inferior vena cava, the system was flushed with 10ml of pre-warmed saline solution (0.9% NaCl, at 37°C). Once the residual blood was removed from the heart and the coronary vessels, the LAD was re-occluded with the suture under- running the LAD, before the perfusion of 0.5% Evans blue (10ml). Evans blue staining is used to identify the area at risk (AAR) which was achieved by coronary occlusion. Following staining, the hearts were frozen at -20°C. A sharp scalpel was used to section the hearts, into five even transverse slices, starting from the apex to the top of the ventricles. The hearts slices were then incubated in 1% 2,3,5- Triphenyltetrazolium chloride (TTC, in PBS) for 30 minutes, at 37°C. TTC staining is 134 used for the identification of infarcted tissue. The slices were washed with ddH2O, before being fixed in 10% buffered formalin solution, at RT. The sections were imaged with an Epson paper scanner and analysed using Image J (version 1.8 National Institutes of Health, USA). The analysis was carried out by one observer, blinded to the experimental groups, to identify the infarct size (IS) and LV AAR (AAR percentage= (LV area- Evans blue stained area / LV area) x100, and IS percentage = (IS area/ AAR) x100).

Figure 4.3.6.1 Representative images of non-recovery murine acute myocardial Ischaemia-reperfusion in vivo surgery model A) Intubation was carried out by the insertion of the cannula into the trachea, via a small incision in the neck. B) Mice were covered with a sterile disposable surgical blanket, only leaving the site of surgery exposed, to help maintain constant body temperature. C) The LAD was exposed following an incision into the thoracic cavity, which was kept open using retractors. D) A silk suture, passed under the LAD, was used to hold a polyethylene tube tightly against the LAD, to cause occlusion.

135

4.3.6.3 Statistical analysis of MI IS and AAR measurement were calculated and expressed as mean ±SEM. Statistical analysis was carried out using an unpaired t-test (GraphPad Prism 6 Software, La Jolla, CA). Difference between groups and the control condition were considered statistically significant if P<0.05. Statistical significance indicated by asterisks (*P<0.05, **P<0.01 and ***P<0.001). No statistical (NS) significance was reported where P>0.05.

4.4 Results

4.4.1 Cardiac mitochondrial phenotyping of whole body MiD49 KO mice

Electron microscopy of five MiD49 KO and five WT littermate control hearts were performed, to identify the baseline cardiac mitochondrial morphology of these mice (Figure 4.4.1.1 A and B). The length of all interfibrillar mitochondria within randomly selected regions of the LV, five images per heart, were measured to calculate the mean mitochondrial length in WT and MiD49 KO hearts (Figure 4.4.1.1 C). Following this, mitochondrial morphology classifications were carried out based on mitochondrial length in comparison to the relaxed sarcomere length (fragmented mitochondrial morphology <2µm, elongated mitochondrial morphology >2µm, and intermediate mitochondrial morphology =2µm in length, Figure 4.4.1.1 D). This protocol was previously established and validated in our laboratory (Ong et al., 2010; Dongworth et al., 2014). There was a significant difference in mitochondrial morphology between the WT and MiD49 KO mice, when mitochondrial morphology was characterised into the set categories described by Ong et al., 2010, with a higher proportion of elongated mitochondria in MiD49 KO cardiomyocytes, compared to WT cardiomyocytes during basal conditions (WT 11.11±1.7% N=5 and MiD49 KO 17.68±2.0% N=5, P=0.0384, Figure 4.4.1.1D). However, there was no overall significant difference of the mean interfibrillar mitochondrial length between WT and MiD49 KO hearts (WT 1.42±0.05µm N=5 and MiD49 KO 1.54±0.06µm N=5, P=0.1251, Figure 4.4.1.1 C).

136

Figure 4.4.1.1 Cardiac mitochondrial morphology of MiD49 KO mice Representative EM images of WT (A) and MiD49 KO (B) cardiomyocytes, at baseline. C) There was no significant difference in interfibrillar mitochondrial length between WT and MiD49 KO hearts (P>0.05). Mitochondrial morphology was determined by the comparison of mitochondrial length, to sarcomere length during relaxation (2µm). D) There was a significant difference in mitochondrial morphology between the two groups, with a higher proportion of fused mitochondria in MiD49 KO cardiomyocytes, compared to WT cardiomyocytes during basal conditions (P=0.0384). Statistical significance assessed by unpaired t-test of MiD49 KO values (N=5), compared to the WT control (N=5). EM imaging was undertaken with the assistance of Miss Khairunnisa Binte Katwadi (Duke-NUS medical school, Singapore).

137

4.4.2 Cardiac phenotyping of MiD49 KO mice during basal and cardiac stress conditions

Anatomical measurements were made using M-mode echocardiography of the left ventricle, in parasternal short-axis view (Figure 4.4.2.1). Echocardiography of MiD49 KO mice and WT littermate control mice demonstrated that there was no significant difference in the LV anterior wall thickness during end diastole and systole (Figure 4.4.2.2 A and B, N=9). There was no significant difference in the LV posterior wall thickness during end diastole, between WT and MiD49 KO mice; however, MiD49 KO hearts had significantly higher wall thickness during end systole at baseline, compared to WT hearts (Figure 4.4.2.2 D). Isoproterenol treatment (4ng/g) had no statistically significant effect on LV anterior and posterior wall thickness during end diastole, in both groups (Figure 4.4.2.2 A and C); however, there was a significant increase in posterior wall thickness during end systole, compared to the baseline in both groups (Figure 4.4.2.2 B and D). There was no significant increase in LV mass, in MiD49 mice when compared to WT littermate controls, (Figure 4.4.2.2 E).

Figure 4.4.2.1 Representative echocardiographic phenotyping of MiD49 KO mice M-mode echocardiography of the left ventricle, in parasternal short-axis view, of WT and MiD49 KO hearts at baseline (A and C), followed by isoproterenol treatment (4ng/g, B and D). Echocardiography and analysis of raw data was performed with the assistance of Miss Nicole Gui Zhen Tee.

138

Figure 4.4.2.2 LV anatomical measurements of MiD49 and WT hearts Anatomical measurements of the WT and MiD49 KO hearts were made using echocardiography imaging. There was no significant difference in the LV anterior and posterior wall thickness of WT and MiD49 KO mice during end diastole, at baseline and isoproterenol treatment (A and C). There was no significant difference in LV anterior wall thickness during end systole between WT and MiD49 KO mice at baseline (B); however, LV posterior wall thickness during end systole was significantly higher in MiD49 KO mice, compared to WT mice, at baseline (D, P<0.05). Isoproterenol treatment (4ng/g) caused a significant increase in anterior and posterior wall thickness during end systole in both groups, compared to the baseline (B and D, P<0.05). There was no significant difference in LV mass between WT and MiD49 KO hearts at baseline or after isoproterenol treatment (E). Statistical significance was assessed by one-way ANOVA followed by Bonferroni comparison test between WT and MiD49 KO hearts (N=9/per group).

139

There was no significant difference between heart rate, stroke volume, cardiac output and peak aortic flow velocity between the MiD49 KO mice and their WT littermates at baseline. Interestingly, although there were minor changes observed, there was no significant increase in heart rate, stroke volume, cardiac output and peak aortic flow velocity following 4ng/g isoproterenol treatment, in both groups (data summarised in Table 4.4.2.1 and Figure 4.4.2.3).

Heart Rate Stroke Volume Cardiac output Aortic flow velocity (BMP) (µl) (ml/min) (cm/s) WT Basal 450.40±9.1 20.71±1.5 9.33±0.7 112.60±10.8 WT Iso 488.10±9.2 22.69±1.7 11.12±0.9 140.00±8.8 KO Basal 479.20±9.4 26.61±2.5 12.75±1.3 129.00±12.5 KO Iso 493.00±0.5 27.86±2.9 13.80±1.5 142.80±12.6

Table 4.4.2.1 Echocardiography cardiac phenotyping summary

140

Figure 4.4.2.3 Echocardiography Cardiac phenotyping of WT and MiD49 KO mice There was no significant difference in heart rate (A), stroke volume (B), cardiac output (C) and peak aortic flow velocity (D) between the MiD49 KO mice and their WT littermates at baseline or following 4ng/g isoproterenol treatment (N= 9/per group). Statistical significance was assessed by one-way ANOVA followed by Bonferroni comparison test between WT basal with KO basal, WT basal with WT Iso and KO basal with KO Iso (P>0.05). Echocardiography and analysis of raw data was performed with the assistance of Miss Nicole Gui Zhen Tee.

LV fractional shortening and LV ejection fraction measurements were made from M- mode echocardiography in SAX view and 2D measurements made from the PLAX view. There was no significant difference between WT and MiD49 KO LVFS and LVEF at baseline. Measurements from PLAX view, compared to M-Mode SAX view produced significantly lower FS values, however, the same trend was observed with both methods. Isoproterenol treatment caused a significant increase in LVFS and LVEF within each group.

141

Figure 4.4.2.4 LV fractional shortening and LV ejection fraction measurements of WT and MiD49 KO mice A) M-mode echocardiography in SAX view measurements showed that there is no significant difference in LVFS between WT (35.71±2.3%) and MiD49 KO (43.26±3.1%) hearts, at baseline. There was a significant increase of LVFS in WT and MiD49 KO hearts post isoproterenol treatment (WT Iso 49.54±3.3% and MiD49 KO Iso 55.61±1.7%). B) Although lower values of the LVFS were collected from 2D PLAX view images, the same pattern was observed during basal conditions (WT 14.37±1.4% and MiD49 KO 17.62±1.5%), followed by a significant increase in LVFS after isoproterenol treatment (WT Iso 21.25±1.6% and MiD49 KO Iso 25.15±1.5%). Ejection fraction measurements from SAX (C) and PLAX (D) view produced very similar measurements, with no significant difference in LVEF between WT and MiD49 KO mice. Isoproterenol treatment caused a significant increase in EF, in both groups (N= 9/per group). Statistical significance was assessed by one-way ANOVA followed by Bonferroni comparison test between WT basal with KO basal, WT basal with WT Iso and KO basal with KO Iso. Echocardiography and analysis of raw data was performed with the assistance of Miss Nicole Gui Zhen Tee.

4.4.3 Susceptibility of MiD49 KO mice to ischaemia-reperfusion injury

To identify if MiD49 KO mice were less susceptible to acute myocardial IRI, MiD49 KO mice and their WT littermates underwent in vivo non-recovery acute myocardial ischaemia-reperfusion surgery, in which the LAD was occluded for 45 minutes, followed by 2 hours of reperfusion. The genetic ablation of MiD49 had no significant effect on cardiac susceptibility to myocardial infarction upon IRI, compared to WT littermates (Figure 4.4.3.1).

142

Figure 4.4.3.1 MI in MiD49 KO mice Representative images of WT (A) and MiD49 KO (B) short-axis heart slices following dual staining with Evan's blue and TTC to stain the AAR and Using a small needle holder IS, respectively. C) There was no significant difference in infarct size between WT (40.69±0.6%, N=2) and MiD49 KO (43.53±5.75% N=4) hearts, post 45 minutes of ischaemia and 2 hours of reperfusion (P=0.755). Statistical significance was assessed by one-way ANOVA followed by unpaired t-test between WT and MiD49 KO hearts. Surgical procedures described here were undertaken by Dr Sauri Hernandez Resendiz.

143

Aim 2: Investigating the effect of cardiomyocyte MiD49 and MiD51 ablation, as targets of cardioprotection: in vivo studies

4.5 Background

Adeno-associated viruses (AAV) are naturally occurring, non-enveloped capsid parvoviruses, roughly 22nm in size (Hoggan et al., 1966). They were first identified and isolated during electron microscopy studies of human and simian adenoviruses, in 1965 (Atchison et al., 1965; Hoggan et al., 1966). These small DNA containing particles were first thought to be subunits or contaminants of adenoviruses. AAVs were later described as defective viruses, given that they cannot replicate without the presence of other helper viruses, such as herpes or adenoviruses, in order to cause lytic cycle activation (Atchison et al., 1965; Hoggan et al., 1966). So far 13 AAV serotypes have been identified, containing over one hundred different AAV genome sequences, sharing from 65-99% sequence identity and over 95% structural identity (Rayaprolu et al., 2013). Where they differ significantly is tropism for various tissues, which can be used to target specific tissue during genetic therapy (Drouin & Agbandje-McKenna, 2013).

A high percentage of individuals are exposed to AAVs during childhood and adolescence. However, their presence has not yet been associated with any pathology in humans (Boutin et al., 2010). Compared to adenoviruses and lentiviruses, their minimal vector-related toxicity makes them a superior choice for stable, long-term, gene transfer therapeutic applications (Boutin et al., 2010; Mingozzi & High, 2013); however, efficiency is limited in fast dividing tissue as AAVs are replication deficient.

4.5.1.1 The AAV Genome AAVs contain a single strand DNA (ssDNA), approximately 4.7kbs in size. Due to their small capsid size, AAVs’ ability to package some therapeutic vectors is limited. AAVs are capable of packaging vectors of up to 5.2kb. However, the packaging efficiency of vectors of this size, or below 4.1kb is significantly reduced, making them very difficult to produce (Dong et al., 1996).

144

Even with this size limitation, AAV DNA sequence codes for several proteins which are required for replication and structure formation. The open reading frame (ORF) encodes for two genes, Rep and Cap (Drouin & Agbandje-McKenna, 2013). The alternative splicing of the Rep transcript leads to the expression of four proteins, required for AAV DNA replication (Rep78, Rep68, Rep52 and Rep40), gene insertion and expression (Muzyczka, 1992). The Cap gene encodes for three proteins, required for capsid formation (VP1, VP2 and VP3), and of defective of virus production (Hermonat et al., 1984; Van Vliet et al., 2008). Inverted terminal repeats (ITR) flank the ORF at each end (145 nucleotides long), which form a T-shaped hairpin structure by creating self-base pairs (Lusby et al., 1980). The ITRs are required for self-priming and the encapsidation of the AAV genome, making them an essential component for recombinant AAV (rAAV) production (McLaughlin et al., 1988; Yan et al., 2005).

4.5.1.2 AAV Capsid Structure AAVs are non-enveloped viruses of approximately 22nm in size, that exist as several naturally occurring serotypes (Hoggan et al., 1966; Mingozzi & High, 2013). AAV capsids have a T=1 icosahedral symmetry and have a diameter of approximately 250Å (Rayaprolu et al., 2013). Sixty VP monomers, coded for by the Cap gene, are required to form the icosahedral capsid structure. The capsid is formed from three different VP proteins, VP1, VP2 and VP3, made from alternative splicing of the Cap mRNA (Van Vliet et al., 2008). The VP3 monomer is present at a much higher ratio, compared to VP1 and VP2 (≈85%). VP1 and VP2 are essential, as they form projections which are required for receptor interaction and localisation in the nucleus (Grieger et al., 2006; Sonntag et al., 2006). Minor differences in the capsid structure can alter viral infectivity, tissue tropism and cell transduction, which can be further manipulated to improve efficacy and tissue specificity of a genetic- based therapy (Grimm & Kay, 2003; Burger et al., 2004; Grimm et al., 2008; Boutin et al., 2010). Capsid dismantling and Assembly has been indicated to take place within the host nucleus (Hunter & Samulski, 1992; Wistuba et al., 1997).

4.5.1.3 AAV life cycle The AAV serotypes’ tissue tropisms predominantly depend on their capsid interaction with cells’ surface receptors and co-receptor, which leads to their attachment and intake into the cells. This depends on VP1 and VP2 capsid proteins structural motifs’ interaction with the host cell’s surface receptors (Sonntag et al., 2006). Once attached, receptor-mediated endocytosis of the virus occurs from 145 clathrin-coated pits (Duan et al., 1999). A reduction in the endosome pH releases the AAV into the cytosol, and the capsid enters the nucleus soon after. Virus particles in the nucleus can be detected within 2 hours of infection (Bartlett et al., 2000).

In the absence of helper viruses, the viral DNA remains at lysogenic state, during which it cannot replicate (Hermonat & Muzyczka, 1984). The AAV vector remains episomal, and unless transduction has occurred in a terminally differentiated cell, its presence is eventually diluted down in dividing cells. As AAV do not have DNA polymerases, they rely upon the host cells’ intrinsic DNA repair and recombination machinery to form stable, double-stranded episomes (Schnepp et al., 2005). This is possible due to the hairpin shape formation of the self-priming ITRs, which resemble double-strand break repair intermediates (Choi et al., 2006). Integration into the host genome can occur, but these events are sporadic and rare. Some wild-type AAV vectors, such as AAV2, can achieve site-specific DNA integration, which allows long-term vector gene expression (Kotin et al., 1990). Lytic cycle activation occurs in the presence of helper viruses, which allows transcription of the Rep and Cap genes required for virus assembly, in the nuclei (Hermonat & Muzyczka, 1984).

The use of adeno-associated viruses in clinical gene therapy studies has become favourable, due to their non-pathogenic and very mild proinflammatory risk profile (Vandendriessche et al., 2007). Recombinant AAVs lack the viral coding sequence; however, if the capsid structure is unaltered, there is a potential chance of an immune response, in patients that have been previously exposed to the wild-type virus (Mingozzi & High, 2013).

4.5.2 Recombinant AAV production

Recombinant AAVs lack the viral Rep and Cap genes. They are specially designed to deliver transgenes into the target cells’ nuclei. The capsid serotype with the highest tropism for the target tissue is selected for therapy, which can be further modified to optimise infectivity (Buchholz et al., 2015). After the viral capsid dismantles in the nucleus, the vector DNA is replicated to form a double strand and remains episomal (Schnepp et al., 2005). This can be desirable during therapy, as the insertion of the sequence may affect neighbouring genes at the site of integration. Engineering vectors capable of achieving site-specific homologous recombination, using intrinsic DNA repair pathways, allows the long-term expression of the vectors; which are tissue-specific, and expression can be induced or 146 controlled, depending on the selected region (Russell & Hirata, 1998; Vasileva & Jessberger, 2005; Miller, 2011). The production of AAVs containing the therapeutic transgene can be carried out using two different methods, the helper-required and the helper-free method (Figure 4.5.2.1). Both of these methods have several advantages and disadvantages.

4.5.2.1 Helper-required AAV production The helper-required AAV system is dependent on the presence of a helper virus, for AAV production (Geoffroy & Salvetti, 2005). A packaging cell line is co-transfected with the vector plasmid, containing the gene of interest flanked with the AAV ITRs, and a second plasmid containing the Rep and cap genes, deficient of the ITRs. At the same time, the cells are infected with a helper, such as adenoviruses, to provide the genes required for AAV production (Matsushita et al., 1998). The disadvantage of this system is that the final product is a mixture of AAV and adenoviruses. Heat activation is often used to deactivate the adenoviruses; however, their presence may still cause an immune response, and their presence during production can be dangerous (Tenenbaum et al., 2003).

4.5.2.2 Helper-free AAV production A safer approach for rAAV production is the helper-free method, which is carried out using a three-plasmid transient co-transfection system, in a packaging cell line. Human embryonic kidney 293 (HEK293) cells are commonly used as they already contain two of the genes normally provided by a helper virus (E1A and E1b), required for AAV replication (Graham et al., 1977). The transgenic plasmid contains the sequence of interest, flanked by ITRs (Samulski et al., 1989). The second plasmid encodes the Rep and Cap genes, which are not flanked by ITRs; and therefore, they cannot be packed into viral capsids during production, to form WT AAV constructs (Samulski et al., 1989). The third plasmid encodes for the remaining adenoviral helper genes required for AAV replication; E2a, E4 and VA RNA which enhance Rep and Cap gene transcription, and stabilise the viral mRNA, for translation (Matsushita et al., 1998). Although this form of AAV production is highly labour intensive, the method produces high viral titres, free of WT AAV and helper virus contamination (Matsushita et al., 1998).

147

Figure 4.5.2.1 Recombinant AAV production methods Recombinant AAV constructs can be produced in the presence of helper viruses, such as adenoviruses (A), or the absence of a helper virus (B), by which, the helper genes required for AAV production are provided using a helper plasmid.

4.5.3 The cardiotropic AAV9 serotype

AAV2 has been extensively researched in understanding the activity of adeno- associated viruses, and designing new rAAVs. However, their prolonged lag phase before expression and low myocardial transduction is a limiting factor for cardiac genetic therapy (Prasad et al., 2007). The recently isolated AAV serotype from human tissue, AAV9, has a strong tropism for cardiac cells, achieving high transduction in myocardial cells (Gao et al., 2004; Inagaki et al., 2006; Pacak et al., 2006; Prasad et al., 2011). Cardiac gene transfer using AAV9 is achieved at a superior degree, compared to other serotypes, causing global and high-level expression, which remains stable for up to a year (Bish et al., 2008).

The cardiac troponin T (cTnT) promoter sequence has been identified to cause cardiomyocyte-specific expression of exogenous proteins, post transfection or viral infection (Ma et al., 2004). The insertion of the cTnT promoter sequence in the rAAV9 transgene, upstream of the gene of interest, further enhances cardiac specificity (Prasad et al., 2011; Jiang et al., 2013). Gene expression is under the control of the cTnT promoter sequence; which interacts with troponin T specific transcription factors in cardiomyocytes. The use of the promoter sequence has been 148 shown to achieve overexpression or knockdown of the protein of interest, by up to 85% (Prasad et al., 2011; Jiang et al., 2013).

The presence of the enhanced green fluorescence protein (EGFP) reporter sequence in AAV vectors, can be used to detect virus transduction in different cell types (de Fougerolles et al., 2007). Recently produced rAAV9 constructs by Jian et al., in 2013, using AAV-9-cTnT-EGFP-RNAi plasmids (Figure 4.5.3.1), allows the identification of infected cells in the heart. Injection of these rAAV9 constructs, into the thoracic cavity at a concentration of 5×1013vg/kg, induces EGFP expression and RNAi mediated protein KD for up to 12months, exclusively in the heart (Jiang et al., 2013; Jiang et al., 2015). The level of cardiac protein KD was normalised with protein expression in mice injected rAAV9 Lac Z RNAi controls (Jiang et al., 2013; Jiang et al., 2015).

Figure 4.5.3.1 Cardiac-specific rAAV9 vector Schematic representation of AAV-9-cTnT-EGFP-RNAi from Jian et al., 2013 containing a collinear cTnT promoter and EGFP reporter. The presence of the cTnT sequence allows the expression of the vector sequence specifically in cardiomyocytes, as its expression is under the control of the cTnT promoter.

4.6 Materials and Methods

This work was carried out in collaboration with Prof Roger Foo’s research group, from The Genome Institute of Singapore and Assistant Prof Jianming Jiang, from the Cardiovascular Research Institute, National University of Singapore. The identification of the best RNAi oligonucleotide sequences and the production of plasmids were carried out by Dr Kelvin See, and Miss Edita Aliwarga, from The Genome Institute of Singapore.

149

4.6.1 MiD49 and MiD51 RNAi Knockdown

Invitrogen BLOCK-iT™ RNAi Designer web tool (Thermo Fisher), was used to identify RNAi sequences capable of optimal MiD KD, using the mouse exon sequence of MiD49 and MiD51 (targeting exon 2,3 or 4, for MiD49, and exon 3,4,5 or 6, for MiD51). Basic Local Alignment Search Tool (BLAST) was used for the top 10 suggested MiD49 and MiD51 RNAi sequences (Table 4.6.1.1). This was carried out to ensure that the oligonucleotide, sequences would bind correctly to target sequences, with minimal off-target binding. The optimum five specific oligonucleotide sequences were selected for MiD49 and MiD51. These oligonucleotides were subcloned via restriction digest and ligation using BsmBI cut sites into pCAG-mir-RNAi-mCherry vector (Figure 4.6.1.1). The mCherry reporter allowed the identification of transfected cells.

The RNAi-mCherry plasmids, capable of the knockdown of the MiD proteins, were used to transfect MEF cells. Knockdown efficiency of the plasmids was assessed using immunofluorescence staining, to identify RFP expressing cells, with minimal MiD49 and MiID51 expression.

150

Figure 4.6.1.1 MiD pCAG-mir-RNAi-mCherry vector production The ten oligonucleotide sequences were subcloned into pCAG-mir-RNAi-mCherry vectors, via restriction digest and ligation using BsmBI cut sites. Plasmid upscaling was carried out using competent E.coli cells. The purified plasmids were then used for transfection of MEF cells (section 4.6.2.1).

151

Oligonucleotide name Sequence (5' - 3') TGCTGTGTCGTGGGAGAATGAATGGGGTTTTGGCCACTG MiD49 miR 1 top ACTGACCCCATTCACTCCCACGACA CCTGTGTCGTGGGAGTGAATGGGGTCAGTCAGTGGCCAA MiD49 miR 1 bottom AACCCCATTCATTCTCCCACGACAC TGCTGTAGTGCTGAACTTAGGTGTCGGTTTTGGCCACTGA MiD49 miR 2 top CTGACCGACACCTGTTCAGCACTA CCTGTAGTGCTGAACAGGTGTCGGTCAGTCAGTGGCCAA MiD49 miR 2 bottom AACCGACACCTAAGTTCAGCACTAC TGCTGTATAGTGCTGAACTTAGGTGTGTTTTGGCCACTGA MiD49 miR 3 top CTGACACACCTAATCAGCACTATA CCTGTATAGTGCTGATTAGGTGTGTCAGTCAGTGGCCAAA MiD49 miR 3 bottom ACACACCTAAGTTCAGCACTATAC TGCTGATATAGATCCTGCAGCCAGAGGTTTTGGCCACTGA MiD49 miR 4 top CTGACCTCTGGCTAGGATCTATAT CCTGATATAGATCCTAGCCAGAGGTCAGTCAGTGGCCAAA MiD49 miR 4 bottom ACCTCTGGCTGCAGGATCTATATC TGCTGAGTACCAGCATCCTGGTCATCGTTTTGGCCACTGA MiD49 miR 5 top CTGACGATGACCAATGCTGGTACT CCTGAGTACCAGCATTGGTCATCGTCAGTCAGTGGCCAAA MiD49 miR 5 bottom ACGATGACCAGGATGCTGGTACTC TGCTGTTAACTGCCAGTGTGGCAATGGTTTTGGCCACTGA MiD51 miR 1 top CTGACCATTGCCACTGGCAGTTAA CCTGTTAACTGCCAGTGGCAATGGTCAGTCAGTGGCCAA MiD51 miR 1 bottom AACCATTGCCACACTGGCAGTTAAC TGCTGAAACCAGGGACATTCATGATGGTTTTGGCCACTGA MiD51 miR 2 top CTGACCATCATGAGTCCCTGGTTT CCTGAAACCAGGGACTCATGATGGTCAGTCAGTGGCCAA MiD51 miR 2 bottom AACCATCATGAATGTCCCTGGTTTC TGCTGTGAACTTCTAGTGTCAGGGCCGTTTTGGCCACTGA MiD51 miR 3 top CTGACGGCCCTGACTAGAAGTTCA CCTGTGAACTTCTAGTCAGGGCCGTCAGTCAGTGGCCAA MiD51 miR 3 bottom AACGGCCCTGACACTAGAAGTTCAC TGCTGAAGACAGTGTCACCAAGGGTTGTTTTGGCCACTGA MiD51 miR 4 top CTGACAACCCTTGGACACTGTCTT CCTGAAGACAGTGTCCAAGGGTTGTCAGTCAGTGGCCAA MiD51 miR 4 bottom AACAACCCTTGGTGACACTGTCTTC TGCTGAATAGGTTCACCTTAGGGTTCGTTTTGGCCACTGA MiD51 miR 5 top CTGACGAACCCTAGTGAACCTATT CCTGAATAGGTTCACTAGGGTTCGTCAGTCAGTGGCCAAA MiD51 miR 5 bottom ACGAACCCTAAGGTGAACCTATTC

Table 4.6.1.1 MiD49 and MiD51 targetting oligonucleotide sequences

152

4.6.2 MiD49 and MiD51 knockdown in mouse embryonic fibroblasts using RNAi vectors

MEF cells were transfected with MiD49 RNAi or MiD51 RNAi pCAG-mir-RNAi- mCherry plasmids, containing five different oligonucleotide sequences for each protein (Table 4.6.1.1). Ten different plasmids, containing the oligonucleotide sequences, were created and used to transfect MEF cells (Figure 4.6.1.1).

4.6.2.1 Transfection MEF cells were cultured in high glucose DMEM (SH30243.01, HyColne), supplemented with 10% FBS (Sv30160.03, Hyclone™), and 100U/ml of Penicillin/Streptomycin (15140122, Gibco). Cells were seeded onto a 12 well plate at a density of 150,000 24h prior to transfection. One oligonucleotide plasmid was used per well (10 plasmids tested in total), and 2 plasmid null wells. Cell medium (500µl of supplemented DMEM), was removed from wells and replaced with 500 µl of Opti-MEM™ reduced serum medium (31985-062, Gibco), before transfection. The transfection mixture, for each well, consisted of 1.25µg of the pCAG-mir-RNAi- mCherry vector, 2.5µl of Lipofectamine 3000, and 62.5µl of Opti-MEM™ reduced serum medium. The transfection mixture was left to incubate at RT for 15 minutes before it was added to each well. Opti-MEM was replaced with 500µl of supplemented DMEM 24h after transfection.

4.6.2.2 Immunocytochemistry of transfected cells The transfected MEFs were fixed 72 hours after transfection, using 4% paraformaldehyde at RT. Post fixation, the cells were washed twice with PBS and permeabilised with 0.2% Triton X-100 in PBS (30 minutes incubations at RT, T8787, Sigma-Aldrich). Following permeabilisation, the cells were washed twice with PBS. Blocking was carried out at RT using 1ml of 5% BSA in PBS, for 30 minutes. The cells were incubated with the primary antibodies (1:100 MiD49, 16413-1-AP and 1:50 MiD51, 20164-1-AP, Proteintech), 5% BSA in PBS, overnight at 4°C. Subsequently, the cells were washed three times with PBS, for 15 minutes. Cells were incubated with the secondary antibody, 1:200 Alexa 488 (A11034, Novex™ Goat anti-Rabbit IgG) for 45 minutes. Cells were washed three times for 5 minutes. Prior to imaging counterstaining was carried out using DAPI (4',6-Diamidino-2- Phenylindole, Dihydrochloride), for 10 minutes at RT(1µg/ml). The cells were washed twice with PBS prior to imaging.

153

4.6.2.3 Fluorescence imaging Immunofluorescence staining was carried out to identify oligonucleotides capable of inducing the highest level of MiD49 or MiD51 KD. Blinded qualitative analysis was carried out to identify MiD49 and MiD51 KD levels, indicated by Alexa 488 Immunocytochemistry staining, in miR RNAi oligonucleotide transfected cells expressing mCherry. Five regions within each well were randomly selected and imaged at a 40x magnification using a Microscope Zeiss Axio Observer Z1 inverted fluorescent microscope. The level of MiD KD was compared to Lipofectamine only negative controls, expressing higher levels of the MiD49 / MiD51 (green) which and lack mCherry expression. BFP, GFP and RFP channels were used to image DAPI, Alexa 488 and mCherry expression.

4.6.3 Development of MiD49 and MiD51 cardiac-specific AAV9 constructs

Following analysis of protein KD in MEF cells, the two oligonucleotide sequences capable of the greatest degree of MiD49 or MiD51 KD were subcloned, via regular restriction digest and ligation using BsmBI cut sites, into the pENN.AAV.cTnT.PI.EGFP.RBG(p13720)Q vector, for rAAV9 production (Figure 4.6.3.1). These plasmids, along with the helper and capsid plasmids were co- transfected into HEK293 cells to produce cardiac-specific, MiD49 and MiD51 targeting, AAV9 constructs for in vivo infection (Figure 4.6.3.1 and Figure 4.6.3.1).

4.6.3.1 Specific plasmids used for rAAV9 production Three plasmids are required for the creation of the AAV9 constructs, using a helper- free system of AAV production, by large-scale cell culture transfection. The best two for MiD49 and MiD51 KD, or Lac Z oligonucleotide sequences were subcloned into the pENN.AAV.cTnT.PI.EGFP.RBG(p13720)Q vector (4952bp), to create the recombinant transgene plasmids. The pENN.AAV.cTnT.PI.EGFP.RBG(p13720)Q vector contains a cTnT promoter and EGFP reporter sequence (Figure 4.6.3.1 A). The pAAV2-9swSEED (7390bp) and the pAdDeltaF6 (15424bp) plasmids were required to provide the AAV Rep/Cap and helper virus genes, respectively (Figure 4.6.3.1 B and C). Plasmids were sequenced to ensure they had undergone successful recombination and contained the oligonucleotide sequences, before high scale cloning.

154

Figure 4.6.3.1 rAAV9 plasmids used in the helper-free AAV production system The best MiD RNAi oligonucleotide sequences were inserted into the cardiac-specific AAV9 plasmid, containing the EGFP reporter (A). The helper-free system was used for MiD- specific rAAV9 constructs production (B). HEK293 cells were co-transfected using the recombinant AAV9 plasmids, along with the Rep/Cap and helper plasmids.

155

156

Figure 4.6.3.2 Plasmid maps of the vectors required for MiD rAAV9 production Three plasmids were required for large-scale cell culture transfection of HEK293 cells, to produce cardiac-specific MiD49 and MiD51 targeting rAAV9 constructs. Oligonucleotide sequences which caused the highest level of MiD49 and MiD51 KD, or a Lac Z gene sequence (control), were subcloned into the pENN.AAV.cTnT.PI.EGFP.RBG(p13720)Q vector(A). The pAAV2-9swSEED (B) and the pAdDeltaF6 (C) plasmids provide the AAV Rep,Cap and helper virus genes, respectively. Plasmids maps were kindly provided by Assistant Prof Jianming Jiang, from the Cardiovascular Research Institute, National University of Singapore.

157

4.6.3.2 Plasmid upscaling Plasmid upscaling was carried out using competent E.coli cells, suitable for viral DNA cloning, which reduces the frequency of homologous recombination of DNA sequences with long terminal repeats. One Shot™ Stbl3™ chemically competent E. coli cells were used for cloning the transgene and pAAV2-9swSEED plasmids. These competent cells were not suitable for pAdDeltaF6 plasmid cloning (C737303, ThermoFisher); and so, MAX efficiency™ Stbl2™ competent cells were used to produce a higher yield of the plasmids (10268019, ThermoFisher).

Competent E.coli stocks were thawed on ice. The cells were gently mixed with plasmid DNA (10-20ng) and incubated on ice for 30 minutes. Heat shock, at 42°C, was carried out for 45 seconds with One Shot™ E.coli cells and 25 seconds for the MAX efficiency™ E.coli cells. The cells were then immediately placed back on ice for 2 minutes, before the addition of pre-warmed S.O.C. medium (500μL, 15544- 034, ThermoFisher), to each vial. Vials were shaken horizontally at 225 RPM, for 90 minutes in a shaking incubator (incubation temperatures: One Shot™ cells at 37°C MAX efficiency™ cells at 30°C). 100μL of E.coli cells in S.O.C. medium was spread over ampicillin LB agar plates, using an inoculation loop. The plates were inverted and incubated overnight at 37°C. The following day, one bacterial colony was selected to be cultured in Lysogeny broth (LB, L3022, Sigma-Aldrich), containing 135µM ampicillin sodium salt (A9518, SIGMA-ALDRICH). The bacterial colony was initially grown in 10ml of LB with ampicillin for 6-8 hours, in a shaking incubator (200 RPM, at 30°C or 37°C, depending on the competent bacteria used). The bacterial culture was then added to conical flasks, containing 400ml of LB with ampicillin, sealed with aluminium foil and left to continue growing for 24h in a shaking incubator (200 RPM, at 30°C or 37°C, depending on the competent bacteria used).

The following day, frozen stocks of transformed E.coli, 20% Glycerol in LB, were created and stored at -80°C. E.coli cells were harvested, after centrifugation of the broth at 4000 RPM for 5 minutes. If pelleted cells were not immediately processed for plasmid isolation, they were frozen and stored at -80°C. Plasmids were isolated from E.coli cells using the E.Z.N.A. FastFilter Plasmid Maxi Kit (D6924-04, Omega Bio-Tek). Plasmid concentration, in TAE buffer, was measured using a NanoDrop 1000 (Thermo Scientific).

158

Figure 4.6.3.3 E.coli transduction and plasmid production protocol Plasmid amplification requires the presence of the E.coli ORI sequence, within recombinant plasmids, which is recognised by bacterial proteins involved in DNA replication. The presence of ampicillin resistance gene sequence, within recombinant plasmids, provides ampicillin resistance for cells which have been successfully transformed. Plasmid upscaling occurs as the plasmids are replicated during E.coli cells division. Heat shock was carried out to cause the intake of recombinant plasmids, into competent E.coli cells. The transformed cells were left to grow overnight on LB agar plates, containing ampicillin. One E.coli colony is selected from the agar dish, and its growth is upscaled using nutrient-rich LB broth containing ampicillin.

159

4.6.3.3 HEK293 cell culture HEK293 cells were used as the packaging cell lines for rAAV9 production. The cells were received as a frozen stock, at passage 3, stored in DMEM medium (SH30243.01, HyColne), with 10% DMSO. The cryovial, containing HEK293 cells, was placed in a 37°C water bath, and gently shaken until ice crystals were no longer visible. 5ml of pre-warmed supplemented DMEM (10%FBS, Hyclone™ Sv30160.030, and 100U/ml of Penicillin/Streptomycin) was gently added to the cell suspension before they underwent centrifugation to remove DMSO (2000 RPM for 2 minutes). Subsequently, the cells were resuspended in 10ml of supplemented DMEM and cultured in a T-75 culture flask.

Culture medium was changed every 48hours until the flask was confluent. To passage the cells, culture medium was removed from the flask and replaced with 3ml of pre-warmed 0.05% trypsin (0.5% Trypsin, Gibco® 15400054, in DPBS). The flask was incubated for 3 minutes at 37°C, followed by applying mild taps to the side of the flask, to aid cell detachment. Trypsin was deactivated by the addition of DMEM (3ml). A fifth of the cells, in 10ml of fresh medium, were added to a new T-75 flask. Cells reached a constant growth rate after 3 passages and were used for transfection. Cell stocks were created, by freezing the cells in culture medium with 10% DMSO.

For virus production, HEK293 cells were grown in 15cm culture plates (Thermo, 168381), with 20ml supplemented DMEM (kept in a 37 °C humidified incubator, supplied with 5% CO2). At the same time, a separate set of cells were grown in two T-75 flasks, as a maintenance stock, in case of contamination of plates. As virus production was done in bulk, cell counting was not carried out prior to cell seeding.

4.6.3.4 AAV9 production in HEK293 cells AAV9 constructs were created for all 4 MiD virus oligonucleotides, as well as the Lac Z control. The transfection of ten to twelve plates was on average required to produce a sufficient virus titre to inject a 3-week old mouse (≈15g). To ensure that a high virus titre was produced to inject three mice/ per construct, 40 plates were transfected, for each specific AAV9 construct. Cells from two confluent T-75 flasks were pooled and seeded onto eight 15cm plates. Plates were confluent after 60 hours. Cells from the eight plates were pooled and seeded onto forty new 15cm plates. After 24h, the cells were usually at an 85-90% confluence, and ready for

160 transfection. A high cell density was required due to the high rate of cell death post transfection. Viral production was carried out over a period of 4 days:

Day 1 – Transfection for each plate was carried out using the transfection mix of: 20µg of pADδF6, 7µg of pAAV9 plasmid, 7µg of the transgene plasmid (MiD49 miR 1, MiD49 miR 4, MiD51 miR 4, MiD51 miR 5 or Lac Z), 136µl of 1µg/µl Polyethylenimine HCl MAX (PEI) in DMEM (medium added to give a total volume of 800µl/plate). This transfection mixture was incubated for 20 minutes at RT before it was added to each plate (Table 4.6.3.1).

Material Amount required per plate pAAV2-9swSEED (capsid) 7µg pAdΔF6 (helper plasmid) 20µg AAV-9-cTnT-EGFP-RNAi (target plasmid) 7µg PEI (1µg/µl) 136µl HEK293 cell culture medium Added to give a total volume of 800µl

Table 4.6.3.1 Helper-free transfection mixture for rAAV9 production in HEK293 cells

Day 2 – The plates’ medium was changed, 24h after transfection, and returned to the 37°C humidified incubator, until 72h post transfection (day 4).

Day 4 – The transfected cells, containing viral constructs were harvested. Due to the high cell density by day 4, cells could be readily detached, without the use of trypsin. Cell medium, collected from the plate, was gently expelled from a 10ml glass strippette to free the cells. The detached cells were transferred into 50ml falcons. Each plate was washed with 3ml of DPBS, to maximise the collection of cells containing rAAV9. Cells were pelleted down (3 minutes of centrifugation at 7800 RPM), resuspended in DPBS to remove residual cell medium, and underwent a final centrifugation step at 7800 RPM for 3 minutes, before being frozen (-80°C). The pellets from the pooled cells were equivalent to cells from 10 plates/50ml falcon tube.

161

4.6.3.5 Transfection efficiency of different MiD rAAV9 oligonucleotide plasmids in HEK293 cells Plates were imaged 2h, 24h, 48, and 72h after transfection to monitor the transfection efficiency of the different AAV9 transgene plasmids, (6 plates per AAV construct). Four regions of a 15cm plate were randomly selected and imaged, in BF and using the GFP filter of a Leica DMi8 fluorescent microscope (viewed on Leica application suite X 3.0.1.15878). In addition to this, a small aliquot of the harvested cells was used to measure transfection efficiency with an Automated Cell Counter (Invitrogen™ Countess™ II FL).

4.6.3.6 Virus purification The transfected HEK293 cell pellets, containing the rAAV9 constructs, were thawed in a 37°C water bath and resuspended in lysis buffer (150mM NaCl, 20mM Tris- HCl). Cells underwent three freeze-thaw cycles, as the falcon tubes were placed in liquid nitrogen for 5 minutes, followed by 15 minutes at 37°C. 10µl of Mgcl2 (1M) and Benzonase (E1014-25KU, Sigma) was added to each falcon tube and incubated at 37°C for an additional 10 minutes. Lysed cells containing the same viral constructs were pooled and homogenised (Wheaton 40ml dounce homogeniser-tight), followed by centrifugation at 6900 RPM for 20 minutes (fixed angle rotor centrifuge), to pellet down the cell debris. AAV9 constructs were isolated by the addition of the supernatant to an Optiprep iodixanol density gradient column (Figure 4.6.3.4). The columns underwent ultracentrifugation at 45000 RPM, at 16°C, for 90 minutes (OptimaTML-100 XP Ultracentrifuge, Beckman Coulter). Subsequently, the 40% iodixanol layer, containing AAV9, was extracted using a 21G needle. The extracted 40% iodixanol layers were added to 100kDa centrifugal filter columns (UFC 910024, Amicon Ultracel, Millipore). Iodixanol was removed by several cycles of dilution in DPBS, followed by centrifugation (4000 RPM for 30 minutes). The concentrated AAV9 samples were collected and stored at -80°C.

162

Figure 4.6.3.4 Optiprep iodixanol density gradient column

4.6.3.7 Viral titreing The viral titre was measured using quantitative polymerase chain reaction (qPCR). The isolated AAV9 constructs were thawed on ice, and 1µl of each sample was sufficient to measure the viral titre. The standard curve was created using the EGFP PCR product (248bp), which was isolated and purified by gel extraction. Serial 10x dilution with autoclaved H2O was carried out using the PCR product, to create the standards (2gn/µl to 0.00000022gn/µl). The Ct (cycle threshold) values of the standard curve were used to determine the viral titre.

The samples underwent a 5x dilution, by the addition of 4µl of autoclaved ddH2O. Further dilution was carried out by the addition of 10x DNase reaction buffer (5µl, B43, ThermoFisher), DNase I (10µl, EN0521 from Thermo Scientific), and 30µl of autoclaved ddH2O. Sample tubes were incubated at 37°C, followed by 10 minutes incubation at 95°C, to denature DNase. This step is essential for removing residual plasmids and DNA fragments that may still be present after viral extraction. Subsequently, 50µl of Proteinase K mastermix was added to each sample (5µl of 10X DNase buffer, 44µl of sterile water, 1µl of Proteinase K (EO0491, Thermo Scientific)). The samples were incubated at 37°C, to break down residual protein and the viral capsid, followed by 10 minutes incubation at 95°C, to denature Proteinase K.

163

qPCR mixture Volume required per sample (µl) Mastermix 1x 10 Forward Primer (10µM) 1 Reverse Primer (10µM) 1 Template DNA 1

Autoclaved ddH2O 7

Table 4.6.3.2 AAV9 virus titre qPCR reaction mix

Subsequently, 1µl of the resultant product was added to 19µl of the qPCR master mix (SensiFAST SYBR no ROX kit BIO-98020, Bioline) with primers (Table 4.6.3.2). Primers were designed based on the EGFP sequence of the rAAV9 transgene, fwd:5'-CGAAGGCTACGTCCAGGAGC-3' and rev:5'-CGATGTTGTGGCGGATCT TG-3' (Axil Scientific, Singapore). Each reaction condition was sampled in triplicates, including the cDNA free ddH2O control. Quantitative PCR conditions are as follows: initial polymerase activation at 95°C for 10 minutes; DNA denaturation at 95°C for 15 seconds; primer annealing at 60°C for 15 seconds; extension at 72°C for 20 seconds (40 cycles). The Ct value of each qPCR reaction was used to determine the number of viral DNA strands, using the formula in Figure 4.6.3.5.

DNA concentration ( ) × (6.022 × 1023) × (Dilution Factor of 100) 1.5376 × 1011

Figure 4.6.3.5 Viral titre calculation formula The molecular weight (Mw) of the 248bp PCR product is calculated by the multiplication of 248 by 2 (due to the 2 DNA strands of the product), multiplied by 310 (average Mw of A,T,C,G nucleotides, in Da). This gives the value of 156,760Da. The DNA concentration, obtained from the values of the standard curve, divided by 1.5376×1011. This value is then multiplied by Avogadro’s constant (6.022x1023) and the dilution factor (100), to give the viral genome concentration/ml (Vg/ml).

4.6.4 Knockdown of MiD49 and MiD51 in C57BL/6 mice

Purified rAAVs were injected into the thoracic cavity, at three weeks of age, to ensure that most of the viral constructs infected the heart.

4.6.4.1 Injection technique Prior to injection, mice were anesthetised with 5% vaporised isoflurane in 1l/min of oxygen, using an induction chamber. Whilst unconscious the mice were weighed, and body hair covering the site of injection was removed using hair removal cream. The injection was carried out while the mice were still under the influence of isoflurane. The needle was inserted into the thoracic cavity from below the right

164 armpit, at a 45° angle, to minimise the risk of puncturing the heart (0.5ml insulin syringe with a 6mm needle, 31 gauge, BD Vevo). The volume injected into the heart was within 100-150µl, as space within the chest cavity is limited. Injections at a higher volume place a high degree of pressure on the heart and lungs, which can be fatal. The rAAV samples were concentrated (section 4.6.3.6), to ensure that of 5x1013 viral genomes/kg could be injected, without exceeding a volume of 150µl. The mice were kept under observation for 20 minutes post-procedure, to ensure there were no signs of injury or distress from the injection.

4.6.4.2 Identifying Protein Knockdown The hearts of mice, injected with the different rAAV9 constructs, were excised two weeks after rAAV9 injection, and snap frozen using liquid nitrogen, before being stored at -80°C. The frozen hearts will be processed for SDS-PAGE, followed by Western Blotting, to identify the MiD49 and MiD51 RNAi oligonucleotide sequences capable of achieving the highest level of MiD KD, compared to the Lac Z controls.

165

4.7 Results

4.7.1 Identifying the best RNAi oligonucleotide to knockdown MiD49 and MiD51 in mouse embryonic fibroblasts

The identification of the best RNAi oligonucleotide sequences and the production of plasmids were carried out by Dr Kelvin See, and Miss Edita Aliwarga. This work was carried out in MEFs, as it was essential to identify RNAi sequences which are capable of knocking down mouse MiD49 and MiD51 proteins. The five RNAi oligonucleotide sequences targeting MiD49 and MiD51 were subcloned into pCAG- mir-RNAi-mCherry vectors. Qualitative analysis of immunostained MEFS, transfected with these plasmids identified that the oligonucleotide MiD49 miR 1, MiD miR 4, MiD51 miR 4 and MiD51 miR 5 were capable of causing the highest degree of protein KD in transfected MEFs. These sequences were subcloned into pENN.AAV.cTnT.PI.EGFP.RBG(p13720)Q vectors and used for viral production. As expected the negative control cells did not express RFP, as they were only treated with lipofectamine 3000.

4.7.2 Transfection efficiency of rAAV9 vectors

Following MiD KD in MEFs, the two oligonucleotides capable of inducing the highest level of MiD49 or MiD51 protein KD were subcloned into pENN.AAV.cTnT.PI.EGFP.RBG(p13720)Q vectors. These recombinant rAAV9 plasmids and the Lac Z control rAAV9 plasmid was used for rAAV9 production in HEK293 cells. The successful transfection of these plasmids causes the expression of the EGFP reporter. The transfection efficiency was used as an indication of the level of rAAV produced. High transfection efficiency will result in a higher virus titre, as more HEK293 cells are involved in the production process.

166

Figure 4.7.2.1 Transfection efficiency of MiD HEK293 cells co-transfected with pAAV2-9swSEED, and pAdDeltaF6 plasmids and AAV-9- cTnT-EGFP-RNAi plasmids, containing the MiD miR sequences, as part of the helper-free AAV production system. Transfection efficiency was measured based on EGFP expression, 72h after transfection, using an automated Cell Counter. There was no significant difference in transfection efficiency of Lac Z (27.67±4.3%), compared to MiD49 miR 1 (20.33±3.5%), MiD49 miR 4 (20.67±1.0%) and MiD51 miR 5 20.33±1.4%. MiD51 miR 4 achieved a significantly lower transfection efficiency, compared to the Lac Z control vector (13.50±1.7% P<0.005). Statistical significance was assessed by one-way ANOVA followed by Bonferroni comparison test between the Lac Z control and other groups, P=0.0243, N=1 (SEM± of six transfected 15cm plates, from the same transfection).

167

Figure 4.7.2.2 Representative image of transfection efficiency of HEK293 cells during rAAV9 production For the production of rAAV9, HEK cells were co-transfected with pAAV2-9swSEED, and pAdDeltaF6 the AAV-9-cTnT-EGFP-RNAi plasmids, contain the different oligonucleotide sequences A) MiD49 miR1, B) MiD49 miR4, C) MiD51 miR4, D) MiD51 miR5, E) Lac Z control. Scale size indicated 0-100µM. 168

4.7.3 rAAV9 purification and injection

Following viral production and purification (methods described in section 4.6.3.4 and 4.6.3.6), qPCR was used measure the virus titre of each sample. The viral genome concentration/ml (Vg/ml), was then used to calculate the volume of rAAV9 required, to inject 5x1010 of viral genomes/g (Table 4.7.3.1).

MiD49 MiD49 MiD51 MiD51 Lac Z miR 1 miR 4 miR 4 miR 5 Virus titre (vg/µl) 7.760x109 1.079x1010 1.007x1010 6.107x109 1.155x1010 vg to inject/g 5x1010 5x1010 5x1010 5x1010 5x1010 volume to inject 6.443 4.634 4.967 8.187 4.331 (ul/g)

Table 4.7.3.1 volume of rAAV9 constructs used for in vivo injection into the thoracic cavity

Viral constructs were injected into the thoracic cavity of 3-week old C57BL/6 mice, under anaesthesia. Hearts were excised two weekss after injection and snap frozen using liquid nitrogen and stored at -80°C. The hearts will be processed for SDS- PAGE followed by Western Blotting, to identify the best MiD49 and MiD51 RNAi oligonucleotide sequences, capable of achieving the highest level of MiD KD, compared to the Lac Z controls.

169

4.8 Discussion

The primary objective of this chapter was to investigate the effect of global MiD49 genetic ablation, or MiD49 and MiD51 cardiac-specific KD on the adult heart and their potential role as a therapeutic target, against myocardial IRI. In this chapter, we identified that the KD of MiD49 had no significant effect on cardiac mitochondrial morphology, cardiac phenotype and did not cause cardioprotection following IRI. For this reason, rAAV9 constructs were created to allow us to investigate the outcome following the dual KD of both proteins, specifically in the heart.

The dual KO of MiD49 and MiD51 is embryonically lethal; therefore, a transgenic dual KO mouse colony cannot be established. Due to the high expression of MiD49 in the heart (Liu et al., 2013), compared to MiD51, we believed that the use of transgenic MiD49 KO colony was a more suitable animal model for the initial stage of our investigation, to provide a better insight into the potential role of the proteins as cardioprotective targets. The KD of MiD49 or both proteins were also required to cause a significant delay in MPTP opening, in HL-1 cells (Figure 3.5.3.1). Despite this, most of our findings from in vitro experiments indicated that both proteins are required to be targeted to cause a significant change in mitochondrial morphology and protect cells against IRI. As MiD49 and MiD51 whole body KO is embryonically lethal, we went on to developed RNAi rAAV9 cardiac-specific constructs to KD both proteins in the murine heart.

4.8.1 The role of MiD49 in determining the cardiac mitochondrial morphology

We know that mitochondrial dynamic proteins are highly expressed in cardiomyocytes and manipulation of their expression or activity cause significant changes in mitochondrial morphology, which is detrimental over long periods (Piquereau et al., 2013; Dorn, 2015). The tamoxifen-induced cardiac-specific KO of Drp1 heterozygous or Opa1+/− transgenic mice display the presence of elongated mitochondria; whereas, Mfn1, Mfn2 or Mfn1 and Mfn2 KO leads to the formation of fragmented interfibrillar mitochondria (Ong et al., 2010; Papanicolaou et al., 2011; Papanicolaou et al., 2012; Piquereau et al., 2012; Ikeda et al., 2015; Hall et al., 2016).

In chapter 3, we were able to demonstrate that the KD of MiD49 leads to mitochondrial fusion, in cardiac cell lines, similar to other cell lines in previously

170 published studies (Palmer et al., 2011; Loson et al., 2013). It is important to note that although an increase in elongated mitochondria was observed with MiD49 KD, there was only a significant increase in mitochondrial elongation following the KD of both MiD49 and MiD51 in cardiac cell lines (Figure 3.5.2.2 and Figure 3.5.2.3). Given that the knockdown of both proteins was required in vitro, the genetic ablation of only MiD49 was not expected to increase mitochondrial elongation in adult cardiomyocytes. We were able to show for the first time that the genetic ablation of MiD49 in mice, had no significant impact on the average length of interfibrillar mitochondrial length at basal conditions, compared to WT littermates (Figure 4.4.1.1 C). Using the established and validated model of characterising mitochondrial morphology in mouse cardiomyocytes (Ong et al., 2010), we identified that there was a significant increase in elongated mitochondria in MiD49 KO mice, compared to their WT littermates (Figure 4.4.1.1 D). Despite this finding, it is difficult to conclude that the KD of MiD49 alone can significantly increase the proportion of elongated mitochondria in adult cardiomyocytes, as there was no significant difference in the absolute length of interfibrillar mitochondria between the two groups. Although the MiD49 expression is higher in cardiomyocytes than MiD51 (Liu et al., 2013), from our in vitro studies and published data, we believe that the KO of both MiD49 and MiD51 is required to cause a significant change in mitochondrial length.

Measuring mitochondrial length using 2D EM images is commonly used to determine mitochondrial morphology (Ong et al., 2010; Papanicolaou et al., 2011; Papanicolaou et al., 2012; Dongworth et al., 2014; Hall et al., 2016), however, this method of analysis is limited as there is still no universal agreement within the scientific community over the most robust process of determining mitochondrial morphology in adult cardiomyocytes. There are publications which suggest that other parameters should also be measured when determining mitochondrial morphology as well as the advantage of using 3D imaging, which has recently revealed the presence of vast mitochondrial networks in cardiomyocytes and their interaction with other organelles (Hayashi et al., 2009; Beikoghli Kalkhoran et al., 2017; Kalkhoran et al., 2017).

4.8.2 Cardiac phenotype of MiD49 KO mice

The long-term disruptions of mitochondrial dynamics in the heart, which induce extreme changes in mitochondrial morphology, significantly affect cardiac phenotype

171 and function (Papanicolaou et al., 2011; Piquereau et al., 2012). Although EM analysis of MiD49 KO hearts revealed a significant increase in the proportion of fused mitochondria (mitochondria >2µM in size), there was no significant increase in the overall mitochondrial length when compared to WT cardiomyocytes (Figure 4.4.1.1 C) Given that MiD49 KO did not induce an extreme, hyper-fused, mitochondrial morphology, as observed following Drp1 KO (Ikeda et al., 2015), the level of fusion achieved was not expected to cause any overt changes in the cardiac phenotype or function.

Indeed, the genetic ablation of MiD49 caused no significant change in the overall cardiac function at baseline or during stress, of 8-13 week old mice, compared to WT littermates. The only significant difference observed between the WT and MiD49 KO cardiac phenotype was the increase in LV posterior wall thickness during end systole, in MiD49 KO mice (Figure 4.4.2.2 D). This is not likely to be an outcome of LV hypertrophy in MiD49 KO mice but instead due to the increased force of contraction; as there was no significant difference in LV anterior or posterior wall thickness during end diastole, the LV posterior wall thickness during end diastole and there was no significant increase in LV mass, when compared to WT hearts (Figure 4.4.2.2). It would be interesting to see if cardiac phenotype and function remains unaffected in aged MiD49 KO mice. Chronic genetic ablation of mitochondrial dynamic proteins has repeatedly been shown to be detrimental, (Papanicolaou et al., 2011; Papanicolaou et al., 2012; Piquereau et al., 2012; Ikeda et al., 2015; Hall et al., 2016), highlighting the importance of their activity in quality control and mitochondrial homeostasis. Disruption of the cardiac mitochondrial fission machinery by the genetic ablation of Drp1 inhibited mitochondrial mitophagy, leading to a significant loss of cardiomyocytes. Echocardiography imaging showed the development of cardiac hypertrophy within 4 weeks, and mice died by 13 weeks after KO, due to HF (Ikeda et al., 2015). MiD49 KO may also affect the mitophagy machinery of cardiomyocytes, due to a reduction in Drp1 recruitment, however as MiD51 and Mff are still expressed in these mice, development of hypertrophy may take a longer period of time, or not occur at all. Changes in cardiac phenotype are likely to be more extreme in the absence of both MiD proteins, as we have shown that the KD of both proteins achieved a higher level mitochondrial fusion in cardiac cell lines (Figure 3.5.2.3).

4ng/g of isoproterenol was used during echocardiography imaging to help identify irregular cardiac functions, which may not be evident during basal conditions 172

(Rottman et al., 2007; Dongworth et al., 2014; Hall et al., 2016). There was a significant difference between echocardiography measurements of the LV anterior and posterior wall thickness during systole, LVFS and LVEF at baseline and stress- induced isoproterenol treatment, in both WT and MiD49 KO hearts (Figure 4.4.2.2 and Figure 4.4.2.4). There was no significant difference in cardiac activity post isoproterenol treatment, between WT and MiD49 KO mice. Although the mice did show to have a significant response to 4ng/g isoproterenol treatment, the response was not as severe as previously observed in WT C57BL/6J and other transgenic mice (Dongworth et al., 2014; Hall et al., 2016). The reduction of cardiac isoproterenol sensitivity phenotype may be due to the mixed C57BL/6J and C57BL/6NTac background of the MiD49 KO colony. Several genetic variations have been identified between the two lines, which have resulted in phenotypical differences between C57BL/6J and C57BL/6NTac mice (Simon et al., 2013).

The chronic KO of MiD49 did not cause any evident changes in cardiac phenotype and function. This suggests that the acute inhibition of the protein upon reperfusion, as a target for cardioprotection, is not likely to have any detrimental effect, as the protein does not seem to mediate any critical cellular activities under physiological conditions. The absence of any overt myocardial pathologies as a result of MiD49 KO, allowed the investigation of cardiac ischaemia-reperfusion susceptibility by LAD occlusion, compared to WT littermates, without the influence of confounding factors.

4.8.3 Susceptibility of MiD49 KO mice to ischaemia-reperfusion injury

MiD49 ablation did not cause a significant change in the cardiac phenotype or function in mice, compared to their WT littermates (section 4.3.5). From our in vitro studies, we believed that the KO of MiD49, on its own may not provide significant protection against acute IRI in the murine heart. Indeed there was no significant difference if MI size between MiD49 KO hearts and WT hearts, following in vivo LAD occlusion for 45 minutes and 2 hours reperfusion.

One possible explanation for this outcome may be due to MARCH5 activity. MARCH5 has been identified to cause MiD49 lysis during cellular stress conditions; subsequently reducing mitochondrial fragmentation, by preventing MiD49-Drp1 interaction (Xu et al., 2016). This intrinsic protective pathway may also be activated during IRI and could explain why there is no significant difference in infarct size between WT and MiD49 KO hearts. This could be further investigated by comparing

173

MiD49 expression or MARCH5 activation during basal condition and following acute IRI surgery.

Another possible explanation for this outcome may be due to the adaption of the heart to the chronic KO of MiD49. Transgenic mice are very commonly used in research for various disease models, with an alteration to the expression of the protein of interest. These models do not necessarily represent the expected phenotype in the absence or overexpressed state of the protein of interest. The observed phenotype may be an adaptive effect to a chronic condition. The absence of MiD49 may have caused the overexpression of MiD51 or other mitochondria fission proteins as a compensatory effect. Unfortunately, we were not able to identify changes in the expression of the MiD proteins in the heart due to difficulties in establishing a robust SDS-PAGE and western blotting assay with the most commonly used and commercially available antibodies for MiD49 and MiD51 (Proteintech 16413-1-AP and 20164-1-AP). In our hands, we found that these polyclonal antibodies lead to the expression of many non-specific bands, which could not be used as a true representation of the proteins’ expression in the heart. To reduce these non-specific interactions, we are currently optimising this assay using isolated cardiac mitochondria samples.

Chronic changes in mitochondrial dynamic proteins are often found to be detrimental in the setting of IRI. The KO of Drp1 significantly increased mitochondrial elongation, however, this also increased IRI susceptibility in mice due to the severe defects in mitochondrial quality control which lead to the formation of cardiac hypertrophy. A similar outcome in our in vivo IRI studies would have highlighted the importance of MiD49 in the heart, as a Drp1 receptor, however, the KO of the protein had no significant effect on IRI susceptibility, and it also did not induce the cardiac dysfunction observed following Drp1 ablation (Ikeda et al., 2015). It is still not clear if there is no interplay between MiD49 and MiD51 proteins’ activity. Our finding from the KD of MiD49 and MiD51 in HL-1 and H9c2 cells suggest that both proteins should be targeted to significantly reduce Drp1-mediated mitochondrial fragmentation and reduce cell death following SIRI (chapter 3). MiD49 and MiD51 should only be acutely inhibited in the setting of IRI, to prevent cardiac dysfunction as a result of an imbalance in the cardiac fusion and fission machinery.

Interestingly the genetic ablation of Mfn1 and Mfn2 has been identified to be cardioprotective and reduces susceptibility to cell death following cardiac IRI,

174 despite the formation of fragmented mitochondria (Papanicolaou et al., 2011; Papanicolaou et al., 2012; Hall et al., 2016). This is believed to be due to the changes in the proteins pleiotropic properties, which reduced ER/SR-mitochondria tethering and therefore prevent mitochondrial calcium overload during IRI (Hall et al., 2016). MiD49 and MiD51 have both been shown to interact with the ER (Elgass et al., 2015). Acute inhibition of the proteins during IRI could also be cardioprotective as a result of reducing ER interaction.

4.8.4 Cardiac-specific knockdown of MiD49 and MiD51 in mice

The primary objective of section two of this chapter was to develop rAAV9 constructs capable of MiD49 and MiD51 knockdown, in the adult heart. The genetic ablation of MiD49 had no significant effect on the reduction of MI post myocardia IRI (section 4.8.3). It was not possible to investigate the KO of both proteins as the genetic ablation of MiD49 and MiD51 is embryonically lethal. The use of rAAV9 constructs would allow the simultaneous KD of both MiD49 and MiD51 in the adult heart. This is important as the use of these viral constructs would allow us to investigate the effect of inhibiting both proteins in the heart, as potential targets of cardioprotection, which would otherwise not be possible. The cardiac-specific targeting nature of the constructs also allows us to create a better animal model for our investigation, as the proteins’ expression is unaffected in other organs; therefore, reducing any confounding factors which may arise from pathologies developed from their global ablation (Jiang et al., 2013; Jiang et al., 2015). Another advantage of using rAAV9 constructs is that protein KD can be carried out after development, which would reduce the negative impact of long-term changes in mitochondrial morphology to cellular health and development (Papanicolaou et al., 2011; Chan, 2012; Piquereau et al., 2012; Ikeda et al., 2015).

Several MiD targeting oligonucleotide sequences were tested in MEF cells, to identify their ability to KD MiD49 and MiD51 in mouse. The two oligonucleotide sequences, capable of inducing the highest level of MiD49 or MiD51 KD were identified, following immunohistochemistry of transfected cells. These sequences were used to form recombinant rAAV9 vectors (pENN.AAV.cTnT.PI.EGFP.RBG). The rAAV9 plasmids, along with the helper and Rep/Cap plasmids were co- transfected into HEK293 cells, to produce cardiac specific MiD49 or MiD51 targeting rAAV9 constructs. rAAV9 vectors, containing the Lac Z sequence, were also used to create control rAAV9 constructs. These constructs were individually injected into the

175 thoracic cavity of three week old C57BL/6NTac mice, to assess their ability to knockdown MiD49 or MiD51 in the mouse heart, compared to mice injected with the Lac Z control constructs. Unfortunately due to ongoing work to establish a robust western blotting assay using the commercially available MiD antibodies this work is still ongoing. Once this assay has been established, the rAAV9 infected frozen hearts will be used to identify the best MiD49 and MiD51 constructs, capable of inducing the highest level of protein KD.

Once the best oligonucleotide sequences have been identified, both MiD49 and MiD51 rAAV9 constructs will be simultaneously injected into the thoracic cavity. The mice will undergo echocardiography imaging at 3 weeks of age, prior to injection. Subsequently, the same imaging protocol will be repeated at 5 and 7 weeks of age, followed with cardiac stress induced isoproterenol treatment (at 7 weeks), to identify any changes in cardiac phenotype or function. The same mice will then undergo in vivo MI surgery at 8 weeks, to assess their susceptibility to myocardial IRI, compared to the Lac Z control mice. These rAAV9 constructs are capable of achieving KD of the target proteins within 72h of injection (Jiang et al., 2013; Jiang et al., 2015). In vivo MI surgery carried out within a few days of injection could provide a better understanding of the expected outcomes from acute MiD inhibition.

4.8.5 Summary

The primary objective of this chapter was to investigate the effect of global MiD49 genetic ablation in the adult mouse heart, and if this could reduce susceptibility to myocardial IRI, followed by investigating the effects of dual MiD KD in the mouse heart.

The loss of MiD49 caused no significant effect on cardiac phenotype or function. Subsequently, there was no significant difference in MI size following in vivo IRI surgery, between MiD49 KO and WT hearts. These results confirm that the inhibition of MiD49 activity, by genetic ablation does not protect the heart against IRI. The global MiD49 KO model does have its limitations, as it is a de novo model of genetic ablation. As mitochondrial dynamics play an essential role in development and mitochondrial homeostasis, the absence of the protein from birth may compromise the health of the mitochondrial network, and therefore conditional KO of the proteins after development is required (Chan, 2012; Dorn, 2015). Another limitation is that that MiD51 is still expressed in these mice. Although MiD51 expression is lower in the adult heart, the loss of MiD49 may cause a compensatory effect on MiD51 176 expression and activity (Liu et al., 2013), to minimise the effect of MiD49 ablation. The MiD proteins are mitochondrial specific recruiters of Drp1, however, their distinct roles have not yet been determined. The knockdown or knockout of either protein has been shown to reduce Drp1 recruitment to the mitochondria; however, it is still not clear whether both proteins must be inhibited to induce a significant change in mitochondrial morphology, similar to Drp1 null cells (Palmer et al., 2011; Loson et al., 2013). From our in vitro and in vivo findings, we believe that both proteins must be acutely inhibited in the heart to protect against acute IRI.

177

CHAPTER 5: Identification of a chemical inhibitor of the Drp1-MiD49/MiD51 interaction

5.1 Introduction

MiD49 and MiD51 are mitochondrial specific Drp1 receptors essential for mitochondrial fission. These proteins can thereby act as precise therapeutic targets for peptides or small molecules, to inhibit Drp1-mediated mitochondria fragmentation (Simpson et al., 2000; Palmer et al., 2011). In this regard, the identification of their mouse crystal protein structures and ex vivo functional analysis of MiD49 and MiD51 have identified essential regions which could be targeted to achieve this goal (Loson et al., 2014; Richter et al., 2014; Loson et al., 2015).

Both MiD proteins have a nucleotidyltransferase fold, however, only MiD51 is capable of interacting with nucleotides (Loson et al., 2014; Richter et al., 2014; Loson et al., 2015). Fluorescence-based shift assay of MiD51 showed the highest level of stability with ADP binding, and it also weakly interacted with GDP at its nucleotidyltransferase fold (Loson et al., 2014; Richter et al., 2014). ADP binding is suggested to act as an essential cofactor, required for stabilising the Drp1 spiral formation around mitochondria; however, MiD51-Drp1 recruitment can still occur in the absence of ADP (Loson et al., 2014; Richter et al., 2014). It is still not clear if ADP binding to MiD51 is essential for mitochondrial fragmentation. Loson et al., 2014 found that MiD51-ADP binding promotes Drp1-Drp1 interaction, and its GTP hydrolysis activity at basal level (Loson et al., 2014). However, Richter et al., showed that Drp1 binding to MiD51 mutants that are unable to bind ADP or GDP, can still form rings around mitochondria and cause fission (Richter et al., 2014).

A single exposed loop on the surface of both proteins was identified to be essential for Drp1 binding (Loson et al., 2014; Richter et al., 2014; Loson et al., 2015). The residues within the Drp1-binding loop are well conserved between MiD51 and MiD49. Both research groups identified that the deletion of the highly conserved surface loops, or their destabilisation, inhibits Drp1 interaction (Loson et al., 2014; Richter et al., 2014; Loson et al., 2015). The mouse surface loop residues were identified as LEFHP (MiD49 aa 230-234) and PEYFP (MiD51 aa 238-242) (Loson et al., 2014; Richter et al., 2014; Loson et al., 2015).The MiD binding loop within the dynamin recruitment region interacts with four Drp1 molecules at four possible Drp1

178 surfaces, to form cofilaments. GTP hydrolysis by Drp1 results in MiD49 or MiD51 dissociation before constriction of the Drp1 rings (Kalia et al., 2018). MiD association was found to be structurally incompatible with the Drp1 rings during constriction, further indicating that receptor dissociation is required prior to constriction (Kalia et al., 2018).

Therapeutic inhibition of MiD49 and MiD51, by inhibiting their surface loop interaction with Drp1, upon reperfusion may provide a novel treatment strategy for reducing myocardial infarct size and preventing heart failure in patients presenting with acute myocardial infarction. There are currently no peptides or small molecules capable of specifically inhibiting the binding of Drp1 to MiD49 and MiD51 as a strategy for inhibiting mitochondrial fission. As such, this is a novel therapeutic strategy for inhibiting mitochondrial fission and has the potential to reduce myocardial infarct size and preserve left ventricular function following AMI.

5.1.1 Biophysical fragment-based approaches for drug discovery

5.1.1.1 Thermal shift assay Thermal shift assay (TSA), is an inexpensive technique often used as part of initial high-throughput drug discovery screens to identify protein-ligand interactions. Thermofluor assays use fluorescent dyes, such as SYPRO Orange to quantify the rate of protein thermal denaturation. These assays are efficient in identifying purified protein or recombinant protein stability, using real-time PCR instruments (Lo et al., 2004). The natural folding of proteins in an aqueous environment consists of hydrophobic residues being unexposed and encapsulated within exposed hydrophilic residues. SYPRO Orange has a high affinity for hydrophobic amino acids, which emits fluorescent signals upon binding to these residues. Thermal denaturation of proteins leads to the exposure of the hydrophobic core, allowing for the binding of the dye to the exposed hydrophobic residues. The fluorescent signal continues to increase until the protein is fully denatured (Lo et al., 2004; Niesen et al., 2007). The thermal shift melting curve is used to measure the Tm, which is the temperature where the concentrations of folded and unfolded proteins are equal (Figure 5.1.1.1). The successful binding of a ligand in most cases increases protein stability and therefore causes an increase in the Tm (Matulis et al., 2005; Senisterra et al., 2006).

179

Figure 5.1.1.1 Representative schematic of protein thermal shift assay Thermal shift melting curve recordings are used to show the change in fluorescence intensity due to thermal denaturation of proteins, in the presence of SYPRO Orange. Fluorescence is very low while the tertiary protein structure is conserved in aqueous solution, at lower temperatures (excitation wavelength 492nm). The increase in temperature initiates protein unfolding, which slowly exposes the hydrophobic residues. SYPRO Orange interaction with these residues results in fluorescent light emission which can be detected at 610nm. The melting temperature, Tm, is the temperature at which the concentration of folded and unfolded proteins is at equilibrium. After full denaturation, fluorescence signal slowly decrease due to protein precipitation or aggregation (Rodrigues et al., 2011)

5.1.1.2 Surface Plasmon Resonance Surface plasmon resonance (SPR) is a label-free biophysical assay which is commonly used in pharmaceutical research to identify binding interactions between proteins to peptides or small molecules (Otto, 1968). SPR involves the transfer of light waves into electron waves (plasmons) at a thin metal layer (Otto, 1968; Daghestani & Day, 2010). SPR sensors detect changes in plasmon waves, which is affected by changes in the refractive index of the medium near the surface through which they travel (Figure 5.1.1.2).

The sensor chip is composed of a glass slide, covered by a thin gold coating (Otto, 1968). The gold film is coated with a matrix most suitable to immobilise ligands, such as the protein of interest. The SPR detector can measure binding events, specificity, affinity and kinetics of analytes, such as peptides or small molecules, in real-time (Myszka DG and Rich RL, 2003). 180

Polarised light is directed through a prism, directly under the surface of the gold film (Muzyczka, 1992; Cooper, 2002). Light is reflected off the back of the gold film, from the glass side of the sensor chip. Binding events during the flow of analytes across the sensor chip cause a change in the refractive index, altering the angle of minimum reflective intensity (θ) (Daghestani & Day, 2010). Molecular interaction on a sensor chip is monitored by the SPR detector and recorded to produce a sensorgram (Daghestani & Day, 2010).

Figure 5.1.1.2 Representative schematic surface plasmon resonance Photons are directed onto the back of the gold-coated sensor chip, through a prism. The detector is able to detect changes in electron waves (plasmons) at a thin metal layer due to the binding of analytes to the immobilized ligands. This interaction causes a change in the refractive index of the medium near the surface through which the electrons travel. Image from Myszka DG and Rich RL, 2003.

181

5.2 Research objective and Aims

Specific hypothesis: Small peptides or small molecules, structurally similar to the MiD-Drp1 binding loop, will compete for Drp1 interaction, and will therefore significantly reduce MiD-Drp1 binding, and consequently reduce Drp1 mediated mitochondrial fission.

The primary objective of this chapter was to identify small peptides capable of preventing the binding of Drp1 to MiD49 and MiD51, as a novel therapeutic strategy for inhibiting mitochondrial fission and protecting the heart against acute IRI. For achieving this objective, the principle aims were:

1) To synthesise and purify Drp1 and MiD49 and MiD51 proteins.

2) To synthesise small peptides resembling the MiD49 and MiD51 loop regions involved in Drp1 interaction.

3) To develop biophysical assays to screen peptides’ ability to prevent Drp1 binding to MiD49 and MiD51 using thermal shift assay and surface plasmon resonance biophysical techniques.

5.3 Materials and Methods

5.3.1 Synthesis of pure Drp1 and MiD proteins

Recombinant plasmids for human Drp1, MiD49 and MiD51 genes were fused with tobacco etch virus protease cleavage site proceeded by a six Histidine (His) N- terminal tag, inserted into the bacterial expression vector pNIC28-Bsa4 (GenBank® accession number EF198106, ordered from Genscript Biotech). Drp1 proteins were produced with N-terminal or C-terminal His tags (Drp1-vc003 with His tag at N- terminus, Drp1-vc008 and Drp1-vc0013.with His tag or Avitag at C-terminus, respectively). The gene sequence encoding residues 1-710 of Drp1 was cloned from the full-length human DNML1 gene, residues 126-454 of MiD49 was cloned from the human SMCR7 gene, and residues 119-463 of MiD51 was cloned from the human SMCR7L gene (GenBank® accession number BC024590.1, BC014973.1 and BC008327.1, respectively). The plasmid was transformed into an E.coli BL21 (DE3) Rosetta strain (developed by the Protein Production Platform (PPP) of Nanyang

182

Technological University). Protein purification was carried out by PPP (Koh-Stenta et al., 2014).

Transformed E.coli cells were cultured overnight in Modified Terrific Broth (TB medium, pH7.2) with Kanamycin (50µg/ ml) and Chloramphenicol (34µg/ml), 37°C, 230 RPM. The E.coli cells were treated with 0.5mM of Isopropyl β-D-1- thiogalactopyranoside (IPTG) and left to grow overnight at 18°C, to trigger transcription and thereby induce protein expression,. The harvested cells were resuspended in 50ml of lysis buffer (20mM HEPES pH7.5, 0.5M NaCl, 10mM Imidazole, lysozyme (1mg/mL), 0.5mM TCEP and DNase cocktail inhibitor), followed by sonication (20Amp). The cell lysate was centrifuged at 18,000 RPM, at 4cC for 20 min. Protein purification was carried out by affinity chromatography, using Ni-NTA beads or Ni-NTA super flow Cartridges (Qiagen) on the AKTAxpress chromatography system (General Electric Healthcare), to harvest the His-tagged proteins. The columns were washed with 500ml of wash buffer 1 (20mM HEPES pH7.5, 50mM NaCl, 10mM Imidazole, 0.5mM TCEP) followed by 250ml of Wash buffer 2 (20mM HEPES pH7.5, 50mM NaCl, 25mM Imidazole, 0.5mM TCEP), to remove unwanted proteins weakly bound to the nickel beads. Elution buffer (20mM HEPES pH7.5, 50mM NaCl, 500mM Imidazole, 0.5mM TCEP) was used to detach proteins of interest from the nickel beads.

Immobilized-metal affinity chromatography (IMAC) fractions were analysed by SDS- PAGE. Fractions containing the proteins of interest were pooled and concentrated using Amicon® Ultra-15 centrifugal filter units at 3900 RPM, 4°C, followed by size exclusion chromatography using gel filtration (gel filtration buffer 20mM HEPES pH7.5, 300mM NaCl, 0.5mM TCEP). The gel filtration fractions were analysed by SDS-PAGE. Protein purity of the pooled products was examined by mass spectrometry. Purified samples were snap frozen using liquid nitrogen and stored at -80°C.

5.3.2 Computational modelling of MiD proteins’ Drp1 binding loops

Based on computational modelling of the human and mouse MiD proteins, we identified that the most exposed amino acids forming the loops structure are LEFHP on the surface of MiD49 and PEYFP on the surface of MiD51.The identification of these short sequences allowed us to design small peptides based on the MiD49 LEFHP loop and MiD51 PEYFP loop, to act as competitive inhibitors of the Drp1-

183

MiD interaction (Table 5.4.2.1). The synthesised peptides (Genscript) were tested for binding to Drp1 using several biophysical assays.

5.3.3 Thermal shift assay

TSA was used to assess the stability of the purified proteins. This was assessed by measuring the Tm, at a final concentration of 5 µM and 8 µM in buffer (20mM HEPES, 150mM NaCl at pH 7.5) containing 15x SYPRO Orange dye (S6650, Thermo Fisher). Subsequently, the stability of the MiD proteins was measured in the presence of ATP, ADP, and GDP (250µM to 1mM).

To assess if the integration of Drp1 (Drp1-vc003) with nucleotides or small peptides had a significant effect its Tm, Drp1 protein and SYPRO Orange dye were diluted to the working concentration of 12.5µM and 150x, respectively in 20mM HEPES, 150mM NaCl at pH 7.5 buffer. The test nucleotides or small peptide were added at a concentration of 50µM, 100µM, 500µM and 1mM to a 384 well-plate, sealed with a transparent film sticker and centrifuged at 2800 RMP. The final concentration of Drp1 and dye were 5µM and 15x respectively. Control samples did not contain nucleotides or small peptides. All conditions were tested in triplicates, and results were used to calculate the average Tm. Samples were analyzed using a Roche lightcycler 480 II. Tm calculation was carried out using LC 480 Gene Scanning software, and the temperature shifts were analyzed using GraphPad Prism Software.

5.3.4 Surface Plasmon Resonance

SPR studies were performed using a Biacore T200 biosensor (GE Healthcare). The resonance units (RU) is used to measure protein interaction, as one RU is the binding of approximately 1pg protein/mm2 (Jonsson et al., 1991). 3,500 RU of full- length human MiD51 was immobilised by amine coupling to the carboxymethylated dextran matrix of a Series S CM5 chip. 3,000 RU of BSA was immobilised to the reference cell by amine coupling, to block interactions with the reference cell. The immobilisation was carried out at a flow rate of 5 μl/min, using a 20mM HEPES running buffer (pH 7.5), 0.1% P20. Following immobilisation of MiD51, single-cycle kinetics analysis of Drp1 binding was performed at 25°C across five concentrations of Drp1 (0.2, 0.4, 0.8, 1.6 and 3.2μM). Association time and flow rate were set at 60 seconds and 30μl/min, respectively. For binding analysis, running buffer 1X PBS

184

(pH 7.2) (Gibco 20012-043) was used. KD determination was performed using the Biacore T200 Evaluation software (Version 2.0, GE Healthcare).

5.4 Results

5.4.1 Protein purification of human MiD49, MiD51 and Drp1

Protein purification was carried out by Protein Production Platform (PPP) of Nanyang Technological University, Singapore.

Protein production of truncated MiD49 and MiD51 (residues 126-454 and residues 119-463, respectively), lacking the N-terminal membrane-proximal region and Drp1 (residues 1-710) was carried out in competent E.coli cells. Proteins were purified from cell lysates by affinity chromatography and size exclusion chromatography by gel filtration (Figure 5.4.1.1). The isolated samples were pooled, and protein purity was assessed using SDS-PAGE (Figure 5.4.1.1). The MiD49 and MiD51 sample elution by gel filtration only produced one peak on the gel filtration chromatogram (Figure 5.4.1.1 A and B). Gel filtration chromatogram Drp1 indicated the presence as higher-ordered structures such as dimers and trimers, shown by the presence of more than one peak (Figure 5.4.1.1 C), as elution of larger molecules occurs at a faster rate due to their inability to travel through the gel filtration beads (Bai, 2015).

185

Figure 5.4.1.1 Purified MiD49, MiD51 and Drp1 proteins Gel filtration chromatogram of protein elutions (left) and SDS-PAGE of pooled purified protein products in kDa (right). Purified MiD49 (A), MiD51 (B) and Drp1 (C) proteins (theoretical molecular weights 37.148 kDa, 41.321 kDa and 81.994 kDa respectively). Gel filtration chromatogram Drp1 indicated the presence of Drp1 as higher-ordered structures as well as monomers.

186

5.4.2 Computational analysis of Drp1 and MiD proteins

Computational analysis was carried out by Dr Edith Chan (Wolfson Institute for Biomedical Research, University College London) and Dr Anna Jansson (A* Experimental Therapeutics Centre, Singapore).

BLAST was used to compare MiD49 and MiD51 protein sequences in Mice and Humans (Database Resources of the National Centre for Biotechnology Information). The Human MiD49 and MiD51 proteins have a 46% sequence identity (Figure 5.4.2.1 C). The Mouse and Human MiD49 are both composed of 454 amino acids, with an 80% sequence identity (Figure 5.4.2.1 A). Mouse and Human MiD51 are both composed of 463 amino acids with a 98% sequence identity (Figure 5.4.2.1B). Computational analysis of the Drp1-recruitment regions previously identified in mice (Loson et al., 2014; Richter et al., 2014; Loson et al., 2015) were compared to the human proteins and revealed that the most exposed amino acids of human MiD49 are LEFCP and the most exposed amino acids of human MiD51 are PEYFP (Figure 5.4.2.1 D). The human and mouse sequences of the MiD51 Drp1 binding loop are identical in sequence however the MiD49 Drp1 binding loop contains a cysteine residue, instead of histidine found in the mouse structure. Human MiD49 proteins were identified to be capable of forming dimers, but dimer formation has not been detected between mouse MiD49 proteins (Loson et al., 2015). Small peptide sequences to inhibit Drp1 interaction were based on MiD49 and MiD51 Drp1-binding loop amino acid sequences (Table 5.4.2.1 and representative image Figure 5.4.2.1).

Small peptide sequences Species LEFHP Mouse MiD49 LEFCP Human MiD49 PEYFP Human and Mouse MiD51 RTQLEFCPRGSSPDRFLVG Human MiD49 RENPEYFPRGSSYMDRCVVG Human MiD51 RENPEYFPRGSSYMD Human MiD51

Table 5.4.2.1 MiD49 and MiD51 small peptide sequences

187

Figure 5.4.2.1 MiD49 and MiD51 amino acid sequence Human and mouse MiD protein sequence comparison, with the loop sequences highlighted in red (NCBI BLAST). A) MiD49 human (query) and mouse (subject) sequence, B) MiD51 human (query) and mouse (subject) sequence. C) Sequence identity of human MiD49 (subject) and MiD51 (query) sequence. D) Representative image of computational analysis of MiD51 protein structure (by Dr Edith Chan).

188

5.4.3 Thermal Shift Assay of Drp1 and MiD proteins

TSA was carried out to assess the structural stability of the purified truncated MiD proteins and purified Drp1 (Drp1-vc003), as well as their stability in the presence of ATP, ADP, and GDP (250µM to 1mM). This method is often used to identify the stabilisation of proteins against thermal denaturation and the influence of interaction with small molecules during incremental increases in temperature (Niesen et al., 2007).

The presence of varying concentrations of ATP, ADP or GTP caused no significant increase in MiD49 stabilisation (5µM), compared to the control TSA of the protein in the absence of nucleotides, P>0.05 (Table 5.4.3.1 and Figure 5.4.3.1. Similarly, ATP cause no significant increase in structural stabilisation of MiD51 (5µM), P>0.05 (Table 5.4.3.1). There was a significant increase in protein stability during TSA of MiD51 in the presence of ADP or GDP, compared to the nucleotide deficient control (Table 5.4.3.1 and Figure 5.4.3.1). This is in accordance to published literature which has shown the interaction of mouse MiD51 with ADP and GDP (Loson et al., 2014; Richter et al., 2014), and the inability of mouse MiD49 interaction with nucleotides, at its nucleotidyltransferase fold (Loson et al., 2015).

Nucleotide Tm In the presence of nucleotides (°C)

concentration ATP ADP GDP Control (0µM) 50.85 50.85 50.85 250µM 49.23 49.55 50.92 MiD49 500µM 49.23 49.55 50.92 1000µM 50.18 49.12 50.29 Control (0µM) 54.50 54.50 54.50 250µM 54.44 56.56 54.86 MiD51 500µM 54.44 57.30 55.92 1000µM 54.44 58.00 57.62

Table 5.4.3.1 Thermal shift assay of human MiD proteins in the presence of nucleotides

Consequently, TSA of purified Drp1 was carried out in the presence of GDP, GTP or the small peptides (Table 5.4.2.1). As expected, the interaction of GDP or GTP with Drp1 improved protein stability and caused a positive shift in the TSA curve (Stowell et al., 1999; Praefcke & McMahon, 2004). The MiD small peptides caused no significant increase in Drp1 stability (Figure 5.4.3.2). The only significant effect

189

observed was a negative shift in Drp1 stability observed in the presence of 1mM human MiD51 peptide (RENPEYFPRGSSYMD-RCVVG). Additionally, the presence of GTP and the small peptide caused no significant change to Drp1 stability (Figure 5.4.3.3). Interestingly, the addition of Mdivi-1 also had no significant effect on Drp1 stability, in the presence or absence of GTP (Figure 5.4.3.4), further supporting previously published work indicating that Mdivi-1 does not directly interact with Drp1 (Bordt et al., 2017).

190

Figure 5.4.3.1 Thermal shift assay of purified MiD49 and MiD51 in the presence of nucleotides ATP, ADP or GDP at concentrations of 250µM, 500µM and 1mM cause no significant increase in MiD49 stability. Average Tm values shown in Table 5.4.3.1, N=3. There was a significant increase in MiD51 protein stability in the presence of ADP or GDP (P<0.05), but no significant increase in protein stability was observed in the presence of ATP, compared to the nucleotide deficient control (P>0.05). Statistical significance assessed by unpaired t-test of each concentration compared to the control (GraphPad Prism 6 Software, La Jolla, CA).

191

Figure 5.4.3.2 Thermal shift assay of purified Drp1 in the presence of small MiD peptides Treatment of 5µM Drp1 with GDP or GTP, at concentrations of 500µM and 1mM, caused a significant increase in Drp1 stability (A and B, P<0.05). Drp1 treated with MiD small peptides at various concentrations (C-H, N=3). The peptides did not cause a significant increase in Drp1 stability (P>0.05). The only significant outcome observed was the negative shift in Drp1 stabilisation in the presence of Human MiD51 peptide, at a 1mM concentration (F, P<0.05). Statistical significance assessed by unpaired t-test of each concentration compared to the control (GraphPad Prism 6 Software, La Jolla, CA). 192

Figure 5.4.3.3 Thermal shift assay of purified Drp1 in the presence of GTP and small MiD peptides 5µM of Drp1 was pre-treated with 1µM of GTP before the addition of the MiD small peptides at various concentrations (A-F, N=3). There was no significant increase in Drp1 stability following treatment with GTP and MiD small peptides (P>0.05). Statistical significance assessed by unpaired t-test of each concentration compared to the control (GraphPad Prism 6 Software, La Jolla, CA).

193

Figure 5.4.3.4 Thermal shift assay of purified Drp1 in the presence of Mdivi-1 Mdivi-1 treatment caused no significant change in Drp1 stability. The presence of 1mM GTP within the assay caused a positive shift, as previously observed (Figure 5.4.3.2 B); however, the presence of Mdivi-1 had no additive effect, indicating that it does not interact with Drp1 (N=3, P>0.05). Statistical significance assessed by unpaired t-test of each concentration compared to the control (GraphPad Prism 6 Software, La Jolla, CA).

5.4.4 Optimisation of surface plasmon resonance assay

Following TSA, we tried to investigate the small peptides and human MiD proteins interaction with human Drp1, using SPR. The assay allows RT identification molecules’ interaction and kinetics with our protein of interest (Drp1), in a label-free system. Amide coupling was used to immobilise Drp1 to the sensor surface. This chip was first used with human MiD51 as the analyte, prior to combining the sample with the small lipids, to see if their competitive interaction with Drp1 can reduce Drp1-MiD binding. The affinity of interaction was measured by calculating the dissociation constant (KD). Smaller KD values represent a higher binding affinity to the ligand.

Drp1 was successfully immobilised to create the active cell, using immobilisation buffer (10mM sodium acetate, pH 5), MiD51 was introduced to the SPR system in 1XPBS at pH 7.2 (single-cycle kinetics analysis of MiD51 binding). A reference cell was used to eliminate sensorgram values which are caused as a result of MiD51 interaction with the sensor surface (sensor deficient of Drp1 ligands). The presence of MiD51 produced negative binding response sensorgram, when the values from the reference cell were taken away from the active cell sensorgram values (Figure 5.4.2.1 A). This indicated that there was a high level of MiD51 interaction with the sensor surface. To prevent this interaction, BSA (50µg/ml) was used to block the 194

reference cell at a flow rate of 5ul/min, prior to the introduction of MiD51. This led to a positive binding response, but no binding events were observed between Drp1 and MiD51. No significant change to the assay was observed when MiD51 was complexed with ADP (at a saturating concentration or 100µM, Figure 5.4.4.1B), prior to SPR analysis (Figure 5.4.4.1C). Theoretical response values indicated in black represent the maximum theoretical analyte response. The observed experimental analyte interaction is shown in red (Figure 5.4.4.1 B,C and Figure 5.4.4.2).

195

Figure 5.4.4.1 Drp1 surface plasmon resonance assay optimisation The negative SPR curve indicates the presence of MiD51 interaction with the sensor surface (A). Blocking the reference cell with BSA inhibited this interaction of the reference cells; however, interaction with immobilised Drp1 could still not be detected in the absence (B) or presence of ADP (C). SPR calculated theoretical response indicated in black and experimental observed response indicated in red (B and C).

Subsequently, MiD51 was immobilised to the sensor surface instead, and single- cycle kinetics analysis of Drp1 binding was carried out. Binding kinetics Drp1 with

196

MiD51 was confirmed in the presence or absence of ADP (100µM ADP complexed with Drp1 prior to binding analysis). The binding of Drp1 to MiD51 assay was reproducible when MiD51 was immobilised. Interestingly, the binding affinity between the proteins was higher in the absence of ADP (KD=334nM, Figure 5.4.4.2

A), than when Drp1 was complexed with ADP (KD=465nM, Figure 5.4.4.2 B).

Figure 5.4.4.2 MiD51 surface plasmon resonance assay optimisation Binding of Drp1 to immobilised MiD51 in the absence (A) or presence of ADP (B). The binding affinity of immobilised MiD51 was higher in the absence of ADP than in the presence of ADP. There was a high level of overlap between the SPR calculated theoretical response indicated in black and experimental observed response indicated in red.

197

5.5 Discussion

Thermal shift assay was used to identify the stability of purified truncated MiD49, MiD51 and Drp1. We identified that the purified proteins were stable and could be used in biophysical assays. The addition of MiD small peptides with the same amino acid sequence as to the mouse and human Drp1-binding loop regions, identified in previous publications and our own computation studies, caused no significant increase in Drp1 stability. We set up an SPR assay to determine if the small peptides were capable of binding to Drp1 and could reduce its interaction with Drp1.

5.5.1 Thermal Shift Assay of Drp1 and MiD proteins

TSA was first used to ensure that the truncated proteins were stable and maintained the same tertiary structure as the native proteins. MiD49 and MiD51 were truncated as previous studies were unable to create stable protein crystals in the presence of their membrane proximal region, which is believed to lack a secondary structure (Loson et al., 2014; Ma & Sun, 2014; Richter et al., 2014; Loson et al., 2015). The deletion of this region does not affect the proteins’ ability to interact with Drp1 or prevent MiD51 to interact with nucleotides; however, the deletion of this region may affect MiD49 dimerisation as the group were not able to detect dimerization of mouse MiD49, which occurs with human MiD49 (Loson et al., 2015).

The addition of ATP, ADP or GDP caused no significant increase in human MiD49 stabilisation compared to the control TSA of the protein in the absence of nucleotides (Table 5.4.3.1 and Figure 5.4.3.1).This was an expected outcome, as the mouse MiD49 structure, which has a sequence identity of 80% with human MiD49 (Figure 5.4.2.1 A), has been shown to be incapable of binding nucleotides at its NTase fold (Loson et al., 2015). Human MiD51 and mouse MiD51 share a 98% aa sequence identity, and the addition of ATP caused no significant change to MiD51 stability, indicating that the protein does not have a significant binding affinity for ATP at its NTase fold (Figure 5.4.3.1). This was in accordance with previously published data of mouse MiD51 (Richter et al., 2014; Loson et al., 2015). Human MiD51 structural stability significantly increased in the presence of ADP and GDP (Table 5.4.3.1 and Figure 5.4.3.1), which has been previously identified in mouse MiD51 to be due to its high affinity binding of these nucleotides at the NTase fold

198

(Loson et al., 2014; Richter et al., 2014). Comparison of this data with structurally similar mouse MiD protein and TSA results indicated that the human MiD proteins have a similar interaction with ADP, ATP and GDP as their mouse homologues. This was the expected outcome, due to the high structural similarities of the human and mouse proteins, indicating that the purified proteins had folded in a similar way to the native structure and stable enough to be used for our investigation.

Purified Drp1 proteins gained stability in the presence of GTP and GDP, which are known to be essential for Drp1’s GTPase activity during mitochondrial fission (Stowell et al., 1999; Praefcke & McMahon, 2004), and caused a positive shift in the TSA curve. Unfortunately, the addition of the small peptides, with the aa sequence of the MiD-Drp1 binding loops, caused no significant increase in Drp1 stability (Figure 5.4.3.2). Additionally, the presence of GTP with the small peptide had no additive effect on Drp1 stability, compared to the addition of GTP alone (Figure 5.4.3.3). This data suggest that if any interaction takes place between the small peptides and Drp1, it causes no significant change in Drp1 stability. Interestingly, the addition of Mdivi-1 also had no significant effect to Drp1 stability, in the presence or absence of GTP (Figure 5.4.3.4), further supporting a recent study which questions whether Mdivi-1 is a direct inhibitor of Drp1 (Bordt et al., 2017).

The limitation with TSA is that unless there is a significant change in protein stability, the data cannot be used to identify if any interaction does take place between the small peptides and the target protein. As the expected interactions with the small peptides are to occur on the surface of Drp1 and not within nucleotide binding pockets, the interaction is indeed less likely to change the stability of the protein during TSA. Another limitation of using TSA in the presence of small peptides is that the peptide structure is not able to withstand similar melting temperatures as Drp1 and therefore the peptide‘s secondary structure is quickly lost during the early stages of the assay. We next tried to use SPR to identify interactions between Drp1 and the small peptides.

5.5.2 Optimisation of surface plasmon resonance assay

We decided to used SPR to identify any interactions between the small peptides of the MiD loop structures and Drp1. To test this, we needed to first establish the

199

binding events and kinetics between Drp1 and the MiD proteins, using sensor chips with immobilised Drp1. Amide coupling was used to immobilise Drp1 to the sensor surface. We first used MiD51 as our analyte to measure interactions in RT, unexpectedly, the comparison of the sensorgram results with a reference cell revealed that MiD51 had a high interaction with the sensor surface (Figure 5.4.4.1A). To prevent this interaction, the reference cell was blocked with BSA, which allowed the measurement of the binding response (Figure 5.4.4.1B). Unfortunately, in such a setting, no interaction with Drp1 could be identified. MiD51 was complexed with ADP prior to SPR, as some research groups believe that ADP binding is essential for MiD51-Drp1 interaction (Loson et al., 2014), however, this had no significant increase in binding events (Figure 5.4.4.1C).

Interestingly immobilised MiD51 to the sensor surface allowed the measurement of interactions with Drp1 and identified that the binding affinity between MiD51 and

Drp1 was higher, in the absence of ADP (absence of ADP KD=334nM, with ADP

KD=465nM, Figure 5.4.4.2). A reduction in binding affinity following nucleotide binding may be essential in the regulation of mitochondrial fragmentation. A reduction in Drp1 association during high levels of cellular ADP may help to promote mitochondrial fusion, which has been associated with maximising the mitochondrial oxidative capacity, in response to increased energy demands (Bach et al., 2003; Ishihara et al., 2003; Chen et al., 2005; Tondera et al., 2009). If a similar reduction in affinity occurs following GDP binding, this could be an important step in mitochondrial fragmentation, allowing for faster MiD51-Drp1 dissociation, following Drp1 GTP hydrolysis during mitochondrial constriction (Stowell et al., 1999; Praefcke & McMahon, 2004; Mears et al., 2011). Further investigation is required to identify if MiD51 is more responsive to metabolic changes due to its ability to bind nucleotides, and may help explain its lower expression levels in the heart, compared to MiD49 (Liu et al., 2013).

5.5.3 Future work

We have been able to purify human MiD49, MiD51 and Drp1 in their native folded state and identified small peptides mirroring the MiD loop structure required for Drp1 binding that may be capable of inhibiting MiD-Drp1 interaction. Unfortunately, we were not able to show their interaction with Drp1 by TSA, as no significant positive

200

shift could be observed. We decided to use SPR to further investigate the possibility of their interaction with Drp1. SPR allows the flow of multiple analytes through the system, and binding interactions can be identified on the sensorgram depending on the molecules’ size. This allows us to distinguish the binding events of peptides and purified MiD proteins to Drp1, in the same system (Mehand et al., 2015). As we are interested in identifying the interaction of small peptides with Drp1, and their competitive inhibition of MiD interaction, the immobilised Drp1 model requires further optimisation. This assay still requires optimisation due to the interaction of MiD51 with the sensor surface. A similar outcome is likely to be observed with MiD49, due to the two proteins high structural similarities (Loson et al., 2014; Ma & Sun, 2014; Richter et al., 2014; Loson et al., 2015).

Drp1 immobilisation by amide coupling of the vc008 construct, with a C-terminus His tag, may allow the identification of binding events with MiD49 and MiD51. Additionally, other SPR immobilisation techniques such as Ni-NTA and Streptavidin- biotin capture will be investigated (Kim & Herr, 2013).

Alongside these experiments, we plan to conduct a fragment-based high throughput assay screen of known compounds using TSA. The compounds can subsequently be used in the optimised SPR assay with immobilised Drp1 to identify any reduction in MiD-Drp1 binding. The use of small fragments can be advantageous as they tend to have a longer half-life and therefore can be administrated in lower doses. Small molecules also allow for a broader range of drug delivery methods as well as being more cost-effective to produce (Cardoso et al., 2014). The identified small molecules will be used in a cell-based assay to monitor the changes in mitochondrial morphology. We hope to be able to identify small molecules which will be capable of acutely inhibiting cardiac MiD-Drp1 interaction upon reperfusion. The efficacy of the molecules’ ability to reduce myocardial infarction will be investigated by comparing infarct size of treated and non-treated mice, post in vivo acute IRI.

201

CHAPTER 6: Overall Discussion and future work

The overall aim of this thesis was to investigate MiD49 and MiD51, newly described mitochondrial-specific fission proteins, as potential targets of cardioprotection. This has been investigated in cardiac cell lines and a mouse in vivo model of MI, using transgenic MiD49 KO mice. The cardioprotective efficacy of targeting the MiD proteins has been discussed in detail in chapter three, four and five. Overall the modification of MiD49 and MiD51 by the overexpression or the knockdown of the proteins caused a significant increase in mitochondrial elongation within cardiac cell lines, similar to other cell lines previously investigated (Palmer et al., 2011; Loson et al., 2013; Palmer et al., 2013).

The elongated mitochondrial morphology observed after the overexpression of MiD49 and MiD51 was distinct from the morphology of the positive control (Mfn2 overexpression), as well as the elongated morphology after MiD KD in cell lines. MiD overexpression caused the formation of hyperfused, collapsed clusters, which were often perinuclear. HL-1 and MEF cells expressing this morphology appeared larger in size. This observation needs to be further investigated and may be caused as a result of preventing mitochondrial distribution during cellular division (Frazier et al., 2006; Chan, 2012). The overexpression of the MiD proteins is believed to cause a significant increase in mitochondrial susceptibility to stress (Loson et al., 2013). However, cells overexpressing MiD49 or MiD51 were less susceptible to MPTP opening and cell death post SIRI. This may be due to the cardioprotective outcomes of reducing mitochondrial fragmentation during IRI, which is believed to enable a higher accumulation of MM ROS before the opening of the MPTP, providing the cells with a longer period of time to recover (Ong et al., 2010). Another possibility is that the cells may return to their mitochondrial hyperfused state soon after reperfusion (Tondera et al., 2009; Gomes et al., 2011; Rambold et al., 2011), or that the collapsed networks are incapable of undergoing mitochondrial fragmentation, due to a reduction of ER and actin filament interaction (Friedman et al., 2011; Ji et al., 2015). To gain a better understanding of this outcome, changes in mitochondrial morphology and interactions with other organelles/ cytoskeletal structures should be further investigated using the RT model of SIRI.

202

As the effects observed by the overexpression of MiD49 or MiD51 proteins are not physiological, we continued our studies by KD of the proteins. The KD of the MiD proteins in HL-1 and H9c2 cardiac cells promoted mitochondrial elongation, however, this change in mitochondrial morphology was only statistically significant following the KD of both proteins, when compared to vector control cells. The KD of MiD49 and MiD51 significantly improved cell survival post SIRI, which was further investigated in real-time.

The KD of the proteins resulted in a significant reduction in mitochondrial fragmentation during ischaemia and a faster recovery of mitochondrial morphology upon reperfusion. Interestingly these elongated mitochondria were also less likely to undergo calcium overload upon reperfusion, which is one of the contributing factors of MPTP opening (Duchen et al., 1993; Ruiz-Meana et al., 2006; Yellon & Hausenloy, 2007). This outcome may be due to a reduction of the MiD proteins’ ER/SR interaction (Elgass et al., 2015). A reduction in SR-mitochondrial interaction has been shown to protect cardiomyocytes against IRI, by reducing calcium overload upon reperfusion (Hall et al., 2016).

Mitochondrial fusion, achieved by MiD KD delayed MPTP opening in the presence of high MM ROS, supporting previously published data suggesting that elongated mitochondria are protected against the oxidative burst which occurs upon reperfusion (Ong et al., 2010; Wang et al., 2011; Disatnik et al., 2013; Zepeda et al., 2014).Changes in ROS levels during RT SIRI, as a result of MiD49 and MiD51 KD, remains to be investigated. We predict that elongated mitochondria will be able to accommodate higher levels of ROS, which may extend the recovery period upon reperfusion, and will therefore, protect against MPTP opening and the release of pro-apoptotic factors soon after reperfusion (Braunwald & Kloner, 1985; Crompton et al., 1987; Griffiths & Halestrap, 1995). It is also important to investigate the proteins’ effect on respiration and mitochondrial membrane potential which affect the level of oxidative stress and calcium overload upon reperfusion (Halestrap et al., 2004).

These findings indicate that targeting both MiD49 and MiD51 activity may be potentially cardioprotective. To further investigate their role in cardioprotection, experiments were continued in vivo. Unfortunately, as the dual KO of MiD49 and 203

MiD51 is embryonically lethal, investigations were carried out using the whole body MiD49 KO transgenic mice. MiD49 expression is higher in the adult heart, and for this reason, we believed that using a MiD49 KO mouse colony was a better animal model for our initial studies (Liu et al., 2013).

Analysis of 2D EM images of MiD49 KO mice and their WT littermates showed no significant difference in interfibrillar mitochondrial length between the two groups. Analysis of 2D images has limitations, and there is currently no universal method of analysing mitochondrial morphology in cardiomyocytes. Recent 3D imaging studies have shown the presence of vast interconnected mitochondrial networks that cannot be identified using 2D imaging (Hayashi et al., 2009; Beikoghli Kalkhoran et al., 2017; Kalkhoran et al., 2017).

Long-term disruptions of mitochondrial dynamics in the heart, which induce extreme changes in mitochondrial morphology, significantly affect the cardiac phenotype and function (Papanicolaou et al., 2011; Piquereau et al., 2012; Ikeda et al., 2015) No significant change in cardiac structure or function could be detected from echocardiography imaging of MiD49 KO mice, compared to their WT littermates. Cardiac isoproterenol stress induction did not reveal a significant difference in the cardiac phenotype of MiD49 KO mice, which may not be evident during basal conditions, compared to their WT littermates (Rottman et al., 2007; Dongworth et al., 2014; Hall et al., 2016). Chronic modifications to the cardiac fission machinery by Drp1 KO in mice resulted in a significant increase in mitochondrial fragmentation, induced HF, and caused death within only 13 weeks of Drp1 KO. These effects consequently cause a significant increase in infarct size following IRI (Ikeda et al., 2015). The absence of any overt myocardial pathologies as a result of MiD49 KO, allowed the investigation of cardiac ischaemia-reperfusion susceptibility by LAD occlusion, compared to WT littermates, without the influence of confounding factors. There was no significant difference in MI size between MiD49 KO hearts and WT hearts, following in vivo LAD occlusion.

204

6.1 Ongoing and Future work

6.1.1 Cardiac-specific knockdown of MiD49 and MiD51

Results from our in vitro studies showed that both MiD49 and MiD51 activity must be inhibited to cause a significant change in mitochondrial morphology and protect the cells against SIRI. In collaboration with Assistant Prof Jianming Jiang and Prof Roger Foo’s research group, we have developed cardiac-specific MiD49 and MiD51 targeting rAAV9 constructs (Jiang et al., 2013; Jiang et al., 2015). These constructs will be used to simultaneously KD both proteins in the mouse heart. This model will allow us to further investigate changes in mitochondrial and cardiac morphology and function in mice, and identify if targeting both proteins is essential to achieve cardioprotection against IRI, using the murine in vivo model of MI. We predict that rAAV9 mediated MiD49 and MiD51 KD will not induce a detrimental cardiac phenotype, observed following chronic Drp1 KO (Ikeda et al., 2015), as mitochondrial fragmentation is downregulated but not fully inhibited, due to the presence of Mff, Fis1 and low-level expression of the MiDs (Stojanovski et al., 2004; Otera et al., 2010; Palmer et al., 2011). Another advantage of this model is that up to 85% KD of target proteins can be achieved within 72h of infection (Prasad et al., 2011; Jiang et al., 2013; Jiang et al., 2015). For this reason, our experimental model is a more representative model of acute MiD inhibition, during in vivo MI surgery, and also less likely to cause confounding effects as a result of the chronic alteration of mitochondrial dynamics (Papanicolaou et al., 2011; Papanicolaou et al., 2012; Ikeda et al., 2015; Hall et al., 2016; Zhou et al., 2017a).

6.1.2 Development of small molecule inhibitors of MiD49 and MiD51

The MiD proteins interact with Drp1 via a highly conserved surface loop (Loson et al., 2014; Richter et al., 2014; Loson et al., 2015; Kalia et al., 2018). The destabilisation of the loop structure or its deletion blocks MiD49 and MiD51 recruitment of Drp1 to the mitochondria, leading to unopposed mitochondrial fusion (Loson et al., 2014; Richter et al., 2014; Loson et al., 2015; Kalia et al., 2018). As Drp1 has a predominantly globular structure, the site in which the MiD proteins interact with Drp1 has been challenging to identify. A recent study has identified that dynamin recruitment region of MiD49 and MiD51, containing the binding loop, is capable of interacting with four Drp1 molecules, at four possible surfaces within the

205

Drp1 rings (Kalia et al., 2018). In collaboration with Dr Joma Joy, Dr Jeffrey Hill and Dr Anna Jansson (Experimental Therapeutics Centre, Agency for Science, Technology and Research, Singapore), we aim to identify small peptides capable of competitive inhibition of the Drp1-MiD interaction. The identified peptides/ small molecules will be administrated during ex vivo and in vivo models of MI, to identify if the acute inhibition of MiD49 and MiD51 is cardioprotective.

6.1.3 Investigating the role of MiD49 in human fibroblasts and induced Pluripotent Stem Cell-Derived Cardiomyocytes

We have been fortunate to obtain human fibroblasts lacking MiD49 expression, due to homozygous nonsense mutation, as a kind gift of a collaborator, Prof Rita Horvath from Newcastle University. This mutation was identified to cause progressive muscle weakness and exercise intolerance. The authors do not mention the presence of a particular cardiac phenotype but this may develop over time, as chronic alterations in the mitochondrial dynamics machinery lead to the development of detrimental cardiac conditions (Papanicolaou et al., 2011; Papanicolaou et al., 2012; Ikeda et al., 2015; Hall et al., 2016). Isolated patient fibroblasts expressed a highly elongated mitochondrial network, and a higher expression of ETC complexes (Bartsakoulia et al., 2018). We hope to use these fibroblasts, and control fibroblasts isolated from healthy individuals, to develop human induced pluripotent stem cell-derived cardiomyocytes (iPSC-CMs). The use of iPSC-CMs cells in cell models of IRI will provide vital transitional evidence of the human cardiac response to acutely targeting the MiD proteins’ activity during IRI, and their potential link to changes in cellular metabolism (Bartsakoulia et al., 2018).

6.1.4 Conclusion

From the findings of in vitro and in vivo studies of MiD49 and MiD51, we believe that both proteins must be targeted to cause a significant change in mitochondrial morphology. Due to their mitochondrial specific nature of Drp1 recruitment, we believe that these proteins should only be inhibited acutely to protect the heart against IRI, as chronic manipulation of the dynamic machinery leads to the development of adverse structural and functional cardiac conditions (Papanicolaou et al., 2011; Papanicolaou et al., 2012; Ikeda et al., 2015; Hall et al., 2016). Due to this, we are currently investigating methods for specifically knocking down MiD49

206

and MiD51 proteins in the heart using rAAV9 constructs, capable of inducing up to 85% protein KD within 72h of infecting cardiac tissue (Jiang et al., 2013; Jiang et al., 2015). These constructs can be used to monitor the short-term and long-term effects of dual cardiac-specific MiD49 and MiD51 KD in the heart. In parallel to this work, we are using pharmaceutical screening techniques to identify small molecules or peptides capable of inhibiting MiD-Drp1 interaction, which could be administrated upon reperfusion. The role of the MiD49 and MiD51 proteins as potential targets of cardioprotection will also be investigated in iPSC-CMs to generate transitional evidence of the human cardiac response to the acute inhibition of MiD49 and MiD51.

207

CHAPTER 7: References

Akram M. (2014). Citric acid cycle and role of its intermediates in metabolism. Cell biochemistry and biophysics 68, 475-478.

Al-Mehdi AB, Pastukh VM, Swiger BM, Reed DJ, Patel MR, Bardwell GC, Pastukh VV, Alexeyev MF & Gillespie MN. (2012). Perinuclear mitochondrial clustering creates an oxidant-rich nuclear domain required for hypoxia- induced transcription. Science signaling 5, ra47.

Anderson SE, Cala PM, Steenbergen C, London RE & Murphy E. (1991). Effects of hypoxia and acidification on myocardial Na and Ca. Role of Na-H and Na-Ca exchange. Annals of the New York Academy of Sciences 639, 453-455.

Atchison RW, Casto BC & Hammon WM. (1965). ADENOVIRUS-ASSOCIATED DEFECTIVE VIRUS PARTICLES. Science (New York, NY) 149, 754-756.

Avkiran M & Marber MS. (2002). Na(+)/H(+) exchange inhibitors for cardioprotective therapy: progress, problems and prospects. Journal of the American College of Cardiology 39, 747-753.

Bach D, Pich S, Soriano FX, Vega N, Baumgartner B, Oriola J, Daugaard JR, Lloberas J, Camps M, Zierath JR, Rabasa-Lhoret R, Wallberg-Henriksson H, Laville M, Palacin M, Vidal H, Rivera F, Brand M & Zorzano A. (2003). Mitofusin-2 determines mitochondrial network architecture and mitochondrial metabolism. A novel regulatory mechanism altered in obesity. The Journal of biological chemistry 278, 17190-17197.

Bai Y. (2015). Detecting protein-protein interactions by gel filtration chromatography. Methods in molecular biology (Clifton, NJ) 1278, 223-232.

Baines CP, Kaiser RA, Purcell NH, Blair NS, Osinska H, Hambleton MA, Brunskill EW, Sayen MR, Gottlieb RA, Dorn GW, Robbins J & Molkentin JD. (2005). Loss of cyclophilin D reveals a critical role for mitochondrial permeability transition in cell death. Nature 434, 658-662.

208

Baines CP, Kaiser RA, Sheiko T, Craigen WJ & Molkentin JD. (2007). Voltage- dependent anion channels are dispensable for mitochondrial-dependent cell death. Nature cell biology 9, 550-555.

Bartelds B, Knoester H, Smid GB, Takens J, Visser GH, Penninga L, van der Leij FR, Beaufort-Krol GC, Zijlstra WG, Heymans HS & Kuipers JR. (2000). Perinatal changes in myocardial metabolism in lambs. Circulation 102, 926- 931.

Bartlett JS, Wilcher R & Samulski RJ. (2000). Infectious entry pathway of adeno- associated virus and adeno-associated virus vectors. Journal of virology 74, 2777-2785.

Bartsakoulia M, Pyle A, Troncosco D, Vial J, Paz-Fiblas MV, Duff J, Griffin H, Boczonadi V, Lochmuller H, Kleinle S, Chinnery PF, Grunert S, Kirschner J, Eisner V & Horvath R. (2018). A novel mechanism causing imbalance of mitochondrial fusion and fission in human myopathies. Human molecular genetics.

Baughman JM, Perocchi F, Girgis HS, Plovanich M, Belcher-Timme CA, Sancak Y, Bao XR, Strittmatter L, Goldberger O, Bogorad RL, Koteliansky V & Mootha VK. (2011). Integrative genomics identifies MCU as an essential component of the mitochondrial calcium uniporter. Nature 476, 341-345.

Beikoghli Kalkhoran S, Hall AR, White IJ, Cooper J, Fan Q, Ong SB, Hernández‐ Reséndiz S, Cabrera‐Fuentes H, Chinda K, Chakraborty B, Dorn GW, Yellon DM & Hausenloy DJ. (2017). Assessing the effects of mitofusin 2 deficiency in the adult heart using 3D electron tomography. Physiological Reports 5.

Bers DM. (2002). Cardiac excitation-contraction coupling. Nature 415, 198-205.

Bhamra GS, Hausenloy DJ, Davidson SM, Carr RD, Paiva M, Wynne AM, Mocanu MM & Yellon DM. (2008). Metformin protects the ischemic heart by the Akt- mediated inhibition of mitochondrial permeability transition pore opening. Basic research in cardiology 103, 274-284.

Bhatnagar P, Wickramasinghe K, Williams J, Rayner M & Townsend N. (2015). The epidemiology of cardiovascular disease in the UK 2014. Heart (British Cardiac Society) 101, 1182-1189.

209

Bi W, Yan J, Stankiewicz P, Park SS, Walz K, Boerkoel CF, Potocki L, Shaffer LG, Devriendt K, Nowaczyk MJ, Inoue K & Lupski JR. (2002). Genes in a refined Smith-Magenis syndrome critical deletion interval on chromosome 17p11.2 and the syntenic region of the mouse. Genome research 12, 713-728.

Bish LT, Morine K, Sleeper MM, Sanmiguel J, Wu D, Gao G, Wilson JM & Sweeney HL. (2008). Adeno-associated virus (AAV) serotype 9 provides global cardiac gene transfer superior to AAV1, AAV6, AAV7, and AAV8 in the mouse and rat. Human gene therapy 19, 1359-1368.

Bolli R, Patel BS, Jeroudi MO, Lai EK & McCay PB. (1988). Demonstration of free radical generation in "stunned" myocardium of intact dogs with the use of the spin trap alpha-phenyl N-tert-butyl nitrone. The Journal of clinical investigation 82, 476-485.

Bonora M, Bononi A, De Marchi E, Giorgi C, Lebiedzinska M, Marchi S, Patergnani S, Rimessi A, Suski JM, Wojtala A, Wieckowski MR, Kroemer G, Galluzzi L & Pinton P. (2013). Role of the c subunit of the FO ATP synthase in mitochondrial permeability transition. Cell cycle (Georgetown, Tex) 12, 674- 683.

Bordt EA, Clerc P, Roelofs BA, Saladino AJ, Tretter L, Adam-Vizi V, Cherok E, Khalil A, Yadava N, Ge SX, Francis TC, Kennedy NW, Picton LK, Kumar T, Uppuluri S, Miller AM, Itoh K, Karbowski M, Sesaki H, Hill RB & Polster BM. (2017). The Putative Drp1 Inhibitor mdivi-1 Is a Reversible Mitochondrial Complex I Inhibitor that Modulates Reactive Oxygen Species. Developmental cell 40, 583-594.e586.

Boutin S, Monteilhet V, Veron P, Leborgne C, Benveniste O, Montus MF & Masurier C. (2010). Prevalence of serum IgG and neutralizing factors against adeno- associated virus (AAV) types 1, 2, 5, 6, 8, and 9 in the healthy population: implications for gene therapy using AAV vectors. Human gene therapy 21, 704-712.

Brand M D & Nicholls D G. (2011). Assessing mitochondrial dysfunction in cells. Biochemical Journal 435, 297-312.

Braunwald E & Kloner RA. (1985). Myocardial reperfusion: a double-edged sword? The Journal of clinical investigation 76, 1713-1719.

210

Brini M & Carafoli E. (2011). The Plasma Membrane Ca(2+) ATPase and the Plasma Membrane Sodium Calcium Exchanger Cooperate in the Regulation of Cell Calcium. Cold Spring Harbor Perspectives in Biology 3, a004168.

Brodie BR, Stuckey TD, Wall TC, Kissling G, Hansen CJ, Muncy DB, Weintraub RA & Kelly TA. (1998). Importance of time to reperfusion for 30-day and late survival and recovery of left ventricular function after primary angioplasty for acute myocardial infarction. Journal of the American College of Cardiology 32, 1312-1319.

Bround MJ, Asghari P, Wambolt RB, Bohunek L, Smits C, Philit M, Kieffer TJ, Lakatta EG, Boheler KR, Moore ED, Allard MF & Johnson JD. (2012). Cardiac ryanodine receptors control heart rate and rhythmicity in adult mice. Cardiovascular research 96, 372-380.

Buchholz CJ, Friedel T & Büning H. (2015). Surface-Engineered Viral Vectors for Selective and Cell Type-Specific Gene Delivery. Trends in Biotechnology 33, 777-790.

Bunting JR, Phan TV, Kamali E & Dowben RM. (1989). Fluorescent cationic probes of mitochondria. Metrics and mechanism of interaction. Biophysical journal 56, 979-993.

Burger C, Gorbatyuk OS, Velardo MJ, Peden CS, Williams P, Zolotukhin S, Reier PJ, Mandel RJ & Muzyczka N. (2004). Recombinant AAV viral vectors pseudotyped with viral capsids from serotypes 1, 2, and 5 display differential efficiency and cell tropism after delivery to different regions of the central nervous system. Molecular therapy : the journal of the American Society of Gene Therapy 10, 302-317.

Campanella M, Casswell E, Chong S, Farah Z, Wieckowski MR, Abramov AY, Tinker A & Duchen MR. (2008). Regulation of mitochondrial structure and function by the F1Fo-ATPase inhibitor protein, IF1. Cell Metab 8, 13-25.

Campello S & Scorrano L. (2010). Mitochondrial shape changes: orchestrating cell pathophysiology. EMBO reports 11, 678-684.

Cardoso F, Lewis R, Vetter I, Inserra M & F King G. (2014). Does Nature do Ion Channel Drug Discovery Better than Us?, vol. 2015. 211

Carll AP, Willis MS, Lust RM, Costa DL & Farraj AK. (2011). Merits of non-invasive rat models of left ventricular heart failure. Cardiovascular toxicology 11, 91- 112.

Castelli WP, Anderson K, Wilson PW & Levy D. (1992). Lipids and risk of coronary heart disease. The Framingham Study. Annals of epidemiology 2, 23-28.

Celsi F, Pizzo P, Brini M, Leo S, Fotino C, Pinton P & Rizzuto R. (2009). Mitochondria, calcium and cell death: a deadly triad in neurodegeneration. Biochimica et biophysica acta 1787, 335-344.

Cereghetti GM, Stangherlin A, Martins de Brito O, Chang CR, Blackstone C, Bernardi P & Scorrano L. (2008). Dephosphorylation by calcineurin regulates translocation of Drp1 to mitochondria. Proceedings of the National Academy of Sciences of the United States of America 105, 15803-15808.

Chan DC. (2012). Fusion and fission: interlinked processes critical for mitochondrial health. Annual review of genetics 46, 265-287.

Chance B & Williams GR. (1955). A method for the localization of sites for oxidative phosphorylation. Nature 176, 250-254.

Chang CR & Blackstone C. (2007). Cyclic AMP-dependent protein kinase phosphorylation of Drp1 regulates its GTPase activity and mitochondrial morphology. The Journal of biological chemistry 282, 21583-21587.

Chang CR, Manlandro CM, Arnoult D, Stadler J, Posey AE, Hill RB & Blackstone C. (2010). A lethal de novo mutation in the middle domain of the dynamin- related GTPase Drp1 impairs higher order assembly and mitochondrial division. The Journal of biological chemistry 285, 32494-32503.

Chen H & Chan DC. (2009). Mitochondrial dynamics--fusion, fission, movement, and mitophagy--in neurodegenerative diseases. Human molecular genetics 18, R169-176.

Chen H, Chomyn A & Chan DC. (2005). Disruption of fusion results in mitochondrial heterogeneity and dysfunction. The Journal of biological chemistry 280, 26185-26192. 212

Chen H, Detmer SA, Ewald AJ, Griffin EE, Fraser SE & Chan DC. (2003). Mitofusins Mfn1 and Mfn2 coordinately regulate mitochondrial fusion and are essential for embryonic development. The Journal of Cell Biology 160, 189-200.

Chen KH, Dasgupta A, Lin J, Potus F, Bonnet S, Iremonger J, Fu J, Mewburn J, Wu D, Dunham-Snary K, Theilmann AL, Jing ZC, Hindmarch C, Ormiston ML, Lawrie A & Archer SL. (2018). Epigenetic Dysregulation of the Drp1 Binding Partners MiD49 and MiD51 Increases Mitotic Mitochondrial Fission and Promotes Pulmonary Arterial Hypertension: Mechanistic and Therapeutic Implications. Circulation.

Cherok E, Xu S, Li S, Das S, Meltzer WA, Zalzman M, Wang C & Karbowski M. (2017). Novel regulatory roles of Mff and Drp1 in E3 ubiquitin ligase MARCH5–dependent degradation of MiD49 and Mcl1 and control of mitochondrial dynamics. Molecular Biology of the Cell 28, 396-410.

Choi VW, McCarty DM & Samulski RJ. (2006). Host cell DNA repair pathways in adeno-associated viral genome processing. Journal of virology 80, 10346- 10356.

Chouchani ET, Pell VR, Gaude E, Aksentijević D, Sundier SY, Robb EL, Logan A, Nadtochiy SM, Ord ENJ, Smith AC, Eyassu F, Shirley R, Hu C-H, Dare AJ, James AM, Rogatti S, Hartley RC, Eaton S, Costa ASH, Brookes PS, Davidson SM, Duchen MR, Saeb-Parsy K, Shattock MJ, Robinson AJ, Work LM, Frezza C, Krieg T & Murphy MP. (2014). Ischaemic accumulation of succinate controls reperfusion injury through mitochondrial ROS. Nature 515, 431.

Clarke SJ, McStay GP & Halestrap AP. (2002). Sanglifehrin A acts as a potent inhibitor of the mitochondrial permeability transition and reperfusion injury of the heart by binding to cyclophilin-D at a different site from cyclosporin A. The Journal of biological chemistry 277, 34793-34799.

Claycomb WC, Lanson NA, Jr., Stallworth BS, Egeland DB, Delcarpio JB, Bahinski A & Izzo NJ, Jr. (1998). HL-1 cells: a cardiac muscle cell line that contracts and retains phenotypic characteristics of the adult cardiomyocyte. Proceedings of the National Academy of Sciences of the United States of America 95, 2979-2984.

Cooper MA. (2002). Optical biosensors in drug discovery. Nature reviews Drug discovery 1, 515-528. 213

Cortassa S. (2009). Control and Regulation of Integrated Mitochondrial Function in Metabolic and Transport Networks. 10, 1500-1513.

Cribbs JT & Strack S. (2007). Reversible phosphorylation of Drp1 by cyclic AMP- dependent protein kinase and calcineurin regulates mitochondrial fission and cell death. EMBO reports 8, 939-944.

Crompton M, Costi A & Hayat L. (1987). Evidence for the presence of a reversible Ca2+-dependent pore activated by oxidative stress in heart mitochondria. Biochemical Journal 245, 915-918.

Crompton M, Ellinger H & Costi A. (1988). Inhibition by cyclosporin A of a Ca2+- dependent pore in heart mitochondria activated by inorganic phosphate and oxidative stress. Biochemical Journal 255, 357-360.

Cung TT, Morel O, Cayla G, Rioufol G, Garcia-Dorado D, Angoulvant D, Bonnefoy- Cudraz E, Guerin P, Elbaz M, Delarche N, Coste P, Vanzetto G, Metge M, Aupetit JF, Jouve B, Motreff P, Tron C, Labeque JN, Steg PG, Cottin Y, Range G, Clerc J, Claeys MJ, Coussement P, Prunier F, Moulin F, Roth O, Belle L, Dubois P, Barragan P, Gilard M, Piot C, Colin P, De Poli F, Morice MC, Ider O, Dubois-Rande JL, Unterseeh T, Le Breton H, Beard T, Blanchard D, Grollier G, Malquarti V, Staat P, Sudre A, Elmer E, Hansson MJ, Bergerot C, Boussaha I, Jossan C, Derumeaux G, Mewton N & Ovize M. (2015). Cyclosporine before PCI in Patients with Acute Myocardial Infarction. The New England journal of medicine 373, 1021-1031.

Daghestani HN & Day BW. (2010). Theory and applications of surface plasmon resonance, resonant mirror, resonant waveguide grating, and dual polarization interferometry biosensors. Sensors (Basel, Switzerland) 10, 9630-9646.

Davel AP, Brum PC & Rossoni LV. (2014). Isoproterenol Induces Vascular Oxidative Stress and Endothelial Dysfunction via a Giα-Coupled β(2)-Adrenoceptor Signaling Pathway. PLoS ONE 9, e91877.

de Brito OM & Scorrano L. (2008). Mitofusin 2 tethers endoplasmic reticulum to mitochondria. Nature 456, 605-610.

214

de Fougerolles A, Vornlocher HP, Maraganore J & Lieberman J. (2007). Interfering with disease: a progress report on siRNA-based therapeutics. Nature reviews Drug discovery 6, 443-453.

De Vos KJ, Allan VJ, Grierson AJ & Sheetz MP. (2005). Mitochondrial function and actin regulate dynamin-related protein 1-dependent mitochondrial fission. Current biology : CB 15, 678-683.

Dedkova EN & Blatter LA. (2012). Measuring mitochondrial function in intact cardiac myocytes. Journal of molecular and cellular cardiology 52, 48-61.

Del Dotto V, Mishra P, Vidoni S, Fogazza M, Maresca A, Caporali L, McCaffery JM, Cappelletti M, Baruffini E, Lenaers G, Chan D, Rugolo M, Carelli V & Zanna C. (2017). OPA1 Isoforms in the Hierarchical Organization of Mitochondrial Functions. Cell reports 19, 2557-2571.

Delettre C, Lenaers G, Griffoin JM, Gigarel N, Lorenzo C, Belenguer P, Pelloquin L, Grosgeorge J, Turc-Carel C, Perret E, Astarie-Dequeker C, Lasquellec L, Arnaud B, Ducommun B, Kaplan J & Hamel CP. (2000). Nuclear gene OPA1, encoding a mitochondrial dynamin-related protein, is mutated in dominant optic atrophy. Nature genetics 26, 207-210.

Denton RM. (2009). Regulation of mitochondrial dehydrogenases by calcium ions. Biochimica et Biophysica Acta (BBA) - Bioenergetics 1787, 1309-1316.

Din S, Mason M, Volkers M, Johnson B, Cottage CT, Wang Z, Joyo AY, Quijada P, Erhardt P, Magnuson NS, Konstandin MH & Sussman MA. (2013). Pim-1 preserves mitochondrial morphology by inhibiting dynamin-related protein 1 translocation. Proceedings of the National Academy of Sciences of the United States of America 110, 5969-5974.

Disatnik MH, Ferreira JCB, Campos JC, Gomes KS, Dourado PMM, Qi X & Mochly‐ Rosen D. (2013). Acute Inhibition of Excessive Mitochondrial Fission After Myocardial Infarction Prevents Long‐term Cardiac Dysfunction. Journal of the American Heart Association: Cardiovascular and Cerebrovascular Disease 2, e000461.

Dong JY, Fan PD & Frizzell RA. (1996). Quantitative analysis of the packaging capacity of recombinant adeno-associated virus. Human gene therapy 7, 2101-2112. 215

Dongworth RK, Mukherjee UA, Hall AR, Astin R, Ong SB, Yao Z, Dyson A, Szabadkai G, Davidson SM, Yellon DM & Hausenloy DJ. (2014). DJ-1 protects against cell death following acute cardiac ischemia-reperfusion injury. Cell death & disease 5, e1082.

Dorn GW, 2nd. (2015). Mitochondrial dynamism and heart disease: changing shape and shaping change. EMBO molecular medicine 7, 865-877.

Drouin LM & Agbandje-McKenna M. (2013). Adeno-associated virus structural biology as a tool in vector development. Future virology 8, 1183-1199.

Du C, Fang M, Li Y, Li L & Wang X. (2000). Smac, a mitochondrial protein that promotes cytochrome c-dependent caspase activation by eliminating IAP inhibition. Cell 102, 33-42.

Duan D, Li Q, Kao AW, Yue Y, Pessin JE & Engelhardt JF. (1999). Dynamin is required for recombinant adeno-associated virus type 2 infection. Journal of virology 73, 10371-10376.

Duchen MR, McGuinness O, Brown LA & Crompton M. (1993). On the involvement of a cyclosporin A sensitive mitochondrial pore in myocardial reperfusion injury. Cardiovascular research 27, 1790-1794.

Duchen MR, Surin A & Jacobson J. (2003). Imaging mitochondrial function in intact cells. Methods in enzymology 361, 353-389.

Elgass KD, Smith EA, LeGros MA, Larabell CA & Ryan MT. (2015). Analysis of ER- mitochondria contacts using correlative fluorescence microscopy and soft X- ray tomography of mammalian cells. Journal of cell science 128, 2795-2804.

Elmore S. (2007). Apoptosis: a review of programmed cell death. Toxicologic pathology 35, 495-516.

Ferris JAJ & Friesen JM. (1979). Definitions of ischaemia, infarction and necrosis. Forensic Science International 13, 253-259.

216

Finegold JA, Asaria P & Francis DP. (2013). Mortality from ischaemic heart disease by country, region, and age: statistics from World Health Organisation and United Nations. International journal of cardiology 168, 934-945.

Fisher SG & Marber MS. (2002). An in vivo model of ischaemia-reperfusion injury and ischaemic preconditioning in the mouse heart. Journal of pharmacological and toxicological methods 48, 161-169.

Forman MB, Puett DW & Virmani R. (1989). Endothelial and myocardial injury during ischemia and reperfusion: pathogenesis and therapeutic implications. Journal of the American College of Cardiology 13, 450-459.

Fox TD. (2012). Mitochondrial Protein Synthesis, Import, and Assembly. Genetics 192, 1203-1234.

Franco A, Kitsis RN, Fleischer JA, Gavathiotis E, Kornfeld OS, Gong G, Biris N, Benz A, Qvit N, Donnelly SK, Chen Y, Mennerick S, Hodgson L, Mochly- Rosen D & Dorn GW. (2016). Correcting mitochondrial fusion by manipulating mitofusin conformations. Nature 540, 74-79.

Frank S, Gaume B, Bergmann-Leitner ES, Leitner WW, Robert EG, Catez F, Smith CL & Youle RJ. (2001). The role of dynamin-related protein 1, a mediator of mitochondrial fission, in apoptosis. Developmental cell 1, 515-525.

Frazier AE, Kiu C, Stojanovski D, Hoogenraad NJ & Ryan MT. (2006). Mitochondrial morphology and distribution in mammalian cells. Biological chemistry 387, 1551-1558.

Frezza C, Cipolat S, Martins de Brito O, Micaroni M, Beznoussenko GV, Rudka T, Bartoli D, Polishuck RS, Danial NN, De Strooper B & Scorrano L. (2006). OPA1 controls apoptotic cristae remodeling independently from mitochondrial fusion. Cell 126, 177-189.

Friedman JR, Lackner LL, West M, DiBenedetto JR, Nunnari J & Voeltz GK. (2011). ER tubules mark sites of mitochondrial division. Science (New York, NY) 334, 358-362.

Frostegard J, Nilsson J, Haegerstrand A, Hamsten A, Wigzell H & Gidlund M. (1990). Oxidized low density lipoprotein induces differentiation and adhesion 217

of human monocytes and the monocytic cell line U937. Proceedings of the National Academy of Sciences of the United States of America 87, 904-908.

Gao G, Vandenberghe LH, Alvira MR, Lu Y, Calcedo R, Zhou X & Wilson JM. (2004). Clades of Adeno-associated viruses are widely disseminated in human tissues. Journal of virology 78, 6381-6388.

Geoffroy MC & Salvetti A. (2005). Helper functions required for wild type and recombinant adeno-associated virus growth. Current gene therapy 5, 265- 271.

Gertz EW, Wisneski JA, Stanley WC & Neese RA. (1988). Myocardial substrate utilization during exercise in humans. Dual carbon-labeled carbohydrate isotope experiments. The Journal of clinical investigation 82, 2017-2025.

Gomes LC, Di Benedetto G & Scorrano L. (2011). During autophagy mitochondria elongate, are spared from degradation and sustain cell viability. Nature cell biology 13, 589-598.

Graham FL, Smiley J, Russell WC & Nairn R. (1977). Characteristics of a human cell line transformed by DNA from human adenovirus type 5. The Journal of general virology 36, 59-74.

Grieger JC, Snowdy S & Samulski RJ. (2006). Separate basic region motifs within the adeno-associated virus capsid proteins are essential for infectivity and assembly. Journal of virology 80, 5199-5210.

Griffiths EJ & Halestrap AP. (1995). Mitochondrial non-specific pores remain closed during cardiac ischaemia, but open upon reperfusion. Biochemical Journal 307 ( Pt 1), 93-98.

Grimm D & Kay MA. (2003). From virus evolution to vector revolution: use of naturally occurring serotypes of adeno-associated virus (AAV) as novel vectors for human gene therapy. Current gene therapy 3, 281-304.

Grimm D, Lee JS, Wang L, Desai T, Akache B, Storm TA & Kay MA. (2008). In vitro and in vivo gene therapy vector evolution via multispecies interbreeding and retargeting of adeno-associated viruses. Journal of virology 82, 5887-5911.

218

Grover GJ, Atwal KS, Sleph PG, Wang FL, Monshizadegan H, Monticello T & Green DW. (2004). Excessive ATP hydrolysis in ischemic myocardium by mitochondrial F1F0-ATPase: effect of selective pharmacological inhibition of mitochondrial ATPase hydrolase activity. American journal of physiology Heart and circulatory physiology 287, H1747-1755.

Gumina RJ, Buerger E, Eickmeier C, Moore J, Daemmgen J & Gross GJ. (1999). Inhibition of the Na(+)/H(+) exchanger confers greater cardioprotection against 90 minutes of myocardial ischemia than ischemic preconditioning in dogs. Circulation 100, 2519-2526; discussion 2469-2572.

Gustafsson AB & Gottlieb RA. (2008). Heart mitochondria: gates of life and death. Cardiovascular research 77, 334-343.

Hahn BH, Grossman J, Chen W & McMahon M. (2007). The pathogenesis of atherosclerosis in autoimmune rheumatic diseases: roles of inflammation and dyslipidemia. Journal of autoimmunity 28, 69-75.

Halestrap AP. (2006). Calcium, mitochondria and reperfusion injury: a pore way to die. Biochemical Society transactions 34, 232-237.

Halestrap AP. (2010). A pore way to die: the role of mitochondria in reperfusion injury and cardioprotection. Biochemical Society transactions 38, 841-860.

Halestrap AP, Clarke SJ & Javadov SA. (2004). Mitochondrial permeability transition pore opening during myocardial reperfusion--a target for cardioprotection. Cardiovascular research 61, 372-385.

Halestrap AP, Connern CP, Griffiths EJ & Kerr PM. (1997). Cyclosporin A binding to mitochondrial cyclophilin inhibits the permeability transition pore and protects hearts from ischaemia/reperfusion injury. Molecular and cellular biochemistry 174, 167-172.

Hall AR, Burke N, Dongworth RK & Hausenloy DJ. (2014). Mitochondrial fusion and fission proteins: novel therapeutic targets for combating cardiovascular disease. British journal of pharmacology 171, 1890-1906.

219

Hall AR, Burke N, Dongworth RK, Kalkhoran SB, Dyson A, Vicencio JM, Dorn GW, II, Yellon DM & Hausenloy DJ. (2016). Hearts deficient in both Mfn1 and Mfn2 are protected against acute myocardial infarction. Cell death & disease 7, e2238.

Hatano A, Okada J, Washio T, Hisada T & Sugiura S. (2015). Distinct functional roles of cardiac mitochondrial subpopulations revealed by a 3D simulation model. Biophysical journal 108, 2732-2739.

Hausenloy DJ, Duchen MR & Yellon DM. (2003). Inhibiting mitochondrial permeability transition pore opening at reperfusion protects against ischaemia-reperfusion injury. Cardiovascular research 60, 617-625.

Hausenloy DJ & Yellon DM. (2003). The mitochondrial permeability transition pore: its fundamental role in mediating cell death during ischaemia and reperfusion. Journal of molecular and cellular cardiology 35, 339-341.

Hausenloy DJ & Yellon DM. (2007). Reperfusion injury salvage kinase signalling: taking a RISK for cardioprotection. Heart failure reviews 12, 217-234.

Hausenloy DJ & Yellon DM. (2013). Myocardial ischemia-reperfusion injury: a neglected therapeutic target. The Journal of clinical investigation 123, 92- 100.

Hausenloy DJ, Yellon DM, Mani-Babu S & Duchen MR. (2004). Preconditioning protects by inhibiting the mitochondrial permeability transition. American journal of physiology Heart and circulatory physiology 287, H841-849.

Haverkate F, Thompson SG, Pyke SD, Gallimore JR & Pepys MB. (1997). Production of C-reactive protein and risk of coronary events in stable and unstable angina. European Concerted Action on Thrombosis and Disabilities Angina Pectoris Study Group. Lancet (London, England) 349, 462-466.

Hayashi T, Martone ME, Yu Z, Thor A, Doi M, Holst MJ, Ellisman MH & Hoshijima M. (2009). Three-dimensional electron microscopy reveals new details of membrane systems for Ca<sup>2+</sup> signaling in the heart. Journal of cell science 122, 1005.

220

He J, Ford HC, Carroll J, Ding S, Fearnley IM & Walker JE. (2017). Persistence of the mitochondrial permeability transition in the absence of subunit c of human ATP synthase. Proceedings of the National Academy of Sciences of the United States of America 114, 3409-3414.

Hengartner MO. (2000). The biochemistry of apoptosis. Nature 407, 770-776.

Hermonat PL, Labow MA, Wright R, Berns KI & Muzyczka N. (1984). Genetics of adeno-associated virus: isolation and preliminary characterization of adeno- associated virus type 2 mutants. Journal of virology 51, 329-339.

Hermonat PL & Muzyczka N. (1984). Use of adeno-associated virus as a mammalian DNA cloning vector: transduction of neomycin resistance into mammalian tissue culture cells. Proceedings of the National Academy of Sciences of the United States of America 81, 6466-6470.

Herzenberg LA, Sweet RG & Herzenberg LA. (1976). Fluorescence-activated Cell Sorting. Scientific American 234, 108-118.

Hoggan MD, Blacklow NR & Rowe WP. (1966). Studies of small DNA viruses found in various adenovirus preparations: physical, biological, and immunological characteristics. Proceedings of the National Academy of Sciences of the United States of America 55, 1467-1474.

Hoit B, U Khan Z, M Pawloski-Dahm C & Walsh RA. (1997). In vivo determination of left ventricular wall stress-shonening relationship in normal mice, vol. 272.

Hollenbeck PJ & Saxton WM. (2005). The axonal transport of mitochondria. Journal of cell science 118, 5411-5419.

Hom J & Sheu SS. (2009). Morphological dynamics of mitochondria--a special emphasis on cardiac muscle cells. Journal of molecular and cellular cardiology 46, 811-820.

Hunter LA & Samulski RJ. (1992). Colocalization of adeno-associated virus Rep and capsid proteins in the nuclei of infected cells. Journal of virology 66, 317-324.

221

Huser J & Blatter LA. (1999). Fluctuations in mitochondrial membrane potential caused by repetitive gating of the permeability transition pore. Biochemical Journal 343 Pt 2, 311-317.

Ikeda Y, Shirakabe A, Maejima Y, Zhai P, Sciarretta S, Toli J, Nomura M, Mihara K, Egashira K, Ohishi M, Abdellatif M & Sadoshima J. (2015). Endogenous Drp1 mediates mitochondrial autophagy and protects the heart against energy stress. Circulation research 116, 264-278.

Inagaki K, Fuess S, Storm TA, Gibson GA, McTiernan CF, Kay MA & Nakai H. (2006). Robust systemic transduction with AAV9 vectors in mice: efficient global cardiac gene transfer superior to that of AAV8. Molecular therapy : the journal of the American Society of Gene Therapy 14, 45-53.

Ingerman E, Perkins EM, Marino M, Mears JA, McCaffery JM, Hinshaw JE & Nunnari J. (2005). Dnm1 forms spirals that are structurally tailored to fit mitochondria. The Journal of Cell Biology 170, 1021-1027.

Ishihara N, Fujita Y, Oka T & Mihara K. (2006). Regulation of mitochondrial morphology through proteolytic cleavage of OPA1. The EMBO Journal 25, 2966-2977.

Ishihara N, Jofuku A, Eura Y & Mihara K. (2003). Regulation of mitochondrial morphology by membrane potential, and DRP1-dependent division and FZO1-dependent fusion reaction in mammalian cells. Biochemical and biophysical research communications 301, 891-898.

Javadov S, Karmazyn M & Escobales N. (2009). Mitochondrial permeability transition pore opening as a promising therapeutic target in cardiac diseases. The Journal of pharmacology and experimental therapeutics 330, 670-678.

Jeffrey FM, Diczku V, Sherry AD & Malloy CR. (1995). Substrate selection in the isolated working rat heart: effects of reperfusion, afterload, and concentration. Basic research in cardiology 90, 388-396.

Ji WK, Hatch AL, Merrill RA, Strack S & Higgs HN. (2015). Actin filaments target the oligomeric maturation of the dynamin GTPase Drp1 to mitochondrial fission sites. eLife 4, e11553.

222

Jiang J, Burgon PG, Wakimoto H, Onoue K, Gorham JM, O'Meara CC, Fomovsky G, McConnell BK, Lee RT, Seidman JG & Seidman CE. (2015). Cardiac myosin binding protein C regulates postnatal myocyte cytokinesis. Proceedings of the National Academy of Sciences of the United States of America 112, 9046-9051.

Jiang J, Wakimoto H, Seidman JG & Seidman CE. (2013). Allele-specific silencing of mutant Myh6 transcripts in mice suppresses hypertrophic cardiomyopathy. Science (New York, NY) 342, 111-114.

Jonsson U, Fagerstam L, Ivarsson B, Johnsson B, Karlsson R, Lundh K, Lofas S, Persson B, Roos H, Ronnberg I & et al. (1991). Real-time biospecific interaction analysis using surface plasmon resonance and a sensor chip technology. BioTechniques 11, 620-627.

Junge W, Sielaff H & Engelbrecht S. (2009). Torque generation and elastic power transmission in the rotary F(O)F(1)-ATPase. Nature 459, 364-370.

Kaandorp TA, Lamb HJ, Bax JJ, van der Wall EE & de Roos A. (2005). Magnetic resonance imaging of coronary arteries, the ischemic cascade, and myocardial infarction. American heart journal 149, 200-208.

Kageyama Y, Hoshijima M, Seo K, Bedja D, Sysa-Shah P, Andrabi SA, Chen W, Höke A, Dawson VL, Dawson TM, Gabrielson K, Kass DA, Iijima M & Sesaki H. (2014). Parkin-independent mitophagy requires Drp1 and maintains the integrity of mammalian heart and brain. The EMBO Journal 33, 2798-2813.

Kalia R, Wang RY, Yusuf A, Thomas PV, Agard DA, Shaw JM & Frost A. (2018). Structural basis of mitochondrial receptor binding and constriction by DRP1. Nature 558, 401-405.

Kalkhoran SB, Munro P, Qiao F, Ong SB, Hall AR, Cabrera-Fuentes H, Chakraborty B, Boisvert WA, Yellon DM & Hausenloy DJ. (2017). Unique morphological characteristics of mitochondrial subtypes in the heart: the effect of ischemia and ischemic preconditioning. Discoveries (Craiova, Romania) 5.

Kim D & Herr AE. (2013). Protein immobilization techniques for microfluidic assays. Biomicrofluidics 7, 041501.

223

Kirichok Y, Krapivinsky G & Clapham DE. (2004). The mitochondrial calcium uniporter is a highly selective ion channel. Nature 427, 360-364.

Kloner RA & Jennings RB. (2001). Consequences of brief ischemia: stunning, preconditioning, and their clinical implications: part 2. Circulation 104, 3158- 3167.

Koch A, Thiemann M, Grabenbauer M, Yoon Y, McNiven MA & Schrader M. (2003). Dynamin-like protein 1 is involved in peroxisomal fission. The Journal of biological chemistry 278, 8597-8605.

Koh-Stenta X, Joy J, Poulsen A, Li R, Tan Y, Shim Y, Min JH, Wu L, Ngo A, Peng J, Seetoh WG, Cao J, Wee JL, Kwek PZ, Hung A, Lakshmanan U, Flotow H, Guccione E & Hill J. (2014). Characterization of the histone methyltransferase PRDM9 using biochemical, biophysical and chemical biology techniques. Biochemical Journal 461, 323-334.

Kokoszka JE, Waymire KG, Levy SE, Sligh JE, Cai J, Jones DP, MacGregor GR & Wallace DC. (2004). The ADP/ATP translocator is not essential for the mitochondrial permeability transition pore. Nature 427, 461.

Kornberg H. (2000). Krebs and his trinity of cycles. Nature reviews Molecular cell biology 1, 225-228.

Korobova F, Gauvin TJ & Higgs HN. (2014). A role for myosin II in mammalian mitochondrial fission. Current biology : CB 24, 409-414.

Korobova F, Ramabhadran V & Higgs HN. (2013). An actin-dependent step in mitochondrial fission mediated by the ER-associated formin INF2. Science (New York, NY) 339, 464-467.

Kotin RM, Siniscalco M, Samulski RJ, Zhu XD, Hunter L, Laughlin CA, McLaughlin S, Muzyczka N, Rocchi M & Berns KI. (1990). Site-specific integration by adeno-associated virus. Proceedings of the National Academy of Sciences of the United States of America 87, 2211-2215.

Kovács M, Tóth J, Hetényi C, Málnási-Csizmadia A & Sellers JR. (2004). Mechanism of Blebbistatin Inhibition of Myosin II. Journal of Biological Chemistry 279, 35557-35563. 224

Kresge N, Simoni RD & Hill RL. (2005). Otto Fritz Meyerhof and the elucidation of the glycolytic pathway. The Journal of biological chemistry 280, e3.

Kroemer G, Galluzzi L & Brenner C. (2007). Mitochondrial membrane permeabilization in cell death. Physiological reviews 87, 99-163.

Kuznetsov AV, Javadov S, Sickinger S, Frotschnig S & Grimm M. (2015). H9c2 and HL-1 cells demonstrate distinct features of energy metabolism, mitochondrial function and sensitivity to hypoxia-reoxygenation. Biochimica et biophysica acta 1853, 276-284.

Kwong JQ, Lu X, Correll RN, Schwanekamp JA, Vagnozzi RJ, Sargent MA, York AJ, Zhang J, Bers DM & Molkentin JD. (2015). The mitochondrial calcium uniporter selectively matches metabolic output to acute contractile stress in the heart. Cell reports 12, 15-22.

Labrousse AM, Zappaterra MD, Rube DA & van der Bliek AM. (1999). C. elegans dynamin-related protein DRP-1 controls severing of the mitochondrial outer membrane. Molecular cell 4, 815-826.

Legesse-Miller A, Massol RH & Kirchhausen T. (2003). Constriction and Dnm1p Recruitment Are Distinct Processes in Mitochondrial Fission. Molecular Biology of the Cell 14, 1953-1963.

Legros F, Lombes A, Frachon P & Rojo M. (2002). Mitochondrial fusion in human cells is efficient, requires the inner membrane potential, and is mediated by mitofusins. Molecular Biology of the Cell 13, 4343-4354.

Leung AWC, Varanyuwatana P & Halestrap AP. (2008). The Mitochondrial Phosphate Carrier Interacts with Cyclophilin D and May Play a Key Role in the Permeability Transition. Journal of Biological Chemistry 283, 26312- 26323.

Levi F, Chatenoud L, Bertuccio P, Lucchini F, Negri E & La Vecchia C. (2009). Mortality from cardiovascular and cerebrovascular diseases in Europe and other areas of the world: an update. European journal of cardiovascular prevention and rehabilitation : official journal of the European Society of Cardiology, Working Groups on Epidemiology & Prevention and Cardiac Rehabilitation and Exercise Physiology 16, 333-350. 225

Li K, Li Y, Shelton JM, Richardson JA, Spencer E, Chen ZJ, Wang X & Williams RS. (2000). Cytochrome c deficiency causes embryonic lethality and attenuates stress-induced apoptosis. Cell 101, 389-399.

Liu T, Yu R, Jin SB, Han L, Lendahl U, Zhao J & Nister M. (2013). The mitochondrial elongation factors MIEF1 and MIEF2 exert partially distinct functions in mitochondrial dynamics. Experimental cell research 319, 2893-2904.

Liu X, Kim CN, Yang J, Jemmerson R & Wang X. (1996). Induction of apoptotic program in cell-free extracts: requirement for dATP and cytochrome c. Cell 86, 147-157.

Lo MC, Aulabaugh A, Jin G, Cowling R, Bard J, Malamas M & Ellestad G. (2004). Evaluation of fluorescence-based thermal shift assays for hit identification in drug discovery. Analytical biochemistry 332, 153-159.

Lopaschuk GD, Collins-Nakai RL & Itoi T. (1992). Developmental changes in energy substrate use by the heart. Cardiovascular research 26, 1172-1180.

Lopaschuk GD, Ussher JR, Folmes CD, Jaswal JS & Stanley WC. (2010). Myocardial fatty acid metabolism in health and disease. Physiological reviews 90, 207-258.

Loson OC, Liu R, Rome ME, Meng S, Kaiser JT, Shan SO & Chan DC. (2014). The mitochondrial fission receptor MiD51 requires ADP as a cofactor. Structure (London, England : 1993) 22, 367-377.

Loson OC, Meng S, Ngo H, Liu R, Kaiser JT & Chan DC. (2015). Crystal structure and functional analysis of MiD49, a receptor for the mitochondrial fission protein Drp1. Protein science : a publication of the Protein Society 24, 386- 394.

Loson OC, Song Z, Chen H & Chan DC. (2013). Fis1, Mff, MiD49, and MiD51 mediate Drp1 recruitment in mitochondrial fission. Molecular Biology of the Cell 24, 659-667.

Lusby E, Fife KH & Berns KI. (1980). Nucleotide sequence of the inverted terminal repetition in adeno-associated virus DNA. Journal of virology 34, 402-409. 226

Ma H, Sumbilla CM, Farrance IK, Klein MG & Inesi G. (2004). Cell-specific expression of SERCA, the exogenous Ca2+ transport ATPase, in cardiac myocytes. American journal of physiology Cell physiology 286, C556-564.

Ma J & Sun F. (2014). Expression, purification, crystallization and preliminary crystallographic study of the cytoplasmic domain of the mitochondrial dynamics protein MiD51. Acta crystallographica Section F, Structural biology communications 70, 596-599.

Maack C, Cortassa S, Aon MA, Ganesan AN, Liu T & O'Rourke B. (2006). Elevated cytosolic Na+ decreases mitochondrial Ca2+ uptake during excitation- contraction coupling and impairs energetic adaptation in cardiac myocytes. Circulation research 99, 172-182.

Maack C & O’Rourke B. (2007). Excitation-contraction coupling and mitochondrial energetics. Basic research in cardiology 102, 369-392.

Malka F, Guillery O, Cifuentes-Diaz C, Guillou E, Belenguer P, Lombes A & Rojo M. (2005). Separate fusion of outer and inner mitochondrial membranes. EMBO reports 6, 853-859.

Marsboom G, Toth PT, Ryan JJ, Hong Z, Wu X, Fang YH, Thenappan T, Piao L, Zhang HJ, Pogoriler J, Chen Y, Morrow E, Weir EK, Rehman J & Archer SL. (2012). Dynamin-related protein 1-mediated mitochondrial mitotic fission permits hyperproliferation of vascular smooth muscle cells and offers a novel therapeutic target in pulmonary hypertension. Circulation research 110, 1484-1497.

Mathers CD & Loncar D. (2006). Projections of global mortality and burden of disease from 2002 to 2030. PLoS medicine 3, e442.

Matsushita T, Elliger S, Elliger C, Podsakoff G, Villarreal L, Kurtzman GJ, Iwaki Y & Colosi P. (1998). Adeno-associated virus vectors can be efficiently produced without helper virus. Gene therapy 5, 938-945.

Matulis D, Kranz JK, Salemme FR & Todd MJ. (2005). Thermodynamic stability of carbonic anhydrase: measurements of binding affinity and stoichiometry using ThermoFluor. Biochemistry 44, 5258-5266.

227

McLaughlin SK, Collis P, Hermonat PL & Muzyczka N. (1988). Adeno-associated virus general transduction vectors: analysis of proviral structures. Journal of virology 62, 1963-1973.

Mears JA, Lackner LL, Fang S, Ingerman E, Nunnari J & Hinshaw JE. (2011). Conformational changes in Dnm1 support a contractile mechanism for mitochondrial fission. Nature structural & molecular biology 18, 20-26.

Mehand MS, Srinivasan B & De Crescenzo G. (2015). Optimizing Multiple Analyte Injections in Surface Plasmon Resonance Biosensors with Analytes having Different Refractive Index Increments. Scientific reports 5, 15855.

Michael LH, Entman ML, Hartley CJ, Youker KA, Zhu J, Hall SR, Hawkins HK, Berens K & Ballantyne CM. (1995). Myocardial ischemia and reperfusion: a murine model. The American journal of physiology 269, H2147-2154.

Miller DG. (2011). AAV-mediated gene targeting. Methods in molecular biology (Clifton, NJ) 807, 301-315.

Mingozzi F & High KA. (2013). Immune responses to AAV vectors: overcoming barriers to successful gene therapy. Blood 122, 23-36.

Minicucci MF, Azevedo PS, Polegato BF, Paiva SAR & Zornoff LAM. (2011). Heart Failure After Myocardial Infarction: Clinical Implications and Treatment. Clinical Cardiology 34, 410-414.

Mitchell P. (1961). Coupling of phosphorylation to electron and hydrogen transfer by a chemi-osmotic type of mechanism. Nature 191, 144-148.

Mootha VK, Arai AE & Balaban RS. (1997). Maximum oxidative phosphorylation capacity of the mammalian heart. The American journal of physiology 272, H769-775.

Murphy E, Perlman M, London RE & Steenbergen C. (1991). Amiloride delays the ischemia-induced rise in cytosolic free calcium. Circulation research 68, 1250-1258.

228

Murphy E & Steenbergen C. (2008). Mechanisms underlying acute protection from cardiac ischemia-reperfusion injury. Physiological reviews 88, 581-609.

Muzyczka N. (1992). Use of adeno-associated virus as a general transduction vector for mammalian cells. Current topics in microbiology and immunology 158, 97-129.

Nallamothu BK, Bradley EH & Krumholz HM. (2007). Time to Treatment in Primary Percutaneous Coronary Intervention. New England Journal of Medicine 357, 1631-1638.

Nickel A, Loffler J & Maack C. (2013). Myocardial energetics in heart failure. Basic research in cardiology 108, 358.

Niesen FH, Berglund H & Vedadi M. (2007). The use of differential scanning fluorimetry to detect ligand interactions that promote protein stability. Nature protocols 2, 2212-2221.

Noll T, Koop A & Piper HM. (1992). Mitochondrial ATP-synthase activity in cardiomyocytes after aerobic-anaerobic metabolic transition. American Journal of Physiology - Cell Physiology 262, C1297-C1303.

Onay-Besikci A. (2006). Regulation of cardiac energy metabolism in newborn. Molecular and cellular biochemistry 287, 1-11.

Ong SB, Subrayan S, Lim SY, Yellon DM, Davidson SM & Hausenloy DJ. (2010). Inhibiting mitochondrial fission protects the heart against ischemia/reperfusion injury. Circulation 121, 2012-2022.

Osellame LD, Singh AP, Stroud DA, Palmer CS, Stojanovski D, Ramachandran R & Ryan MT. (2016). Cooperative and independent roles of the Drp1 adaptors Mff, MiD49 and MiD51 in mitochondrial fission. Journal of cell science 129, 2170-2181.

Otera H, Miyata N, Kuge O & Mihara K. (2016). Drp1-dependent mitochondrial fission via MiD49/51 is essential for apoptotic cristae remodeling. The Journal of Cell Biology 212, 531-544.

229

Otera H, Wang C, Cleland MM, Setoguchi K, Yokota S, Youle RJ & Mihara K. (2010). Mff is an essential factor for mitochondrial recruitment of Drp1 during mitochondrial fission in mammalian cells. The Journal of Cell Biology 191, 1141-1158.

Ottani F, Latini R, Staszewsky L, La Vecchia L, Locuratolo N, Sicuro M, Masson S, Barlera S, Milani V, Lombardi M, Costalunga A, Mollichelli N, Santarelli A, De Cesare N, Sganzerla P, Boi A, Maggioni AP & Limbruno U. (2016). Cyclosporine A in Reperfused Myocardial Infarction: The Multicenter, Controlled, Open-Label CYCLE Trial. Journal of the American College of Cardiology 67, 365-374.

Otto A. (1968). Excitation of nonradiative surface plasma waves in silver by the method of frustrated total reflection. Zeitschrift für Physik A Hadrons and nuclei 216, 398-410.

Pacak CA, Mah CS, Thattaliyath BD, Conlon TJ, Lewis MA, Cloutier DE, Zolotukhin I, Tarantal AF & Byrne BJ. (2006). Recombinant adeno-associated virus serotype 9 leads to preferential cardiac transduction in vivo. Circulation research 99, e3-9.

Pacher P, Csordas P, Schneider T & Hajnoczky G. (2000). Quantification of calcium signal transmission from sarco-endoplasmic reticulum to the mitochondria. The Journal of physiology 529 Pt 3, 553-564.

Pacher P, Thomas AP & Hajnóczky G. (2002). Ca<sup>2+</sup> marks: Miniature calcium signals in single mitochondria driven by ryanodine receptors. Proceedings of the National Academy of Sciences 99, 2380.

Palmer CS, Elgass KD, Parton RG, Osellame LD, Stojanovski D & Ryan MT. (2013). Adaptor proteins MiD49 and MiD51 can act independently of Mff and Fis1 in Drp1 recruitment and are specific for mitochondrial fission. The Journal of biological chemistry 288, 27584-27593.

Palmer CS, Osellame LD, Laine D, Koutsopoulos OS, Frazier AE & Ryan MT. (2011). MiD49 and MiD51, new components of the mitochondrial fission machinery. EMBO reports 12, 565-573.

230

Palty R, Silverman WF, Hershfinkel M, Caporale T, Sensi SL, Parnis J, Nolte C, Fishman D, Shoshan-Barmatz V, Herrmann S, Khananshvili D & Sekler I. (2010). NCLX is an essential component of mitochondrial Na+/Ca2+ exchange. Proceedings of the National Academy of Sciences of the United States of America 107, 436-441.

Papanicolaou KN, Khairallah RJ, Ngoh GA, Chikando A, Luptak I, O'Shea KM, Riley DD, Lugus JJ, Colucci WS, Lederer WJ, Stanley WC & Walsh K. (2011). Mitofusin-2 maintains mitochondrial structure and contributes to stress- induced permeability transition in cardiac myocytes. Molecular and cellular biology 31, 1309-1328.

Papanicolaou KN, Ngoh GA, Dabkowski ER, O'Connell KA, Ribeiro RF, Jr., Stanley WC & Walsh K. (2012). Cardiomyocyte deletion of mitofusin-1 leads to mitochondrial fragmentation and improves tolerance to ROS-induced mitochondrial dysfunction and cell death. American journal of physiology Heart and circulatory physiology 302, H167-179.

Park YM, Febbraio M & Silverstein RL. (2009). CD36 modulates migration of mouse and human macrophages in response to oxidized LDL and may contribute to macrophage trapping in the arterial intima. The Journal of clinical investigation 119, 136-145.

Pham AH, McCaffery JM & Chan DC. (2012). Mouse lines with photo-activatable mitochondria (PhAM) to study mitochondrial dynamics. Genesis (New York, NY : 2000) 50, 833-843.

Pike MM, Kitakaze M & Marban E. (1990). 23Na-NMR measurements of intracellular sodium in intact perfused ferret hearts during ischemia and reperfusion. The American journal of physiology 259, H1767-1773.

Piquereau J, Caffin F, Novotova M, Lemaire C, Veksler V, Garnier A, Ventura- Clapier R & Joubert F. (2013). Mitochondrial dynamics in the adult cardiomyocytes: which roles for a highly specialized cell? Frontiers in physiology 4, 102.

Piquereau J, Caffin F, Novotova M, Prola A, Garnier A, Mateo P, Fortin D, Huynh le H, Nicolas V, Alavi MV, Brenner C, Ventura-Clapier R, Veksler V & Joubert F. (2012). Down-regulation of OPA1 alters mouse mitochondrial morphology, PTP function, and cardiac adaptation to pressure overload. Cardiovascular research 94, 408-417.

231

Praefcke GJ & McMahon HT. (2004). The dynamin superfamily: universal membrane tubulation and fission molecules? Nature reviews Molecular cell biology 5, 133-147.

Prasad K-MR, Xu Y, Yang Z, Toufektsian M-C, Berr SS & French BA. (2007). Topoisomerase Inhibition Accelerates Gene Expression after Adeno- associated Virus-mediated Gene Transfer to the Mammalian Heart. Molecular Therapy 15, 764-771.

Prasad KM, Xu Y, Yang Z, Acton ST & French BA. (2011). Robust cardiomyocyte- specific gene expression following systemic injection of AAV: in vivo gene delivery follows a Poisson distribution. Gene therapy 18, 43-52.

Pryde KR & Hirst J. (2011). Superoxide is produced by the reduced flavin in mitochondrial complex I: a single, unified mechanism that applies during both forward and reverse electron transfer. The Journal of biological chemistry 286, 18056-18065.

Qiagen. (2010). Taq PCR Handbook, 3rd edn.

Rambold AS, Kostelecky B, Elia N & Lippincott-Schwartz J. (2011). Tubular network formation protects mitochondria from autophagosomal degradation during nutrient starvation. Proceedings of the National Academy of Sciences of the United States of America 108, 10190-10195.

Randle PJ, Denton RM, Pask HT & Severson DL. (1974). Calcium ions and the regulation of pyruvate dehydrogenase. Biochemical Society symposium, 75- 88.

Randle PJ, Garland PB, Hales CN & Newsholme EA. (1963). The glucose fatty-acid cycle. Its role in insulin sensitivity and the metabolic disturbances of diabetes mellitus. Lancet (London, England) 1, 785-789.

Rayaprolu V, Kruse S, Kant R, Venkatakrishnan B, Movahed N, Brooke D, Lins B, Bennett A, Potter T, McKenna R, Agbandje-McKenna M & Bothner B. (2013). Comparative Analysis of Adeno-Associated Virus Capsid Stability and Dynamics. Journal of virology 87, 13150-13160.

232

Reid RA, Moyle J & Mitchell P. (1966). Synthesis of adenosine triphosphate by a protonmotive force in rat liver mitochondria. Nature 212, 257-258.

Reimer KA, Hill ML & Jennings RB. (1981). Prolonged depletion of ATP and of the adenine nucleotide pool due to delayed resynthesis of adenine nucleotides following reversible myocardial ischemic injury in dogs. Journal of molecular and cellular cardiology 13, 229-239.

Reimer KA, Lowe JE, Rasmussen MM & Jennings RB. (1977). The wavefront phenomenon of ischemic cell death. 1. Myocardial infarct size vs duration of coronary occlusion in dogs. Circulation 56, 786-794.

Richter V, Palmer CS, Osellame LD, Singh AP, Elgass K, Stroud DA, Sesaki H, Kvansakul M & Ryan MT. (2014). Structural and functional analysis of MiD51, a dynamin receptor required for mitochondrial fission. The Journal of Cell Biology 204, 477-486.

Rodrigues JV, Prosinecki V, Marrucho I, Rebelo LP & Gomes CM. (2011). Protein stability in an ionic liquid milieu: on the use of differential scanning fluorimetry. Physical chemistry chemical physics : PCCP 13, 13614-13616.

Rojo M, Legros F, Chateau D & Lombes A. (2002). Membrane topology and mitochondrial targeting of mitofusins, ubiquitous mammalian homologs of the transmembrane GTPase Fzo. Journal of cell science 115, 1663-1674.

Ross R. (1999). Atherosclerosis--an inflammatory disease. The New England journal of medicine 340, 115-126.

Ross R & Glomset JA. (1973). Atherosclerosis and the arterial smooth muscle cell: Proliferation of smooth muscle is a key event in the genesis of the lesions of atherosclerosis. Science (New York, NY) 180, 1332-1339.

Rottman JN, Ni G & Brown M. (2007). Echocardiographic evaluation of ventricular function in mice. Echocardiography (Mount Kisco, NY) 24, 83-89.

Roubille F, Lairez O, Mewton N, Rioufol G, Ranc S, Sanchez I, Cung TT, Elbaz M, Piot C & Ovize M. (2012). Cardioprotection by clopidogrel in acute ST- elevated myocardial infarction patients: a retrospective analysis. Basic research in cardiology 107, 275. 233

Rouslin W & Broge CW. (1996). IF1 function in situ in uncoupler-challenged ischemic rabbit, rat, and pigeon hearts. The Journal of biological chemistry 271, 23638-23641.

Rouslin W, Broge CW & Grupp IL. (1990). ATP depletion and mitochondrial functional loss during ischemia in slow and fast heart-rate hearts. The American journal of physiology 259, H1759-1766.

Rubanyi GM, Romero JC & Vanhoutte PM. (1986). Flow-induced release of endothelium-derived relaxing factor. The American journal of physiology 250, H1145-1149.

Ruiz-Meana M, Garcia-Dorado D, Miro-Casas E, Abellan A & Soler-Soler J. (2006). Mitochondrial Ca2+ uptake during simulated ischemia does not affect permeability transition pore opening upon simulated reperfusion. Cardiovascular research 71, 715-724.

Russell DW & Hirata RK. (1998). Human gene targeting by viral vectors. Nature genetics 18, 325-330.

Samulski RJ, Chang LS & Shenk T. (1989). Helper-free stocks of recombinant adeno-associated viruses: normal integration does not require viral gene expression. Journal of virology 63, 3822-3828.

Schafer C, Ladilov YV, Siegmund B & Piper HM. (2000). Importance of bicarbonate transport for protection of cardiomyocytes against reoxygenation injury. American journal of physiology Heart and circulatory physiology 278, H1457- 1463.

Schinzel AC, Takeuchi O, Huang Z, Fisher JK, Zhou Z, Rubens J, Hetz C, Danial NN, Moskowitz MA & Korsmeyer SJ. (2005). Cyclophilin D is a component of mitochondrial permeability transition and mediates neuronal cell death after focal cerebral ischemia. Proceedings of the National Academy of Sciences of the United States of America 102, 12005-12010.

Schnepp BC, Jensen RL, Chen CL, Johnson PR & Clark KR. (2005). Characterization of adeno-associated virus genomes isolated from human tissues. Journal of virology 79, 14793-14803.

234

Senisterra GA, Markin E, Yamazaki K, Hui R, Vedadi M & Awrey DE. (2006). Screening for ligands using a generic and high-throughput light-scattering- based assay. Journal of biomolecular screening 11, 940-948.

Shaw JM & Nunnari J. (2002). Mitochondrial dynamics and division in budding yeast. Trends in Cell Biology 12, 178-184.

Sheffer R, Douiev L, Edvardson S, Shaag A, Tamimi K, Soiferman D, Meiner V & Saada A. (2016). Postnatal microcephaly and pain insensitivity due to a de novo heterozygous DNM1L mutation causing impaired mitochondrial fission and function. American journal of medical genetics Part A 170, 1603-1607.

Shen Q, Yamano K, Head BP, Kawajiri S, Cheung JT, Wang C, Cho JH, Hattori N, Youle RJ & van der Bliek AM. (2014). Mutations in Fis1 disrupt orderly disposal of defective mitochondria. Molecular Biology of the Cell 25, 145- 159.

Shiota T, Traven A & Lithgow T. (2015). Mitochondrial biogenesis: cell-cycle- dependent investment in making mitochondria. Current biology : CB 25, R78- r80.

Shu S, Liu X & Korn ED. (2005). Blebbistatin and blebbistatin-inactivated myosin II inhibit myosin II-independent processes in Dictyostelium. Proceedings of the National Academy of Sciences of the United States of America 102, 1472- 1477.

Silverman HS & Stern MD. (1994). Ionic basis of ischaemic cardiac injury: insights from cellular studies. Cardiovascular research 28, 581-597.

Simon MM, Greenaway S, White JK, Fuchs H, Gailus-Durner V, Wells S, Sorg T, Wong K, Bedu E, Cartwright EJ, Dacquin R, Djebali S, Estabel J, Graw J, Ingham NJ, Jackson IJ, Lengeling A, Mandillo S, Marvel J, Meziane H, Preitner F, Puk O, Roux M, Adams DJ, Atkins S, Ayadi A, Becker L, Blake A, Brooker D, Cater H, Champy MF, Combe R, Danecek P, di Fenza A, Gates H, Gerdin AK, Golini E, Hancock JM, Hans W, Holter SM, Hough T, Jurdic P, Keane TM, Morgan H, Muller W, Neff F, Nicholson G, Pasche B, Roberson LA, Rozman J, Sanderson M, Santos L, Selloum M, Shannon C, Southwell A, Tocchini-Valentini GP, Vancollie VE, Westerberg H, Wurst W, Zi M, Yalcin B, Ramirez-Solis R, Steel KP, Mallon AM, de Angelis MH, Herault Y & Brown SD. (2013). A comparative phenotypic and genomic analysis of C57BL/6J and C57BL/6N mouse strains. Genome biology 14, R82. 235

Simpson JC, Wellenreuther R, Poustka A, Pepperkok R & Wiemann S. (2000). Systematic subcellular localization of novel proteins identified by large-scale cDNA sequencing. EMBO reports 1, 287-292.

Sitia S, Tomasoni L, Atzeni F, Ambrosio G, Cordiano C, Catapano A, Tramontana S, Perticone F, Naccarato P, Camici P, Picano E, Cortigiani L, Bevilacqua M, Milazzo L, Cusi D, Barlassina C, Sarzi-Puttini P & Turiel M. (2010). From endothelial dysfunction to atherosclerosis. Autoimmunity reviews 9, 830-834.

Skulachev VP. (2001). Mitochondrial filaments and clusters as intracellular power- transmitting cables. Trends in biochemical sciences 26, 23-29.

Slager RE, Newton TL, Vlangos CN, Finucane B & Elsea SH. (2003). Mutations in RAI1 associated with Smith-Magenis syndrome. Nature genetics 33, 466- 468.

Smirnova E, Griparic L, Shurland DL & van der Bliek AM. (2001). Dynamin-related protein Drp1 is required for mitochondrial division in mammalian cells. Molecular Biology of the Cell 12, 2245-2256.

Smirnova E, Shurland DL, Ryazantsev SN & van der Bliek AM. (1998). A human dynamin-related protein controls the distribution of mitochondria. The Journal of Cell Biology 143, 351-358.

Smith AC, McGavran L, Robinson J, Waldstein G, Macfarlane J, Zonona J, Reiss J, Lahr M, Allen L & Magenis E. (1986). Interstitial deletion of (17)(p11.2p11.2) in nine patients. American journal of medical genetics 24, 393-414.

Smolina K, Wright FL, Rayner M & Goldacre MJ. (2012). Determinants of the decline in mortality from acute myocardial infarction in England between 2002 and 2010: linked national database study. BMJ (Clinical research ed) 344, d8059.

Song M, Mihara K, Chen Y, Scorrano L & Dorn GW. (2015). Mitochondrial fission and fusion factors reciprocally orchestrate mitophagic culling in mouse hearts and cultured fibroblasts. Cell metabolism 21, 273-285.

236

Sonntag F, Bleker S, Leuchs B, Fischer R & Kleinschmidt JA. (2006). Adeno- associated virus type 2 capsids with externalized VP1/VP2 trafficking domains are generated prior to passage through the cytoplasm and are maintained until uncoating occurs in the nucleus. Journal of virology 80, 11040-11054.

Steenbergen C, Murphy E, Watts JA & London RE. (1990). Correlation between cytosolic free calcium, contracture, ATP, and irreversible ischemic injury in perfused rat heart. Circulation research 66, 135-146.

Stojanovski D, Koutsopoulos OS, Okamoto K & Ryan MT. (2004). Levels of human Fis1 at the mitochondrial outer membrane regulate mitochondrial morphology. Journal of cell science 117, 1201-1210.

Stowell MH, Marks B, Wigge P & McMahon HT. (1999). Nucleotide-dependent conformational changes in dynamin: evidence for a mechanochemical molecular spring. Nature cell biology 1, 27-32.

Taegtmeyer H, Sen S & Vela D. (2010). Return to the fetal gene program: a suggested metabolic link to gene expression in the heart. Annals of the New York Academy of Sciences 1188, 191-198.

Taguchi N, Ishihara N, Jofuku A, Oka T & Mihara K. (2007). Mitotic phosphorylation of dynamin-related GTPase Drp1 participates in mitochondrial fission. The Journal of biological chemistry 282, 11521-11529.

Tenenbaum L, Lehtonen E & Monahan PE. (2003). Evaluation of risks related to the use of adeno-associated virus-based vectors. Current gene therapy 3, 545- 565.

Thygesen K, Alpert JS, White HD, Jaffe AS, Apple FS, Galvani M, Katus HA, Newby LK, Ravkilde J, Chaitman B, Clemmensen PM, Dellborg M, Hod H, Porela P, Underwood R, Bax JJ, Beller GA, Bonow R, Van der Wall EE, Bassand JP, Wijns W, Ferguson TB, Steg PG, Uretsky BF, Williams DO, Armstrong PW, Antman EM, Fox KA, Hamm CW, Ohman EM, Simoons ML, Poole-Wilson PA, Gurfinkel EP, Lopez-Sendon JL, Pais P, Mendis S, Zhu JR, Wallentin LC, Fernandez-Aviles F, Fox KM, Parkhomenko AN, Priori SG, Tendera M, Voipio-Pulkki LM, Vahanian A, Camm AJ, De Caterina R, Dean V, Dickstein K, Filippatos G, Funck-Brentano C, Hellemans I, Kristensen SD, McGregor K, Sechtem U, Silber S, Tendera M, Widimsky P, Zamorano JL, Morais J, Brener S, Harrington R, Morrow D, Lim M, Martinez-Rios MA, Steinhubl S,

237

Levine GN, Gibler WB, Goff D, Tubaro M, Dudek D & Al-Attar N. (2007). Universal definition of myocardial infarction. Circulation 116, 2634-2653.

Tondera D, Grandemange S, Jourdain A, Karbowski M, Mattenberger Y, Herzig S, Da Cruz S, Clerc P, Raschke I, Merkwirth C, Ehses S, Krause F, Chan DC, Alexander C, Bauer C, Youle R, Langer T & Martinou JC. (2009). SLP-2 is required for stress-induced mitochondrial hyperfusion. The EMBO Journal 28, 1589-1600.

Tousoulis D, Kampoli AM, Tentolouris C, Papageorgiou N & Stefanadis C. (2012). The role of nitric oxide on endothelial function. Current vascular pharmacology 10, 4-18.

Traaseth N, Elfering S, Solien J, Haynes V & Giulivi C. (2004). Role of calcium signaling in the activation of mitochondrial nitric oxide synthase and citric acid cycle. Biochimica et biophysica acta 1658, 64-71.

Tsukamoto K, Kinoshita M, Kojima K, Mikuni Y, Kudo M, Mori M, Fujita M, Horie E, Shimazu N & Teramoto T. (2002). Synergically increased expression of CD36, CLA-1 and CD68, but not of SR-A and LOX-1, with the progression to foam cells from macrophages. Journal of atherosclerosis and thrombosis 9, 57-64.

Twig G, Elorza A, Molina AJA, Mohamed H, Wikstrom JD, Walzer G, Stiles L, Haigh SE, Katz S, Las G, Alroy J, Wu M, Py BF, Yuan J, Deeney JT, Corkey BE & Shirihai OS. (2008). Fission and selective fusion govern mitochondrial segregation and elimination by autophagy. The EMBO Journal 27, 433-446.

van Laake LW, Passier R, Monshouwer-Kloots J, Nederhoff MG, Ward-van Oostwaard D, Field LJ, van Echteld CJ, Doevendans PA & Mummery CL. (2007). Monitoring of cell therapy and assessment of cardiac function using magnetic resonance imaging in a mouse model of myocardial infarction. Nature protocols 2, 2551-2567.

Van Vliet KM, Blouin V, Brument N, Agbandje-McKenna M & Snyder RO. (2008). The role of the adeno-associated virus capsid in gene transfer. Methods in molecular biology (Clifton, NJ) 437, 51-91.

Vandendriessche T, Thorrez L, Acosta-Sanchez A, Petrus I, Wang L, Ma L, L DEW, Iwasaki Y, Gillijns V, Wilson JM, Collen D & Chuah MK. (2007). Efficacy and 238

safety of adeno-associated viral vectors based on serotype 8 and 9 vs. lentiviral vectors for hemophilia B gene therapy. Journal of thrombosis and haemostasis : JTH 5, 16-24.

Vanstone JR, Smith AM, McBride S, Naas T, Holcik M, Antoun G, Harper ME, Michaud J, Sell E, Chakraborty P, Tetreault M, Majewski J, Baird S, Boycott KM, Dyment DA, MacKenzie A & Lines MA. (2016). DNM1L-related mitochondrial fission defect presenting as refractory epilepsy. European journal of human genetics : EJHG 24, 1084-1088.

Vasileva A & Jessberger R. (2005). Precise hit: adeno-associated virus in gene targeting. Nature reviews Microbiology 3, 837-847.

Wan B, LaNoue KF, Cheung JY & Scaduto RC, Jr. (1989). Regulation of citric acid cycle by calcium. The Journal of biological chemistry 264, 13430-13439.

Wang JX, Jiao JQ, Li Q, Long B, Wang K, Liu JP, Li YR & Li PF. (2011). miR-499 regulates mitochondrial dynamics by targeting calcineurin and dynamin- related protein-1. Nature medicine 17, 71-78.

Wang JX, Li Q & Li PF. (2009). Apoptosis repressor with caspase recruitment domain contributes to chemotherapy resistance by abolishing mitochondrial fission mediated by dynamin-related protein-1. Cancer research 69, 492-500.

Waterham HR, Koster J, van Roermund CWT, Mooyer PAW, Wanders RJA & Leonard JV. (2007). A Lethal Defect of Mitochondrial and Peroxisomal Fission. New England Journal of Medicine 356, 1736-1741.

Watt IN, Montgomery MG, Runswick MJ, Leslie AG & Walker JE. (2010). Bioenergetic cost of making an adenosine triphosphate molecule in animal mitochondria. Proceedings of the National Academy of Sciences of the United States of America 107, 16823-16827.

Wentz AE, d'Avignon DA, Weber ML, Cotter DG, Doherty JM, Kerns R, Nagarajan R, Reddy N, Sambandam N & Crawford PA. (2010). Adaptation of myocardial substrate metabolism to a ketogenic nutrient environment. The Journal of biological chemistry 285, 24447-24456.

239

Wistuba A, Kern A, Weger S, Grimm D & Kleinschmidt JA. (1997). Subcellular compartmentalization of adeno-associated virus type 2 assembly. Journal of virology 71, 1341-1352.

Wong ED, Wagner JA, Gorsich SW, McCaffery JM, Shaw JM & Nunnari J. (2000). The Dynamin-Related Gtpase, Mgm1p, Is an Intermembrane Space Protein Required for Maintenance of Fusion Competent Mitochondria. The Journal of Cell Biology 151, 341-352.

Xiao D, Wang Z & She M. (1999). Minimally modified low-density lipoprotein induces monocyte chemotactic protein-1 expression in vivo and a novel model for monocyte adhesion to arterial intima. Chinese medical journal 112, 438-442.

Xiao X, Hu Y, Quiros PM, Wei Q, Lopez-Otin C & Dong Z. (2014). OMA1 mediates OPA1 proteolysis and mitochondrial fragmentation in experimental models of ischemic kidney injury. American journal of physiology Renal physiology 306, F1318-1326.

Xu S, Cherok E, Das S, Li S, Roelofs BA, Ge SX, Polster BM, Boyman L, Lederer WJ, Wang C & Karbowski M. (2016). Mitochondrial E3 ubiquitin ligase MARCH5 controls mitochondrial fission and cell sensitivity to stress-induced apoptosis through regulation of MiD49 protein. Molecular Biology of the Cell 27, 349-359.

Yan Z, Zak R, Zhang Y & Engelhardt JF. (2005). Inverted Terminal Repeat Sequences Are Important for Intermolecular Recombination and Circularization of Adeno-Associated Virus Genomes. Journal of virology 79, 364-379.

Yellon DM & Hausenloy DJ. (2007). Myocardial reperfusion injury. The New England journal of medicine 357, 1121-1135.

Yoon Y, Pitts KR & McNiven MA. (2001). Mammalian dynamin-like protein DLP1 tubulates membranes. Molecular Biology of the Cell 12, 2894-2905.

Yvan-Charvet L, Ranalletta M, Wang N, Han S, Terasaka N, Li R, Welch C & Tall AR. (2007). Combined deficiency of ABCA1 and ABCG1 promotes foam cell accumulation and accelerates atherosclerosis in mice. The Journal of clinical investigation 117, 3900-3908.

240

Zepeda R, Kuzmicic J, Parra V, Troncoso R, Pennanen C, Riquelme JA, Pedrozo Z, Chiong M, Sanchez G & Lavandero S. (2014). Drp1 loss-of-function reduces cardiomyocyte oxygen dependence protecting the heart from ischemia- reperfusion injury. Journal of cardiovascular pharmacology 63, 477-487.

Zhao J, Liu T, Jin S, Wang X, Qu M, Uhlen P, Tomilin N, Shupliakov O, Lendahl U & Nister M. (2011). Human MIEF1 recruits Drp1 to mitochondrial outer membranes and promotes mitochondrial fusion rather than fission. The EMBO Journal 30, 2762-2778.

Zhou H, Hu S, Jin Q, Shi C, Zhang Y, Zhu P, Ma Q, Tian F & Chen Y. (2017a). Mff‐ Dependent Mitochondrial Fission Contributes to the Pathogenesis of Cardiac Microvasculature Ischemia/Reperfusion Injury via Induction of mROS‐ Mediated Cardiolipin Oxidation and HK2/VDAC1 Disassociation‐Involved mPTP Opening. Journal of the American Heart Association: Cardiovascular and Cerebrovascular Disease 6, e005328.

Zhou L, Li R, Liu C, Sun T, Htet Aung LH, Chen C, Gao J, Zhao Y & Wang K. (2017b). Foxo3a inhibits mitochondrial fission and protects against doxorubicin-induced cardiotoxicity by suppressing MIEF2. Free radical biology & medicine 104, 360-370.

Zou H, Li Y, Liu X & Wang X. (1999). An APAF-1.cytochrome c multimeric complex is a functional apoptosome that activates procaspase-9. The Journal of biological chemistry 274, 11549-11556.

241